DEVELOPMENTS IN FOOD ENGINEERING

Proceedings of the 6th International Congress

on Engineering and Food
Proceedings of the Sixth International Congress on Engineering and Food
23-27 May 1993, Makuhari Messe, Chiba, Japan.

ORGANIZING COMMITfEE

Chairman YANO, Toshimasa Prof. Emeritus, Univ. of Tokyo

General Secretary NAKAMURA, Kozo Prof., Univ. of Tokyo
Program Chairman MATS UNO, Ryuichi Prof., Kyoto Univ.
Advisor TOEI, Ryozo Prof. Emeritus, Kyoto Univ.
Advisor FUJIMAKI, Masao Prof. Emeritus, Univ. of Tokyo

ADACHI, Shuji Kyoto Univ.

ARAI, Soichi Univ. of Tokyo
FUJIO, Yusaku Kyushu Univ.
FUKUSHIMA, Masayoshi Snow Brand Milk Products Co., Ltd.
IMAMURA, Kazuo FAIS
KAWADE, Hiroyuki Fuji Oil Co., Ltd.
KIMURA, Shoji Univ. of Tokyo
KOBAYASHI, Takeshi Nagoya Univ.
KUBOTA, Kiyoshi Hiroshima Univ.
MIYA W AKI, Osato Univ. of Tokyo
NAKANISHI, Kazuhiro Okayama Univ.
NOGUCHI, Akinori National Food Research Institute
OHKUBO, Yukima Ajinomoto Co., Inc.
OHSAKI, Katsumichi Kikkoman Corp.
SAGARA, Yasuyuki Univ. of Tokyo
WATANABE, Atsuo TOTO Ltd.
WATANABE, Hisahiko Tokyo Univ. of Fisheries

INTERNATIONAL ADVISORY COMMITfEE

Bimbenet, J.J. (France)

Bruin, S. (The Netherlands)
Earl, RL. (New Zealand)
Farkas, D. (USA)
Filka, P. (Czech)
Hallstrom, B. (Sweden)
Jowitt, R (UK)
LeMaguer, M. (Canada)
Lewicki, P. (poland)
Unko, P. (Finland)
Lund, D. (USA)
Martin, A.M. (Canada)
McKenna, B. (Ireland)
Schubert, H. (Germany)
Spiess, W.E.L. (Germany)
Yano, T. (Japan, Chairman)
 DEVELOPMENTS IN FOOD ENGINEERING

Proceedings of the 6th International Congress

on Engineering and Food

Edited by

Professor Toshimasa Yano (Professor Emeritus, University of Tokyo),

Department of Bioengineering, Faculty of Engineering,
Yokohama National University,
Tokiwadai, Hodogaya-ku, Yokohama 240, Japan

Professor Ryuichi Matsuno,

Department of Food Science and Technology,
Faculty of Agriculture, Kyoto University,
Sakyo-ku, Kyoto 606-01, Japan

and

Professor Kozo Nakamura

Department of Agricultura/ Chemistry,
Faculty of Ag riculture, University of Tokyo,
Yayoi, Bunkyo-ku, Tokyo 113, Japan

Springer-Science+Business Media, B.V.

 First edition 1994

© 1994 Springer Science+ Business Media Dordrecht

Originally published by Chapman aud Hali 1994
Softcover reprint of the hardcover 1st edition 1994

ISBN 978-1-4613-6149-7 ISBN 978-1-4615-2674-2 (eBook)

DOI 10.1007/978-1-4615-2674-2
1 work in 2 parts, not available separately

Apart from any fair dealing for the purposes of research or private study, or
criticism or review, as permitted under the UK Copyright Designs and Patents
Act, 1988, this publication may not be reproduce<!, stored, or transmitted, in
any form or by any means, without the prior permission in writing of the
publishers, or in the case of reprographic reproduction only in accordance with
the terms of the licences issued by the Copyright Licensing Agency in the UK,
or in accordance with the terms of licences issued by the appropriate
Reproduction Rights Organization outside the UK. Enquiries conceming
reproduction outside the terms stated here should be sent to the publishers at
the Glasgow address printed on this page.
The publisher makes no representation, express or implied, with regard to
the accuracy of the information contained in this book and cannot accept any
legal responsibility or liability for any errors or omissions that may be made.
A catalogue record for this book is available from the British Library
Library of Congress Catalog Card Number: 94-70986

@ Printed on acid-free text paper, manufactured in accordance with ANSIINISO

Z39.48-1992 (Permanence of Paper).
 PREFACE

The necessity of prediction and fine control in the food manufacturing process is
becoming more important than ever before, and food researchers and engineers
must confront difficulties arising from the specificity of food materials and the
sensitivity of human beings to taste. Fortunately, an overview of world research
reveals that the mechanisms of the many complex phenomena found in the food
manufacturing process have been gradually elucidated by skilful experiments
using new analytical tools, methods and theoretical analyses.
This book, the proceedings of the 6th International Congress on Engineering
and Food (ICEF6), held for the first time in Asia - in Chiba, Japan May 23 - 27,
1993 - summarizes the frontiers of world food engineering in 1993. Congress
was joined by the 4th International Conference on Fouling and Cleaning. There
were 476 active members from 31 countries participating in the Congress. The
editors hope that readers will find this book to be a useful review of the current
state of food engineering, and will consider future developments in this research
field.
The editors extend thanks to the members of the organizing committee of
ICEF6, and the advisors, Dr. Ryozo Toei, Professor Emeritus of Kyoto
University and Dr. Masao Fujimaki, Professor Emeritus of the University of
Tokyo. They also acknowledge the international advisory board members who
helped the organizing committee in many ways, and the 10 foundations and 66
companies that financially supported the ICEF6. Finally, the editors are indebted
to the reviewers of the manuscripts of these proceedings.

September, 1993
Toshimasa Yano
Ryuichi Matsuno
Kozo Nakamura
 ACKNOWLEDGEMENTS
The Sixth International Congress on Engineering and Food has been financially
supported by the following foundations and companies.
The Commemorative Association for
the Japan World Exposition (1970)
Chiba Convention Bureau
Hosokawa Powder Technology
Foundation
Nestle Science Promotion Committee
The Kajima Foundation
The Iijima Memorial Foundation for
the Promotion of Food Science and
Technology
The Iwatani Naoji Foundation
The Naito Foundation
TOKYO OHKA Foundation for the
Promotion of Science and
Technology
The Asahi Glass Foundation
Advance Co., Ltd.
Ajikan Co., Ltd.
Ajinomoto Co., Inc.
APV Crepaco Far East, Inc.
Asahi Breweries Ltd.
Calbee Foods Co., Ltd.
Dai Nippon Printing Co., Ltd.
Daido Steel Co., Ltd.
Dainippon Ink & Chemicals Co., Ltd.
The Federation of Electric Power
Companies
Fuji Electric Co., Ltd.
Fuji Oil Co., Ltd.
Gekkeikan Sake Co., Ltd.
Godo Steel Ltd.
Hayashibara Co., Ltd.
Hokkaido Sugar Co., Ltd.
House Food Industry Corp., Ltd.
Izumi Food Machinary Co., Ltd.
Japan Automobile Manufacturer's
Association, Inc.
The Japan Steel, Ltd.
Japan Tobacco Inc.
K F Engineering Co., Ltd.
Kagome Co., Ltd.
Kamewada Technical Consulting
Engineers Office
Kaneka Corporation
Kawasaki Steel Corporation
Kikkoman Corporation
Kirin Brewery Co., Ltd.
Kobe Steel Ltd.
Kohjin Co., Ltd.
Kubota Corporation
Kurimoto, Ltd.
Kyowa Hakko Kogyo Co., Ltd.
Matsushita Electric Industrial Co., Ltd.
Meiji Milk Products Co., Ltd.
Meiji Seika Kaisha, Ltd.
Membrance Researcn Circle of Food
Mitsubishi Steel Manufacturing Co.,
Ltd.
Miura Co., Ltd.
Momoya Co., Ltd.
Morinaga Engineering Co., Ltd.
Morinaga Milk Industry Co., Ltd.
Morishita Jintan Co., Ltd.
Nakayama Steel Works Ltd.
Nippon Dairy Co., Ltd.
Nippon Kokan K. K.
Nippon Lever B. V.
Nippon Steel Corporation
viii
Nippon Sui san Kaisha, Ltd.
Nisshin Flour Milling Co., Ltd.
Nisshin Steel Co., Ltd.
NOF Corporation
Research Conference of Membrane
Application
Sapporo Brewery Co., Ltd.
Satake Corporation
Senbatoka Industry Co., Ltd.
Snow Brand Milk Products Co., Ltd.
Sumitomo Metal Industries Ltd.
Suntory Ltd.
Tokyo Banker's Association, Inc.
Topy Industries, Ltd.
Toshiba Corporation
TOTO Ltd.
Toyo Seikan Kaisha, Ltd.
Wado Doctor's Group
Yamazaki Baking Co., Ltd.
Yodogawa Steel Works Ltd.
 CONTENTS

Keynote Lecture

Food Engineering - From arts to science/ T. Yano 1

Plenary Lectures

Mechanical emulsification! H. Schubert, H. Karbstein 9

Applications of NMR to food science and engineering! P.S. Belton 15

Progress in pasteurization and sterilization! T. Ohlsson 18

Fouling and cleaning: Mechanisms and models/ PJ. Fryer,

M.T. Belmar-Beiny, PJ.R. Schreier 24
Supercritical fluids: Fundamentals and applications/ P.G. Debenedetti 30

Neural networks in bioprocess engineering! Y.-H. Zhu, S. Linko,

T. Eerikainen, P. Linko 36

Implementation of model based optimization of product quality

parameters in dairy processes/ 1. van der Linden 42
Physiological functions of foods/ S. Arai 48

Physical and Physicochemical Properties of Food

Estimation of double layers flow model for fluidity of milk in a

capillary/ T. Nakamura, A. Yamamoto, K. Sakanishi, T. Mineshita 54
Effect of moisture content on flow characteristics of SPI melt at an
elevated temperature/ N. Hayashi, I. Hayakawa, Y. Fujio 57

Novel approach to characterize the flow properties of fine dry food

x
powders/ H.D. Kono, Y. Itani 60

Viscosity measurements of fresh and heat stressed frying oil at high

temperatures/ KS. Miller, B.E. Farkas, R.P. Singh 63

Viscosity properties by amylograph of rice/ H. Kurasawa, I. Shoji 66

Physicochemical properties and hypolipidemic function of levan and its

partial hydrolysate/ M. Takahashi, R. Takayama, K Ishihara 69

The use of xylanolytic enzymes in processing of cereals/ K Poutanen,

H. Harkonen, T. Parkkonen, K Autio, T. Suortti, A. Kantelinen,
L. Viikari 72

Viscoelasticity of butter/ H. Hayashi 75

Viscoelastic characteristics of bread crumb and graphic analysis of

compression creep curve/ Y. Wang, H. Morishima, Y. Seo, Y. Sagara,
K Imou 78

Viscoelastic behavior of parboiled rice/ S.M. Saif, D.A. Suter 81

Stress relaxation characteristics of alfalfa cubes/ R.T. Patil,

S. Sokhansanj 84

Studies on rheological properties of foods/ Y. Masumoto, K Kubota,

y. Hagura 87

Rheology of set type and stirred type yoghurt: build-up, break-down

and recovery; the effects of pH, temperature and starter/ E. Ronnegard,
J. Jiang, P. Dejmek 90

Monitoring of milk curd formation by means of physical properties/

H. Nakanuma, T. Doi, R. Watanabe 93

Reological analysis of jelly strength in heat induced gel like kamaboko/

H. Miki, J. Shindo, J. Nishimoto 96

Relationships of rheological and hydrodynamic properties of

biopolymeric ingredients to the composite surimi gel texture/ C.M. Lee 99

Hypocaloric jams from grape juice/ E.C. Matias, 1M.N. Sousa,

D. Laureano 102

Mechanical properties of foods near the sol-gel transition point!

H. Kumagai, T. Inukai, T. Fujii, T. Yano 105

Effects of sugars on the gel-sol transition of agarose and

K-carrageenan! K Nishinari, T. Takaya, K Kohyama, M. Watase 108
 Xl

Rheological changes during sol-gel transition of pectin-sucrose/fructose

dispersions/ M.A. Rao, J.P. Van Buren, H.J. Cooley 111

Critical behavior of agarose near the sol-gel transition point! T. Fujii,

H. Kuma~ai, T. Yano 114

Microstructure of hydrogel and its relationship with molecular-sieve

characteristics in gel electrophoresis/ S. Miura, T. Fujii, H. Kumagai,
T. Yano 117

Physicochemical studies on gelation of soybean 7S and 11S proteins by

glucono-()-Iactone/ K. Kohyama, K. Nishinari 120

Rheological models for shelled maize en-masse/ L.O. Gumbe 123

Nondestructive texture measurement of apples/ S. Prussia, K. Morita,

Y. Hung, C. Thai, W. Tollner 126

Description of bread textural deterioration by stress-strain curves,

recoverable work, and thermal transitions/ P. Chinachoti 129

Fracture stress of frozen soybean curd/ H. Watanabe, C.Q. Tang,

T. Mihori 132

The influence of moisture content on the failure mode of biscuits/

L. Piazza, R. Bringiotti, P. Masi 134

Changes of physical properties of drying materials/ P.P. Lewicki,

D. Witrowa, W. Pomaranska-Lazuka 137

Effects of baking and storage conditions on mechanical properties of

white bread/ Y. Wang, H. Morishima, Y. Seo, Y. Sagara, K. Imou 140

Impedance spectroscopic analysis in agricultural products/ K. Toyoda 143

Finite element analysis on the effective thermal conductivity of

dispersed systems/ T. Sakiyama, T. Yano 146

Measurement of fraction of frozen water and thermal conductivity in

frozen food materials/ R. Pongsawatmanit, O. Miyawaki 149

Thermophysical properties of raspberries and apples/ V. Bifani,

D. Borcoski, P. Moyano, F. Osorio 152

A thermomechanical method for determining collapse temperatures of

food powders/ Y.T. Huang 155

Thermal studies of natural fruits/ M.M. Sa, A.M. Sereno 158

Thermal conversion of starch at low water activity/ K. Poutanen,

xii
O. Myllymaki, K. Autio, P. Myllarinen, T. Suortti 161

Characterization of W/O/W emulsions with a view to possible food

applications/ S. Matsumoto 164

Application of membrane emulsification method for preparing food

emulsions and emulsion characteristics/ K. Suzuki, 1 Shuto,
Y. Hagura 167

Evaluation of model food emulsion stability by hygrometric

measurements/ C.R. Lerici, C. Corradini, P. Pittia 170

Addition of soy lysophospholipids enhances emulsifying ability of

food emulsifiers/ S. Fujita, A. Suzuki, H. Kawai 173

Changes in water activity and functional properties of protein

hydrolysates/ H. Kumagai, H. Seto, H. Kumagai, H. Sakurai,
K. Ishii, S. Arai 176

Water activity and water behavior ·of soy sauce, dehydrated soy sauce
and the improvement on hygroscopicity of dehydrated soy sauce/
M. Hamano 179

Sorption isotherms and heats of sorption for fruit jams/ M.M. Sa,
A.M. Sereno 182

Water sorption isotherms of food model mixtures/ P.P. Lewicki,

W. Pomarariska-Lazuka 185

Glass transitions and physical state of dehydrated milk products/

Y. Roos, K. Jouppila 188

Stickiness and caking of infant formula food powders related to the

glass transition phenomena! L.E. Chuy, T.P. Labuza 191

Tentative of inducing fruit sugar crystallization during freezing to

reduce the hygroscopicity of the corresponding freeze-dried powders/
S. De Melo, T. Giarola , J. Cal-Vidal 194

Diffusivity of water in flour/raisin mixtures/ A.E. Kostaropoulos,

V.T. Karathanos, G.D. Saravacos 197

The use of digitized video images for monitoring color and color
evolution of Jonagold apples during shelf life/ F. Vervaeke,
E. Schrevens, J. Verreydt, K. Portier, J. De Baerdemaeker 200

Computerized image analysis of gas cell distribution in several foods/

S. Gohtani, N. Ariuchi, S. Kawasome, Y. Yamano 203

Functional properties of Lupinus luteus proteins/ 1M.N. Sousa,

 xiii
M.L. Beirao da Costa, S.E. Hill, J.R. Mitchell, S.E. Harding 206
Stability of glutathione in solution! T. Yamamoto, K Ishihara 209
Aroma of Thymus zygis. Chemical and sensorial analysis/
M. Moldao-Martins, G. Bernardo-Gil, M.L. Beirao da Costa 212

Application of NMR to Food Engineering

Measurement of material properties in heterogeneous food systems by

magnetic resonance imaging! M.J. McCarthy, KL. McCarthy,
R.L. Powell, J.D. Seymour, T.-Q. Li 215

Analysis of the flow field in a single screw extruder using magnetic

resonance imaging! C.K Agemura, R.I. Kauten, KL. McCarthy 218

Determination and calculation of freezing equilibria in frozen foods/

M. l..otz, H. Weisser 221

The characteristics of the hydration water of protein in relation with the

freezing and drying denaturation! N. Hanafusa 224

NMR study of the dynamic behaviour of water in food systems: Guar

galactomannan solutions and pectin gels/ E. Brosio, A. D'Ubaldo,
B. Verzegnassi 227

NMR-measurements to study binding states and mobility of water in

cellulose packaging materials/ F. Liebenspacher, H. Weisser 230

Heat capacity modeling of frozen food gels from NMR determination

of water states/ P. Comillon, J. Andrieu, J.e. Duplan, M. Laurent 233

Measurement of moisture diffusion in soybean seed by pulsed-field-

gradient NMR method/ H. Watanabe, M. Fukuoka, S. Shimada 236

Moisture diffusion in fresh and dehydrated fish flesh measured by

PFG-NMR/ M. Fukuoka, W. Wen, T. Mihori, H. Watanabe 239

IH-NMR images and localized spectra of foods/ N. Ishida, H. Kano,

H. Ogawa 241

Mechanical Processing of Food

Investigation of jet agglomeration processes/ H. Schuchmann,

H. Schubert 244

The effect of interparticle forces on the separation of fine powders from

xiv
gas-solid two phase flow/ H.O. Kono, T. Hikosaka 247

An application of cryo-shattering for separating low-fat meat from fatty

meat! Y. Hagura, K. Suzuki, K. Kubota 250

Utilization of brittle fracture properties of frozen foodstuffs/

K. Okamoto, Y. Hagura, K. Suzuki, K. Kubota 253

Mechanical disintegration of microorganisms by wet milling and

high-pressure homogenization - a comparative study/ M. Pittroff,
H. Schubert 256

Development of micro cut blender/ T. Hosokawa, H. Omura,

T. Takeuchi 259

Simulation in food plant design using SIMAN/ J.P. Clark, S.M. Hyman,
R. Symns, R Kelley 262

Intelligent mixer for bread dough! Y. Torikata, N. Ban 265

Effects of L-ascorbic acid and related compounds on gluten and

starch during breadmaking in a home baking machine/ L. Zhang,
N. Morita, M. Takagi 268

Analysis of continuous whipping of cream/ M. Kikuchi, M. Endo,

N. Yanagihara, T. Miyamoto, R. Watanabe, S. Matsumoto 271

Controlled biaxial extension of food polymer gels/ M.A. Tung,

1.1. Britt, J. Tang 274

Process of formation of porous structure of baked foods during heating!

Y. Shimiya, K. Sasaki 277

Reduction in damage to dried sultana during removal of cap-stems/

M.R Mollah, RJ. Hayes, P.R. Franz, I.V. Gould 280

Thermal and Mass Transfer Operations of Food

Thermal property measurements of fried foods using differential
scanning calorimeter/ A.B. Buhri, R.P. Singh 283

Colorimetric measurements of meat products cooked using different

systems/ P. Pittia, M. Anese, C. Orlando, A. Sensidoni 286

Interactions between maillard reaction products and lipid oxidation

during the roasting process and storage of hazel-nuts (Cory/us
aveUana)/ C. Severini, P. Pittia, M.C. Nicoli, G.G. Pinnavaia 289

A new method of monitoring particle motion and temperature in a

 xv
liquid/particle system/ H. Sawada, R.L. Merson 292

Co-rotating disc scraped-surface heat exchanger for food processing!

A Friis, J. Adler-Nissen, O. Hassager 295

Heat transfer in bakery ovens/ R.E. Altomare 298

Modelling and simulation of food frying processes/ G.S. Mittal,

P. Ateba 301

Modelling pasta cooking processes/ L. Piazza, M. Riva, P. Masi 304

Ohmic thawing of shrimp blocks/ M.O. Balaban, T. Henderson,

A Teixeira, W.S. Otwell 307

Ohmic heating of food materials - Effects of frequency on the heating

rate of fish protein! K. Uemura, A Noguchi, SJ. Park, D.U. Kim 310

Heat transfer analysis in food heated by far-infrared radiation!

N. Sakai, T. Hanzawa 313

Optical characteristics of vegetables in far-infrared irradiation heating!

M. Shimizu, A Hashimoto, S. Oshita 316

Temperature distribution in microwave oven heating - The influence

of different cavity modes/ T. Ohlsson, P.O. Risman 319

In-flow microwave heating of pumpable foods/ T. Ohlsson 322

Mathematical modeling of batch heating of liquids in a microwave

cavity/ AK. Datta, H.M. Prosetya, W. Hu 325

Measurement of evaporation coefficient of water during vacuum cooling

of lettuce/ AH. Tambunan, Y. Seo, Y. Sagara, H. Morishima,
y. Kawagoe 328

Heat and mass transfer modelling during beef carcass chilling for
quality control! G.S. Mittal, P. Mallikarjunan 331

Com drying: Modelling the quality degradation! F. Courtois,

A Lebert, J.e. Lasseran, J.J. Bimbenet 334

Puff drying of foods in a f1uidised bed/ N.C. Shilton, K. Niranjan 337

Estimation of the effective moisture diffusivity from drying data.

Application to some vegetables/ C.T. Kiranoudis, Z.B. Maroulis,
D. Marinos-Kouris, G.D. Saravacos 340

Effect of freezing conditions on vapor permeability of freeze-dried

foods/ K. Nakamura, K. Matagi, T. Odawara, T. Hoshino 343
xvi
Moisture content distribution and moisture diffusion in foods during
heating! C. Skjoldebrand, K. Thorvaldsson 346

Mass transfer in osmotic processes/ Z. Yao, M. Le Magner 349

The sorption of water in the agricultural materials/ T. Kameoka,

K. Horibe 352

The diffusion of seasoning substances in various slurries - Application

of nondestructive measurement using ultrasonic technique-/ T. Torii,
S.Odake 355

Diffusion of salt in dry-salted feta cheese/ S. Yanniotis,

J. Zarmpoutis, E. Anifantakis 358

Extraction rate for soluble solids during coffee brewing! S. Oshita,

A. Hashimoto, T. Miyata, T. Okado 361

The measurement of the losses of soluble solids in legume seeds (Lablat

purpureos) during the process of soaking at different temperatures and
in the process of cooking! A. Jadan, M. Silva, M. Moscoso 364

Evaluation of steam distillation of roasted coffee for quality

improvement of soluble coffee/ N. Imura, O. Matsuda 367

Phase Change Operation

Studies on the recrystallization of ice in frozen food systems/ S.G. Min,

W.E.L. SpieS 370

Ice structure and its control in frozen food gels/ O. Miyawaki, S.K. Bae 373

A comparative study of freezing time prediction for food products/

F. Kluza, W.E.L. SpieS 376

Variation of correction factors in an approximate equation for freezing

and thawing time prediction! A. Rubiolo de Reinick 379

Progressive freeze-concentration and its relationship with the ice

structure at freezing front! S.K. Bae, O. Miyawaki 382

Concentration and Dehydration Processes

Freeze concentration of some heat sensitive solutions/ O. Flesland,

X. Song, I. Str<jlmmen 385

Studies on the freeze concentration of foods. - Determination of

eutectic temperatures of amino acids in aqueous ethanol solution -/
 xvii
M. Shibata, K. Ishikawa, Y. Fukuta, S. Ishikawa, M. Shimoyamada,
K. Watanabe 388

Ice crystals agglomerated in freeze concentration! Y. Shirai,

K. Nakanishi, R. Matsuno 391

Thermodynamic modeling of the osmotic concentration process/

R.N. Biswal, M. Le Maguer 394

Recent advances in dewatering and impregnation soaking processes/

AL. Raoult...;Wack, R. Saurel, G.M. Rios, S. Guilbert 397

Dewatering through immersion in sugar/salt concentrated solutions

("Osmotic dehydration"). An interesting alternative for seafood
stabilisation! AL. Raoult-Wack, A Collignan 400

Osmotic dehydration of peas and its effect on drying! F. Kaymak,

T. Cakaloz 403

Studies on the sorption characteristics of bulk wheat and canol a/

W. Lang, S. Sokhansanj 406

Diffusivity of water and solubles during rehydration of osmo-convection

dried plant tissue/ A Lenart, B. Iwaniuk 409

Study of rehydration before eating on dehydrated vegetables/ M. Zhang,

C. Wang, X. Ma, C. Li 412

Effect of drying conditions on effective moisture diffusivity in "chufa"

tubers (Cyperus esculentus L.)/ A Altarriba, A Nunez-Lemos, D. Vidal 415

Simultaneous heat and moisture transfer with hygrostrain-stress

formation in food undergoing drying! K. Hayakawa, T. Tsukada,
T. Furuta, N. Sakai 418

Dynamic modelling of fruit dehydration! F. Giroux, S. Guilbert,

G. Trystram 421

Determination of structural parameters for the dried layer of coffee

solution undergoing sublimation dehydration! J. Ichiba, Y. Sagara,
Y. Kawagoe, Y. Seo 424

Quality aspects in drying of food granular materials/ K. Kramer,

I. Str<jlmmen, X. Song 427

Considerations on the diffusivities of moisture and aroma components/

W.J. Coumans, AAJ. Ketelaars, PJ.AM. Kerkhof 430

Flavour retention in different methods of spray drying! A Senoussi,

B. Bhandari,. E. Dumoulin, Z. Berk 433
xviii
Improvement of I-menthol retention by cyc10dextrin during drying!
T. Furuta, H. Yoshii, A. Yasunishi 436

Spray drying leaflash technique: An investigation on pulverization and

flow profiles/ L. Clement, E. Dumoulin, S. Lagarde, C. Bourlier 439

Research of far-infrared drying on white mushroom! M. Zhang, N. Xu 442

Convective drying of Lactobacillus plantarum/ LJ.M. Linders,

G. Meerdink, K. van't Riet 445

Convective drying of lactic acid starter cultures with high survival rates/
U. Kessler, W. Bauer 448

Drying of enzymes : Relationships between thermal stability of enzymes

and water concentration! S. Yamamoto, Y. Kasuga, Y. Sano 451

Reaction Kinetics In Food Processing

Prediction of kinetic parameters for food quality changes using

equivalent time at standard temperature/ H.-Y. Cho, Y.-R Pyun 453

Prediction of quality change during food storage/ D.R Heldman,

RR Chao 456

Correlations between physicochemical properties and kinetic parameters

of different cultivars of rice/ Y.-H. Chang 459

Development and validation of prediction models for texture change

under varying temperature conditions/ B.E. Verlinden,
J. De Baerdemaeker 462

The effect of cook-chill processing on the texture development of

different types of vegetables and fruits/ B.E. Verlinden,
x.K. Vandewalle, J. De Baerdemaeker 465

The effect of water activity and temperature on the retrogradation rate of

gelatinized wheat flour/ C. Rhee, S.W. Lee 468

Diffusion and reaction in wheat grains/ A.G.F. Stapley,

M.P. Hollewand, L.F. Gladden, PJ. Fryer 471

Studies on interaction between non enzymatic browning reaction and

lipid oxidation in food processing! M. Dalla Rosa, D. Barbanti,
S. Pizzirani, F. Bressa 474

Oxidation rate of polyunsaturated fatty acid and antioxidation effect of

gel layer/ K. Nakamura, K. Nagai, T. Inoue, M. Hakoda 477
 xix
Simulation of oxidation processes of liquid lipids/ S. Adachi, 1. Imagi,
T. Ishiguro, R. Matsuno 480

Isomerization reaction kinetics of catechins in packaged tea drinks

during processing! Y. Komatsu, S. Suematsu, Y. Hisanobu, H. Saigo,
R. Matsuda, K. Hara 483

Citrus fruits combination improving taste and flavor of marmalade/

T. Morishita, K. Nishio 486

Kinetics of aspartame degradation in liquid dairy beverages/ L.N. Bell,

D. Shoeman, M. Tsoubeli, T.P. Labuza 489

Esterification in micro aqueous organic phase/ M. Inoue, M. Imai,

M. Shimizu 492

Thermal inactivation of pectinesterase in papaya pulp (PH 3.8)/

P.R. Massaguer, M.A. Magalhaes, R.M. Tosello 495

Heat gelation and reaction kinetics of whey protein concentrates/

S.M. Taylor, L.F. Gladden, PJ. Fryer 498

Effect of microwave heating on non-enzymatic browning in banana

products/ M.P. Cano, V. Lizarraga 501

Effect of microwave heating rate on maltose production in sweetpotato/

T. Nakai, J. Sawai, A. Hashimoto, T. Honda, M. Shimizu 504

Studies on the microwave heating-rate equations of foods/ K. Kubota,

M. Kurokawa, H. Araki, T. Okazaki, L.T. Lu, Y. Hagura 507

Studies on the cooking-rate equations of foods/ K. Kubota,

Y. Masumoto, G. Zhang, Y. Hagura 510

Kinetic modelling of chicken muscle thermal conductivity during deep-

fat frying! M.O. Ngadi, L.R. Correia 513

Cold extrusion - Tribochemistry/ S.S. Wang, X. Zheng, C.T. Ho, D. Ou 516

Fermentation Processes

Chemical and physical changes of tofuyo (fermented tofu) during

fermentation! M. Yasuda, N. Kobamoto 519

Statistical characterization of tempeh starter from the aroma components

of soybean tempehi Suprinyanto, Y. Fujio, I. Hayakawa 522

Manufacturing of solid culture media in film bags for shiitake

cultivation! K. Ohsaki, S. Yamada, O. Fuse, T. Motegi, G. Kawai,
xx
Y. Fukushima 525

Sorghum brewing with facultative anaerobic Rhizopus in practical

operations/ H.-H. Wang, J.-e. Tsao 528

Production of enzymes for food processing by solid state fermentation!

M. Taniguchi, K. Ooshima, M. Fujii 531

Cell growth and a-amylase production characteristics of Bacillus subtilis

ATCC 601/ T.J.G. Salva, 1.0. Moraes 534

Fermentation processes for the production of natural colour compounds

for the food industry/ AM. Martin 537

Effect of castor oil hydrolysate on the production of y-decalactone and

cis-6-dodecen-4-olide by Sporobolomyces odorus/ S.-L. Lee,
C.-e. Chou 540

Modelling of transport phenomena during the cultivation of Bacillus

thuringiensis for the production of bioinsecticides/ A Perez-Galindo,
e.G. Velazque, H. Medrano-Roldan, e.M. Robles,
e. Rodriquez-Padilla, R. Tamez-Guerra, IL. Galan-Wong 543

Measurement of the main industrial fermentation parameters governing

the production of bioinsecticides by Bacillus thuringiensis var.
kumamotoensis/ H. Medrano-Roldan, CG. Velazquez, C.M. Robles,
A Perez-Galindo, C.P. Correa, e. Rodriguez-Padilla,
R. Tamez-Guerra, J.L. Galan-Wong 546

Use of sweet potato processing wastewater as substrate for fermentations/

C.AA Ribeiro, M.E. Cavenagui, AE.R. De Pontes, G.L. Barros,
AL. Pessoa, AC. Barana, F.R. Guidolin, V.L. Del Bianchi, 1.0. Moraes 549

Fermentative production of L-Iactate from xylose/ T. Ueda, K. Tanaka,

A Ishizaki 552

Optimization studies of a lactic acid fermented beverage process/

T. Moe, J. Adler-Nissen 555

General method for lactic acid batch fermentation processes monitoring!

E. Latrille, D. Picque, G. Corrieu 558

Anaerobic continuous ethanol fermentation using a computer coupled

medium feeding system which has DDC control pH of the culture broth!
S. Tripetchkul, M. Tonakawa, A Ishizaki, Z. Shi, K. Shimizu 561

Method for on-line prediction of the alcoholic fermentation rate in

wine-making! B. Perret, G. Corrieu 564

Measurement of cell density in the broth of aggregative organism

 xxi
by continuous-dilution-photometric-assay/ T. Yano, A. Masduki,
Y. Nishizawa, H. Ohtake 567

Dynamic modelling of a large scale air lift fermenter/ G. Trystram,

S. Pigache 570

Bioreactors Using Enzymes and Cells

Biocatalytic production of cellobiose containing oligosaccharide mixture/

M. Rossi, Y.-Y. Linko, P. Linko, T. Vaara, M. Turunen 573

Enzymatic production and membrane concentration of oligosaccharides

from milk whey/ M.H. LOpez Leiva, M. Guzman 576

Construction of integrated bioreactor system for biologically active

peptides from isolated soybean protein! K. Sonomoto, Y. Okamoto,
S. Kohno, K. Kano 579

Trypsin-catalyzed synthesis of oligopeptides containing hydrophilic and

essential amino acid, L-lysine/ Y. Kimura, M. Shima, T. Notsu,
S. Adachi, R. Matsuno 582

Integration of reaction and recovery by a continuous emulsion enzyme

reactor with in-line removal of the oil and water phase by membrane
separation! C.G.P.H. Schroen, A. Van der Padt, K. van't Riet 585

Interesterification of triglycerides in organic solvent using modified

lipase/ K. Mogi, S. Basheer, A. Yamaoka, F.B. Padly, K. Fujiwara,
M. Nakajima 588

The effect of lipids oxidation on the activity of interesterification of

triglyceride by immobilized lipase/ T. Nezu, S. Kobori, W. Matsumoto 591

Triacylglycerol synthesis from fatty acid and glycerol using immobilized

lipase/ Y. Kosugi, N. Tomizuka, N. Azuma 594

NAD(P)H regeneration in yeast cells using ethanol as energy source/

T. Kometani, H. Yoshii, E. Kitatsuji, R. Matsuno 597

Loose RO membrane reactor for L-amino acid production with

coenzyme recycling! T. Harada, O. Miyawaki 600

Studies on the biological reduction of the nitrate content of carrot juice

and of a model solution using Paracoccus denitrificans DSM 65/
S.H. Min, W.E.L. SpieS 603

Development of a bioreactor for semi-solid fermentation purposes:

bacterial insecticide fermentation! D.M.F Capalbo, 10. Moraes,
R.O. Moraes 606
xxii
Continuous ethanol production by Zymomonas mobilis in a bioreactor
with flame making ceramic carriers/ A. Mori, T. Aoki 609

Design and scale-up of bioreactors for microbial polysaccharide

fermentation! Y. Kawase, M. Tsujimura 612

Application of a bioreactor system to soy sauce production! T. Hamada,

Y. Fukushima, H. Motai 615

Substrate removal characteristics in an immobilised cell fluidized bed

reactor/ S.H.R Bhamidimarri 618

Separation and Purification Processes

Solvent extraction of flaxseed/ F. Shahidi, P.KJ.P.D. Wanasundara,

R Amarowicz 621

Large scale preparation of highly-purified egg yolk phosphatidylcholine

by HPLC/ H. Kobayashi, H. Narabe, M. Hasegawa, F. Harada,
K. Nomura, K. Kitagawa 624

Extraction and concentration of omega-3 fatty acids of seal blubber/

F. Shahidi, RA. Amarowicz, J. Synowiecki, M. Naczk 627

Separation of EPA and DHA from fish oil using supercritical extraction
with Ag complex pretreatment! K. Nagahama, T. Suzuki, S. Kikuchi,
Y. Tanaka, K. Nakano, H. Noritomi, S. Kato 630

Selective extraction of proteins from complex solutions by reverse

micelles/ K. Naoe, M. Imai, M. Shimizu 633

Fractionation of B-lactoglobulin in bovine whey by intermittent up

flow system with multi-stage ion exchange column! H. Ohtomo,
T. Kuwata, I. Kurihara, T. Kikuchi, S. Furusho 636

Recovery of lysozyme and avidin from egg white by ion-exchange

chromatography/ S. Yamamoto, T. Suehisa, Y. Sano 639

Development and application of cycIodextrin polymer/ H. Okemoto,

H. Hashimoto 641

Membrane Processes

Turbulent high-Schmidt number mass transfer in UF/ C. Rosen,

Ch. Tragardh, P. Dejmek 644

The pore size of ultrafiltration membranes - a novel approach!

K.M. Persson, G. Tragardh 647
 xxiii
Reverse osmosis system for highly concentrated juice/ H. Nabetani,
H. Igami, M. Nakajima, K. Hayakawa, Y. Yamada, Y. Ishiguro 650

Factors affecting the performance of crossflow filtration of yeast cell

suspension! T. Tanaka, K. Nakanishi 653

Cross-flow membrane filtration of soy sauce lees/ T. Furukawa,

K. Katou, K. Kumakura, K. Osaki 656

Applications of the ceramic membranes to soy sauce/ N. Kanekuni,

H. Nogaki, H. Tabata, A. Watanabe 659

Ceramic filters for food and beverage applications/ M. Fushijima 662

Recovery and functional properties of protein from the wastewater of

mungbean starch processing by ultrafiltration! W.C. Ko, W.J. Chen,
T.H. Lai 665

Improved extraction and ultrafiltration process for protein recovery from

soy flour/ J.L. Harris, S.K. Razavi, F. Sherkat 668

The production of protein isolates from Chinese rapeseed/ L.L. Diosady 671

Separation and concentration of polyunsaturated fatty acids by a

combined system of liquid-liquid extraction and membrane separation!
Y. Sahashi, H. Ishizuka, S. Koike, K. Suzuki 674

Refining vegetable oils by membrane technology/ M. Chery an,

L.P. Raman, N. Rajagopalan 677

Membrane separation of sunflower oil hydrolysates in organic solvents/

S. Koike, M. Yokoo, H. Nabetani, M. Nakajima 680

Catalyst removal from hydrogenated oils using membrane technology/

C. Vavra, S.S. Koseoglu 683

Continuous synthesis of aspartame precursor with membrane enzyme

reactor - Membrane extractor system/ Y. Isono, H. Nabetani,
M. Nakajima 686

Electro-ultrafiltration bioreactor for enzymatic reaction in reverse-

micelles/ M. Hakoda, Y. Ogawa, T. Akashi, K. Nakamura 689

Pasteurization and Sterilization Processes

Evaluation of the integrated-time-temperature effect of thermal

processes on foods: state of the art! M. Hendrickx, G. Maesmans,
S. De Cordt, P. Tobback 692
xxiv
Inactivation kinetics of bacterial spores in skim milk, concentrates, in
dependence on water activities and under sealings/ H.G. Kessler,
R. Behringer, J. Pfeifer 695

Determination of the thermal death kinetics of bacterial spores by

indirect heating methods/ J. Haas, R. Btiltermann, H. Schubert 698

A first order probabilistic perturbation analysis of the growth and

thermal inactivation of Lactobacillus cells during cold storage and
reheating of lasagna/ B.M. Nicolai, J.F. Van Impe, J. De Baerdemaeker 701

Variability of thermal process lethality/ K. Hayakawa, J. Wang,

P. de Massaguer 704

Microwave heating of liquid foods/ M. Kyereme, R.c. Anantheswaran 707

Selection of heating media for thermal processing of retort pouches/

P.R. Massaguer, C.F. Cardelli, H.G. Aguilera 710

Numerical analysis of heat transfer in the sterilizing process of canned

foods by vacuum pack! X.J. Huang, T. Hanzawa, N. Sakai 713

Heat transfer in liquid-filled containers during end-over-end rotation!

I.J. Britt, AT. Paulson, R. Stark, M.A Tung 716

Development of high speed batchwise sterilizer by steam fusion!

K. Aono 719

Computer aided design and optimization of sterilization of canned tunal

J.R. Banga, AA Alonso, J.M. Gallardo, R.1. Perez-Martin 721

Different strategies for controlling pressure during the cooling stage in

batch retorts/ AA Alonso, J.R. Banga, R.1. Perez-Martin 724

The influence of uncertainties in processing conditions on thermal

process calculations/ B.M. Nicolai, 1. De Baerdemaeker 727

ICRSIDS: a computer package for the optimization of batch processes

and its applications in food processing! J.R. Banga,
R.I. Perez-Martin, R.P. Singh 730

Disruption of microbial cells by flash discharge of high pressure gas/

K. Nakamura, A Enomoto, M. Hakoda, H. Fukushima, K. Nagai,
T. Mukae, Y. Masuda 733

Sterilization of beverages under normal temperature by a high-voltage,

pulsed discharge/ M. Sato, K. Kimura, K. Ikeda, T. Ogiyama, K. Hata 736

Comparative effects of gamma-rays and electron beams on film

dosimeters and bacterial spores/ T. Hayashi, H. Takizawa,
 xxv

S. Todoriki, T. Suzuki, M. Furuta, K Takama 739

Growth-inhibitory effect of ceramics powder slurry on bacterial

1. Sawai, H. Igarashi, A. Hashimoto, M. Shimizu 742

Aseptic Processes

Recent studies on aseptic processing of particulate foods/ N.G. Stoforos 745

Laminar tube flow of a Newtonian fluid containing large spheres -

Application to aseptic processing! 1. Fregert, C. Tragardh 748

Flow velocities of particles in holding tube/ H. Tozuka, T. Fujiwara,

M.Mori TIl
Flow of solid-liquid food mixtures/ S. Liu, 1.-P. Pain, P.J. Fryer 754

Modeling of time-variant heat transfer in two-phase system/ KH. Park,

RL. Merson 757

Heat generation and transfer in electric heating of a laminar flow of

food/ L. Zhang, P.J. Fryer 760

A new nondestructive system to evaluate quantitatively the efficiency of

food antiseptics/ K Takahashi 763

Development and application of aseptic new materials (ASEPLA)/

S. Kunisaki, K Noda, T. Saeki, T. Amachi 766

Continuous sterilization of particulate foods by ohmic heating: critical

process design considerations! S.K Sastry 769

Quality changes of aseptically packed apple juice during storage and the
prediction of its shelf-life/ S.-H. Yuo, S.~S. Lin, S.-Y. Chen,
C.-c. Chen, c.-C. Chen 772

Comparison of the immersion biotest and the spore-test method for

evaluating the integrity of seals of flexible retort packages/
G. Wirtanen, E. Hurme, R Ahvenainen, L. Axelson-Larsson,
T. Mattila-Sandholm 775

Packaging Science and Technology

Food packaging and shelf-life/ 1. Varsanyi 779

Modified atmosphere and modified humidity packaging to extend the

shelf life of fresh mushrooms (Agaricus bisporus)/ S. Roy,
RC. Anantheswaran, RB. Beelman 781
xxvi
Design of perforated polymeric packages for the modified atmosphere
storage of broccoli/ J.D. Mannapperuma, R.P. Singh 784

Protection against perishableness - New packaging materials and

principals of permeation measurement/ J. Hertlein, H. Weisser 787

Edible wheat gluten films: optimization of the main process variables

and improvement of water vapor barrier properties by combining gluten
proteins with lipids/ N. Gontard, S. Guilbert, S. Marchesseau, J.-L. Cuq 790

Selection of laminated film for a food packaging! M. Shimoda,

~~~ m
Continuous microwave drying of polyethylene terephthalate (PEn/
C.AR. Anjos, J.AF. Faria, A Marsaioli Jr. 796

Sterilization of packaging containers by gasified hydrogen peroxide and

the application to flexible aseptic packaging machine/ Y. Shibauchi,
T. Tanaka, K Hatanaka 799

Aseptic filling of beverages in glass bottles - New developed processes

~~~R~~~~ ~

Fouling and Cleaning

A kinetic model for fouling in milk processing! PJ.R. Schreier,

I. Toyoda, M.T. Belmar-Beiny, PJ. Fryer 805

Initial events in surface fouling! M.T. Belmar-Beiny, PJ.R. Schreier,

P.J. Fryer 808

Adsorption of protein onto stainless steel particle surface and its

desorption behavior/ H. Itoh, T. Nagai, T. Saeki, T. Sakiyama,
K Nakanishi 811

Processing under Unusual Conditions

Supercritical fluid extrusion - A new process/ S.S.H. Rizvi,

S.J. Mulvaney 814

Rate processes in supercritical fluid extraction! BJ. McCoy 817

Membrane separation of supercritical fluid mixture/ K Nakamura,

T. Hoshino, A Morita, M. Hattori, R. Okamoto 820

Adsorption isotherms for supercritical carbon dioxide on proteins

and polysaccharides/ K Nakamura, T. Hoshino, Y. Suzuki, N. Yosizawa 823
 xxvii
Application of supercritical CO2 for food processing! Z. Knez,
M. Skerget 826

Supercritical fluid extraction of aroma compounds from aromatic herbs

(Thymus zygis and Coriandrum sativum)! AL. Cardoso,
M. Moldiio-Martins, G. Bernardo-Gil, M.L. Beiriio da Costa 829

Separation of aroma components from soy sauce by continuous

supercritical CO2 extraction! Y. Kitakura, H. Imamura, S. Hayakawa,
M. Hamano, H. Hashimoto 832

Supercritical carbon dioxide extraction of carotenoids from carrots/

M. Goto, M. Sato, T. Hirose 835

The performance of preparative supercritical-fluid chromatography of

lipids and related materials! S. Yamamoto, Y. Sano 838

Continuous processing of milk fat with supercritical carbon dioxide/

S.S.H. Rizvi, AR. Bhaskar 840

Separation of antioxidative ferruginol from Japanese cedar bark by

supercritical carbon dioxide! H. Ohinata, N. Inoue, Y. Yonei 843

Moisture adsorption isotherms and reaction characteristics of immobilized

lipase in supercritical carbon dioxide! K. Nakamura, T. Sugiyama,
H. Sudo 846

Enzyme reaction in supercritical fluid/ Y. Endo, K. Fujimoto, K. Arai 849

Supercritical carbon dioxide as processing medium for enzymatic

interesterification! B. Moshammer, R. Marr, A Biladt, F. Froschl,
W. Preitschopf 852

Enzyme inactivation by pressurized carbon dioxide! M.O. Balaban,

S. Pekyardimci, C.S. Chen, A. Arreola, M.R. Marshall 855

Purification of organic acids by gas anti-solvent crystallization!

A. Shishikura 858

Effect of high pressure on activity of some oxidizing enzymes!

Y. Aoyama, M. Asaka, R. Nakanishi 861

Application of sterilization technique by hydrostatic high pressure for

green tea drink! T. Takeo, H. Kinugasa, M. Ishihara, K. Fukumoto,
T. Shinno 864

Study of high pressure effect on inactivation of Bacillus stearothermophilus

spores/ I. Hayakawa, T. Kanno, Y. Fujio 867

Development of fiber-containing non-expanded products by extrusion

xxviii
cooking process/ W. Chiang, M.-J. Shih, A-R Yiao, J.-H. Weng 870
The heat denaturation process of soybean protein by twin screw
extruder/ Y. Akiyama 873
The development and control of colour in extrusion cooked foods
simulated using a model reaction celV L. Bates, J.M. Ames,
D.B. Macdougall 876

Application of extrusion-cooking for feed premixes stabilization!

L. Moscicki, S. Matyka 879

Extrusion of Castanea sativa/ F. Silva, A Choupina, 1M.N. Sousa,

M.L. Beirao da Costa 882

Transportation and Preservation of Food

Effects of preparation procedures and packaging materials on quality
retention of Cut Chinese cabbage/ E. Hurme, R Ahvenainen, E. Skytta,
M. Hagg, M. Mattila, R-L. Heinio, A-M. Sjoberg 885

Effects of preparation procedures and packaging materials on quality

retention of grated carrots/ E. Hurme, R Ahvenainen, E. Skytta,
M. Mattila, M. Hagg, R.-L. Heinio, A-M. Sjoberg 888

Storage technology on dehydrated sword bean! M. Zhang, B. Kong,

C. Wang, Y. Yang 891

Effects of several treatments on the quality of cold stored "clemantine

mandarins"/ F. Pazir, M. Azak 894

Postharvest treatments on quality of mango fruits (Mangifera indica (L)

var. Zill) stored at low temperature/ M.AB. Kanesiro, J.F. Durigan,
RRS. Faleiros, D.M. Andriolli 897

Flavor change of grape juice during processing! H. Ohta, Y. Nogata,

K. Yoza 900

Production of powdery products out of concentrated fruit-juices and

other food and bio-products/ T. Gamse, R Marr, F. Froschl,
E. Moschitz 903

Physiological roles of membrane alteration in gamma irradiated potato

tuber/ S. Todoriki, T. Hayashi 906

Development of a time-temperature indicating device using

phospholipase/ S.H. Yoon, C.H. Lee, D.Y. Kim, lW. Kim, K.H. Park 909

Heat treatment experiments on hay to eliminate possible contamination

 xxix

with Hessian fly/ S. Sokhansanj, H.C. Wood 912

Bacteriocin producing lactic acid bacteria used for biopreservation of

food/ B. lelle, N. Peitersen 915

Studies on the control of the growth of Saccharomyces cerevisiae by

using response surface methodology to achieve effective preservation at
high water activities/ A.-E. King 918

Sensors, Process Control, and Factory Automation

Theory and applications of a new viscometer based on annulus liquid
flow/ K. Suzuki 921

An extended hot-wire method for monitoring fluid viscosity and the

onset of gel formation! T. Hori, Y. Shiinoki, K. Itoh 924

Use of an in-line viscometer in the manufacture of skim milk powder/

C. O'Donnell, N. Herlihy, B. McKenna 927

Development of shear stress based sensor to measure drying rate and its
application to snack drying automation! S.C. Shin, 1.1. Yoo, 1.K. Chun 930

Measurement of fluid thermal conductivity with a steady state hot wire

method/ Y. Shiinoki, T. Hori, K. Itoh 933

Optical sensors (UV, VIS & NIR) for the determination of connective
tissue, lipid and protein functionality in meat! H.l. Swatland 936

Quick determination of fat content in beef longissimus by near-infrared

spectroscopy with a fiber optic probe/ M. Mitsumoto, S. Ozawa,
T. Mitsuhashi, K. Shinohara, K. Tatsubayashi 939

Individual sugar content control by the use of F.T.-I.R. spectroscory

coupled with an A.T.R. accessory/ V. Bellon, C. Vallat, O. Pauwels 942

Electrical conductivity in avocado as maturity index/ M. Montoya,

V. LOpez-Rodriguez, J.L. De La Plaza 945

New electrical method for density sorting of spherical fruits/ K. Kato 948

Highly sensitive and rapid determination of diacetyl in liquid foods by

electrochemical method/ N. Horikawa, K. Hayakawa, Y. Yamada,
O. Miyawaki 951

Nondestructive quality evaluation of melons by acoustic transmission

characteristics/ S. Hayashi, 1. Sugiyama, K. Otobe 954

Nondestructive internal defect detection of agricultural products using

xxx
secondary ultrasonic wave/ S. Tanaka, K Morita, S. Taharazako 957

Wireless image transmit unit and high speed inspection system for inner
side of container/ M. Hoshino 960

Application of a biogas bubble counter to fermentation processes/

Y.J. Lee, N.E. Choi, D.H. Woo, KM. Kim, lK Chun 963

Computer aids in flexible food manufacturing! C. Skjoldebrand 966

Automation of cheese and yogurt manufacturing processes/ G.S. Mittal 969

Automation of shrimp quality evaluation! M. Balaban, S. Yeralan,

Y. Bergmann, W. Steven Otwell 972

Automatic control of the biscuit baking oven process/ G. Trystram,

M. Allache, F. Courtois 975

Study of handling techniques for the soft and plastic substance such as
food/ S. Mammoto 978

Dynamic modelling and simulation of food processes/ J.J. Bimbenet,

G. Trystram, A. Duquenoy, F. Courtois, A. Lebert, M.L. Lameloise,
F. Giroux, M. Dec;loux 981

Neural network control for extrusion cooking! K Uemura, S. lsobe,

A. Noguchi 984

Fuzzy techniques for process state estimation! V.J. Davidson,

R.B. Brown, G.L. Hayward 987

JIT and CIM concepts, applied to a large scale catering production

plant! T. Martens 990

Innovation In Equipment Design and Plant Operation

Development of novel agitating device for bioreactors/ K Maruyama,

M. Ohmi, M. lmai, S. Urushiyama 993

Total energy management system (fEMS) at foodstaff factory moderated

environmental impact production system/ S. Sakashita 996

A blackboard and object oriented fermentation plant model! V. Niviere,

P. Grenier, J.-M. Roger, F. Sevila, M. Oussalah 999

Environmental Problems in Food Industry

Treatment of sweet-potato processing wastewater using UASB reactors.

 xxxi
I - Start-up of the process/ M.D. De Maria, V.L. Del Bianchi,
1.0. Moraes 1002

Two stage controlled anaerobic bio-reactor for digesting mixed dairy

waste/ c.L. Hansen, S.H. Hwang 1005

Activated sludge treatment of meat processing wastewater/

AP. Annachhatre, S.M.R Bhamidimarri 1008

Anaerobic treatment of high strength dairy processing waste in an

upflow anaerobic sludge blanket reactor/ B. Grauer,
S.M.R Bhamidimarri, RL. Earle 1011

Anaerobic treatment of distilled waste water of wheat 'shochu'

using two-phase fermenter/ K. Monta, S. Taharazako, Y. Nishi 1014

Pleurotus production in coffee bean wash water/ J .E. Sanchez

V, A.M. Martin 1017

Byproducts from food industries: utilization for bioinsecticide

production! 1.0. Moraes, D.M.F. Capalbo, RO. Moraes 1020

NOx removal in water by biochemical reaction! T. Oku, M. Kondo,

H. Sato, T. Nishio, T. Ito, M. Hoshino, H. Seki 1023

Innovation in Traditional Food Processing

RIPFADI: !bero American network on physical properties of foods/

J.M. Aguilera 1026

Relationships between physical properties and chemical components of

Okinawan brown sugar/ T. Akinaga, Y. Kohda 1029

The effect of processing variables on the product quality of soya

plantain baby food/ P.O. Ogazi, H.O. Ogundipe, F.A Oyewusi,
S.AO. Adeyemi 1032

Predicting tofu productivities of soybean varieties by tofu gel

centrifugation! S.-J. Tsai, T.L. Hong, S.C.S. Tsou 1035

Use of the modem technology in improvement of Chinese tranditional

food Tou-nao processing! G.-Q. Shen, C.-J. Cheng, S.-J. Li 1038

Frozen stabilized mince as a source of pacific whiting surimil

R Simpson, E. Kolbe, G.A MacDonald, T.G. Lanier, M.T. Morrissey 1041

Production of Brasilian imitation cheese with a partially hydrogenated

vegetable fat/ M.L. Gigante, S.M. Roig 1044
xxxii
Design of Physiological Functions of Foods from Engineering Viewpoints

Improvement of bread flavor by a dairy substrate treated with enzymes

and lactic acid bacterial M. Yamamoto, R. Watanabe, T. Kaneko,
H. Suzuki 1047
Interaction between inhibitors of calcium phosphate formation and
calcium ion! K. Yamamoto, H. Kumagai, T. Sakiyama , H. Ogawa,
T. Yano 1050
Studies on the optimum conditions to utilize biologically active peptides
derived from food proteins/ M. Yoshikawa, H. Fujita 1053
Anti-platelet principle found in the essential oil of garlic (Allium
sativum L.) and its inhibition mechanism/ T. Ariga, T. Seki, K. Ando,
S. Teramoto, H. Nishimura 1056
Lipid peroxide decreasing activity of microbial cells/ M. Ito, K. Ishihara 1059
Effect of water and alcohol on the formation of inclusion complex
between cyclodextrins and d-limonene by twin screw kneader/
H. Yoshii, T. Furuta, T. Kobayashi, T. Nishitarumi, H. Hirano,
A. Yasunishi 1062
Entrapment of liquid lipids into powdery matrixes of saccharides and
proteins! R. Matsuno, J. Imagi, S. Adachi 1065

AUTHOR INDEX 1068

SUBJECT INDEX 1075
 FOOD ENGINEERING ---- FROM ARTS TO SCIENCE

TOSHIMASA YANO
Department of Bioengineering, Faculty of Engineering
Yokohama National University
Tokiwadai, Hodogaya-ku, Yokohama-shi, Kanagawa 240, Japan

When persons having a background in engineering start work in the food industry, they
encounter various interesting practices which are based more on elaborate experience
accumulated over a long period of time than on recent engineering principles. I like to call such
routines a kind of practice art. There are many arts in the food industry. In fact, any food
process is an ensemble of arts, and often they are quite pleasing to observe.
But what happens if any of the processes must be revised for some reason -- say -- to
change equipment, or operational methods, or production scale, or when introducing new
products or labor-saving techniques? In such cases, the personnel involved must acquire new
experiences. To acquire new experiences quickly, as much as possible avoiding unnecessary
trials, they have to make logical predictions before they begin. And this amounts to the
application of engineering and scientific principles.
However, in food processing, doing this is not so simple. For instance, regardless of
food processing method, the finished product must be palatable to its consumers. The
processing method can be changed to one thought perfectly logical, but if the consumer does
not like the product, the food industry cannot accept the logic. Is there any known engineering
or scientific principle that can make food more delicious? I wish there were. Engineers, as
you can see, face difficulties in the food industry never before experienced.
Another problem engineers face in food processing is one of biological materials. In
any type of food processing, raw materials must be of biological origin. Why are biological
materials a problem? Frankly, most engineers and scientists failed to recognize the problem
correctly. Instead they believed that theories successfully applied to non-biological materials
should also apply to biological materials. But in many cases it just doesn't work.
Engineers are often surprised when they try to use proven engineering theories in food
processing, knowing that the physical properties of materials with which they are familiar
are not clear. Didn't anybody analyze the physical property yet? Of course! A great many
measurements were taken, but no one could tell quantitatively what the property of a given
food was! Things like this happen because the chemical composition and the dispersed state of
components can vary according to material. With food, its physical properties even change
during processing. Take, for example, dough used to bake bread or cakes. Its water content
can vary during processing; and, more importantly, air or carbon dioxide contained in the
dough changes its volume fraction and dispersion state even under constant temperature.
In addition to uncertainty among physical properties, it is not unusual that known
theories fail to predict physical food behavior. This is not because known basic principles do
not apply to biological materials; rather, certain assumptions from which known theories derive
are not valid when it comes to biological materials. Among these assumptions we find
2
homogeneity, constant parameters, simplified initial and boundary conditions, linearity, and
things like that.
In spite of the difficulty, engineers were forced to act, and they did, little by little, step
by step. Following their efforts over the past two decades, which obviously supported and
stimulated developments within the food industry, a branch of chemical engineering destined to
be called Food Engineering appeared. Food Engineering is now stimulating food processes to
take off the ensemble of arts, to involve solid engineering methods, and to land the engineering
methods in science. This is what is meant in the title of my talk here today.
Let's look at some of the developments we are concerned with, those I've selected
mainly from among the activities of our Japanese food engineers. I apologize for any bias in
my selection, and for including some of topics presented at ICEF4 in Edmonton and at ICEF5
in KOln. In this review, I'd like to separate the developments into two categories: Food
Engineering in Practice, and Food Engineering in Theory.

FOOD ENGINEERING IN PRACTICE

1. Before I get into its developments, let me show you what the food section of a Tokyo
department store looks like. Here you see the final phase of food processing. I took this
picture nearly six years ago, but things are pretty much the same now as they were then. The
main feature is that a large variety of foods are offered in relatively small quantities each. This
is done to accommodate our contemporary consumer market, which consists largely of smaller
families, single and married persons who live apart from their families, and senior citizens
whose children have left home. Eating a variety of food is not only enjoyable, but it's also
good for the health. The Japanese sign reads "Individuality in Variety".
How did Japan's food industry cope with the trend toward providing a greater variety
of food in smaller amounts? For one thing, the concept of production had to be changed, at
least in part, from large scale manufacture of a few items to small scale output of many. How
was this accomplished? A clever man suggested a model which was later applied.
2. Here you see a man working in a Sushi bar near The University of Tokyo --- some
might even call him an artist at his trade. I took this picture too six years ago, but he's still as
active as he was then. I didn't tell him that he had the honor to be chosen as a model for our
food industry. In this model there are three important points: First, the sushi artist does a
variety of unit operations by himself. Second, he makes his products only after receiving an
order. And, third, he never stocks the finished products.
As many of you may have noticed, the sushi artist's modus operandi is the same as a
housewife's routine in her kitchen.
Taking a tip from the sushi artist and from their own wives, Japanese food engineers
started converting specialists in one operation to persons capable of doing many things. But. ..
to err is human, to forgive divine. And not to forgive is consumerism. Chances for human
error vastly increase when the specialist is converted into a jack of all trades. The answer lay
in the introduction of micro-computers, which vastly reduced human errors. Thanks to the
model provided by a man making sushi and by our own better half. Garnishing it with a bit of
science topped with a micro-computer, we succeeded in meeting the new market need, while
our inventory of finished products was greatly reduced in the bargain. We owe much to the
efforts of our food engineers.
3. The demand for a variety of foods stimulated commercialization of new products.
Today, even in the development stage of new products, experiments and tests are becoming
more and more scientific.
This slide was kindly provided by the Kirin Beer Company, a firm that not long ago
developed a light beer. At first they succeeded in making a distinction between three kinds of
beer --- Japanese, German, and American --- by applying the so-called multivariate analysis.
The two orthogonal axes show two combined contributions of chemical components peculiar
to the respective tastes of beer. Next they set up a design target, the character of which
differed from those of the three types of beer in question, but lay between Japanese lager and
 3
American Light. Lastly, they developed a process called dual fermentation to arrive at the
targeted beer, but only after they met with failure. The lesson to be learned here is that the
targeted beer had been arbitrarily designed in terms of chemical components, prior to
development of the new fermentation process.
4. While on the subject of brewing, you might like to know that the aging of Japanese
Sake can be controlled during its production stage, taking into account the chemical character
of the finished product and the expected temperature change during its distribution. This
control became possible based on extensive work by Dr. Iwano. After finding a marker
substance for aging, he studied the reaction kinetics of its formation and its subsequent change,
and included the effects of the product's chemical character on the reaction rate constants,
using multivariate analysis (this type of data is especially important for food engineers). He
then studied the heat transfer characteristics of pasteurizer used, made many computations to
solve differential equations, and finally he proposed a way to predict and control the aging of
Sake. I had the honor of being one of the judges for his doctorate thesis. It was about twenty
years ago.
5. Another member of Japan's traditional food industry is also changing: Tofu, or
soybean curd, is one of our nation's most popular foods. But it was always considered
unsuitable for mass production, one of the reasons being that it spoiled easily. Today,
however, it can be kept without spoiling for several days under refrigeration, thanks to recent
advancements in pasteurization techniques. And this has made possible large scale production.
An extreme case of application of the advanced sterilization technique is the appearance
of Long Life Tofu. Long Life Tofu will stay fresh for a whole year without refrigeration, or
even longer. The principle in making it is the same as for Long Life Milk: high temperature
short-time sterilization (HTST sterilization). But how can HTST sterilization be applied to a
semi-solid substance? Simply put, it cannot. The process is applicable only to liquids. So, to
sterilize tofu while it's still in liquid state, engineers changed its coagulation agent from
traditional Ca salts to glucono-o-Iactone, which slowly coagulates soy protein after the
sterilized soy milk is aseptically packaged. An important problem to be solved was how to
prevent the browning reaction that inevitably occurs during HTST sterilization. The answer
lay in extracting the carbohydrates before the beans were milled. The texture change resulting
from carbohydrate extraction was adjusted by an appropriate combination of heating
temperature and time. In this manner, Long Life Tofu was born. But, ironically, its sale is
prohibited in Japan for political reasons, so that thousands of small, cottage industry type Tofu
producers may survive. Long Life Tofu is sold only out of Japan.
6. Japan's cheese industry is changing, too. Dr. Hori, of Snow Brand Milk Products Co.,
invented a simple yet highly useful sensor to monitor cheese coagulation. Watching the
coagulation of cheese to pick the optimum moment for cutting curd so as to remove whey has
long been the task of experienced cheese-makers. Dr. Hori's invention automates this exacting
process. The principal component of his sensor is a thin wire in which electrical current flows.
The heat in the wire transfers to the outer layer of defatted milk where coagulation is in
progress. The rate of heat transfer depends on the degree of natural convection taking place
near the surface of the sensor. When coagulation begins, the natural convection is suppressed,
raising the temperature of the wire. Thus, by measuring the temperature increase of the wire,
the coagulation of milk can be monitored. At present, the sensor is used for in-line processing,
and already is enjoying wider application in viscosity-change monitoring (refer to his paper in
this Proceedings). With Dr. Hori's thesis too, I had the honor of being one of its judges.
7. Another recent topic is the application of high static pressure up to 600 MPa (about
6000 atm). The combination of high pressure and mild heating was applied to produce a new
type of fresh jam.
S. Genetic engineering and protein engineering are expected to improve food materials. In
recent years, certain individuals have become highly sensitive as regards substances common
in daily life. Rice is one example, pollen is another. The upper picture of this slide shows the
allergic reaction of a fourteen year old girl after eating ordinary rice. The crucial amino acid
sequence forming the allergen of rice was identified by a food scientist who happened to be a
classmate of mine. Later, Professor Arai of The University of Tokyo used his enzyme to
hydrolyze the crucial amino acid sequence. After deleting the allergen from protein, he applied
4
the perboiled rice technique used in Asian countries to reconstitute the hypoa11ergenic rice
grains to almost the same appearance and texture of the normal rice. As the lower picture
shows, the patient can now enjoy Dr. Arai's rice without suffering from an allergic reaction.
Details of this will be given by Dr. Arai himself (refer to his paper in this Proceedings).
9. Before ending my review of Food Engineering in Practice, I'd like to introduce a
wonderful machine invented by a confectioner about thirty years ago. This machine
completely encloses one material inside another, like a four-dimensional operation. Its
production capacity is one per second. The machine was invented to make traditional Japanese
sweets. But later, its function was combined with flattening and folding, as such it found new
application in making delicate pies. You'll meet the inventor this evening at the Welcome
Reception of the Japan Food Machinery Manufacturers' Association, since he's now the
president of the Association. You'll also see an advanced model of this machine at the
Exhibition slated to run concurrently with our Congress starting tomorrow.

FOOD ENGINEERING IN THEORY

Let's now tum to Food Engineering in Theory. To move from an ensemble of arts into the
realm of science, Food Engineering must have a solid theoretical foundation. The past two
decades have seen a lot of developments in Food Engineering theory. Once again, with your
permission, I'll draw my data mainly from my own research at The University of Tokyo over a
period of twenty-one years unti11992. I might add that what I'm about to say represents only
a tiny facet of the vast range of activities observed among Japan's many universities.
Until 1971 I was a microbiologist, at which time I was invited to take charge of the
newly founded Food Engineering laboratory. As I looked here and there among the ensemble
of arts in food processing back then, I foresaw my new career in Food Engineering as riddled
with difficult problems peculiar to food processing which could not be resolved by simple
application of known theories. And I made it my initial task to find out why. It wasn't easy,
since practically no one knew the problems in my mind.
After asking questions and getting answers that added up to nothing, I had the good
fortune of meeting a wonderful chemical engineer retained by Ajinomoto Co., who solved my
question. Mr. Sagara was his name. The three important points he gave me were: (1)
physical properties are uncertain, (2) any unit operation must be with multi-purpose, and (3)
no engineering trial is successful until the finished product wins approval via the human
sensory test. This is what I mentioned at the start of my lecture. With Mr. Sagara's answer in
mind, I started my Food Engineering. After the darkest five years, I found a gleam of hope in
the effective thermal conductivity of food materials.
10. Uncertainty in effective thermal conductivity stems from versatility among food
materials with respect to chemical composition and physical structure. This being the case, I
found that the best way to predict effective thermal conductivity is to find a mathematical
formula that describes effective thermal conductivity as a function of the chemical composition
and the physical structure of material. In the mathematical formula, chemical composition
should lie in the volume fractions of each component together with their respective intrinsic
thermal conductivity, and physical structure should occupy the functional form.
In my research, however, two unknowns awaited me. One was that the intrinsic
thermal conductivities of proteins and carbohydrates were not known. This is because the
proteins and carbohydrates become powdery when purified, and the intrinsic thermal
conductivity of powdery material cannot be directly measured. The other unknown was the
functional form applicable to food materials of highly irregular structure, although many
functional forms had been derived theoretically from models of regular structures. To make a
long story short, we succeeded in clearifying the two unknowns. The intrinsic thermal
conductivities of typical proteins and carbohydrates were roughly estimated, and four
functional forms of theoretical bases were found to be applied to the foods of irregular
structures (see Reference (1) for the four functional forms and the intrinsic thermal
conductivity of wet soy protein).
 5
How to predict effective thermal conductivity of foods is summarized in my book (2).
Several years ago, a professor emeritus of Kyoto University, who was concerned with
a medical matter, asked me about the effective thermal conductivity of human skin. I didn't
know if anybody had ever measured it, but I answered him based on my experience,
considering the average composition of human skin in terms of water, protein and fat contents,
plus the extent of variation in fat content.
11. Meanwhile, a wonderful student came to me, one who was interested in rheology. I
knew from literatures that rheological behaviors of many foods deviated from the predictions
of known theories, and I surmised that one of the causes behind the discrepancy was fme air
bubbles impregnated in food materials. The known theories assume a constant volume of
material under a given stress, but the volume of bubble-impregnated material may significantly
change, even if the matrix material does not change in volume. We had no theory applicable to
the elastic behavior of volume variable material. So I asked the student to research the elastic
behavior of air-impregnated food gels. Starting from nothing, he first detoured a bit, probably
because he couldn't find any relevant literature, but he finally returned to the problem. He
thoroughly studied the phase of rheology relevant to the problem, introduced the newest Finite
Element Method of computation, deleted the crucial bugs from the computer program, and
discovered a way to predict the three-dimensional elastic behavior of volume variable material,
within the range of linear deformation (3).
To briefly summarize the student's work, the three-dimensional elastic behavior is
represented by four parameters: Young's modulus, shear modulus, Poisson's ratio, and bulk
modulus. Among the four parameters, two equations hold theoretically. For a volume
invariable matrix material, the bulk modulus is infinity, Poisson's ratio is 0.5, and one of the
two equations becomes uncertain, leaving only one available equation. Thus, measuring either
Young's modulus or the shear modulus gives us all the information necessary. With volume
variable material, however, we must measure two parameters at the same time, because only
two equations hold. For air-impregnated gel, a total of nine parameters are involved: four each
for the matrix and whole body, plus the volume fraction of fine air bubbles. Since we have
seven kinds of information after measuring two parameters, we need two more equations to
predict the whole. But we didn't have the two necessary equations. Thus the elastic behavior
of many types of food remained a mystery.
Among the four parameters of the whole body, only Poisson's ratio is a function of the
volume fraction of air. Shortly before we began our work, Kyoto University scholars showed
that Poisson's ratio of a wooden plate with a hole in it can be calculated by applying the Finite
Element Method of computation. My student confirmed that the result approximated the
measured Poisson's ratio of our air-impregnated gels. However, taking into account that the
computed result was for a plate with a hole, we proposed an empirical equation, one which
might serve until rigorous three-dimensional computation is done. Although we proposed this
empirical equation, we expect that the equation will be developed from purely theoretical
principles in the future.
Now, Mr. Shiinoki, this was the name of my brilliant student, did his best to discover
the final equation that we needed, and found the most suitable one. The prototype for the
equation, which is quite complicated but most generally applicable, was theoretically worked
out by Professor Okano of The University of Tokyo during his youth (4). In this manner, we
were able to propose a comprehensive method for predicting the three-dimensional elastic
behavior of air-impregnated material. In other words, by measuring two elastic parameters
simultaneously within the range of linear deformation, we can calculate the five other unknown
parameters (5,6).
12. Young ladies came to my laboratory, too, ones who had no knowledge of engineering.
They wanted to work in the field of Home Economics after finishing their doctorate theses.
Each of them did a very good job. One Korean lady worked on whipping.
Before the lady even began to experiment, I foresaw the kind of results she would get.
If a similarity exists among the whippings of various whippers, and the formed bubble size
depended mainly on three kinds of force, such as inertial, viscous, and surface tension, then
the normalized mean bubble diameter should be a function of Reynolds Number and Weber
Number. I also added the effect of gravity as Froude Number. The lady in question obtained
6
her experimental data mainly using a single rod whipper, and she compared the data with the
results obtained from four industrial whippers and two home size whippers.
Here you see the results obtained with the single rod whipper. The empirical equation
obtained was generally applicable to the variety of materials shown, under diverse operating
conditions. The industrial size whipper produced smaller bubbles, down to half the diameter
predicted by the equation. However, I still suggest the use of this equation, keeping in mind
that smaller bubbles will result from use of industrial whippers. How to calculate the
whipping parameters is summarized in a book (7).
13. Other ladies studied the growth and shrinkage of bubbles in flour dough, under
temperature increase as well as under constant temperature. In both cases, the rate of growth
or shrinkage was controlled by diffusion of dissolved gas into, or out of, bubbles (8). Even
though the change of the distribution in bubble size has yet to be quantitatively predicted, the
diffusion-controlled growth and shrinkage of bubbles in dough will give those in the food
industry an idea as to how they can control food texture. The change of distribution in bubble
size is often overlooked in food processing and in the transportation of so called "half-cooked"
foods, but it plays an important role in controlling food texture. The ladies are now bringing
new flavor into Home Economics and Food Science.
14. From the very outset of my Food Engineering career, one of the major topics of
research was a systematic analysis of traditional Kori-tofu processing. In this research, I was
more interested in the linkage of unit operations than the analyses of each unit operation. The
Kori-tofu industry is small as regards production scale, but to me it was an interesting example
of food processing. Following the ftrst energy crisis, back in 1973, my research took on the
added aspect of energy saving. Fortunately, my colleagues and I had a chance to conduct an
energy analysis of the second-largest Kori-tofu plant. Our analysis revealed an unexpected
characteristic of the process (9).
Prior to our analysis, we expected the tofu drying operation to consume the greatest
amount of energy in view of the heat required. But extraction and denaturation of soy protein
used twice the thermal energy than the drying process did. We also expected the freezing and
aging process to consume the greatest amount of electrical energy. But again we were wrong
in that the waste treatment aspect consumed more electrical energy than the entire production
line. The reason why the extraction and denaturation of soy protein consumed so much
thermal energy lay in the large amount of water used in the process, plus the fact that relatively
dilute soy milk had to be heated to approximately the boiling point in order to denaturate
protein. So, if the amount of extraction water used were cut to one-half, for example, thennal
energy consumption should also be reduced to one-half. By observing that recipe, oil
consumption in the entire production line will be lowered to two-thirds.
The simple reduction of extraction water lowered the yield of protein, and that was
unacceptable to the producer. Overcoming this was hardly a problem, since by introducing
two-stage extraction, we easily cut the extraction water to one-half and got an even better
protein yield. The real question was how to change the operational conditions of subsequent
processes. With soy milk now having double its former concentration, coagulation and
consolidation conditions had to be revised, in order to maintain the quality of the consolidated
fresh tofu.
Our analysis focused on a subsystem comprising four operations: extraction of soy
protein, denaturation of the protein, its coagulation with Ca, and consolidation. The
rheological property was chosen as representative of the quality of the middle product,
consolidated fresh tofu. The fresh tofu was to be frozen for several weeks so as to make its
protein insoluble. The linkage between extraction-denaturation and coagulation became clear
through our analysis of the binding equilibrium of Ca and soy protein. In the coagulation of
our thick soy milk, the amount of Ca bound to soy protein should be controlled to a certain
ratio. Insufftcient or excessive binding of Ca to protein not only lowered the yield of
coagulated protein, it also had an adverse effect on the consolidation process. Control of
particle size was also recommended to assure proper strength in the tofu against breakage.
A theory advanced by Professor Shirato of Nagoya University found application in the
consolidation of coagulated soy protein. The relationship between coagulation and
consolidation was studied as the effects of bound Ca and coagulation temperature on the
 7
coefficient of consolidation (10). The rheological property of the consolidated fresh tofu was
analyzed using a six-element rheological model, each of its parameter being a function of water
content alone. However, the break strength of fresh tofu was linked to the particle size
distribution of coagulated soy protein.
Since all the links between and among the various unit operations in the overall process
became clear, thermal energy consumption in the Kori-tofu plant could be reduced to two-
thirds of its former rate, without changing the rheological property of its fresh tofu output.
Moreover, if anyone wants to make a small change in the rheological property of fresh tofu, or
in its consolidation or coagulation, with constraint given, we can predict how to make it.
Although this type of research may not be attractive to many of us, I firmly believe that such
studies are necessary if an overall solid foundation in Food Engineering is to be laid
15. Last on my agenda of Food Engineering topics is fractal.
Here you see a scanning electron micrograph of dried alginate gel. Within this irregular
porous structure, lies a strange world of non-integer dimension. It's an infinitesimal world,
having 2.7 dimension. In it, the surface area of material is not proportional to the square of
particle diameter, rather it is proportional to the diameter raised to the power of non-integer.
The concept of surface area is not defmite, because the surface area fluctuates depending on the
scale of measurement. Moreover, the differential equations from which many theories derive
don't apply to this world, since its very differentiation is undefinable. A structure having non-
integer dimension is called fractal. Two young colleagues of mine studied the curious fractal
character of certain food materials, and they learned that some of the materials they chose
indeed have a fractal nature. The surface area could be proportional to the particle diameter
raised to the power of any non-integer real number in the range of between 2.0 and 3.0 (11).
The surface area of food material was recognized differently depending on kind of molecule, as
the fractal nature suggested. One of my two colleagues found that the three empirical equations
proposed for a relation between the energy requirement for milling and the reduced mean
diameter of solid particles do not conflict with one another, provided the surface structure of
the solid particles is fractal (12).
16. Such fractal behavior is understandable if the self-similarity exists in the irregular
porous structure. Parallel behavior is seen in the so called critical phenomena that occur near
the secondary phase transition. Secondary phase transition is known as that between normal
and strong magnetism, and between normal and super-conductivity. However, De Gennes, a
recent Nobel Prize awardee, suggested a similarity between sol-gel transition and secondary
phase transition, based on the concept of percolation transition.
This gives you an idea of percolation transition on a square lattice. The group of small
circles connected by a solid line is called a cluster. The cluster of solid circles, which percolate
the lattice from its uppermost end through its lowermost, is called an infinite cluster.
Secondary phase transition occurs at the moment the infinite cluster appears, or disappears.
The upper left figure corresponds to states of normal magnetism, normal conductivity, and
probably sol. Conversely, the upper right figure corresponds to states of strong magnetism,
super-conductivity, and probably gel. Young colleagues of mine at The University of Tokyo
are now working on sol-gel transition from this viewpoint, searching for a similarity among
the three phenomena (13).
This, then, has been a review of developments in Food Engineering here in Japan.
Before ending my talk, let's move out a bit into the world, and try to see what we can look
forward to in Food Engineering developments. Among very recent topics, which I did not
include, are NMR application and glass transition, both of which are attracting a great deal of
attention (refer to this Proceedings). However, even in seemingly well-established segments
of Food Engineering, there are new ways of approach. In this context, I was deeply
impressed when attending an international symposium, one whose name was abbreviated to
ISOPOW-V (8-14 November 1992, Pe~iscola, Spain). At the Symposium, I learned that Latin
Americans had been conducting a big Thero-American research program, its purpose being to
find optimal combinations of well known methods for preserving food, without using
refrigeration or other forms of high-level technology. Dr. Aguilera of Chile may comment on
this research program in his oral presentation (refer to his paper in this Proceedings). At all
8
events, whatever infonnation you need to make forecasts about future developments most
certainly will be aired during this Sixth International Congress on Engineering and Food.
This is the herald of our congress. Its designer is a sister of my young colleague. It's
rather abstract, so you're free to imagine it as anything you like. To me, the small circle
shown here seems to represent Food Engineering as it is today. It will develop through future
years into the darkness of space, in all directions, and by so doing most likely will take on the
form of an ellipse.
The Sixth International Congress on Engineering and Food is underway. Learn
everything you can from it. And above all, enjoy yourselves!

REFERENCES
1. Yano, T., Kong, J.Y., Miyawaki, O. and Nakamura, K., The "intrinsic" thennal
conductivity of wet soy protein and its use in predicting the effective thennal conductivity
of soybean curd. J. Food Sci., 1981,46, 1357-1361.
2. Yano, T., Shokuhin Kogaku no Kiso. Vol. 1 of Shokuhin Kogaku Kiso Koza, ed. T.
Yano and R. Toei, Korin Publishing Co., Tokyo, 1992, Chapter 6.
3. Shiinoki, Y. and Yano, T., Rheological properties of dispersed two-phase systems. J.
Texture Studies, 1986,17,175-185.
4. Okano, K., The propagation constant for a plain longitudinal wave in the disperse system
of viscoelastic material, II. Rep. Prog. Polym. Phys. Japan, 1961,4, 11-13.
Okano, K., On the concentration dependence of the effective modulus and viscosity of
disperse systems. Rep. Prog. Polym. Phys. Japan, 1962,5, 79-81.
5. Yano, T., Shiinoki, Y., Miyawaki, O. and Sakiyama, T., Instantaneous elastic
defonnation of a porous gel analyzed by the finite element method. J. Food Eng., 1987,
6, 217-230.
6. Yano, T., Matsumoto, S., Hayashi, H. and Kako, M., Nyuuka to Bunsan. Vol. 9 of
Shokuhin Kogaku Kiso Koza, ed. T. Yano and R. Toei, Korin Publishing Co., Tokyo,
1988, Chapter 3.
7. Reference (6), Chapter 2.
8. For example: Shimiya, Y. and Yano, T., Diffusion-controlled shrinkage and growth of
an air bubble entrained in water and in wheat flour particles. Agric. Bioi. Chem., 1987,
51, 1935-1940.
9. Iibuchi, S., Yano, T., Kawashima, M. and Nakagawa, K., Energy analysis of a Kori-
Tofu plant. J. Food Eng., 1982, 1, 17-29.
10. Torikata, Y. and Yano, T., Effects of coagulation condition on consolidation of soybean
protein coagulate. Agric. Bioi. Chem., 1988,52, 2209-2213.
11. Nagai, T. and Yano, T., Fractal structure of deformed potato starch and its sorption
characteristics. J. Food Sci., 1990,55, 1334-1337.
12. Suzuki, T. and Yano, T., Interpretation of the relationship between energy and size
reduction on the bais of the fractal concept (in Japanese). J. Soc. Powder Tech. Japan,
1989, 26, 99-102.
13. Kumagai, H., Fujii, T., Inukai, T. and Yano, T., Concentration dependence of
mechanical properties of gelatin near the sol-gel transition point. Biosci. Biotech.
Biochem., 1993,57, 532-535.
 MECHANICAL EMULSIFICATION

HELMAR SCHUBERT, HElKE KARBSTEIN

Institute of Food Process Engineering, University of Karlsruhe, Kaiserstr. 12, D-76128 Karlsruhe, Germany

ABSTRACT

Emulsification is governed by droplet disruption due to mechanical energy input and re-coalescence of unstabi-
lized droplets. Hence, the emulsification result is determined by the volumetric energy input as well as the ad-
sorption kinetics of the emulsifier. The present knowledge concerning droplet disruption and stabilization is
summarised and new theoretical and practical results are presented. Changes in the droplet size distribution during
storage of emulsions with high oil fractions are calculated using the population balance theory. Thus, the long-
term stability of o/w-emulsions can be predicted.

FUNDAMENT ALS

Emulsions consist of at least one hydrophilic and one lipophilic liquid, e.g. water and oil. Depending on the
liquid that forms the continuous or the dispersed phase they are classified into o/w- (oil-in-water-) or w/o- (water-
in-oil-) emulsions. If the dispersed phase is an emulsion, too, the system is called a polyphase emulsion. The
mean droplet diameter is usually between 0.1 and 100 JllIl. These so-called macro-emulsions are thermodynami-
cally unstable and tend to separate unless stabilized. Micro-emulsions with droplet diameters below 300 nm are
thermodynamically stable, but they will not be discussed here. Since this communication is only a brief sum-
mary, the interested reader is advised to consult the general literature on emulsions (e.g. [1,2,3]).

MECHANICAL EMULSIFICATION

Emulsification machines
Rotor - Statac Dlspcr Ing Machin Hlgb Pr ure Homogenizer
coUoid miU toothed discs dispersing macbine force

-
rotor and stator:
- variou teeth
-
various jets
smooth or toothed (design. number) c.g. fiat jet

Figure 1. Emulsification machines (schematically).

10
Stirrer: The disperse phase is mainly disrupted by shearing in turbulent flow. The energy input per unit
volume is very low, therefore stirrers are only suitable for coarse emulsions, which oUght to be further processed
if a fine emulsion is desired.
Colloid-mills: The droplets are disrupted in a conical gap between rotor and stator, each of which can be
either smooth or toothed (Fig. 1). Gap width, speed and flow rate are decisive for the shearing intensity in lami-
nar (smooth surface) or turbulent (toothed surface) flow.
Toothed discs dispersing machine: Rotor and stator consist of several concentric arranged splitted discs
of various designs (Fig. 1). Passing the rotor-stator-system the emulsion is repeatedly subjected to stress result-
ing from tangential acceleration and deceleration.
High pressure homogenizers: The emulsion is pumped by a three- to five-piston pump through a gap of
10 to 200 IJlIl (Fig. 1). Thus, a pressure of several hundred bar builds up before the gap resulting in high veloci-
ties in the gap. Droplet disruption takes place due to cavitation and turbulent flow.

Disruption of droplets
Droplets are disrupted, if the shape-preserving interfacial forces (capillary pressure) are locally exceeded by the de-
formation forces (stresses), and if the deformation time exceeds a critical breakage time lbr,cr the droplets need to
deform and break up [4,5].
Droplet disruption in laminar flow' The ratio between the deforming shear stress and the shape-preserv-
ing capillary pressure is given by the laminar Weber number. Droplets are disrupted, if the laminar Weber num-
ber Wei exceeds a critical value Wel,cr:

't d 't d max TId

Wei = 4y > Wel,cr =- 4 y and t br > t br,cr = 't-4y/d (I)

where 't is the shear stress, d the droplet diameter, y the interfacial tension and TId the viscosity of the
disperse phase. The laminar critical Weber number is a function of the viscosity ratio between the disperse phase
and the emulsion (TId/TIe) [6]. The laminar critical Weber number in minimal for 0.1 < TId/TIe < 1 and descends
steeply for TId/TIe> 5. Therefore, it is difficult to produce emulsions with high viscosity ratios in laminar flow.
Droplet disruption in turbulent flow: In turbulent flow the droplets are subjected to stress resulting from
fluctuating pressure and liquid velocity. Deformation results from Reynolds tensions. Two dimensionless num-
bers describe droplet deformation and break-up in turbulent flow, the Weber and the viscosity group [7]:

C p lf3 d 5/ 3 f.2/3
We = c and Vi _ TId (2)
t 4y /d - (y Pd d)lf2

where Pd is the density of the dispersed phase and f. the power density (power per dispersing volume).
There exists no analytical function to determine the maximum droplet diameter in turbulent flow, but limits can
be found [8]:
Vi small (TId <lOmPa·s): d max ocf.-2/5
(3)
Vi high (TId »lOmPa·s): d max ocf.-lf4T1r4

According to experiments, the mean droplet diameter is also a function of the residence time t [9].
Droplet disruption by cayitation- High velocities in the high pressure homogenizer valve result in
vapour bubbles, which flow at supersonic speed. After a normal shock the bubbles collapse, which causes high
power dissipation in local regions Get streams). According to current theories emulsification is effective, if there
are favourable conditions for the formation and growth of the vapour bubbles, an intense normal shock, a re-
stricted area in which the collapse occurs and a narrow range of collapse times [10].

Emulsification in different emulsification machines

In the gap of colloid mills there is laminar flow, if rotor-stator-systems with smooth surfaces are used, whereas
turbulence occurs, if rotor and stator are toothed. Droplets are also disrupted by turbulent flow in toothed discs
dispersing machines. In order to compare the emulsification results of different equipment, it is necessary to de-
termine the volumetric energy density Ev (energy per dispersing
volume or power per flow rate). The emulsification result, indicated by the mean droplet diameter dl,2, is propor-
tional to Ev-b :
 11
d1,2 = C. E y - b

with Ey oc t (laminar) and Ey oc E' t (turbulent) (4)

It is required, that a critical value of t or E and lbr is exceeded. The emulsification result is therefore
governed by the viscosity ratio TId/TIe for colloid mills with smooth rotor-stator-systems (Eq. (1». In turbulent
flows the mean droplet diameter of the emulsions is only a function of the energy density depending neither on
the viscosity ratio TId/TIe nor on the type of emulsification machine (Fig. 2). The factor b in Eq. (4) should be in
the order of 0.25 ...0.5. These results are only valid for droplet disruption. If there is re-coalescence of disrupted
droplets, the emulsification result is totally changed. Hence it is necessary to use emulsifiers which are able to
stabilize new droplets within milliseconds.

~~------------------~-------------------+
o toothed discs dispersing machin
J.Ull
11l'l. =0.005 ... 500
• colloid mill, cross-toothed
11/'1. = 0.005 ... 20
30
o colloid mill, smooth
11/'1. = 0.005 ... 1
• colloid mill, smooth
20 11/11. =5
o colloid mill, smooth
11/11. = 20

10

•
o
107
energy density Ev
Figure 2. Mean droplet diameter dl,2 of emulsions produced with different emulsification machines as a function
of the energy density Ey (first results, which have to be proved in further experiments).

ST ABILIZATION OF EMULSIONS

Stabilizers and Emulsifiers

Instability processes in emulsions are due to sedimentation (creaming) and coalescence. As a result the disperse
state may disappear partly or completely. The complete separation of the phases is called "breaking of an emul-
sion". Two kinds of additives are used to stabilize droplets:
Stabilizers: Stabilizers are macromolecules which increase the viscosity of the continuous phase. Hence,
the Brownian motion and creaming velocity of the droplets are reduced, which results in lower collision and coa-
lescence frequencies. Typical food stabilizers are modified starches, gelatine and cellulose derivatives.
Emulsifiers: Emulsifiers or surfactants are molecules with hydrophilic and lipophilic components. They
adsorb at the interface between the continuous and the dispersed phase. Their chemical structure, mode of action
and guidelines for their selection have been thoroughly discussed in the literature [11,12]. In order to prevent re-
coalescence of droplets during emulsification a sufficient number of emulsifier molecules have to occupy newly
formed interfaces in short time. Therefore, the emulsifier adsorption kinetics is of crucial importance for the sta-
bilization process. Measurement and results for different emulsifiers are published in [6]. The special device per-
mits to measure adsorption times, the influence of demicellisation and diffusion is excluded. Thus the results are
applicable to the emulsification process. In this case emulsifier molecules are transported to the new interfaces
by convection (laminar or turbulent flow). It is interesting, that food emulsifiers (e.g. egg yolk or whey protein)
adsorb at new interfaces at a very low speed, whereas technical surfactants (e.g. ethylenoxids) stabilize new inter-
12
faces within seconds or less (Fig. 3). The adsorption kinetics of proteins is also a function of the salt concentra-
tion and the pH-value [13].

Interfacial tension waterlparaffln 011

50
concentration of the emulsifiers Ce > erne
mN
m a: Ce = 10.2 molll b: CE =10 gil
40

c
0 ~_+ __...;La=c~prodan 50@)
·enc 30
.!
:i(,)
20
~
.!
.5
10 OK·Ester F 140 @)

0
0 5 10 s 15 2 4 8 min 12
Interface age

Figure 3. Adsorption kinetics of different emulsifiers (LEO-lO: dodecyl alcohol-IO-glycol ether, CEO-lO: cethyl
alcoool-IO-glycol ether, DK-ester: ester of saccharose and fatty acids, Lacprodan-60: spray-dried whey protein).

Influence of the emulsifier adsorption kinetics on the emulsification result

Calculation of the dispersity assumin~ no coalescence· The emulsifier adsorption kinetics determines the
interfacial tension between disperse and continuous phase during emulsification. Two limits are assumed:
1) instantaneous re-adsorption of emulsifier molecules at new interfaces (fast kinetics): New droplets
are stabilized instantaneously. The value of the interfacial tension corresponds to the equilibrium value at the
critical micelle concentration (maximum surface load). The hydromechanic limit of dispersity is reached.
2) no re-adsorption during droplet disruption (slow kinetics): During emulsification the surface load
decreases while the interfacial tension increases. As a result the minimum droplet diameter, which can be
achieved by droplet disruption, is greater than the hydromechanic limit of dispersity.
Experimental results using emulsifiers with different adsorption kinetics at low oil fractions, which pre-
vent droplet coalescence, corroborate this model [6).
Calculation of the dis.persity assl!min~ disDJption and re-coalescence: If slow emulsifiers are used and the
fraction of the disperse phase is reasonably high, re-coalescence of unstabilized new droplets occurs. This coales-
cence takes place immediately after the droplet disruption, e.g. in the pipe following the dispersing zone. In lam-
inar flow the number of collisions depends on the number of droplets and the shear rate [4). Assuming
monomodal distributions the mean droplet diameter can be predicted as a function of the emulsifier adsorption
kinetics, the pipe's dimensions and the flow rate. Droplets do not collide during transportation, if the continuous
phase has a sufficiently high yield stress (bulb flow).

LONG-TERM STABILITY

Mter production emulsions have to be stable over a given period of time (months to years). Instability may re-
sult from sedimentation (creaming) and coalescence. Factors that influence the long-term stability are: the droplet
size distribution after emulsification, filling and transport; the pH-value and the ionic strength of the continuous
phase; the zeta-potential of the droplets; the viscosity of the phases; the density difference; the volume fraction of
 13
the disperse phase and the temperature as well as physical properties. The behaviour of the droplet size distribu-
tions during storage can be monitored (storage tests) or simulated using the population balance theory [14].
0.1
J.I.IIl- 1 It = 168 h I
C''''

§
'.:1
:::I
.J::i
0_08
11em"',iooI
'E 0_06 population balance theory
:a'" A =5 . 10"20 J (Lyklema, 1968)
E =68 (Kruyt,1952)
~ r
5:::I
C'
Q)
0_04
.t: experiment
Q) x1•3 = 10.4 Ilm
§
'0 0.02
>

o ~~~~~~~;-------------~----~~~~~
0.1 1 10 100
droplet diameter d

Figure 4. Comparison of measured (experiment) and calculated ~pulation balance theory) droplet size distribu-
tions q3(d): 10 % soybean oil-in-water, T =6°C, Cion = 10- M, pH II, y = 15 cm, yges = 18 cm, t = 168 h.

Basics of the model

Initially, the total interaction energy Vt,ij between two droplets of different radius ri and rj is needed [15]. Re-
garding o/w-emulsions without emulsifier, both van der Waals and electrostatic interactions occur. They are added
according to the DL VO theory [16,17]. Two droplets coalesce only if they collide. Even in an emulsion at rest
droplets move. Collisions result from different creaming velocities and from the Brownian motion of the
droplets. The coalescence frequency ~i,j is calculated as product of the collision frequency and the coalescence
probability pi,r

1' .. 2kT -1 +r -1 )(r· +r·)+-----f

= ( ---Cr· 2xApg ((n ) (r·2 -r·2 )(r· +r·) 2J .p ..
~I,J 31']c 1 J 1 J 91']c 'Y 1 J 1 J I,J (5)

with :

(6)

where k is the Boltzmann constant, T the temperature, L\p the density difference between disperse and
continuous phase, and <p the oil fraction. The storage vessel is assumed to be subdivided into yges height layers.
The droplet collective is divided into n classes, where each class corresponds to a certain size. The number of
droplets Nk of class k, which are at time t in height layer y, is calculated. During the time interval L\t droplets
of this size do not only cream into this layer but leave it as well. During the same time droplets of this size re-
sult from coalescence of smaller droplets. If droplets of size k coalesce with other droplets, they disappear from
class k. All these separate processes are balanced for each droplet size class, each height layer and each time step:

(7)

Emulsions relevant to the food industry have droplets with diameters d > 1 J.I.IIl and oil fractions <p up to
80 %. Creaming is of crucial importance. The creaming velOCity Uk(<p) cannot be calculated using Stokes' law as
it is done in [14] (conditions: d "" 1 J.I.IIl, <p < 0,1 %). Indeed, it is a function of the oil fraction <p (Eq. (8) [18])
and therefore has to be calculated for each emulsion layer and each time step separately.
14

[5
1- cp
f(cp) =
(1 - cp 1/3) ·exp
T\c+ 2,5T\d
cp 2,5(T\c+T\d)
1 (8)
3(1- cp)

Changes in the droplet size distribution of unstable emulsions

In order to observe instability processes in unstable emulsions experiments are carried out. Soybean oil-in-water
emulsions without any emulsifier and stabilizer (unstable model emulsions) are produced varying the pH value
and the ionic strength Cion of the continuous phase as well as the droplet size distribution after emulsification
and the storage temperature. These emulsions are stored in a special storage vessel which allows sampling at dif-
ferent heights after different times [19]. Experimental and calculated values of the distribution at 16g h of storage
are shown in Fig. 4. The zeta-potential of the droplets is measured, the values of the dielectric constant E and the
Hamaker constant A are taken from the literature [20,21]. There are no fitting parameters. Not only mean droplet
diameters but also the exact shape of the droplet size distributions can be calculated. Using these calculations, the
long-term stability of emulsions can be predicted. The long-term stability of soybean oil-in-water emulsions
without emulsifier improves with increasing pH, decreasing ionic strength, increasing viscosity of the continu-
ous phase and decreasing temperature.

ACKNOWLEDGEMENT

The authors like to thank the Arbeitsgemeinschaft Industrieller Forschungsvereinigungen (AlP) and the For-
schungskreis der Erniihrungsindustrie e.V. (FEI) for their financial support.

REFERENCES

[1] Becher, P.: Encyclwedia of emulsion technolo2Y. Vol. 1-3. Marcel Dekker Inc.: New York, Basel
(1983, 1985, 1988).
[2] Larsson, K.; Friberg, S.: Food emulsions. Marcel Dekker Inc.: New York, Basel (1990).
[3] Dickinson, E.; Stainsby, G.: AdYances in food emulsions and foam. Elsevier Appl. Sci. Publ. Ltd.:
London, New York (1988).
[4] Walstra, P.: Formation of emulsions. In [1].
[5] Walstra, P.: Chern. Eng. Sci. 48 (1993), No.2, pp. 333-349.
[6] Armbruster, H.: PhD thesis. University of Kalrsuhe (1990).
[7] Hinze, J.O.: Am Inst Chern. En2. 1 (1955), No.3, pp. 289-295.
[8] Arai, K.; Konno, M.; Matunaga, Y.; Saito, S.: J Chern En2 Jap. 10 (1977), No.4, pp. 325-330.
[9] Koglin, B.; Pawelowski, J.; SchOnring, H.: Chem.-Ing.-Tech. 53 (1981), No.8, pp. 641-647.
[10] Holley, W.: in: Hochschulkurs Emulgiertechnik. University of Karlsruhe (1988).
[11] Schuster, G.: Emul2atoren filr Lebensmittel. Springer: Berlin, Heidelberg, New York, Tokyo (1985).
[12] CharaIambous, G.; Doxastakis, G.: Food emulsifiers. Elsevier: Amsterdam, Oxford, New York, Tokyo
(1989).
[13] Taiwo, K.; Karbstein, H.; Schubert, H.: Wiss AbschluBher 28 Internat Seminar. University of
Karlsruhe (1993).
[14] Reddy, S.R.: PhD thesis. University of Michigan (1981).
[15] Suzuki, A.; Ho, N.F.H.; Higuchi, W.I.: J Colloid Imerf Sci. 36 (1971), pp. 552-564.
[16] Dejarguin, B.V.; Landau, L.D.: Acta Pbysicochim USSR 14(1941), pp.633 ff.
[17] Vervey, E.J.w.; Overbeek, J.Th.G.: Theory of the stahility of IYCWhobjc colloids. Elsevier: Amsterdam
(1948).
[18] Bamea, E.; Mizrahi, J.: The Can J Chem En2. 53 (1975), pp. 461-468.
[19] Karbstein, H.; Schubert, H.: Chem.-ID2.-Tech. (will be published).
[20] Kruyt, H.R.: Colloid Science. Vol. 1. Elsevier: Amsterdam (1952).
[21] Van Leeuwen, H.P.; Lyklema, J.: Modem AslX'cts ofEJectrochemislIy 17 (1986), pp. 411-483.
 Applications of NHR to Food Science and Enqineerinq

P.S. Belton
AFRC Institute of Food Research, Norwich Research Park,
Colney Lane, Norwich, NR7 4UA

Abstract

NMR measurements are well suited for application to

heterogenous systems such as foods. They give information
on chemical environment, molecular motion and position over
a wide range of length and time scales. Typically the
length scales range from 10-1'1n to 10-2 m and time scales from
10-12 s to 10-1 s.

NMR has two very important advantages over other physico

chemical methods when applied to food science and
engineerinq. These are: that it is non-invasive and that
information may be obtained over a wide variety of length
and time scales. Because of these two properties NMR is
well suited to probing the complexities of heterogenous
systems such as food.
The range of time and length scales that are available
derive from the different sorts of information that NMR can
give. In general terms three types of information are
available which relate to chemical environment, molecular
dynamics and position in space. Often environmental results
contain mixtures of these types of information and detailed
analysis is required before separation is possible.
The primary source of information on the chemical
environment is the NMR spectrum. The chemical shift is a
reflection of the nature of the valence electron cloud and
is thus very sensitive to the detailed nature of chemical
bonds. Exchange between chemical environments is reflected
in changes in lineshape and analysis of these changes
16
enables much useful kinetic information to be obtained.
Even in circumstances where no line shape changes are
apparent useful information about chemical exchange may be
obtained by the use of saturation transfer methods.
Reference [1] gives a general discussion of these effects.
Molecular dynamics impinge upon all aspects of NMR but
they are most clearly manifested in relaxation time
measurements. At the simplest level such measurement may be
used to discriminate between solids and liquids. In solids
transverse relaxation is of the order of microseconds whilst
in liquids it may be of the order of seconds. Thus very
simple measurement of solid to liquid ratio are possible and
are widely used in the food industry [2]. More subtle
variations together with detailed analysis can give a much
fuller picture of the motion, both translational and
rotational of molecules [3]. positional information arises
from the applications of magnetic field gradients to
generate NMR images [4]. However if applied appropriately
much may be found out about translational motion such as
diffusion and flow.
Not all experiments are practical for all samples but the
developments both in instrumentation and new experimental
approaches is continually increasing the amount of
information available. Table 1 summarises the sorts of
measurements that are possible and indicates the length and
time scales about which information is given. The values
given are indicative rather than absolute limits and are
estimates for the ranges likely to be measurable in food
systems. For other systems different ranges of values may
apply.
 17
TABLE 1

Length and timescales available from NMR measurements

NMR Measurement Length Scale Time Scale
(Metres) (Seconds)
Chemical shift 10-10 - 10-8 10-3 - 10 1
Scalar coupling 10-10 - 10-9 10-3 - 10-1
Transverse relaxation 10-10 - 10-5 10-12 - 10-2
Spin lattice relaxation 10-10 - 10-4 10-12 - 10 1
Imaging 10-5 - 100 10-3 - 10 1
Diffusion and flow 10-5 - 10-2 10-3 - 10-1

References
1. Pfeffer, P.E. and Gerasimowiez, W.V. Introduction to high
resolution NMR spectroscopy and its application to in
vivo studies of agricultural systems. in NMR in
Agriculture, ed P.E. Pfeffer and W.V.Gerasimowiez, CRC
Press, Boca Raton, 1989 pp 3-70.

2. Waddington, D. Application of wide line NMR in the oils

and fats industry. In Analysis of oils and Fats, Ed.
R.J. Hamilton and J.B. Rossel, Elsevier, London 1986, pp
341-400.

3. Belton, P.S. Comments Agric. Food. Chem., 1990, ~,

179-209.

4. Andrew, E.R., Bydder, G., Griffiths, J., lIes, R. and

Styles, P., Clinical Magnetic Resonance Imaging
Spectroscopy, John Wiley and Sons, Chichester, 1990.
 PROGRESS IN PASTEURIZATION AND STERILIZATION

T. OHLSSON
SIK, Box 5401, S-402 29 GOTEBORG, Sweden

ABSTRACT

Pasteurization and sterilization are well established methods for preserving foods. The
dominating technique is in-container thermal processing where retorts with steam or hot
water are used. Modem retorts are microptocessor-controlled, with possibilities to store
processing cycles for a range of products. "Real time" process controllers are gradually
coming. Modem retorting systems have automated equipment for pre- and post-retort
handling of the containers. On-line integrity control for container leakage is under
development.

Pasteurization and sterilization can also be done in-flow in indirect heat exchangers and
direct heating equipment, using steam or electricity. Much development is taking place
in using direct electric resistance (Ohmic) heating or dielectric microwave and high
frequency heating of pumpable foods. There is a fair number of industries where
microwaves are used for rapidly heating prepared foods to pasteurization temperatures
and a few for microwave sterilization. Finally, a number of new non-thermal pasteuri-
zation and sterilization methods are being tried out at present; e.g. high pressure tech-
nique, high voltage pulses, light discharges.

INTRODUCTION

Pasteurization and sterilization are well established methods for preserving foods, with
routes back to 1810 and Nicholas Appert. Pasteurization and sterilization has been
industrialized during this century and it is today one of the dominating food preser-
vation methods. Although, the method has been under attack from other preservation
methods and from the consumer trends towards fresh foods, pasteurization and
sterilization is maintaining its very large product volume on most markets in the in-
dustrialized world.

Progress in pasteurization and sterilization must therefore be seen in the light of

a very well established industrialized technology with a wide application area in the
food industry world wide. Developments take place in a number of different tech-
nologies that each contribute to improve pasteurization and sterilization technology in
 19
terms of processing efficiency, process control and product quality. Pasteurization and
sterilization is also central in food engineering and food technology research. In my
presentation, I will review developments in the areas of:
* Packaging
* Process control
* In-package sterilization
* In-package pasteurization
* Liquid processing
* New technologies
- Electric methods
- High pressure technologies
- Emerging methods

PACKAGING

Traditional packaging materials for pasteurization and sterilization are cans made of
metal and jars made of glass. Recent developments with the new environmental driven
laws on reuse and recycling of packaging in Europe has lead to renewed interest in
glass and metal packaging. This is both related to the attitudes of the consumers and
politicians and the costs imposed on packaging materials, primarily plastic material, for
the recycling systems. Technology for metal and glass packaging continues to develop,
with improved graphics, an important aspect of packaging, as well as with reduction in
material thickness.

Retortable plastic containers have shown a great growth on the markets during
the last couple of years. Most of these containers are based on polypropylene or PET
and contain a high oxygen barrier material, sandwiched between the bulk materials.
Developments in the barrier material are very intense. Of most interest today is the
silicone dioxide barriers that is so thin that they do not limit the recyclability of the
plastic material as a monopolymer.

Flexible aluminium packaging, so called retorts pouches, have preliminary found

a market in Japan whereas both the European and American markets are small. An
interesting development here is that the so called standing pouches are gaining in
popularity because of their increased convenience and capacity.
It should be pointed out that although, all of these packaging materials are very
different in terms of their design and materials composition, retorting is done in
traditional batch retorts, using overriding air pressure.

There has been a rather substantial research on finding the optimal sterilization
temperatures for these new packaging designs, that are thinner, thus allowing for high
temperature short time processing. Typically a package with 25 mm thickness has an
optimal sterilization temperature in the range of 124 C, as evaluated by minimal
0

chemical and sensory quality changes . Thinner packages, such as retort pouches with
10 mm thickness have optimal temperatures closer to 1300 C.
20
PROCESS CONTROL

With the background of stricter legislation on product liability and the strong emphasis
on quality control and quality standards, the progress in the area of controlling all
aspects of pasteurization and sterilization processing is rapid. The increased use of
plastic, paperboard and aluminium foil based material has led to developments of non-
destructive, all-product inspection methods for testing the packaging integrity, including
video inspection techniques, pressure drop measurements and helium leakage tech-
niques.

There is also an active development in methods for evaluating and ensuring the
processing safety, using foodlike carrier materials of bacterial spores or enzymes or
even solid state "process memory cells". The carriers are mixed with the product and
the efficiency of the thermal processing is evaluated after separating them from the
product after processing,

Process control aspects of the influence of process irregularities using various

forms of computer based simulation techniques have been used to investigate the
sensivity to variations in processing and product parameters on the sterilization safety
of the process. Often, it has been demonstrated that the physical properties of the food,
that is thermal properties and the geometrical thickness, is the most important aspect to
control. The process has turned out to be surprisingly robust in terms of variations in
processing temperature.

On the equipment side, modern retorts come equipped with PLC-controllers and
small PCs for controlling the thermal processing. The controllers are primarily
sequential time event controllers, controlling opening and closing of valves, temperatu-
re in processing vessel, etc. They also have possibilities for storing processing pro-
grams for a range of products. A number of attempts have been made to commercialize
real time process control systems where the computer is keeping track of the develop-
ments during the thermal process and actively adjusts processing time or temperature to
compensate for variations happening during the cause of production.

IN-PACKAGE STERILIZATION

There is a wide range of batch and continuous retorts available for sterilization of
hermetically sealed packages. Sprays of water that covers the entire retort basket is an
increasingly popular heat transfer system for batch retorts. Some of these retorts come
with systems for indirect heating of the sterilizing and cooling water by an external
heat exchanger reducing risks of contamination through the can seams during the
cooling process from cooling water. Research into variability of processin& in modem
type of retorts has shown that the variability is normally less than +/- 0.5 C or in
terms of Po-values within +/- 1 minute.
 21
The Vatech steam vacuum process has recently been presented in France. It is a
vegetable sterilization technique where the cans are partially clinched and the steam
produced during high temperature sterilization inside the can is used to create a vacuum
after it has been double seemed and cooled. Improvements in the forms of increased
sensory quality and improve vitamin retention are claimed to be advantages of the
method.

Another major trend in sterilization is the increased automation of not only the
control of the retorts but also of pre- and post process handling of retort baskets and
individual packages. Newer installations with retorts are often highly automated with
very little personnel involved in the entire operation from can seaming to delivery of
the cases to the warehouse.

IN-PACKAGE PASTEURIZATION

Progress in the area of pasteurization of ready made meals is much coupled to the
development of the so called "sous vide" technique. This is a term that involves
vacuum packing high quality raw materials and cooking products under well controlled
processing conditions. In contrast to modem HTST-sterilization, sous vide cooking
involves long time - low temperature cooking, with the objective to avoid reaching
temperatures where irreversible damage to water holding capacity of foods occur. The
low internal temperatures give small pasteurization values. Shelf life between 0 and
3 C is usually between one week and up to five weeks. The importance of maintaining
0

limited shelf lives and storage temperatures close to 0 C, is demonstrated by recent

0

research on the heat resistance characteristics of psychotrofic strains of Clostridium

botulinum in sous vide products.

An important contribution to improved process control in sterilization and

pasteurization is the development of good manufacturing practice manuals. These are
excellent sources of relevant scientific backgrounds as well as the established industrial
practices. With the strong pressure on the market to produce more and more "like
fresh" thermal processed foods as well as lowering the amounts of additives that
contribute to protecting the food such as salt, acid and sugars, it is important that these
good manufacturing practices are adhered to strictly.

LIQUID PASTEURIZATION AND STERILIZATION

Among the traditional heat exchanger systems for thermal processing of liquids, tubular
heat exchanger has attracted much interest in recent years. These offer lower costs,
allow high pressure drops and can handle a wide range of viscosities of liquids. So
called spiral tubes have been increasingly popular. The tube is manufactured in ~ spiral
configuration of the tube walls, which improves turbulence and thus good heat transfer
by lowering the critical Reynold-number when turbulence develops in comparison to
smooth tube heat exchangers.
22
In the area of aseptic processing the primary interest is in two-phase flow with
solid food particles suspended in viscous, liquid carriers. For the control of the
process, it is important to know at what speed the particles are being travelling through
the processing equipment. The safety of the process will be determined by the fastest
moving particle and the particle that will require the longest for sterilization of the
interior, due to low surface heat transfer coefficient or/and slow interior heat conduc-
tion. Research on flow visualization of particles and residence time distribution indicate
that particles are generally travelling at lower speeds than the maximum centre line
velocity, which according to classical laminar flow theory is twice the average liquid
velocity. The general practice of using this ratio in the design of aseptic in-flow
sterilization plans gives adequate safety. The problems seem more to be one of too long
residence time for many of the particles leading to too much heat processing and
thermal damage to the particles which limit the acceptability of the process.

To overcome this limitation of single stream aseptic processing, a number of

equipment manufacturer have developed two-stream aseptic processing systems. In
these the particles are sterilized separately in specially designed steam or hot liquid
sterilizers. The particles are then mixed with the liquid carrier that is pre-sterilized in
conventional heat exchanger systems. However, food products produced both by the
single-stream and the two-stream aseptic processing systems have not met with any
greater success on the markets in Europe or Japan. Many of the products have been
withdrawn from the market.

NEW TECHNOLOGIES
Electrical methods.
Microwave heating is attracting interest both for in-package and on-line pasteurization
and sterilization of foods. The important application areas for microwave pasteurization
are prepared foods and fresh pasta in Europe. Typically a shelf life of about 3-4 weeks
is achieved with the microwave processes. Target temperatures are usually 75 0 C in the
pasteurization step. Processing to pasteurization temperatures close to 100 OC, requires
overpressure processing, e.g. in modified atmosphere for fresh pasta products. It is
customary to combine microwave pasteurization and sterilization with traditional
heating methods, where microwave heating is used for rapidly reaching the target pro-
cessing temperature, and conventional heating is used for the first step of the process.

Microwave heating is also used for in-package sterilization, for prepared food
products. Two different methods are commercially utilized in Europe today. Microwa-
ve processing offers excellent opportunities to rapidly reach the target processing
temperatures, but recent research has indicated that it is important to be able to control
the heating uniformity during the microwave heating process to be able to capitalize on
the advantages of the high temperature/short time sterilization of the microwave
process.

Microwave as well as high-frequency heating at 27 MHz is utilized in in-flow

heat treatments where the ability to rapidly and directly heating foods of complex
 23
composition is utilized. Heating time is short and temperature increase can easily be
controlled. Products such as meat slurries and prepared foods are treated. Another
electrical method is Ohmic heating, where the food is heated by their electric resistan-
ce.

Irradiation
The use of ionizing radiation for pasteurization and sterilization of foods is gradually
increasing in a number of countries. Ionizing treatments with relatively low dosage are
given to fresh products to prolong shelf life in distribution., e.g. strawberries. Another
application area is the reduction of the risks of microbiological contaminants, e.g.
Salmonella in e.g. poultry and frog legs. Thus, the improvements in quality and safety
that the ionized products offer are more and more appreciated by the consumers, in
spite of the strong emotional objections to the method heard over the years.

High pressure technology

The method of subjecting foods to high pressures in the range of thousands of at-
mospheres has been much published in last few years. To inactivate vegetative microor-
ganisms and enzymes, pressures in the range of 4000 to 6000 atmospheres are needed.
For sterilization, pressures in range of 10000 to 12000 atmospheres as well as elevated
temperatures are needed. An interesting aspect is that textural changes also take place
during high pressure treatment. As it does take some considerable time before a suffici-
ent scientific background material is established for a new processing method to be
accepted by regulatory bodies, the modifications of product characteristics might be the
most interesting aspect of high pressure technologies in the near future.

Emerging technologies
Among the emerging technologies, intensive light pulses are being investigated in the
United States. Electrodes are charged and discharged in milliseconds emitting light
pulses with extremely high power levels. The method is primarily seen as a sterilization
method for food packaging surfaces. A similar process is the high voltage pulses
method. The food is placed between two electrodes that are alternatively charged and
discharged. A high electromagnetic field strength is needed to puncture the cell mem-
branes of microorganisms, which is believed to be the principle inactivation mecha-
nisms.

CONCLUDING REMARKS

Progress in pasteurization and sterilization in the future will involve continued empha-
sis on process control and product safety. On-line sensors and integrated production in-
formation systems that allow for more flexibility and increased productivity will be
seen. The contradictory trends for increased safety and less visible preservation
methods and reductions in salt, sugar and acid content that we find on the market
today, point towards the need for more research and development on the engineering
aspects of food pasteurization and sterilization.
 FOULING AND CLEANING: MECHANISMS AND MODELS

PI FRYER, MT BELMAR-BEINY and PlR SCHREIER.

Department of Chemical Engineering,
University of Cambridge, Pembroke St,
Cambridge, CB2 3RA.
ABSTRACT
The problems of fouling of heat transfer surfaces are reviewed, with specific reference to milk
processing. Although the chemical processes which result in fouling are known the kinetics of
fouling are not. At pasteuriser temperatures, the final deposit has a mineral rich layer next to the wall,
but the initial layer of deposit is protein. Protein denaturation and aggregation reactions in the bulk
and on the surface lead to the build-up of deposit Computational models which take account of bulk
and surface reactions have been successful in fitting experimental data. The processes of cleaning are
also reviewed; an optimal concentration of cleaning agent exists which minimises the cleaning time.
THE PROBLEM OF FOULING
The thermal instability of food ingredients results in the formation of solid fouling deposit on the
inner surfaces of food process plant. Fouling deposit (i) increases resistance to heat transfer, (ii)
increases pressure drop across the exchanger, (iii) offers places where microbial growth and
corrosion can occur, and thus threatens sterility. In membrane and separation systems, fouling can
also limit mass transfer; this paper considers only hard surface fouling.
Fouling in the food industry is severe; in the petrochemical industry it is common to clean only
once a year or less, in comparison to the daily cleaning needed in food plants. In engineering practice,
a fouling resistance, Rp, is included in the equation relating the clean overall heat transfer coefficient,
UO, to that after fouling, U:
1 I
U = va + RF (1)

Fouling is a transient process. In practice, there may be an induction period in which U = UO,
followed by ajouling period in which the heat transfer coefficient decreases. Figure 1 shows ways in
which the fouling resistance changes; in food processing, fouling is so severe that it generally does
not reach a constant value before cleaning is needed.
If models for the increase in RF with time were available it might be possible to optimise the
design and operation of process plant. Fouling research [1-3] seeks to relate fouling to engineering
parameters such as time, fluid flow, temperature, heat transfer rates and materials selection and to
create rate equations. Models consider the process as a balance between deposition and removal
dRF
(h=<Pd -<Pr (2)
where <Pd and <Pr are deposition and removal fluxes; if fouling is controlled by a chemical reaction, <Pd
will be an Arrhenius function of temperature. Most fouling work has studied the milk system. The
thermal response of milk is highly complex; for an engineering analysis of fouling, including heat
transfer and flow, it is necessary to use simplified models. Information obtained by food chemists is
important in understanding the fouling behaviour of milk, but must be combined with a knowledge of
the flows and temperatures in the heat exchanger. The literature on milk fouling is now extensive [4-
6], and this paper reviews the progress made by a number of groups.
 25
MILK FOULING
Fouling problems associated with milk processing are well known. It was found [7] that milk fouling
could be reduced by preholding, and it was suggested that fouling was due to denaturation of proteins
and the decrease in solubility of milk salts with increasing temperature. This has been widely
confirmed. It is now known that the denaturation of whey proteins, especially B-Iactoglobulin, during
preheating reduces the formation of proteinaceous deposits later in the process. In practice, the
selection of preheating temperatures is a matter of compromise between reducing downstream fouling
without causing deposition in the preheater [8].
Dairy fluid fouling is a function of many variables, both chemical and physical, and seasonal
changes in milk composition are reflected in fouling. Fouling is a function of milk pH, air and urea
content of milk, and milk solids concentration. Fouling is also temperature and velocity dependent. It
has been found [9] that the activation energy for demineralised whey fouling was 315 kJ/mol, similar
to the activation energy for protein denaturation (300 kJ!mol). However, significantly lower
activation energies (ca. 90 kJ/mol) were found [10-11] for milk fouling. Both the amount and extent
of fouling decreases with increasing flow rates. This is clearest in a tubular heat exchanger, in which
it is possible to relate flow rate to the shear stress in the equipment; it is more difficult to relate fouling
to flow in plate heat exchangers because of their geometrical complexity, making it much more
difficult to extract kinetic data.

Composition of deposit. A number of chemical analyses of milk deposits have been reported:
(i) Type A: a soft, voluminous material. Starts to form at temperatures above 75°C, is maximum in
the range 95-1 10°C, and then decreases at higher temperatures. The deposit is 50-70% protein and
30-40% minerals [12-14]. At the lower end of the temperature range, most of the protein is B-
lactoglobulin (B-Ig), but at the higher end of the range it is predominantly casein [14-15].
(ii)Type B: forms at higher temperatures than type A, above 120°C. It is hard and granular; 70-80%
minerals and 10-20% protein, largely caseins. [14-15].
The amount of protein, especially B-Ig in the deposit is much higher than in raw milk; B-Ig forms
at least half of the deposit but less than 10% of the protein content of milk. Similarly, calcium and
phosphate typically account for 90 % of the mineral content of deposit, but only represent 30 % of the
mineral content in raw milk. The calcium! phosphate ratio for Type A deposit is ca. 1.5. [15-16],
indicating calcium triphosphate. Fat is a negligible constituent of both Type A and Type B deposits;
even in the processing of 36 % fat cream, fat is still a negligible component of fouling deposits [17].
MECHANISMS AND MODELS FOR FOULING

Mechanism of fouling. The induction period is characterised by unchanged heat transfer and
pressure drop; if the processes which govern the induction period were understood, it might be
possible to extend it. Several studies (e.g. [18]) have found that the nature of the surface does not
affect the fouling rate once the first layers are adsorbed. Different coatings do affect the adhesion of
deposit, suggesting that the first material to adsorb defines the strength of the deposit-surface bond.
The final deposit has a mineral-rich layer at the wall; however, different workers have found that
different species can adsorb first. Experiments described elsewhere [19] demonstrate that protein
films are the first formed at pasteuriser temperatures. The lag phases observed in protein adsorption
[20] may thus apply to fouling, with mineral diffusing through the deposit as suggested by [21,22].
Proteins are very surface active, and a clean metal surface has a large free surface energy gradient, so
it would be expected that a protein layer would form rapidly.
The thermal stability of B-Ig and other milk proteins is crucial in fouling. When heated above
about 700C, the structure of B-Ig is irreversibly altered by two separate and consecutive processes of
denaturation and aggregation. Protein first unfolds, in molecular denaturation, and then polymerises
with other denatured molecules in intermolecular aggregation. Although denaturation is reversible at
low temperatures, as molecules can return to their unfolded state on cooling, B-Iactoglobulin
aggregation is irreversible, and the resulting aggregates are insoluble. One difficulty in interpreting
the literature is that the two processes are very difficult to separate: the whole process is often
described as denaturation.
After the induction period, a number of chemical reactions contribute to fouling. Addition of
chemicals to suppress aggregation decreases fouling, whilst increasing the concentration of reactive
SH groups increases fouling [16]. B-Iactoglobulin can also form mixed aggregates with other heat-
labile proteins, such as K-casein, which may be involved in fouling [23], although the processes
involved are not yet well understood. The other main whey protein, a-lactalbumin, is also heat labile,
with denaturation reversible to about 85°C, and is present in milk fouling deposits [24].
26
Both denaturation and aggregation could be the controlling protein reaction. Aggregation has been
suggested [25-26] as the governing reaction; the effect of pH on fouling was opposite to the effect of
pH on denaturation but the same as the effect of pH on aggregation. The role of calcium and
phosphate ions in milk fouling is complex and uncertain. Addition of calcium increases both the
amount of fouling and whey aggregation [27]. Precipitation of phosphate will be a function of both
the local temperature and the resulting supersaturation of the material [28]. In the bulk, calcium ions
promote B-Ig aggregation [29], whilst on the surface they will form bridges between adsorbed protein
and adhering aggregates. The precise interactions between mineral and protein adsorption are not yet
known.
Kinetic models for fouling. B-Iactoglobulin, as the most important protein in the deposit, has
formed the basis for a number of models, in which the amount of deposit at any point in a heat
exchanger is compared with the amount of denaturation, determined by the local temperature
[15,26,30]. The amount and type of deposit formed in pilot-scale plate heat exchangers has been
compared [15] with the predicted rates of B-Iactoglobulin and immunoglobulin denaturation. The
temperature at which maximum deposition occurred corresponded to the temperature of maximum
protein denaturation.
Although the final deposit forms on the surface, protein aggregation and salt precipitation will
occur throughout the fluid wherever it is hot enough. Bulk and surface effects in reaction fouling are
well-known, and have been seen in soya protein fouling [31]. Evidence for bulk involvement in milk
fouling was found by [32]: the initial rate of fouling, after an induction period, was a simultaneous
function of surface temperature and tube Reynolds number. This was explained in terms of reaction
in the thermal boundary layer of the fluid [33]. The Reynolds number dependence was due to
changes in the volume of the thermal layer and the sticking probability of foulant, i.e. the likelihood
that a particle of foulant will adhere to the surface. A distribution of surface residence times is found.
If a turtmlent burst brings a protein to the wall, it will adhere in a surface reaction; the shorter the
adhesion time, the more likely it is that fouling will occur [34-35]. It is reasonable to suppose that
denatured or aggregated protein will react faster than native protein; reactive protein will adhere faster
than native protein, explaining the bulk dependence. Figure 2 shows experiments for the same
temperature profile and different flow rates. In all cases an increase in fouling takes place when the
bulk temperature exceeds about 75°C [36]. Similar effects are seen in other types of heat exchangers
[37]. The non-uniformity of the deposit indicates that mass transfer does not control; simple models
[38] can demonstrate that wall reactions cannot be solely responsible for fouling. Bulk processes
cannot be the only ones involved, however; the final stage in fouling must be the incorporation of
deposit into the surface, which must involve a surface reaction.
To predict fouling it is necessary to model the behaviour of heat exchangers. Any kinetic model
for fouling could be used both in the design and control of heat exchangers [39,40]. A number of
exchanger simulations have been produced [15,26,41]. The first two considered only surface
processes, whilst [41] studied B-Iactoglobulin denaturation and fouling in a plate heat exchanger for
temperatures between 70°C -122°C. The model included bulk reactions and mass transfer; reasonable
fits between experiment and theory found for temperatures up to 100°C. More details of models
showing the importance of bulk processes are given in [42].
CLEANING
Complex and expensive CIP (Cleaning-In-Place) techniques have been developed to remove deposit,
though the kinetics and processes involved are poorly understood [43]. Two types of CIP treatment
are found in milk processing, (i) two stage cleaning using alkali, commonly sodium hydroxide, and
an acid wash of nitric or phosphoric acid, and (ii) single stage cleaning using formulated detergents.
Two stage cleaning involves different doses of chemicals and rinses, and may not achieve a clean
surface [44]. Single stage methods decrease cleaning times, but need more expensive chemicals.
Cleaning is multistage [45], and involves steps that may be controlled by mass transfer, diffusion
or reaction. Cleaning chemical must be transported to the solid-liquid interface, and there make
contact with, and subsequently penetrate the deposit. The solution then reacts with the deposit, which
is then removed. The removal step depends on the fluid mechanics of the system. Any step may
control the overall cleaning process; for example, mass transfer in low shear areas will be slow,
leading to slow cleaning. It has been seen that [46,43] (i) on contact with sodium hydroxide, protein
deposit swells, and (ii) cleaning is not uniform, but occurs via the uneven removal oflumps of
swelled deposit from the surface. For kinetic studies, it is necessary to study removal of a uniform
deposit under quantified conditions, such as plates [46] or tubes [43]. Figure 3 gives a typical
cleaning curve, showing the rate of removal of deposit as a function of time.
 27
Factors which affect cleaning. The time taken to clean is a function of temperature, flow rate
and the cleaning chemical concentration. Oeaning is also affected by the finish on the heat exchanger
surface, the geometry of the heat exchanger and the overall process design. Reaction, diffusion, mass
transfer and viscosity can all be Anhenius functions of temperature; it was found [43]that the cleaning
rate becomes significantly higher above 50°C, suggesting that a chemical reaction is taking place.
Anhenius models have been fitted to cleaning data, and activation energies in the order of 80 kJ/mol
found [47-48]. The effect of concentration is complex. An optimum cleaning concentration exists
which minimises the cleaning time [49,43]. Figure 4 shows the effect of changing sodium hydroxide
concentration on the cleaning of a uniform deposit; a clear minimum is found in the region of 0.5
wt%. Above that concentration, sodium hydroxide appears to modify the surface of the deposit to
resist removal. The same effect has been seen in other systems. The use of concentrations above the
optimum both increases cleaning times and raises the effluent produced.
Models for cleaning. Despite the non-uniform nature of removal, models for cleaning have
treated the surface as uniform. Commonest are rate laws that are first order in both deposit mass per
unit area, m, and cleaning chemical concentration c, [50]:
an
dt =-kcm (3)

which has been found useful for comparing different cleaning chemicals. The removal of fouling
deposits from 50 mm pre-fouled discs was studied [46]; a first-order rate constant was fitted but
varied with time, suggesting that such a simple model is insufficient. A four-stage cleaning model,
which assumes that deposit is changed by hydroxyl ions from an initial state to an intermediate state
prior to removal has been fitted [49,18]. Deposit removal was found to control, however the
activation energy fitted to this reaction (138 kJ/mol) is high enough to suggest a chemical reaction. A
two-stage model has been fitted to data, giving activation energies in the region 40-50 kJ/mol [51].
No model currently exists which can reliably predict cleaning times in process plant. The unsteady
nature of removal suggests that a statistical approach may be necessary.
DISCUSSION: RESEARCH NEEDS
Fouling from dairy fluids results from deposition of both proteins and minerals. The initial stages of
fouling involve surface protein deposition within the induction period, and both bulk and surface
protein reactions are involved during the fouling period. Type A deposit appears to be formed by the
adhesion to the wall of proteins which have partially reacted in the bulk of the fluid. However, the
interaction between proteins and calcium phosphate is still unclear, as are the processes which control
adhesion and the rate of adhesion. The controlling kinetic step in UHT fouling is unclear, and the role
of minerals in the fouling process is not fully understood.
Three approaches to reduce or eliminate fouling are possible; modifications (i) to the heat treatment
which the fluid receives, (ii) to the design of the heat exchanger, either by changing its configuration
or to the surface finish (iii) to the food fluid. Options (i) and (iii) may be impossible because of
quality and sterility requirements, although it is possible to change the mineral content, for example in
milk-based spreads. Option (ii) may offer most scope for engineering advance. Understanding the
mechanisms of fouling is vital here. For example, if fouling is bulk-controlled, surface modifications
will not prevent deposition. However, if fouling is a surface process, it is necessary to determine
which species are deposited first, and design the surface to resist such deposition or to extend the
induction period for as long as possible. It may be possible to use protein-resistant coating to prevent
the first layer of adhesion - but processing temperatures are such, particularly in UHT processing,
that calcium phosphate will still precipitate and may adhere to the surface. Further surface
investigations will be necessary to determine the interactions of protein with stainless steel, and to see
how the surface influences the deposit.
A total solution to the fouling problem is unlikely. If fouling models are to be used to predict the
behaviour of industrial plants then some kinetic expression is needed. Without accurate models for
the local temperatures and shear stresses in equipment, it is very difficult to interpret fouling data in
terms of the basic kinetics. More work is needed in a number of areas; e.g, there are a number of
processes, such as evaporation, which have not been thoroughly studied, and in which the
mechanism of fouling is not known. Equally, there are many food products which have not been
studied and whose behaviour may be significantly different from those discussed here.
More work to understand cleaning is needed. As with fouling, the interaction between chemistry
and fluid mechanics is unclear; for example, it is not known how, when the chemical concentration is
higher than the optimum, the deposit is difficult to remove. More work is needed to investigate the
28
processes of cleaning and to define kinetic models which reflect its non-unifonn nature. Models for
the operation of plate heat exchangers, such as [41], potentially represent a significant advance; if
better models for both fouling and cleaning can be incorporated into computer programs which
describe process plant then significant advances can be made.
Acknowledgements. MTBB is supported by AFRC. PJRS is supported by a DTI LINK Scheme
with AEA Harwell, APV Baker and Unilever. PJF wishes to thank other sponsors; SERC, AEA
Harwell (SM Gotham), Cal Gavin Ltd (SE Bradley), Lever Industrial Maarsen (MR Bird).

REFERENCES
1. Somerscales, E.F.C. (1981), in: Fouling of Heat Transfer Equipment, Eds. Somerscales, E.F.C.
and Knudsen, J.G., McGraw-Hill, pp. 1-39.
2. Epstein, N. (1988), in [3].
3. Melo, L.F., Bott, T.R. and Bernardo, C.A. (1988), Fouling Science and Technology, NATO ASI
Series E, Vol. 145, Kluwer Academic Publishers.
4. Hallstriim, B., TrligArdh, C. and Lund, D.B. (1981), Fundamentals and Applications of Surface
Phenomena associated with Fouling and Cleaning in Food Processing, Univ. of Lund, Sweden.
5. Lund, D.B., Sandu, C. and Plett, C. (1985), Fouling and Cleaning in Food Processing,
University of Madison, USA.
6. Kessler, H.G and Lund, D.B. (1989), Fouling and Cleaning in Food Processing, Univ. of
Miinich, Gennany.
7. Bell, KJ. and Sanders, C.F. (1944), J. Dairy Sci., 27, 499-504.
8. Burton, H. (1988), Ultra-high-temperature Processing of Milk and Milk Products, Elsevier,
London.
9. Kessler, H.G., Fiedler, J. and Hege, W.U. (1986), Chemie Ingenieur Technik ,58,475-485.
10. Fryer, P.J. (1986), PhD. Thesis, University of Cambridge.
11. Lalande, M. and Corrieu, G. (1981), in [4].
12. Burton, H. (1968), J. Dairy Res., 35,317-330.
l3. Lalande, M., Tissier, J.P. and Corrieu, G. (1984), J. Dairy Res., 51, 557-568.
14. Tissier, J.P., Lalande, M. and Corrieu, G. (1984), in Engineering and Food. Vol.i. Engineering
Sciences in the Food Industry, ed. McKenna, B.M., Elsevier, London.
15. Lalande, M., Tissier, J.P. and Corrieu,G. (1985), Biotechnol. Prog., 1, 131-l39.
16. Skudder, P.J., Thomas, E.L., Pavey, lA. and Perkin, AG. (1981) J. Dairy Res., 48,99-113.
17. Hiddink, J., Lalande, M., Maas, AJ.R. and Streuper, A. (1986), Milchwissenschaft, 41, 542-
546.
18. Yoon, J., and Lund, J.B. (1989) in [6].
19. Belmar-Beiny, T.M, Schreier, P.J.R., and Fryer, P.I. (1993) This conference.
20. Arnebrant, T., Barton, K. and Nylander, T. (1987), J. Coli. Inter! Sci., 119,383-390.
21. Tissier, I.P., and Lalande, M. (1986), Biotechnol. Prog., 1, 218-229.
22. Foster C.L. and Green M.L. (1990), J. Dairy Res., 57, 339-348.
23. leurnink, T.I.M. (1991), Neth. Milk Dairy J., 45,23-32.
24. Arnebrant, T., Barton, K. and Nylander, T. (1986), Abs. of Papers of Amer. Chem. Soc., 191,
Apr., 196.
25. Lalande, M. and Rene, F. (1988), in [3].
25. Gotham, S.M., Fryer, P.J. and Pritchard, AM. (1992), Int.l.Food.Sci.Technol., 27, 313-327.
27. de Wit, J.N. (1981), Neth. Milk Dairy J., 35,47-64.
28. Sandu, C. (1989), Ph. D. Thesis, Univ. Madison-Wisconsin.
29. Xiong Y.L. (1992), J. Ag. Food Chem., 40, 380-384.
30. Hege, W.U. and Kessler, H.G. (1986), Internationale Zeitschriftfur Lebensmittel-Technologie
und Verfahrenstechnik, 37, 92-100.
31. Obeng, E.D.A, and Hoare, M. (1988), IChemE Symp. Ser., 88, 1213-1222.
32. Fryer, P.I. and Slater, N.K.H. (1984), in: Fouling in Heat Exchange Equipment, ASME HDT-
35, ASME, New York.
33. Paterson, W.R. and Fryer, P.I. (1988), Chem. Eng. Sci., 43, 1714-1717.
34. Fryer, P.l, and Slater, N.K.H. (1987), Chem. Eng. Commun., 57, 139-152.
35. Vatistas, N. (1989), Chem. Eng. Sci., 44, 1603-1608.
36. Gotham., S.M. (1990), PhD. Thesis, University of Cambridge.
37. Bradley, S.E. and Fryer, P.I. (1992), Biofouling, 5, 295-314.
38. Belmar-Beiny, M.T., Gotham, S.M., Fryer, PJ. and Pritchard, AM. (1993), J. Food Eng.,
19, 119-l39.
39. Fryer, PJ. and Slater, N.K.H. (1986), Chem. Eng. Sci., 41, 2363-2372.
40. Fryer, P.I., Hobin, P.J. and Mawer, S.P. (1988), Can. J. Chern. Eng., 66, 558-562.
 29
41. de Jong, P., Bouman, S., van der Linden, H.J.LJ. (1992),1. Soc. Dairy Tech., 45,3-8.
42. Schreier, P.J.R., Toyoda, I., Belmar-Beiny, M.T. and Fryer, P.J. (1993). This conference.
43. Bird, M.R. and Fryer, P.J. (1991), Trans. [ChernE. C,69, 13-21.
44. Timperley, D.A. and Smeulders, C.N.M. (1987),1. Soc. Dairy Tech., 40, 4-7.
45. Sandu, c., Lund, D.B. and Plett, E.A. (1985), in [5].
46. GraBhoff, A. (1989), in [6].
47. Perlat, M. N. (1986), Ph. D. Thesis, Univ. Lille, France.
48. Gallot-Lavalee, T. and Lalande, M. (1985), in [5].
49. de Goerderen, G., Pritchard, N.J., and Hasting, A.P.M. (1989), in [6].
50. Hoffman, W. and Reuter, H. (1984), Milchwissenschajt, 39,645-647.
51. Bird, M.R. (1993), Ph.D. Thesis, Univ. Cambridge.

Constant-rate fouling
t
Figure 1: Possible variation of
Possible Fouling RF with time. showing
Induction Period induction and fouling periods.
Period Asymptotic fouling

- Time---+

2S
c
c
Figure 2: Effect of Reynolds l'
1!20 c
c
Re = 1800
number on fouling [38]. c c
~ c
Fluid: whey protein concentrate,
Fluid inlet and outlet ~
Q 15 c
c
C

temperatures: 73, 83°C. ~ Re =5200

!2 00

~ 10
Re =7500

1.0 1.5 2.0

LENGTH OF THE TUBE (m)

Figure 3: typical cleaning curve. Figure 4: variation of cleaning time with

concentration [51].

,
0.5 . , . - - - - - - - - - - - - - - - , 60
Peak removal
rale ~
---+ ••• •
..
50

:s§ 40
~ 0.4
N~
:§9 .,
0.3
• ..§
~ Time to 30
•
bIl
peak
.~ •
j
Cil 0.2
• removal rate •
20
•
• •• •
jl
Cleaning time U
•
.~ •
0.1 10
••
0.0 +--...---.----.--,---.-~~......__r-.,.--t 0
o 2 4 6 8 10 0.0 0.5 1.0 1.5 2.0 2.5
Time (mins) Sodium hydrQxide concentration (wt%)
 SUPER CRITICAL FLUIDS: FUNDAMENTALS AND APPLICATIONS

PABLO G. DEBENEDETTI
Department of Chemical Engineering
Princeton University
Princeton, NJ 08544-5263, USA

ABSTRACT

Fluids in the approximate range of temperature and pressure 1 to 1.1 and 1 to 2,

respectively (referred to critical conditions) are commonly called supercritical. The
thermophysical properties in the supercritical region can be exploited in such applications
as selectivity enhancement in enzymatic reactions [1] to formation of biologically active
protein powders [2]. Supercritical mixtures of practical interest are characterized by a
pronounced asymmetry between conditions in the bulk, and in the solvation region
around solute molecules. This solvation region tends to be solvent-rich with respect to the
bulk. One of the most promising applications of supercritical fluids is particle formation
[3]. Two techniques exist: rapid expansion [3] and anti-solvent precipitation [2]. We have
used both methods to produce devices for the controlled release of therapeutic drugs and
biologically active protein powders.

PHYSICAL PROPERTIES

A supercritical fluid is any substance whose temperature and pressure are simultaneously
higher than their respective critical point values (Tc, Pc; Figure 1; [4]). In practice, the
term is used to describe fluids in the approximate range 1 to 1.1 (Tffc), and 1 to 2 (PlPc).
Figure 2 is a schematic density-pressure projection of the phase diagram of a fluid in the
vicinity of its critical point [4]. The fluid's isothermal compressibility is defined by (p =
molar density)

KT=.!._(a
p ap
p)
T
(1)
 31
The compressibility of any fluid diverges at its critical point. Thus, supercritical fluids are
very compressible: in sharp contrast to liquids, their density can be varied continuously
between gas- and liquid-like extremes with moderate changes of pressure.
Figure 3 compares some thermophysical properties of ideal gases, liquids, liquid
metals, and supercritical fluids [5]. Supercritical fluids can be almost as dense as liquids,
but they are considerably less viscous. Furthermore, diffusivities of small-to-medium
molecules tend to be higher in supercritical fluids than in liquids (typically 10-4 cm2/sec
versus 10-5 cm2/sec) [4]. Thus, high mass transfer rates can in principle be achieved at
supercritical conditions [5].

w
a:
:::J
~ SOLID LlOUID
w
a:
0..

GAS

TEMPERATURE

Figure 1: The supercritical region. (Reprinted from [4], © 1986 Butterworths).

2.0

1.0

o~~--~~~~~~~~~~~
0.1 1.0 10.0
PR = ~c

Figure 2: Schematic phase diagram of a fluid near the critical region. All quantities have
been scaled by their value at the critical point. (Reprinted from [4], © 1986 Butterworths)
32

'0']
10. 1 10.3 10 2 SCF
Hg

~HO
104
r 1(j2 10. 4 10

Hg Air
H2 O
10 3 1(j3 H2 O 10. 5 H2 O
SCF

10. 4 10- 6 H:!O 10

·1
SCF

Air
Hg
10 10. 5 10- 7 10-2
SCF

Air

Air

Figure 3: Comparison of thermophysical properties of air, water and mercury (at 298 K
and 1 bar) with those of supercritical carbon dioxide (Tc = 304.2K; Pc = 73.8 bar) at 310
K and 150 bar [5].

SOLVATION

A useful classification scheme divides supercritical solutions into attractive, repulsive, or

weakly attractive [6]. In an attractive mixture, the local environment around solute
molecules is solvent-rich, the solute's partial molar volume and enthalpy are large and
negative, and the system exhibits large enhancements in solubility with respect to ideal
gas conditions at the given temperature and pressure. All supercritical mixtures of
practical interest are attractive. In general, the size and characteristic interaction energy of
the solute are larger than those of the solvent in an attractive mixture, and smaller in a
repulsive one. The classification criterion [6] is given by (k, Boltzmann's constant; 1,
solute; 2, solvent; x, mole fraction)
 33
8 < 0 ~ attractive

pkT>8>0~weaklyattractive

8> pkT ~ repulsive (2)

where

o=x~oH:~L.J (3)

Figure 4 shows a comparison between the ratio of local-to-bulk solvent density as

determined by fluorescence spectroscopy and molecular dynamics simulations, for pyrene
in supercritical carbon dioxide [7]. This density augmentation is typical of attractive
behavior. Preliminary evidence [8] suggests that solubility enhancements brought about
by addition of small quantities of co-solvents are also due to a local density augmentation
of co-solvent around solute molecules.
2.5~-----------------------------'

...
>
> 2.0 0
0 •
D ••
W t:: 0
•• •
zw en
zw o •

.-
c 1.5
..J C
0
•••
< ~
o o••• •
U ...J
1.0
0 :::J
m
~
..J

0.5
0.000 0.005 0.010 0.015 0.020
BULK DENSITY (MOLlCC)
Fi!:ure 4: Comparison between local density augmentation for pyrene in supercritical
carbon dioxide, as deduced by fluorescence spectroscopy (dots), and computed by
molecular dynamics (squares). Both curves are for Tffc = 1.02. The arrow denotes the
critical density of carbon dioxide. (Reprinted from [7], © 1992 American Chemical
Society).

PARTICLE FORMATION

One of the most promising novel applications of supercritical fluids is particle formation
[3]. Particles can be formed in two complementary ways: by rapid expansion [3], and by
anti-solvent precipitation [2]. The former route is used for substances that dissolve
appreciably in supercritical solvents; the latter, for sparingly soluble substances.
34
In the rapid expansion of supercritical solutions (RESS), the idea is to dissolve a
substance at high pressure, exploiting the supercritical fluid's liquid-like solvent power,
and to precipitate it by partial decompression, exploiting its high compressibility. Rapid
decompressions in nozzles and capillaries (faster than 10- 5 sec.) give rise to high
supersaturations, and hence to small particles. Pressure reduction, a mechanical
perturbation, travels at the speed of sound and favors the rapid attainment of uniform
conditions. RESS is well-suited to produce small and uniform particles. Figure 5 shows a
composite particle formed by RESS co-precipitation of two solutes [9]. It consists of a
lovastatin needle (lovastatin is an anti-cholesterol drug) coated by poly (DL-lactic acid), a
bioerodible polymer. Such particles are of interest in controlled release of therapeutic
drugs.

Figure 5: Scanning electron micrograph of lovastatin needle embedded in a spherical poly

(DL-lactic acid) matrix, formed by RESS in supercritical carbon dioxide (Reprinted from
[9], © 1992 American Chemical Society.)

In the supercritical anti-solvent (SAS) route, the solute of interest is dissolved in a

liquid solvent, and a supercritical fluid, having low solvent power with respect to the
solute but miscible with the liquid, is added to precipitate the solute. This can be carried
out by spraying droplets into a supercritical continuum. Conditions can then be reached
such that, after an initial dissolution of the supercritical fluid in the dispersed phase, the
droplets evaporate, leaving behind a finely divided powder of the solute. Figure 6 shows
biologically active powders of insulin generated by this route [2].
 35

Figure 6: Scanning electron micrograph of insulin powders formed by the SAS route,
with carbon dioxide as anti-solvent and dimethyl sulfoxide as solvent. (Reprinted from
[2], © 1993 John Wiley & Sons, Inc.)

REFERENCES
[1] Randolph, T.W., Clark, D.S., Blanch, H.W., and Prausnitz, J.M., Enzymatic
Oxidation of Cholesterol Aggregates in Supercritical Carbon Dioxide. Science,
1988, 239, 387.

[2] Yeo, S.-D., Lim, G.-B., Debenedetti, P.G., and Bernstein, H., Formation of
Microparticulate Protein Powders Using a Supercritical Fluid Antisolvent.
Biotech. Bioeng., 1993,41,341.

[3] Tom, J.W., and Debenedetti, P.G., Particle Formation with Supercritical Fluids-
A Review. J. Aerosol Sci., 1991,22, 555.

[4] McHugh, M. A, and Krukonis, V.J. Supercritical Fluid Extraction. Principles

and Practice. Butterworths, Stoneham, MA, 1986.

[5] Debenedetti, P.G., and Reid, R.C., Diffusion and Mass Transfer in Supercritical
Fluids.AIChEJ., 1986,32, 2034.

[6] Petsche, LB., and Debenedetti, P.G., Influence of Solute-Solvent Asymmetry

Upon the Behavior of Dilute Supercritical Mixtures. J. Phys. Chern., 1991,95,
386.

[7] Knutson, B.L., Tomasko, D.L., Eckert, C.A, Debenedetti, P.G., and Chialvo,
AA, Local Density Augmentation in Supercritical Solutions. A Comparison
Between Fluorescence Spectroscopy and Molecular Dynamics Results. ACS
Symp. Ser., 1992,488, 60.

[8] Kim, S., and Johnston, K.P., Molecular Interactions in Dilute Supercritical Fluid
Solutions, Ind. Eng. Chern. Res., 1987,26, 1207.

[9] Tom, J.W., Lim, G.-B., Debenedetti, P.G., and Prud'homme, R.K.,
Applications of Supercritical Fluids in the Controlled Release of Drugs. ACS
Symp. Ser., 1993,514, 238.
 NEURAL NETWORKS IN BIOPROCESS ENGINEERING

YI-HONG ZHU. SUSAN LINKO. TERO EERIKAINEN and PEKKA LINKO

Laboratory of Biotechnology and Food Engineering.
Helsinki University of Technology
SF-02150 Espoo. Finland

ABSTRACT

Neural networks are computer programs initially developed to mimic the function of brain. They are
characterized by their ability to learn from past experiences. The superwised backpropagation learning
procedure presented in the late 1980s opened up the way to practical engineering applications involving
dynamic. highly non-linear systems. Neural networks have been demonstrated to be a powerful tool in control.
estimation. fault detection. time-series analysis. etc. The paper discusses briefly the principles of mUltilayer.
backpropagation networks and describes a program package written in Microsop VisualC++'" for Windows"'.
Determining the end point of /3-galactosidase fermentation by multistep ahead prediction. estimating of
substrate and product in lysine fermentation. and dynamic control of an extrusion cooker are given as examles
of neuroengineering applications.

INTRODUCTION

Conventional bioprocess control strategies frequently present difficulties owing to uncertainties. complex
nature of biological processes. and lack of suitable sensors. In such situations fuzzy reasoning has been shown
to be helpful [1-3]. Although fuzzy chips and development tools have recently become available [4]. the
construction of fuzzy knowledge based systems can be costly and time consuming. Novel estimation and
prediction algorithms have recently become the object of great interest as so called software sensors [5].
Particularly. knowledge representation in the form of neural network models [6.7] offer a unique solution to a
number of problems typical of bioengineering. such as extimation [8-10] and control [11. 12]. Neural
networks are computer programs operating under parallel distributed paradigm. Just like brain a neural
network program is composed of a large number of interconnected processing elements called neurons. The
first binary model for a single neuron in brain was presented by McCulloch and Pitts already in 1943. and the
first neural network structure as model of brain by Rosenblatt 15 years later [13]. The relatively simple system
was capable of solving only problems that are linearly separable. with limited application potential [14]. Only
after more complex systems with supervised backpropagation learning algorithm were introduced in the late
1980s [15] the great potential of neural networks in engineering applications involving dynamic. non-linear
systems was realized [7]. The paper presents a brief introduction to neural networks in general and describes a
few examples of bioengineering applications. Of the vast amount of general literature on neural network and
their application only a few can be refered to here [16-18].
 37
HOW ARE NEURAL NETWORKS CONSTRUCTED?

A neural network model is an abstract specification of the neural network paradigm. In this paper only
software applications of multilayer backpropagation networks are considered. Figure I shows a general
graphical representation of a feedforward, multilayer neural network.

Architecture
The architecture of a neural network describes how the network is constructed from layers of interconnected
neurons as processing elements. A feedforward, backpropagation network (Figure 1) has been most often
applied to non-linear engineering problems [6, 19,20]. All neurons do not have to be connected at all points,
and some may receive a feedback from others. A recurrent network is a feedback network in which the current
activation state is a function both of the previous activation state and of the inputs. A spatiotemporal network
can deal with inputs and outputs that are explicit functions of time.

Topology
Neural network topology is given by the number of neurons in respective layers. According to the convention
followed in this paper the topology of the network in Figure 1 is L-M-N. However, neural network
terminology is yet to be standardized. There is usually one input layer, one or more hidden layers, and one
output layer. The number of neurons in the input and output layers is determined by the number of variables
involved in the problem solving. The number of hidden layers and the number of neurons in each hidden layer
is related to the converging performance of the output error function during the learning procedure described
below. The number of hidden neurons is usually detennined by trial and error.

Function
Each neuron receives scaled inputs from other neurons, or from the outside. The neurons in the input layer
transfer input values scaled for example to within a range of [0,1] to the appropriate neurons in the hidden
layer through a set of weighted connections. Each connection between two neurons has a weight coefficient.
A weight vector is given by the set of weights in a given layer. Each neuron in the hidden and output layers
calculates a weighted sum of its inputs and passes the result through a non-linear or, in some cases, linear
transfer function. A sigmoid transfer function is most commonly employed, and was also used in the present
work.

Input Layer Hidden Layer Output Layer

~
.i
0'1

e ot1
e
Weight acljuslment ot2
by delta rule of the
bGckpropagation algorithm
e otN

Figure I. The architecture and training principle of a three layer neural network.
38
Learning
A neural network is trained by iteration, repeated presentation of representative exemplar input/output vector
pairs (or pair patterns) in an analogy to a child learning through repeated trial and error. After repeated
learning cycles an internal model is obtained which can then be used in similar but unknown situations. The
knowledge is acquired by adjusting the weights of neural connections through minimizing the error between
the actual and desired outputs of the network. The backpropagation learning procedure illustrated in Figure I,
and used in this work, proceeds essentially as follows:

Step 1 Small initial random values to all weights are assigned;

Step 2 A selected input vector (1) of scaled elements from one sample data me to the input layer is
pr~sented;
Step 3 The inputs are trasfered to the hidden layer, and each neuron in the hidden layer calculates the
weighted sum of its inputs;
Step 4 The output of the hidden layer OJ is calculated as follows:

OJ - f[ ~ W'j ·1, •• j ] lSjSM (I)

StepS The output O'k of the output layer is calculated as follows:

O'k=f[ I)=1
Wjk 'Or S'k 1 15,k5,N (2)

Sj and S~ in the equations (I) and (2) are internal bias terms, and f was in the present work a
sigmoid non-linear transfer function of the type 1('0) = (I + e-U)-I;
Step 6 The difference !!1 representing the error between the actual output (O~ ) and the desired output
(O'k) is then obtained and the error factors for generating the !!1 weights in the output and
hidden layers are calculated by the least mean square principle;
Step 7 The weight matrices are updated by the learning rule adopted;
StepS Next iteration cycle is began at Step 2, and the training is stopped either after a predetermined
number of iterations or after a defined minimum error is reached.

By following the above procedure the weights are gradually adjusted by shifting towards their own minimum
along the error surface in the weight space, the height of which at any point is equal to the error function
value. The weight space is a hyperspace of (Z + 1) dimensions, where Z is the number of weights in the
network. The bias term ( S ), scaled in this work to unity with a weight attached to it determines the
coordinate space of non-linearity, and minimizes the chance of getting trapped in local minimum in the
gradient descent procedure. There are many ways of improving the training and performance of a neural
network. A careful selection of the input and output variables, and the type of information used in training is
of almost importance. Adding recurrent loops and time delays may be helpful. An increase in the number of
the hidden neurons and layers is often helpful, but too many easily results in a poor performance.
Reinitializing weights or making random step changes in weights during training can be useful. Perhaps the
easiest and most effective way is adjusting the momentum term a, implemeted by adding a fraction of the last
weight change to the next set of weights (included in Step 6).

PROGRAMMING ENVmONMENT

A novel neural network programming package based on the back-propagation training algorithm with a
momentum term was written in Microsoft® VisualC++'" for Windows'" 3.1. A number of attractive features
make the program useful and user-friendly in a variety of applications such as simulation, state estimation,
multi-step ahead prediction, and dynamic modelling and control. The user interface has a main application
window and special child windows for example for selection of the topology and epoch length, for editing of
the running parameters and for error history in training, and for output. Several training experiments can be
simultaneously rnn, and displayed graphically and dynamically. The program was written to support on-line,
 39
real-time applications. For most applications a personal computer with an Intel 486DX66 processor, 200 MB
hard disc, and 8 MB memory on the main board was employed. The superboard 486 VLB with VESA Local
Bus design provided direct 32-Bit data transfer between a video card, a controller card and the CUP at CPU
speed. The data files used for training and testing of the neural network models were prepared in text mode in
MicrosoP Excel.

LYSINE PRODUCfION

In industrial fermentations substrate and product concentrations are usually monitored off-line. For on-line
control an accurate state estimation model would be invaluable. A well trained neural network model offers a
novel way for on-line state estimation on the basis of routinly measured parameters such as oxygen uptake rate
(OUR) and carbon dioxide evolution rate (CER) [8-10]. Lysine fermentation is presented as an example case.
Lysine is an esential amino acid, which is one of the limiting nutrients in most plant proteins. Consequently,
lysine is an important additive in animal feeds. The annual production of lysine has been estimated at about
200,000 tonnes. Figure 2a shows a feedback network of 9-12-2 topology employed in simultaneous estimation
of sugar and lysine in industrial lysine fermentation by a Brevibacterium flavum strain. The input vector
consisted of integral OUR and CER, RQ, and 3 time delays of the output vector. Figure 2b shows an example
case where consumed sugar and total lysine were simultaneously estimated with a trained neural network in
the case of a fed-batch fermentation. Except for some oscillation the results were clearly quite promising.

18 4

L(t,.3) co 15
L(t,.2)
L(t,.l) ~ 12
e,AI
'-'
I. OUR
LeER ~ 9 e_, "~

o.:~rP~ L(t) =' -. ,~

RQ '" ~'II'.
S(t,.l)
O"'::,..<:,J~,.". S( t)
~=' 6 --. . .
,"'.
S(t,.2)
'"c
0
/ '". e
/'.
S(t,.3)
u 3
'
Ie

0 ...
~~~~----~----~----~ 0
0 , 20 40 60 80
S = Consumed ugar
L = Total Iyai De Tirre (h)
(a) (b)

Figure 2. Neural network architecture (a) for the simultaneous estimation of consumed sugar and total
lysine(b) in a fed-batch lysine fermentation.

p-GALACTOSIDASE PRODUCTION

p-Galactosidase is an important enzyme used for example by the dairy industry in the hydrolysis of lactose in
milk and whey. A novel process for the production of p-galactosidase by the autolytic strain of the lactic acid
bacterium Streptococcus salivarius subsp. thermopphilus IlF has been recently described [21]. In this case the
intracellular enzyme is rapidly released into the medium after a relatively short growth period, and an
accurate advance prediction of the end point would be of great importance. In this case a neural network of a
relatively simple topology of 3-8-3 gave good results. The input vector consisted only of ammonia feed
together with 2 time delays, and the output vector of biomass expressed as turbidity, together with values
predicted ahead for 2 time intervals of 0.25 hours each. An excellent fit to experimental data can be seen in
Figure 3. The maximum p-galactosidase activity was always reached about one our after the maximum
turbidity.
40
120 12 12

90 9 9
] 0')
....
><
60 6
~ ~ 6
~ -<
30 3 3

0 0 0 •
0 2 4 6 8 5 6 7 8 9 10
Time (h) Time (h)
Figure 3. Training data (left, shadowed area; - , ammonia consumption and 0 and • measured turbidity values
of 2 fermentations) for a neural network used in the estimation and prediction in bacterial ~-galactosidase
fermentation, and neural estimation(- at time 1 ) and prediction (-- at time t+ 1, --- at time 1+2) of turbidity.

EXTRUSION COOKING

Extrusion cooking is one of the most versatile food processes. However, owing to the complexity of the basic
phenomena taking place during processing efficient on-line control presents problems [11]. Neural networks
of varying architecture and topology were investigated in the control of extrusion baking of flat bread.

Statu of tile data flies used

File Name of The Sample Data
The Sample Dill : 75 Patterns
File Nlme of The Weights:

Topology of The Neurll Network

5 -10 - 1 Neuron" In The Neurll
WIth Z Output TIme Delays
~
~j Running Information - - - --4
!t "crating Loops 31
Current ErrQr Sym ; 0 . 73~6Z

~
Prevfuoa Error Sum : 0.13Z61

111111···· ····-------······ · ··· ·:

Elapsed TIme (seq : 10.16
Rool tr.4ean Square(RMS) : 5.Z~
lO

Figure 4. Dynamic neural network control of special mechanical energy (SME) in twin-screw extrusion
cooking of flat bread on the basis of screw speed (SS) and its 3 time delays, together with display child
windows showing the neural network structure, and the application window.
 41
Figure 4a illustrates the excellent results obtained in dynamic control of specific mechanical energy (SME) on
the basis of screw speed (SS) alone, employing the neural network shown in Figure 4b.

EPLIOGUE

The few presented example cases, representing food related fermentations and extrusion cooking clearly
illustrate the power and great potential of neural networks in bioengineering. There is no doubt that neural
networks are going to play an increasing role in control and a number of other applications in the near future.

REFERENCES

1. Linko, P., Uncertainties, fuzzy reasoning and expert systems in bioengineering. Ann. New York Acad.
Sci. 1988,542,83-101.
2. Eerikliinen, T. and Linko, P., Extrusion cooking modelling, control and optimization. In Extrusion
Cooking, eds C. Mercier, P. Linko and J. M. Harper, American Association of Cereal Chemists, St. Paul,
MN.
3. Siimes, T., Nakajima, M., Yada, H., Asama, H., Nagamune, T., Linko, P. and Endo, I., Object-oriented
fuzzy expert systems for on-line diagnosing and control of bioprocesses. Appl. Microbiol. Biotechnol.
1992,37,756-761.
4. Borron,l 1 Putting Fuzzy logic into focus. Byte 1993, 18(4), 111-118.
5. Montague, G. A., Morris, A. 1 and Tham, M. T., Enhancing bioprocess operability with generic
software sensors. J. Biotechnol. 1992, 25, 183-201.
6. Linko, P. and Zhu, Y.-H., Neural network programming in bioprocess variable estimation and state
prediction. J. Biotechnol. 1991,21,253-270.
7. Linko, P. and Zhu, Y.-H., Neural networks in bioengineering. Kemia-Kemi 1992, 19,215-220.
8. Linko, P. and Zhu, Y.-H., Neural network modelling for real-time variable estimation and prediction in
the control of glucoamylase fermentation. Process Biochem. 1992,27,275-283.
9. Simutis, R., Havlik, I. and Liibbert, A., Fuzzy-aided neural network for real-time state estimation and
process prediction in the alcohol formation step of production-scale beer brewing. J. Biotechnol. 1993,
27, 203-215.
10. Linko, P., Uemura, K., Zhu, Y.-H. and Eerikliinen, T., Application of neural network models in fuzzy
extrusion. Trans. I. Chem. E. 1992, 70 Part C, 131 -137 .
11. Di Massimo, c., Lant. P. A., Saunders, A., Montague, G. A., Tham, M. T. and Morris. J. A., Bioprocess
applications of model-based estimation techniques. J. Chem. Tech. Biotechnol. 1992,53,265-277.
12. Chtourou, M., Najim, K., Roux, G. and Dahhou, B., Control of a bioreactor using a neural network.
Bioprocess Eng. 1992,8,251-254. NY.
13. Rosenblatt. F., Principles of Neurodynamics, Spartan, New York, 1962.
14. Minsky, M. and Papet, S., Perceptrons, MIT Press, Cambridge, MA, 1969.
15. Rumelhart, D. E. and McClelland, J. E. eds, Parallel Distributed Processing: Explorations in the
Microstructure of Cognition, MIT Press, Cambridge, MA, 1986.
16. Hecht-Nielsen, R., Neurocomputing, Addison-Wesley Pulblishing Company, Reading, MA, 1990.
17. Hertz, 1, Krogh, A. and Palmer, R. G., Introduction to the Theory Of Neural Computing. Addison-
Wesley Publishing Company, Reading, MA, 1990.
18. Miller, W. T III, Sutton, R. S. and Werbos, P. 1 eds. Neural Networks for Control, MIT Press,
Cambridge, MA, 1990.
19. Hoskins, J. C. and Himmelblau, D. M., Artiftfical neural network models of knowledge representation in
chemical engineering. Comput. Chem. Eng. 1988,9,881 -890.
20. Lippman, R. P., An introduction to computing with neural nets. IEEE AASP Mag. 19874(2).4-22.
 IMPLEMENTATION OF MODEL BASED OPTIMIZATION OF PRODUCT
QUALITY PARAMETERS IN DAIRY PROCESSES

Joost van der Linden

Netherlands Institute for Dairy Research (NIZO)
P.O. Box 20, 6710 BA Ede, The Netherlands

ABSTRACT

As a consequence of the continuing diversification of food and increased product

quality demands, process complexity has increased to such an extent that nowadays
process development in the food industry is the limiting and conditioning framework
for the development and production of foodstuffs.
Process developments are primarily based on generalised scientific knowledge. But the
implementation of these complex mathematical modelling results in practice is unique
and specific for each application and has to guarantee the realisation of the required
outcome.
The implementation of modelling results in SME's (Small and Medium Enterprises) in
the food industry is significantly facilitated by the development of user-oriented
computer programs. The optimal outcome can be determined in relation to the
demanded product quality or the required cash flow.

INTRODUCTION

At the meeting "Strategies 2000" of all the working parties of the European Federation
of Chemical Engineers (EFChE) at Karlsruhe in Germany [1], the Food Working Party
described the activities in the food industries as follows: "The primary goal of the food
industry is the production of high quality food at acceptable prices under the conditions
of minimizing waste and emissions".
Achieving this goal is nowadays complicated on the one hand by the complexity of
food composition, increased product quality demands, reduced product life-cycles and
 43
the continuing diversification of food products [2]. On the other hand the design and
operation of a competitive production plant requires a rationalization of the production
activities, which means not a qualitative but a quantitative approach. A quantitative
approach requires mathematical models to optimize the investment of new equipment
and the operation of flexible installations. The development of software based on
mathematical models creates the opportunity to transfer generalised scientific
knowledge to industrial practice by implementing unique and specific program
application results. Examples of user-oriented software developed for process and
equipment design in the dairy industry will be presented.

COMPETITIVENESS

The small margins in the food industry require well-considered investment decisions
and an optimal control of return on sellings and on costs of production. The relation
between these challenges and the final profits in respect of time can be demonstrated by
the realised cashflow during the life time of the production installation.
The accumulated cashflow is an accepted (and excellent) tool to present the profitability
of a financial activity and is defined as the difference between incomings and outgoings
in the future of a company project.
Before the start of an activity an investment decision for an optimal design of an
installation can be tuned by the application of an adequate mathematical model. After
the start of the operation, competitiveness requires improvement of the operation
efficiency, which can be achieved by enhancing the product quality and reducing the
production costs. Therefore the development of mathematical models also plays a
dominant role in the optimization and control of the operation and additionally gives, in
many cases, the opportunity for indirect measurement of product quality parameters
that can only be measured off-line.

CASHFLOW

accumulated c
cashflow

time
(years)

Figure 1. Cumulative cash flow of an investment in respect of time.

44
Generally speaking, no process is mature. The aim of process and equipment
development is directed to a number of improvements:
to increase a product life-cycle the relevant product quality parameters have to be
improved or a constant product quality level has to be maintained during a
production run,
the investment costs can be reduced by avoiding overdimensioning of the
equipment, by retrofitting of existing equipment or by process upgrading,
to reduce operation costs the production capacity of a plant has to be increased or
made more flexible and
operation costs can be decreased by reduction of production losses, energy
consumption or environmental waste and emissions.

MODELLING

To achieve competitiveness in business, rationalisation of activities by the application of

scientific knowledge by generalised models in user-oriented software appears to be a
fruitful approach.
If a mathematical model is defined as a system in which the relations between the
system elements are described by a set of equations, advantages but also restrictions of
modelling can be mentioned:
nowadays complex systems can be managed with a computer and process behaviour
under various and dynamic conditions can be predicted without or with only a few
experiments. By intelligent use of materials and energy, costs can be reduced,
modelling also saves time and costs in learning to understand a process by allowing
a faster information response between man and process model and
the exploration of the process beyond the constraints can be done in a safe way.
Generally the application of models is restricted by a lack of physical and (bio)chemical
data of mostly empirically developed processes in the food industry and by the dilemma
between the perfection and the completeness of a model. Detailed models give
moderate accuracy and have to be modified or simplified. But a simple model with an
accurate fit often gives an over-simplified impression of reality.
Within the large variety of models we can distinguish the white-box and the black-box
model as the extremes. White-box models are characterised by their deterministic
origin. By an analysis of the elements of a process system and the relations between
these elements, a mathematical process description is derived using first principles:
mass and component balances, state and transport equations such as kinetic and Navier-
Stokes equations. The developments of white-box models is rather time consuming but
has a general validity and mostly a limited number of fitparameters. White-box models
 45
are suitable for new equipment and for retrofitting and give a lot of understanding of
the process behaviour under stationary and dynamic conditions.
Black-box models are based on input/output analysis of existing installations and have a
stochastic character by using a statistical approach. The model development takes less
time but the results are specific in relation to the installation investigated and the
product used, and only valid around the tested operation condition. The application is
more or less restricted to control and tuning actions, requires handling of a large
number of fitparameters and does not give an understanding of process behaviour.
For practical reasons "grey-box" models are the most satisfactory for flexible
application in computer programs.

SOFTWARE DEVELOPMENT

The implementation of modelling results in small and medium enterprised (SME's) in

the food industry by the transfer of generalised scientific knowledge, aggregated by
finite differences methods and dynamic optimization methods, is significantly facilitated
by the development of user-oriented software.
The process engineering research in the dairy industry is directed to the physical,
chemical and biological transformations in the various production processes. The
interactions between the conversion and selectivity of the selected relevant key
components, such as the most important product quality parameters of the process, and
the residence time, residence distribution time and remaining process conditions, can be
modelled. These quantitative process descriptions have to verified experimentally and
supplemented with data from the literature.
Two categories of processes can be distinguished. Stationary models describe processes
in equilibrium or in steady state. Dynamic models represent the unsteady state
occurring during the start-up of a process, in non-isothermal processes or after a
modification of the process settings. Depending on the desired calculation options such
as process simulation, process and equipment optimization, process control or
equipment design, modules of the model equations, the data input, the data output and
the data files are developed and put together in a menu-driven program.

SOFTW ARE APPLICATION

In table 1 a survey of the applications and the outcomes of a number of software

programs developed by NIZO is presented.
46
TABLE 1

NIZO software

program dairy process application outcome

HeatCARD direct and indirect simulation, process reduced fouling,

heat treatment design, retrofit and optimal conversion
optimization of key components [3]
EvaSim multistage falling- simulation, process reduced fouling,
film evaporation design and retrofit re-arrangement of
stages [4]
EvaDes multistage falling- pre-engineering tube diameter/length
film evaporation equipment design ratio, number of
tubes, reduced energy
consumption
STREAM multistage falling- dynamic process flexible process
film evaporation control development operation [5] opti-
mization of start-stop
operations
DrySpec2 spray drying simulation, process In-Solubility-Index
management control, reduced
energy consumption
[6]
DrySim spray drying 2D-simulation, process agglomeration
and equipment design conditions, component
transformations
PMS-MP membrane separation optimization, process optimal cashflow [7],
control reduced energy
consumption
ProdWaste start-stop and cleaning optimization, process product losses, energy
(CIP) and piping design and emissions
reduction, optimal
tube diameter/length
ratio
ClearSys start-stop and cleaning process analysis and energy and emissions,
(CIP) optimization reduction, GMP
DairyBase dairy products and engineering database physical data
processes presentation in
graphical form, tables
and equations
 47
CONCLUSIONS

The complexity of food processes and the complex composition of food products
means that the empirical approaches to exisiting and new production processes are
inadequate. Competitive operation of production installations therefore requires
quantitative process descriptions by mathematical modelling based on first principles.
- Investment decisions and the determination of optimal stationary and dynamic plant
operation conditions can be specified by the application of versatile computerised
models.
- The implementation of modelling results in small and medium enterprises is
significantly facilitated by the development of user-oriented software.

REFERENCES

1. Strategy 2000, Proceedings of the Fourth World Congress of Chemical Engineering,

Karlsruhe, June, 1991, pp. 725.

2. Bilderbeek, J., Brinkman, S., de Leeuw, A.C.J., Poly Bedrijfskundig Zakboek,

Koninklij ke PBNA, Arnhem, 1990.

3. De Jong, P., Van der Linden, H.J.L.J., Process design based on heat-induced
transformations of milk components. Proceedings of the International Dairy
Federation Seminar Munich, August 1992.

4. Bouman, S., Waalewijn, R., De Jong, P., Van der Linden, H.J.L.J., Design of
falling-film evaporators in the dairy industry. Journal of the Society of Dairy
Technology, submitted January 1993.

5. Quaak, P., Gerritsen, J .B.M., Modelling dynamic behaviour of multiple-effect

falling-film evaporators. European Symposium on computer applications in chemical
engineering, The Hague, May 1990.

6. Straatsma, J., Van Houwelingen, G., Meulman, A.P., Steenbergen, A.E.,

DrySpec2: a computer model of a two stage dryer. Journal of the Society of Dairy
Technology, 1991,4, 107-111.

7. Van Boxtel, A.J.B. and Otten, Z.E.H., New strategies for economic optimal
membrane fouling control based on dynamic optimization. Desalination,
90 (1993) 363-377.
 PHYSIOLOGICAL FUNCTIONS OF FOODS

SOl CHI ARAI

DEPARTMENT OF AGRICULTURAL CHEMISTRY,
THE UNIVERSITY OF TOKYO
BUNKYO-KU, TOKYO 113, JAPAN

ABSTRACT

Food has three categories of function. Besides primary and

secondary functions that contribute to nutrition and prefer-
ence , respectively, the "third" function exists which is in-
volved in modulation of physiological systems of our body. The
intake of some physiologically functional substances of foods
can be effective in preventing us from diseases that may be
caused by some disorders of the physiological systems.
Technologically , it is possible to design physiologically func-
tional foods which would satisfy our today I s demand for pre-
venting diseases by eating. Currently, a three-year national
project (Grant-in-Aid for Scientific Research on priority Area
No. 320 sponsored by the Japan Ministry of Education, Science
and Culture) is proceeding on the title "Analysis and Molecular
Design of Physiologically Functional Foods". Administratively,
the Japan Ministry of Health has initiated a policy of approv-
ing several foods of this category in term of "food for speci-
fied health use". Our laboratory is engaged in studies of de-
signing rice-based foods with antiviral and antiallergic func-
tions. Detailed data are presented of these topics.

INTRODUCTION

The quality of food is evaluated often by its function to our

body. Since any food is basically a nutrient supplier to us,
its nutritional function is understood to be of primary impor-
tance. This understanding may be accepted almost in any seg-
ment of the world in any period of the history. It is also
understood that any food should have a palatable taste.
Actually, the eating quality of most foods depend on their
function to our sensory systems. Such being the case, the
function of food has long been studied mostly from the aspects
of both nutrition and preference.
In the meantime, a different situation arose in Japan par-
 49
ticularly in the 1980 I s when people began to seriously recog-
nize the problem of a rapid increase in advanced age popula-
tion. Accordingly, social attention began to be paid to
prevention of adult and geriatric diseases through daily food
life. In connection with this situation, basic research was
largely activated to investigate the "third" function of foods
that may be contribute to disease prevention.

PHYSIOLOGICALLY FUNCTIONAL SUBSTANCES OF FOOD ORIGIN

A great many studies conducted in recent years have shown that

there are food-originating substances with the "third" function
involved in preventing diseases by modulation of immune, endo-
crine, neuron, circulation, and digestion systems. Conceptu-
ally, the cellular system may be included. Actually, a number
of substances possibly contributing to modulation of these
physiological systems have been elucidated for chemical struc-
ture and physiological function (TABLE 1).

CONCEPT OF "PHYSIOLOGICALLY FUNCTIONAL FOOD" AND ITS

REALIZATION IN TERMS OF "FOOD FOR SPECIFIED HEALTH USE"

In principle, physiologically functional foods are designed and

constructed based on substances with the "third" function.
They lie in a position between conventional foods andmedicines,
targeting the intermediate, semi~healthy state which is under-
stood as a premonitory (FIGURE 1). To be brief, any physiolog-
ically functional food should thus be used for prevention of a
particular disease at the stage of premonitory, not for remedy
of the disease at the stage of development. This concept was
proposed and followed up in two national projects sponsored by
the Japan Ministry of Education, Science and Culture (1)(2).
Meanwhile, the Japan Ministry of Health has initiated an in-
vestigation of the physiologically functional food from an
administrative standpoint and is going to officially approve
several items in terms of "food for specified health use".

DESIGNING THE PHYSIOLOGICALLY FUNCTIONAL FOOD

In Japan, the third three-year national project (Grant-in-Aid

for Scientific Research on Priority Area No. 320 sponsored by
the Ministry of Education, Science and Culture), chaired by the
author himself, has started which is entitled "Analysis and
Molecular Design of Physiologically Functional Foods". Part of
the study in this project is directed toward investigating
methodologies for the design. In principle, the design is
based on using conventional foods or their constituents as
starting materials. In particular, those which we have cus-
tomarily been consumed in our daily menus may be preferable.
Indeed, many foods per se contain substances with the "third"
function. However , it is probably a general case that their
quantities are little if any and thus no expected function may
be expr~ssed to any significant extent. One method for solving
this problem is to separate and purify the substances. It is
also an effective method to concentrate these substances by re-
moving other constituents including conventional nutrients.
50
TABLE 1
Food-originating substances with physiological functions

Food Substance Possible function or effect

For the immune system

Milk Immunoglobulin immunoactivation
Wheat Lipopolysacccharide immunostimulation
Lobster Chitin immunostimulation
Milk Casein peptide Macrophage activation

For the endocrine system

Red pepper Capsaicin Antiobesic function*
Soybean Trypsin inhibitor Cholecystokinin secretion
Soybean Glycinin Antidiabetic function*

For the nerve system

Milk Casein peptide Opioid function
Milk Casein peptide Opioid antagonism
Wheat Gluten peptide Opioid function

For the circulatory system

Milk Casein peptide Hypotensive effect
Corn Zein peptide Hypotensive effect
Soybean Glycinin peptide Hypotensive effect
Fish Myofibril peptide Hypotensive effect
Soybean Glycinin Lowering cholesterol
Lobster Chitosan Lowering cholesterol
Oil seeds ,,(-Linoleic acid Lowering cholesterol
Fish Eicosapentanoic acid Regulating coagulation
Cattle Heme Producing hemoglobin

For the digestive system

Milk Cpp** Accelerated Ca intake
Ser\lm albumin Albutensin A Peristaltic effect
Food proteins Oligopeptide mixtures Efficient a.a.*** intake
Wheat Gluten peptide Intestine activation
Many foods Oligosaccharides Microflora modulation

For the cellular system

Milk Lactoferrin Antibacterial effect
Milk Ganglioside Antibacteriotoxic effect
Egg Chicken cystatin Antiviral effect
Rice Oryzacystatin Antiviral effect
Rice ,,(-Oryzanol Antioxidative effect
Sesame Sesaminol Antioxidative effect
Ginger Gingerol Antioxidative effect
Buckwheat Lutin Antioxidative effect
Carrot Carotenoids Antioxidative effect
Orange Oreoresin Anticarcinogenic effect*
Wakame Fucosterols Anticarcinogenic effect*

* Speculative ** Casein phosphopeptide *** Amino acid

 51
Another method for enriching physiologically functional sub-
stances is to add them to conventional foods. The enrichment
can also be attained with the aid of enzymatic reactions and by
the use of gene technologies. On the other hand, a unique con-
cept is available for the design, which is based on removing
some unfavorably functional factors. The removal of food al-
lergens would offer a good example as discuused in the example
of producing hypoallergenic rice. Anyway, physiologically
functional foods can bedesigned either by maximizing favorable
functions (TABLE 1) or by minimizing unfavorable functions.

Semi-healthy state Disease A b-

Health Semi-healthy state Disease B J- .--

Semi-healthy state Disease C ~

i
Functional food A I I Medicine A ~

Foods
Functional food 8 I I Medicine 8
as
nutrient sources
~. Functional food C J l Medicine C J-
-- Non-specific -- -- Specific -- -- Specific --

To be taken To be taken To be applied

for health for disease prevention for remedy
day after day as occasion damands as occasion demands

Figure 1. Positioning of the physiologically functional food.

THE CASE OF RICE-BASED FOODS WITH THE RTHIRD" FUNCTION

1. Breeding an oryzacystatin-enriched rice cultivar
with an antiviral function -- a long-range study

Cystatin is a general term denoting a superfamily of proteins

which specifically inhibit the enzymes belonging to the cyst-
teine proteinase class (EC 3.4.22). In the field of medical
sciences, a variety of animal cystatins have long been ex-
tensively investigated for structure, function, and physiolog-
ical significance. Some cystatins occur in various tissues of
our body, probably acting as biodefense substances against
microbial and viral infection. Oryzacystatin we found (3) is
just the first well-defined cystatin of plant origin (4). We
have also disclosed the gene and protein structures of two
species of oryzacystatin by molecular cloning (5)(6). Both are
as small as trypsin inhibitor and are highly heat-stable.
Treated under cooking conditions, each does not lose its orig-
inal papain-inhibitory activity. It follows that we take
oryzacystatin in the active form by way of boiled rice in our
daily menus. Incidentally, this protein is contained in rice
52
seeds in a quantity of 0.1-0.2 mg/lOO g. Different from tryp-
sin inhibitor, oryzacystatin has no effect on our digestion
process because no cysteine proteinase exists as a digestive
enzyme in the human small-intestine.
In the seed of rice, oryzacystatin has target enzymes of
its own (7). It can also recognize some exogenous cysteine
proteinases as its target enzymes. These include the enzymes
of insect, microbial, and viral origins. Some human viruses
require cysteine proteinses for processing of their
polyproteins during the prolification in infected cells. This
suggests that the application of a cystatin could show an anti-
viral effect as it enters the cells. Recently, we observed
potent inhibitory effects of two species of oryzacystatin on
the replication of poliovirus in infected Vero cells (8). Pre-
sently, we are challenging to breed an oryzacystatin-enriched
cultivar of rice. Transgenic rice plants are already regener-
ated from protoplasts to which a plasmid containing an
oryzacystatin cDNA insert has been electroporated. We have
also confirmed that both oryzacystatin and its papain-inhibi-
tory activity are actually expressed in the seeds harvested.

2. producing hypoallergenic rice using an

enzymatic reaction -- a short-range study

While a long-range study is being carried out based on the con-

cept of maximizing a favorable factor (oryzacystatin), a short-
range study has been conducted to produce a physiologically
functional food from rice by minimizing its unfavorable factor,
allergenic protein (9). In Japan, the problem has arisen that
the number of patients with rice-associated allergy in the form
of atopiC dermatitis is unexpectedly increasing. Though the
reason has not been clarified yet, this actual state is indeed
a nuisance for those whose staple food is rice, since stopping
consuming such a traditional food may greatly affect their
accustomed dietary styles.
We have developed a unique process for producing hypo-
allergenic rice. The process is comprised of the first step
for decomposing the salt-soluble proteins (including a maj or
rice allergen) with a protease (Actinase) and of the second
step for drying the protease-treated rice grains by surface
parboiling. The product resulting from the first step
treatment was found to have retained no detectable amount of
the salt-soluble proteins, with glutelin (principal seed pro-
tein) remaining half or more. The use of the second step
treatment was effective in producing rice grains which resemble
normal rice grains in terms of physical and sensory qualities
(10) (11) • Clinical tests have been conducted in many
hospitals, with the result that the application of this hypo-
allergenic rice in place of ordinary rice is sometimes dramti-
cally effective in improving the skin inflammatory conditions
of more than 40 volunteered patients with rice grain-associated
atopic dermatitis (12).
This rice has been industrially produced and suppled as an
example of the physiologically functional food. It is and will
be of great benefit for the patients in Japan as well as pos-
sibly in any other rice-consuming countries of the world(13).
 53
REFERENCES

1. Reports of "Systematic Analysis and Development of Food

Functions" (1984-86 Grant-in-Aid for Special Research Pro-
ject, The Ministry of Education, Science and Culture of
Japan). Abstracts of Papers (in English), ed. M. Fujimaki,
Gakkai Shuppan Center, Tokyo, March 1988.
2. Reports of "Body-modulating Functions of Foods" (1988- 91
Grant-in-Aid for Scientific Research on Priority Areas, The
Ministry of Education, Science and Culture of Japan).
Abstracts of Papers (in English), ed. H. Chiba, Gakkai
Shuppan Center, Tokyo, February 1992.
3. Abe, K., Emori, Y., Kondo, H., Suzuki, K., and Arai, S.,
Molecular cloning of a cysteine proteinase inhibitor of
rice (oryzacystatin). J. BioI. Chem., 1987, 262, 16793-97.
4. Abe, K., Kondo, H., Watanabe, H., Emori, Y., Arai, S.,
Oryzacystatins as the first well-defined cystatins of plant
origin and their target proteinases in rice seeds. Biomed.
Biochim. Acta, 1991, 50, 637-41.
5. Abe, K., Emori, Y., Kondo, H., Arai, S., and Suzuki, K.,
The NH2-terminal 21 amino acid residues are not essential
for the papain-inhibitory activity of oryzacystatin, a mem-
ber of the cystatin superfamily. J. BioI. Chern., 1988,
263,7655-59.
6. Kondo, H., Abe, K., Nishimura, I., Watanabe, H., Emori, Y.,
and Arai, S., Two distinct cystatin species in rice seeds
with different specificities against cysteine proteinases.
J. Biol. Chem., 1990, 265, 15832-37.
7. Watanabe, H., Abe, K., Emori, Y., Hosoyama, H., and Arai,
S., Molecular cloning and gibberellin-induced expression of
mul tiple cysteine proteinases of rice seeds (oryzains) .
J. Biol. Chern., 1991, 266, 16897-902.
8. Kondo, H., I j ir i, S., Abe, K., Maeda, H., and Ara i , S.,
Inhibitory effect of oryzacystatins and a truncation mutant
on the replication of poliovirus in infected cells. FEBS
Lett., 1992, 299, 48-50.
9. Izumi, H., Adachi, T., Fujii, N., Matsuda, T., Nakamura,
R., Tanaka, K., Urisu, A., and Kurosawa, Y., Nucleotide
sequence of a cDNA clone encoding a major allergenic pro-
tein in rice seeds - homology of the deduced amino acid
sequence with members of amylase/trypsin inhibitor family.
FEBS Lett., 1992, 302, 213-16.
10. Watanabe, M., Miyakawa, J., Ikezawa, Z., Suzuki, Y., Hirao,
T., Yoshizawa, T., and Arai, S., Production of hypoallerg-
enic rice by enzymatic decomposition of constituent
proteins. J. Food Sci., 1990, 55, 781-83.
11. Watanabe, M., Yoshizawa, T., Miyakawa, J., Ikezawa, Z.,
Abe, K., Yanagisawa, T., and Arai, S., Quality improvement
and evaluation of hypoallergenic rice grains. J. Food
Sci., 1990, 55, 1105-07.
12. Ikezawa, Z., et al., Clinical effect of hypoallergenic rice
on atopic dermatitis (in Japanese). Allergy, 1991, 40,
633-42.
13. Watanabe, M., Development of hypoallergenic rice - award
lecture (in Japanese). J. ~ Soc. Nutr. Food Sci., 1992,
45, 479-84.
 ESTIMATION OF DOUBLE LAYERS FLOW MODEL
FOR FLUIDITY OF MILK IN A CAPILLARY
Teruko Nakamura, Akemi Yamamoto,
Kishiko Sakanishi and Takeshi Mineshita
Dept. of Food Science, Tezukayama College, Nara Japan

SUMMARY
Flow properties of human milk, cow's milk and homogenized milk were
studied by using various bore-size of low shear capillary viscometer.
Each milk shows characteristic shear dependence on the visccosity and
viscosity values changed with the change of capillary bore-sizes. From
these both shear dependence on viscosity and capillary bore-size depend-
ence on viscosity values, the change in the dispersion state of each milk
might be attributed to the formation change of the aggregate structure
of milk fat globules in a capillary. From these experimental results, flow
model was derived and assumed the wall layer composed of double layers
in the flowing liquid in a capillary tube.

INTRODUCTION
It is significant to investigate rheological properties of milk in a capillary
to explain the milk flow on tongue and through throat for the consumer
preference. Milk shows two significant characteristic flow properties in a
capillary, one is the nature of non-Newtonian flow behavior at low shear
range, and another one is the viscosity value decreased with decreasing of
capillary bore-size of the viscometer. In this paper, effects of various
bore-sizes of glass capillary tube on flow properties of milk were dealt
with a theoretical treatment and were elucidated with rheological results
obtained experimentally in a glass capillary tube.

EXPERIMENTAL
Samples used throughout this experiment were human fresh milk, a single
cow's fresh milk, without further pasturization and homogenized milk.
Viscosity measurements were carried out with various bore sizes of
capillary viscometer with continuous varying pressure head of shear
stress range from 0.2 to 30 dynes / cm2• To take a photograph of the
image of milk fat droplets flowing in a capillary, photomicroscope com-
bined with a capillary viscometer was used. The result of photomicro-
scopic image of flowing milk was analyzed by a densitometric method.
 55
RESULTS AND DISCUSSION
Experimental results were represented by the following the oretical treat-
ment. The physical function Q due to the number distribution of fat
globules in a capillary will change with each position, therefore, Q is
expressed as the function of x and y. Under the photomicroscopic
observation, Q* depends on the function of x, and relationship between
Q and Q* is expressed as follows;
Q"'(x) = f~R2-,,2 Q(x,y)dy 1
j -'/R2-,,2
Where R is the radius of a capillary tube. To obtain the average value
divided by numbers of K layers in x cross section of a tube, Eq.!
becomes Eq.2. (Xi+~ '" 2R
Q(r/<= j tJQ (x)dx, lJ=- 2
Xj-- K
In the case of total distribution numbers of fat globules dispersed in
2

whole of the cross section in a capillary is constant value of no, Eq.2

will be calculated, and distribution numbers of fat globules divided by
K layers are expressed as follows;

NdN= Yi+I - )'i

Xi
y.=-
J R
J
1- (X
--.!..
R
.)2 . Xj
+510- 1-
R
The value of N represents the total numbers of fat globules at the cross
section in a capillary tube, and it is expressed as follows;
N=no·]{·R2
The value of N is also expressed as the ratio of flux: and flow rate as
follows;
N = f/ iT
Where, f is the flux: and v is the flow rate of milk fat globule in a
capillary. To obtain the N value experimentally, total amounts of fat
globules at the cross section were accounted by a photomicroscopic mea-
surement. Photomicrogram of the image of fat globules was taken conti-
nuously' and was developed to account these fat globules. This developed
photomicrogram of the cross section was divided six thin layers and, the
fat globules in each layer were accounted and represented as N. The ratio
ofLJ.N / N means numbers of fat globules occupied by a certain area
against total numbers of them on an unit area at the cross section in a
capillary. From the experimental results of flow curves of each milk
and photomicrographs of flowing milk in a capillary, the characteristic
flow behavior of each milk will be represented as the change in aggre-
gation of milk fat globules in a capillary. To identify this difference
between the dispersion state of each milk in a capillary, the relationship
between distribution of particle numbers of fat globules (6N / N) and
relative position of each milk was calculated. From this theoretical
treatment, it is shown that the largest amount of fat globules in homo-
genized milk gathered at the center core of a capillary tube. On the
contrary, distribution curve of cow's fresh milk shows the smallest amount
56
of fat globules gathered at the near side of the center core and the
largest amount of fat globules forms aggregate structures at the near side
of the capillary wall. This fact means that the protein layer surrounding
with fat globules in homogenized milk might be cut off by the treatment
of homogenization, so, the repulsion force between the fat globule and
the capillary wall occured and, as a result, fat globule-globule interaction
increased and the flow became easier at the center of a capillary tube.
In the case of cow's fresh milk, fat globules are surrounding with thick
layer of protein molecules like casein and, as a result, an attractive force
occured between hydrophobic binding of milk surface and capillary wall.
In a case of human milk, some of the fat globules formed aggregate
structures at the both side of the center of a capillary tube and the
capillary wall. From the flow distribution profile of human milk in
various capillary bore sizes, the curve in the widest capillary bore size
shows the largest numbers of fat globules gathered at the center in a
tube. The flow distribution curve in the narrowest capillary bore size
shows the smallest numbers of fat globules flowing in the center of a
capillary tube but, shows the largest numbers of them gathered and
formed aggregate structures at the near side of the capillary wall. This
fact means that the fat globules of human milk in the widest capillary
bore size will not be influenced by the steric hindrance of the capillary
wall and will form large aggregate structures easier at the near side of
the center of the tube, but when these globules flow in the narrowest
capillary bore, they will be influenced by the steric hindrance of the
capillary wall, then, the aggregate structures would be formed easier at
the near side of the capillary wall. Thus, the flow properties of milk
will change with the formation change of aggregation of milk fat globu-
les in a capillary. As a result, non-Newtonian flow behavior and the
dependence of the capillary bore-size on viscosity of milk in a capillary
were caused by the formation change of aggregate structure of milk
fat globules in a capillary: Consequentry, a flow model of milk in a
capillary is suggested to compose from three layers, one is the layer of
the center core of a tube, and another one is the wall layer. However,
the wall layer must be estimated to compose of double layers (do and di)
in a flowing liquid. One is the nearest side of the wall and, another
one is a close to this wall, and this layer will be influenced by the
capillary wall. From this flow model of double layers, the thickness of
the wall layers will be calculated.
REFERENCE
l. T. Mineshita et aI, Biorheology 29, p.85 (1992).
2. T. Mineshita and A. Yamamoto, Proceedings of Int.
Congress on Food Hydrocolloids p.42 (1992).
3. T. Mineshita et aI, Biorheology 19, p.376 (1982).
4. A. Yamamoto et aI, Biorheology 20, p.623 (1983).
5. T. Mineshita et aI, J.Texture Studies 17, p.205 (1986).
 EFFECT OF MOISTURE CONTENT ON FLOW CHARACTERISTICS OF
SPI MELT AT AN ELEVATED TEMPERATURE

N. HAYASHI, I. HAYAKAWA, Y. FUJIO

Department of Food Science & Technology, Faculty of Agriculture, Kyushu University,
6-10-1, Hakozaki, Higashi-ku, Fukuoka 812 JAPAN

ABSTRACT

At 140 ·C, the moisture content of SPI melt (soy protein isolate with 20 to 71 %d.b. moisture)
was the most dominant factor that affected its flow properties wherein the Herschel-Bulkley
power-law model was discontinued due to the alteration of the physical state of water in
sample. Below the critical moisture content, SPI melt flowed as a non-Bingham plastic fluid,
otherwise, it was shifted to pseudoplastic fluid by the effect of free water in SPI melt. At this
time, the free water in SPI melt played the role as lubricant.

INTRODUCTION

The flow properties of protein melt at an elevated temperature is considered to be affected not
only by its moisture content but also by the physical state of water (1). In the present study, the
flow properties of SPI melt was analyzed and their relation with the water state in sample was
investigated.

MATERIAL AND METHODS

SPI was moisturized from 20 to 71 %d.b. by mixing with pulverized ice powder at - 20 ·C.
Using a newly developed completely air-tight capillary tube viscometer, molten flow behaviour
was investigated at 140 ·C which is enough as a flow starting temperature of the lowest
moisturized SPI (20 %d.b.). Regression analysis was done by adopting the Herschel-Bulkley
power-law model (HB model). The water state in SPI heat-treated at 140·C for 20 min under
15 MPa was analyzed by the methods of DSC and NMR.
58
RESULTS AND DISCUSSION

Shear rate and shear force were calculated from measured flow rate and pressure drop,
respectively, and plotted as shown in Fig. 1. Solid lines in Fig. 1 show the regression flow
curves analyzed using the HB model. This figure reveals satisfactory fitness of measured data
to HB model and the flow properties of the SPI melt were markedly dependent on its moisture
content. The regression coefficients in the HB model was discontinuously changed at ca. 40
%d.b. of moisture content and this fact suggested the difference in melt flow mechanisms at the
opposite sides of this moisture range (2).

10 1 ~--------------------------,

10-2~~~~~~~~~~~~~~

10 1 10 2 10 3 10 4 10 5
Y (sec- 1 )
Fig. 1 Flow curves for SPI melt with various moisture content at 140 ·e.
plot; measured data, solid line; regression curve
moisture content (%d.b.) : 0; 20, .; 30, 6; 41, .A.; 54, 0; 70

60

40

20

o 0~~--~~--~--~~
o 40 60 80 100
Total moisture content (%d.b.)

Fig. 2 Freezable water content of heat treated SPI plotted as a function of its total moisture
content.
 59
Figures 2 and 3 show the state of the water in SPI, heat treated at 140°C for 20 min under 15
MPa, analyzed by measurement of the freezable water content with DSC and by spin-spin
relaxation time (f2) measurements with pulsed NMR.

6 r-----------------------------~
5

o ~~~~~--~--~~--~~--~~
o 20 40 60 80 100
Total moisture content (%d.b.)
Fig. 3 Relationship between total moisture content and spin-spin relaxation time (T2) of
water proton at - 20°C (.), 0 °C (D) and 30°C (0) for heat treated SPI.

Close agreement on the critical moisture content of ca. 40 %d.b. were obtained between the
minimum freezable water content and the moisture content of T2 inflection point. This result
indicated that the physical state of water in heat treated SPI changed at this moisture content.
The fact that the moisture content, wherein the flow properties changed, was more or less
similar to that of the point where the physical state of water changed suggested that the water
state affects dominantly on the molten flow properties of SPI (3). By the effect of free water in
SPI melt, it was a pseudoplastic fluid above the critical moisture content, but otherwise it was a
non-Bingham plastic fluid. At this time, the free water in SPI melt played the role as lubricant
and caused the decrease in strain history effect as viscosity decreased (4).

References
1. Y Fujio, N. Hayashi & I. Hayakawa (1991) Effect of moisture content on flow behaviour of
molten soy-protein isolate under an elevated temperature. Internal. J. Food Sci. Techno!.,
26, 45-51.

2. N. Hayashi, I. Hayakawa & Y Fujio (1993) Flow behaviour of soy protein isolate melt with
low and intermediate moisture levels at an elevated temperature.J. Food Engin., 18, 1-11.

3. N. Hayashi, I. Hayakawa & Y Fujio (1992) Hydration of heat treated soy protein isolate
and its effect on the molten flow properties at an elevated temperature. Internat. J. Food Sci.
Technol., 27, 565-571.

4. N. Hayashi, I. Hayakawa & Y Fujio (1992) Influence of time-temperature history and strain
history on the melt rheology of soy protein isolate at an elevated temperature. Internat. J.
Food Sci. Technol., 27, 297-304.
 NOVEL APPROACH TO CHARACTERIZE THE FLOW PROPERTIES OF
FINE DRY FOOD POWDERS

H.O. KONO and Y. !TANI

Chemical Engineering Department, West Virginia University

P.O. Box 6101, Morgantown, WV 26506-6101, U.S.A.

ABSTRACT

A measurement method was developed by using the powder aeration technique to

measure the flow properties of fine powders in terms of fracture strength of "dry powder
structure" .

INTRODUCTION

The powder structures are generally formed in fine and dry powders, which include
the fine particle fraction smaller than 20,um, because of the relatively strong inter-particle
forces.
When the bulk powders that can form this powder structure are processed in flow,
mixing, transporting, weighing etc., the properties of powders significantly affect the
operational characteristics of the concerned process.
To measure the tensile strength of these packing structures, the concept of tensile
strength was introduced by Rumpf [1] and various measurement methods have been
developed, such as split-cell methods by Jimbo [2]. On the other hand, the shear stress
test was defined and developed by Jenike [3].
We found a new accurate measurement method using the deformation and fracture
of aerated fine powders under a specifically well defined condition. In this study, the
results between this method and the conventional split-cell method were compared.
 61
BASIC CONSIDERATION

When a gas flows through a bed of particles, the deformation of the fine powder
structure occurs between Umt", the minimum fluidization velocity, and ~nb' the minimum
bubbling velocity. When the gas velocity reaches Umf' the gravitational force is
compensated with the drag force as described in Ergun's equation.
As shown in Figure I (b), in accordance with the gas velocity, LlP, the pressure
drop through the column, in the range of Umf and U mb remains constant but the bed height
increases from Hmf to Hmb and the void age increases from Emf to Emb' On the other hand,
the fracture strength of the bed ({Tr) changes from O"f,mf to O"r,mb •
The tensile strength of the fine powder structure as defined in the previous section
(O"J can be expressed by Rumpf's equation.
Using the principle of virtual work (or power), the following equations can be
written.

Umb

ED = ~ JdP(uo) duo (1)

Umf

Umb

Eo J(]
Umf
f lUo ) duo (2 )

Here, ED and Eo represents the dissipation and cohesion powers respectively. N is

the number of powder layers in the bed.
The physical meaning of the principle of virtual work (or power) can be illustrated
in Figure I(c).

EXPERIMENTAL and RESULTS

As shown in Table 1, eight different sample powders were used in the experiments.
The tensile strength of relatively non-sticky powders (5,8) can be determined in the split-
cell by only compressing to a high pressure.
The fracture strength (0" f mb) and tensile strength (0"J of our sample powders were
experimentally measured and shown in Figure 2, using both the conventional split-cell
method and the newly developed fracture strength method of this paper.
For the measurements of sticky powder with O"t > 10 Pa, the split cell method is
a better method. But for the less sticky powders with O"t < 10 Pa, the above method can
not be used and only the fracture strength method can be useful. The starch sample with
CTr,mb = 10-2 Pa order could also be obtained.
62
E

E mb

E mf

..
U mb Uo Umf

(bl (e)
(0)

Figure 1. Schematic diagram of concept of the Fracture Strength

lE4y----------------------------------------~
~

--I ~
...... 1000 c.bon Black

(Jlmbo .1 all
( Con.....tional ".thod )

Spent FCC
<Thilo ....Ihod )

CGttoon Block
e . .+
...
CaCO;! Yidu_·

" • starch
S~nt FCC
• Claa Beada

lE-2+---------~--------~--------~--------~
o 500 1000 1500 2000
Bulk Density [kg/m.3]

Figure 2. Fracture and Tensile Strengths of various powders measured by the conventional and new method.

Table 1. Physical Properties of Powders

Powder __________dy __________ U rnf U rnb ar,m( 0
--.?.!--- ----- ----- --
!L m kg/m3 cm/s cm/s Pa

1 Spent FCC 65 (20<~< 125) 1850 0.61 0.9 0.19

5 Spent FCC + 5.0wt% CaC03 62 ( 2<~<125) 1890 0.71 1.8 0.50

6 Glass Beads 30 (10<~< 60) 2500 0.69 0.8 0.04

7 Starch 15 ( 5 <dp < 30) 1550 0.33 0.4 0.02

8 Carbon Black 30 (20<~< 50) 1800 0.37 1.9 0.13

9 Calcium Carbonate (CaC0 3) 3 (0.05<d p < 5) 2700 - - -

REFERENCES

1. Rumpf, R., Chern. lng. Tech., 42, 538 (1970)

2. Jimbo, G., R. Yamazaki, 1. Tsubaki and R.Kamiya, Memoirs of the Faculty of Engim
(Nagoya Univ.), 40 (1), 1 (1988)
3. Jenike, A.W., Bull. Univ. Utah, No.108, 52 (1961)
VISCOSITY MEASUREMENTS OF FRESH AND HEAT STRESSED FRYING
OIL AT HIGH TEMPERATURES

K.S. MILLER, B.E. FARKAS, AND R.P. SINGH

Department of Biological and Agricultural Engineering
University of California, Davis
Davis, CA 95616 USA

ABSTRACT

A capillary viscometer and a computer controlled, convective air bath were used to
examine the kinematic viscosity of four frying oils at various levels of degradation. Viscosity
was measured at 170, 180, and 190°C after 0, 12, 24, and 36 hours of frying. All oil samples
showed an increase in viscosity with increased time of frying. A decrease in viscosity was
seen with increasing temperature, within each degradation time. Free fatty acid (FFA) content
and total polymer content were tested as a means of measuring the levels of oil degradation.
FFA values were highly correlated with viscosity.

INTRODUCTION

Literature review
Extensive research has been conducted in the field of oil chemistry as related to oil degradation,
out of this work a number of tests for oil degradation level have emerged. A review of oil
chemistry has recently been presented [1] and covers the development of free fatty acids and
polar materials during frying, dielectric constant and other topics. Most recently [2,3] gel
strength has been studied with its affect on oil uptake and an oil uptake ratio, UR, has been
proposed. While there has been considerable research in oil chemistry, the literature on oil
viscosity, particularly at frying temperatures is notably devoid. In research [4] on lowering oil
content of fried foods, oil viscosities in the temperature range of 121-205°C were measured. In
the development of instrumentation, a sensor to record frying oil viscosity [5] by damping with
a piezocrystal was developed and ultrasonics was used to measure oil viscosity [6]. Oil density
and viscosity for the temperature range of 25-140°C have been reported [7].
The objectives of this research were to measure oil viscosity at frying temperatures,
study the effects of degradation level on oil viscosity, and determine the implications of
viscosity change on rate of heat transfer.
64
MATERIALS AND METHODS
Four types of oil (canola, com, palm, and soybean) were tested at four levels of degradation,
fresh, 12 hours, 24 hours, 36 hours and their viscosities were measured at 170°C, 180°C,
190°C. Oil degradation was induced by frying 200 g of Russet potatoes for 5 minutes at 190°C
at hourly intervals for a total of 36 hours. At the end of each 12 hour period viscosity
measurements were taken and the oil stored at 25°C for 12 hours. This was repeated three
times for a total time of 36 hours. Viscosity was determined using a size 75 Cannon
Ubbelhode (Cannon Instrument Company, State College, PA, USA) capillary viscometer.
Viscometer temperature was maintained with a custom built convective air heater which
complied with the ASTM 1992, D445-88 testing procedures. Oven temperature range was 25-
250°C ±O.1 °C. Free fatty acid analysis was determined using AOCS method Ca-5a-40 (1964)
and percent polymers were found by high performance size exclusion chromatography [8].

RESULTS AND DISCUSSION

Com oil showed the largest increase in viscosity caused by degradation followed by soybean,
palm and canola oils. The changes in viscosity suggested that com oil was the least heat stable
oil and canola the most heat stable oil. Results are summarized in Table 1.

TABLE 1
Kinematic viscosity (cSt) of canola, com, palm, and soybean oils measured at 170, 180, and
190°C and four levels of degradation.

o(hr) 12 (hr)
170°C 180°C 190°C 170°C 180°C 190°C
ICanolaoil 3.2030 2.8827 2.6064 3.2341 2.9649 2.7050
Com Oil 3.0881 2.7993 2.5697 3.4449 3.1217 2.8333
Palm Oil 3.1510 2.8800 2.6142 3.3248 3.0440 2.7240
Soybean Oil 3.0294 2.7558 2.5005 3.2952 3.0488 2.7459

It was found that %FFA and %Polymers increased with increasing degradation time.
When correlated with changes in viscosity r-squared values in the range of 0.958 - 0.999 were
found for %FFA and 0.745 - 0.997 for %Polymers. This indicates a strong correlation
between the chemical changes and physical changes produced during oil degradation.

An empirical correlation for Nusselt, Grashof, and Prandtl numbers was used to
determine the convective heat transfer coefficient (eqn. 1), physical properties of oil were
found in standard text. Sample results are shown in Table 2 for com oil, other oils gave
similar results.

Nu = {2+0.43(GrPr)O.25} [1]
 65
TABLE 2
Comparison of theoretical and measured heat transfer coefficient for corn oil at 170°C.
Degradation Theoretical h @ 170°C Measured h @ 170°C
Time (hr) (W/m2K) (W/m2K)

°12 262
256
265
259
24 251 251
36 245 248

CONCLUSIONS
1. It was found that the oil degradation time significantly effects oil viscosity.
2. Heat transfer coefficient may be predicted using empirical correlations with an
acceptable error.
3. The strong correlation between viscosity and %FFA suggests that viscosity could be
used as an indicator for oil degradation under these experimental conditions.

REFERENCES
1. Miller, K.S., Physical and Thermal Properties of Edible Ftyin~ Oils. Masters Thesis,
University of California, Davis, 1992.
2. Pinthus, E.J., Weinberg, P., and Saguy, I.S., Gel-strength in restructured potato
products affects oil uptake during deep-fat frying. Journal of Food Science. 1992, Vol
57, no. 6, 1359-1360.
3. Pinthus, E.J., Weinberg, P., and Saguy, I.S., Criterion for oil uptake during deep-fat
frying. Journal of Food Science, 1993, Vol 58, no. 1,204-205,222.
4. Shaw, R and Lukes, A.C., Reducing the oil content of potato chips by controlling
their temperature after frying. 1968, Agricultural Research Service report no. ARS-73-
58. U.S. Department of Agriculture, Washington D.C.
5. Kress-Rogers, E., Gillat, P.N., and Rossell, J.B. Development and evaluation of a
novel sensor for the in situ assessment of frying oil quality. Food Control, 1990,
July: 163-78.
6. Lacey, RL. and Payne, F.A., Ultrasonic properties of frying oil as a measure of
viscosity. Paper no. 916518, presented at 1991 International Winter Meeting of the
American Society of Agricultural Engineers, Chicago, IL, December 17-20, 1991.
7. Clements, L.D., Shafer, R, Noureddini, H., and Mammel, W., Density and viscosity
for vegetable oils and fatty acids. Paper no. 916522, presented at 1991 International
Winter Meeting of the American Society of Agricultural Engineers, Chicago, IL,
December 17-20,1991.
8. Christopoulou, C.N. and Perkins, E.G., High performance size exclusion
chromatography of fatty acids, mono-, di-, and triglyceride mixtures. J. of the Amer.
Oil Chern. Soc. 1986,63:679-84.
 VISCOSITY'" PROPERTIES BY' AMYLOGRAPID OF RICE

HUMIO(F) KURASAWA and IeHIRO SHOJI

Department. of food, Faculty of Home Economics,
Koriyama Women's University.
Koriyama-oity, Fukushima Prefecture, 963, Japan

ABSTRACT

There are superior and inferior varieties in non-glutinous oooked rice.

For the oohesiveness of oooked rioe, glutinous showed high, non-gluti-
nous superior showed middle and non-glutinous inferior showed low
values. Maximum visoosities of amylogram of rioe starches were in the
same order as desoribed above. But in rioe flours glutinous rioe show~
ed lower value than non-glutinous rice. It is due to the fact that
gelatinization of glutinous rioe starohes is affected definitely
8:reater by protein, fat and J..-amylase than non-glutinous rice starches.
By using amylograph, high qualities of non-glutinous rice bread, non-
glutinous rice noodle and non-glutinous rice cake were obtained.

INTRODUCTION, METHODS, RESULTS

PROPERTIES OF GLUTINOUS AND NON~GLUTINOUS COOKED RICES

There are superior varieties ( stickiness, color and gloss, flavor are
high value ) and inferior varieties ( stickiness, oolor and gloss,
flavor are low value ) in non-glutinous cooked rioe produced in Japan.
And it was possible to estimate those varieties by Midometer.
The former showed high value and the latter showed low value in the
maximum viscosity by Brabender- amylograph of rioe flours. Rice
sta.rches prepared from both varieties showed similar result. For the
cohesiveness of oooked rioe by the mOdified-balanoe method, glutinous
rice was 70g, non-glutinous superior rice was 49g and non-glutinous
inferior rioe was l3g (1). The result is shown in Table 1.

AMYLOGRAMS OF RI CE FLOUR AND RI CE STARCH

The maximum viscosities by amylograph of starches prepared from
glutinous, non-glutinous superior and non-glutinous inferior rices were
650, 550 and 450BU. The maximum VJiscosities of rioe starches were in
the same order- as desoribed above. The maximum viscosities by amylo-
graph of rice flours were 130, 470 and 370BU. The maximum viscosities
by amylograph of non-glutinous rice flour were in the same order as
 67
TABLE 1
Eating and cooking qualities of glutinous and non-glutinous rices

Rice Palatability of cooked rice Cohesiveness of Amylose

stickiness color flavor cooked rice(g) oontent(t1l)

Glutinous rice
(Koganemochi) 70 0
Non-glutinous R.
Superior(Honenwase) 61 60 59 49 18.7
Inferior(Okumasari) 44 44 44 13 22.1

Correlation coeff. between palatability and Midometer(tester)=0.996

described above. But in rice flours, glutinous rioe showed lower

value than superior and inferior non-glutinous rices. The break down
values of rice starohes were 300, 260, 35BU and the those values of
rice flours were 50, 220, 160BU (2). The result is shown in Table 2.

TABLg 2
Amylogram viscosities of rioe starch and rice flour (rm)

Rioe Maximum Break down Set baok

visoosity (Disintegration) (Retrogradation)

Rice starch
Gluti. R. (Koganemoohi) 650 300 -200
Non-glutinous rioe
Superior (Honenwase) 550 260 0
Inferior (Okumasari) 450 35 60
Rice flour
Gluti. R. (Koganemochi) 130 50 10
Non-glutinous rice
Superior (Sasanishiki) 470 220 60
Inferior (Toyonishiki) 370 160 150

AMYLOGRAMS OF RICE FLOURS 0131'/\.INED BY l'RE\TING WITH E'rH~R, ALKALI AND

CuS04
fhe maximum viscosities of glutinous and non-glutinous rioe flours
obtained by treating with ether were 190, 390BU. The those of the
both rioe flours ~(starohes) obtained by treating with dilute alkali
were 990, 980BU. The those of both rioe flours obtained by treating
with 1% CuSO (inhibitor of enzyme) were 570, 450BU. As for the lower
value of the 4maximum visoosity by amylography of glutinous rice flours,
it is due to the fact that the gelatinization of glutinous rice starch
is affeoted definitely greater by protein, fat· and J..-amyl3.se than non-
glutinous rioe starch is. The result is shown in Table 3.
d.."7AMYLASE ACTION OF GLUTINOUS AND NON-GLUTINOUS RICES
68
TABLE 3
Amylogram visoosities of rioe flours obtained by treating dil.
alkali, ether and l~ CUS04 (BU)

Rioe Rioe Dil. alkali Ether 1~ CuSOA (Inhibi-

flor treatment; treatment tor of enzyme)

Glutinous rioe
( Koganemoohi ) 130 990 190 570
Non-glutinous rioe
( Sasanishiki ) 470 980 390 450

Enzyme aotiv! ty of glutinous rioe J.-amylase to glutinous rioe starch

was loll. E:nzyme activity of non. . .glutinous rioe ~-amylase to non-
glutinous rioe was 0.19. For' the cl-amylase action, glutinous riCl'8
has more J.-amylase activity, and the starch is more deoomposed than
non-glutinous rice. So it is due to the fact that; the starch is
deoomposed during,the measurement and the visoosi ty or'st'arah solution
deoreases. In the case of glutinous rice, it. is necessary for
measurement to add the enzyme-inhibitor ( 1% CuS04 ) in the measure-
ment solution. The result. is shown in' 'rable 4.

TABLE 4
Enzyme activities of Gluti. R. enzyme to Gluti. R. starch and non-
Gluti. R. enzyme to non-Gluti. R. starch (620 nm)

Rice starch Gluti. R. J. -amylase lfon-Gl uti. R. d.-amylase

Native rice starch 1.11 0.19
Gelatinized rioe staroh 1.31 0.28

Gluti. R. has more ~-amylase activity, its staroh is more deoomposed.

MANUFACTORES OF NON-GLUTI. R. NOODLE, BREAD, CAKE,AND AMYLOGRAM

In the oase of rioe noodle, the mixture of rioe flour' size 250 mesh
( the maximum visoosi ty of amylogram 660BU ) 9 parts and rioe flour
size 80 mesh ( the maximum viscosity of amylogram 5:00BU ) 1 part;
showed high quality. In the case of rice bread, the compound of rioe
flour size 250 - 200 mesh ( the maximum visoosity of amylogram 660 -
630BU ) 65~, active gluten 2~, wheat staroh 15~ showed high quality.
In the oase ofaFioe cake, the cake ( the dough l7'OBU ) obt:ained by'
heating at 110 cr: showed higher value than the oake ( the dough 280BU )
obtained by heating at 99°C (3).

REFERli.:NCillS
1. Kurasawa, H., Eating quality of non-glutinous oooked rioe. Science
of Cookery, 1979, 12, 128-37.
2. Shoji, I. and Kurasawa, H., Amylogram properties between GlutL R.,
non-Gluti. R. and the starches. J. Home Econ. Japan, 1979, 30, 292-4.
3. ShOji, I., Kurasawa, H.t. The relation between conditions inproducing
and the qual! ty of non-lie R. cake. J. Home Econ. Jpn. 1979, 39, 666-71.
 PHYSICOCHEMICAL PROPERTIES AND HYPOLIPIDEMIC
FUNCTION OF LEVAN AND ITS PARTIAL HYDROLYSATE

MASAO TAKAHASHI, RIE TAKAYAMA, and KAZUOKI ISHIHARA

Institute for Intestinal and Environmental Microbiology
Advance Co.Ltd., 1-35-2, Shimoishihara, Chofu, Tokyo, Japan

ABSTRACT
Levan of Str.saLivarius and partially hydrolized levan( PHL ) suppressed
the elevation of serum lipid levels in rats and hamsters. Viscosity of PHL
having molecular weight( M ) less than 10· was low, and intrinsic viscosity
[1/] of PHL was well-approximated by the equation [1/ ]=KW. In hydrolysis
of levan, the rate could be calculated by increase of reducing sugar and the
activation energy was almost constant at pH 3.0 to 5.0 and the frequency
factor was pH dependent.

INTRODUCTION
Many studies on hydrolysis of levan at low temperature and the hydrolized
products of which molecular weight( M.W.) was more than 10 5 have been
reported[l] , [2]. However, few papers deal with physiological significance
of levan of Str.salivarius. In this study, hydrolysis of levan at high
temperature was analyzed and physicochemical properties and hypolipidemic
activity of partially hydrolized levan( PHL ) were examined in view point
of application of levan and PHL to food stuff.
MATERIALS AND METHODS
Preparation of levan and PHL
Levansucrase from a culture medium of Str.salivarius AD10002 was added into
O.OlM phosphate buffer( pH6.0 ) containing 5% sucrose and incubated at 37~.
Levan produced in the reaction mixture was concentrated and purified by
ultrafiltration. The purity of the levan was more than 99%. The levan was
hydrolyzed in water acidified to pH 3.0±0.1, 4.0±0.1 and 5.0±0.1 with
O.lN HCI at 70, 80, 90 and 98~. The hydrolysate was freeze-dried.
Fractionated PHL( FPHL ) with narrow M.W. range was obtained by gel
filtration and ethanol precipitation.
Assay for reducing sugar and fructose
Reducing sugar was measured with 3,5-dinitrosalicylic acid. Fructose was
measured with F-kit( Boehringer Mannheim Yamanouchi, Japan ).
70
Measurement of molecular weight and viscosity
Number-average molecular weight( Mn ) and weight-average molecular weight
( Mw ) were measured by HPLC( column:Asahipac GFA-30F, Asahi Chemical
Industry Co.Ltd., Japan) and GPC PROGRAM( Shimazu, Japan ). Viscosity was
measured at 25~ in aqueous solution by Cannon-Fenske viscometer[3].
Animal experiments
Ten weeks old male F-344 rats and seven weeks old Sirian male hamsters were
used. Serum triglyceride( TG ) and total cholesterol ( TC ) were measured
with enzyme kit( Kyowa Medex Co.Ltd., Japan ).
(1) The rats were given a 20%-fructose diet added 1.5% levan, PHL( Mn =1000,
Mw =7000±3000, containing about 7% fructose) or FPHL( Mn =830, M>1220
) for 4 weeks. Control rats were given the 20%-fructose diet.
(2) The hamsters were given a 1%-cholesterol diet added 3% levan or 2.5%
PHL for 4 weeks. Control hamsters were given the l%-cholesterol diet.
RESULTS AND DISCUSSION
Hypolipidemic effect of levan and PHL
Serum TG and TC levels in experimental animals given levan, PHL or FPHL were
significantly lower than control group( Table 1 ).

Table 1
Effect of levan and PHL on serum TG and TC in experimental animals
TG (mg/dl) in rats TC (mg/dl) in hamsters
Groups
( n=5 ) levan PHL FPHL levan PHL
Control 481±11 412 ±37 351±10 350±22 307 ±10
Administered 320±35*** 307 ±18** 274±31** 298±11 * 278± 8*
Val ues are means ±S. E.. ***p<O.OI, ** p<0.025, *p<0.05

Viscosity of levan and PHL

Viscosity of levan and PHL was shown in Table 2. The intrinsic viscosity[ n]
of FPHL depended on the M.W.( Mw/Mn=1.1~1.3 ) as shown in Table 3. These
data were well-approximated by the equation [n ]=KM!, when the values of K
and a shown in Table 3 were used. The measured intrinsic viscosi ty[ n] of
PHL and the value calculated by using the equation and molecular weight
distribution of PHL agreed well( measured value: 0.065±0.008( dl/g ),
calculated: 0.063±0.003(dl/g) ).

Table 2
Viscosity( cP of levan and PHL in aqueous solution
Conc. ( g/dl 1.0 2.0 5.0 6.0 6.7 10
Levan 1.1 1.6 4.6 7.6 11. 7
PHL 1.0 1.1 1.2 2.0
 71
Table 3
Intrinsic viscositY[T/]( dl/g) and Mw of fractionated PHL
Mw of FPHL
K a
910 2070 4390 8720

[ 71 1t 0.035 0.048 0.064 0.080 2.7xl0-3 0.38

[ 71 12 0.035 0.050 0.070 0.090 1.6xl0- 3 0.45

1 71 reI -1= T/ reI [ 71 ]C, 2T/sp=[1l]C

Hydrolysis of levan
The ratio of reducing sugar produced to initial levant R.S. (%) ) was
proportional to hydrolizing time and reciprocal of Mn( Table 4 ). Decrease
of hydrolysate having M.W. more than 10· was rapid and the products in range
of lower M.W. increased in hydrolysis process of levan. The temperature
dependence of hydrolysis rate( reducing sugar produced(mg)/initial levan(mg)
/sec ) was of the Arrhenius type. The activation energy at pH 3.0 to 5.0
was 25 to 26 kcal/mol and the frequency factor at pH 3.0, 4.0 and 5.0 was
2.7xl011, 7.6xl0 IO , 2.7xlO IO ( sec-I), respectively.
These findings show that it is possible to obtain PHL with low fructose
content and low viscosity by control of hydrolizing time, pH and temperature.

Table 4
Physicochemical parameter in hydrolysis process of levan ( pH 4.0, 98"(; )
Hydro. Distribution(%) of M.W.
time Mn R. S. (%) Fru. (%)
min ) 20> 20--100 100--400 >400(x10 2)
20 2080 9.8 2.8 22 38 36 7.9
40 910 20.8 7.0 49 45 8 0.1
60 590 29.6 13.7 70 30 1.6 0
80 440 39.4 24.0 84 17 0.1> 0
100 350 47.7 30.5 92 9 0 0

REFERENCES
1. STIVALA.S.S. and ZWEIG.J.E., Physicochemical Parameters of Partially
Hydrolyzed S.salivarius Levan Fractions.
Biopolymers, 1981, 20, 605-609.
2. LAUREN. M.D. , STIVALA.S.S., BAHARY.W.S. and LONG.L.W., Levans. \T.
Kinetics of the Acid Hydrolysis of StrePtococcus salivarius Levan.
Biopolymers, 1975, 14, 2373-2385.
3. T.TANGLERTPAIBUL(nee SORNSRIVICHAI) and M.A.RAO, Intrinsic Viscosity
of Tomato Serum as Affected by Methods of Determination and Methods
of Processing Concentrates, Journal of Food Science, 1987, 52,
1642-1645
 THE USE OF XYLANOL YTIC ENZYMES IN PROCESSING OF CEREALS.

K.POUTANEN,H.HARKONEN,T.PARKKONEN,K.AU110,T.SUORTTI,
A. KANTELINEN*, L. VIIKARI*.
VTT Food Research Laboratory, P.O.Box 203, 02151 Espoo, Finland,
*VTT Biotechnical Laboratory, P.O.Box 202, 02151 Espoo, Finland

ABSTRACT

Treatment of whole meal rye flour slurries with pure endoxylanase caused
fragmentation of the cell walls and increased extraction of the pentosans. Due to the
heterogenous structure of rye cell walls also B-glucanase and protease were needed for
their complete degradation. About 30 % of the rye flour pentosans remained insoluble
even after treatment with a high dosage of Trichoderma reesei culture filtrate rich
with xylanolytic enzymes and B-glucanases. Pure xylanases effectively reduced the
viscosity of polymeric rye flour water-soluble polysaccharides.

INTRODUCTION

Xylans are heteropolysaccharides which constitute an essential cell wall component of

many cereal grains, and especially in rye and wheat they may contribute considerably
to the water-binding and viscous properties during processing. The importance of
these pentosans in baking was recently reviewed by 0' Appolonia and Schwarz (1).
The untinutritive effect of xylans in cereal animal feed has been clearly demonstrated
(2). In cereal arabinoxylans the backbone chain of l,4-linked B-D-xylopyranose units
is substituted by L-arabinofuranoside units bound by a-1,2- and a-1,3-linkages, and
part of the arabinosyl units are esterified with feruloyl or coumaroyl acids. Xylan-
degrading enzymes can accordingly be used as processing aids ~ in the baking and
feed industries.

Both depolymerizing and substituent-cleaving enzymes are needed for the

complete hydrolysis of heteroxylans (3). Endo-1,4-B-xylanases (E.C.3.2.1.8) and B-
xylosidases (E.C.3.2.1.37) act synergistically with a-arabinosidase, a-glucuronidase
and acetyl and feruloyl esterases for the degradation of xylans. The substituent-
cleaving enzymes could also affect the solubilization of xylans from the cell walls, or
the oxidative gelation suggested to happen via free feruloyl groups. We have
previously isolated and characterized many of the xylanolytic enzymes of Trichoderma
reesei, including both endoxylanases and different substituent-cleaving enzymes (4-6).
 73
In this paper the action of T. reesei endoxylanases on rye arabinoxylans is described.

MATERIALS AND METHODS

Whole meal rye flour from the Finnish cultivar Voima was produced by milling with a
Universal Laboratory Mill to pass a 0.8 mm screen. In addition to the purified major
endoxylanases of T. reesei (5) a culture filtrate of Trichoderma reesei was used. The
extraction experiments were carried out at 30°C for 90 min using a flour:water ratio 1:4
(w/w). The enzymatic depolymerization of the water extracts of whole meal rye flour
was monitored by GPC-HPLC methods using pHydrogel columns (Millipore Waters),
and 50 mM NaoH or water as eluent. Refractive index detection or on-line Calcofluor
staining and detection at 415 and 445 nm were used. Changes in viscosity were detected
by Bohlin viscometer (88BV). The effect of enzymes on rye flour microstructure was
studied using fluorescence microscopy with Calcofluor and Acid Fuchsin staining (7).

RESULTS AND DISCUSSION

Without enzyme addition about 30% of rye pentosans were extracted as high molecular
weight polymers during water extraction. Xylanase addition doubled the amount of
solubilized pentosan, which also was hydrolyzed to smaller molecular weight oligo-
saccharides. The purified xylanases were almost as effective in solubilization of pentosan
as the crude culture ftltrate containing also a.-arabinosidase and other hydrolases. About
30% of the rye pentosans remained insoluble irrespective of the high xylanase level used
(Fig. 1).

1 2 0 r - - - -- - - - - - - - - - - ,

r777"7'77;>7777TFJ ' •• • •••••••••

o 20 60
11mB, mIn

Figure 1. The amount of pentosan Figure 2. The effect of xylanase

extracted from rye flour with a) no
enzyme b) T. reesei pI 9 xylanase, c)
T. reesei pI 5,5 xylanase, d) T.reesei
culture filtrate. Xylanase dosage 2500
nkat/g flour.
treatment on the viscosity of rye
water extracts. Xylanase dosage
500 nkat/g d.m.
74
In rye cell walls xylan is in matrix with B-glucan and protein. Fluorescence
microscopic examination indicated that the cell walls were clearly degraded by the
addition of xylanase, but some cell walls especially of aleurone cells were still visible.
The addition of protease or the use of the whole culture filtrate caused remarkably more
pronounced degradation of the cell walls, and disappearence of the blue fluorescence
(results not shown).

The rye water extracts obtained without added enzymes contained much more
pentosan than B-glucan, and arabinoxylans also contributed more than B-glucan to the
viscosity of rye extracts. Rapid viscosity reduction of rye water extracts by the isolated
T.reesei endoxylanases was observed (Fig. 2), although the complete enzymatic
degradation of arabinoxylan was hindered by the high degree of arabinosyl substitution.

Earlier studies showed that xylanase addition enhanced the fragmentation of cell
walls during baking (7). The water-binding capacity of flours decreased and the dough
proofmg time reduced markedly. The increased availability of water increased starch
swelling during heating of the xylanase-treated doughs, and caused release of proteins
from aleurone cells.

REFERENCES

1. D'Appolonia, B.L. and Schwartz, P.B., Importance of cereal non-starchy poly-saccha-

rides in end-products. In: Cereal Chemistry and Technology: a long past and a bright
future, ed' P. Feillet. 9th International Cereal and Bread Congress, June 1-5, 1992. Paris.

2. Petterson, D. and Aman, P., Effects of enzyme supplementation of diets based on

wheat, rye or triticale on their productive value for broiler chickens. Anim. Feed Sci.
Technol., 1988,20, 313 - 324.

3. Poutanen, K., Tenkanen, Korte, H. and PuIs, J., Accessory enzymes in the hydrolysis
of xylans. In Enzymes in Biomass Conversion. eds. G. Leatham and M. Himmel, ACS
Symp. Ser. 460, WaShington 1991, pp. 426 - 436.

4. Poutanen, K. and PuIs, J., The xylanolytic enzyme system of Trichoderma reesei. In
Plant Cell Wall Polymers. Biogenesis and Biodegradation. eds. N.G.Lewis and M.G.
Paice, ACS Symp. Ser. 399, Washington 1989, pp. 630-640.

5. Tenkanen, M., PuIs, J., Poutanen, K., Two major xylanases of Trichoderma reesei.
Enzyme Microb. Technol.. 1992, 14: 566 - 574.

6. Tenkanen, P., Schuseil, J. PuIs, J. and Poutanen, K., Production, purification and
characterization of an esterase liberating phenolic acids from lignocellulosics.
J. Biotechnol., 1991, 18: 69 - 84.

7. Autio, K., Harkonen, H., Aman, P., Parkkonen, T., Frigilrd, T., Siika-aho, M. and
Poutanen, K., Effects of purified xylanase and B-glucanase on structural and baking
characteristcs of rye doughs. Food Structure. 1993, submitted.
 VISCOELASTICITY OF BUTTER

HIROMICHI HAYASHI
Department of Food Science, College of Bioindustry,
Tokyo University of Agriculture,
196 Yasaka Abashiri, Hokkaido, Japan

ABSTRACT
The relationship between temperature and viscosity was expressed by an
equation of linear reciprocal with the accuracy of the coefficient of
correlation r=0.98 as follows:
7J = 1/ (-7.527 x 10- 3 + 664.4 x 10 - 6 t)
where 7J: viscosity of butter (Pa's), t : temperature of butter ("C)
Tano (loss tangent) decreased so steeply with decreasing temperature in the
vicinity of 14.9 "C that a transformation of butter probably occurred from
viscous body to elastic one. From the above points it is at least necessary
for butter to be held at higher temperature than this temperature in order
that the improvement of its spreadability be expected.
INTRODUCTION
Fat of butter consists of many kinds of fatty acid glyceride, 97-98% of which
is triglycerides. In general butter is understood as an plastic material,
exhibiting instantaneous elasticity, on one hand, at lower temperature, while
viscosity, on the other hand, above a certain temperature.!) Viscoelasticity
of butter depends upon the fractional crystallization (ratio of solid fat to
liquid fat), size of crystallization fat and molecular arrangement of
triglycerides.
Regarding how to measure the physical properties of butter there are
penetration method, cutting method and compression method etc.,however, they
yield only relative values. Accordingly, using Hakke's Rotovisco Viscometer,
the viscosity and viscoelasticity of butter in relation to temperature are
to be expressed by the absolute values. On this basis the suitability for
processing and handling easiness for cooking are discussed in this paper.
MATERIALS AND METHOD
2.1 Preparation of samples: Commercial butter manufactured one month ago (225
g, December product) is used; after being kept at about 10°C in refrigerator,
the butter is formed into a cylinder about 45 mm in diameter and about 1mm
thick.
76
2.2 Measurement of viscosity and viscoelasticity: Hakke Rotovisco RV 20 type
Viscometer is used. Using the PO 45 sensor, the viscosity is measured at the
shear rate of 0.77 sec- 1 , while the viscoelasticity is measured at the
frequency of 1 hertz, and the angular velocity of 6.3 sec- 1 • The sample
temperature ranges from 11.89C to 26 9C adjusted by Hakke F3 thermostat.
Measurement and calculation are carried out by IBM PS 552 computer and the
result is recorded by means of Epson printer.
RESULTS AND DISCUSSION
3.1 Viscosity: Shear stress and viscosity are measured by raISIng temperature
of butter in the range from 13.69C up to 26.09C or by lowering temperature
both continuously.
The viscosity decreases with increasing temperature, i.e. from 862.8 Pa·s
down to 112.9 Pa·s. Some typical equations for viscosity are examined to
fulfil those numerical relationship between temperature and viscosity, and as
a result the coefficient of correlation, r, is given as in Table 1 from which
a linear reciprocal equation turns out to be a best fit.
7J = 1/(-7.527x 10- 3 +664.4x 10- 6 t)
where 7J :viscositY(Pa·s), t:temperature(9C)
The above relationship is again depicted in Fig.1, from which the viscosity
of butter within the temperature range decreases rapidly with increasing
temperature. Davis pointed out that an appropriate temperature at which one
readily spreads butter on bread was 16-179C and the viscosity was 3x10 6
Poise(3,000 kPa·s). On the other hand, the present data of viscosity are
344.5 Pa·s at 169C and 278 Pa's at 179C, which are only of the order of
1/10 000 times as much the above value. Viscosity of butter depends usually,
to a considerable extent, on the working method, setting method, summer or
winter product etc. A remarkable discrepancy between the present viscosity
data and those of Davis 1 ) might be due to the use of different types of
viscometer.

Table 1 Fitness of related equation for viscosity

Name of related equation equation r

Ostwald r=aD7) O. 94
Herschel-Bulkley r=ro+aD7) O. 84

Casson O. 83
Exponential function 7) = a expbl O. 89
1 O. 98
Linear reciprocal 7) = a+b.t

r : Shear stress(Pa) a, b: Viscosity Index

D Shear rate (S-l)
7) : Viscosity CPa •• )
 77
3.2 Viscoelaticity: Looking at G' (storage modulus) - temperature relation in
Fig 2, G' increases gradually with decreasing temperature from 28"C down to
16°e, but it increases rapidly from 16"C to 14"C, and at 14.6"C, G' is
greater than G" (loss modulus). G" is always greater than G' and indicated
similar behavior to G' in connection with temperature, but both curves
intersect at 14.6"C, below which G" decreases rapidly. Elastic modulus at
14.6"C is about 900 Pa. Tano varies with temperature in accordance with such
an interrelationship between G' and G" as above. Namely, remains constant in
the range of 28"C to 15"C it decreases ,rapidly with lowering temperature.
From the above two points one may consider;
1) The spreadability of globular fat of butter decreases below 15"C.
2) Energy consumption is lowered in an irreversible manner, so that a change
is taking place from viscous body to elastic one. Butter exhibits high
elasticity at lower temperature. This arises from the decrease of loss
modulus and the increase of storage modulus through a relaxation of the
viscous elements at a stress application. From this meaning butter is
understood to transform from viscous to elastic in the vicinity of 15"C. This
behavior roughly agrees with the exothermic peak temperature at the melting
point of butter fat, which is measured with DSC( Differential Scanning
Calorimeter) .
The transformation mentioned above is therefore assumed to be due to the
phase change of crystalline fat affecting markedly the structure of butter.
Davis 1 ) reported of the elasticity modulus of butter to be around 10 dyne/cm 2
( 100 kPa ) at 16 - 17"C, which is roughly 500 times as much as compared with
the present data of G" ranging from 296 Pa at 16.1"C to 215 Pa at 17 .. 4"C.
(Pal
1000 10000
900
800 measured line o G· (pal
700 • G' (Pa)
1000 <>O<>:lo<>.

/ ~---...
600 X tanlo)

.
,;
500

~
-...
"- 100 :.; 100
~
~

~.
~
300 linear reciprocal line ~
15
~
0 10
~
..._x-~x ",,-x~ A
.,,"***'x~
'"0 200 fY;><-'I;-'IIi'<
~
>- X
1

.-x
y(

>;(-?
100
10 15 20 30 18 22 26 30
14
('e)

Temperature of butter ("C) Temperature of butter

Fig. I Relation between temperature of ~lg. 2 Relation between temperature of

butter and viscosi ty butter and viscoelasticity

LITERATURE CITED
1) Davis,J.G. ,The rehology of cheese, butter and other milk products, J.Dairy
Resarch, 1937,8, 25
2) Kawanari,M. et al.,J. of the Japanese Society for Food Science and
Technology, 1992, vo1.39,pp-:-227- - --
3) Timmes,R.E., Australian ~ of Dairy Technology, 1980, June, 47-53
4) Ruegg,V.M. ,Moor U und Blanc B, Milchwissenschaft, 1983, 10, 601-605
 VISCOELASTIC CIUffiACTERlSTICS OF BREAD CRUMB AND
GRAPIIIC ANALYSIS OF COMPRESSION CREEP CURVE

YIPING WANG
Toyohashi Works, Shinko Electric Co.,Ltd.,
150 Motoyashiki,Sanya-cho,Toyohashi 441-31,Japan
HIROSHI MORISHIMA, YASUHISA SEO, YASUYUKI SAGARA
Department of Agricultural Engineering,
Faculty of Agriculture,The University of Tokyo,
1-1-1 Yayoi,Bunkyo-ku,Tokyo 113,Japan
KENJI IMOU
Department of Agricultural Environmental Engineering,
Faculty of Agriculture,Utsunomiya University,
350 Mine-machi,Utsunomiya 321,Japan

ABSTRACT

An optimum loading condition for the creep test of bread crumb has been
established experimentally, and a four-element rheological model has
appeared to be appropriate for describing the viscoelastic behavior
of crumb. Both creep and recovery of crumb were analyzed by a simple
graphic method. All of the elastic moduli and viscosity coefficients
have been found to increase with increasing storage time indicating
marked change within 12 hours after baking and with a decrease in stor-
age temperature. They were also affected by various baking conditions
such as molding method, baking temperature and baking loss.

INTRODUCTION
The viscoelastic characteristics of solid foods are of great importance
to textural evaluation and structural analysis. Most of solid foods are
considered as composite uniform viscoelastomers and their viscoelastic
behaviors have been discussed. However, little work has been conducted
on the viscoelasticity of bread crumb from the viewpoint of rheology!1].
This paper describes an optimum loading condition in creep test and
characterizes the viscoelastic behavior of crumb with a 4-element rheo-
logical model by a simple graphic method. The effects of storage time
and temperature on the viscoelasticity of crumb will also be reported.
 79
MATERIALS AND METHODS

White breads of 230 G in weight were baked at 230 0 C by the sponge-dough

procedure. After baking, loaves were wrapped in polyethylene bags and
then stored at 20 0 C for at most 7 days. Viscoelastic characteristics
were measured with a developed creep test system by using 4cm cubic
crumbs sampled from 16x8x8 cm loaves.

RESULTS AND DISCUSSION

In order to perform creep measurement successfully, an optimum loading

condition for bread crumb was investigated experimentally. It was sug-
gested that a constant stress from 300 to 2500 Pa was to be applied and
kept for 100 s in creep test. The 5 g initial load was used to avoid
incomplete compression and lessen the effect of unevenness of specimen
caused in preparing procedure.
Figure 1 shows a typical compressive creep curve of crumb exhibit-
ing the ranges of instantaneous elasticity(ab), retarded elasticity(bc)
and viscous flow(cd). The instantaneous deformation occupied 64% of
total deformation and the viscous flow had 12 % of that.
It was found that the creep behavior could be represented by the
4-element Burgers model. Creep strain f can be calculated as follows.
u0 u0 (J u0
f = - (l-exp(- - ) ) + - ( J
+ - (1)
E0 E1 r kv 77 2
here u 0 means constant stress, (J Is time, E0 and 8 1 are elastic modu-
lus, 771 and 772 .are viscosity coefficient. rkv is defined by 771/Fl '
To determine' the coefficients, the method of successive residuals
and the simple graphic method(2J were applied and the latter was found
to be convenient ahd useful indicating the relative errors within 3 % in
creep(Figure.l) and recovery curve for curve fitting.

0.125
Thickness 40mm
Age; 12h
Constant load 50g
0.100 Initial load 5g
Calculated d. __
:2....-r--~=='-"1
r::: 0,075
'j;
b Observed
en
0.050 b Eo;5.9xI03 Pa
EI
EI;1.4XIO' Pa
'1 1 ;8.7xIO' Pa·s
0.025 '11; 3.0 X 10 8 Pa·s
r\.; 6.0s
a
0
0 60 80 100 J 20 140
Time (s)

Figure 1. A typical compressive creep curve of crumb and curve fitting

80
......,. !j
til
p. Instantaneous elastic modulus Eo
~4
,-,.

~3

~2
:p0
. \{J 1
r-i
J1l

0 24 48 72 96 120 144 168 192

storage time (h)
Figure 2. Change in the elastic modulus E0 of crumb during storage

The instantaneous elastic modulus E0 and other viscoelastic coeffi-

cients were pr6~ortional to storage time(Fig. 2), indicating marked
change within 12 hours after baking. They increased with the decrease
of storage temperature ranging from 5 to 20 oC. They were also affected
by molding methods of dough, baking temperature and baking loss.
The data showed that the values of E0 ranged from 3.4x10 3 to
2.6x10 4 Pa, E1 from 0. 8x10 4 to 8. 5x10 4 Pa, 771 from 4. 0x104 to 4.lx105
Pa· sand 77 2 from 1.6x10 6 to 1. 8x10 7 Pa· s. The values of E1 and 77 2
were larger than those of E0 and 771' respectively'. The same relation
was observed in various gel foods. Staleness of bread can be predicted
from its viscoelastic coefficients.

CONCLUSIONS

Creep test of crumb was conducted under the optimum testing condition
established. In general, crumbs behaved like the 4-element model, whose
viscoelastic parameters were accurately determined by the simple graphic
method.
The viscoelastic behavior of crumb was significantly affected by
storage time and temperature. The creep testing conditions and analysis
procedures for the crumb were proposed, and they were recommended as a
useful tool for evaluating viscoelastic characteristics of the crumb.

REFERENCE

1. LasztIty, R., Rheological studies on bread at the Technical UniversIty

of Budapest. ~ Texture. Studies., 11, 81-115.
2. Akabane, II., and Nakahama, N., RheologIcal measurements on cookery(part
1). Chori Kagaku. , 1988, 21(4), 245-254.
 VISCOELASTIC BEHAVIOR OF PARBOILED RICE

SAIFULLAH M. SAIF
Bangladesh Agricultural Research Council
Fanngate, Dhaka 1215, Bangladesh

DWAYNE A. SUTER
Department of Agricultural Engineering
Texas A & M University
College Station, Texas 77843, U.S.A.

ABSTRACT

Two varieties of rice were parboiled and their stress relaxation properties were investigated.
A four-element Maxwell model best described the viscoelastic behavior of parboiled rice at
a storage moisture of 10.5% and above. Bold type medium grains (Rico-I) had greater
relaxation time constant than slender long grain (Lemont) signifying more gelatinization in
the latter. Relaxation time and decay stress values increased during the first two to three
hours of storage compared to sealed storage (control) which decreased over the following
24 h period. Exposure of the dried samples to a 65% relative humidity storage environment
increased the relaxation time constant compared to control. A relative humidity of above
75% gradually decreased the relaxation time constant. Prediction models for the stress
relaxation parameters have been proposed with respect to relative humidity and storage time
for given types of grains.

INTRODUCTION

Parboiling process improves the nutritional quality of rice and head rice yield. However,
improper processing and storage conditions may drastically reduce milling yield,
particularly in partially parboiled rice. Breakage of raw rice during milling is caused by
several factors related to physical properties of the grain [1]. Much of the kernel damage
during milling might occur due to previous weakening of the kernel induced by moisture
stress causing stress cracks or fissures. However, when rice is properly parboiled most of
the existing shortcomings such as partial and full fissures are eliminated [2, 3].
Nevertheless, researchers have found that parboiled rice also fissures like raw rice when
exposed to moisture adsorbing environment [3].
During the course of mechanical handling and storage parboiled rice kernels are
exposed to a series of static and dynamic stresses, both internal and external. The
mechanical and viscoelastic behaviors are expected to dictate the breakage phenomena in
parboiled rice. The objectives of the present research were: 1. Compressive stress
relaxation behavior would be studied with reference to two varieties of rice representing
medium and long grains, 2. Evaluate the effects of storage relative humidity and duration of
storage on the stress relaxation parameters.
82
MATERIALS AND METHODS

Rice samples of two varieties, Rico-l (a medium grain) and Lemont (a long grain) were
used. Only mature plump kernels were selected and soaked in 55 C water of neutral pH for
five hours followed by 10 min exposure to saturated steam at 103 kPa. Parboiled samples
were then dried with 40 C air at the rate of 0.15 Vs per I of grain. Dried kernels at 10.5%
moisture (w.b.) were immediately transferred to four different relative humidity chambers,
namely, sealed vial, 65%, 86%, and -100% rh, maintained at room temperature. Such
stored kernels were subjected to compressive stress relaxation tests at every 0, 1, 2, 3, 9,
and 24 h intervals. An INSTRON Universal Testing Machine linked to a personal computer
equipped with Series IX AMTS Software, version 4.14, was used to carryout the tests. A
deformation rate of 1.5 mm/min and an initial load of 40 N were used. Running of the test
and data acquisition were accomplished directly through the computer. Later these data
were analyzed by Successive Residual Technique using a modified FORTRAN program
originally developed by Bashford and whitney [4]. The experimental design was a Split-
Plot Factorial. Effects of relative humidity and storage durations on the viscoelastic
parameters were determined by split-plot factorial analyses of variance.

RESULTS AND DISCUSSION

A four element Maxwell model best described the viscoelastic behavior of parboiled rice at
a storage moisture of 10.5% and above. Such models are shown in TABLE 1 for the two
types of grains corresponding to four storage environment conditions. Relaxation time
constant values were mostly higher in bold type medium grain (Rico-I) than in slender long
grain (Lemont) attributing to the presence of more ungelatinized starch granules in the
latter. Storage of dried kernels in 65% relative humidity environment increased the
relaxation time constants compared to control (sealed vial). A relative humidity of75% and
above gradually decreased relaxation time. Effects of relative humidity, storage duration,
and interaction between them were found to be highly significant on the relaxation time
constants. Relaxation time constants were always found to be significantly higher in the
first term than in others (TABLE 1). Therefore, these values in the first terms can be used
for the comparison purposes. Such a comparisons between the varieties, storage relative
humidities, and storage times are shown in TABLE 2. A peak value of relaxation time
occurred during storage time of 1 and 3 hours. On the average, the peak value of 246.6 s
occurred in Lemont after one hour storage, while the same occurred in Rico-l at two hours
of storage yielding a value of 386.6 s. The relaxation time constants obtained in this
research are much lower than those obtained by Husain et al [5] (376.5 s) and Nguyen [6]
(515.0 s) for raw rice. The reason apparently seems to be the state of structural
modification and the configuration of starch and other molecules in the grain. Indications
were that the stress relaxation behavior was improved as a result of parboiling. Higher
stress relaxation times in bold kernels possibly attributed to less uniformity in gelatinization
of starch molecules in the inner endosperm.

CONCLUSIONS

A four-element Maxwell model will best describe the viscoelastic behavior of parboiled
rice at storage moisture of 10.5% and above. Bold type medium grain has greater relaxation
time and decay stress than slender long grain , signifying more gelatinization and
homogeneity in the structure of the kernel in the latter. Relative humidities above
equilibrium relative humidity would cause decrease in stress relaxation parameters.
Combined effect of relative humidity and storage time is highly significant on the relaxation
time constant.
 83
TABLE 1
Mean decay stress (A) and stress relaxation times (min) for four tenn Maxwell models
corresponding to storage relative humidity for Lemont and Rico-I parboiled rice

Relative humidities (%)

Sealed vial 65 86 -100
Variety Tenn
--------
A
--------
A
--------
A
----------
A t
------------------------------------------------------------------------------
Lemont: 1 32.45 327.22 32.40 433.56 32.90 317.83 31.76 292.38
2 3.77 5.10 3.77 5.82 3.87 4.28 4.39 4.28
3 2.45 2.02 2.54 1.35 2.32 1.00 2.73 1.13
4 1.46 0.35 1.49 0.37 1.24 0.34 1.33 0.35

Rico-I: 1 32.39 393.16 32.82 435.57 32.13 408.61 31.52 293.73

2 2.63 5.15 3.48 5.56 3.61 5.90 4.45 4.22
3 2.22 1.38 2.35 2.39 2.30 1.37 4.48 1.04
4 1.03 0.36 1.33 0.38 1.45 0.40 1.24 0.32

TABLE 2
Mean stress relaxation times (min) corresponding to storage relative humidity and
storage time for Lemont and Rico-l parboiled rice

Storage Relative humidities (%)

Variety time (h) Sealed vial 65 86 -100
------------------------------------------------------------------------------
393.25 426.25 391.41 371.34
Lemont: 0
1 430.03 362.78 358.50
2 436.28 396.69 375.83
3 321.06 458.21 347.73 326.50
9 340.82 445.83 297.93 279.64
24 343.74 448.87 267.19 122.88

Rico-I: 0 436.07 456.06 427.26 437.84

1 471.78 381.82 367.11
2 440.00 443.18 328.50
3 423.23 424.47 393.88 287.93
9 366.67 417.91 401.57 303.28
24 346.66 406.09 403.94 143.93
------------------------------------------------------------------------------

REFERENCES
1. Spadaro, J. J., Mathews, ]" Wadsworth, J. 1. ~: Production ill!d Utilization (B. S.
Luh ed). AVI Publishing Company, Westport, CT., U.S.A 1980.
2. Bhattacharya, K. R. Breakage of rice during milling, and effect of parboiling. Cereal
~, 1969, 46, 478-85.
3. Chattopadhyay, P. K. and Kunze, O. R. Fissuring characteristics of parboiled and raw
rice. Transactions Qf ~ ASAE, 1986,29, 1760-66.
4. Bashford, l.l., and Whitney, R. W. A computer model for determining creep and
relaxation model. ASAE paper No. 73-302. St. Joseph, MI:ASAE, U.S.A
5. Husain, A, Agrawal, K. K., Ojha, T. P., and Bhole, N. a.Viscoelastic behavior of
rough rice. Transaction Qf ~ ASAE, 1971, 14, 313-314,318.
6. Nguyen, C. N., and Kunze, O. R. Fissures related to post drying treatments in rough
rice. ~~, 1984,61,63-68.
 STRESS RELAXATION CHARACTERISTICS OF ALFALFA CUBES

R. T. PATIL and S. SOKHANSANJ

Department of Agricultural and Bioresource Engineering,
University of Saskatchewan,
Saskatoon, S7N OWO, CANADA

ABSTRACT
Alfalfa is processed in the fonn of cubes, pellets and dense bales. The relaxation data
of two types of cubes subjected to 750 N load was analyzed by the method of large
deformation as well as by applying a simple Maxwell model. The cubes having different
levels of moisture content had significantly different relaxation times in Maxwell model. This
parameter could be used for grading alfalfa cubes based on their stiffness.

INTRODUCTION
Alfalfa cubes are subjected to compressive loads resulting from the static weight of
other cubes leading to the deformation in storage. It is useful to know the rheological
behavior of the cubes under these static storage loads.
Stress relaxation behaviors have been conducted to analyze rheological characteristics
of several food materials [1, 2]. The objective of this study was to characterize stress
relaxation in alfalfa cubes so as to grade them depending on their stiffness.

MATERIALS AND METHODS

The regular and large alfalfa cubes were used in the study. The regular cubes were cut
uniformly to get a cube of 35 mm x 35 mm x 43 mm and large cube was cut into 35 mm x 36
mm x 42 mm size. The moisture content of the cubes was adjusted at three levels by exposing
them to the atmosphere of 220 C and 90% RH in controlled atmosphere chamber for 24 and
36 hours.
The cubes were compressed on the plain surface under a plunger by Instron 1011
universal testing machine (Instron Corporation, 100 Royall street, Canton, Ma) at cross head
speed of 20 mm/min. The Instron Series IX Automated Materials Testing System software
package (Version 5.2) was used to conduct the tests. For relaxation test three choices were
available for the control ; load, % strain and displacement. Since this data was collected as
part of ASTM D- 575 [3] method to determine the % strain in cube under specified load, the
limit type of load level was used for relaxation tests. The cube was placed on an anvil facing
plain side up. The specific load of 750 N was applied at 20 mm/min stress relaxation data
was acquisitioned by the computer at the rate 10 points per second.

Data analysis
Large deformation analysis:
 85
Since the strain are considered large, the Henkey's strain (natural) or true strain was
used and stress values were corrected for change in the contact area. According to
Purkayastha and Peleg [cited by Stroshine at al.] the relaxation stress at any time t can be
obtained by following equation

(EO true - E true) / EO tru e = (tikI +k2t) (1)

where kI and k2 are constants. The relaxation modulus when stress decays to the value of (1-
lie) of its initial value was obtained by following equation.

Er=Ee true+ l/e(E{) true-Ee true) (2)

The analysis was performed for data on individual cube and average value and
standard error of mean for 15 cubes was determined. The values of relaxation modulus and
time required to reach this value were computed from the average values.

Small deformation analysis:

The stress relaxation behavior of viscoelastic bodies has been explained by Maxwell
model consisting of spring and dashpot in series by Mohsenin [1] and Stroshine et al. [2].
The simple Maxwell model is represented by following equation:

(3)

where at is true stress k Pa at any time t, ad is the decay stress, k Pa and Trel is relaxation
time, min. The relaxation data was analyzed by Proc NUN on SAS to find out the value of
ad, and Trel. For this analysis the data on 15 cubes was pooled and average value was used
against time.

RESULTS AND DISCUSSION

The moisture contents of the cubes were 7.0, 12.7, 15.4%, and 11.3, 15.6, 18.5% for
regular and large cubes, respectively. The relationship of t(Eo true/CEo true-Ee true» versus t
was straight line with slope k2 and intercept k l' The results in Table 1 show that the
calculated values of relaxation modulus and relaxation time varied with moisture content of
the cubes. For both sizes, the relaxation modulus decreased with increase in moisture content.
The t test analysis performed to compare the same cube at two moisture contents indicated
that there was significant difference in stiffness due to moisture content of the large cubes.
For regular size cube the difference was not significant between the cubes at 7.0% moisture
and 12.7% moisture. Further increase in moisture content of cubes to 15.4% reduced the
relaxation modulus significantly. This shows that the relaxation modulus could be used as
criteria for only large cubes.
Table 2 gives the values of constants for simple Maxwell model. The equation was
found to fit accurately to the experimental data. The comparison of the mean relaxation time
values with t test showed that there was significant difference between the relaxation times
for cubes with different stiffness as represented by their moisture contents. The relaxation
time is higher for the cubes with lower moisture content. It is obvious because at low
moisture content, the cube is stiffer than at a higher moisture content and hence longer time is
required for the applied stress to relax. This criteria could be used for grading the cubes to
access their stiffness.
CONCLUSIONS

1. The relaxation time by simple Maxwell model was significantly different for different
moisture content in both sizes of cubes. Hence this parameter could be used for numerical
grading of cubes based on their stiffness.
86
REFERENCES

1. Mohsenin, N. N., Physical Properties of Plant and Animal Materials, Gorden and Breach
Science Publication, New York, 1986.
2. Stroshine, R., Pitt, R. and Hemman, D., Physical Properties of Agricultural Materials and
food Products, Dept of Agricultural Engineering, Purdue University, West Lafayatte,
Indiana, 1992, pp. 136-141.
3. ASTM, Annual Book of ASTM standards. Vol 9.01, 1992,105 -108.

TABLE 1
Result of large deformation analysis of alfalfa cubes

Moisture K1 K2 E otrue Eetrue Er t, min

content, %wb kPa kPa kPa

Large cube at 750 N

11.3 0.052 3.31 3900 2789 a 3195 0.023
(0.002) (0.069) (59) (45)
15.6 0.037 2.26 3332 1955 b 2459 0.023
(0.001) (0.02) (54) (31)
18.5 0.040 2.40 1969 1203c 1483 0.025
(0.001) (0.02) (13) (7)
Regular cubes at 750 N
7.0 0.13 6.39 2694 2287 d 2436 0.028
(0.01) (0.12) (23) (20)
12.7 0.04 2.68 2774 2278 d 2459 0.028
(0.001) (0.02) (70) (47)
15.4 0.05 3.00 2392 1660e 1928 0.027
(0.001) (0.03) (54) (38)

Means with the same letter are not significantly different at P=0.05 Values in parentheses
are the standard errors

TABLE 2
Result of simple Maxwell model analysis of alfalfa cubes

Moisture content, %,wb ad' kPa Trel' min R2

Large cube at 750 N

11.3 387.83 0.77 0.999
(2.02) (0.06)
15.6 335.08 0.48 0.998
(2.49) (0.09)
18.5 350.22 0.45 0.998
(2.86) (0.10)
Regular cubes at 750 N
7.0 424.98 1.56 0.999
(1.04) (0.03)
12.7 382.96 0.56 0.998
(2.39) (0.08)
15.4 63.32 0.07 0.999
(0.00) (0.00)

Values in parentheses are the standard errors

 STUDIES ON RHEOLOGICAL PROPERTIES OF FOODS

YORIMASA MASUMOTO, KIYOSHI KUBOTA, and YOSHIO HAGURA

Department of Food Science, Faculty of Applied Biological Science, Hiroshima University
1-4-4 Kagamiyama, Higashi-Hiroshima, 724 Japan

ABSTRACT
Mochi, a traditional waxy rice cake consumed widely in Japan, behaves as a viscoelastic body.
The producing of Mochi has the optimum processing time to produce a favorably texture
Mochi. Then, it is necessary to do on-line texture measurement during processing of Mochi. In
this study, we tried to predict the texture properties of Mochi during stamping from the pestle's
load change (PLC). As result, a work load due to lifting pestle and the relaxation time of
damped oscillation when pestle dropped changed as preparation time of samples.
Consequently, continuous measurement of PLC is an available method to predict the optimum
processing time during stamping Mochi.

INTRODUCTION
Texture of Mochi depends on process conditions, e.g., processing methods, processing time.
In the processing of Mochi, there is the optimum processing time to produce a favorably texture
Mochi. Because, under processing products are less extendible, or over processing products
are too soft. It is difficult to control processing of Mochi automatically, because the optimum
processing time depends on the difference of lots. Then, on-line texture measurement is needed
to decide the optimum processing time automatically. We supposed that there were some
relations between the texture properties of Mochi and forces loading on the product during the
stamping process, and tried to measure the texture properties of Machi from the pestle's load
change (PLC) during stamping Mochi.

MATERIALS AND METHODS

Preparation of experimental samples
Preparing procedure of Mochi were; washing glutinous rice 3 times with water, soaking in
88
water for 12 hours, draining for 30 min, steaming for 30 min, and then kneading for 0,3,5,
10, 15, and 20 min. A commercial electric kneading machine (Toshiba AFC-166, Japan) was
used for steaming and kneading processes.

Texture measurement
Measurements of viscosity and adhesiveness were made on a meometer (Sun Science Co., R-
VOJ-OM II) equipped with a cylindrical plunger (<I>l0mmx50mm). The disc-shaped samples
(</>5Ommx25mm) were compressed under a constant rate of 6Ommlmin.

Measurement of load change during stamping

A special designed stamping machine consisting of pestle (50 and 60mm in diameter) and
mortar (62.3mm in LO. and 200mm in depth) was used for measuring the texture properties of
Mochi. One side of the pestle was connected to a round bar, which was installed with a load
cell. The kneading samples were put into the mortar and then stamped by dropping the pestle
repeatedly under controlled stroke. The forces acting on the products during the stamping
process were measured by the load cell.

RESULTS AND DISCUSSION

Pestle's load change
PLC during the stamping process was recorded continuously. One cycle of typical PLC was
shown in Figure 1. We investigated the relation between a work load due to lifting pestle (y./)
and the adhesiveness of Mochi, and the relation between a relaxation time of damped oscillation
when pestle dropped (T) and the viscosity of Mochi. Since, it was considered that W was

equivalent to the adhesiveness of Mochi and the viscosity of Mochi affected T.

tI
II /JX..Il T

~ IW Iv\F~-v--
TIME
»

Figure 1. One cycle of typical pestle's load change. I and V, load of pestle on
sample; II, load when lifting pestle that adhered sample; III, load when sample
separated from pestle; IV, load when pestle dropped on sample; W, a work load
due to lifting pestle; T, relaxation time of damped oscillation when pestle dropped.
 89
Relation between Wand adhesiveness of Mochi
W60 (<I> 60mm pestle), W50 (<I> 50mm) and adhesiveness of Machi were plotted vs. kneading
time in Figure 2. The maximum adhesiveness was observed at 5 min kneading time. Inflection
points of W60 and WSO were observed at 3 min and 5 min, respectively.

Relation between T and viscosity of Mochi

T60 (<I> 60mm pestle), TSO (<I> 5Omm) and viscosity of Machi were plotted vs. kneading time in
Figure 3. The viscosity of Machi decreased exponentially with an increase in kneading time.
While, T60 and TSO increased with an increase in kneading time except 0 min kneading time.

Namely, T increased with a decrease in viscosity.

~ '";;'15000 60000 '";;' ~ 0.60 r------.-----, 3e+6

~~ ~ ~ 0.50
~ ~ 10000 40000 ~ ~ 0.40 2e+6~
~

8~
~~ 5000
~
20000 e;
~ 0.30
~

~ 0.20 I:i-__---t-~t_____r le+6~

b
~§ ~
~ 0.10 U
~
00 10 20 0 ~ 0.000 10 20 O~
~~ KNEADING TIME [min.] KNEADING TIME [min.]
Figure 2. Kneading time changes in Figure 3. Kneading time changes in the
a work load due to lifting pestle and relaxation time of damped oscillation when
adhesiveness. . , <l>50mm pestle; pestle dropped and viscosity. • , <l>50mm
. , <I>6Omm.;'&, adhesiveness. pestle; . , <I>6Omm; ~, viscosity.

CONCLUSIONS
We examined the measurement of the texture properties of Machi from PLC during the
stamping process. The values of W and T changed as preparation time of samples. This result
shows the relations between PLC and the texture properties of Machi, and suggests that
continuous measurement of PLC enable us to predict its texture properties.

ACKNOWLEDGEMENT
The authors are very grateful to Mr. S. NAKAI, Technical Official, the Hiroshima University,
for his kind cooperation.

REFERENCES
1. Y Isono, E. Okamura, T. Fujimoto, and T. Watanabe, Linear viscoelastic properties and
tissue structure of mochi cake, Agric. BioI. Chem, 1990,54, 2941-2947.
2. N. Nagashima, Physical characteristics of mochi. Denpun Kagaku, 1992, 39, 23-31.
 RHEOLOGY OF SET TYPE AND STIRRED TYPE YOGHURT:
BUILD-UP, BREAK-DOWN AND RECOVERY;
THE EFFECTS OF pH, TEMPERATURE AND STARTER

ERIKA RONNEGARD, nN nANG AND PETR DEJMEK

Department of Food Engineering,
Lund University, PO Box 124, S-221 00 Lund Sweden, fax +46 46104622

ABSTRACT

Skim milk was heat treated (l00·C/5min) and fermented with ropy and non-ropy
commerical starters, in beaker or in rheometer cup. Yoghurt properties were followed by
dynamic rheometry (Bohlin Universal Rheometer) and by penetrometry (Stevens LFRA
Texture Analyzer) and the measuring methods were compared.
Set yoghurt storage modulus depended on pH but not on fermentation time. A
maximum in loss angle was observed during the gel development. Yield stress increased
with falling pH but yield strain was pH-independent.
Set yoghurt could be broken down in the rheometer, with no observed slip, by large
strain oscillating movement. The amount of residual structure as measured by storage
modulus was generally proportional to its initial value. The relative breakdown could be
controlled by the choice of strain and frequence but was independent of pH and temperature.
Break-down kinetics were complex, 4 to 5 constants being needed to describe the time
course of the storage modulus. No general difference was observed between the structure
breakdown of yoghurts containing exocellular polysaccharides and standard non-ropy
starters.
After moderate break-down levels, the character of gel as measured by the loss angle
was rapidly restored. The recovery of storage modulus resembled breakdown in reverse; -
stiffer gels recovered more of their stiffness, there was an initial fast increase but the total
increase occurred on several timescales.

INTRODUCTION

In a set type yoghurt, milk is fermented in a consumer package. In a stirred type

yoghurt, the fermentation is performed quiescently in a vat and subsequently the gel is
partially broken down by stirring, pumping and passage through coolers and fillers.
Additional strainers or orifice plates may be used to assure homogenity. Drinking yoghurts
may be homogenised in a high pressure piston homogenizer.
Little is known about the relationship between the rheological properties of the set-
type yoghurt gel and the properties of the whey-gel particle suspension produced by stirring
and the eventual recovery of gel character on standing. We have developed [1,2,3] a method
for the measurement of intact gel break-down and recovery under defined conditions, and
report here further data.
 91
MATERIALS AND METHODS

Yoghurt was made from pasteurized market skim milk additionally heated and kept for
5 min at 100°C. Starters were freeze-dried B3 (Chr. Hansen, Copenhagen, Denmark) and
Y2, Y 4, VI and V2 (Wiesby Gmbh, NiebiilI, Germany). For details, see [2,3].
Dynamic rheological measurements (oscillation) were made in Bohlin Universal
Rheometer (Bohlin Rheologi, Lund, Sweden) in Couette geometry (bob diam 25 mm, 0.31
mm gap) at 0.0125 strain. For in-situ break-down, the strain was increased.
For empirical measurements, Stevens LFRA Texture Analyzer, (C. Stevens & Son, St
Albans, UK) was used to record the force vs position in a penetrometer experiment :
indenting the surface of a set yoghurt gel in a 80 mm diameter beaker with a 25 mm diam
cylinder at 0.5 mm/s.

RESULTS AND DISCUSSION

YOl!hurt build-up: We found the gel build-up unaffected by the amount of starter
used and only dependent on the attained pH. At 1 Hz, start of gelation defined as elastic
shear modulus G' > 1 Pa was observed at about pH 5.5. The gel build-up was not uniform, at
pH 5.2, a plateu in G' and a maximum in phase angle 0 was found, Fig 1. The repeatability
of the rheometer measurement and starter differences are shown in Table 1.

TABLE 1
Rheological characteristics of yoghurts at 1 Hz, pH 4.5, 43°C
Starter # Elastic modulus G'[Pa] Loss angle o[degrees]
VI (viscous) 4 265 ± 12 15.4 ± 0.5
V2 (viscous) 3 331 ± 6 14.5 ± 0.4
Y2 4 249 ± 16 15.6 ± 0.3
Y4 2 268 ±6 15.8 ± 0.1

There was no general difference between viscous (ropy) and nonviscous starters. The
viscous starter V2 showed significantly higher elastic modulus, but did not differ in loss
angle.

Yoghurt break-down : Penetrometry measurement, Fig 2, showed an increased

modulus and increased yield stress with decreasing pH, but yield strain was unaffected.
With high strain oscillatory measurement, gel could be broken down gradually and the
apparent moduli measured concomitantly, Fig 3. At first glance, the moduli tend to an
apparent asymptotic value, but in fact decrease continued as long as we measured, up to tens
of hours. A simple model for the decrease is a sum of exponentials,

(1)

where t time, t 1,'c2, characteristic breakdown time constants, G' 1,G' 2 amount of fast
and slow breakdown and G'inf the estimated G' at infinite time.
The relative residual modulus, G'inVG'initial' could be simply modelled as
92
G'.m fIG'·Inttl
. ·al= exp (- 52·
. s· £D.8) (2)

where s strain and f frequency. The relative residual modulus was independent of the
breakdown temperature and the pH.

y Q2hurt structure recovery: The kinetics of recovery followed a course similiar to

that of break-down. A surprising feature was the fast recovery of phase angle to almost its
set-yoghurt value, Fig 3. There was no general difference between viscous and non-viscous
starters, but the yoghurt based on the viscous starter V2 suffered less structure loss and its
recovery was no slower, Fig 4. The initial rate of recovery did not significantly depend on
the pH.
500.---. .~----~-------,25 80~------------------~40 ~

';j' 450 70 B ~ 35 ~
~ 400 Z 60 Iii g fo!;c; 30 ~
i
0

~ ~~g 50 'b 25 ~
"S 250 ..9l cS 40 20 ~
f!
'b
S 200
~ 150
10 il 30 15 ~
gj 100 .~ 20 , d:forritatio·; -., 10 ~
4l 50 10 5 4l
0~~~~~~~~~~40 0~~~~--~--~--~-40 .~
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 4.1 4.2 4.3 4.4 4.5 4.6 4.7
time [h] pH
Fig l. Yoghurt build-up, 43 ·C, 1 Hz Fig 2. Yield force and yield deformation in
penetrometer test

350.--------------------.35 100%
90%
~ 300 30 Ul

t
'" 80%
~
250
200
25
20 ~ ~e 70%
60%
50%
S 150 15 'bh
W § l 40%

l4l
u 'lil 30%

..s~
WO f 20%
50 5 10%
0%
00 10 20 30 40 50 60 700 y2 y4 v1 v2
time [min] starter

Fig 3. Structure break-down and recovery at Fig 4. Residual elastic modulus at 43·C
43 ·C, 1 Hz, successive strains 0.33, 0.50, after 5 min break-down at 1Hz, 0.16 strain
0.67,0.0125 (filled bars), and after 40 min recovery
(open bars).

REFERENCES
l. Arshad, M., Ronneglird, E. and Dejmek, P., In situ build-up, breakdown and rebodying
of acid casein gel, IDF seminar Protein and Fat Globule Modifications, Munich,
August 1992
2. Ronneglird, E. and Dejmek, P., Development and breakdown of structure in yoghurt
studied by oscillatory rheological measurements, Le Lait, 1993, in press
3. Jiang, J., Fundamental and empirical measurements of set and stirred yoghurts based
on viscous and non-viscous starters, MSc thesis, International Postgraduate Dairy
Course, Lund, 1993
 MONITORING OF MILl CURD FORMATION BY lEANS OF PHYSICAL PROPERTIES

H.NAKANUMA. T.DOI. R.WATANABE

Engineering Research Center.
Morinaga Milk Industry CO .• LTD
4-515~ Tateno. Higashiyalato-shi. Tokyo 207. Japan

ABSTRACT
Diffuse reflectance by the optical fiber sensor. curd tension. and near
infrared spectra were measured during the enzymatic coagulation of lilk.
Electron microscope observation of milk curd was also carried out.
Increasing of diffuse reflectance. near infrared absorbance. and curd
tension was observed. The optical change during milk curd forlation was
very similar to the change of curd tension. Thus. it was found that the
state of milk curd could be continuously monitored by measuring diffuse
reflectance using the optical fiber sensor.

INTRODUCTION
[n the cheese making process. the first step is enzYlatic coagulation of
milk followed by cutting the coagulum into cubes. It is known that the
firmness of the curd at the cutting point critically affects the recovery
of milk curd and quality of the product. Thus. it is very important to
determine the optimum cutting time. Conventionaly. cheesemakers determine
the cutting point by visual observation of the coagulum. But it is very
difficult even for an expert.
A sensor which can monitor the state of milk curd would allow the
coagulum to be cut at the optimul point for cheese making. Recently.
several sensors that monitor the optical change during milk curd
forlation have been reported ll2l • In this study. with the objective of
understanding the milk curd formation in detail. we lonitored the milk
coagulation by optical fiber sensor. curd tension meter, near infrared
spectrometer and electron licroscope.
MATERIALS AND METHODS
Milk curd formation was carried out as follows. 200ml of raw milk was
heated to 33~ in a water bath. and then prescribed amount of rennet
solution was added. Diffuse reflectance was measured continuously using
E3XA-CC4A fiber sensor(OMRON. wavelength;660nm), and a recorder. Curd
94
tension was measured using SUN RHEO METER(SUN SCIENTIFIC). Near infrared
spectra were measured from the wavelength of 680 to 1235nm by NEOTEC-
6250(NlRECO) using a quartz cell. Electron microscope observation of the
section of frozen milk curd was carried out using JSM-5400(JEOL).
RESULTS
Fig.1 shows the changes of diffuse reflectance and curd tension with time
after adding rennet(30ppm) for .ilk curd formation. Diffuse reflectance
increased with time and reached a plateau at 40min. After the plateau,
a secondary peak appeared, and syneresis of the curd was observed. Curd
tension started increasing when diffuse reflectance reached the plateau,
and then increasing rate became lower at 110 .inutes. After the secondary
peak of diffuse reflectance, increasing rate of curd tension became
higher again because of syneresis.

60
~
"
~
60 ~
~ "§
-
~
u ~
~
00

~ 40 ~
~ ~

~ ~

B
00 ~
20
~
~
A
20
• •
0
·0
0 20 40 60 80 100 120 140 160 180 200 220
Time / min
Fig.1 The changes of diffuse reflectance and curd tension for milk
coagulation using 30 PPI rennet concentration.

We also measured the changes of diffuse reflectance and curd tension

using various rennet concentrations of 30-120ppm. The shapes of the
curves were similar to those of 30pPI. The higher the concentration was,
the earlier the time when diffuse reflectance and curd tension started
increasing was. Also, curd tension started increasing at the point when
diffuse reflectance reached the plateau at any rennet concentrations. By
visual observation, curd cutting just after the points was optimUm.
Absorbance of near infrared transmission also increased during the
coagulation of milk(30ppm rennet) at any wavelength. However, no change
in the shape of the spectra due to the enzyme reaction could be found.
By electron microscope observation, it was found that with time
after adding rennet, casein micelles in raw milk disappeared, and the
surface of the section became uniform when the curd was formed.
DISCUSSION
Diffuse reflectance can detect the point when curd tension starts
increasing at any rennet concentrations. Further, the points when
syneresis occurs agree with the secondary peak. Thus, we assume that
 95
diffuse reflectance is suitable for lonitoring the state of milk curd.
Near infrared absorbance can also lonitor curd for.ation, but
transmission is not suitable for a sensor fixed in a process line
because of the cOMPlexity of the structure.
Fig.2 shows a lodel of curd and light paths nearby the optical
fiber sensor tip. Taking the results of curd tension leasurement and
electron microscope observation into consideration, the relationship of
the change of diffuse reflectance and the state of lilk curd can be
explained as follows. First, lilk solid disperses and forls a colloidal
system in raw milk. So the intensity of diffuse reflectance is weak
because of wide space as indicated by lodel "a". Second, the curd is
formed by addition of rennet. In lodel "b", the space where light
transmit become narrow. So diffuse reflectance increases. When the curd
is formed sufficiently, the coagulul covers the e.itter of the sensor
tip. Thus, diffuse reflectance stops increasing. It is the plateau of
diffuse reflectance as shown in Fig. 1. Third, when syneresis occurs as
shown in model "c", the curd shrinks, and a very narrow space between
the sensor tip and the .ilk curd is forled. Then sufficient light emits
and diffuse reflectance starts increasing again. This point agrees with
the secondary peak of the diffuse reflectance. Finally, when syneresis
proceeds, the space grows wide because the curd shrinks as shown in
Model "d" . It becomes easier for I ight to translit. Therefore, the output
of the sensor also starts decreasing after the secondary peak.

eIIitter
't} e
a) before adding rennet ====~~~.~~ c) at the secondary peak
of reflectance
receptor
J

d) after the secondary peak

of reflectance

Fig.2 A model of milk curd and light paths nearby the sensor tip.

CONCLUSIONS
By measuring diffuse reflectance, milk curd forMation could be monitored
continuouslY, and we could know the optimum point for cutting the
coagulum, further, syneresis could also be detected. If the optical fiber
sensor will be fixed in the cheese laking process line, we will be able
to control the cheese production auto.atically and ROSt suitably.
REFERENCES
1. Mcmahon, D. J., Richardson, G. H., and Brown, R. J., EnzYlic milk
coagulation: Role of equations involving coagulation time and c~rd
firmness in describing coagulation, J. Dairy Sci., 1984. 67. 1185-93.
2. Hardy, J., and Fanni, J., Application of reflection photometry to the
measurement of milk coagulation, J. Food Sci.. 1981, 46. 1956-57.
RHEOLOGICAL ANALYSIS OF JELLY STRENGTH IN HEAT INDUCED GEL LIKE KAMABOKO

HIDEMASA MIKI, 10 SHINDO, JUN-ICHI NISHIMOTO

Laboratory of Food Preservation,
Faculty of Fisheries, Kagoshima University,
4-50-20, Shimoarata, Kagoshima 890, Japan

ABSTRACT

The relationships between jelly strength and rheological properties by the creep test were investigated
for heat-induced gels to estimate the rheological characters from the values of the breaking strength
and breaking deformation which is composed of jelly strength.
The breaking strength showed similar characteristics to the elastic modulus. The breaking
deformation was similar to the viscous modulus. The jelly strength properties could be applied to
estimating the rheological character so that it can improve the estimation of texture on surimi-based
products.

INTRODUCTION

Elasticity is an important factor in deciding the quality of surimi-based products such as Kamaboko.
The texture of these products has been estimated from jelly strength (JS) by the penetrating test [1, 2],
but it is difficult to know the rheological characteristics. In this paper, the relationships between the JS
and the rheological properties by the creep test are discussed to estimate the rheological character from
the values of JS.

MATERIALS AND METHOD

The fish paste was prepared from surimi (walleye pollack, grade SA) which was ground, adjusting the
final water contents to 70 to 90% with soluble starch contents from 5 to 20% and salt content of 3% to
get typical values of JS components. The heat-induced gels (HIG) like Kamaboko were prepared from
the fish paste, which was charged into the sample holder of stainless steel pipe (30 mm i.d. x 25 mm)
and heated at 90°C for 30 minutes in the water bath. The JS properties of the breaking strength (BS[gD
and the breaking deformation (BD[mmD were measured by Yamamoto Food-checker (San Kagaku-
3A2B) with the spherical plunger 8 mm o.d. Rheological experiment and analysis were carried out by
Rheoner (Yamaden-RE3305) with a disk type plunger 30 mm o.d. connected by a software program,
which by applying the experimental data gives the 6 elements model due to the theory of linear visco-
elasticity[3]. Analysis of relationships between the JS and rheological properties were carried out by
factor analysis following principal factor analysis.
 97
RESULTS AND DISCUSSION

Figure 1 shows the prepared values of BS and BD for all the RIG. The line BSI (No.7, 6, and 3) and
the line BS2 (No. 10, 9, 8, 11, and 12) indicated the changes of the BS with invariable values of the BD
as about 8.01 and 10.35 mm. The line BD (No.5, 1, 3,2, and 4) indicated the changes of the BD with
invariable values of the BS as about 556 g.

1000

-:g,
......
0"1

800 6 9

~
.-~
0"1
C
600 5 SO
f 1 4
m 12
400
5 6 7 8 9 10 11 12
Breaking deformation (mm)

Figure 1 The values of breaking strength and breaking deformation for all the heat induced gels.

Table 1 The factor loadings of breaking strength

and the 6 creep elements by factor analysis for
heat-induced gels
Factor 1 Factor 2 Factor 3 Factor 4
BS 0.834 -0.525 0.118 -0.117
Eo 0.846 -0.487 0.212 0.050
E, 0.897 0.271 -0.056 0.334
n, 0.878 0.422 0.228 0.001
E2 0.957 -0.135 -0.258 0.009
n2 0.960 -0.111 -0.210 -0.147
nn 0.625 0.755 0.034 -0.194

Table 1 shows the factor loadings of the BS and the 6 creep elements by factor analysis for the RIG.
In the second factor, the factor loadings of the BS, the instantaneous elastic modulus (Eo), and the 2nd
retardation elastic modulus (E 2) indicated negative values. Also, factor loadings of the 1st retardation
elastic modulus (E l ) indicated the low values as 0.271. The BS was classified into the same group as
elastic modulus.
Table 2 shows the factor loadings of the BD and the 6 creep elements for the RIG. In the fourth
factor, the factor loadings of the BD, viscous moduli in the 1st, 2nd retardations and the steady flow
region (respectably nb n2, and nn) indicated negative values. The BD was classified into the same
group as viscous modulus.
The three-dimensional objections were illustrated for the relationships among the BS, the invariable
BD, and the instantaneous elastic modulus (Eo) and the relationships among the BD, the invariable
BS, and the viscous modulus in the steady flow region (nn) using regression equations.
98
Table 2 The factor loadings of breaking
deformation and the 6 creep elements by factor
analysis for heat-induced gels
Factor 1 Factor 2 Factor 3 Factor 4
BD -0.785 -0.392 0.456 -0.148
Eo 0.734 -0.637 0.010 0.234
El 0.983 -0.151 0.086 -0.056
nl 0.978 -0.121 0.136 -0.099
E2 0.714 -0.595 -0.328 0.170
n2 0.060 0.997 0.013 -0.047
nn 0.956 -0.060 -0.107 -0.266

CONCLUSIONS

The breaking strength showed similar characteristics to the elastic modulus. The breaking deformation
was similar to the viscous modulus. The jelly strength properties could be applied to estimating the
rheological character so that it can improve the estimation of texture on the surimi-based products.

REFERENCES

1. Kohyama, K., Nishinari, K., and Shimizu, Y. Relationship between texture of Kamaboko and its
mechanical properties. l. lpn. Soc. Food Sci. Tech., 1990, 37, 612-8.
2. Montejano, J. G., Hamann, D. D., and Lanier, T. C. Comparison of two instrumental methods
with sensory texture of protein gels. 1. Texture Stud., 1985, 16, 403.
3. Nakagawa, T. and Kanbe, H. Rheology, Misuzu Publishers, Tokyo, 1959, pp. 493-529.
 RELATIONSHIPS OF RHEOLOGICAL AND HYDRODYNAMIC PROPERTIES OF
BIOPOLYMERIC INGREDIENTS TO THE COMPOSITE SURIMI GEL TEXTURE

CHONGM.LEE
Department of Food Science & Nutrition,
University of RhodeIsland, Kingston, RI 02881, U.S.A.

ABSTRACT

Surimi(fish myofibrillar protein)-based composite gels were prepared by incorporating various

bioploymeric ingredients (starch, gum, protein). Hydrodynamically active biopolymers
underwent thermally-induced swelling and water binding, where the more hydrodynamically
active a dispersed biopolymer, the greater the gel strengthening effect it exhibited. The
centrifugally-measured physically bound water contributed more than the DSC-measured
unfreezable bound water(-30·C) to gel strengthening effect of added proteins. Firmness
(compressive force) and elasticity(relaxation rate) of dispersed gum globules highly correlated
with compressive strength of composite gels after dispersed gum was thermally activated.

INTRODUCTION

Protein gel-based composite foods are prepared by incorporating various types of particulates,
usually ofbiopolymeric nature, in a continuous matrix. The structural and material properties of
such composite foods are affected by the physicochemical properties of ingredients and their
interaction with a protein gel matrix. In the present study, we focused on hydrodynamic
properties (dynamics of water binding and distribution of moisture between dispersed particulates
and matrix) (1,2) and rheological properties of thermally activated dispersed particulates (3) in
relation to their effects on the composite gel texture.

MATERIALS AND METHODS

Surimi(fish myofibrillar protein)-based composite gels were prepared by solubilizing surimi

(600g) with salt (2%) and water (to adjust the final moisture to 78%) and incorporating various
bioploymeric ingredients (starch, gum and protein) using a silent cutter with a total of 10 min
comminution. Compressive force of the resulting composite surimj gels (cooked at 90·C for 20
min) was measured as gel strength using the compression test (4). Hydrodynamic properties of
dispersed protein particulates were evaluated by the combination of centrifugation (3000g for 10
min after heated at 85·C for 20 min for water binding) (1) and DSC (unfreezable water at -30·C
as bound water) (2). Compressive and elastic properties of thermally activated starch and gum
particulates were determined by compressing the gel placed in a cylinder ( 3.15cm diameter x
3.85cm height) to 2mm depth with a plunger having a flat disc(31.4mm diameter) and relaxing
upon holding. The initial compressive force at 2mm deformation was used as firmness and the
100
relaxation rnte (slope from a double log graph of force over time) as elasticity. The starch and
gum samples were hydrnted with the corresponding amount of water which was measured by
DSC as free water in the surimi gel matrix after heated to 90' C for 20 min.

RESULTS AND DISCUSSION

As seen in Fig. 1, water binding of various proteins measured by centrifugation showed a good
correlation (r=0.93) with compressive force (gel strength) of protein-incorpornted surimi gels
where soy protein isolate(SPI) and spray-dried egg white(EW) had the highest WBA and
lactalbumin(LA) and whey protein concentrnte(WPC) the lowest. Whereas the bound water
measured by DSC correlated poorly (r=0.63). When the physically bound water was computed
by substrncting the DSC-bound water from the centrifugation-bound water, it correlated well
(r=0.91). This indicates that physically bound water of incorpornted protein particulates
contributed to surimi gel strengthening more than the DSC-bound water. It should be noted that
the water retained by protein particua1tes after centrifugation was mostly made up with physically
bound water which amounted to five times greater than DSC-bound water.

60~---------------------'
........
§2
.........
40 50 (r=O.63)
L.U
~ 30 40
o
L1.. 30
~ 20 (r=O.93) 20 LA III eMPI
o 10 10 4-~--.---r""~--..----r---r~---.---'---'...----t
U 0 1 2 3 4 0.0 0.1 0.2 0.3 0.4 0.5 0.6
CENTRIFUGATION WBA(g WATER/g SOLID) DSC-BOUND WATER (g WATER/g SOLID)
Figure 1. Relationship of water binding of various proteins to composite surimi gel strength.
(WG : wheat gluten; MPI: milk protein isolate)

Along with the role of water binding, the moisture distribution between dispersed particulates and
matrix was evaluated in terms of moisture retention measured by centrifugation. As seen in Fig.2,
depending upon the water binding capacity of each incorpornted protein at a set moisture level, the
added moisture can be retained entirely or partially by the protein. For partial retention, the
unretained water will be available to the surimi protein gel matrix. This, in tum, weakens the gel
due to the dilution effect of added moisture. Fig. 2 also shows the extent of water retained by
incorportated proteins at varying levels of added moisture. Complete retention of added moisture,
less complete retention, and little retention were found in SPI, EW and WPC, respectively. Even
at a high moisture level, SPI exhibited complete retention, while the retention got slightly weaker

~z
z - 100
-0 COMPLETE
Q~
I-(!)
Z::J INCOMPLETE
UJ L1..
1--
80
-0-- SPI

0-
UJC::
c::~ -0- EW
UJUJ
c:: U
60 PARTIAL
::Jc:: • WPC
I-UJ
!!!t
0« 40+-~.-~.-~r-~~~~---,---.---.---.---.~~
:::E ....... 74 76 78 80 82 84
88 90 86 92
MOISTURE CONTENT OF PROTEINS (%)
Figure 2. Changes in moisture retention in various proteins and moisture distribution model
 101
for EW and no increases in retention despite increased moisture for WPC. This suggests that gel
strengthening effect would be less affected for SPI and EW, but greatly for WPC by moisture.
Gum and starch become hydrodynamically active upon hydration and heating and exhibit gel
strengthening effect when incorporated in the surimi gel (3). Such gel strengthening effect by
gum and starch was believed to be related to their viscoelasticity. In our earlier work done on
potato starch (5), a maximum gel reinforcing effect was reached at 75·C heating, followed by
90·C and 60·C. This coincided with the starch paste viscosity (10% solid) peaked at 76·C. A
similar observation was made with 75% moisture level where the highest gel reinforcement was
achieved, and the gel reinforcement decreased steadily with an increase in moisture. The results
suggest the important role of viscoelasticity of thermally activated dispersed particulates. When
carrageenan gums of different types (lambda, iota and iota with synergists-K2C03, Na2C03 and
CaS04) were thermally activated at a predetermined moisture(96%), initial compressive force
(finnness) and relaxation rate(elasticity) highly correlated with gel strength of gum-incorporated
surimi gels (Fig. 3). This clearly indicates that the textural properties of composite gels are
affected by rheological properties, namely, firmness and elasticity of dispersed particulates that are
thermally activated. However, when pastes of different starches were tried, a poor correlation

50~--------------------------------------------------~ 50~-------------------------------------------------------~
""'" (iota +
~ 40 (r=0.93)
synergist) 40
UJ
~ 30 30
o
u. 20 20
c::
Q..
:::E
10 10 (r=-O.88)
8 0 +-r_;""",-r-r--r-T-r-"T'""T""'T'"'"'"r"""T""T""'T'""'T'"'"'"r-4 O;---r---,,--..---.--r-----r-........-,--..~
o 50 100 0.12 0.14 0.16 0.18 0.20 0.22
150 200
INITIAL COMPR FORCE OF CARR GUM(G) RELAXATION RATE OF CARR GUM
(LOG FORCE,g/LOG SEC)
Figure 3. Firmness and elasticity of dispersed carrageenan gums on composite gel strength.

was observed. Such a lack of correlation is believed to be due to the differences in the physical
and structural states between paste and swollen starch granules.

REFERENCES

1. Chung, K.H. and Lee, C.M. Relationships between physicochemical properties of nonfish
protein and textural properties of protein-incoroporated surimi gel. J. Food Sci. 1990, 55,
972-975 & 988.
2. Chung, K.H. and Lee, C.M. Water binding and ingredient dispersion pattern effects on
surimi gel texture. J. Food Sci. 1991,56, 1263-1266.
3. Lavery, S. and Lee, C.M. The relationship of physical and water binding properties of
gelling biopolymers to their gel-strengthening effect in composite surimi gel. 1991. Master
thesis, Dept. of Food Science & Nutrition, University of Rhode Island, Kingston.
4. Lee, C.M. and Chung, K.H. Analysis of surimi gel properties by compression and
penetration tests. J. Text. Stud. 1989, 20, 363-317.
5. Kim, J.M. and Lee, C.M. Effect of starch on textural properties ofsurimi gel. J. Food Sci.
1987,52, 722-725.

ACKNOWLEDGMENT

Contribution # 2856 of the College of Resource Development, University of Rhode Island, with
support from Rhode Island Agricultural Experiment Station.
 HYPOCALORIC JAMS FROM GRAPE JUICE

ELSA C. MATIAS, ISABEL M.N. SOUSA and OLGA LAUREANO

S.A.C.T.A., Instituto Superior de Agronomia, Universidade Tecnica de Lisboa
Tapada da Ajuda, 1399 Lisboa Codex; Portugal

ABSTRACT

To find alternative uses of grapes of low enological potential and increasing the offer of dietetic
products, the production of hypo caloric jams from grape juice is studied. To optimise the formulation
of the jams, the response surface methodology (RSM) uses as independent variables sugar content
(200Brix to 500Brix), low methoxyl pectin (0.5%-1.5%), and processing temperature (55°C-90°C).
The dependent variable chosen to find the optimal formula was global appreciation (GA) from sensory
analysis. The resulting polynomial equation (R2 = 0.92) showed the maximum value for GA to be
38°Brix of sugar content, 1.2% of pectin addition, and 69°C processing temperature. The optimised
jams had 8.5 x 10-3 kg of hardness and 2.6 of cohesiveness determined from textural profile analysis.

INTRODUCTION

A reduction in wine consumption has led wine producers to search for alternative uses for their grape
production. On the other hand, the worries over healthy food are increasing as is the subsequent
demand for dietetic products. The production of hypocaloric jams from grape varieties with low
enological potential or production surplus will accomplish the above objectives. The pectins are one of
the major gelling agents used in the food industry. The use of low methoxyl (LM) pectins may reduce
the content of sugar for gelling if the level of ester groups is sufficiently low (1). The characteristics of
LM pectins and their gel characteristics have been previously studied (2).
The objective of the present work is to find a mathematical model for optimisation of the grape juice
jam in what concerns the level of pectin added, sugar concentration and temperature of the process.

MATERIALS AND METHODS

Materials: Grape juice with a pH value of 3.4, total solids of 18.8°Brix and 125.7 mg Ca++/dm3 and an
ami dated low-methoxyl pectin, 325NH95 (Sanofi Bio-Industries).

Preparation of jams: Preliminary experiments showed that the calcium concentration in the juice will
be enough for a good gelling performance of the pectin. So, the calcium level was not considered in this
study. The jams were prepared with different formulations following a central composite rotatable
experimental design (CCRD) with 6 repetitions of the central point (3). The three dependent variables
studied were used at the following range: sugar concentration (Xl) from 200Brix to 50o Brix, pectin
concentration (X2) from 0.5% to 1.5% and processing temperature (X3) from 55°C to 90°C. The
response (dependent variables) were: global appreciation from sensorial panel, hardness and
cohesiveness. The RSM technique (4) was adopted to optimise the formulation from the CCRD.
 103
Methods: Sensory analysis was performed by a test panel of twenty members. An unstructured scale
was used considering the attributes: aroma intensity (AI), sweetness (S), acidity (A), taste intensity
(TI), hardness (H), spreadability (SP), textural appreciation (TA) global appreciation (GA). Textural
parameters were calculated from a texture profile analysis (TPA) (S) using a TAX-TI (Stable
Microsystems) texturometer with a 1/1 cylinder probe. The probe was set to move 0.4 cm into the jam at
a speed of 2 mm s-', back to surface, wait for S s and repeat the test. A Rotronic-Hygroskop DT was
used for a w determinations. The results presented are the average of at least five repetitions for textural
determinations and three repetitions for the chemical results.

RESULTS

The overall quality of the jams can by evaluated by the attribute 'global appreciation'. The sensory
data studied by the principal component analysis (Figure 1) showed that the first component (axis 1) is
related with the hedonic appreciation and the second (axis 2) represents mainly the intensity of 'aroma'
and 'taste'. The 'textural appreciation' is highly correlated with the 'global appreciation' (GA).

is 2
13

14 10

7 18
Axis 1 1
-2 2

-1

Figure 1 Principal component analysis of the data from sensory evaluation (projection of the variables
and the twenty formulations from CCRD).

The GA was used to optimise the jam formulations. These results showed a good fit (Table 1) to a
second order polynomial equation (the coefficients are at least significant p<O.OS):
y=-28.97 + 1.3Sx, + 14.S9x2 - O.Olx/ - 2.767xl + o.oOlxl - 0.202xlX2 - 0.004XjX3 (1)
This equation shows that the maximum GA of the jam is obtained for 38.4% of sugar content,
1.32% of LM pectin added and 68.8°C of temperature processing. The corresponding polynomial
equations for cohesiveness, hardness and aware respectively:
y=-10.371 + 0.402Xl + 3.439x2 + 0.087x3 - 0.002Xj2 - 0.092x,X2 - 0.002X1X3 (2)
y=0.S70 - 0.023xl + 0.27SX2 - 0.01lx3 - 3E-4X1 2 - 0.6E-4X22 - 2 X 1O-3xjx2 (3)
+ 1.6E-4xjX3 - 2E-3X2X3
y=0.902 + O.003xj - 5.8E-5X12 - lE-5xjx3 + lOE-3x2x3 (4)
From the equations (1), (2) and (3) the textural characteristics of the optimised jam are 8.S x
10'-3 kg for hardness and 2.5S for cohesiveness with an aw of 0.93.
The correlation matrix between instrumental determination of the textural jam parameters and the
sensory analysis data showed that textural hardness is positively correlated with sensorial hardness
(r = 0.87; p<O.OOl) and negatively correlated with spreadability (r = 0.79; p<O.OOl). Our results did
not show significant correlation between textural hardness and global appreciation from the panel. On
the other hand the textural cohesiveness is highly correlated with sensorial hardness and global
appreciation (r = 0.77; p<O.OOl).
If it is desirable to reduce further the sugar content «38%) one could use model (1) to predict the
respective content of pectins to add and the process temperature.
104

Table 1 Statistics of the fitted second order

polynomial models
Equation R2 F residual stand error
·(1) 92.5 21.0 *** 0.467
(2) 79.2 8.3 *** 0.219
(3) 91.9 130.1 *** 0.016
(4) 75.0 41.3 *** 0.006
*** (p<O.OOl)

CONCLUSIONS

From a Portuguese panel test, the best grape juice jam has 38% sugar, 1.2% low-methoxyl pectin
added, processed at 69°C. Its textural characteristics are 8.53 x 10-3 kg of gel hardness and 2.55 of
cohesiveness. The aw value was 0.93, meaning the necessity of preservative addition. Cohesiveness was
the instrumentally evaluated characteristic that presented significant correlation with the global
appreciation of the jam.

REFERENCES

1. Morris, V. J., in Functional Properties of Food Macromolecules, ed. J. R. Mitchell and D. A.

Ledward, Elsevier Applied Science, London, 1987, pp. 143-50.
2. Chang, K. C. and Miyamoto, A. Gelling characteristics of pectin from sunflower head residues. f.
Food Sci., 57(6). 1435-1438 (1992).
3. Gacula, M. C. and Singh, J. Statistical Methods in Food and Consumer Research, Academic Press,
New York, 1984, pp. 241.
4. Mitchell, J. R., Back, H., Gregson, K., Harding, S. and Mather, S., Optimisation of products and
processes, in Chemistry and Physics of Baking, ed. Blanshard et al., Royal Society of Chemistry,
London, 1986, pp. 237-50.
5. Szczesniak, A. S., Brandt, M. A. and Fridman, H. H. Development of standard rating scales. f.
Food Sci., 1963, 28. 397-406.
 MECHANICAL PROPERTIES OF FOODS NEAR THE SOL-GEL
TRANSITION POINT

HITOSHI KUMAGAI, TETSUO INUKAI, TOMOYUKI FUJII AND

TOSHIMASA YANO
Department of Agricultural Chemistry, The University of Tokyo,
1-1-1 Yayoi, Bunkyo-ku, Tokyo 113, Japan

ABSTRACT

The mechanical properties of polyacrylamide (model substance), gelatin, and 1(-

carrageenan near the sol-gel transition point were analyzed by the percolation theory.
The values of the critical exponent for sol viscosity, s, and that for gel elasticity, t,
were evaluated by fitting the experimental data to the scaling law. The value of s for
polyacrylamide was estimated to be about 0.7, being similar to that predicted by the
percolation theory assuming the superconductor/normal mixture model, 0.7 - 0.8. The
estimated values of s for gelatin were larger than the predicted value. The values of t
for polyacrylamide, gelatin, and 1(-carrageenan were 1.7 - 2.0, being similar to that
predicted by the percolation theory.

INTRODUCTION
It is important to confirm that percolation theory is applicable to food gels with
complicated gelation mechanism, not only for controlling the mechanical properties of
food gels but also for examining the universality of the percolation theory for
describing the gelation phenomena. In this study, the mechanical properties of
polyacrylamide (model substance), gelatin, and 1(-carrageenan near the sol-gel
transition point were analyzed by the scaling equation derived from the percolation
theory.

MATERIALS AND METHODS

Percolation Theory [1]

Sol viscosity, T], and gel elasticity, G, are expressed by the following scaling equations
derived from the percolation theory, near the sol-gel transition point.

T]ee IXg-xl-s (1)

Gee /X-Xg/t (2)

106
where s is the critical exponent for sol viscosity, t is the critical exponent for gel
elasticity, x is a parameter such as temperature (T) and monomer concentration (I/J),
and ~g is X at the sol-gel transition point.
Viscosity 11 or correlation length ~ of the bonded monomer cluster diverges at the
sol-gel transition point x g. The parameter x g for gelation was estimated by
extrapolating 1/11 to zero. As for K-carrageenan, Xg was taken as x where the
correlation length diverged. The correlation length ~ was measured with a Malvern
4700 photon correlation spectroscopy system, using a 30mw, 633nm He-Ne laser
(NEC).
The critical exponent s was estimated from the slope of the double logarithmic
plot of 11 vs. 1xg - x I. The critical exponent t was estimated from the slope of the
double logarithmic plot of G vs. 1X - Xg I.

Measurements of sol viscosity and gel elasticity

Sol viscosity was measured with a cone and plate viscometer, gel elasticity being
measured at a frequency of 2 Hz.

RESULTS AND DISCUSSION

Figure 1 shows the dependence of correlation length ~ on K-carrageenan concentration.

The correlation length ~ diverged both on sol and gel side. The concentration at the
sol-gel transition point, I/Jg, was estimated, as shown in the Fig. 1.

10

~
o Sol
S 8
:::t
........ • Gel
UJ>
..c
.....
Q/j 6

- c::
C1)

c::
.....
-
0
..... 4
~

~ 0
u 2 lPg

0
/
0.10 0.20 0.30 0.40 0.50
lC-Carrageenan concentration t/J [xlO-2 kg/kg solution]

Fig. 1 Dependence of correlation length on K-carrageenan concentration at 25 "C.

The concentration of potassium ion was 0.169 x 10-2 kg/kg-solution.
 107
Table 1 shows the values of the critical exponent s. The value of s for
polyacrylamide was about 0.7, being similar to that predicted by the percolation theory
assuming the superconductor/normal mixture model, 0.7 - 0.8 [1]. The values of s for
gelatin were larger than the predicted value. The value of s has been estimated by
computer simulation assuming that the entanglement between adjacent molecules is
neglected [2]. During the gelation of gelatin, coil-helix transition occurs and the
formation of helices implies as aggregation of the chain. Therefore, as for gelatin, the
assumption would not be satisfied, causing the discrepancy between the calculated and
experimental values of s. On the other hand, polyacrylamide is a copolymer, and the
bond is relatively simple. Then the above assumption may be satisfied.

TABLE 1
Critical exponents for viscosity s

Parameter X
Materials Concentration q> Temperature T

Polyacrylamide 0.7
Gelatin 1.1 0.9

s (calculated) =0.7 - 0.8

Table 2 shows the values of the critical exponent t. The values of t for
polyacrylamide, gelatin, and K-carrageenan were 1.7 - 2.0, these values of t being
similar to that predicted by the percolation theory [1].

TABLE 2
Critical exponents for elasticity

Parameter X
Materials Concentration q> Temperature T

Polyacrylamide 2.07
Gelatin 2.05 1.9
K-Carrageenan 1.74 1.8

t (calculated) = 1.7 - 2.0

REFERENCES

1. de Gennes, P. G., Scaling Concepts in Polymer Physics, Cornell University Press,

Ithaca and London, 1979.
2. de Gennes, P. G., Viscosite. Viscosite pres d'une transition sol-gel, C. R. Acad.
Sc. Paris, 1978,8286, 131 - 3.
EFFECTS OF SUGARS ON THE GEL-SOL TRANSITION OF AGAROSE AND K-CARRAGEENAN

KATSUYOSHI NISHINARI 1 , TOKOHISA TAKAYA 1 , KAORU KOHYAMA 2 and MINEO WATASE 3

IDepartment of Food and Nutrition, Faculty of Science of Living,
Osaka City University, Sumiyoshi, Osaka 558, Japan
2National Food Research Institute, Tsukuba 305, Japan
3Chemical Research Laboratory, Faculty of Liberal Arts,
Shizuoka University, Ohya, Shizuoka 422, Japan

ABSTRACT
The effects of various sugars such as ribose, glucose, fructose, man-
nose, galactose, sucrose, maltose and raffinose, on the gel-sol transition
of agarose and /\, -carrageenan were studied by differential scanning calo-
rimetry (DSC) and by viscoelastic measurments. The addition of sugars
except ribose shifted the melting temperature Tm and the setting tempera-
ture Ts to higher temperatures, and increased the elastic modulus of gels.
However, the excessive addition of sugars decreased the elastic modulus of
agarose and /\, -carrageenan gels. This suggests that the hydrogen bonding
between hydroxyl groups in agarose or /\'-carrageenan and sugar is newly
created, and that it stabilizes the structure of junction zones.

INTRODUCTION
Agarose and carrageenan have been studied as a model for gelling sub-
stance in the food industry and in biology. Many investigations have aimed
at clarifying the gelation mechanism and gel properties of agarose and
carrageenan [1 J.

MATERIALS AND METHODS

K-Carrageenan was extracted from Eucheuma cottonii produced in Korea
in 1989 using potassium hydroxide, as described previously [2,3J. The
sulfur content by elemental analysis was 6.3% and the molecular weight by
gel permeation chromatography (GPC) at 40°C was about 1. 8 x 10 6 • Dextran
was used as a standard material in GPC, and 4M urea was used as a solvent.
The contents of the inorganic ions K, Na, Ca and Mg were determined using
 109
an SAS-760 atomic absorption spectrum frame method as K: 7.93%, Na: 0.31%,
Ca: 0.6% and Mg: 0.05%, respectively.
Agarose was extracted at l29°e from Gelidium amansii produced in the Izu
Suzaki region in 1989 in the same way as reported previously [4,5J. The
molecular weight determined by gel permeation chromatography at 400C using
0.1 M KSCN solution as a solvent to prevent gelation was 1. 41 x 10 5.
Dextran was used as a standard material. The 3, 6-anhydro-L-galactose con-
tent determined by the quantitative colorimetric method using fructose as
the standard was 41.8%. Sulfate content determined by elemental analysis
for powdered specimen was 0.0035 moljCsHlo05.
Ribose, glucose, fructose, mannose, galactose, sucrose, maltose, raffi-
nose of the extra reagent grade were used without further purification.
DSC measurements were carried out by a sensitive differential scanning
calorimeter 5600 (Seiko Electronics Inc.). A sample of 45 ± O. 1 mg of gel
kept at 2°e for two days was put into a silver pan of 70 /11. It was
heated at 2OC/ min from 5OC. The endothermic peak accompanying the gel-to-
sol transition was observed at the temperature Tm. Then, it was kept at a
temperature 200C higher than Tm for 10 min, and it was cooled at 2OC/min.
The exothermic peak accompanying the sol-to-gel transition was observed at
the temperature Ts.

RESULTS AND DISCUSSION

A sharp endothermic peak was observed in the heating DSC curves around
40---60OC for 2 wt% f(, -carrageenan gels and around 80---90OC for 2 wt% aga-
rose gels, which is attributed to the gel-to-sol transition. The endo-
thermic peak temperature Tm in a heating DSC curve is higher than the exo-
thermic peak temperature Ts observed in a cooling DSC curve. Both Tm and
Ts shifted to higher temperatures with increasing concentration of added
sugars, and Tm was always higher than Ts. This has been explained by a
zipper model approach[6J.
The differences ~ Tm :: Tm - Tmo and ~ Ts :: Ts - Ts 0, where Tmo and Ts 0
are the melting temperature Tm and the setting temperature Ts respectively
of agarose and f(, -carrageenan gels without sugars, as funct ions of the
dynamic hydration number nDHN of added sugars, were found to increase line-
arly with increasing nDHN as well as the mean value of the number n(e-OH)
of equatorial OH groups existing in the various conformers of sugar usual-
ly found in solutions of added sugars for f(, -carrageenan gels with the
addi tion of 0.5 or 1. 0 Msugars. The increase in ~ Tm or ~ Ts was in the
following order: raffinose> maltose> sucrose> glucose> galactose>
mannose > fructose> ribose. The increase in Tm with increasing concentra-
tion of added sugars has been attributed to the stabilization of junction
zones in agarose or f(, -carrageenan gels by newly created hydrogen bonds
between hydroxyl groups in the sugar molecules and in the agarose or f(,-
carrageenan molecules.
The elastic modulus of agarose and f(,-carrageenan gels was increased by
the addition of sugars. The elastic modulus of thermoreversible gels was
proposed to be a function of the number N of junction zones, the bonding
energy c, the number n of segments liberated from the junction zone, and
the ceiling number j), which is the upper limit number of segments which
can be liberated from the junction zone before the gel-to-sol transition
110
occurs[7J. The addition of sugars seems to increase the number of junc-
tion zones, which is equivalent to the number of zippers in a zipper model
approach. The endothermic peak accompanying gel-to-sol transition became
larger with increasing concentration of these sugars. Since the DSC peak
height is mainly determined by the number of zippers ~ this experimental
result is explained by the increase of N by the addition of sugars. The
shift of Tm or Ts to higher temperatures suggests that the structure of
agarose and kappa-carrageenan gels has become thermally stable. This may
be mainly due to the decrease of the rotational freedom of parallel links
constituting a zipper by the addition of sugars.
As was discussed recently, the number of zippers N increases, the rota-
tional freedom of parallel links G decreases and the number N of parallel
links decreases with the addition of sucrose to agarose gels [8J. Other
sugars may have a similar influence also on agarose and tc -carrageenan gels.
The rotational freedom of parallel links G in agarose and tc -carrageenan
gels will decrease with increasing nDHN or n(e-OH) of added sugars and, as
a result, the rotational motion of parallel links in a zipper is impeded.
Therefore, Tm is shifted to higher temperatures with increasing nDHN or
n(e-OH) of added sugars.
However, the excessive addition of sugars decreases the elastic modulus
of agarose and tc -carrageenan gels [2,4J. The excessive sugar seems to
decrease G, hence Tm or Ts is shifted to higher temperatures. On the other
hand, the excessive sugar cannot increase N probably because this could
reorganize the structure of the agarose and tc -carrageenan gels; it might
promote the aggregation of zippers. This explains why the endothermic peak
in the heating DSC curves for agarose and tc -carrageenan gels became
smaller with the excessive addition of sucrose.
It is not possible to conclude which factor is more decisive for the
conformational change of agarose and tc-carrageenan molecules; the struc-
ture of water as a solvent is changed by the addition of sugars and this
may then change the conformation of agarose and tc -carrageenan. The other
possibility is the direct interaction between the hydroxyl groups of sugars
and those of agarose and tc -carrageenan molecules. Structural studies
which may shed light upon this point are urgently required.

REFERENCES
1. Clark,A.H. and Ross-Murphy, S.B., Adv. Polym.Sci., 1987, 83, 57-192.
2. Nishinari,K., Watase,M., Williams, P. A and Phillips,G.O., 1.Agric.Food
Chern., 1990, 38, 1188-1193.
3. Nishinari,K. and Watase,M., Thermochim.Acta, 1992, 206, 149-192.
4. Watase,M., Nishinari,K., Williams, P. A. and Phillips, G. O. , 1.Agric.Food
Chern., 1990, 38, 1181-1187.
5. Watase,M., Kohyama,K. and Nishinari,K., Thermochim. Acta, 1992, 206, 163-
173.
6. Nishinari,K., Koide,S., Williams,P.A. and Phillips,G.O., 1.Phys.France,
1990, 51. 1759-1768.
7. Nishinari,K., Koide,S. and Ogino,K., 1.Phys.France, 1985, 46, 793-797.
8. Nishinari,K., Watase,M., Kohyama,K., Nishinari, N., Oakenfull, D., Koide,
S., Ogino,L, Williams, P. A. and Phillips,~O., Polym. 1., 1992, 2t 871-877.
 RHEOLOGICAL CHANGES DURING SOL-GEL TRANSITION OF PECTIN-
SUCROSE/FRUcrOSE DISPERSIONS

M. A. RAO, J. P. VAN BUREN, H. J. COOLEY

Cornell University-Geneva, Geneva, NY 14456-0462, USA.

ABSTRACT

Frequency sweeps over 20-40°C revealed weak-gel behavior of the pectin gels. At 20°C,
during sol-gel transition, both G' and G" increased with time, rapidly in the initial stages
and slowly later. Pectin gels made with sucrose showed higher G' than those made with
fructose. Structure development rates during 24 h were higher initially.

INTRODUCTION

The stability of junction zones in high-methoxyl (HM) pectin/sugar gels is influenced by

hydrogen bonds and hydrophobic interactions. They in turn are influenced by the chemical
nature of sugar and other constituents [1]. At the molecular level, pectin gels may be
considered to be homogeneous and may be described as "association networks" as opposed
to particulate nature of many denatured protein gels [2]. Dynamic rheological (DR) tests
provide quantitative measures of the elastic (storage modulus, G') and the viscous (loss
modulus, Gil) characteristics of materials; the complex viscosity (11 *) also can be
determined, 11*=0*/(0, where G*=~(G')2 + (G")2 and (0 is angular frequency. Therefore,
they are well suited for following the network development during gelation. Earlier, we [3]
determined gel state in HM pectin-fructose dispersions using the criterion of Winter and
Chambon [4] and structure development (SD) rates [5]. The objectives of this work were
to determine the role of sucrose and fructose on DR properties of HM pectin-sugar gels,
and on SD rate.

MA TERIALS AND METHODS

HM pectin-sugar dispersions were made in a round bottom flask with a reflux condenser
and a magnetic stirrer. Crystalline NaCI (AR grade) was dissolved in sodium citrate buffer
112
(0.05 M, 3.0 pH) under agitation followed by sugar at room temperature; NaCl was added
to control the ionic strength. The mixture was heated to 100±1°C and the pectin (Genu
Pectin, type BB, rapid set 150°, Hercules Inc., Newark, DE) was added slowly and held
under agitation for 8 min. Dispersions of pH 2.68 and 3.0 were made. DR properties were
measured with a cone (4 cm dia, 2°) and plate system of a Carri-Med CSL 100 rheometer
(Valley View, OH) and Carri-Med 50 software. A solvent trap was placed on top of the
cone to minimize moisture loss [3). SD rates in gels at 20°C, after cooling a hot dispersion
from 60°C to 20°C, were calculated as described earlier [5).
RESULTS AND DISCUSSION

Frequency sweep data obtained at 40, 30 and 20°C as the pectin-sugar dispersion was
cooled revealed that sucrose gels were firmer (higher 0') than fructose gels; magnitudes of
G" were not affected as much as those of G' (Figure 1). The difference in the gel strength
of the two gels is attributed to hydrophobic effects [1). Because of the strong dependence
of G' and G" on frequency (00), all the dispersions and gels exhibited behavior common to
weak gels and biopolymer solutions [2). Gels containing both sugars showed intermediate
values of G' and G" (not shown).

5
•• D
• !.!~~-
D ...

4 • • • • ••• ••
• •D D D ....
D
-..
cu • • •
D
D

• • • o
D 0f0I0
D 00 00
~
• D
• 00

•
00
D 0

~ 3 • 0
0
0

-- 2
~ 0
0

C
0 CI G' Sucrose
~ 0
• Gil Sucrose
0 G' Fructose
1
• Gil Fructose

0
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Ln (Frequency, rad/s)

Figure 1. Ln (G', G") vs. In 00 data on 60% fructose/sucrose, 1% pectin gels at 20°C
 113
At 20°C, during sol-gel transition, both G' and Gil increased with time. SD rates
during the ftrst 24 h of structure development were calculated using complex viscosity, 11 *
(poise), (Figure 2). The SD rates in the gels were high initially and decreased with time;
they were not signiftcantly different. Both positive and negative SD rates were calculated;
the latter were attributed to the weak gel nature and breakage of the bonds in the tenuous
gel network [5].

0.4
o Fructose 60 %, pectin 1 %
0.3 • Sucrose 60%, pectin 1 %

0
..- 0.2 •
....c •• • •
e 0.1 0·
....
0
~!;f.)

0 o·
• 0 Ii .0
00 ! .0

-
C- D a a 0
'-'
0.0 • a
•
QJ
0
CU 0 0 a
~
-0.1 a •• •
•
•
0

-0.2- •
-0.3 I

0 400 800 1200 1600

Time (min)
Figure 2. Rate of structure development in 60% sucrose/fructose gels at 20°C

REFERENCES

1. Oakenfull, D. G. The chemistry of high-methoxyl pectins. In The ChemistI)' and

Technology of Pectins. ed., R. H. Walter, Academic Press, New York, 1991, pp. 87-
108.
2. Doublier, J. L., Launay, B., and Cuvelier, G. Viscoelastic properties of food gels. In
Viscoelastic Properties of Food. ed. M. A. Rao and J. F. Steffe, Elsevier Applied
Science, London, 1992, pp. 371-434.
3. Rao, M. A., Van Buren, J. P., and Cooley, H. J. Rheological changes during gelation
of high-methoxyl pectin/fructose dispersions: Effect of temperature and aging. L
Food Sci., 1993, 58, 173-176,185.
4. Winter, H. H., and Chambon, F. Analysis of linear viscoelasticity of a crosslinking
polymer at the gel point. J. Rheol., 1986,30, 367-382.
5. Rao, M. A., and Cooley, H. J. Dynamic rheological measurement of structure
development in high-methoxyl pectin/fructose gels. J. Food Sci., 1993,58, (In
Press).
 CRITICAL BEHAVIOR OF AGAROSE NEAR THE SOL-GEL
TRANSITION POINT

TOMOYUKI FUm, HITOSHI KUMAGAI,

AND TOSHIMASA YANO
Department of Agricultural Chemistry, The University of Tokyo
1-1-1 Yayoi, Bunkyo-ku, Tokyo 113, JAPAN

ABSTRACT
It was possible to determine the correlation length of agarose gel by using the dynamic light
scattering method. The correlation length showed a tendency to diverge when the
concentration decreased and approached the sol-gel transition point. The log of the correlation
length was a linear function of the log of deviation in concentration from the sol-gel transition
point. The correlation length was described by scaling law, and the scaling law between the
critical exponent of the correlation length and that of the elastic shear modulus was valid.
From the result, it was suggested that by measuring the correlation length, scaling law was
shown to be effective when observing the physical property changes near the sol-gel transition
point.

INTRODUCTION
Recently it has been noted that sol-gel transition is analogous to critical phenomena [1], that
is, the percolation theory may be applicable to the sol-gel transition. In this connection, the
elastic shear modulus G' of gel and its exponent t (G' oc e f) have been measured at near the
sol-gel transition point [2]. However, not only the macroscopic property of a sol-gel
transition system, such as viscosity or elastic modulus, but also the microscopic property,
such as the molecular weight distribution of clusters or the correlation length that may be used
as a parameter to describe effectively the microstructure of gel network, should be measured
to learn whether the sol-gel transition is described by the percolation theory.
In this paper, the network mesh size in gel network, that is, the correlation length was
attempted to be estimated by using the dynamic light scattering (DLS) method, and the sol-gel
transition behavior was discussed from the viewpoint of scaling concept.

MATERIALS AND METHODS

Material
Agarose (Nacalai Tesque, Tokyo) was dissolved in distilled water filtered through a Millipore
filter of 0.025 !lm pore diameter.
 115
Dynamic light scattering (DLS) method [1,3]
The normalized autocorrelation function of the scattering intensity from gel network, gl(t),
can be written with a distribution function, as following:

g'(") = L- G(l) exp(-n) dr

(r7= [ fG(l}df

(r) = DcoopQ2
Q = 41tn sin ft
A 2
The correlation length in gel network, ~, is given by

~ = kBT
61tTJDcoop
The scattered light intensity from a sample was measured with the Malvern 4700 photo!'
correlation spectroscopy system (Malvern, UK), utilizing a 30 mW, 633 nm He-Ne laser
(NEC, Tokyo). The scattering angle was fixed to 90°.

RESULTS AND DISCUSSION

By using the DLS method it was possible to estimate the correlation length of gel network of
agarose that forms cross-links with non-covalent bonds. The correlation length showed a
tendency to diverge when the concentration decreased and approached the sol-gel transition
point. Therefore, it was suggested that the correlation length estimated was effective as a
parameter for the microstructure of gel network. Fig.l shows the effect of the concentration
on the correlation length of agarose gel at 27.0°C. The log of correlation length was a linear
function of the log of deviation in concentration from the sol-gel transition point. It was found
that the correlation length was described by scaling law from this result:

(1)

where v is the critical exponent of the correlation length. The sol-gel transition point and the
critical exponent were determined to be 0.071 wt% and 1.53 at the temperature of 27.0°C,
respectivel y.
Fig. 2 shows the effect of the concentration on the shear modulus of agarose gel at
27.0°C. The log of shear modulus was also a linear function of the log of deviation in
concentration from the sol-gel transition point. It was possible to fit the data with scaling law:
G'oc (C - Cg)t (2)
where t is the critical exponent of the shear modulus. The sol-gel transition point and the
critical exponent were determined to be 0.046 wt% and 2.59 at the temperature of 27.0°C,
respectively.
De Gennes [4] derived the scaling law between the critical exponent of the correlation
length and that of the elastic shear modulus:
t = 1 + v (d - 2) (3)
116

where d is spatial dimension. Therefore, the critical exponent t was predicted from the result
of Fig.l. The experimental value of t in Fig.2 was close to the predicted value. It was found
that the scaling law between the critical exponent of the correlation length and that of the elastic
shear modulus was valid.

10000 1000
E
oS

.r:.
C,
1000 "',€ 1 00
c: ~
J!1
c:
o
~ 100 (!) 10
~
....
o
()
ISlope = -1.531 1Slope = 2.591
56 23456 5 6 7 8 9 2 3
0.1 0.1
C - Cg [wt.%] Cg - C [wt.%]

Figure 1. Effect of the concentration on the Figure 2. Effect of the concentration on the
correlation length of agarose gel at 27.0 dc. shear modulus of agarose gel at 27.0 DC.

CONCLUSIONS
In this study, the correlation length of agarose gel could be measured by using the DLS
method. The correlation length was suggested to be effective as a parameter for the
microstructure of gel network, and was described by scaling law. The scaling law between the
critical exponent of the correlation length and that of the elastic shear modulus was valid with
the concentration at a constant value. By measuring the correlation length, scaling law was
shown to be effective when observing the physical property changes near the sol-gel transition
point.

NOMENCLATURE
d, spatial dimension [I]; Dcoop , cooperative diffusion coefficient [nm2 /s]; G', elastic shear modulus [pa];
G(I), normalized distribution function of relaxation rate [s]; gl(t), normalized autocorrelation function of the
scattering intensity [I]; kB, Boltzmann's constant [JIK]; n, refractive index of the solution [1]; Q, scattering
vector [nm- I ]; T, absolute temperature [K]; t, critical exponent of G' [I]; t, relaxation rate [s-I]; e, deviation
from the sol-gel transition point [I]; 11, solvent viscosity [pa s]; e, scattering angle [rad]; A, wavelength of the
incident light [nm]; ~, correlation length [nm]; v, critical exponent of ~ [1]

REFERENCES
1. de Gennes, P.G., Scaling concepts in polymer physics, Cornell Univ. Press, Ithaca, NY,
1979.
2. Tokita, M. and Hikichi, K., Mechanical studies of sol-gel transition: Universal behavior of
elastic modulus. Phys. Rev., 1987, A35, 4329.
3. Chu, B., Laser light scattering, Academic Press, San Diego, CA, 1991.
4. de Gennes, P.G., On a relation between percolation theory and the elasticity of gels. J.
Physique, 1976,37, L-1.
 MICROSTRUCTURE OF HYDROGEL AND ITS RELATIONSHIP WITH
MOLECULAR-SIEVE CHARACTERISTICS IN
GEL ELECTROPHORESIS

SUSUMU MIURA, TOMOYUKI FUnI, lllTOSHI KUMAGAI,

AND TOSHIMASA YANO
Department of Agricultural Chemistry, The University of Tokyo
1-1-1 Yayoi, Bunkyo-ku, Tokyo 113, JAPAN

ABSTRACT

Gel electrophoresis of DNA fragments was carried out to study the relationship between the
microstructure of hydrogel and molecular-sieve characteristics in gel electrophoresis. To
analyze the behaviour of DNA during electrophoresis, two parameters were introduced;
normalized mobility, defined as the mobility of DNA in gel divided by the free mobility of
DNA, and normalized diameter, defined as the diameter of DNA fragment divided by the
correlation length of gel determined from light scattering measurement. The double logarithmic
plot of the normalized mobility in gel matrices with different concentrations against the
normalized diameter of DNA fragment was expressed by a single curve for DNA fragments
with different size. The present result suggests that the mobility of molecules in gel
electrophoresis can be predicted by using the correlation length.

INTRODUCTION

Recent advances in the field of biochemistry has seen extensive usage of hydrogels as
separating media in such technology as gel electrophoresis for separating molecules like DNA
and proteins. The key factor governing the separation of these molecules seems to be their
diffusion within the gel network, or the microstructure of hydrogel. However, the correlation
between the microstructure of hydrogel and the molecular diffusion within has not yet been
extensively studied. This is thought to be due to the lack of knowledge and methods for
evaluating the microstructure of hydrogel. Recently, it has been suggested that correlation
length, which is known to represent the mesh-size of network, may be used as a parameter to
describe effectively the microstructure of hydrogel [1].
In this paper, we have devised a new method to analyze the relationship between the
mobility of polymer chains (DNA in this case) and gel network, and carried out agarose gel
electrophoresis of DNA to discuss the applicability of the method.

MATERIALS AND METHODS

118
Analysis of DNA gel electrophoresis
a) For a given DNA fragment, its mobility ~ was obtained by dividing the migrating distance
by the length of time and the strength of electric field during electrophoresis.
b) The free mobility 110, which is defined as the mobility of the fragment in free solution, was
determined by extrapolating the Ferguson plot [2] to the gel concentration of 0 wt%.
c) The radius of gyration Rf of the fragment was calculated using the following equation:

(1)

Where p is the persistence length and L is the contour length of the DNA fragment. Eq. (1)
was obtained by Benoit and Doty [3] by assuming a worm-like coil model for molecular
chains.
d) The mesh-size, or the distance between the two neighbouring cross-links within the gel
network was measured as the correlation length ~ by using the dynamic light scattering method
[4].
e) To analyse the mobility of DNA fragments within the gel network during electrophoresis,
two parameters were introduced; normalized mobility (w'I1O), defined as the mobility ~ of each
DNA fragment in gel divided by its free mobility 110, and normalized diameter (d/~), defined as
the diameter d of DNA fragment divided by the correlation length ~. By taking the double
logarithmic plot of the above two normalized values, the relationship between the mobility of
molecules and the microstructure of gel network around them was studied.

Gel electrophoresis of DNA

Two different kinds of agarose (Agarose ME and LE) were purchased from Nacalai Tesque
Inc. The two agarose samples contain different ratios of negatively charged sulphate groups
bound to their polysaccharide backbones (ME:0.5%, LE:O.4%), and the higher rate of negative
charges in agarose ME results in stronger electro-osmosis flow during electrophoresis.
Agarose powder was added to distilled water, and was dissolved by heating and
stirring the mixture at over 90°C. After agarose has dissolved completely, the solution was
poured into a gel tray for electrophoresis, and was allowed to cool down to 27°C. The gel tray
containing the mixture was kept at 27°C for 24 hours, for the gelation to complete.
The gel tray containing the hardened agarose gel was set into the electrophoresis
apparatus purchased from Takara Shuzo (Takara HE-13), and enough TAE electrophoresis
buffer (Tris-Acetate, EDTA, pH 8.0) was poured into the apparatus to cover the gel. Then the
sample of HindIII digested A.-DNA was loaded in the well placed along the top side of the
agarose gel, and the electrophoresis was carried out at 27°C. The strength of electric field
during electrophoresis was 0.61 V/cm, and the electrophoresis was run for 18 hours at gel
concentrations varying from 0.3 to 1.0 wt%.

RESULTS AND DISCUSSION

Figure shows the double logarithmic plots of the normalized mobility against the normalized
diameter. As shown in the Figure, the plot was expressed by a single curve that could be
divided into two regions at the (d/~) value of approximately 1. The fact that the plot could be
expressed by a single curve suggests that the mobility ~ of polymers during gel electrophoresis
can be predicted by using the correlation length ~ to evaluate the network structure of hydrogel.
Moreover, the fact that the curve could be divided into two parts at the (d/~) value of 1
suggests that the relative size of the molecules and the mesh-size of hydrogel plays an
important role in the separation mechanisms of gel electrophoresis; i.e. at (d/~» 1, the
interaction between gel network and molecules within is comparatively larger than that at
(M)<1. Therefore much more effective separation can be expected at (dI~» 1.
The double logarithmic plots of the normalized mobility against the normalized diameter
obtained for Agarose LE was nearly identical to that obtained for Agarose ME (data not
 119
shown). This result shows that our method for correlating the behaviour of molecules and the
microstructure of hydrogel is applicable for agarose gels in general.

>-
I-
::i
6
5
• •• -. ,..ot-
1Xi 4 0 23130bp
0
::2: 3 I:::. 9416bp
Cl 0 6557bp
•
W 2
N
::i
4361bp
« & 2322bp o
::2: 0.1
c:
0
z
• 564bp o

2 3 4 5 6 2 3 4 5 6 2
0.1 1 10
NORMALIZED DIAMETER (1)

Figure. Double logarithmic plots of the normalized mobility of DNA fragments against their
normalized diameter in agarose gel electrophoresis.

CONCLUSIONS
In this study, we have shown that the mobility I! of molecules during gel electrophoresis could
be predicted by using correlation length ~ to represent the microstructure of hydrogel. Also, we
have shown that the relative size of molecules against the mesh-size of hydrogel plays an
important role in the separation mechanisms of gel electrophoresis; at (d/~» 1, more effective
separation could be expected compared to that at (~)<1.
By carrying out further studies, it may become possible to design hydrogels that have
the best correlation length ~ to molecular diameter d combination for the most effective
separation of given polymers.

REFERENCES
1. de Gennes, P.G., Scaling concepts in polymer physics, Cornell Univ. Press, New York,
1979.
2. Ferguson, K.A., Starch-gel electrophoresis - Application to the classification of pituitary
proteins and polypeptides. Metabolism, 1964,13,985.
3. Benoit, H. and Doty, P., Light scattering from non-Gaussian chains. J. Phys. Chern,
1953, 57, 958.
4. Oikawa, H. and Murakami, K., Dynamic light scattering of swollen rubber vulcanizates
and the swelling mechanism. Macromolecules, 1991,24,1117.
 PHYSICOCHEMICAL STUDIES ON GElATION OF SOYBEAN 7S AND llS
PROTEINS BY GLUCONO-o-lACTONE

KAORU KOHYAMA· and KATSUYOSHI NISHINARI

·National Food Research Institute, Kannondai, Tsukuba, Ibaraki 305, Japan,
and Department of Food and Nutrition, Faculty of Science of Living, Osaka City University,
Sumiyoshi, Osaka 558, Japan

ABSTRACT

Dynamic viscoelasticity measurements were carried out to examine the gelation process of
soybean proteins, storage and loss moduli being observed as a function of time after adding
glucono-c5-lactone (GDL). Both storage and loss moduli at the final stage of the gelation for
7S-GDL gels showed almost the same value as those of llS gels. However, breaking stress
and breaking strain for 7S after the completion of gelling were smaller than those for llS
gels. The rate of gelation for 7S was much slower and the gelation time was longer than for
llS.

INTRODUCTION

More than 50 % of soybean yield in Japan has been used for tofu (soybean curd) making.
Recently, glucono-c5-lactone (GDL) became a more popular coagulant for tofu-processing.
Soybean proteins contain two major globulins, 7S and llS, which show different
thermal transition temperatures and gel forming properties. The gelation of 7S and llS
proteins in the presence of GDL is investigated by dynamic viscoelastic measurements and
compression tests.

MATERIALS AND METHODS

Soybean 7S and llS proteins were isolated by the method of Nagano et al. (1) and freeze-
dried. The protein contents of these powders were 95 % for llS and 92 % for 7S globulin
determined by Kjeldahl method with a conversion factor of 6.25.
A 7S or an llS protein solution (2.0 mL) was heated in boiling water for 10 min and
then cooled to 60°C. Freshly prepared GDL solution (0.05 mL) was added to the protein
solution. The storage (G') and loss (Gil) moduli at 60°C were recorded as a function of time
 121
using a Rheolograph Sol (Toyoseiki Seisakusho, Tokyo). The details of this method were
described elsewhere (2, 3).
The protein solution preheated at 100°C for 10 min was kept in 60°C bath. GDL
solution was added to the solution. The mixture was immediately poured into molds with 16
mm diameter and 10 mm height. They were allowed to stand for various periods at 60 0c.
Compression tests for gels were carried out using a Rheoner RE-33005 (Yamaden Co. Ud.,
Tokyo). A cylindrical gel sample on the stage was vertically compressed until rupture with a
flat plunger of 40-mm diameter at 27°C at the compression rate of 1.0 mm/s. Breaking
stress and breaking strain of gels were determined.

RESULTS AND DISCUSSION

Saturated Shear Modulus

Both storage and loss moduli began to rise at a certain time (gelation time) after the addition
of GDL (t=O). The gelation curves fitted well with first-order reaction kinetics as follows.

G(t)=Gsat[l-exp{ -k(t-tJ}]

where Gsat is the saturated value of the storage or loss modulus, k is the rate constant of gela-
tion, to is the gelation time, and t is time.
The saturated storage modulus (G'sat) for 4 % 7S with 0.4 % GDL was slightly larger
than that for 4 % 11S, while the saturated loss modulus (G" sat ) for the 7S system was almost
the same to that of 11S. Therefore, tano of completely gelled 7S (G"sa/G'sat) became a little
smaller than that of 11S.
The shear modulus of polymer gels is often described by an exponential function of
polymer concentration Cpo In many cases, the exponent becomes 2 (4). The relationships
between G'sat and protein concentration are examined. The exponent for 7S gels with 0.4 %
GDL was 2.43 and that for 7S gels with Cp:Cg = 10:1, where Cg is the concentration of GDL,
was 1.89, which was smaller than the former. Exponent values for 11S gels with 0.4 % GDL
and 11S gels with Cp:Cg =10:1 became 3.52 and 3.19, respectively. In the latter case, the
exponent value was again smaller than in former one. The protein concentration dependence
of G'sat for both 7S and 11S was more pronounced in gels with 0.4 % GDL than in those with
a fixed ratio of GDL to protein. This fact suggests that GDL decreases the saturated modulus
slightly. Since the exponent values for 7S were smaller than those for 11S, the saturated
storage modulus of 7S gels was larger than that for 11S at lower C p than 4-5 %, which is
typical protein concentration for Kinugoshi tofu (5), and the inverse relationship was ob-
served at higher Cpo
122
Gelation Rate
A large difference in the rate of gelation was observed between 7S and 11S. The gelation
time to of 7S was much longer than that of 11S at the same protein concentration Cp' GDL
concentration Cg, and temperature. Gelation rate of 7S was also slower than that of 11S;
therefore, it took a long time before storage and loss moduli of 7S reached equilibrium
values.
Gelation proceeded faster in systems of higher GDL concentrations, and a similar
tendency was observed for 11S systems. However, the value of k for 11S was 2-3 times
larger than that for 7S. Gelation rate decreased with increasing protein concentration for both
7S and 11S systems. A slower gelation rate was observed again for 7S than for 11S in the
fixed GDL concentration systems. The gelation time to decreased with increasing GDL
concentration at a constant protein concentration (4.0 %).

Comparison between Saturated Storage Modulus and Breaking Stress

The breaking stress and the breaking strain for 7S and 11S gels increased with time and then
seemed to level off. However, the breaking strain for 7S gels was slightly increasing even
after heating for 240 min at 60 cC. Faster gelation in 11S than in 7S was clearly observed
again. Therefore, the time required to saturate mechanical properties for 7S gels became
longer from several to 10 times than that for 11S. When enough time had passed and me-
chanical properties were almost saturated, 11S gels showed larger breaking stress and strain
than 7S gels. Breaking stress of 11S gels after heating for 60 min was much larger than that
of 7S after heating for 240 min. Stress values of both gels were almost the same at small
strain, but 11S gels showed larger stress than 7S when a large strain was applied. 7S gels
was more brittle than 11S gels, even though 7S gels had a Young's modulus similar to that of
the 11S gels as observed in the dynamic viscoelasticity measurement.

REFERENCES
1. Nagano, T., Hirotsuka, M, Mori, H., Kohyama, K. and Nishinari, K., Dynamic viscoelastic
study on the gelation of 7S globulin from soybeans. J. Agric. Food Chem., 1992, 40, 941-4.

2. Yoshida, M., Kohyama, K. and Nishinari, K., Gelation properties of soymilk and soybean
11S globulin from Japanese-grown soybeans. Biosci. Biotech. Biochem., 1992,56,725-8.

3. Kohyama, K. and Nishinari, K. Rheological studies on the gelation of soybean 7S and 11S
proteins in the presence of glucono-o-Iactone. J. Agric. Food Chem., 1993,41,8-14.
4. Fukada, E. and Kaibara, M., The dynamic rigidity of fibrin gels. Biorheology, 1973,10,
129-38.

5. Resources Council, Science and Technology Agency. Standard Tables of Food Composi-
tion in Japan, 4th revised ed., Science and Technology Agency, Tokyo, 1982, pp.98-9.
 RHEOLOGICAL MODELS FOR SHELLED MAIZE EN-MASSE

LAWRENCE o. GUMBE
Department of Agricultural Engineering
university of Nairobi, Kenya.

ABSTRACT

The mechanized handling, processing and storage of grain is an

indispensable part of modern agriculture. A knowledge of the
rheological behaviour of en-masse grains under static and dynamic
loads is a crucial step in the characterization of the mechanical
damage, design of storage structures and processing machinery.

This paper review various rheological models for shelled maize

en-masse and presents experimental results based on elastic,
viscoplastic and elastoplastic models.

INTRODUCTION

Rheology has been defined as "a science devoted to the study of

deformation and flow". Therefore, when the action of forces
result in deformation and flow of the material, the mechanical
properties will be referred to as rheological properties (1).

Rheological models are usually presented as stress-strain

equations. The modelling of systems of bulk grain, such as
shelled maize en-masse, usually involves the idealization of the
grain as a continuum, even though it is composed of discontinuous
kernels. Some aspects of constitutive behaviour are common to
all particulate media en-masse, e.g, the importance of frictional
and/or cohesive bonding as particle contracts and the dilatancy
of densely packed particulate materials (2). For this reason
theories and constitutive equations developed for cohesionless
soils have been applied to grain.

This paper
presents summary of models based on elastic,
v~scoplastic, and elastoplastic models that have been applied or
may apply to shelled maize en-masse.

Models

In the application of models, the material is normally idealized

as an isotropic, homogeneous continuum.
124
Elastic Models

Gumbe and Kuria (3) have evaluated elastic parameters for shelled
maize en-masse as outlined by Hardin et al (2). Elastic moduli
were determined using triaxial tests. The model used was:
(1)

Where {dee} is the elastic increment vector, [C) is the elastic

constitutive matrix and ida'} is the effective stress increment
vector. The experiments were conducted at 13% (wb) moisture
content, bulk density 725 kg/m 3 , void ratio 0.63 and ambient
temperature 23 degree C. Young's modulus (E) and Bulk modulus
(k e ) were evaluated, the results are in Table 1.

Table 1: Elastic Moduli for Shelled Maize

Young's Modulus (E) Bulk Modulus (K)
Confining Isotropic
Pressure (kPa) E x 10 6 Pa Stress (kPa) Ke x 10 6 Pa
98.1 0.96 0 0
196.2 1.08 98.1 5.70
294.3 1. 53 147.2 7.32
196.2 7.32
245.3 6.86

Parameters such as shear modulus, G, can be evaluated from k e and

E. The preceding parameters can be used to determine material
behaviour in the elastic range. If other parameters such as
plastic, structural and volumetric ones are evaluated, they can
be used in elastoviscoplastic models (4).

Viscoelastic Models

viscoelastic models, both linear and non-linear, have been

applied to biological materials. Bock et al (5) used the
triaxial test to evaluate data for the characterization of the
relaxation behaviour of en-masse wheat. A similar procedure
could be applied to shelled maize en-masse. Balastreire (7)
evaluated the temperature and moisture shift functions for horny
corn (maize), he obtained an expression for relaxation modulus
E(t) as:
E( t) =b +mlogt (MPa) (2)

Where t is the time in second, and band m constants which are

1619.14 and -281.19 at 30° C respectively.

Elastoplatic Models

Elastoplatic models have been used to model the behaviour of en-

masse wheat successfully (6). Gumbe and Salano (8) have also
applied it to sorghum, it may be a promising model for the
behaviour of shelled maize en-masse.
 125
REFERENCES

1. Mohsenin, N. N., Physical Properties of Plant and Animal

Materials. Gordon Breach Science Pub. 1986, pp. 89

2. Hardin, B. O. et al. Triaxial Compression. Simple Shear.

and Strength of Wheat En-Mass. Trans of ASAE 33 (3), pp 933
- 943.

3. Gumbe, L. O. and Kuria, M., Elastic Parameters for Shelled

Maize En-Masse. African Journal of Science and Tech. Paper
under publication.

4. Li, Y et al. Elastic-Viscoplastic Cvclic Model Parameter

Determination and Evaluation for Wheat En-Masse. Trans of
ASAE Paper 1987. No. 1984-1995

5. Bock et al. Modelling Stress Relaxation of Wheat En-Masse

Using the Triaxial Test. ASAE Paper No. 87 - 4525.

6. Zhang, Q., et al. Determination of Elastoplastic

Constitutive Parameter for Wheat En-Masse. Trans of ASAE
1986. 29(6), pp. 1739-1746.

7. Balastreire, L. A. Relaxation Modulus for Corn Endosperm in

Binding. Transactions of ASAE 21(4), pp 767 - 772.

8. Gumbe, L. O. and Salano, P. A. Elastoplastic Constitutive

Parameters for Sorghum En-Masse. 1993. Unpublished
manuscript. Department of Agricultural Engineering.
University of Nairobi.
 NONDESTRUCTIVE TEXTURE MEASUREMENT OF APPLES

STANLEY PRUSSIA
Professor /Biological and Agricultural Engineering Department
University of Georgia
Georgia Station, Griffin, Georgia 30223 USA

KAZUO MORITA
Associate Professor/Agricultural Systems Engineering Laboratory
Faculty of Agriculture, Kagoshima University
21-24 Korimoto I, Kagoshima 890 JAPAN

YEN-CON HUNG
Associate Professor/Food Science and Technology Department
University of Georgia
Georgia Station, Griffin, Georgia 30223 USA

CHI THAI
Associate Professor/Biological and Agricultural Engineering Department
University of Georgia
Georgia Station, Griffin, Georgia 30223 USA

WILLIAM TOLLNER
Professor /Biological and Agricultural Engineering Department
University of Georgia
Driftmier Engineering Center, Athens, Georgia 30602 USA

ABSTRACT

Postharvest market decisions for apples could be improved if nondestructive tests were
available for accurately predicting eating texture which is the ultimate test. Texture is
commonly predicted by human touch and color evaluation and by an instrument to
measure penetration force. In this study the nondestructive methods of CT-Scan X-Ray
and coefficient of restitution along with eight other quality measures were correlated
to eating texture and to each other. The best correlations to eating texture were for the
puncture test without skin (0.70) followed by firmness to the touch (0.54). Poor
 127
correlations were obtained for the new methods (CT-Scan = 0.31, coefficient of
restitution = 0.028). The highest overall correlation was between instrument and
sensory hue angle (0.87), however neither correlated well with eating texture (0.54 and
0.47). The results indicate that more research is needed.

INTRODUCTION

The ability to predict eating texture of apples and other fruit would help with decisions
involving final purchase, retail display, packinghouse sorting, storage selection, and
harvest date. Existing tests by humans are touch, color evaluations, and eating texture.
Instrument color measurements are cornman for laboratory experiments. Especially
needed is a nondestructive method which would allow all items to be evaluated rather
than only a sample as required for the standard Magness-Taylor penetrometer test.
Some possible alternatives are coefficient of restitution, nuclear magnetic resonance,
and CT-Scan X-Ray. The purpose of this study was to compare the suggested new
methods with existing methods and with other factors related to fruit quality.

MATERIALS AND METHODS

Differences in the texture of 'Golden Delicious' apples (78 fruit) were caused by storing
them for 0, 7, 14, and 21 days at 1, 10, 18, and 27 0 C. Nondestructive tests were mass,
color (sensory and Hunter L, a, b), firmness to the touch (expert judges), coefficient of
restitution (bounce), and CT-Scan X-Ray [1] (water distribution, only on days 7 and 14).
The bounce test was a drop test onto a force plate with three load cells. Destructive
tests were eating texture, soluble solids, moisture content, puncture force (with and
without skin), nuclear magnetic resonance [1] (NMR, data from free induction decay
and pulse echo). Nondestructive tests were completed on each test day on all
remaining fruit followed by destructive tests on a preselected sample. Pearson
correlation coefficients were found for all combinations of test methods.

RESULTS

Future publications will present the results of analyses of test data over time. This
report focuses on correlations of the texture data regardless of the date of the test.
The NMR results did not show any differences among the samples. Tables 1 and 2
show correlations with special interest.
128
TABLE 1
Correlations of selected measurements to eating texture

r MEASUREMENT

0.70 PUNCTURE FORCE WITHOUT SKIN

0.54 FIRMNESS TO THE TOUCH
0.54 HUE ANGLE FROM INSTRUMENT
0.47 HUE ANGLE FROM SENSORY
0.31 CT-SCAN X-RAY (WATER DISTRIBUTION)
0.12 SOLUBLE SOLIDS
0.028 COEFFICIENT OF RESTITUTION
0.008 MOISTURE CONTENT

TABLE 2
Correlations for selected combinations

r MEASUREMENT 1 VS. MEASUREMENT 2

0.87 HUE ANGLE (INST.) HUE ANGLE (SENSORY)

0.60 PUNCT. FORCE (SKIN) PUN CT. FORCE (NO SKIN)
0.41 HUE ANGLE (INST.) SOLUBLE SOLIDS
0.50 HUE ANGLE (INST.) PUNCTURE FORCE (SKIN)
0.38 HUE ANGLE (INST.) PUNCT. FORCE (NO SKIN)
0.11 COEFFICIENT OF REST. PUNCTURE FORCE (SKIN)

CONCLUSIONS

The new methods were not suitable for measuring apple texture. Firmness to the touch
is not a good predictor of eating texture. Puncture force (no skin) remains the best.
More research is needed for developing a nondestructive texture sensor.

REFERENCES

1. Tollner, W.E, IK. Brecht, and B.L. Upchurch. 1993. Nondestructive

Evaluation: Detection of External and Internal Attributes Frequently Associated
With Quality or Damage. IN Postharvest Handling: A Systems Approach. (R.L.
Shewfelt and S.E. Prussia, Editors). Chapter 2. Academic Press, Orlando, FL.
 DESCRIPTION OF BREAD TEXTURAL DETERIORATION BY STRESS-STRAIN
CURVES, RECOVERABLE WORK, AND THERMAL TRANSITIONS

PAVINEE CHINACHOTI
Department of Food Science
University of Massachusetts
Amherst, MA 01003, U.S.A.

ABSTRACT

Standard white (SWB) and intermediate moisture (MRE) breads are

characterized here in terms of stress-strain curves analysis and
thermal transitions. Observation of changes in these parameters can be
useful for studying the mechanisms of bread staling in relation to
textural deterioration. Glass transition temperature was found to
increase in SWB but not in MRE breads. Relationships with moisture
loss and amylopectins crystallization are discussed.

INTRODUCTION

Textural deterioration of bread (firmness and crumbliness) has been one

of the major factors leading to relatively short shelf-life of most
breads. The molecular mechanisms related to bread staling, although
not well understood, are thought to be closely related to changes in
thermomechanical properties of bread polymers.
Standard white and intermediate moisture (MRE or Meal,
Ready-to-Eat) breads were subjected to this bread staling study.
During storage, compressive stress strain curves and recoverable work
were measured and the rates of textural change were obtained. A
sigmoid compressive stress-strain relationship is a familiar
characteristic of many spongy products. The empirical model and model
constants have been described by Peleg et al [2] to provide a
qualitative expression and even some quantitative changes in the
stress-strain relationships as follows:

C1 (l-exp -(C 2- £H))

(C 3 11)
where ~ is the engineering stress and £H the Hencky's strain.
130

RESULTS

Compressive stress-strain errors were analysed giving C1 , C2 and

C3 . The values C1 was found to increase with hardness as it
reflects the degrees of the magnitude of the applied stress. C1 was
thus found to be increasing most significantly with storage time. C2
was found to increase with time as well but it reflects more of the
cellular structure relating to the curve shape and the shoulder.
However, C3 , the densification strain showed no change at all as
staling progress. Thus, it was concluded that C1 and C2 is more
appropriate to be used as an indicator of staling of bread.
Bread recoverabi1ity was found to decrease significantly with storage
time of standard white bread. In MRE bread, little change in C1 ,
C2 and recoverable work was found over storage of up to 1-3 years at
25°C.
Thermal transitions of SWB and MRE bread have been shown elsewhere
[4]. The storage study of these two breads are shown in Fig. 1 and 2
for SWB and MRE breads, respectively. SWB thermograms by DMA (Fig. 1)
gave a more drastic change in the magnitude of the main transition in
the temperature range between -2.S o C in fresh bread (confirmed by the
DSC results for ice melting) and 2.S o e (in this case no ice melting
was observed) in 19 month old bread. This was accompanied by a drastic
decrease in the tan 6 peak height during staling. There was also an
emergence of another peak approximately at 70 0 e, possibly be due to
amylopectin crystallization. Because there was a significant moisture
loss, it was speculated that the drastic change in the main transition
was due to moisture loss.

E"
FRESH
E' MeWB 0.381

'ii 'ii
tan S e:. c a.
W 1;;
!

1X105 3X10S[
-42.5 15.0 72.5 130.0 o. 2 -100.0
Temp.(e) Heating Temp.(e) Heating

S 9
1.Sx10 O.OS 2x1

E" 9 MONTHS 19 MONTHS

MeWB 0.303 MeWB 0.178
E'

I
I
r tanS
J
3x1 -.,,=",-~"-':-~~"ti"_~_~----::~":
0.02 -100.0 -42.5
Temp.(e) Heating Temp.(e) Heating

Figure 1. DMA Thermograms of SWB breads stored at room temperature.

 131
MRE bread samples stored at room temperature for up to three years
gave thermograms (Fig. 5, Ref. 4) the fresh sample and showed a sharp
transition at a tan 8 peak temperature of -9.7 0 C, with a very small
right shoulder. As the bread aged for three months, the T1 transition
became less prominent (from 0.45 tan 0 for fresh bread to 0.25 tan 0
for three month old bread). However, there were no significant changes
in the main transition Tl or T3 temperature with storage time.
Evidently it was found that there were dramatic differences in
amylopectin crystallization in SWB and MRE breads (Fig. 2).
0.8 r----r-----.---.----r--~

0.7
AMYLOPECTIN CRYSTAlLIZATION
BREADS STORED AT 25 C
b 0.6
ffi
z 0.5
W
"I
ZU STANDARD WHITE BREAD
~~ 0.4
~(/)
z~ 0.3
§s
"- 0.2
~
~ 0.1
'IRE BREAD
o.O&====:::::I===:c:~r=:::J
o 10 15 20 25
STORAGE TIME (MONTHS)

Figure 2 Amylopectin crystallization during storage of SWB and MRE

breads.

According to Slade and Levine [3], a maturation process of the

polymer network occurred during bread staling leading to another glass
transition ("effective Tg") at some higher temperature (max. 60 0 C).
Our data in Fig. 2 indicated no development of such "effective Tg"
transition over time in MRE bread. In addition the sample firmness as
measured by the stress-strain curve was found to increase only slightly
during the first three months and then leveled off for the rest of the
storage period. Over the three-year period, the sample remained
relatively soft and acceptable, unlike staled SWB bread. Extensive
firming in SWB was accompanied by a drastic change in tan 8 peak. This
change occurred concurrently with both moisture loss and amylopectin
crystallization.

REFERENCES

1. Attenburrow, G.E., Goodbrand, R.M., Taylor, L.J. and Liford, P.J.

1989. Structure, mechanics and texture of a food sponge. J. of
Cereal Sci., 9, 61-70.
2. Peleg, M., Roy, R., Campanella, O.H. and Normand, M.D. 1989.
Mathematical characterization of the compressive stress-strain
relationships of spongy bakery products. J. Food Sci. 54:947.
3. Slade, L., Levin, H. 1991. Beyond water activity: Recent advances
based on an alternative approach to the assessment of food quality
and safety. Crit. Rev. Food Sci. Nutr. 30, 115.
4. Hallberg, L.M. and Chinachoti, P. 1992. Dynamic mechanical
analysis for glass transitions in long shelf-life bread. J. Food
Sci. 57:1201.
 FRACTURE STRESS OF FROZEN SOYBEAN CURD

Hisahiko Watanabe, Cun Qi Tang, and Tomoo Mihori

Department of Food Science and Technology
Tokyo University of Fisheries
Konan 4, Minato, Tokyo 108, JAPAN

ABSTRACT

Fracture stress of soybean curd (tofu) was measured in a wide range of low temperature by
compression tests. Fracture stress of frozen soybean curd increased as the temperature
decreased from -20°C. At -65°C, fracture stress of soybean curd reached to that of pure water
ice, and it continued to increase as temperature decreased until it reached a characteristic
temperature (-80°C); below this temperature the fracture stress was constant (35MPa).
Soybean curd with less moisture content, prepared by dehydrating before freezing, showed
larger fracture stress. When a soybean curd was dehydrated from 9.93 g H20/g DM to 1.0 g
H20/g DM, its fracture stress increased up to 2.5 times large. The measured fracture stress
was successfully analyzed using a mathematical model, a two component system consisting
of pure water ice and concentrated amorphous solution (CAS).The estimated value of fracture
stress of CAS was nearly five times larger than that of pure water ice.

INTRODUCTION

Although a number of reports has been published on the mechanical properties of unfrozen
food, only a little attention has been paid concerning mechanical properties of frozen food.
Recently, rapid freezing by spraying cryogen such as liquid nitrogen is applied in fishery
industry. Spraying liquid nitrogen sometimes causes cracks on fish due to thermal stress
expansion generated by non-uniform cooling. Hence an appropriate procedure for spraying
needs to be set up on the basis of mechanical properties of frozen fish. Application of cryo-
mechanical separation method to separating low fat meat from fatty fish [1] also requires the
knowledge of mechanical property of frozen food. In this study, the fracture stress of frozen
soybean curd was measured in a range between -20°C and -196°C. The measured fracture
stress was analyzed with a model in which frozen soybean curd was assumed to consist of
pure water ice dispersed in concentrated amorphous solution (CAS).

MATERIALS AND METHOD

 133
Sample
Soybean curd (tofu) as was purchased in the market (9.93 g H20/g DM) and dehydrated to
the moisture content of 3.51, 2.52, 1.49 and 1.00 g H20/g DM was used. Soybean curd
stuffed into aluminum tubes (10 mm i.d. X 20 mm) was frozen in a freezer kept at -30°C for
12hr. Cylinders removed from the tubes were used as test pieces for compression.

Apparatus and Procedure

A testing machine (Model UTM4-200, Toyo Baldwin Co.) was used for uniaxial
compression tests. A test piece, placed between the bearing plates which had been cooled in
advance with cold nitrogen gas, was cooled to a preset test temperature between -20°C and
-196°C. The test piece was cooled slowly and carefully at a rate of about 2.0°C/min to avoid
cracking due to thermal shock.

RESULTS AND DISCUSSION

The maximum stress recorded in a stress-time curve was referred to as fracture stress. The
fracture stress of soybean curd increased (10 MPa at -20°C) as the temperature decreased until
it reached a plateau (35MPa) when the samples cooled down over a characteristic temperature
(-80°C). The fracture stress depended upon the moisture content of the soybean curd before it
was frozen. The lower the moisture content, the larger the fracture stress. The fracture stress
at the plateau varied from 35 MPa (9.93 g H20/g DM) to 90 MPa (1.00 g H20/g DM). On the
other hand, the fracture stress of pure water ice (25MPa) was constant over the whole range
between -20°C and -196°C.
We analyzed the fracture stress of the frozen soybean curd using a mathematical model in
which frozen soybean curd was regarded as a two component system consisting of pure
water ice and concentrated amorphous solution (CAS).
Since volumetric fraction of pure water ice in the system is required for the analysis, the
degree of freezing of soybean curd with varied moisture content was estimated as a function
of temperature using a hypothetical phase diagram for soybean curd. Based on these data,
fracture stress of CAS was calculated using series model and parallel model. The calculated
fracture stress of CAS was found to be a unique function of temperature without regard of
moisture content before freezing. This result is supported by the fact that the protein
concentration of CAS is not affected by initial moisture content of soybean curd but is
determined uniquely by the temperature.

REFERENCE

1.Hagura, Y. & Watanabe, H (1991). Factors affecting separation of low fat flesh from
fatty fish by cryo-shattering. 1. Food Sci., 56, 1567-1571.
 THE INFLUENCE OF MOISTURE CONTENT ON THE FAILURE MODE
OF BISCUITS

L. PIAZZA, R. BRlNGIOTTI AND P. MASI*

DIS TAM, Universita' degli Studi di Milano, via Celoria, 2 - 20133 Milano, Italia
* Istituto di Industrie Agrarie, Universita' degli Studi di Napoli "Federico II"
Parco Gussone - 80055 Portici (Napoli), Italia

INTRODUCTION

Friability is a salient textural characteristic for most fresh biscuits. Its loss during storage
because of moisture absorption is a major cause of rejection by consumers of these baked
products (I). It may therefore result of great interest to identity a water content limit at
which the textural quality of biscuits becomes organoleptically unacceptable.
This paper presents the failure behaviour of biscuits at various moisture contents. The
relationship between consumer acceptance and the evolution of mechanical properties of bi-
scuits is discussed.

MATERIALS AND METHODS

Commercial short-dough-rotary-mouldered biscuits were used (2). Samples utilized in

sensory analysis were conditioned at different moisture content in the range of 0.02 - 0.16
grams of water per grams of dry matter. A Climatic Chamber (Hereaus-Votsch Mod. HC
0020) was utilized. The same moisture range was obtained by equilibrating samples over
saturated salt solutions in dessicators to constant weight for the pre treatment of biscuits
used in instrumental test. Inspection of biscuits at different water contents was performed
with a Cambridge Stereo Scan 250 MKZ electron light microscope, after freeze drying and
metallization treatments of the samples.
As for mechanical force/deformation analysis, a textural test was performed by means of an
Instron UTM (Mod. 430 I) interfaced with a PC for automatic data collection and
elaboration. The apparatus was equipped with a three points bending fixture. The test was
carried out at a crossbar speed of 50 mm/min. The sensory analysis was carried out with a
panel of 15 subjects. They evaluated friability and the overall product acceptability. A score
test was performed, using a structured rating scale. The experimental design was based on
three repetitions. The data were submitted to the ANOV A analysis.
 135
RESULTS AND DISCUSSION

One of the most important textural properties of biscuits is friability. This is the parameter
that better characterizes the mechanical properties of the product.
Electron light scattering investigations reveal a cellular structure of the biscuit. When
biscuits undergo a force during mastication they break into many pieces depending on the
size of the air cells and on the thickness of the cell wall. From a structural point of view, they
can be therefore considered as a rigid foams and their friability may be correlated to their
brittle type failure that can be observed for these alveolar structures (3).
In this research a flexural test has been performed to express friability in physical terms
on the basis of instrumental measurements.
A typical stress/strain diagram obtained with a flexural method for a low moisture
product is characterized by a large number of peaks which correspond to the progressive
failure of cellular walls. For small deformations biscuits show an elastic behaviour.
Therefore, the elastic modulus can be calculated in the linear elastic region and assumed as
an index of friability. Another index of the failure mode of friable materials is the energy to
failure. It clearly depends on friability: the more deformable the structure, the higher the
energy required to break the product.
Moreover, one must take into consideration that the the assessment of friability evaluated
in a flexural mode reproduces what really happens during mastication. Here the first impact
with the biscuit is carried by incisive teeth operating a flexure breaking action. Later on the
compression of the product is due to the action of the molar teeth.
Therefore, in order to compare mechanical and sensory evaluation of friability, the
flexural test has been referred as to a useful instrumental approach.
Mechanical properties of biscuits are modified when the hydration of the product varies.
Water acts in the structure as a plasticizer, that softens the rigid matrix. It has been
demonstrated that as the moisture content of biscuits increases, their structure become softer
and their failure mode varies: the modulus of elasticity decreases when the water content
increases. It has been also demonstrated that the higher moisture content, the higher the
structure resistance to deformation caused by small stresses. The experimental evidence
shows that the failure mode of biscuits changes from brittle to ductile.
Since the aim of this research is to establish in objective terms critical moisture conditions
for textural acceptance of biscuits, an effective correlation between instrumental and sensory
evaluations of friability must be assessed. To this aim fifteen judges evaluated friability
intensity of the samples conditioned at different moisture contents.
The sensory acceptability is higher for the driest biscuits and increases when friability
increases. Figure 1 shows a linear correlation between instrumental and sensory assessments.

CONCLUSIONS

The results indicate that the textural properties of friable products can be described in
explicit physical terms and that a linear correlation exists between the flexural modulus and
the sensory perception of friability. It can be therefore recognized a threshold of modulus of
elasticity at which the textural quality of biscuits becomes organoleptically unacceptable. The
relevant threshold of water content has been identified equal to 0.031 grams of water per
grams of dry matter, which corresponds to a flexural modulus equal to about 6200 g/mm 2 .
136
7000

• •
6000
•
-
5000
N

§ 4000
C,
UJ
3000 •

2000
• Y=-910.785+1011.007X
R2= 0.961
1000

•
0
1 2 3 4 5 6 7 8
Sensory Score

Figure 1. Correlation between sensory and instrumental evaluation of friability

ACKNOWLEDGMENTS

Research supported by National Council of Italy, Special Project RAISA, Sub-Project N.4,
PaperN.926

REFERENCES

l. Kats, E.E. and Labuza, T.P., Effect of water activity on the sensory crispness and
mechanical deformation of snack food products. L Food. Sci., 1981,46,403-409.

2. Bringiotti, R., Prodotti da fomo tipo biscotteria: messa a punto di metodologie di

carattere fisico per la valutazione dei parametri di qualita' del prodotto. Thesis,
University of Milano, 1992.

3. Romano, V., Demma, G., Acierno, D. and Masi, P., Rheological behaviour of low
moisture bakery products. In: Engineering and food. Proceedings ICEF ~ Cologne,
1989. W. Spiess and H. Schubert (cds). Vol. 1: Physical properties and processes.
Elsevier Sci. Pub. Ldt, Barking, England, 1990, pp. 89- 97.
 CHANGES OF PHYSICAL PROPERTIES OF DRYING MATERIALS

P.P.Lewicki, D.Witrowa, W.Pomaranska- ~zuka

Department of Food Engineering
Warsaw Agricultural University, SGGW
02-766 Warszawa, ul.Nowoursynowska 166

ABSTRACT

Apple and potato were cut in cubes and dried by convection to different
final water contents. The shrinkage of the material was measured, and a
linear function between shrinkage and water content has been found.
Rehydration of both materials show the differences in changes incurred to
the material by the drying process. Measurement of water imbibition and
solubles leakage shows the moments of drying which are most detrimental to
the material.

INTRODUCTION

Drying, as a method of food preservation, causes many changes in material

being dried. The effect of drying procedures, of drying methods, and of
drying conditions on the chemical and sensory changes in the material, are
well known and described in the literature [1,2].
Little work has been done to recognize changes of physical properties
of the material which are induced by drying. Modifications of physical
properties brought about by drying process are expressed. among others, by
the ability of dry material to imbibe water, swell, and to loose solubles
to the rehydrating water.
No information is available about the course of rehydration in
relation to the state of dryeness of the material. The aim of the present
work is to investigate the changes induced to the material by drying,
expressed by its ability to imbibe water, and to loose solubles to the
surrounding medium.

MATERIALS AND METHODS

Apple of Antonawka variety and potato of Kalina variety were cut in 1 cm

cubes and dried in convection dryer at air velocity of 2 m/s. Potato cubes
were blanched in boiling water for 2 min, and apple cubes were immersed in
1% citric acid solution prior to drying. Drying was done at 60°C for
potato, and at 70°C for apple. The drying process was concluded at variable
138
final water contents.
Rehydration was done in distilled water at 20°C for 5 hours. At
specific time intervals samples of the material were taken from the
rehydrating water. Mass and dry matter content were measured.

RESULTS

Drying process. Both, apple and potato, show constant and falling drying
rate periods. Drying rate in the constant rate period is equal to
1.74'10-3 kg H2 0/(kg d.m. s) for apple and 1.06·10-~g H2 0/(kg d.m. s) for
potato. During this period 30.7% of water is removed from apple and 36.6%
from potato.
Apple and potato shrink during drying and the shrinkage is linearly
related to the water content.
Both materials differ
substantially as the rate and
IMBIBED WATER AT EQUILIBRIUM, kg/kg idm the extent of shrinkage are
3,0
concerned. However, the largest
difference appears when
experimental shrinkage is
compared with theoretical
1,5
shrinkage calculated on the
basis of the amount of
1,0 f- .........................•............. evaporated water. Potato
0,5 f- ............................ , ......................... , ......................................•....................",
shrinks less, and apple shrinks
more than is predicted by the
theory.
0,0 t===:::L:===:::J===:::t====t:=~=:::J
o 2 3 Rehydration. Dry apple and
WATER CONTENT, kg/kg d.m. potato immersed in water imbibe
it and the amount of water
taken in depends on the
Figure 1. Relationship between absorbed dryeness of the material. The
water and initial water content lower the water content the
higher imbibition of water is
observed. The amount of imbibed
;...:R=EL=A...:..:J:..:..IV.::...:E~D.::...:R:..:..Y-=M.:..::A-'-J=-T:....::E::..:...R=-C.::...:O.::...:N::..:...T.:....:E=.cN.:....:T~_--,
water is always less than the
1,1 amount removed during drying.
This relationship is linear
until the content of 1 kg/kg
i.d.m. is reached in dry
material. Both apple and potato
0,9 f- ......................... ....................,...................... f··········· /c ... ;.......................... I
~""""
with moistness less than 50%
behave similarly, and they
O,Bf-................................................................... , ..............
H .................. ,·, ................ .
imbibe less water when the
lower the initial water content
exists in the material (Fig.1).
0,7 L--_ _---L_ _ _"----_ _-"--_ _---'_ _- - - - '
Leakage of solubles is
o 2 3
very different for both
WATER CONTENT, kg/kg d.m. materials studied. Apple looses
increasingly more solubles the
Figure 2. Relationship between relative lower initial water content is
dry matter content and initial water in the rehydrating material.
content This relationhip is linear for
 139
the whole range of water contents studied. Up to 75% of initial dry matter
content is lost during rehydration of apple. Potato looses solubles easily
up to water content of about 2.5 kg/kg i.d.m. This amounts up to about 20%.
Below this water content solubles are immobilized in the material and
potato containing about 10% water looses about 7% of solubles during
rehydration (Fig.2).

DISCUSSION

Apple and potato are parenchymateous tissues but their structure and
chemical composition is very different. Potato contains about 25% d.m. and
most of it is starch. Other compounds constitute about 26% of d.m., and
only some of them are soluble in water. The intercellular free space is
some 3% of total tissue volume [3,4]. Apple contains about 15% d.m. and
over 85% of it are solubles. Its intercellular free space is much larger
than that of potato.
The course of drying and shrinkage are directly related to the
structure and chemical composition of the material being dried. At high
share of insoluble material in total d.m. the drying rate is slow and
shrinkage is less than that predicted. Tissue which contains mostly
solubles, and has an open structure (e.g. apple) dryes quickly and its
shrinkage is large, even larger than predicted.
Drying causes changes in the material from the very beginning. Even at
low temperature drying (constant drying rate period) changes are so
advanced that the material does not imbibe as much water as evaporated.
These changes are probably due to concentration of the cell sap. The more
water is removed the changes are more pronounced. This may arise from both
concentration and thermal effects. These changes are expressed in
rehydration process, but the water imbibition and solubles leakage must be
simultaneously analysed.

CONCLUSIONS

The following conclusions are drawn from this experiment:

1. The structure and chemical composition of the material affect
strongly the course of drying and the shrinkage of the material.
2. The advancement of changes caused by concentration and thermal
effects in the material can be observed by rehydrating materials dried to
different final water contents.

REFERENCES

1. Food Dehydration, ed. Van Arsdel, W.B., AVI, Westport, Conn. 1973,
Vol.2., second edition.

2. Karel,M., Optimization of quality of dehydrated foods and biomaterials.

In Drying-92 ed. Mujumdar,A.S., Elsevier, London 1992, pp. 3-16.

3. Sterling,C., Anatomy and histology of the tuber with respect to

processed quality. Froc.Plant Sci.Symp. 1966, 11,25.

4. Crapiste,G.,H. and Rotstein,E., Prediction of sorptional equilibrium

data for starch containing food-stuffs. J.Food Sci 1982, 47, 1501-7.
 EFFECTS OF BAKING AND STORAGE CONDITIONS
ON MECllANICAL PROPERTIES OF flUTE BREAD

YIPING WANG
Toyohashi Works, Shinko Electric Co. ,Ltd.,
150 Motoyashiki,Sanya-cho,Toyohashi 441-31,Japan
HIROSHI MORISHIMA, YASUHISA SEO, YASUYUKI SAGARA
Department of Agricultural Engineering,
Faculty of Agriculture,The University of Tokyo,
1-1-1 Yayoi,Bunkyo-ku,Tokyo 113,Japan
KENJI IMOU
Department of Agricultural Environmental Engineering,
Faculty of Agriculture,Utsunomiya University,
350 Mine-machi,Utsunomiya 321,Japan

ABSTRACT

A force-compression behavior of, bread crumb has been characterized from

th~ viewpoint of rheology. The experimental results indicate that struc-
ture, mechanical strength and elastic modulus of crumb change by loading
directions showing its anisotropy. Firmness and breaking energy of crumb
increase with the storage time and change markedly within 12 hours after
baking. They are affected by storage and measuring conditions. An
appropriate measuring method was recommended for obtaining more uni ver-
sal data of breaking characteristics.

INTRODUCTION

Measurement of mechanical properties of bread crumb is closely related

to evaluation of staling and to general crumb quality control. Effects
have been made to measure firmness of bread crumb. The official AACC
method has been used to evaluate staleness of bread with the Baker com-
pressimeter. Instron factors involved in measuring crumb firmness were
discussed. The 25 % compression was considered as an optimum depth for
measuring bread crumb firmness[l).
The objectives of this work were to investigate the direction
dependence of mechanical properties and the effects of storage and
measuring conditions on firmness of crumb.
 141
MATERIALS AND IIlETIiODS

Whi te breads of approximately 230 g were prepared as the samples by the

sponge-dough procedure. All 16x8x8cm loaves baked at 230°C were wrapped
in polyethylene bags and stored at 20 <t until compression test. An
uniaxial compression testing machine was developed, and the mechanically
sliced ~ em thick breads were compressed by a 3 cm diameter plunger and
the 4 em cubic crumbs by a 5.5 em one in evaluating breaking properties
and anisotropy of crumb. Firmness was defined at 25 % compression of the
force-compression curve.

RESULTS AND DISCUSSION

A typical force-compression curve for white bread is shown in Figure I.

The curve showed the common characteristics of baked goods[2] and was
found to consist of four stages. The initial toe(ab) up to possibly 4 %
compression was caused by the initial compression of the partially rup-
tured surface; namely, nonuniform surface against the plunger. In the
linear stage(bc), crumb behaved elastically indicating a rather rapid rise
in the force. When compression reached the range from 12 % to 40 %(cd),
the solid matrix of crumb distorted remarkably and broke partially. When
compression became larger than 40 %(de), the force increased and a lot of
pores were ruptured, showing densification in the texture of crumb.
The experimental results indicated that the structure and mechani-
cal strength of crumb depended on storage time and the direction of
applied force. The equivalent Young's moduli increased with storage time
and varied by direction and showed largest values in the direction of
vertical axis and smallest ones in long axis during storage. Their range
were between 0.5x10 4 to 1.5x10 5 Pa, comparing with 2xl0 2 Pa reported[3].

50r------------------------r--------~
2cm slice
Age: 24h
40
I I II I III IV
~ 30
Z
(I)

~ 20
0
~ I ; Initial toe
II ; Linear stage
10 Ill; Collapse
IV; Densification
a
0 25 50 75 100
Compression (%)

Figure 1. A typical force-compression curve for white bread crumb

142
20r------------------------------.

CIS
0... 15
0 ......... 0

9....·.. ·..
.......... Ill····
_.~~'
.....
...... ,,0· ..''···
.....o ,,".. .
... 0" -'•
~.~~~.

./6 ............ .-Storage temperature 5·C

.' .•.. 0 ...... Storage temperature 13'C
........ Storage temperature 20'C
O~L_L_L_L_L_~L_L_L_L_~~~~~~~

24 48 72 96 120 l44 168 216

Age (h)

Figure 2. The effect of storage temperature on the firmness of crumb

As shown in Figure 2, The firmness increased as the crumb age and

changed markedly with in 12 hours after baking. Crumb produced s ignif i-
cant difference in firmness it was when stored at 5, 13 and 20 0 C and the
firmness increased as a decrease in storage temperature. The same tend-
ency was also found in breaking energy, and thus the storage temperature
was suggested to be an important factor for bread quality control.
For obtaining more universal data of breaking characteristics, an
appropriate measuring method was recommended as follows.
(1) Use the plunger which surface area is between 8 to 20 % of that
of slice. (2) Firmness is determined at 25 % compression with deforma-
tion rate between 5 to 10 mm/s.

CONCLUSIONS

Force-compression curve of crumb has the common characteristics as same

as shown by other baked goods. The values obtained for the breaking
characteristics of crumb at any particular time after baking depend
markedly on the direction of applied force as well as storage tempera-
ture. In order to know the mechanism of anisotropy of crumb, further
work on relation between microscopic structure and strength of loaf is
required.

REFERENCE

1. Baker, A.E., Walker, C.E. and Kemp, K., An optimum compression depth
for measuring bread crumb firmness, Cereal Chemistry., 1988, 65(4), 302-7
2. Peleg, M., Roy. I., Campanella, O.H. and Normand, M.D., Mathematical
characterization of the compressive stress-strain relationships of
sponge baked goods., ~ Food Sci., 1989, 54(4),947-49.
 IMPEDANCE SPECTROSCOPIC ANALYSIS
IN AGRICULTURAL PRODUCTS

KIYOHIKO TOYODA
Department of Agricultural Engineering
Kobe University, Nada, Kobe 657, Japan

ABSTRACT

The relationship between the impedance characteristics and the

physical and physiological properties of agricultural products was
investigated by an equivalent electrical model. The impedance of
several agricultural products, measured over a range of
alternating-current frequencies from 40 Hz to 110 MHz, was
analyzed by the modified Hayden model taking account of the four
parameters (apoplasmic resistance Ra, symplasmic resistance Rs,
membrane capacitance Cm and constant-phase angle 11').
The results show that the modified model accurately predicts the
impedance characteristics of all the specimens and that impedance
spectroscopy is useful in estimating the changes in the properties
of agricultural products, e. g. maturation or gelatinization of
starch.

INTRODUCTION

Impedance spectroscopy (IS) is a relatively new and effective

method of characterizing many of the electrical properties of
materials and it can be used in detecting the changes in physical
and physiological properties of biological materials which
accompany the changes in their electrical properties.
Equivalent Electrical MOdel
Cellular tissues like those of agricultural products can be
macroscopically regarded as a system in which cell membranes
electrically isolate intracellular (symplasmic) electrolytes from
extracellular (apoplasmic) ones. Hence, the impedance
characteristics can be described by an appropriate and equivalent
electrical model.
Hayden et al. and others described an electrical model of four
elements, i.e. apoplasmic resistance Ra, symplasmic resistance Rs
and resistance Rm and capacitance Cm of cell membranes (Fig.l.a).
However, for most biological materials, Hayden model does not
144

Ra

Figure 1. Equivalent Electrical Model

=
N, wO Z=Z' + j Z" cell impedance :z=-------
1 IRa + 1 I (Rs + Zm)

cos", + jsinlp
membrane impedance: Z m= - - - - - - -
Cmw

Figure 2. Cole-Cole Plot and Equation of Modified Model

agree with their impedance characteristics. Therefore, the author
used a modified model incorporating a capacitor with a constant-
phase angle 1p for membrane capacitance (Fig. 1. c) . The equation
and the Cole-Cole plot of the modified model are shown in Fig.2.

MATERIALS AND METHODS

The impedance of potato tubers at various temperatures (20 to SO

oC), banana fruits at various stages of maturation and others
(carrot roots, persimmon fruit:;s, horse radish roots, Japanese
radish roots and calluses of melons) was measured with LCR Hi-
tester (HIOKI Corp. 3520) and Vector Impedance Meter (HP-4193A) by
four electrode measurements.

RESULTS AND DISCUSSION

90

h
""""
I'-- -
---
./"
---.....
--- - - -
- ,.,- ---:::
. , .
t'--
. ,
T. . .
l~ l~ l~ l~
frequency, w (Hz)
b_ Bode diagram

Figure 3. Typical Impedance Characteristics of Plant Tissues

The following was observed on the Cole-Cole plot (Fig.3)

 145
i) The first measured arc was similar to that calculated from the
modified model. ii) the second smaller arc appeared at a
frequency higher than 1 MHz.
The modified model was applied to the first measured arc and
Ra, Rs, em and '" were determined. For all specimens, the values
calculated from the model agreed with the values measured in the
first arc.
with a rise in the temperature from 50 to 60 0C (corresponding
to 3.0 x 10- 3 to 3.1 x 10- 3 of l/Tabs), the effect on potato
tubers appeared as a decrease in the radius on the Cole-Cole plot.
In the Arrhenius plot of the parameters, Ra decreased rapidly, and
Rs changed from a gradual decrease to a rapid increase (Fig. 4) .
Microscopic observation confirmed that these changes related to
gelatinization of starch grains in potato cells, and it was
reasonable to assume that the symplasmic electrolytes were forced
out of the cells with the swelling of starch grains in
gelatinization and manifested the above changes in Raj Rs also
showed these changes. Therefore, the ratio of Rs to Ra responded
to the changes in and was conducive to estimating the degree of
gelatinization. By similar analysis, banana fruits at different
stages of maturity showed that the development of maturation
contributed to the decrease in apoplasmic resistance.

g::j
2"2.0/
Ra

J-'---..l-.-_-'-_ _1.... ,10. 3

2.0 LJ.._ _..l-._ _-'-_ _ ~, 10. 3
2.8 3.0 3.2 3.4 2.8 3.0 3.2 3.4
Iffabs Iffabs

~ 0.0 o~ rat;o,(RslRa)

~L
E:S
o•
..9
-1.0

I-J-_ _L-_-'---'::=~ ,I o'}

2.8 3.0 3.2 3.4
Iffabs

Figure 4. Relationship between the model parameters

and the temperature.

CONCLUSIONS

The modified model can interpret the impedance characteristics of

the tested agricultural products successfully, and IS is effective
in detecting the maturation of banana fruits and gelatinization of
potato tubers through the modified model.
 FINITE ELEMENT ANALYSIS ON THE EFFECTIVE THERMAL
CONDUCTIVITY OF DISPERSED SYSTEMS

T AKAHARU SAKIYAMA and TOSHIMASA Y ANO

Department of Agricultural Chemistry, The University of Tokyo,
1-1-1 Yayoi, Bunkyo-ku, Tokyo 113, Japan

ABSTRACT

Effective thermal conductivity (lte) of dispersed systems of spheres of unequal sizes and/or
materials was computed by the finite element method. When spheres of two different sizes,
but of a single material, were dispersed in the system, calculated values of Ite was well
approximated by the Maxwell-Eucken model. When two kinds of materials were dispersed
as spheres, Ite was in good agreement with the value expected from the extension of the
Maxwell-Eucken model for multi-phase systems. These results suggested the applicability of
the Maxwell-Eucken model and its extension for the prediction of Ite of foods in dispersion
state. The applicability of these models was confirmed by experimental measurements of Ite
for gels containing paraffin particles and/or air bubbles.

INTRODUCTION

To predict the effective thermal conductivity (lte ) of a heterogeneous food, it is necessary to

know the most suitable heat conduction model for its structure. For dispersions, a lot of
theoretical models have been proposed. However, little consideration has been given to the
effect of size distribution of the dispersed spheres on Ite. In addition, few studies have been
undertaken for multi-phase systems.
In this study, Ite of dispersed systems of spheres of unequal sizes is computed by the finite
element method (FEM) for two- and three-phase systems to discuss the applicability of
known models. Experimental measurements of Ite are also performed for several dispersions.

MATERIALS AND METHODS

Measurement of Effective Thermal Conductivity

Agar gels impregnated with solid paraffin particles were prepared as two-phase model
systems containing dispersed spheres of unequal sizes. Gelatin-glycerol gels (moisture
content=12.4%) impregnated with paraffin particles and air bubbles were prepared as three-
 147
phase model systems. These samples were subjected to the measurement of Ae by the steady
heat flow method [1]. In each measurement, the mean temperature of the sample was 20o e.

Finite Element Analysis

Figure 1 shows regular arrangements of spheres used as dispersion models in the numerical
analysis. Model (A), a simple cubic lattice, contains spheres of a uniform size, whereas
model (B) contains spheres of two different sizes. If the size of the spheres is uniform, model
(B) is identical with model (A). In these two models, dispersed spheres are composed of a
single material. Model (C) is similar to model (B), but with spheres of two different
materials. For each model, a three dimensional steady heat conduction problem was solved
by FEM to calculate Ae.

RESUL TS AND DISCUSSION

For two-phase systems, a lot of theoretical heat conduction models for sphere dispersions
have been proposed. The Maxwell-Eucken model [2] is an example of those models.

Ae = <H2-2¢(1-cr) (1)
Ac cr+2+¢(1-cr)

where Ac is the thermal conductivity of continuous phase, cr the thermal conductivity ratio of
dispersed spheres and continuous phase, and ¢ the volume fraction of dispersed spheres. In
Fig.2, the results ofFEM computation for models (A) and (B) are compared with Eq.(1).

(A) (B) (C)

Figure 1. Dispersion models used in the finite element analysis.

1.0 10.0 , . . - - - - - - - - - - r - - - - - ,
(a) 0"=0.01 (b) 0"=100

0.8 8.0

6.0
~

0.6
«" «"
--
«
0> 0.4 --
«0> 4.0 0

Key Model Eq.(1) Key Model

0.2 0 A II 2.0 0 A
0 B 0 B

0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Volume fraction of dispersed phase (1) Volume fraction of dispersed phase (1 )

Figure 2. Effective thermal conductivity of models (A) and (B) calculated by FEM.
148
In the case where 0=0.01, values of Ae of models (A) and (B) were approximately equal with
the values expected from Eq.(l) in a wide range of l/J. In the case where 0=100, however, Ae
of model (B) was significantly smaller than that of model (A) for l/J>0.4. Equation (1) gave a
fairly good approximation for Ae of model (B). Thus, the applicability of Eq.(1) was
suggested for the prediction of Ae of foods containing dispersed spheres of unequal sizes.
Equation (1) also gave a good approximation for the experimental values of Ae of gels
containing paraffin particles no matter how the size distribution of particles varied.
For multi-phase systems, limited number of heat conduction models have been proposed.
One of those is the extension of the Maxwell-Eucken model derived by Hamilton [3].

Ae = 1-2Il/Ji(1-0'j}/(2+O'i)
(2)
Ac 1+Lf>i (l-O'i)/(2+O'i)

where suffix i represents each dispersed phase. Figure 3 shows the results of FEM
computation for model (C) in the case where 0'1=0.1 and 0'2=0.01, in comparison with Eq.(2).
The values of Ae were in good agreement with those expected from Eq.(2), suggesting the
applicability of Eq.(2) for the prediction of Ae of three-phase systems. Experimental values
of Ae of gelatin-glycerol gels containing paraffin particles and air bubbles differed by no
more than 10% from the values expected from Eq.(2).

1.0
Key </11
0.8 0 0.167
0 0.217
~

0.6
«0
--
«OJ 0.4

0.2

0.0 '--_-'---_....J......_---'-_--'_----'

0.0 0.1 0.2 0.3 0.4 0.5

</12 (1)

Figure 3. Effective thermal conductivity of model (C) calculated by FEM for the case where
0'/=0.1 and 0'2=0.01.

REFERENCES
1. Sakiyama, T. and Yano, T., Effective thermal conductivity of porous gel. In Engineering
and Food, VoU, ed. by W.E.L. Spiess and H. Schubert, Elsevier Applied Science,
London, 1990, pp.424-431.
2. Eucken, A., Allgemine GesetzmaBigkeiten fOr das Warmeleitvermogen verschiedener
Stoffarten und Aggregatzustande. Forsch. Geb. Ingenieurwes., 1940, A 11, 6-20.
3. Hamilton, R.L, Thermal conductivity of two-phase materials. Ph.D. Thesis, Univ.
Oklahoma, U.S.A., 1960. (cited in Cheng, S.C. and Vachon, R.I., The prediction of the
thermal conductivity of two and three phase solid heterogeneous mixtures. Int. J. Heat
Mass Transfer, 1969, 12,249-264.)
 MEASUREMENT OF FRACTION OF FROZEN WATER AND
THERMAL CONDUCTIVITY IN FROZEN FOOD MATERIALS

RUNGNAPHAR PONGSAWATMANIT,* AND OSATO MIYAWAKI

Department of Agricultural Chemistry, The University of Tokyo,
1-1-1 Yayoi, Bunkyo-ku, Tokyo 113 Japan,
*Department of Agro-Industry, Faculty of Agricultural Technology,
King Mongkut's Institute of Technology Ladkrabang, Bangkok 10520, Thailand.

ABSTRACT

Fraction of frozen water as a function of temperature, was determined from phase diagram or
from differential scanning calorimetry method for glucose, sucrose and gelatin solutions. Effective
thermal conductivities of these solutions were measured by the steady state method in the
temperature range from -20 to 20°e. At unfrozen state, thermal conductivity was not much
dependent on temperature, and then intrinsic thermal conductivities of the solute materials were
determined based on the structure models (parallel, series and Maxwell-Eucken models). Below the
freezing point, however, thermal conductivity was strongly dependent on temperature reflecting the
change in the fraction of frozen water with temperature. Theoretical analysis on the temperature
dependency of the thermal conductivity was made again by applying the structure model analysis for
frozen state. Among the three models tested, Maxwell-Eucken model with ice as the dispersed phase
gave the best prediction for effective thermal conductivity.

INTRODUCTION

Thermal conductivity of food materials is very important for improvement in design of heating
and cooling processes. The most basic characteristics of a frozen food necessary for the design of
freezing process is the relationship between the ice fraction and temperature. The amount of ice is a
function of chemical composition of the system and temperature. Accurate know ledge of this allows
calculation of many factors required in the design of the system for freezing of food. The ice fraction
strongly affects on the effective thermal conductivity of frozen food materials because thermal
conductivity of ice is about four times that of water.
In this paper, phase diagram method and DSC method are applied to determine ice fraction for
binary water system with glucose, sucrose,and gelatin. Effective thermal conductivity of frozen
foods measured from steady state method is analyzed by structure models combined with intrinsic
thermal conductivity.
150
MATERIALS AND METHODS

Ice fraction determination

Phase dia~ram method: The prepared sample solutions of glucose, sucrose or gelatin gels
contained in the brass tube with sensor for temperature measurement were immersed in the coolant
(ethylene glycol 50%) until the sample was completely frozen. Then, the temperature of the coolant
was increased for heating process to determine melting point.
Differential Scanning Calorimetry (DSC) method: L-DSC (Rigaku Corporation) was used to
measure enthalpy changes in samples. Each solution, 15-25 mg, was transferred into an aluminium
DSC pan and the pan was sealed. The aluminium oxide (AI203) was used as a reference sample.
After the samples were cooled down to -70°C or below to ensure the complete freezing, the samples
were warmed up at a constant rate. The enthalpy obtained from the DSC endothermic curve was
used to determine ice fraction as a function of temperature following the method by Maeda and Koga
[1].

Measurement and prediction of effective thermal conductivity of food materials

Effective thermal conductivity of food sample was determined from the coaxial concentric
cylinder with steady state method in the temperature range from -20 to 20°C [2]. The predicted value
was obtained from the analysis of the structure model: series, parallel and Maxwell-Eucken models.

RESULTS AND DISCUSSION

Ice fraction determined from phase diagram and DSC

Phase diagram determined from the melting point agrees well with that from literature [3,4] for
glucose and sucrose as shown in Fig. 1 and was used to calculate ice fraction. For that of gelatin gel,
there is no literature data to compare with. In principle, phase diagram method provides the best value
of ice fraction because it is an equilibrium method based on thermodynamics. However, in many
cases, an accurate phase diagram is not available especially in the system involving macromolecules
like water-gelatin system. Therefore, the DSC method is used by correcting the enthalpy obtained
with temperature-dependent latent heat of ice and empirical equation proposed by Schwartzberg [5]
as the following equation
5~-----------------r--~----.
O • Glue-exp.
Glue-literature
_5r-....~~.. I';J --
... Sue-expo
••••••. Sue-literature
-1 0 I';J Gelatin-exp
-15 '.
-20
-25
-30~~~~~~~~~~~
o 1 0 20 30 40 50 60
Concentration (%)
Fig. 1 Phase diagram of glucose and sucrose solutions and gelatin gels obtained.

(1)
where, x i is ice fraction in total system, x w is water content, x b is bound water content and T f is the
initial freezing point. This can be applicable to determine ice fraction of macromolecule system where
phase diagram is not available.
 151
Prediction of effective thermal conductivity of frozen food materials
For the prediction of effective thermal conductivity, ice fraction was obtained from the phase
diagram method for glucose and sucrose, and from the DSC method for gelatin.The intrinsic thermal
conductivities of glucose, sucrose and gelatin determined from effective thermal conductivity of
unfrozen state with Maxwell-Eucken models were 0.295, 0.293 and 0.280, respectively [2]. The
prediction of effective thermal conductivity for frozen sample by the Maxwell-Eucken model,
assuming ice as dispersed phase and thick solution as continuous one showed the better result than
the series and parallel models. The prediction showed a good agreement with the experimental
results as shown in Figs. 2 and 3 within ±1O%.
1.8 r--------;:::====:::::;-i sz 1.2 r---------~
• 20.206% E • 28.4%-exp
1.6 • 29.323% ~ --28.4%-ME
•
1.4
• 38.33%
• 50.11%
~
.:;
1_0 • • .- • 38.0%-exp
·······38.0%-ME
--20.2%-ME • 45.0%-exp
...
-.
12 ··-·_·-29.3%-ME ~ ...... -··4S.0%-ME
-g
...,..
0.80
. _ . - 38.3%-ME
• '--

.. ...
...... -~..... - - SO.l%-ME
--. "'"
o
o
.... iii 0.60 ~ 'IO"--'~."-:
0.8
0.6 <t..
~~""" ....
- \
i
-5 \\:::u::.:::::-~. . =::y
.\.~: __ .!-:-:__,,",."P.,=-:-:"W."..=~~
0.4
..........
- - .... - g! 0.40

0.2
~
w 0.20 L.....~..........~~--'--~~..........~-o-J..~~--'--~.............J
-30.0 -20.0 -10_0 0.0 10.0 20.0 -30.0 -20.0 -10.0 0.0 10.0 20.0 30.0
Temperature (oG) Temperature (oG)
Fig. 2 Comparison of thermal conducllvities of glucose solutions Fig. 3 Comparison of thermal conductivities of gelatin
from the experiment with those from the calculation. gels from the experiment with those from the calculation.

CONCLUSIONS

Effective thermal conductivity of frozen sample was strongly dependent on temperature due to the
change of ice fraction with temperature. Ice fraction was determined by phase diagram and DSC
methods for solutions with low molecule and high molecule materials, respectively_ The structure
model analysis for frozen state showed that the Maxwell-Eucken model with ice as dispersed phase
combined with the intrinsic thermal conductivity obtained from unfrozen state gave a good
agreement within ±1O% difference from the experimental results.

REFERENCES

1. Maeda, Y. and Koga,S., Thermal properties of water in Bacillus cereus and vegetative cells.
Netsusokutei., 1981,8,57-61.

2. Pongsawatmanit, R., Miyawaki, O. and Yano, T. Measurement of the thermal conductivity of

unfrozen and frozen food materials by a steady state method with coaxial dual-cylinder apparatus.
Biosci. Biotech. Biochem., 1993,57,1072-1076.

3. Young, F.E., D-Glucose-water phase diagram. J. Phys. Chern., 1957,61,616-619.

4_ Young, F.E. and Jones, F.T., Sucrose hydrates. J_ Phys. and ColI. Chern., 1949,53, 1334-1350.

5. Schwartzberg, H.G., Effective heat capacities for the freezing and thawing of food. 1. Food Sci.,
1976, 41,153-156.
 THERMOPHYSICAL PROPERTIES OF RASPBERRIES AND APPLES

VALERIO BIFANI(l), DAVID BORCOSKI(l) , PEDRO MOYANO (2)

AND FERNANDO OSORIO(3)
1. Chem.Eng.Dept. U. de La Frontera, P.O. Box 54 D Temuco
2. Chem.Eng.Dept. U. de Santiago, P.O.Box 10233 Santiago
3. Food Eng.Dept. U. de La Serena, P.O.Box 599 La Serena
CHILE

ABSTRACT

Density, specific heat and heat diffusivity of raspberry and

apple puree were experimentally determined above and below the
freezing point (fp) and heat conductivity was calculated from
them.
Densi tl varied from 950.4 kg/m3 for apples below fp to
1054.0 kg/m for raspberries above fp. Specific heat varied from
1985 J/kg K for apples below fp to 4900 J/kg K for raspberries
above fp. Values of heat diffusivity ranged from 1.063*10- 7 to
2.375*10- 7 m2 /s for apples above and below fp respectively. These
values are consistent with those reported for apple puree in the
literature; no references on thermal data for raspberries was
found.

INTRODUCTION

Food preservation processes are linked wi th equipment and methods

where different operations, like dehydration, evaporation,
refrigeration or freezing, are involved. All these operations are
related with thermal properties of foodstuff and their
effectiveness is a function of each thermal property involved in
the specific operation. The knowledge of thermophysical data of
foodstuffs and its dependence on temperature and on food
composition is of great importance for heat load calculations,
heat flux calculations and food quality assesments. Since water
is the main component of foodstuffs, the study of thermophysical
properties above and below of freezing point (fp) is necessary
to satisfy the needs of food processors, because during freezing,
thermal properties of water show dramatic changes [1,2].
 153
During the last 10 years, chilean fruit production for
export showed a constant increase; this development needs a
similar one on fruit industry, but there is scarce information
for thermal properties of fruits and fruit products in Chile.
Based on COST 90 Project on Physical Properties of Foods, the
objectives of this work were to carry out experimental studies
wi th the aim of adapting simple methods to measure thermal
properties of fruits.

MATERIALS AND METHODS

Apple puree was prepared from washed, peeled, seeded, steam

blanched and pulped Granny Smith apples. Raspberry puree was
obtained from raw Meeker variety.
Above freezing point, density was determined by picnometry,
and below freezing point it was determined using the
relationship reported by Hsieh et al. [3] where density is a
function of the weight ratio of water, ice and solids. Specific
heat was determined using an aluminium sample container inserted
into a polystyrene block at room temperature, as described by
Moline et al. [4,5]. Heat diffusivity measurements were made
using the thermal diffusivity apparatus des- cribed by Dickerson
[6], with an aluminium sample container placed into an insulated
and well stirred silicone oil bath. With the results obtained for
density, specific heat and heat diffusivity, heat conductivity
was calculed applying the existing relationship.

RESULTS

Tables 1 and 2 show the results obtained for apple and rasp-
berry puree, respectively, averaged over 0 and 25°C for values
above the freezing point, and averaged over -25 and 0 °c for
values below the freezing point.
Specific heat values for apples are similar to those
reported in the literature [7,8]. Polley et al. [7] reported
lower values for raspberries than those obtained in this work,
but they don't give any information about variety.

TABLE 1
Thermophysical properties of Granny Smith apple puree
Moisture content 88.79 %

Property Units Above fp Below fp

Density kg/m 3 1016.7 950.4

Specific heat J/kg K 3640 1985
Heat diffusivity m2 / s (* 1 0- 7 ) 1. 063 2.375
Heat conductivity W/m K 0.393 0.448
154
For heat diffusivity, results obtained with apples are
similar to those reported elsewhere [8] as well as values
calculated for heat conductivity. No reference on other thermal
properties of raspberries was found on literature.

TABLE 2
Thermophysical properties of Meeker raspberry puree
Moisture content 85.63 %
Property Units Above fp Below fp
Density kg!m 3 1054.0 980.1
Specific heat J/kg K 4900 3462
Heat diffusivity m2 Is (* 1 0- 7 ) 1. 335 2.036
Heat conductivity wlm K 0.689 0.691

Thermophysical properties of apple and raspberry puree were

determined wi th cheap and easy methods wi th a good accuracy. Data
reported is one of the first obtained for chilean fruit and it
is useful for local processors.

REFERENCES

1.- Meffert, H. F. Th. History, Aims, Results and Future of

Thermophysical Properties Work within COST 90. In: Physi-
cal Properties of Foods, ed. R. Jowitt, F. Escher, B.
Hallstrom, H.F.Th. Meffert, W. Spiess and G. Vos. Applied
Science Publisher, London, 1983, pp. 229-267.
2.- Mellor, J.D. Thermophysical Properties of Foodstuffs. I.
Introductory Review. Bulletin IIF/IIR, 1976, N°3.
3.- Hsieh, R.C.; Lerew, L.E. and Heldman, D.R. Prediction of
Freezing Times For Foods as Influenced by Product Proper-
perties. J. Fd. Process Eng., 1977, 1, (2); 183.
4.- Moline, S.W.; Sadwye, J.A.; Short, A.J. and Rinfret, A.P.
Thermal Properties of Foods at Low Temperatures. I. Speci-
fic Heat. Food. Technol. 1961, 15, (5); 228.
5.- Mohsenin, N.N. Thermal Properties of Foods and Agri-
cultural Materials, Gordon & Breach Science Publi-
shers. New York, 1980, pp. 75.
6.- Dickerson, R.W. An Apparatus for the Measurements of Ther-
mal Diffusivity of Foods. Food Technol., 1965, 19, (5);
198.
7.- Polley, S.L., Snyder, o.P. and Kotnour, P. A Compilation
of Thermal Properties of Foods. Food Technol., 1980, 34,
(11); 76-94.
8.- Ramaswamy, H.S. and Tung, M.A. Thermophysical Properties
of Apples in Relation to Freezing. J. Fd. Sci. 1981, 46,
724-728.
Project number 91/0492, sponsored by FONDECYT, CHILE
 A THERMOMECHANICAL METHOD FOR DETERMINING COLLAPSE
TEMPERATURES OF FOOD POWDERS

VICTOR T. HUANG
Grand Metropolitan Food Sector,
330 University Ave. SE, Minneapolis, Minnesota 55414, USA

ABSTRACT

A Thermomechanical Analyzer (TMA), modified by adding an oscillating load

with a frequency of 0.08 Hz, was used to measure the softness of samples as
a function of temperature during heating. The onset temperature of flow under
0.5 g weight is related to the glass transition temperature(Tg) as measured
by DSC. Sticky point (Ts) from literature is 7°C higher than Tg for the
amorphous sucrose/fructose at 7:1 ratio. Ts, or collapse, occurs when the
sample is 0.5-0.8 log cycle softer than that at the onset of flow.

INTRODUCTION

Physical changes of amorphous powders, such as structural collapse, sticking

or caking, and recrystallization, have been recognized to result from
increased diffusion-related mobility in the rubbery state at temperatures
above Tg [1]. Various thermal analysis techniques, such aa DSC[2,3], DMTA[3],
NMR[3], TMA[4,5,6], TSC[5], and dielectric spectroscopy[5] have been adapted
to analyze Tg of food systems. Structural collapse and stickiness are two
terms describing the same physical change; thus, they can be used
interchangeably. The objective of this study was to determine if the modified
TMA method can be used to measure collapse temperatures of food powders.

MATERIALS AND METHODS

Sucrose and fructos~, both Sigma grade I reagents, at a weight ratio of 7 to

1, were dissolved in distilled water (20% solution). The mixture was gently
heated and well-stirred until a clear solution was obtained. Ten grams of the
solution in an Aluminum container were immediately frozen in liquid nitrogen
and subsequently freeze-dried at below 0.1 mbar using a laboratory freeze-
dryer (Denton Vacuum, model DFD-2). After freeze-drying, containers were
transferred into a vacuum desiccator. The moisture content of the initial
san"le was determined by vacuum oven method. Samples were incubated in
various relative humidity chambers to achieve different moisture contents.

Previously reported method of TMA modification and testing procedure were

used [4,5,6]. Solutions with concentration from 90 to 100% were run in the
temperature range from -100° to 60°C at a heating rate of 2°C/min. The width
of the first-order derivative of the TMA curve, reflecting the response of
the sample to the oscillating stress, was measured as a function of
temperature. The width has units of p,m/sec; thus, the relative width at
different temperatures to that at Tg is defined as relative softness. The
inverse of relative softness can be viewed as relative viscosity.
156
RESULTS & DISCUSSION
Fig. 1 shows the plasticizing effect of water on the Tg of amorphous
sucrose/fructose at 7:1 ratio as measured by DSC[7] and TMA methods. The
average temperature values of oriset and end of transition from DSC were used.
Tg decreases as moisture content increases. A regression line was determined
using the Gordon and Taylor equation

where T9 1 : Tg of moisture-free sucrose/fructose co-glass at 7:1 ratio 333K

T92 : Tg of water = 135K
WI solid fraction
W2 moisture fraction
k : = 4.97
70~--------------------------------------------~

60

50

40
•
A A
G
'-' 30 •
~
+
~
+' 20

l
~

10
....
0

-10

-20
+
-3~

0 2 6 8 10
Mol stlTe content (~

Figure 1 Effect of moisture content on Tg and Ts. (+ Tg from TMA . • midpoint

Tg from from DSC [7]. ~ Ts from force measurment [8]).

The Ts values [8] in Fig. 1 also show the Ts-lowering effect of water. These
Tg and Ts regression lines show that Ts is about 7°C higher than Tg.
Figure 2 shows the TMA thermogram of an amorphous sucrose/fructose at 7:1
ratio with 6% moistuFe. Below the Tg of 13°C, the sample is hard and in the
glassy region. At T > Tg, relative softness increases. Figure 3 shows the
plot of log (relative softness) vs l/T for 3 samples. The activation energy
(Ea) for these samples has an average of about 35 kcal/mole. At 7°C higher
than respective Tg's, softness values are about 0.5 to 0.8 log cycles lower
than those at Tg.
If the viscosity at Tg is about 10A11 Pa.s and about 10A6 to 10A8 Pa.8 at
Ts[9], then the difference in viscosity should be much larger than the 0.5-
0.8 log cycle observed in this study, unless the viscosity at Tg for fructose
containing system follows that of pure fructose with viscosity at Tg ~ 10A8
Pa.s[l]. This assumption is reasonable, since at Ts= 7+Tg, viscosity ~ lOA
(7.2) to 10 A (7.5), if the Ea is 35 kcal/mole.
This preliminary study demonstrated that product collapses and becomes sticky
at a viscosity value 0.5-0.8 log cycle lower than that at Tg. However, the
magnitude of softness change needs to be confirmed in other systems before it
can be broadly applied.
 157

24
Tamp (Oeg C)

Figure 2. Original and first-order derivative of the TMA thermogram of

sucrose/fructose (7:1 ratio) at 6% moisture.

0.2
0.1
0
+ •
" (fl -0.1
(fl
-0.2 •
~ +
• •
(>
-0.3
....+-'0
III
Q)
-0.4
-0.5
• Q Q

> -0.6 +
+
+-' -0.7
Il
-o.e
u
Q)
L • •
01
-0.9
.,
0
-1
...J
-1.1
-1.2
• Q

-1.3
-1.4
0.00336 0.0034 0.00344 0.00348 0.003:52 0.003:56 0.0036
11T (k~-1)

Figure 3. Effect of temperature on sample softness at T > Tg.

REFERENCES
1. Levine, H. and Slade L., A polymer physico-chemical approach to the study
of commercial starch hydrolysis products (SHPs). Carbohydr. Polym., 1986,
6:213-244.
2. Roos,Y., Effect of moisture on the thermal behavior of strawberries
studied using DSC. J. Food Sci., 1987, 52, 146.
3. Kalichevsky,M.T., Jaroszkiewicz,E.M., Ablett,S., Blanshard,J.M.V.,
Lillford,P.J., The glass transition of amylopectin measured by DSC, DMTA,
and NMR. Carbohydr. Polym., 1992, 18:77-88.
4. Le Meste,M., V.T.Huang, Thermomechanical analysis of frozen sucrose
solutions. J. Food Sci., 1992, 57(5) ,1230-1233.
5. Huang,V.T., F.McIntyre, H.Levine, L.Slade, L.Hynes, Glass transition
temperatures of starch, gluten, and bread as determined by TSC, TMA, and
Dielectric Spectroscopic Methods, presented at AACC Annual Meeting,1992.
6. Le Meste, M., Huang, V.T., Panama, J., Anderson, G., Lentz, R., Glass
transition of bread. Cereal Foods World, 1992, 37(3) :264.
7. Roos, Y., Karel,M., Plasticizing effect of water on thermal behavior and
crystallization of amorphous food models. J. Food Sci., 1991, 56(1), 38.
8. Downton,G.E., Flores-Lunn, J.L., and King, C.J., Mechanism of stickiness
in hygroscopic, amorphous powders. Ind. Eng. Chern. Fundam. 1982, 21:447.
 THERMAL STUDIES OF NATURAL FRUITS

M.M.sA,A. M.SERENO*
University of Porto, Faculty of Engineering, Department of Chemical Engineering
Rua dos Bragas, 4099 PORTO, Portugal
Tel: 351-2-2007437, Fax: 351-2-319280, E-mail: sereno@fe.up.pt

ABSTRACT

Differential Scanning Calorimetry (DSC) was used to measure phase transitions and unfreezable
water in fresh samples of onion, grape and strawberry and after equilibration at different water
activities. From each DSC curve, glass transition and melting temperatures, specific heat
capacities, latent heats of melting and unfrozen water contents were obtained. Preliminary state
diagrams for those fruits and vegetable were established. Gordon-Taylor equation proved to be
able to predict glass transition temperature for the water-food systems studied based on values
corresponding to pure components.

INTRODUCTION

Freeze-dried foods are easily rehydrated due to the formation of a low density bulk structure of a
highly viscous metastable amorphous material, with either "rubber" or "glass" characteristics,
depending on the final temperature and moisture content (1). Current interpretations for this
behaviour view the material as a water plasticized food polymer and are well documented on a
recent and comprehensive review by Slade and Levine (2). The change from glass to rubber
occurs at the so called glass transition temperature (Tg) and its value corresponding to a
maximally freeze-concentrated amorphous matrix (Tg'), a characteristic of each material, is
becoming a leading physicochemical parameter with respect to product properties and stability
(2). All the possible physical states of the material are well described by a temperature vs.
concentration state diagram as proposed by Franks et al. (3). Recent studies report state
diagrams obtained for solutions and model systems including namely low-sugar solutions (2,4),
starch and related polymers (5). However the phenomena has not yet been so widely studied
and discussed with experimental data for natural foods.

MATERIALS AND METHODS

Materials and Sample Preparation

Three common fresh foods were used in this study: Povoa Red onion (Allium cepa) , a local
cultivar; white grape (Vitis vinifera) of the Azal wine-making cultivar; common garden strawberry
(Fragaria x ananassa). Round slices of onion and strawberry, 7 mm thick, were cut perpendicular
to their axis and submitted to freezing followed by freeze-drying. Freeze-dried onion and
strawberry were powdered and 5-30 mg samples were equilibrated over saturated salt solutions
(Aw = 0.12 - 0.90). The same procedure was used for slices of freeze-dried onion and for fresh
non-dried grape. Strawberry and onion were frozen at -40°C fOllowed by freeze-drying under 65
Pa in a TELA BE mod. LF10 plate freeze-dryer.

Moisture Determination and Differential Scanning Calorimetry

Moisture contents were determined in a vacuum-oven at 70°C and 13.3kPa. A SHIMADZU Mod.
DSC-50 + LTC50 differential scanning calorimeter was used with 30 JLI aluminium pans both for
 159
the sample and as reference. The samples were scanned at 5°C/min from -,20°C to 100°C with
helium as carrier gas. The thermograms were analysed for glass transition (Tg), devitrification
and melting temperatures (Td, Tm), change of specific heat capacity due to glass transition (a.cp)
and latent heat of melting of ice (a.Hm). At least five independent thermograms were obtained for
each level of moisture content of each material.

Prediction of Glass Transition Temperatures and Unfreezable Water Content

Gordon-Taylor (6) equation was used to predict glass transition temperatures;

(, )

where Tg s ' Tgw are the glass transition temperatures (OC) and X s ' Xw the weight fractions of
solids an water respectively, and K is an empirical parameter obtained by numerical regression. A
value of Tg=-135°C was used for pure water (7). Unfrozen water content (Xg') corresponding to
the maximally freeze concentrated material was calculated by extrapolating to zero a linear plot
of .6.Hm vs. moisture content.

RESULTS AND DISCUSSION

Figure 1 shows DSC thermograms for grape; similar plots were obtained for onion and
strawberry. Analysis of such plots indicated that samples humidified in atmospheres of
Aw<O.85 showed no ice formation by freezing and, upon rewarming, only one glass transition
was observed, decreasing Tg with increasing moisture content. Melting of ice was found in
samples humidified at Aw20.85, for all products. The samples humidified at Aw=O.85 showed
an exothermic peak after the glass transition and before the melting endotherm. According to
Flink (8), this is due to water crystallization on rewarming. Dependence of onset glass transition
temperature on water activity was adequately described by a linear relationship. Figures 2 and 3
are the state-diagram plots obtained for onion and grape and show glass transition and melting
temperatures against moisture content. Assuming that the total heat of melting of ice is linearly
dependent on the moisture content of the sample, values of Xg' were estimated.

An extended version of this communication, with full tabular data and a complete set of plots is
available and can be supplied on request.

CONCLUSIONS

In this study it was possible to determine. reproducibly, glass transition temperatures and
associated phase characteristics for three natural fresh and partially dehydrated foods. From this
information state diagrams for those materials were ploned. Those diagrams have the same
general characteristics as those previously reported for pure solutions and model food systems.
The empirical Gordon-Taylor equation was able to predict the dependence of Tg on moisture
content with K parameters determined by least squares regression. Good linear correlation
between heats of melting of fresh and partially dehydrated material vs. moisture content and
onset glass transition temperature vs. water activity were determined.
A single Tg line valid for both powdered and flaked onion having distinct sorption isotherms is a
new evidence of the shortcomings of the Aw concept to describe water management in foods.
Annealing was found to be necessary to achieve maximum ice formation and corresponding Tg'
and Xg' experimentally determined. These studies are currently under way.

REFERENCES

1. Roos, Y., Karel, M., Differential scanning calorimetry study of phase transitions affecting the quality of
dehydrated foods. Biotechnol. Prog., 1990, 159-163.
2. Slade, L. and Levine, H., Beyond water activity. In Critical Reviews in Food Science and Nutrition, ed.
F.M. Clydesdale, CRC Press, Boston, 1991. pp. 115-360.
160
3. Franks, F., Asquith, M.H., Hammond, C.C., Skaer, H.B. and Echlin, P., Polymeric cryoprotectants in the
preservation of biological ultrastructure (I). J. Microscooy, 1977, 110, 223.
4. Roos, Y. and Karel, M., Phase transitions of amorphous sucrose and frozen sucrose solutions. ~
~, 1991, 56, 266-267.
5. Blanshard, J.M. V., Starch granule structure and function: physicochemical approach. In Starch: Prooerties
and Potential, ed. T. Galliard, John Wiley & Sons, New York, 1987, pp. 16-54.
6. Gordon, M. and Taylor, J.S., Ideal copolymers and the second-order transitions of synthetic rubbers. I.
Non-crystalline copolymers. J.Aool. Chem., 1952, 2, 493-500.
7. Johari, G.P., Halbrucker, A. and Mayer, E., The glass-liquid transition of hyperquenched water. ~,
1987, 330, 552-553.
8. Flink, J.M., in Physical Prooerties of Foods, ed. M. Peleg and E.B. Bagley, AVI Pub. Co., Inc. Westport,
Conn., 1983, 17, pp. 473-521.

ACKNOWLEDGMENTS

The authors acknowledge the helpful suggestions received from Profs. Marcus Karel and John Blanshard
when planning this study and the financial support received from NATO Scientific Affairs division, project
PO-PORTOFOOD and JNICT (Portugal). Funda!,:30 Calouste Gulbenkian (Lisbonl and University of Porto have
also contributed to attend this conference.

25
- - - -_ _--.:-15.23°C Aw
25

i
= ~-2~3~,Ol~~----__
Aw

..J
- -_______ ..J
~ ~
::E ::E 0.85
a:
w
I
....
0
:::::=: a:
w
I
....
o
x x
w w 0.90
S
.......
0
..J
0,53
~
w 0,61
I Fresh

0.76
(
-90

TEMPERATURE (OC)
-12 14 . -100 -75 -SO -25 o
TEr·..1PERATURE (OC)

Figure 1. Thermograms for grapes humidified at different water activities.

50 TempE~ature loc) 50 Te~erature ~CI

Empil'lcal pararreter of Empirical pararreter of
Gordan-Taylor equation Gordan-Taylor equation
K' 4.0 K = 3.3
o

" -50

-100
~'= c.iB%

m
-150~~--~--~~--~--~~--~~--~
o ID ~ E ~ 50 00 00 ~ ~
MOisture I%J

Figure 2. Preliminary state diagram for onion Figure 3. Preliminary state diagram tor
( 0 Tg (flakes), c. Tg (powder), 0 Tml. grape ( 0 Tg and 0 Tml.
 THERMAL CONVERSION OF STARCH AT LOW WATER ACTIVITY.

K. POUTANEN, O. MYLLYMAKI, K. AUTIO, P. MYLLARINEN, T. SUORTfI.

VTT Food Research Laboratory, P.O.Box 203, 02151 Espoo, Finland.

ABSTRACT

The water solubility and swelling power of barley and oat starches were increased by
treatment in aqueous ethanol at 150-170 °C. Viscosity curves of 8 % water dispersions
of the partially depolymerized granular starch products at room temperature were
comparable to those of native starch heated at 98 °C.

INTRODUCTION

The structural changes of starch during thermal processing are very much dependent on
the starch:water ratio, as reviewed recently by ~ Kokini et al. (1). The effect of water
and other plasticizing solvents on the conversion kinetics and stability of thermally
processed starch materials has been discussed frequantly in the literature (2). The
minimum amount of water needed for starch gelatinization at atmospheric pressure has
been reported to be near 30 %. High pressure and shear forces, however, can
accomplish starch gelatinization and plastification also at lower water levels (3).

Thermal modification at elevated pressures in the presence of low

molecular weight solutes offers possibilities of producing starches with very much
altered physical properties and functionality. For example cold-swelling granular starch
has been produced in the presence of selected aqueous alcohols (4). Extrusion in the
presence of a suitable plasticizer, such as glycerol, can be used to produce a
thermoplastic melt with properties resembling those of synthetic polymers (5).

We have studied thermal conversions of starch in the presence of ethanol

or glycerol. In this paper some results on the production and properties of a granular
cold-swelling starch are reported.

MATERIALS AND METHODS

Commercial barley and oat starch manufactured by Alko Ltd (Finland) were used.
Thermal modification was carried out in a 600 ml pressure reactor (Parr Company,
USA) with neglible shear. 80 g starch was suspended in 320 ml of 80 or 94% w/w
ethanol solution and heated to the desired temperature (150-180 0c) with muffle
162
resistance heating at the rate of 7°C/min. After the thennal treatment the suspension
was cooled to 30°C with an internal cooling coil at the rate of 25°C/min. The
suspension was fIltered through a Whatman no 4 fIlter paper in a Buchner funnel and
the solids were washed in 94% ethanol and air-dried.

Gelatinization enthalpy was determined by differential scanning

calorimetry at 70% moisture (Mettler T A 4000 thennal analyzer system with DSC 30S
cell, heating rate 10 °C/min). Swelling power and solubility were determined using the
principle of Leach et al. (6). Viscosity of 8% starch dispersions was measured by
Bohlin VOR rheometer. The microstructure of the starch dispersions was studied by
light microscopy with iodine staining as described by Autio (7). Changes in molecular
size distribution were monitored by GPC-HPLC using pHydrogel columns (Millipore
Waters) and 50 mM NaOH as eluent as described earlier (8).

RESULTS AND DISCUSSION

During a 3 h treatment in 94% ethanol at ISO-170°C starch retained its granular fonn
but its physical properties changed remarkably. The decrease in gelatinization enthalpy
(Fig.l) was accompanied by increased swelling and solubility at 25°C (Fig.2). Starch
depolymerization, especially noticeable in the amylopectin moiety of starch, increased
with processing temperature (results not shown). Treatment of starch in 80% ethanol at
160°C yielded a granular cold-swelling product which did not show any endothennic
transitions in the DSC-analysis.

12r-------------------------,
MJ
Jig

170'C 1eOC

a b c d e a b c d e

Fig. 1. Gelatinization enthalpy of Fig. 2. Swelling power of native and

native and thennally modified barley thennally modified barley starches at
starches. a) native, b-d) processed 25°C. a) native, b-d) processed for 3h
for 3h in 94% ethanol, e) processed in in 94% ethanol, e) processed in 80%
80% ethanol. ethanol.
 163

The viscous properties of the cold-swelling starches at 25°C resembled those of

native starch at 80°C (Fig.3). Because thermal treatment in ethanol increased the
solubility of amylopectin, 8% starch dispersions made of thermally modified starches
were composite networks in which the continuous phase was composed of both
amylose and amylopectin.
1.6:..------------------,

:
o W ~ ~ 00 100 1W 1~ 1~
Shear Rate (lis)

Fig. 3. Viscosity curves of 8% starch dispersions. (+) Native barley starch measured at
80°C after 15 min preheating at 100°C, (x) modified barley and (_) modified oat
starch measured at 25°C. Modification was in 80 % ethanol at 160°C for 2h.

REFERENCES

1. Kokini, J.L., Lai, L.-S. and Chedid, L.L., Effect of starch structure on starch
rheological properties. Food Techno!. 1992,46: 124-139.

2. Levine, H. and Slade, L., Influences of the glassy and rubbery states on the thermal,
mechanical, and structural porperties of doughs and baked products. In Dough
Rheology and Baked Product Texture, eds. Faridi, H. and Faubion, J.F., AVI, Van
Nostrand Reinhold, New York 1990, pp. 157 - 330.

3. Vainionpiiii, J., Forssell, P. and Virtanen, T., High-pressure gelatinization of barley

starch at low moisture levels and elevated temperatures. Starke 1993, 45: 19-24.

4. Eastman, J.E. and Moore, C.O., Cold-water soluble granular starch for gelled food
compositions. U.S. Patent No: 4,465,707, 1984.

5. Wittwer, F. and Tomka, I., Polymer composition for injection molding. U.S. Patent
No: 4,673,438, 1987.

6. Leach, H. W., McCowen, L.D., Schoch, T.J., Structure of the starch granule. I.
Swelling and solubility patterns of various starches. Cereal Chern., 1959, 36: 534-544.
7. Autio, K., Rheological and microstructural changes of oat and barley starches during
heating and cooling. Food Structure, 1990, 9: 297-304.

8. Suortti, T. and Pessa, E., The OPC-analysis of starches with alkaline eluents.
J. Chromatogr., 1991, 536: 251-254.
 CHARACTERIZATION OF W/O/W EMULSIONS WITH
A VIEW TO POSSIBLE FOOD APPLICATIONS

SACHIO MATSUMOTO
Department of Agricultural Chemistry,
College of Agriculture,
The University of Osaka Prefecture
Gakuen-cho, Sakai-shi, Osaka 593, Japan

ABSTRACT

Electrostatic interactions between the dispersed vesicular globules of W/O/W

emulsions can be controlled by the magnitudes of isoelectric point of pro-
teins immobilized in the inner aqueous phase of the globules. A certain
amount of saccharides dissolved i~ the two aqueous phases of the emulsions
plays a role in an increase in hydration repulsion between the globules
due to the formation of thick hydration layer on the globules. An attempt
at estimating total potential energy of interactions between the vesicular
globules has been made in consideration of the hydration force.

INTRODUCTION

The dispersed globules in W/O/W emulsions can be characterized by a vesic-

ular structure with single or multiple inner aqueous phase separated from
the aqueous suspending fluid by a thin oil layer [I]. The W/O/W emulsions,
therefore, could find applications within the food science and technology,
as for example; the inner aqueous phase could immobilize certain amounts of
ingredients, so that the emulsions could possibly be used as vehicles for
nutrient administration in special dietary situations. The W/O/W emulsions
also may be one of the suitable morphological models for foods, since many
of foods could be generalized as disperse systems with a multi-phase cellu-
lar structure. The formation and properties of W/O/W emulsions were already
studied with a view to elucidating the conditions for high emulsion yields
when using the two stage procedure [2] or the phase inversion technique [3],
and also to ascertain the structural and functional properties of the oil
layer surrounding the inner aqueous phase of the vesicular globules [4,5].
It is the purpose of this study to obtain further insights into the
interactions between the dispersed vesicular globules in W/O/W emulsions in
the light of the electric charge and the hydration force of the globules
in the presence of proteins and saccharides in the aqueous phases.
 165

MATERIALS AND METHODS

W/O/W emulsions to be tested were divided into liquid-paraffin and olive

-oil systems, respectively. The former was prepared by use of Span 80
(sorbitan monooleate) and Tween 80 (polyoxyethylene sorbitan monooleate),
while the latter was stabilized by TGCR (tetraglyceryl condensed ricinolate)
and DGMO (decaglyceryl monooleate). Each system provided by the two stage
procedure [2] was also subdivided into four samples according to a kind of
proteins immobilized in the inner aqueous phase with a concentration of 1%
w/w, i.e. control (without protein), BSA (bovine serum albumin, pI:4.4-4.9),
chymotrypsin (pI:8.1-8.8), and lysozyme (pI:10.5-ll.0) samples. One of a
series of saccharides (ribose, xylose, fructose, glucose, sucrose, maltose)
was introduced into the two aqueous phases of the control sample with
various concentrations in the case for examining hydration of the globules
using zeta potential measurement. When the hydration force was examined,
0.5 mol/Q glucose was ressolved in the two aqueous phases of all the
samples. An optical microscopic observation indicated that all of the
dispersed globules in freshly prepared samples have the vesicular structure
with single or multiple compartments of the inner aqueous phase within a
range of diameters from 2 to 30 ~m. An assembly with a rectangular cell
was employed for evaluating the electrophoretic mobility of the vesicular
globules in each sample at neutral pH at room temperature, while the
steady-flow viscosity of the samples was measured as a function of aging
time by means of a cone-and-plate viscometer at 25.0°C.

RESULTS AND DISCUSSION

The effect of proteins on the zeta potential of the globules appears in re-
lation to the different magnitudes for the isoelectric point of the proteins
immobilized in the inner aqueous phase, although the values obtained with
the liquid-paraffin system are always smaller than those with the olive-oil
system. The zeta potential of the globules, however, can be ranked with
the magnitudes of the isoelectric point of the proteins, irrespective of the
kinds of emulsifiers used, while a rapid coagulation occurs irreversively
among the vesicular globules in the lysozyme sample showing the lowest zeta
potential in a series of the samples to be tested. The surface activity
of protein molecules may result so as to form an adsorbed phase on the
inside surface of the oil layer in the vesicular globules; the adsorbed
protein molecules thus seem to playa role in the surface potential of the
globules according to the value of pH in the two aqueous phases. The author
tried to estimate the potential energy of interactons between the two vesic-
ular globules in each sample as a function of the separation of the slipping
plane on the basis of the DLVO theory, assuming that there is no energy
barrier against the coagulation between the globules encapsulating lysozyme
in each system because of the instabiiity in the dispersion state, i.e.

v VR + VA = 211: £ oDr ~ 2 In[1+exP(-KH)] - rA/12H (1)

where VR and VA are the potential energies of repulsion and attraction, £ 0

is the permittivity of space, D is the relative permittivity of the aqueous
phases, r and ~ are the radius and zeta potential of the globules, K is
the Debye's parameter, H is the separation, and A is the Hamaker constant.
Thus. the value for the Hamaker constant in the van der Waals attraction
166
could be obtained as being 4 X 10- 20 J for the liquid-paraffin system and
1.3 X 10- 19 J for the olive-oil system, respectively.
The zeta potential of the vesicular globules in the control sample
reduces systematically with increasing concentration of saccharides dis-
solved in the two aqueous phases. This seems to be brought about by an
expansion of the slipping plane on the surface of the globules according
to an increasing thickness of the hydration layer. An emprical relation-
ship between the zeta potential of the globules and the concentration of
saccharides ranging from 0 to 1 mol/Q in the two aqueous phases suggests
that with increasing concentration the slipping plane moves asymptotically
to about 7.5 nm outside from the standard position of slipping plane at
which the concentration of saccharides is zero. An increasing thickness
of hydration layer calculated in the above will be more or less far from
the location where the van der Waals attraction appears significantly
between the globules. The fact that the viscosity of lysozyme sample con-
taining 0.5 mol/Q glucose in the two aqueous phases becomes to be a steady
state against aging time after the emulsification, while the creamed glob-
ules are ready to be redispersed by a gentle agitation with container of the
sample. The author, therefore, assumed 300kT, 600kT and 900kT for the re-
pulsion at a separation (H) of 5 nm between the globules for estimating
potential energy of the hydration repulsion (Vh) among the globules in the
lysozyme sample in each system according to the following relation [6]:

Vh = (nr A,2 Po/kT) exp(-H/A,) (2)

where A, is the thickness of hydration layer on the globules, and Po is

the hydration force. Then, the total potential energy of interaction (VR
+ VA + Vh) was calculated for comparing the potential energy curves against
separation between the vesicular globules. The results obtained suggest
that a range of values, from 7 X 10- 17 (600kT at maximal Vh) to 1.1 X 10- 16 N
(900kT at maximal Vh), should be necessary for the hydration force (Po)
so as to stabilize the lysozyme sample.

REFERENCES

1. Matsumoto, S., W/O/W-type mUltiple emulsions. In Nonionic Surfactants,

Physical Chemistry, ed. M.J. Schick, Marcel Dekker, Inc., New York,
1987, pp. 549-600.
2. Matsumoto, S., Kita, Y. and Yonezawa, D., An attempt at preparing
water-in-oil-in-water multiple-phase emulsions. ~ Colloid Interface
Sci., 1976, 57, 353-61.
3. Matsumoto, S., Development of W/O/W-type dispersion during phase
inversion of concentrated W/O emulsions. J. Colloid Interface Sci.,
1983, 94, 362 - 6 8 •
4. Matsumoto, S., Inoue, T., Kohda, M. and Ikura, K., Water permeability
of oil layers in w/O/W emulsions under osmotic pressure gradients.
~ Colloid Interface Sci., 1980, 77, 555-63.
5. Ueda, K. and Matsumoto, S., Effect of sugars on the physicochemical
properties of W/O/W emulsions. ~ Colloid Interface Sci., 1991, 147,
333-40.
6. Rand, R.P. and Parsegian, V.A., Hydration forces between phospholipid
bilayers. Biochim. Biophys. Acta, 1989, 988, 351-76.
 APPLICATION OF MEMBRANE EMULSI}'ICATION
METHOD FOR PREPARING FOOD EMULSIONS
AND EMULSION CHARACTERISTICS

KANIcm SUZUKI, IKUKO SHUTO, and YOsmo HAGURA

Department of Food Science,
Faculty of Applied Biological Science Hiroshima University,
1-4-4 Kagamiyama, Higashi-Hiroshima, 724 Japan

ABSTRACT

A new method was applied for preparing food emulsions with sharp particle diameter
distribution. Mean particle diameters of emulsions were about twofold of the mean pore
sizes of the membranes used. Increasing the flux of dispersed phase through membrane
or pressure applied decreased the diameter. Because of high stability of emulsion
particles, little change in particle diameter distribution occurred. Concentration of
emulsifying agents influenced the creaming rate and volume of cream layer even if the
emulsions were prepared with the same membrane

INTRODUCTION

Membrane emulsification, a newly developed emulsifying method, prepares emulsions

with sharp particle diameter distribution by dispersing oil or water phase into continuous
phase through a rnicroporous glass membrane I). Since, this method has tested mainly
kerosine-water emulsion systems, only a few studies on preparing food emulsions have
been reported 2 ,3). This paper investigated the usefulness and problems of the membrane
emulsification method for preparing o/w food emulsions. Influences of constituent
concentration, emulsifying pressure, and pore size of membranes on particle diameter
distribution, creaming phenomena, and stability of emulsion were studied.

MATERIALS AND METHODS

Com oil was used for oil phases. Low HLB polyglycerol esters were dissolved in oil
168
phase, and high HLB polyglycerol esters in water. Concentrations of emulsifying agents
ranged from 0.1 wt% to 2.0 wt%. The glass microporous membranes used were the tube
type SPG membranes ( Ise Chemical Corp., Japan) with 10 mm 1.0., 170 mm in length,
and thickness of 0.5 mm. For preparing o/w emulsions by the membrane emulsification
method, a hydrophilic membrane had to be used. Thus, the SPG membranes were dipped
in 2N-HCI solution at 70°C for 2 hours to make membranes hydrophile, then rinsed in
distilled water before emulsifying process. Figure 1 shows the schematic illustration of
the membrane emulsifying process, and the emulsifying apparatus used in this study.

manometer

ntinu module
micrQpOrou
memOrane

Fig. 1 Schematic illustration of the membrane emulsifying process, and the experimental
apparatus.

RESULTS AND DISCUSSION

For our emulsion systems, oil phases did not disperse smoothly into water phase through
membrane, because membrane wetted with com oil immediately after the emulsifying
process started, and lost hydrophilicity of the membrane. However, when o/w emulsions
with larger particle diameter than mean pore size of the membrane prepared by an
appropriate emulsifying method, e.g., mixing, were used as dispersed phases (pre-
emulsified emulsion), o/w emulsions with very sharp particle diameter distribution were
obtained. Figure 2 compares the particle diameter distributions of emulsions before and
after the membrane emulsification. Therefore, we used the rough particle size o/w
emulsions as dispersed phases in our emulsifying procedures. Mean particle diameters
were about twofold of mean pore sizes of the membranes. However, the increase in
dispersed phase flux through membrane or pressure for dispersion decreased the mean
particle diameters to about 1.5 times of the mean pore size. Because of sharp particle
diameter distribution, we could observe creaming phenomena easily. The concentrated
phase separated distinctly from the dilute phase (water phase). Emulsion particles were
very stable, and no appreciable coalescence was observed during keeping stationary
over several weeks, even though they formed the concentrated cream layer. By shaking
the creamed emulsions, we measured almost the same particle diameter distributions of
those of freshly prepared emulsions (Fig.3). Concentration and kind of emulsifier
influenced the creaming rate. Higher the concentrations, lower the creaming rate of
 169
emulsions, even if they were emulsified with the same membrane. Emulsifier
concentration affected also the volume of cream layers in equilibrium state.

100 60
_ _ membrane emulsified ......- fresh
......,
~
80 ,
--/:s- pre-emulsified .......,
~
··6·· after 21 days

....... 40
60
'--'

>. ;;...
C,)
u
=
u 40 Q

= 20
Q)
;:l
C" 0"
u

~ p.",Jr~ Ir"'~" &:

cl: 20
'l>i>
0 0
~ .... 10 20 30 40 50 0 5 10 15 20
Particle Diameter [~m ] Particle Diameter [~m ]

Fig. 2 Comparison of particle diameter Fig. 3 Comparison of particle diameter

distributions of emulsions before (pre- distribution of fresh emulsion with that
emulsified) and after membrane of 21-day standing emulsion (emulsifier
emulsification (mean pore size: 4.7 !-IDl). conc. : 2% in oil and water phases).

CONCLUSIONS

Com oil-in-water emulsions with sharp particle diameter distribution were prepared by
membrane emulsification method. The mean particle diameters were about twofold of the
mean pore sizes of the membranes. Emulsion particles were very stable, even though they
formed the concentrated cream layer. No appreciable coalescence was observed during
standing for several weeks.

ACKNOWLEDGMENT

The author wish to express his sincere thanks to Mr. S. Nakai, Technical Official,
Hiroshima University, who contributed to make the experimental apparatus.

REFERENCES

1) T.Nakashima & M.Shimizu : Proceedings of 21st Autumn Meeting of Kagaku

Kogakukai, 86 (1988).
2) K.Suzuki, I.Shuto & Y.Hagura : Proceedings of 1992 Annual Meeting of Nippon
Nogei Kagakukai, 23 (1992).
3) K.Suzuki, I.Shuto & Y.Hagura: Proceedings of 25th Autumn Meeting of Kagaku
Kogakukai, 180 (1992).
 EVALUATION OF MODEL FOOD EMULSION STABILITY BY HYGROMETRIC
MEASUREMENTS.

C.R. LERICI, C. CORRADINI, P. PITTIA

Department of Food Science, University of Udine
via Marangoni 97 - Udine (Italy)

INTRODUCTION

Many natural and formulated foods, such as milk, cream, mayonnaise, butter
and margarine, consist of an intimate mixture of at least two non-miscible
phases in the form of finely or coarsely dispersed systems defined as
emulsions. Emulsions are not thermodynamically stable systems. In fact, after
a time, the droplets of the inner phase can recombine and 'coalesce', or
cluster toghether (flocculation) and ultimately give two separate phases due to
a density difference between the inner and the outer phase [1-2J. The
stability of an emulsion depends on different factors (degree of dispersion of
the inner phase, outer phase viscosity, a single phase volume to total volume
ratio, specific gravity of the phases and temperature), and on the presence of
emulsifying agents able to form an interfacial layer between the two phases.
In literature numerous methods have been proposed to estimate the emulsion
stability: they evaluate the chemical, physico-chemical and physical changes
that in the global emulsion volume occur spontaneously or -in the aging
tests- due to thermal (heating or freezing) or mechanical stress [3-61.
In this work, the measurement of the rate to reach the Equilibrium Relative
Humidity (%ERH) expressed as d(%RH)/dt vs %RH was used to evaluate the
degree of dispersion between aqueous and lipidic phase in a model food
emulsion.
MATERIALS AND METHODS

For the aqueous phase, a whey protein dispersion (1 % in protein w/v in

distilled water) was prepared and soybean oil for the lipid phase was used. To
prepare the emulsions, soybean oil and whey protein dispersion were mixed
and the mixture (60 ml) was then homogenized with a Politron PT 20
(Kinematica, Switzerland) at speed 5 for 1 minute. Two different types of
model emulsion were obtained: indicating with f the oil volume/total emulsion
volume, the first was an oil-in-water emulsion (ft::0.67); the second was a
water-in-oil emulsion (ft::0.90). In order to avoid microbial spoilage during
storage, 0.4% (w/w) of sodium benzoate was added to the aqueous phase before
the homogeneization.
A prefixed volume (25 m!) of sample was then placed and hermetically sealed
in a special plastic cell. The hygrometer (ROTRONIC HYGROSCOP DT, connected
to a ROTRONIC DMS-IOOH probe, PBI Milan, Italy) and sample sealed cell were
put into a hermetic box in which %RH of the air was maintained at a very low
value (about 10 %). The cell with the sample was opened immediately before
the measurement of the %RH and then transferred into the hygrometer case.
The %RH values were recorded and the data expressed as d(%RH)/dt vs %RH.
 171
In order to estimate changes in the rate to reach %ERH, changes due to
breakdown processes that occur in the model emulsion, %RH measurements of
the samples stored 48 hrs at room temperature (25°C) were carried out.
The considered samples are indicated in the text as: A, separated lipidic
and aqueous phases; B, whey protein dispersion; C, emulsion F=O.67; D,
emulsion F=O.90. Data were the average of five replicates.

RESULTS AND DISCUSSION

In figure 1 the %RH increase vs time of A,B,C and D samples to reach %ERH
value are reported.
100~----~--------------------------~

80

60

40

A
20
\
0
0 5 10 15 20
TIME (min)
Figure 1. %RH increase vs time of A,B,C and D samples.

It is possible to note that B, C, and D samples showed very similar trend

while separated phase (sample A), compared to ot.her samples, presented a very
slow increase of O/ORH vs time due to barrier effect of lipidic phase to water
vapour transfer from dispersion to head space. In Figure 2 the d(%RH)/dt
values, derived from the data of Figure 1, vs %RH are reported.
130,-------------------------------,
120 - A
*8
.C
• 0
100

--
'U

X
a:
M 60
-:v
80

40

20

0
0 20 40 60 80 100
XR.H.
Figure 2. d(O/ORH)/dt values vs %RH.
172
It is possible to observe that the rate of %RH increase (d(%RH)/dt) decreased
very quickly as the %RH approaches the %ERH value. No significant differences
in the decrease rate were observed between B,C and D samples, while sample
A showed a very slow rate, as expected. Nevertheless it is interesting to note
that repeating the measurements on the same model emulsions (samples C and
D) after 48 hrs of storage at room temperature, d(%RH)/dt values vs %RH
decreased significantly in the range 20-80 %RH (Figure 3).
130 .c -0
120 o C 48 h IJ 0 48 h

100

-
.;; 80
:2
ei
'ii
60

40

20

O+---r--.,__-..----=~...... +--.,__-r__-.__:~IIAI ........

o 20 40 60 80 1000 20 40 60 80 100
SR.H.
Figure 3. d(%RH)/dt values vs %RH of C and D samples just
prepared and after 48 hours.

CONCLUSIONS

Measurement of %RH increase in head space of food emulsion vs time, from an

initial very low %RH value -obtained conditioning hygrometric system and
head space of the sample- to the %ERH value, could give interesting
information regarding the physical state of the emulsion and about changes
due to aging or physical treatments. In fact, movement of the lipidic phase
towards the surface causes a significant slowing in the %RH increasing rate in
the head space of the emulsion. This is due to the barrier effect to water
vapor transfer from emulsion to the head space.

REFERENCES

1. Quaglia, G.B, Alessandroni, A., Riccardi, A., Proprieta emulsionanti. In

Proprieta funzionali delle proteine. Monografia CNR P.F. "Ricerca di nuove fonti
proteiche e di nuove formulazioni alimentari", Roma, 1981.
2. Leman, J. and Kinsella, J.E., Surface activity, film formation, and
emulsifying properties of milk proteins. CRC Crit. Rev. Food Sci Nutr., 1989,
28, 115.
3. Pearce K.N. and Kinsella J.E. Emulsifying properties of proteins: evaluation
of a turbidimetric technique.J. Agric. Food Chem.,1977, 26, 716.
4. Kato, A., Fujishige, T., Matsudomi, M., Kobajashi, K. Determination of
emulsifying properties of some proteins by conductivity measurements. J. Food
Sci., 1985, 50, 56.
~Acton, J.C. and Saffle, R.L .. Stability of oil-in-water emulsions. II Effect of
oil phase volume, stability test, viscosity, type of oil and protein additive . .L
Food Sci., 1971, 36, 118.
6. Latreille, B. and Paquin, P.. Evaluation of emulsion stability by
centrifugation with conductivity measurements.J. Food Sci., 1990, 55, 1666.
 ADDITION OF SOY LYSOPHOSPHOLIPIDS ENHANCES
EMULSIFYING ABILITY OF FOOD EMULSIFIERS

SATOSHI FUJITA, ATUSHI SUZUKI* and HITOSHI KAWAI*

S. Fujita Consultant Office, 3-7-1, Tsukusigaoka, Kashiwa-shi, Chiba 277;
*Nagoya Seiraku K.K., 310, Nakasunatyo, Tenpaku-ku, Nagoya, Aichi Japan

ABSTRACT
SOy lysophospholipids (SLP) were added to sucrose fatty acid ester (SE),
for the preparation of O/W emulsions containing milk protein, and the ad-
dition of SLP improved quality and stability of the emulsions. The droplet
size became finer when the ratio of SLP/SE increased. To investigate the
interaction between protein and surfactant, surface tension was measured
for aqueous solutions of soy lysophosphatidylcholine (SLPC), SLP, SLP/SE
mixture, and SE, in the presence or absence of sodium caseinate. Protein-
surfactant complex was formed with SLP, but SLPC and SE had no interaction
with the protein. It seems that the protein/SLP complexes are attached
tight to the droplets and stabilize the emulsion.

INTRODUCTION
Recently utilization of lysophospholipids (LPs) is being explored. Soy
lysophospholipid (SLP), soy lysolecithin, are very promising food surfac-
tants because of their low production cost vis-a-vis high emulsifying ac-
tion. The authors studied the properties (1,2), and synergistic effects of
SLP and SLPC with other food surfactants (3-5). Their mixtures showed im-
proved emulifying ability (6). Since most of food emulsions contain pro-
teins, it is important to know protein-surfactant interactions. This study
deals with interaction of surfactant with protein and protein containing
O/W emulsions prepared with SLP/sucrose fatty acid ester (SE) mixtures.
MATERIALS AND METHODS
Soy LPC (99% pure), from Nihon Shyouji K.K., SLP (90% pure) prepared by
the modified method of Nakai (7), sucrose mono- distearate (Ryouto Ester S
-1170, 98% pure, HLB=11) from Mitsubishi Kasei Shokuhin K.K., Na-caseinate
(89% protein, M-111, Uni Lait France) were used. O/W emulsions (simulated
vegetable coffee whitener) were prepared with 25 wt% of hydrogenated vege-
table oil (MP=32°C), 0.6 wt% of mixed surfactants (SLP/SE=4/0-0/4), 4.0
wt% of caseinate and water. Pre-emulsions prepared by turbine mixer were
further emulsified by homogenizer, 65 atm/cm2, at 65°C. After incubating
the emulsions at 4QC for 16 hrs, the median droplet diameter was determin-
ed by particle size distribution analyzer. Surface tension of aqueous
solutions of emulsifiers and their mixtures with 0.1% Na-caseinate was
174
measured by Wilhelmy's plate method at 25°C.

RESULTS AND DISCUSSION

Figure 1 shows the median droplet diameter (/..lm) and the surface area
(m 2 /g fat) of the emulsions. The addition of SLP to nonionic surfactants
(SE) improved the emulsions. Higher the ratios of SLP, finer the droplet
size was obtained. Our previous data (6) showed that SLP and SLP/SE mix-
tures had better emulsion stability than LPC when used in simulated coffee
whitener containing milk proteins. Figure 2- I "'" IV, show surface tension y

<>:
"" +'
7 SUIIHCI( AIIl(A '{ 0
<>: '"
0::
~
I , SLPC
""
u
<>:
....
be
60
~
0::
"- 0
::::>
N
50 ~

~
6
'"
40
0:: 5 ~

.
""
E-< -II '---'-
""<>:
:E:
MKDIAN nllOPLKT • 70 ~ II, SLP
..... nIAMI(TKlt / 6
A
E-<
8
:!.
1.3 ~ "-
:z:
6
60
....,
"" /e

t::L_
50
"'0::"' -s. 1.2
C)
• 1"-
A
:z: :z:
40
<>: 1. 1 C)
..... 4/0 2/2 1/3 0/4
3/1 l-<
-11--1---'--
A
'"
:z:
""
:E: FIG.1 SLP/SE RATIO (Wt/Wt) ""
E-<
70

""
u 60
....
<>:
0::
::::> 50
'"
40
P-o.
I
-11----'
".0---.
70
Arrows in Fig.2 denote the
ratio at which the emulsions 60
shown in Fig.1 were prepared.
50

40

FIG.2 J~oo -4 -3 -2 -I
LOgl6 Cs of Surfactant Concentration (Wt.%)

Fig.1 Median droplet diameter and surface area of the emulsions, emulsi-
fied with SLP/SE mixtures, after standing at 4°C for 16hr. ---()---
surface area (m 2 /gr fat), • : median droplet diameter (/..lm).
Fig.2 Surface tension y of surfactant + Na-caseinate as a function of
logarithm of surfactant concentration Cs (WU), . , no protein present;
0, 0.1 wU Na-caseinate present. Surfactant: I , SLPC; II, SLP; III, SLP/SE
(1:1 by wt.) mixtures; IV, SE. Abbreviations, SLPC: soy lysophospha-
tydilcholine; SLP: soy lysophospholipid; SE: sucrose mono- distearate.
 175
of surfactant solutions (I :SLPC; II :SLP; III :SLP/SE(1:1 by wt.) and IV :SE)
with or without 0.1% Na-caseinate, as a function of logarithm of surfac-
tant concentration (Cs). In Fig. I and IV, the two sets of results become
essentially coincident at Cs=10- 3 for SLPC and at Cs=10- 2 for SE. Above
this concentration, the protein seems to be displaced completelY from the
surface, and the cmc value is same both in the presence and absence of ca-
seinate. The behaviour shows that there is no interaction between casien-
ate and surfactant. On the other hand, as shown in Fig. II and III, with SLP
or SLP/SE mixture, the situation is more complicated. At high surfactant
concentration, the tension for surfactant/caseinate mixture is same as
that for each surfactant alone, suggesting the complete protein displace-
ment. The tension in the presence of protein was significantly higher than
that of surfactant alone, in the region of surfactant concentration near
to its cmc, and there is a plateau region. Dickinson (8,9) reported simi-
lar phenomena, suggesting that, in the plateau range, it is more favoura-
ble in terms of free energy for surfactant to bind on to protein than it
is for surfactant to displace protein from the surface. At the endof the
plateau range, when there are no more binding sites available on the pro-
tein, free surfactant displace the protein-surfactant complex from the
surface. In the region where surfactant concentration is very low, the
tension is essentially the same as that of the protein alone, and surfac-
tant is bound strongly (by electorostatic interaction) to protein mole-
cules in bulk solution. In the next slope region, it was also suggested
(8) that due to the protein-surfactant complex formation and its conforma-
tion change and increased hydorophobicity, the molecules show increased
surface activity. Jones (10) reported that anionic surfactant unfold pro-
tein by their electorostatic bonding, and after unfolding, hydorophobic
bonding is followed between protein and surfactant. SLP contains choline,
ethanolamin and inositol lipids almost in equal ratio. LPC is nutral, and
other two are acidic lipids. The SLP/caseinate mixture has apparent cmc of
0.1 wt%, where molar ratio of surfactant/protein is about 50. Whereas (SLP
/SE)/caseinate has apparent cmc of 0.01 wt% and it coincide with that of
SE alone, where SLP/protein molar ratio is about 3. It seems that SE has
no interaction with caseinate. Thus, at the concentration above cmc it may
displace the protein-SLP complex. Arrows in Fig.2 denote the ratio at
which the emulsions (Fig.1) were prepared. To prepare stable emulsions, it
may be important that a plenty of protein-surfactant complex exist on an
O/W interface. In the case of LPC, fine emulsion was obtained immediately
after the emulsification, but its stability was poor (6). That might be
due to the excessive additon of LPC which displaced the complex molecules.

REFERENCES
1. S.Fujita,and K.Suzuki, Nippon Nogeikagaku Kaishi.64:1355(1990).
2. S.Fujita,and K.Suzuki, Nippon Nogeikagaku Kaishi.64:1361(1990).
3. S.Fujita,and K.Suzuki, J.Am.Oil Chern. Soc .. 67:1008(1990).
4. S.Fujita,and K.Suzuki, J.Japan Oil Chern. Soc .. 40:1105(1991).
5. S.Fujita,and K.Suzuki, J.Jap.Soc.Food Sci.Tecnol. .39:151(1992).
5. S.Fujita,A.Suzuki,and E.Yahisa, In Proceedings of International Confer-
ence on Food Hydrocolloids, Tukuba, 1992. Plenum Press, in print.
7. E.Nakai,K.Suzuki,S.Sato,and M.Kato, Japan Pat.laid Open.89/16595(1989).
8. E.Dickinson,and C.M.Woskett,in:"Food Colloids,"R.D.Bee,ed. ,p.74, Royal
Soc.Chem. ,Cambridge(19S9).
9. E.Dickinson,in:"Microemulsions and Emulsions inFoog," M.El-Nokaly Ed.
,p.114, Am. Chern. Soc.,Washington,DC(199l).
10. M.N.Jones,and A.Brass,in:"Food Polymers,Gels,and Colloids,"E.Dickinson
,ed.,p.55, Royal Soc.Chem.,Cambridge(1991).
 CHANGES IN WATER ACTIVITY AND FUNCTIONAL PROPERTIES OF
PROTEIN HYDROLYSATES

HITOMI KUMAGAI, HIROTAKA SETO, HITOSHI KUMAGAI*, HIDETOSHI

SAKURAI, KENJI ISHII AND SOICHI ARAI*
Department of Agricultural Chemistry, College of Agriculture and Veterinary Medicine,
Nihon University, 3-34-1, Shimouma, Setagaya-ku, Tokyo 154 Japan
*Department of Agricultural Chemistry, Faculty of Agriculture, the University of Tokyo,
1-1-1, Yayoi, Bunkyo-ku, Tokyo 113 Japan

ABSTRACT
The changes in the interaction between food proteins and water, and those in the surface
functional properties during their enzymatic hydrolysis were investigated. Ovalbumin, soy
protein isolate (SPI), casein were used as protein samples and were hydrolyzed with
trypsin. The enzyme was inactivated by heating at appropriate intervals and each
hydrolysate was subjected to the following measurements: degree of hydrolysis, water
activity, activity coefficient of water, forming capacity, and emulsion capacity. Every
protein once increased foaming capacity and emulsion capacity. Water activity of
ovalbumin showed a minimum whereas those of SPI and casein kept decreasing during
hydrolysis. Every protein had a minimum of activity coefficient of water and the minima
obtained by ovalbumin and SPI almost coincided in degree of hydrolysis with their
maxima of foaming and emulsion capacity. It was suggested that activity coefficient of
water might be a better parameter than water activity to evaluate the interaction with water.

INTRODUCTION
Protein of some kind has surface functional properties like foaming and emulsifying
properties. Limited enzymatic hydrolysis of proteins is sometimes effective to improve
these functional properties [1 - 3]. However, the surface functional properties of protein
have not been so explained as those of low-molecular surfactants because of its large
molecular size and complicated structure. It is probable that interaction of protein with
water would change during hydrolyzing process and affect the surface functional
properties. We found that ovalbumin had a minimum of water activity when its foaming
and emulsion capacities showed maxima during tryptic hydrolysis. In the present study,
we chose soy protein isolate (SPI) and casein in addition to ovalbumin as food protein
samples and further examined the changes in their interaction with water and their surface
functional properties during tryptic hydrolysis.

MATERIALS AND METHODS

Materials
 177
Ovalbumin, SPI and casein were obtained from Tokyo Kasei Kogyo Co., Fuji Oil Co. and
Merck Co., respectively. Trypsin was purchased from Difco Co. and corn salad oil was
from Ajinomoto Co. All the other reagents used were of reagent grade.
Enzymatic Hydrolysis
Each protein was dispersed in distilled water at a concentration of 10% (wt/wt) and held in
a water bath at 37t for 20 min prior to the addition of trypsin amounting to 1% (wt/wt) of
protein. Enzymatic reaction was carried out with magnetic stirring at 37t and was
stopped at a selected time by heating at 80t for 15 min. Each hydrolysate was dried
under vacuum prior to the following measurements.
Measurement of Degree of Hydrolysis
The degree of hydrolysis was expressed as nitrogen percent in hydrolysates dissolvable in
0.2N aqueous trichloroacetic acid per that in total hydrolysate. After obtaining the
calibration curves between absorbance measured by Folin method and nitrogen percent by
Kjeldahl method for each protein, Folin method was used to measure the degree of
hydrolysis.
Measurement of Water Activity (aw)
Approximately equal amounts of hydrolysate and distilled water were homogeneously
mixed and the water activity was measured at 25t with Novasina Thermoconstanter
Humidat TH2 apparatus. The water content of each sample was measured by heating at
105t. The water activity at water content of 100% (dry basis) was calculated by
interpolating the data near 100% (dry basis) water content.
Calculation of Activity Coefficient of Water (yw)
Total moles of amino acids, peptides and protein in each hydrolyzed sample were
measured by trinitrobenzene sulfonate (TNBS) method [4]. Mole fraction of water (Xw) in
each sample of 100% (dry basis) water content was then calculated. Activity coefficient of
water was calculated by the following equation: Yw = aw /X w.
Measurement of Foaming Capacity
Foaming capacity was measured by air delivery method and expressed as foam volume
(ml) per initial volume (ml).
Measurement of Emulsion Capacity
Emulsion capacity was measured by electric conductivity method and expressed as volume
of oil emulsified until phase transition occurred (ml) per weight of protein (mg).

RESULTS AND DISCUSSION

Even after tryptic hydrolysis for 15 hr, ovalbumin was slightly hydrolyzed, while SPI and
casein were hydrolyzed by 50 and 80%, respectively. Figure 1 shows the changes in
foaming capacity, emulsion capacity and water activity during tryptic hydrolysis of three
proteins. The surface properties of ovalbumin and SPI showed maxima, whereas those of
casein kept increasing. The water activity of ovalbumin showed a minimum where its
surface properties were maxima. The water activity of SPI and casein continued
decreasing probably because it was affected by the considerable change in moles of
hydrolysates. Therefore, activity coefficient of water was obtained to remove the influence
of mole change and to examine the interaction with water, with the result that every protein
had a minimum of the activity coefficient of water during hydrolysis (Fig. 2). This result
suggests that a conformational change of protein would affect the interaction with water
and some peptide would strongly bind to water rather than original protein or peptides in
lower molecular size. As for ovalbumin and SPI, their minima of the activity coefficient
of water almost coincided with their maxima of surface properties probably because the
limited hydrolysis was more effective to change the conformation of folded protein such as
ovalbumin and SPI than that of unfolded one such as casein.
178
IOvalbumin I SPI
_0.6
e
~
--
~800
900

-
--
";;;;

-
5 0.5 0
~
... ;:::... 600
~700 o 0.99

'0 (,J
:~
~0.4 ~500 0.98 ~
eu eu ~=---x---=---i
(,J 400 ~eu
=
(,J
~
.5 300 0.97 ~
.20.3
:;
rI.l
e
e ~ 200
~
~ 0.2 100 0
1 2 3 4 o 10 20 30 40 50 0 20 40 60
Degree of hydrolysis (%)
Fig. 1 Effects of enzymatic hydrolysis of ovalbumin, SPI and casein
on foaming capacity, emulsion capacity and water activity.
o Foaming capacity, b. Emulsion capacity, .Water activity

I
-
IOvalbumin ~ SPI
~ Casein

-.- -
1.00 1.00 0.985 1.011-..!-
I
r..

.::-
~
'-' 0.997 1.010 -;
'-'
I r..

.-5-
~
~

-
... 0.99 -;0.99 0.996 1.009 ....
....
~
0.995
0

- -
(,J 0 1.008 ~
eu 0.98 .§0.98 0.994
r..
~ 1.007
eu (,J 0.993
eu ~

8
~ 0.97 c:: 0.97 1.006
...
-
~
'0 1.005]

0.96
~
0.96
0 1 2 3 4
0.982
0 10 20 30 40 50
0.990
0 5 10 15 20
1.004 <
Degree of hydrolysis (%)
Fig. 2 Effects of enzymatic hydrolysis of ovalbumin, SPI and casein on water
activity [Aw], mole fraction [Xw] and activity coefficient of water [y w]
o Water activity, L:::. Mole fraction of water, • Activity coefficient of water

REFERENCES
1. Puski, G., Modification of functional properties of soy proteins by proteolytic enzyme
treatment. Cereal Chern., 1975,52,655 - 64
2. Grunden, L. P., Vadehra, D. V. and Baker, R. C., Effects of proteolytic enzymes on
the functionality of chicken egg albumen. L. Food Sci., 1974,39, 841 - 3.
3. Kuehler, C. A. and Stine, C. M., Effect of enzymatic hydrolysis on some functional
properties of whey protein. L. Food Sci., 1974, 39, 379 - 82.
4. Hazra, A. K., Chock, S. P. and Albers, R. W., Protein determination with
trinitrobenzene sulfonate: A method relatively independent of amino acid
composition. Anal. Biochem., 1984,137,437 - 43.
 WATER ACTIVITY AND WATER BEHAVIOR OF SOY SAUCE,
DEHYDRATED SOY SAUCE AND THE IMPROVEMENT ON
HYGROSCOPICITY OF DEHYDRATED SOY SAUCE

MITSUTOSHI HAMANO
Research & Development Division, Kikkoman Corporation,
399 Noda, Noda-shi, Chiba 278 Japan

ABSTRACT

The state of water in soy sauce (shoyu) containing 65% (w/w) of water, aW 0.803 at
25°C, can be divided into three water regions such as hydration water, eutectic melt-
ing, and ice melting, respectively. On the other hand, dehydrated soy sauce, aW
0.203 at 25°C, water content 2.5% (w/w), has many kinds of hygroscopic components
such as glucose, lactate, nitrogen compounds, and color components. The improve-
ment of caking and hygroscopicity in dehydrated soy sauce by the addition of saturat-
ed fatty acids with 12 to 18 carbon atoms and shoyu oil were achieved. The aroma
components in dehydrated soy sauce was improved by adding soy sauce aroma
components recovered by the use of the vacuum ethanol distillation method.

INTRODUCTION

Water activity and the state of water have a decisive effect on the qualitative stability
in foods(I). During the drying process, the properties of a food are obviously modi-
fied and the extent depends upon the type of drying. The physical properties of
powdered food may be generally affected by changes such as caking, hardening,
stickiness, shrinkage, and hygroscopicity. Recently, dehydrated soy sauce has been
widely used in soup mixes, sauce mixes, and gravy mixes for flavor enhancement.
However, dehydrated soy sauce is difficult to process and store since it is non free-
flowing, develops caking, and becomes solid due to hygroscopic components. During
the dehydration process in spray or freeze drying, aroma components of soy sauce
decrease more than other materials (2). This paper reports the state of water and aWof
soy sauce. The improvement of hygroscopicity and aroma components of dehydrated
soy sauce were discussed in terms of their application to new food additives.
180
METHODS AND MATERIALS

Soy sauce containing 17% (w/v) of sodium chloride, 1.6% of total nitrogen, was
dehydrated by using a spray dryer (Niro Atomizer Co., Copenhagen) and freeze dryer
(Kyowa-shinku Co., Tokyo). High grade reagents on saturated fatty acid and shoyu
oil (by-product of soy sauce fermentation) were used. The water content ofthe
sample was determined by the Karl-Fisher method. Water activity was determined by
the saturated salt solution method and with a hygrometer (Novashina Co.,
Switzerland). Low temperature type of Differential Scanning Calormetory (DSC)
were used for measuring the state of the water in soy sauce. The determination of
volatile components in the samples were done by a gas liquid chromatography
(lO%PEG 20M on Chromo sorb W-AW).

RESULTS AND DISCUSSION

Water behavior of soy sauce and dehydrated soy sauce

The state of water in soy sauce 65% (w/w) water, aW 0.803 at 25°C could be divided
into three water regions: (1) Hydration water (2) eutectic melting, and (3) ice melting.
The hydration water contained a higher level value than other water regions in the soy
sauce. The state of water in soy sauce was affected by the components in the soy
sauce such as NaCl, alcohols, and nitrogen compounds etc. On the other hand, the
behavior of water in dehydrated soy sauce could be divided into water regions from
the results of the water sorption isotherm, heat of sorption, DSC measurement, and
sensory evaluation. Water regions include: (1) Powdery region below aW 0.25 (below
3% of water content), (2) caking to surface wetness region in the aW 0.25-0.55 range
(3-15% of water content), and (3) solid solution region above aW 0.55 (above 15% of
water content).

Improvement of hygroscopicity and aroma of dehydrated soy sauce

Hygroscopic equilibrium of the dehydrated soy sauce obtained by spray dried soy
sauce to which C12-C18 saturated fatty acids were previously suspended decreased
more than that of spray dried soy sauce without the addition of saturated fatty acids.
In addition, the reduction of caking of the dehydrated soy sauce by adding 1%(w/w) of
powdery saturated fatty acids and shoyu oil were achieved. It has been proven in
actual industrial application that the reduction of hygroscopicity and caking in the
dehydrated soy sauce can be successfully achieved by the addition of saturated fatty
acids. During the drying process, volatile components having low boiling points in soy
sauce, such as isobutyl alcohol, isoamyl alcohol, and ethyl lactate decreased more
than those with high boiling points, such as phenolic compounds. The improvement of
aroma components of the dehydrated soy sauce was also achieved with recovery from
soy sauce aroma components recovered by the use of the vacuum ethanol distillation
method.
 181
CONCLUSIONS

(1) The state of water in soy sauce could be divided into three water regions: Hydra-
tion water, eutectic melting, and the ice melting region.

(2) The water behavior in the dehydrated soy sauce could be divided as follows:
Powdery region below aW 0.25 (below 3% of water content), caking to surface wet-
ness region in the aW 0.25-0.55 range (3-15% of water content), and solid solution
region above aW 0.55 (above 15% of water content).

(3) It has been proven in actual industrial application that the reduction of hygrosco-
picity and caking in the dehydrated soy sauce can be successfully achieved by the
addition of saturated fatty acids and shoyu oil.

(4) The enhancement of aroma components of the dehydrated soy sauce was improved
by adding the recovered soy sauce aroma components for industrial application.

REFERENCES

1. Labuza T.P. Interpretation of sorption data in relation to the state of constituent

water. in:t'Water relation of foods>, R. B. Duckworth ed., Academic Press, London,
1974,155-172.

2. Hamano M and Okayasu M. Water sorption and retention of volatile components in

dehydrated soy sauce and ~-cyc1odextrin. in~(Fundamentals and applications of
freeze-drying to biological materials, drugs and food stuff~'. issued., Intemation Insti-
tute of Refrigeration. Paris. France. 1985, 109-112.
 SORPTION ISOTHERMS AND HEATS OF SORPTION FOR FRUIT JAMS

M. M. Sa and A. M. Sereno*
University of Porto, Faculty of Engineering, Department of Chemical Engineering
Rua dos Bragas, 4099 PORTO, Portugal
Telephone: 351 - 2 - 200 7437, Telefax: 351 - 2 - 319 280, E-mail: sereno@fe.up.pt

ABSTRACT

Water sorption isotherms at three different temperatures for two different brands of
"marmelada" quince jam and four brands of peach jam were measured, using a static
equilibration method. Results obtained showed that the typical high sugar content of these
products leads to an inversion of the usual temperature-dependence of water sorption isotherms
for values of water activity above 0.6. This confirmed a behaviour already reported for other high
sugar products like dried fruits. From the set of isotherms obtained for each jam, it was also
possible to estimate net isosteric heats of sorption and their dependence on moisture content.

INTRODUCTION

Temperature has a significant effect on sorption isotherms and its knowledge is essential
for the efficient design and operation of several processing operations and storage. Saravacos et
al. (1) found that sugars may modify significantly the dependence of sorption isotherms on
temperature. Several studies involving dehydrated fruits (raisins, figs, prunes and apricots) found
that a cross-over of the isotherms occurred in the water activity range of 0.5 to 0.7.
The major objectives of this study were then (i) to determine water sorption isotherms of
quince and peach jams at different temperatures and check for any anomalous behaviour due to
the presence of high sugar contents, and (ii) to evaluate the net isosteric heats of sorption and
their dependence on moisture content.

MATERIALS AND METHODS

Two brands of quince jam and four brands of peach jam bought at a local supermarket
were used. The sugar contents as determined by refractive index (2) were:

Peach Jam Quince Jam

LINEA 41.5% FERBAR 63.6%
BEBE 47.0% T.NOVA 71.4%
ZUEGG 56.0%
SUMOl 65.0%

Small samples (2-3g) of the jam were placed in 25 ml weighing bottles and equilibrated
over saturated salt solutions according to COST 90 recommendations (3). The hygrostats were
maintained within .±..1°C in a temperature-controlled cabinet. Six samples of each brand of jam
were used. At each temperature, equilibration was reached when a change in weight of
O.4%/week or less was observed. In most cases this was achieved after three weeks.
With the help of a non-linear regression programme from IMSL, experimental water
sorption data was fitted to the well-proven GAB model:
 183

x X m , C, K - GAB parameters (1)

x - moisture content
RESULTS AND DISCUSSION

Water sorption isotherms at 25°C, 35°C and 45°C and GAB parameters are shown in
Figures 1 and 2 for each jam, based on average moisture contents of at least five
determinations. A clear "cross-over" of the isotherms in the aw range of 0.50 to 0.75 is
observed. This inversion, already observed with dried fruits containing significant quantities of
sugar, may be due to an endothermic dissolution of the sugar at high water activities (1).
Fruit jams are usually stored at room temperature and have moisture contents
corresponding to water activities exceeding 0.6. In this range of aw microbial spoilage may occur
and this inversion of the usual water activity (aw) dependence on temperature IT) at constant
moisture content, is then a relevant factor opposing the normal trend for an increased microbial
activity with temperature.
Following the procedure proposed by Rizvi and Benado (4) and assuming that the net isosteric
heat of sorption, 0st = (Hads.- Hcond), is temperature independent:

(Hads. - Hcond.) 1
In aw =- R T + const. Hads., Hcond. - heats of adsorption (2)
and condensation

The value of 0st may then be calculated from the slope of In aw vs. 1rr plots. Figures 3
and 4 show the net isosteric heat of sorption vs. the moisture content. The observed positive
values of the net isosteric heat of sorption at low moisture contents are supposedly due to an
easy physical sorption of water molecules forming a monomolecular layer (5), while slightly
negative values of 0st at high moisture contents have been explained by the contribution of the
endothermic dissolution of sugars in the sorbed water (1).

An extended version of this communication, with full tabular data and a complete set of
plots is available and can be supplied on request.

ACKNOWLEDGEMENTS

The authors acknowledge the financial support of NATO's Scientific Affairs Division
through Program SfS, Project PO-PORTOFOOD and author MMS the support of JNICT-Junta
Nacional de Investiga9ao Cientffica e Tecnol6gica. Funda9ao Calouste Gulbenkian and University
of Porto have given partial support for its presentation at ICEF6.

REFERENCES

1. Saravacos, G.D., Tsiourvas, D.A. & Tsami, E. (1986). Effect of temperature on the water adsorption
isotherms of Sultana raisins, Journal of Food Science. 51,381-387.
2. AOAC. (1984). Official Methods of Analysis. 14th edition, Association of Official Analytical Chemists,
Virginia, USA.
3. Spiess, W.E. & Wolf, W.R. (1983). The results of the COST 90 project on water activity. In: Phvsical
Prooerties of Foods. (edited by R. Jowitt et aLl. pp. 65-87. London. Elsevier Applied Science Publ.
4. Rizvi, S.S. & Benado, A. L. (1984). Thermodynamic analysis of drying foods, Drying Technologv, 2,
471-502.
5. Tsami, E., (1991). Net isosteric heat of sorption in dried fruits. Journal of Food Engineering. 14, 327-
335.
184
1Moisture Content Ig H;D/g solidsl 08 Moisture Content Ig HfJ/g solidsl
Xm*102 C Xm'102 C k
o 25"C 0.5 10 O} 0 25"C 5.0 8.0 1.06
0.8 ,:,. 35'C 4.1 7.5 0.6 ,:,. 350C 4.0 10.0 1.13
o 45"C 40 098 o 450C 2.8 10.0 1.20
06 - GAB equation 0.5 - GAB equation
04
0.4
0.3
0.2 0.2

00~~0~j~0~.2~0~.3~0~4~0~.5::0~.6=-0~}~0~.8-0~.9~ 0.114~~~:::~~,---,-L-J
00 0.1 0.2 0.3 0.4 0.5 0.6 O} 0.8 0.9
\.\\3ter Activity lawl \.\\3ter Activity lawl

Figure 1. Sorption isotherms of peach jam (LINEA and BE BE) at different temperatures.

08 Moisture Content Ig ftO/g solidsl 0.8 MOisture Content Ig ftO/g solidsl

. Xm'102 C k Xm*102 C k
o To 25'C 5.3 17.7 1.06 o To 25OC 4.8 26.2 1.06

0.6 I:; To 35 DC 9.1 1.05 1.03 0.6 I:; H5 OC 6.3 1.6 1.07
o To 45 DC 13.5 0.49 1.04 o To 45OC 7.3 0.63 1.14
- GAB equation - GAB equation
0.4 04

0.2 0.2

00~~0~j~0~.2~0~.3~0~4~0:.5::0(76~07}~0~B~0~9~
o

0.1 0.2 0.3 0.4 0.5 0[6 O} O.B 0.9 \.\\3ter Activity awl
\.\\3ter Activity awl
Figure 2. Sorption isotherms of quince jam (FERBAR and TAPADA NOVA)
at different temperatures.

30 Net Isosteric Heat of Sorption. Q'L KJ/rml 80 Net Isosteric Heat of Sorption. QsL KJ/rrol

25 PE/lCH JAM 70 QUINCE JAM

0 LINEA 60 o FERBAR
20 0 BEBE o TAPAQA NOVA
50
,:,. SUMOL
15 40
0 ZUEGG
30

~
10
20
5
iD

O~~§$~60
~o~
0
-iDo
=
10 20 30 40 50 ill 20 30 ~ 50 60
Moisture Content 1% dry basisl MOisture Content 1% dry basisl

Figure 3. Net isosteric heats of sorption as Figure 4. Net isosteric heats of sorption as
function of moisture content (Peach jam). function of moisture content (Quince jam).
 WATER ooRPl'ION ISO'l1f.KRMS OF J(X)D tilDE MIXTURES

PIOTR P. LKWICKI, ~ R:IfAlW'.ISKA~AZUKA

Department of Food Engineering
Warsaw Agricultural University. SGGW
02-766 Warazawa, uLNowouraynowska 166

ABSTRACT

Mixtures of potato starch. sodium caseinate, cellulose and salts were stored
over saturated salt solutions, and their sorption isotherma were determ.ined.
Isotherma of individual colllponents were also measured. Isotherms A.re
strongly affected by the mixture composition and cannot be predicted on the
basis of isotherms of individual components.

INTROOOCTION

Many foods are simple mechanical mixtures of dry components. Water activity
in such mixtures can be predicted with the use of the Salwin [1] equation.
At high. water activities, the Ross [2] equation can be used. These equations
predict the equilibrium state of the mixture but do not account for possible
interactions between canponents caused by transfer of water. Johnson and
fuckworth [3) showed that the presence of solute in microcrystalline
cellulose affects the form of sorption isotherm.
The aim of this work waS to investigate sorptional behavior of
mechanical mixtures of dry components and to compare predicted with measured
isotherms.

METHODS

Potato starch, sodium caseinate, cellulose for chromatography (whatman) and

mixture of inorganic salts (NaCl:Na2 S04 :1(2004 :N~P04 :KCl:~P04=1:1:1:1:1:1)
were used as components. Components were dried in an oven and ke"pt over P2 05
before use _ Mixtures were "prepared mechanically. Approximately t g samplp.
was equilibrated in desiccators over sulphuric acid and saturated sal t
solutions at 25°C. Equilibration was run for 3 months.
Isotherms for mixtures were predicted on the basis of water balance at
aasumed water activities. The following equation was used:
186

SA· UA + Ss· Us
U = ---:::---=--
SA + Ss

where: S - mass of dry matter. g d.m.

U - water content of pure component at chosen water activity (read
from the isotherm), g H2 0/g d.m.
subscript A and B refer to components of the mixture.
Statistical methods were used to assay significance of observed
differences.

RESULTS

Predicted isotherms do not match with measured isotherms and the difference
was expressed as a ratio of predicted to measured value at chosen water
activities. The ratio plotted veraua water activity was skewed
substantially, with lower values accumulating mainly in the low water
activity range.
Predicted and measured values were analyzed in two water activity
ranges, Le. <0.23 and. :>0.33. Results of the analysis are presented in Table
L

TABLE 1
Results of statistical analysis

a \,/
< 0.23 a v :> 0.3.1
Mixture
n r t n r t
~tatQ atarcb _ 8
- 2 0.9174 0.9877 1.88 0.9337 0.9837 1.29
sodium caseinate
~tato atarcb _2
- 8 0.7233 0.9512 4.09 0.9532 0.9787 0.78
sodium caseinate
wtato atarch _ tl
- 2 0.9381 0.9870 0.91 0.9044 0.9880 1.35
cellulose
Mato starcll _2 0.71
- 8 0.8696 0.9919 2.26 0.9130 0.9658
cellulose

a < 0.60 a v :> 0.60

'"
~tatQ starcb _85 1.0695 0.9982 3.21 3.2111 0.9707 2.78
salts - 15

n-slope of the line, r-correlation coefficient, t-t test value.

The mixture of 80% starch and 20% sodium caseinate yields a sorption
isotherm which does not differ significantly from the measured one. However,
all predicted water contents are smaller than the measured values. Increase
of content of sodium caseinate in the mixture lead to an isotherm which was
significantly different from the predicted one.
Mixture of potato starch and cellulose behaved similarly to the
mixture of starch and sodium caseinate. The difference between predicted
and measured isotherm is manifested mainly in the low a w range, and is not
 187
~~ large as for the mixture of starch and sodium caaeinate-
When salts are added to starch the predicted isotherm is statistically
different from the measured one. Predicted values are always higher than
the measured ones. The difference is large at water activities higher than
0.6.

DIRCUSSION
Mixtures of potato starch with sodium caseinate or cellulose adsorb more
water than can be predicted from the isotherms of individual components.
This is especially evident at low water activities. These results
contradict
exclude the assumption that the number of sorption centers is constant
(Langmuir, BET, GAB theories) and can be determined only by- the area of
the
internal surface of the adsorbent. Adsorption of water probably induces
conformational or configurational changes of adsorbent which are affected
by- the presence of other components. The sorption conformation coupling
has been considered in the theory of gas adsorption on polymers [4].
Conformational changes of starch chains caused by- water adsorption and
favoring the access to subsequent centers of sorption at low water
activities have been described by- Klimek and Poliszko (5].
The data on sorption of water on the mixture of starch and salts shows
the competitive nature of this process.

roNCLUSION
Water sorption isotherms of mechanical mixtures of potato starch with sodium
caseinate or cellulose predicted on the basis of isotherms of individual
components do not match with experimental ones. The difference is larger the
higher the weight ratio of sodium caseinate or cellulose to potato starch.

REFERENCES
L Salwin,H. Moisture levels required for stability in dehydrated foods.
Food Technol. 17, 1963, (9), 34.
2. Ross K.D. Estimation of water activity in intermediate moisture foods.
Food TechnoL 29, 1975, 26--30.

3. Johnston K.A. , Duckworth,R.B. : The influence of soluble components on

water sorption hysteresis. In Properties of Water in Foods in Relation
to Quality and Stability (ed. D.Simatos, J.L.Mlllton). Martinua Nijhoff
Publ. Dordrecht 1985.

4. Pyda M., Kurzytlski Z. Theory of sorption of gases on polymers. L

Conformational effects and the double-sigmoid shape of sorption
isotherms. Chem. Phys. 67, 1982, 7-11.

5. Klimek D., Poliszko S.: Adsorption of water on freeze-dried starch

geL Properties of Water in Foods. Warsaw Agricultural University
Press. Warsaw 1993. 34-40.
GLASS TRANSITIONS AND PHYSICAL STATE OF DEHYDRATED MILK
PRODUCTS

YRJO ROOS and KIRSI JOUPPILA

Department of Food Technology, P.O. Box 27 (Viikki B),
SF-OOO 14 University of Helsinki, Helsinki, Finland

ABSTRACT

Glass transition temperature (Tg), its dependence on moisture, and crystallization behavior
of lactose were determined for dehydrated milk products. Various water contents were
obtained by rehumidification, and the Tg values were compared with those of lactose.
Physical stability was related to the temperature difference, T-Tg . The state diagram
established can be used to evaluate stability of milk powders.

INTRODUCTION

Deteriorative changes such as caking, stickiness, loss of structure, oxidation, and

nonenzymatic browning affect the quality of dehydrated products (1-4). Many of these
changes can be accounted for the physical state of food components e.g. lactose (3). Lactose
often exists in an amorphous form (1-3), which is stable as a glass, but becomes plasticized
by water (3). A fairly low water content decreases the glass transition temperature (Tg)
below room temperature (3-4). This results in increasing molecular mobility and lactose in
the "rubbery" state above T g may crystallize (3). Crystalline lactose exists as anhydrous 13-
lactose or a-lactose monohydrate and crystallization has been observed from loss of
adsorbed water (2). Therefore sorption isotherms show temperature-dependent water
activity (aw) values, above which the water content decreases (2-3). In this study the T g and
crystallization behavior of lactose and lactose in dehydrated milk products were determined.

MATERIALS AND METHODS

Skim and whole milk (Valio; 0 and 3.9% fat, respectively) were purchased from a local
store and a-lactose monohydrate (Sigma) was dissolved in distilled water (10 w-%
solution). Each material at 7°C was transferred to preweighed glass vials (27 x 60 mm) with
 189
a 5 ml measurement pipette. Sets of two samples were prepared in replicate and the weight
of each sample was monitored. The samples were frozen at -40°C and freeze-dried (Finn-
Aqua GT 2). Residual moisture was removed in a vacuum desiccator over P205 and the
water content was calculated from total weight loss. Stoppers were used at intermediate
stages to avoid uncontrolled evaporation or adsorption of water.

Crystallization
140 • Whole Milk
612 • Skim Milk
... Lactose
~10
w
a:
::)
80

iw
Q.
::::E Glass Transition
W 0 cWholeMilk
I-
·20 o Skim milk
It. Lactose
·40 O.36aw
.60 '--_'--_'--_.&.....JI::;...'----''--_'--_'--........
o 0.2 0.4 0.6 0.8
WATER ACTIVITY

Figure 1. Glass transition (Tg) and crystallization (Tcr) temperatures at various water
activities (aw). Tcr values of the milk products were higher than those of lactose and
decreased only at high aw values indicating delayed crystallization.

Dehydrated samples were equilibrated 24 hrs at 24°C in vacuum desiccators over

saturated salt solutions [LiCl, CH3COOK, MgCI2, K2C03, Mg(N03)z, NaN02, and NaCI
with respective relative humidities (% RH) of 11.5, 23.9, 33.0, 44.4, 53.8, 66.2, and 76.4;
and water activity (aw) of O.OlxRH]. Evacuation of desiccators allowed rapid equilibration.
The average equilibrium moisture content (m) was obtained from the increase of the weight
of the samples. Differential scanning calorimetry (DSC; Mettler 4000 system with TClOA
processor, DSC30 cell, and GraphWare TA72AT.2) was used to determine Tg (onset) and
onset temperatures of crystallization (Tcr). Samples (3 to 10 mg) in 40 JlI aluminium pans
(Mettler) were equilibrated over saturated salt solutions (24 hrs), sealed, and scanned at
5°C/min from T g-30°C above Tcr. Crystallization was also observed from water adsorption.

RESUL TS AND DISCUSSION

The T g values decreased with increasing moisture. The decrease was almost linear with
increasing aw (Figure 1). Delayed crystallization in dehydrated milk products was observed
from thermal behavior and sorption isotherms, which showed loss of moisture due to
190
lactose crystallization above T g (Figure 2). Crystallization was assumed to occur at room
temperature above 0.36 aw (Figure 1).

P'
.......
1oor:i2s~~===:::;-----C
.. Lactose
w o Skim Milk
a: • Whole Milk
:::)

~
a:
If
:::E 0
~
Z
o
j::: -50
en
z
c:c
a:
I- -100 Experimental Data
th Jl.. Lactose
~-'
o0 Skim Milk
Whole Milk
<-' -1500'~~-~----+:~-~:---~=--~
0.2 0.4 0.6 0.8 1
WEIGHT FRACTION OF NONFAT SOLIDS

Figure 2. State diagram showing composition dependence of T g. In milk products stability

decreases as Tg decreases below ambient temperature (m>7 gH20/100g nonfat solids at
25°C). Adsorption isotherms (inset figure) showed loss of water as lactose crystallized.

Tg values at various aw values were close to those of lactose, which seems to govern
the physical state of nonfat solids. The results confirmed that dried milk products should not
be stored at RH>36%, above which Tg falls below 25°C. The decrease of Tg below ambient
temperature results in stickiness, caking, and crystallization of lactose (3). The sorption
isotherm and state diagram can be used to evaluate stability of dehydrated milk products.

REFERENCES
1. Whiw, G.W. and Cakebread, S.H., The glassy state in certain sugar-containing food
products. J. Food Technol.. 1966, 1, 73-82.
2. Berlin, E., Anderson, B.A. and Pallansch, M.J., Effect of temperature on water vapor
sorption by dried milk powders. J. Dairy Sci.. 1970,53, 146-149.
3. Roos, Y. and Karel, M., Applying state diagrams to food processing and development.
Food Technol.. 1991,45(12),66,68-71, 107.
4. Slade, L. and Levine, H., Beyond water activity: Recent advances based on an
alternative approach to the assessment of food quality and safety. Ceit. Rev. Food Sci.
Nutr.. 1991,30, 115-360.
 STICKINESS AND CAKING OF INFANT FORMULA FOOD POWDERS
RELATED TO THE GLASS TRANSITION PHENOMENA

LETICIA E. CHUY AND THEODORE P. LABUZA

Department of Food Science and Nutrition
University of Minnesota
1354 Eckles Avenue, St. Paul, MN 55108, U.S.A.

ABSTRACT
The glass-rubber collapse phenomena as related to physical change was measured for
three infant formula powders. The following methods were used: (1) Lazar method
(advanced caking for 5g, Tae), (2) the ampoule method (surface caking for 1 g, Tsc), and
DSC (Tg for 5-10 mg). Caking of the powders was also studied during storage at 20 D C
for five weeks at different %RH. Tg, Tse and Tae appeared at higher values at faster
heating rates due to the shorter residence time at any temperature. Tae occurred at higher
temperatures than Tg and Tsc at given aw because it measures the advanced stage of
collapse. Linear extrapolation of results predicted that powders would not cake until
reaching a higher %RH than observed for five weeks storage at 20 D C, while Tg gave
good prediction.

INTRODUCTION
Food powders containing amorphous carbohydrates, generally, undergo physical
changes which are related to collapse phenomena [1, 2]. Collapse occurs when a matrix
can no longer support its own weight, leading to structural changes [1]. Recently, the
polymer science approach has been applied to food systems. An understanding of the
glass-rubber physical state of a material may provide information about the mechanisms
to control or prevent changes in the food physical chemical stability [2]. Previous studies
have suggested that both T sc and Tg increase with increasing heating scan rate and that
Tse and Tae are above the Tg [1,2]. The objective of this study were to determine the
effect of heating rate on Tsc and Tae measurement and to determine the relationship among
Tg, Tsc and Tae of food powders.
192
MATERIALS AND METHODS

Three commercial infant formula powders containing approximately 12% protein, 30%
fat and 57 % carbohydrate with different type of carbohydrate were used: Mead Johnson
(Evansville, IN) Enfamil® (lactose) and Prosobee® (com syrup solids); and Carnation
(Glendale, CA) Good Start® (lactoseimaltodextrin). The powders were equilibrated for
five weeks over saturated salt solutions. The surface caking temperature, T sc, was the
temperature at which the powder in a sealed ampoule begins to clump and adhere to the
sides despite shaking and tapping [1]. The advanced caking temperature, Tae, was
determined as the temperature of increased resistance to turning a shaft in a sealed glass
tube during heating [3]. Both Tsc and Tae were determined at 10 m min, 5 and lOoC/min
heating rates; while Tg was determined by differential scanning calorimetry, using a
Dupont 910 DSC, at 5 and 10°C/min heating rates.

RESULTS AND DISCUSSION

Only small differences in T sc were observed (Table 1) comparing the 5 and lOoC/min
heating rate for Good Start®. Tsc measured at 1°C/3 min is lower compared to Tsc values
obtained at 5 and 10°C/min heating rates. At the slower heating rate, the longer residence
time at any temperature allows the sample to undergo surface caking, thus, a larger extent
of caking can be observed. The same trends were observed for Tae. Table 1 also shows
that T sc decreases with increasing storage aw due to the plasticizing effect of moisture.
Linear extrapolation of the predicted a w conditions at which caking will occur based on
Tsc values obtained at lOoC/min heating rate are higher than was actually observed for 5
weeks storage at 20°C, while Tg gave a better prediction (Table 2). Both Tg and Tsc are
determined at 5 and 10°C/min heating rate, while caking during storage was observed at
zero heating rate.

TABLE 1
Tsc (OC) range (average ± s.d.) as function of heating rate for food powders

Powder 10m min

Enfamit® 0.12 76.3 - 83.7 87.4 - 115.0 93.1 - 102.3
0.33 55.0 -55.0 56.3 - 69.1 66.0 - 79.0
GoodStart® 0.12 88.8 - 94.6 99.1 - 114.3 84.1 - 99.3
0.33 65.0 - 65.0 73.8 - 79.6 80.4 - 86.2
0.45 50.4 - 56.2 58.8 - 64.6
Prosobee® 0.33 75.4-81.2 84.2 - 94.4 85.0 - 85.0
0.45 65.0 - 65.0 75.3 - 83.3 75.0 - 75.0
0.55 47.5 - 59.1 65.0 - 65.0 58.8 - 64.6

It was found that Tae is higher than Tse because these temperatures measure different
stages of collapse phenomena; Tsc measures the initial stages, while Tae measures the
advanced stages of collapse. Figure I shows that Tae is higher than Tsc, and Tsc is higher
than T g as was found for the other powders which is a consequence of glass transition
phenomena.
 193
TABLE 2
Predicted water activity vs. storage water activity at which caking was observed for 20D C

Powder predicted from predicted from Tg actual caking

Tsc (DSC)
Enfami1® 0.35 0.44
Prosobee® 0.80 0.65 0.64
GoodStart® 0.75 0.55 0.52

_ 140
~ 120

i
100
80
~
60
t
...
~ 40
20~~--T-~--~~~~
2 4 6 8
moisture (g H20/ 100 g solids)

Figure 1. (0) Tg , ( . ) Tsc, (11) Tac as function of moisture content for Prosobee®

CONCLUSIONS

Lower Tg, T sc and Tac values were obtained at a lower heating rate compared to the faster
heating rates, indicating that these are kinetic phenomena. A slower heating rate provides
a longer residence time for the powder at any temperature, which allows observation of
collapse at lower temperatures. Caking was observed during storage when the
temperature was greater than T g. Tg can be used as an index of relative physical stability
of food powders. Caution must be taken when predicting the aw at which powder cakes
by linear extrapolation of Tsc, since the predicted aw values were higher than the observed
during storage, whereas, Tg value were quite close to the actual values.

REFERENCES

1. Tsourouflis, F., Flink, J.M. and Karel, M., Loss of structure in freeze dried carbo-
hydrate solutions: effect of temperature, moisture content and composition. J. Sci.
Food Agric., 1976, 27, 509-519.

2. Levine, H. and Slade, L., Interpreting the behavior oflow moisture foods. In Water
and Food Qualitv, ed. T.M. Hardman, Elsevier Applied Science Publishers, London,
1989, pp. 71-134.

3. Lazar, M.E. Brown, A.H., Smith, G.S., Wong, F.F., and Lindquist, F.E., Exper-
imental production of tomato powder by spray drying. Food Techn. , 1956,10,
129-133.
 TENTATIVE OF INDUCING FRUIT SUGAR CRYSTALLIZATION DURING FREEZING TO
REDUCE THE HYGROSCOPICITY OF THE CORRESPONDING FREEZE~DRIED POWDERS

S. DE MELO, T. GIAROLA and J. CAL-VIDAL

Department of Food Science, ESAL
37.200-000 Lavras MG, Brazil

ABSTRACT
Freeze-dehydration is a higly recommended process to obtain powders from
fruit juices and extracts. These powders are very hygroscopic materials
which make them very susceptible to the caking problem. Solutions with a
high sugar content, when freeze-dried, yield products with a high degree of
amorphous sugars.
In this investigation, model solutions containing 10% (weight basis) of su-
crose, glucose and fructose were slow frozen. Corresponding sugars for seed-
ing were added as well as methanol, forming two separate systems. Viscosity
determinations and photomicrographs were made at some time intervals and
the results show a significant effect of sample seeding. Crystal formations
were also observed.

INTRODUCTION
In the freeze-dehydration of solutions with a high sugar content -such as
fruit juices- a product with a high degree of amorphous sugars is obtained
(1,2). This type of material is very hygroscopic presenting high suscepti-
bility to the caking problem also found in other food powders (3).
The possibility of reaching a highly organized structure during freez-
ing and drying could reduce the hygroscopic phenomena, considering the fact
that crystalline sugars have a lower water sorption potential. Some previ-
ous literature reports have indicated that adding some specific substances
to sugar solutions, before freezing, a certain crystalline degree may be
obtained. The addition of glucose seeds and alcohol had a positive effect
on sugar crystallization and sugar solubility, respectively.
Freezing is a very fundamental step in the freeze-drying process, being
responsible for many structural characteristics of the final product.
There are several factors affecting the degree of crystalline structure
reached by these molecules. A basic one relates to their ability to achieve
an equilibrium configuration. Fast freezing processes when applied to sugar
solutions promote a very low molecular mobility with a high sample viscosi-
ty and the impossibility of achieving an equilibrium configuration and giv-
ing as result the amorphous state. On the other hand, if the freezing proc-
ess is conducted at lower rates, with stirring, the viscosity increase can
be related in some degree to the nucleation phenomenon.
 195
This study offers a contribution to the problem of sugar crystalliza-
on during slow freezing, using model solutions that had their viscosity
vels monitored and the formation of crystalline phase evaluated through
croscopic analyses.

MATERIALS AND METHODS

gar solutions (10% w/v) containing individually sucrose, glucose and fruc-
se were prepared, under constant stirring (300 RPM) conditions.

eding preparation
edings were prepared from each different sugar, using a disc mill Polytron
d a set of screens (opening = 0.15mm). The particles obtained were then
ied in a drying cabinet set at 55C to reach a final moisture content of
out 3.0%. Treatments received by the 10% sugar solutions, were: sucrose,
ucose and fructose (seedings: 3.0, 6.0 and 9.0% each; methanol 4.0, 8.0
d 12.0% each).

oling/Freezing and Seeding

mples were cooled to 2C and then submitted to slow freezing (rate: 1.5C/
ur), with stirring (300RPM). At regular time intervals viscosity deter-
nations and photomicrographs were performed. Sugar seedings were made
ring cooling when the solution reached -lC. Methanol was also added befo-
the freezing process, in a separate experiment.

scosity and Microscopic Determinations

scosity determinations were made at regular intervals using a viscometer
ookfield, model LVF, with spindle n9 5 and velocity of 10RPM, during 60
conds. Photomicrographs were obtained under polarized light with a Karl
iss Microscope, at each rheological determination. All estimations were
ie in triplicate.

RESULTS AND DISCUSSION

led Seeding Effects

~nges in viscosity of sucrose (A), glucose (B) and fructose (C) solutions
th added respective seedings are presented in Fig. 1. It is remarkable
= viscosity increase for samples with high seeding content. After 250 mi-
tes of observation crystal formations can be detected by the microscopic
~lyses as shown in Fig. 2.

10 VISCOSITY. 1000 CPS

c
ZO;·'~SC~O~S'~TY~.~'~~~C~'S~__________- - ,

.,
Jll.IILIIAJ
-0- CONTROL
...... "to
B 15
mR!!W
-0- CONTROL
.... s"to
..,. ."to -0- ,0,•
10 ...... ,OJ. 10 -+- 90;.

oL-~~~~-=~~~~~~~. oL-~~~C-~__~~~~-=~~
100 1110 zoo zeo lIOO !IIO 0 110 zoo zeo 300:JllO 0 110 100 2110 300 350
TillE. min. TillE. min. TIME. min.

?ig. 1 Changes in viscosity of sucrose (A), glucose (B) and fructose (C)
196

........ ' . " ~'

I.D

Fig. 2 Typical crystal formations in sucrose solutions

As indicated in the literature t viscosity increases may be associated
with possible molecular arrangements taking place on the solution environ-
ment, favoring crystal structuring.
Added Methanol Effects
Viscosity changes in samples having methanol added are shown in Fig. 3.Hig-
her values were obtained for fructose solutions, which offered a clear indi-
cation of crystal formation in all levels of methanol addition (Fig. 4).
Yl'COSITy 1000 C"'
Il ytSCO"'Y . l1OOO C.~I til
I.
., .!!1la!IJIL
IILlAIIQL
B
..
'1 10 ..- C.ON' ''Oi...
....... (ONTROL
10
..... ,.
--+- "'"'..
•s ...... 04,-1
-+- -' .
.
• ...0- 12 '"I. "0- IZ · /.

• 10

40 '20 240 300

TIIIII . mil'&.
1 •• (. ", In. Titil. ""In.
Fig. 3 Viscosity changes in samples having methanol. (Solutions of
sucrose (A), glucose (B) and fructose (C)

Fig. 4 Crystal formations. (Subindices refer to methanol additions; 1: 4%

2: 8%; 3: 12%)

CONCLUSIONS
Fruit sugar crystallization, at low temperatures, is possible with added
seedings, stirring and slow freezing, after certain induction time.
The addition of methanol, under the above conditions, favored crystal for-
mations on fructose solutions.

REFERENCES
1. Makower, B. and Dye, W.B. Equilibrium moisture content of dehydrated ve-
getables. J. Agric. Food Chern., 1956 ~(1):72-8l.
2. Simatos, D. and Blond, G. The porous texture of freeze-dried products.
In Freeze Drying and Advanced Food Technology (S.A. Goldblith, L. Rey
and W.W. Rothmayr, eds.) Academic Press, New York, 1975, pp.40l-ll.
3. Flink, J.M. Application of freeze drying for preparation of dehydrated
powders from liquid food extracts. In Freeze Drying and Advanced Food
Technology (S.A. Goldblith, L. Rey and W.W. Rothmayr, eds.) Academic
Press, New York, 1975, pp. 309-29.
 DIFFUSIVITY OF WATER IN FLOUR/RAISIN MIXTURES

AE. KOSTAROPOULOS\ V.T. KARATHANOSI AND G.D. SARAVACOS2

IDept. of Food Science and Technology, Agricultural University, Athens 11855, Greece
2Dept. of Chemical Engineering, National Technical University, Athens 15773, Greece

ABSTRACT

The dlffusivity of water is an Important transport property in the processing and storage
of food products. Dlffuslvlty data are essential In modeUing and calculations of water transport in
flour/raisin mixtures (bakery products). The water dlffuslvity at 20-60% moisture and 15-7CY'C was
determined from moisture distribution data In two contacted cylinders of wheat flour/raisin pulp,
applying the diffusion equation. Higher water diffuslvltles were found In the flour than in raisin A
sharp discontinuity of moisture distribution was observed at the flour/raisin interface, with higher
concentration in the raisin side, apparently due to differences In water activity.

INTRODUCTION

The diffusion of water In food systems is a fundamental physicochemical process which is

important in the proceSSing, storage stability and quality of various products. Recent advances in
modelling and computing in Food Science and Engineering require reliable thermophyslcal and
transport properties of food materials. Experimental data on water dlffusivlty are necessary, since
theoretical prediction Is not possible, due to the complex structure of foods (I).
Raisins, produced usually by sun-drying of seedless grapes, are consumed either as dried
fruit or as mixtures in cereals and bakery products. The transport of water in flour/raisin systems
Is Important in physicochemical interactions that affect storage stability of these products. Some
data on the water dlffuslvity during air drying of raisins are available (2). The dlffusivlty of water
In starch materials has been Investigated in more detail (3). The dlffuslvlty Is affected strongly by
the physical structure of the material (granular, gelatinized or extruded starch). Sorption Isotherms
of raisins and starch materials are available (4, 5). However, little is known on the transport
(dlffuslvity) of water in flour/raisin mixtures.

MATERIALS AND METHODS

Sun-dried raiSins, produced from seedless Sultana grapes in Southern Greece, and
commercial wheat flour were used in these experiments. The diffuslvity of water was determined in
raisins and in flour/raisin mixtures, using the moisture distribution method (6).
The dried raiSins, containing about 20% moisture (dry basis), were pulped In a laboratory
mill. Two cylindrical samples of raisin pulp, containing different moisture content (e.g. 60% and 20%)
198
were prepared by filling and pressing the pulp Into a plastic tube 12 mm diameter and about 150
mm long. The flour/raisin system was prepared by fining the tube with hydrated flour (dough),
containing 55% moisture, and raisin pulp (20% moisture). The samples were kept at a constant
temperature (150 to 700C) until a characteristic moisture distribution was developed.
The moisture distribution was measured by cutting the cylindrical samples into slices 2-3
mm thick and determining the moisture content gravimetrically. The water dlffuslvlty (D) was
determined by fitting (regression analysis) the experimental moisture distribution to the appropriate
solution of Flck's diffusion equation (I, 6):

- - - - = erfc [ [1]

where, Co, Ce and C are respectively the moisture contents at the beginning, at equilibrium and after
time (t), and z Is the penetration depth.
The sorption Isotherms (water activities) of flour and raisins were determined using a water
activity apparatus (Rotronlc) and the standard gravimetric method of COST-90 (4).

RESULTS AND DISCUSSION

A typical moisture distribution curve for the system raisin/raisin, shown In Fig. I, indicates
an expected transport of water from the 'wet' to the 'dry' sample. The sharp change in moisture
content near the Interface suggests a low water diffuslvlty In the raisin pulp. The computed values
of water dlffuslvlty In the raisin (eqn. 1) range from lxlO-1o rTf/s to 3xlO-lO rTf/s, which are similar
to the diffusivitles obtained from drying data (2).
A completely different moisture distribution was found in the system of flour/raisin, with
a sharp discontinuity near the Interface of the two materials (Fig. 2). Water was transported from
the hydrated flour to the dried raisin and the moisture content near the Interface was considerably
higher In the raisin than in the flour.
The apparently anomalous distribUtion of moisture near the flour/raisin Interface can be
explained on the basis of water activities of the two materials. The water sorption capacity of the
dried fruits Increases sharply at high water activities, due to the high sugar content. On the other
hand, the sorption capacity of flour (mainly starch) levels off at high water activities, approaching
30-35% moisture content at saturation Thus, at a moisture content of 30% the water activity of
flour Is about I, while the water activity of raisin is about O.B, resulting in a driving force of water
activity which transports water from the flour to the raisin, although the difference In moisture
content may be negative.
The mean water diffuslvltles In flour and raisin, computed from the moisture distributions
of the mixture (Fig. 2) were 5xlo-lO rTf/s to 3xlo-10 rTf/s respectively, close to the reported dlffuslvity
values (2, 3).
At low water activities, the sorptlve capacities of the two materials are reversed, I.e. at a
given moisture content, the water activity in the dried raisin Is higher than In the flour. Thus, water
would diffuse from the dried fruit to the flour product-a problem In mixtures of cereals and dried
fruit. The data obtained In our Investigation are applicable to the high moisture region of flour/raisin
mixtures, e.g. raisin bread and cakes. Changes In moisture content of food components during
storage could affect the quality of the product, e.g. texture, chemical and microbial stability.
 199

D·100.------------------------------,
.Ll

,
Raisin
A

00
-
0
0

v
x
ED
f-
Z
W
z 40
f-

0
u Raisin
w 20
a
::J
f-
en
..... 0
0 -50 -40 -30 -20 -10 0 10 20 30 40 50
:L
DISTANCE FROM INTERFACE, mm

Figure 1. Moisture distribution In raisin/raisin cyhnders (T=15°C, t=168 h).

D100,------------------------------,
.D

,
A

0 00
0
....
v
x Flour
60
f-
Z
W
f-
z 40
0
u
w 20
a
::J
f-
en
..... 0
0 -50 -40 -3D -20 -10 0 10 20 30 <10 50
:L
DISTANCE FROM INTERFACE, mm

Figure 2. Moisture distribution In flour/raisin cyhnders (T =7fJ'C, t=48 h).

REFERENCES

1. Saravacos, GoO. Mass transfer properties of foods. In Engineering Properties of Foods, eds. M.A
Rao and S.S.H. Rizvi, Marcel Dekker, New York, 1986, pp. 89-132.
2. Saravacos, G.D. and Raouzeos, G.S. Dlffuslvlty of moisture In alr-drylng of raisins. In 'Drying '86',
vol. 2, ed. AS.Mujumdar, Hemisphere Publ, New York, 1986, pp. 487-491.
3. Marousls, S.N., Karathanos, V.T. and Saravacos, G.D. Effect of physical structure of starch
materials on water diffuslvlty. J. Food Proc. and Preservation, 1991, 15, 183-195.
4. Marouhs, 2.B., Tsaml, E., Marlnos-Kourls, D. and Saravacos, GoO. Apphcatlon of the G.AB. model
to the moisture sorption isotherms of dried fruits. J. Food Eng., 1988, 7, 63-78.
5. Iglesias, H.A and Chlrlfe, J. Handbook of Food Isotherms, Academic Press, New York, 1982.
6. Karathanos, V.T., Vagenas, G.K. and Saravacos, GoO. Water dlffuslvlty In starches at high
temperatures and pressures. Blotechnol. Prog., 1991, 7, 178-184.
 THE USE OF DIGITIZED VIDEO IMAGES
FOR MONITORING COLOR AND COLOR EVOLUTION
OF JONAGOLD APPLES DURING SHELF LIFE

FRANS VERVAEKE - EDDIE SCHREVENS'" - JAN VERREYDT'"

KEN PORTIER"'''' - JOSSE DE BAERDEMAEKER
K.U.Leuven, Agricultural Engineering Department, B-3001 Leuven, Belgium
... K. U.Leuven, Laboratory of Plant Production, B-3001 Leuven, Belgium
...... University of Florida, Department of Statistics, Gainesville, FL 32611-0327

ABSTRACT

Color and color distribution are important quality characteristics, especially for multicolored apples. Color
quantification is evaluated as a basis for improving apple quality recognition and for estimation of the color
evolution.
Shelf life color evolution of Jonagold apples was monitored during ripening under storage conditions of 18°C
and 65 %RH. Using digital image ,processing, the apple surface color was analyzed and the color path from
green to yellow (the background color) and the red (blush) in the RGB space was evaluated. Both classical
Bayesian discriminatory analysis and a robust discrimination technique were used to segment the video images
of the apples in background and blush. The color change during shelf life was quantified: the color path of the
background is an important indicator.

INTRODUCTION

The consumer behaviour towards apples is often based on a subjective perception of quality parameters such as
appearance. The color of an apple provides valuable information in examining the ripeness and the freshness.
In a competitive market, fruit color distribution is a very important factor to growers and consumers.
Apple grading can be based on a refined quantitative description of the color of an individual apple as a non
destructive measurement of quality, but should also take into account the dynamics of the apple color [1].
Models of the fruit color evolution during storage and shelf life must be developed to obtain and maintain a
successful consumer response. Recently, researchers have done a lot of work to use neural networks for the
quantification of human quality assessments [2].
The commercial classification of Jonagold apples in auction takes into account the apple mass, the background
color (green-yellow) and the blush (red surface).
The present work will concentrate on the color classification problem of Jonagold apples during shelf life.

MATERIALS AND METHODS

A set of Jonagold apples was used as the source material for the tests on color evolution and the characterization
of color distribution. These apples have green/yellow to red color variations. Some parallel tests for other non-
destructive quality assessments (e.g. firmness using impact response) were carried out on the same fruits at
harvest, after storage in commercial U.L.O. conditions and during shelf life. The latter implied storage in a
 201
climatic room at Isoe and 65%RH. The color analysis during shelf life was scheduled as: immediately after
U.L.O. storage (measurement I), on day 4 of the shelf life experiment (measurement 2), followed by a color
evaluation on every fourth day after the previous measurement (measurements 3, 4, and 5).
To identify the apple color, the fruit was illuminated by four quartz tungsten halogen lamps outside a cylindrical
diffuser. Three rotations of the apple over 90° with its peduncle-calyx axis at right angles to the video camera
resulted in four images to be combined for a whole apple image. The color image was taken by an RGB camera
(JVe TK-1070E) and grabbed by a pc board (Truevision TARGA+). Spectrophotometric measurements were
used for calibration and for optimizing the number of spots to be used for reliable color estimation [3].
A gray level thresholding procedure was performed on the image to separate the sample from the background.

DATA ANALYSIS AND RESULTS

Assessing the appropriate information for color classification of apples can be based on surface spots or on whole
apple images. Since Jonagold apples are widely heterogeneously colored, the color classification was carried
out on whole apple images. '
On each apple two color classes were defined: the background (depicted as green) and the blush (red). Two
methods for the segmentation of the apple images into color classes were tested: normal and non parametric
discrimination.
Spots of homogeneous green/yellow and red apple surface parts of different apples were selected and
characterized by their mean R G and B values and the covariance matrices for each cluster. The number of
spots in the green cluster was 2 693 and I 697 spots were averaged for the red design set. This training set
was used in a discrimination analysis. In the first discrimination method a linear discriminant function was
calculated to impose a classification scheme (this method will be referenced as 'Id') [4]. A second segmentation
method assumed the distribution of the training set to be different from a multivariate normal distribution: the
color classification criteria for every pixel were derived from an eight-nearest-neighbour method (referenced as
'S-nn') [4,5].
In figure I the evolution of the average R and G values during shelf life is given for the background color of
an apple after a V.L.O. storage period of 5 months, using the allocation criteria of 'S-nn'.

DISCUSSION

Apples are biological material, with the typical high variability for all characteristics. To deal with this immense
breadth, during shelf life the change in the apple surface color map can mainly be quantified by the mean R
value change of the background color. The higher R level correlates with the yellowing during the ripening
process.
The two methods provided results for the distribution of color on an apple, that can describe the evolution of
the apple appearance during ripening.
110

110

110 R

ISO
RAND G VALUE
110

IJO
--------~- ........~..
········n.······----··------.6.

o 1 1 J ! 5 I 1 I I 10 11 11 1J 14 15 16

SHELF LIFE [ DAYS 1

Figure 1. R value (0) and G value (A) evolution during shelf life for the background color of a Jonagold
apple.
202
PERCENT PERCENT
II /I

II
U

II

II
U

II
/I

/I
II
II

I I •• _______________~--~~~~~
o ........ . _. green • . • . - •. - . - . 1 0 - . - - - . - . - . -green - - - . - - - - - - - 1
1 --. -------. red --. -. -. -. -. 0 1 . - . -... -- . - red - - .. - - . - - .. 0
POSTERIORI PROBABILITY

Figure 2. Pixel distribution as a function of the posteriori probability using 'ld' (left) or '8-nn' (right)
( • red 181 green 0 no class )
Figure 2 shows the comparison of the segmentation of an apple surface (for measurement 1 of the shelf life
period), using both the 'Id' and '8-nn' method. The latter procedure indicated two percent (of the 95353 pixels)
not classified pixels, due to a posteriori probability of 0.5 in both classes and more pixels were classified as
red. In this study, this parameters showed no interaction with time.

CONCLUSIONS

The process of classifying apples on laboratory scale using pixel color information resulted in a good
quantification of color dynamics during shelf life. The two discrimination analyses provide valuable information
to classify apples. When faced with the real-time requirements of an on-line realisation, an optimization will
be required to reduce computational efforts. The end result would be a more consistent procedure for color
grading.

ACKNOWLEDGEMENTS

The authors are especially gratefull to EUROFRU, for the financial support of this research.

REFERENCES

1. Verreydt, J., Schrevens, E., Keulemans, J., De Baerdemaeker, J., Vervaeke, F., Evolution of quality
parameters of Jonagold apples by means of non-destructive methods. ISHS Post Harvest 1992, Davis, Aug.
9-14.
2. Thai, C.N. and Shewfelt, R.L, Modeling sensory color quality of tomato and peach: neural networks and
statistical regression. Trans. ASAE 34 (3) 1991 : 950-955 .
3. Schrevens, E. and Raeymaeckers, L, Colour characterization of Golden Delicious apples using digital image
processing. ISHS, Sensors in Horticulture. Acta Hortic., March 1992 (304).
4. SAS Institute, SAS/STAT User's guide release 6.03 edition, SAS Institute, Cary, NC 27511, 1988.
5. HAND, D.J., Discrimination and classification, John Wiley & Sons, Chichester, 1981.
 COMPUTERIZED IMAGE ANALYSIS OF GAS CELL
DISTRIBUTION IN SEVERAL FOODS

SHOICHI GOHTANI, NAOKO ARIUCHI*, SETSUE KA WASOME**

AND YOSHIMASA YAMANO
Department of Bioresource Science, Kagawa University, Miki, Kagawa
761-07
*Tokushima Bunri Junior College, Yamashiro, Tokushima 770
**Kagawa-ken Meizen Junior College, Kameoka, Takamatsu, Kagawa 760

ABSTRACT

Computerized image analysis was very practical for determining both the
gas cell distribution in bread and sponge cake and the hole distribution in
kori tofu because of its rapid, convenient measuring procedure. Both the
number and the area ratio of gas cells correlate to the specific volume in
both bread and sponge cake. Both the diameter and the area of holes de-
creased with swelling by water uptake in kori tofu.

INTRODUCTION

The gas cell distribution is an important factor for determining the texture
of baked cereal products such as bread and sponge cake. In order to esti-
mate the gas cell distribution, a subjective sensory test has been normally
practiced. However, the sensory test requires a skilled sensory testing
panelist and takes time. In this study the usefulness of the computer image
analysis method in food science was investigated in order to develop a
rapid, convenient method to determine both gas cell distribution in baked
cereal products and hole distribution in kori tofu.
204
MATERIALS AND METHODS

A variety of breads were prepared by the method of Goshima et al[l] and

sponge cakes by the method of Kawasome et al[2]' The sliced cross sec-
tion of both the bread and the sponge cake were stained with charcoal ink
and then pressed on to paper to produce a print(print method) or pho-
tographed(photo method). The cross sectional images were analyzed with a
computerized image analyzer(Nireco:LuzexIIIU) and gas cell parameters
were determined[3].
Kori tofu, a freeze-dried form of a traditional Japanese soy bean food,
was obtained from Asahimatsu Shokuhin Co. Ltd. The cross sectional
image was transferred to the image analyzer directly since the ink ran in
water.

RESULTS AND DISCUSSION

CI
w
a::: 30
:::> ..r-
oo' E
25 y=O.848
L5 ~
w
~
00 N 20 a<O.OO1
.....I ~
.....I
W «15
•
z
()
00
«
« w 10
C!:I C!:I
u. ~
«
0
a::: :r: 5
w ~
m ~
~ 0
:::>
z 0 5 10 15 20 25 30
NUMBER OF GAS CELLS
COUNTED MANUALLY(cm-2)

Figure 1. Correlation between the number of gas cells measured with image
analyzer and that counted manually[3].
 205
Figure 1 shows that the number of gas cells determined by the print method
with image analyzer correlates positively to that counted manually(y=O.848,
a<O.OOl). A positive correlation was obtained between the number of gas
cells determined with the photo method and the print method. These re-
sults indicate that computerized image analysis is practical for determining
the gas cell distribution in either the print or photo method[3].
Negative correlations were found between the specific volume and the
number of gas cells for both bread and sponge cake(y=-O.758, a<O.02 and
y=-O.652, a<O.05 respectively), while positive correlations were found
between the specific volume and the area ratio of gas cells for both bread
and sponge cake(y=O.736, a<O.05 and y=O.806, a<O.Ol, respectively).
For sponge cake, the area equivalent diameters correlated positively to the
specific volume(y=O.776, a<O.Ol) and negatively to the number of gas
cells(y=-O.923, a<O.Ol), while no correlation was obtained between these
gas cell parameters for the bread. It is considered that the shape of the
gas cells in bread are deformed from spherical owing to the remarkable
evolution of carbon dioxide during fermentation, while the shape of those in
sponge cake remain spherical during baking[3].
For kori tofu, neither the swelling ratio nor the hole diameter could be
determined below 50% of water content since the uptake of water was too
uneven. Above 50% of water absorption, both the area equivalent diameter
and area ratio of holes decreased remarkably with an increase of swelling
ratio. It is considered that the water absorbed by kori tofu is mostly
taken up by the protein which makes up the wall networks of kori tofu
rather than directly into the holes above 50% of water content.

REFERENCES

1. Goshima, G., Watanabe, Y., Shingu, H., Isosaki, H., Tsuge, H. and
Ohashi, K, Comparison of baking quality of bread flours milled from
domestic and foreign wheats. Nippon Shokuhin Kougyou Gakkaishi,
1986, 39,102-107.
2. Kawasome, S. and Yamano, Y., Effect of butter content on the texture of
sponge cake. 1. Home Econ. Ipn., 1986,37, 759-766.

3.Gohtani, S., Ariuchi, N., Kawasome, S. and Yamano, Y., Computerized

image analysis of sudachi(gas cell distribution) of baked cereal products.
Nippon Shokuhin Kougyo Gakkaishi. 1992, 39, 749-754.
 FUNCTIONAL PROPERTIES OF Lupinus luteus PROTEINS

I. M.N. SOUSA*, M. L. BElRAO da COSTA*, S. E. HILL**, J. R. MITCHELL**

and S.E. HARDING**
*S.A.C.T.A.I Instituto Superior de Agronomia, Technical University of Lisbon
Tapada da Ajuda, 1399 Lisboa Codex. Portugal
** A.B.F.S. Faculty of Agricultural and Food Sciences, University of Nottingham
Sutton Bonington. Loughborough. LE12 5RD U.K.

ABSTRACf

The thickening potential and the gelation ability of lupin protein isolates were studied using
soy isolates as a comparison. Lupin major globulin fractions were characterised by
ultracentrifugation. Three globulins (2.5S, 7.7S and l2.2S) were present and these were
associated with the three peaks seen in the differential scanning calorimetry (D.S.C.)
thermograms. The molecular masses of the two main globulins were found to be 390 kD and
90 kD. The lupin and soy isolates showed similar solubilities. The intrinsic viscosity of the
soy isolates was higher (12 cm3 .g- l ) than the lurin (7 cm3.g- l ). The soy viscosity was
consistently higher (1.2 Pas against 0.2 Pa.s at 50 s- ,23% isolate concentration). The D.S.C.
denaturation temperature of the lupin globulins was higher than the soy globulins. The gelling
behaviour of lupin protein was very poor when compared to soy protein even when slightly
improved by promoting the Maillard reaction. It was concluded that lupin globulins have a
stronger hydrophobic nature which explains the higher thermal stability, poor thickening and
gelling properties.

INTRODUCTION

Lupin, specially Lupinus luteus, is a legume that can be produced in marginal soils and is part
of an environmentally friendly agricultural system traditional in Portugal. The presence of
alkaloids, in bitter varieties, prevents the direct use of the unmodified seeds in human foods.
The isolated protein is alkaloid free and has potential in human food applications currently
employing soy isolates. The success of this concept will depend on how the functional
properties of lupin and soy proteins compare.
In this paper we describe the characterisation of lupin proteins and compare their
solubility, thickening and gelation properties with soy.
 207
MATERIALS AND METHODS

Lupin (from L. luteus) and soy isolates and the major lupin globulins were produced as
previously described (1, 2).
The proteins were characterised by ultracentrifugation (Beckman XL-A analytical
ultracentrifuge pH 7, 11 = 0.01), D.S.C. (Perkin Elmer D.S.C.-2, water, heating rate of
5°C/min.) and by intrinsic viscosity (pH 7,11 = 0.01). Gelation properties were determined by
heating proteins in Universal bottles at a range of times and temperatures. In some cases
xylose was incorporated to promote the Maillard reaction (3). Solubility was measured as
previously described (4) and flow behaviour was determined using a Bohlin CS rheometer
equipped with concentric cylinders geometry.

RESULTS AND DISCUSSION

Protein characterisation
The sedimentation velocity studies for the major lupin globulin fractions gave (Sw 20) values
of 12.2S, 7.7S and 2.5S and molecular masses obtained by sedimentation equilibrium of 390
kD and 90 kD for the fIrst two globulins.
DSC lupin isolate thermograms revealed three peaks (peak maxima 372 K, 381 K and
387 K) which we assign to the denaturation of the three globulins. The soy isolate
thermograms revealed only two peaks (at 368 K and 384 K). The denaturation temperature of
the soy 2S is close to the soy 7S and peak is thus hidden. The lupin peaks were broader (less
cooperative) suggesting that there is aggregation between 7S and lIS globulins following
denaturation.

Functional properties
The solubility of lupin is similar to that of the soy (fIg. 1). A high solubility indicates the
absence of denatured protein and is a requirement for thickening and gelation.

100

• Soy isolate
% 50
A Lupin iso.

2 3 4 5 6 7 8
pH

The shear viscosity of 23% isolate cold suspension at 50s- 1 was 0.2 Pa.s for lupin and
1.2 Pa.s for soy. This is consistent with the intrinsic viscosity which was 6.7 cm3 .g- 1 for
lupin isolate and 12.3 cm3 .g- 1 for soy.
208
The heat gelation properties of lupin isolate were very poor compared with soy over the
whole isolate concentration (20 - 30%), pH (5.00 - 9.00) and salt addition range (0 - 0.5M)
studied (5). Even at the upper concentration limit it was not possible to form a coherent gel
from lupin and in contrast to soy (6) the addition of xylose did not result in a significant
improvement.

CONCLUSIONS
We suggest that the poor gelation properties reflect the greater hydrophobic nature of
the lupin protein system. This causes aggregation rather than gelation. Supporting evidence
for this high hydrophobicity comes from the high thermal stability and low hydrodynamic
volume.

ACKNOWLEDGEMENT

We wish to thank Dr. Peter Morgan for assistance in the ultracentrifugation

measurements. This work was supported by a EEC 'SCIENCE PROGRAM' grant (JNICT,
Portugal).

REFERENCES

1. Wolf, W. J., Soybean proteins their functional, chemical and physical properties. L Agric.
Food Chern., 1970, 18 (6) : 969 - 977.

2. Suchkov, V.V., Popello, 1., Grinberg, V. Ya. and Tolstogusov, V.B., Isolation and
purification of 7S and lIS globulins from broad beans and peas. L ~ Food Chern.,
1990,38 (1) : 92 - 95.

3. Hill, S. E., Mitchell, J.R. and Armstrong, H.J., The production of heat stable gels at low
protein concentration by the use of the Maillard reaction. In Gums and Stabilisers for the
Food Industry Q. Phillips, G.O. et aI., Eds. IRL Press. Oxford University Press. Oxford,
1992, pp. 133 - 136.

4. Shen, J.L., Solubility and viscosity. In Protein Functionality in Foods. Cherry, J.P. , Ed.,.
ACS Symposium Series, 147. Washington D.C., 1981, pp. 89-109.

5. Sousa, 1.M.N., Beirao da Costa, M.L., Hill, S.E. and Mitchell, J.R., Gelation of L. luteus
protein isolate: a response surface analysis. VI International Lupin Conference, Temuco.
Chile. November 1990.

6. Larsen, L. B., Sousa, 1.M.N. and Beirao da Costa, M.L., The use of the Maillard reaction to
improve gelation of soy and lupin proteins. In Advances in Lupin Research. "Proceedings
of the VII International Lupin Conference". In press, 1993.
 STABILITY OF GLUTATHIONE IN SOLUTION

TOSHIO YAMAMOTO. KAZUOKI ISHIHARA

Institute for Intestinal and Environmental Microbiology. Advance Co .. Ltd.
1-35-2. Shimoisihara. Chofu-shi. Tokyo. 182. Japan

ABSTRACT

Decomposition of Glutathione (GSH) in aqueous solution was an apparent

primary order reaction and depended on temperature. Half life of GSH was
lengthened by anaerobiosis and presence of citrate and L-ascorbic acid. The
apparent activation energy of decomposition of GSH was increased in the
presence of L-ascorbic acid.

INTRODUCTION

Glutathione (GSH) has scarcely been added into aqueous food. because it has
been thought instable in aqueous solution. However. few papers have been
published on stability of GSH in aqueous solution[l]. We examined influence
of organic acids and L-ascorbic acid on decreasing rate of GSH in aqueous
solution at l~w temperature (4 to 40 ~) under aerobic and anaerobic
conditions.

MATERIALS AND METHODS

Materials
Glyoxalase I and Methylglyoxal were obtained from Sigma Chemical Co.
All other chemicals used in this experiment were of the highest grade
available.

Deter.ination of GSH
GSH concentration was determined by Glyoxalase method [2] and HPLC [3].

Preparation and storage of reaction .ixture

GSH was dissolved in 0.02 M organic acid buffers (pH 3.5) containing 0 to
0.05 % L-ascorbic acid. GSH solution (3.3 mM or 13 mM) were poured into
210
test tubes and heated at 90 ·C for 10 minutes and quickly cooled in water.
A half of the test tubes was sealed in an anaerobic glove box (H2-10 %,
C02-10 %, N2 -BO %) and stored in anaerobic jar « 1 ppm 02). The other was
sealed under aerobic condition. These test tubes were stored in constant
temperature incubator at 4, 10, 20, 30 and 40 ~.

RESULTS AND DISCUSSION

Influence of organic acids
The semi-logarithmic plots of the ratio of remaining GSH vs storage time
were linear at any conditions in this experiment. Half life of GSH in
0.02 M citrate, malate, acetate, tatarate and formate buffer at 30 ·C under
aerobic condition was 4, 2, 2, 3 and 3 weeks, respectively. As this fact
suggests that citrate contributes to stabilization of GSH, citrate buffer
was used in the following experiments.

Influence of temperature. L-ascorbic acid and anaerobic condition

As shown in TABLE 1, the half life of GSH in anaerobic condition was 1.1 to
1.B times as long as that in aerobic condition. The presence of 0.05 %
L-ascorbic acid lengthened the half life of GSH 10 % to BO % at each
storage temperature and this effect was remarkable at low storage
temperature.

TABLE 1
Half life of GSH under aerobic and anaerobic conditions.

Half life weeks ) of GSH

Storage temperature Aerobic condition Anaerobic condition
( ·C) L-ascorbate (%) 0 0.05 o 0.05

4 IB.7 32.B 27.3 44.4

10 15.5 25.5 16.4 28.1
20 7.0 8.9 B.l 9.4
30 3.4 3.B 5.9 6.2
40 1.6 1.8 2.3 2.5

Apparent activation energy of GSH in citrate buffer without L-ascorbic

acid was 51 kJ/mol on both aerobic and anaerobic conditions. In the
presence of 0.05 % L-ascorbic acid. it was 60 kJ/mol and 57 kJ/mol under
aerobic and anaerobic conditions, respsctively. Addition of L-ascorbic acid
increased the activation energy of GSH in aqueous solution ( Figure 1 ).
The activation energy of anaerobic decomposition of GSH according to Aruga
et al. was B4 kJ/mol at 50 to 80 ·C [1]. This difference of activation
energy may be due to difference of range of temperature at which
experiments were carried out.
From the results obtained in this study, it is suggested that the
addition of GSH to aqueous foods is available if they are stored at 10 ·C
or below.
 211

-0.5

log ko -1. 0

-1. 5

-2.0

3.2 3.4 3.6

lIT x 10- 3

Figure 1. Temperature dependence of decreasing of GSH. ( 0 ) : with 0.05 %

L-ascorbic acid. ( . ) : without L-ascorbic acid under aerobic condition and
( D ): with 0.05 % L-ascorbic acid. ( • ) :without L-ascorbic acid under
anaerobic condition.
ko:the primary order reaction rate constant ( l/week ).

REFERENCE

1. Masayoshi.A .• ShoJi.A. and Manabh.H .. Kinetic Studies on Decomposition

of Glutathione. IT .Anaerobic Decomposition in Aqueous Solution. Chem.
Pharm. Bull.. 1980. 28. 514-520.

2. Bernt. E. and Bergmeyer.H.U .. in Methods of Biochemical Analysis. ed.

Bergmeyer.H.U .. Verlag Chemie. Weinheim. 1974. pp. 1643.

3. Hideo.K .. Eisuke.F. and Kunio.T .. An Epoch of Glutathione Reserch. ed.

Kouichi.N .• Kyoritu Syuppansha Co .. Tokyo. 1988. pp. 1392-1396.
 AROMA OF Thymus zygis. CHEMICAL AND SENSORIAL ANALYSIS

M. MOLDAO-MARTINS, *G. BERNARDO-GIL and M.L. BEIRAo DA COSTA

Dep. of Food Sci. & Tech., Inst. Sup. Agronomia, Tapada da Ajuda, 1399 Lisboa Codex.
*Dep. of Chern., Inst. Sup. Tecnico, Av. Rovisco Pais, 1096 Lisboa Codex, Portugal

ABSTRACT

Thymus zygis is a botanical species commonly found in aromatic herbs. The aim of this paper is to
study the composition of extracts obtained by steam and Clevenger distillation, N2 flow and the
sensorial analysis of the same extracts. Chemical components were evaluated by capillary GC and GC-
MS. Sensorial analysis was done by an untrained test panel of twenty judges. The minimum detection
level, the preferred concentration and the preferred extract were evaluated in a vegetable oil having
different added extracts and compared to the same oil having fresh herb added. Results show that the
minimum detection level for 75% of the judges is about 0.01 % for Clevenger and steam distillation
extracts and 0.04% for N2 flow extract. The preferred extract was the one produced by steam
distillation at 0.03% concentration. The disliking of the extracts seem to be related to the presence of
p-cymene and 'Y-terpinene.

INTRODUCTION

Thymus zygis aroma is used in the food industry. Several methods of extraction have been in literature
[1] and the respective chemical composition assessed. Nevertheless the knowledge of the relationship
between the chemical and sensorial profiles is still quite scarce. The aim of the present work is to look
for this probable effect and to find the kind of extraction procedure more suited in terms of sensorial
response. Different extraction procedures (steam distillation and Clevenger distillation and N2 flow)
were applied to thyme. Chemical characterisation was conducted by capillary GC and GC-MS. The
extracts were added to a sunflower oil enriched in oleic acid at the same concentration and sensory
level compared to the same oil having a fresh herb quantity containing similar values of essential oils.

MATERIALS AND METHODS

Raw materials
Thymus zygis spp silvestris was collected in the north of Portugal during the flowering period, and
stored in the dark at room temperature for a week.

Extraction methods
Steam and Clevenger distillation were carried out at atmospheric pressure for 30 min. N2 flow
extraction was conducted at 333 K, 2.7 kPa at a flow rate of 400 mUmin for 60 min.
 213
Methods of analysis
GC analysis were performed with an HP 5890 instrument equipped with an HP-5 (5% biphenyl, 95%
dimethyl polysiloxane (50 m X 0.32 mm; coating thickness 0.17 /-Lm» and a FID.
Analytical conditions: injector temperature 473 K; detector temperature 523 K; oven temperature
333 KIlO min; oven temperature programmed 337-453 K at 2 Klmin, then 453-473 K at 10 Klmin,
then isothermal at 473 K for 50 min; carrier gas flow 2 mUmin.
Relative concentrations were evaluated using peak areas, without correction for response factor
GC-MS analyses were performed with HP-5890 instrument equipped with a WAX capillary column
(30 m x 0.32 mm; coating thickness 0.1 /-Lm) and 7.00 Mevolts mass selective detector.
Analytical conditions: injector temperature 503 K; oven temperature 333 Kl15 min, oven
temperature programmed 333-493 Kat 2 Klmin, then isothermal for 15 min; carrier gas flow 2 ml He/
min; source 7.00 Mevolts. Standards of most of the components were used for identification of
compounds. When it was impossible to use standards, the mass spectrum of an unknown compound
was compared with the library.

Sensorial Analysis
Minimum detection level and preferred concentration-increasing level of each extract from 0--0.05%
(ml of essential oil/ml of vegetable oil) were added to oleic sunflower oil and tested by an untrained test
panel of twenty judges. The minimum detection level was identified with the concentration detected by
75% of the judges [2]. Preferred concentration corresponds to the one preferred by most of the judges.
Preference analysis was conducted in the same vegetable oil containing selected concentrations of
each extract and of oil added to the fresh thyme in a similar essential oil concentration to that found in
the steam distillation extract. All the samples were randomised and presented simultaneously to the
judges who were asked to order the samples by increasing preference for taste and flavour attributes.

RESULTS AND DISCUSSION

Chemical composition of the different extracts is shown in Figs 1,2,3. It can be seen that extracts of
distillation methods are similar in the kind of components but Clevenger extracts show higher values of
more volatile compounds (ex-thujene, ex-pinene, camphene, myrcene, limonene and mainly in p-cymene
and "{-terpinene). N2 flow extract is quite different showing only the more volatile compounds being
the main constituents p-cymene and "{-terpinene. Minimum detection level for N2 flow is lower than for
steam distillation (0.005% and 0.01% respectively). The preferred concentration is 0.03% for
distillation extracts and 0.02% for N2 flow extract. The evaluation of the extracts by sensorial analysis
is shown in Fig. 4. It is quite apparent that steam distillation extract is more appreciated by the panel.
N2 flow extract is not preferred or is just indifferent for most of the judges. Relating the chemical
composition with sensorial analysis it seems that the presence of p-cymene and "{-terpinene is detri-
mental in what concerns both taste and flavour attributes. The oil added to the fresh plant was not
found to be acceptable by the panel.

~ :~L", ..".. J.L ::L,h"..,J.L,

40 30

><
1 3 5 1 91113151119212325212931 4 7 10 13 16 19 22 2S 28 31

Compound Compound

Figure 1 Essential oil profile from steam dis- Figure 2 Essential oil profile from Clevenger
tilla tion extract. distillation extract.
214
60

~ ~L,"1""H""""''''''
• Like

• Indiff.

o DisJi<e
1 3 5 7 91113151719212325272931
Compound Flavour Taste

Figure 3 Essential oil profile from N2 flow Figure 4 Comparative sensorial analysis of
extract. different extracts.

1- a-thujene; 2- a-pinene; 3- camphene; 4-j3-pinene; 5- myrcene; 6- a-felhandrene; 7- a-terpinene; 8-limonene; 9-

p-cymene; 10- 1,8-cineole; 11- 'Y-terpinene; 12- octanol; 13- terpinene-1-ol; 14- undecane; 15- linalool; 16-
camphor; 17- borneol; 18- terpinene-4-ol; 19- hexyl-butirate; 20- nerol; 21- neral; 22- geraniol; 23-linalyl-acetate;
24- bornyl-acetate; 25- thymol; 26- carvacrol; 27- geranyl acetate; 28- j3-carophylene; 29- dodecanal; 30- a-
humulene; 31- oxide j3-caryophylene; 32- tetradecanal.

REFERENCES

1. Gagnon, M. J. and Marcoux A. M., A comprehensive study of three procedures for the isolation of
volatile flavour composition from a model solution. Acta Aiimentaria, 1989, 18 (3), 283-297.
2. C.O.I., Evaluation organoleptique de l'huile d'olive vierge, 1992, T.20IDoc. 3/rev. 2, 28 Mai.
 MEASUREMENT OF MATERIAL PROPERTIES IN HETEROGENEOUS
FOOD SYSTEMS BY MAGNETIC RESONANCE IMAGING

M.J. McCARTHY, K.L. McCARTHY, R.L. POWELL*, J.D. SEYMOUR*, and T.-Q. u*
Department of Food Science and Technology
*Department of Chemical Engineering
University of California Davis
Davis, CA 95616-8598, U.S.A.

ABSTRACT
Foods in general are unique heterogeneous mixtures. These materials are heterogeneous on
different scales; e.g., microscopic and macroscopic. Magnetic resonance imaging is an
experimental technique recently applied to study the properties of foods and the dynamics of
changes occurring in foods during processing and storage. This paper will relate magnetic
resonance imaging measurements to the development of transport theories for heterogeneous
materials. These concepts will be illustrated by using magnetic resonance imaging velocity
measurements and relating these measurements to fluid properties for suspensions with
particular emphasis on the length and time scales involved.

INTRODUCTION
The need to improve food quality and food quality control is driving the development and
utilization of new analytical techniques. These analytical techniques include nuclear magnetic
resonance (NMR), magnetic resonance imaging (MRI), ultrasonics, x-ray tomography, as well
as new optical techniques. NMR and MRI appear to have the broadest range of applications
because they are noninvasive, nondestructive and can quantify a large range of material and
structural properties.
MRI measurements of material properties provide more information than classical
measurements, which are most often integral measurements. Unlike classical measurements,
the MRI measurement is resolved spatially as well as temporally. For instance in the past, the
porosity of a bed of glass spheres has been characterized by an average value. Now, an MRI
measurement of the porosity can provide a three-dimensional map of porosity variations in the
bed with resolution from 32x32x32 to 512x512x512 elements. The length scale of each
element is determined by the size of the bed, the MRI equipment, experimental parameters, and
fluid properties (1). For interpretation of MRI measurements in multiphase systems one must
address the relationship of the length scale of the measurement to the length scale of the process
being studied (2). In this paper, length scales and time scales of importance for the study of
fluid motion will be discussed.
216
VELOCITY MEASUREMENTS
The classical model for describing the influence of motion on the nuclear magnetic moments
was presented by Torrey (3):

M "+M "
dM = yMxB- Xl yJ + (MZ -MO)k +DV. VM- V.(vM)
dt T2 Tl

where y is the gyromagnetic ratio, M is the net macroscopic magnetization, B is the applied
external field, Tl 2 are relaxation time constants, 0 is the self-diffusion coefficient and v is
velocity. Solving this equation for a spin-echo phase-encoding pulse sequence, the signal is
(4,5):

S(k,p) = ff drdUp(r)r(r,U)exp[-21ti(k.r+p. U)]

t t*
where, per) = p(r)exp{-t / T2 - y2 D dt * f [f G(t' )dt']2}
o 0

f
t t
f
21Ck = y G(t' )dt' 2np = y t' G(t' )dt'
o o
and p(r)r(r, U) is the joint spatial-velocity density distribution. For fluid flows in which
there are fluctuating components of velocity, a solution can be obtained by decomposing the
fluid velocity into an average and an instantaneous deviation (4,6):

S(k,p) = fff drdUdJ.l/;(r,U)l1(r,J.l)exp[ -21ti(k· r + p. U + p. J.l)]

where, U=U +Jl; U is the average velocity; J.l. is the instantaneous deviation from U; l;(r,U)
and l1(r,J.l) are the probability density distribution functions for the mean velocity and the
fluctuating component, respectively.

DISCUSSION
The fluctuating components of velocity in the phase encode direction result in an additional
phase dispersion of the spin magnetization and hence attenuation of the signal. Other factors
influencing the signal intensity include partial volume effects, molecular selfdiffusion, spin
relaxation, magnetic susceptibility variations and inflow/outflow effects. For pipe flow the
experiments can be conducted to eliminate or correct for spin relaxation and self-diffusion by
using the equations presented previously. Partial volume effects and magnetic susceptibility
variations will be important at the fluid/wall interface where part of each voxel may be inside
the tube wall. Variations in the diamagnetic susceptibility between the fluid and the pipe wall
will result in signal attenuation. Partial volume effects can also weight the velocity infonnation
content of a voxel. If there is a large gradient in the velocity field over a small portion of the
voxel, as occurs in systems with almost complete plug flow and a no slip boundary condition,
the velocity encoded by a phase sensitive method will not reflect these velocity components in
the shear region between the wall and the plug. This results because the velocity is a weighted
 217
average of all velocities within the voxel. This has been reported for the flow of paper pulp
suspension at low velocities (7). Inflow/outflow effects become important when velocities are
high enough such that a significant part of the fluid in a voxel is displaced during the
experiment (8). Inflow/outflow effects influence both signal intensity and fluid velocity
values.
The time scale of the MRI measurement influences the MRI signal. This occurs because
MRI is inherently an averaging technique. The time scale of a phase flow encoding experiment
is approximately lOOms, which is on the same order of time scale as turbulent velocity
fluctuations. This leads to an averaging of the contributions from the fluctuations and a
reduction in signal intensity of the average velocity image of water in tube flow, as in Figure 1.

r
pipe center line

10 20 30 40 50 60 70 80
VELOCITY IN ARBITARY UNITS
Figure 1. Contour plot of the joint spatial-velocity density distribution function for water in
tube flow, Reynolds number of 4100,40 signal averages, OAcrn/s/voxel velocity resolution,
198Jlm/voxel spatial resolution.

REFERENCES

1. Maneval, J.E., McCarthy, M.J., and Whitaker, S., Use of NMR as an experimental
probe in multiphase systems. Water Resources Res. 1990, 26, 2807-16.
2. McCarthy, MJ., Lasseux, D, and Maneval J.E., NMR imaging in the study of diffusion
of water in foods. 1. Food Eng. 1993, in press.
3. Torrey, H.C., Bloch equations with diffusion terms. Phys. Rev. 1956, 104,211-15.
4. Callaghan, P.T., Principles 2f Nuclear Magnetic Resonance Microscopy, Oxford
University Press, New York, 1991.
5. Caprihan, A. and Fukushima, E., Flow measurements by NMR. fhn. R«port 1990,
198, 195-235.
6. Nalcioglu, O. and Cho, Z.H., Measurement of bulk and random directional velocity
fields by NMR imaging. IEEE ~ Med. Imaging, 1987, MI-8, 356-60.
7. Seymour, J.E., Maneval, I.E., McCarthy, K.L., McCarthy, M.J., and Powell, R.L.,
NMR velocity phase encoded measurements of fiber suspensions. J. Phys. Fluid A:
Fluid Dyn. in press.
8. Kose, K., NMR imaging of turbulent structure in a transitional pipe flow. J. Phys.D:
Appl. Phys., 1990, 23, 981-983.
 ANALYSIS OF THE FLOW FIELD IN A SINGLE SCREW EXTRUDER USING
MAGNETIC RESONANCE IMAGING

C.K. AGEMURA, R.J. KAUTEN, AND K.L. MCCARTHY

University of California-Davis
Department of Biological and Agricultural Engineering
Davis, CA 95616

ABSTRACT
The flow field in an extruder was evaluated using magnetic resonance imaging (MRl). This
technique offers a non-destructive, non-invasive method for determining velocity profiles.
Images of Newtonian and non-Newtonian solutions in a model single screw extruder have
been obtained. The flow was characterized at extruder speeds in the range of 0-60 rpm for
the two limiting cases: full closed discharge and full open discharge. The circulation pattern
of the flow was seen by examining a series of time elapsed cross sectional images.
Quantitative velocity information derived from the MR images agreed with theoretical
predictions.

INTRODUCTION

Extruders are versatile systems which combine the processes of mixing, shearing, kneading,
and cooking into one operation. Industries are taking advantage of the system's versatility by
forming foods of a variety of shapes and textures not possible with other food machinery. A
plethora of food products such as ready-to-eat cereals, pastas, snack foods, and pet foods are
produced using extrusion technology.
Most of the theoretical analyses of food extrusion have been derived from studies of
plasticating extrusion. Although the research on the metering section is highly developed, it
is, nonetheless, based on simplifying assumptions. Many of the experimental techniques
previously used to verify theoretical models were performed under conditions which were far
removed from industrial practice. Methods which use dye-injection (1) and particle tracers
(2) to obtain velocity information are not only invasive but usually require a system in which
the barrel rotates at very low speeds. The importance of strengthening mathematical models
and developing them for more complex systems is crucial to aid in the design, optimization,
and scale-up of extruders. This requires systems which more closely simulate industrial
extruders and would relate the three types of extrusion parameters - process, system, and
product properties (3). Process parameters such as screw speed, screw geometry, and
throughput rate are controlled by the operator. In contrast, the residence time distribution
(RTD) and total strain which are measures of the history of the material within the extruder
are system parameters that are dependent on the process parameters. Product properties,
 219
most important to industry, are measures of product quality such as rheological properties,
texture, and color. Defming the relationship between product, system and process parameters
in a simple, useful model would greatly benefit the food industry. This research addresses
the development of a model that focuses on describing the system parameters as a function of
process parameters. The manner in which these parameters are all inter-linked is through the
velocity profile which in turn provides infonuation about the degree of mixing, RTD, and
shear history for prediction of final product quality.
Verification of developed models is possible only with an experimental technique that
can obtain data from these more complex systems. Magnetic resonance imaging offers a
method to obtain velocity profiles non-invasively. In addition, the complete velocity profile-
from screw root to barrel-is obtained in one image (5).
RESULTS
In a previous paper (4), the velocity equations for a simplified rectangular channel analysis
were presented for a straight screw. References (4) and (5) can be referred to for a discussion
of the experimental set-up, theoretical explanation of MRI, and expanded analysis of the
equations of motion. The resulting velocity in the cross-channel direction (x-direction) is not
a function of viscosity (4). For open and closed discharge, the velocity in the down-channel
direction, Vz, is also independent of the viscosity of the fluid. This is seen by defining a
dimensionless variable, a, equal to the ratio of pressure to drag flow. The pressure gradient
in the down channel direction, ap/az, can then be defined as a function of this variable.

_ Qp _ H2 ap
a------- (1)
Od 6Vz /J. az
where H is the channel depth, /J. is the viscosity, and V z is the screw velocity in the z-
direction.
For closed discharge, drag flow, Qd, and pressure flow, Qp, are equal and opposite in
sign, therefore, the ratio is exactly 1. For completely open discharge, there is no pressure
gradient in the z-direction and the ratio is equal to zero. Defining the pressure gradient in
tenus of a and substituting it into the equation for the velocity in the z-direction as a function
of position within the channel (y), yields:

(2)

Therefore, for the two limiting cases, the velocity in the z-direction is also
independent of viscosity. From the theory, the velocity profiles obtained for fluids with
different viscosity should be identical. This is verified by examining the images for a
Newtonian and power law fluid at the same screw speed for closed discharge (Figure 1).
Quantitative flow rates can be obtained from the images. The flow rate in the plane
of the image, Qpt' (cm2/s), is calculated by measuring the area under the velocity profile on
an image. The contour plots in Figure 1 show the outline of typical velocity profiles from
which Q l' is obtained.
\J'erification of theory is illustrated by integrating the velocity in the plane of the
image across the channel for three screw speeds. The expected flow rates are compared to
the theoretical predictions for the flow rate in the plane of the image, Qpl, for closed
discharge (Table 1).
220

a) b)

Figure 1. Contour plot of an image with theory overlay for images taken at N = 20 rpm.
a) silicone oil, Il = 4500 cP, b) CMC with powerlaw parameters: n= 0.91, m = 0.104 Pa sn.

TABLE 1
Comparison of theoretical and experimental flow rates

Screw Speed Theory Experiment Error

N QpI=VH Qpl' = area on image IQpI-Qpl'1
(RPM) (sq cms/s) (sq cms/s) Qpl
10 0.44 0.39 10.70%
20 0.88 0.77 11.80%
35 1.54 1.71 11.20%

DISCUSSION
The error in the experimental determination of the flow rate may have come from a variety of
sources. The most significant error is the poor resolution in the image. Presently, there is
approximately a 15% error in measuring the channel width. This error will be reduced to
about 7% by increasing the image resolution. Another source of error may be the inability to
maintain a constant screw speed over the time frame of the experiment.
In conclusion, MRI offers an improved method of studying flow in extruders.
Velocity profiles in an idealized extruder were obtained in order to verify the validity of the
technique as well as identify and quantify sources of error. Comparisons for the cases
studied were in good agreement. The imaging technique can be modified to image other
screw geometries as well as more viscous materials. Therefore, velocity profiles can be
directly studied and incorporated into mathematical models to simulate industrially relevant
extrusion systems.
REFERENCES

1. Mohr, W. D., Clapp, J. B. and Starr, F. C., Flow patterns in a non-Newtonian fluid in a single-screw extruder.
Soc. Plast . Eng.. TranS.. 1960, 1, pp. 113-120.
2. Eccher, S. and Valentinotti, A., Experimental determination of velocity profiles in an extruder screw. llli1.
Eng. Chem .. 1958,50 (5), pp. 829-836.
3. Meuser, F. and van Lengerich, B., System analytical model for the extrusion of starches. In Thennal
Processing and Quality of Food, ed. P. Zeuthen, Elsevier, London, 1984, p. 175.
4. Agemura, C.K., R.I. Kauten and K.L. McCarthy, Flow fields in straight and tapered screw extruders using
magnetic resonance imaging. Submitted.
5. McCarthy, K. L., Kauten, R. J. and Agemura, C. K., Application of NMR imaging to the study of velocity
profiles during extrusion processing. Trends in Food Science and Technology, 1992,3 (819), pp. 215-219.
 DETERMINATION AND CALCULATION OF FREEZING
EQUILIBRIA IN FROZEN FOODS

MARKUS LOTZ and HORST WEISSER

Institute for Brewery Installations and Food Packaging Technology,
Technical University of Munich, 85350 Freising-Weihenstephan, Fed. Rep. of Germany

ABSTRACT
The solid/liquid ratio in partly frozen foods is an important quantity in freezing processes.
The ratio is described in freezing curves. Freezing curves for food products at sub-freezing
temperature can be calculated numerically using several thermodynamical approaches. The base
for such a calculation is the knowledge of water activity coefficients which must be determined
in experiments using binary mixtures. The needed data can be rapidly calculated using NMR
spectroscopy. Some advantages of the NMR spectroscopy over thermoanalytical methods are
based on the facts that one sample of known concentration can be measured over a whole
temperature range, that liquid and solid food products can be examined, that the method
works destruction-free, and that one sample can be analysed several times.

THEORY OF FREEZING CURVES

Freezing curves describe the thermodynamic equilibrium between a pure ice phase and a concen-
trated residual solution at a temperature T. If a hydrous solution with one dissolved compound
is cooled below the freezing point, a pure ice phase and a concentrated residual solution will
form. This process continous until only ice crystals and crystals of the solute are present at the
so-called eutectic temperature.
Freezing curves for binaries can be calculated using thermodynamic quantities. With the
data of several binary mixture freezing curves it is possible to determine the freezing curves of
multy-compound mixtures. For those calculations the solubility curve must not be considered.
The activity coefficient IW of water forms the base for the calculation of freezing curves. The
activity coefficient of water describes the deviation of a multy-compound mixture from the ideal
behavior. The activity coefficient of water can be calculated with a suitable statement for the
excess free energy of mixture gE [1]:

n
gE = RT L Xi In Ii
8(n T gE))
or for IW : RT In IW = ( 8n w (1)
;=1 T,P,nj

Xi:= mole fraction of component i; nT:= total number of all moles

222
Several approaches exist to calculate /w. Some of them are empirical (for example Margules,
van Laar), some semi-empirical (for example Wilson, NRTL, UNIQUAC) [1]. The parameters
for these equations are derived from experimental data. Harz [2, 3] was the first to prove that
the formulations of Margules und Wilson were appropriate for the determination of freezing
equilibria. Wilson's formulation has the advantage of a consistent solution for multy-compound
mixtures. This formulation was therefore chosen for our experiments. For a binary mixture
(example: water/glucose) the activity coefficient of water can be written as [4]:

(2)

(3)

This equation uses two temperature-dependent parameters (AI2' A21 ), which must be de-
termined from experimental data. The molar volumes VI and V2 refer to pure liquids with an
activity of 1. A pure liquid phase does not exist for dissolved components (for example sac-
charides). It is not possible to specify their molar volumes. Instead, the ratio of their molar
volumes is added to the regression. Derived from coefficients determined of binaries the following
equation describes the activity coefficient for a multy-compound mixture:

In, w = -In (t
k=1
XkAwk) + 1- t
k=1
--:::'::-,k:.....A....;:w:..;,:.k_
2: XiAik
(4)

i=1

Interaction parameters of dissolved compounds are not directly accessible and therefore not
taken into consideration in this equation. A pure liquid phase of the dissolved compounds is
again assumed for the calculation. The mole fraction Xw of water can be expressed by:

aw(T)
Xw = ------~~--~ (5)
,w(T, Xl.' .. , xn)
The water activity a w is only a function of temperature T and can be described with an
appropriate accuracy by [5]:

1
-In aw = 27.622 - 528.373 T - 4.579 In T T := absolute temperature (6)

The freezing curve can be calculated stepwise:

1
cr(T) =1- ( ) M 9 := mean molar weight of the dissolved compound (7)
1+ /w_l Mg
aw Mw

RESULTS
Tested mixtures were model solutions of orange and apple juice. Examined binaries were mix-
tures of fructose, glucose, sucrose, and citric acid with water. The computation of the NMR
measurements was improved by adding the ratio of molar volumes to the method described by
 223
Harz. Figure 2 shows that the freezing curve calculated with Wilson's equation fits very well
to measured values (Young et al., NMR). Figure 2 also visualizes that NMR spectroscopy is
suitable to determine freezing curves. Good agreement is achieved when results from NMR
measurements are compared to the thermoanalytical method developed by Young et al. [6].
T (K) T (K)
275 275

270 270

265 265

260 260

255 255
o measured by Young et al.
250 ® measured values ~MR) 250 ® measured values ~MR)
- predicted values ( ilson) - predicted values ( ilson)
correlation r: 0,997 correlation r: 0,998
245 relative error 1,1 % 245 relative error: 0,8 %
240 240~~~~~~~~~~~~~~~
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Concentration Concentration
Figure 2. Freezing curve of the Figure 3. Freezing curve of a
binary fructose/water. model orange juice.

The freezing curve of a model orange juice is shown in figure 3. The freezing curve was
calculated with Wilson's equation. The model juice consisted of water, glucose, fructose, sucrose,
and citric acid.
In addition to the results presented here further experiments are performed with model
foods containing fat and proteins. Again it is the goal to test the applicability of Wilson's
equation to calculate freezing curves for this group of food products.

REFERENCES

1. Prausnitz, J. M., Thermodynamik der Phasengleichgewichte. vt-Hochschulkurs III: Ther-

mische Verfahrenstechnik: Phasengleichgewichte. supplement to Verfahrenstechnik, 1979,
13, No. 12.
2. Harz, H.-P., Untersuchungen zum Gefrierverhalten fiiissiger Lebensmittel im Hinblick auf
das Gefrierlagern, Gefriertrocknen und Gefrierkonzentrieren. Karlsruhe, Technical Uni-
versity, Department of Chemical Engineering, Dissertation, 1987.
3. Harz, H.-P. and Weisser, H., Berechnung von Fest/Fliissig-Gleichgewichten fiiissiger Le-
bensmittel mit Hilfe von Aktivitiitskoeffizienten. Chem.-Ing.-Tech., 1988, 60, No.4,
316-17.
4. Wilson, G. M., A new expression for the excess free energy of mixing. Journal American
Chemical Society, 1964,86, 127-30.
5. Lenci, C. R., Piva, M. and Dalla Rosa, M., Water activity and freezing point depression
of aqueous solutions and liquid food. Journal of Food Science, 1983,48, No.6, 1667-69.
6. Young, F. E., Jones, F. T. and Lewis, H. J., D-fructose-water phase diagram. Journal of
Physical and Colloid Chemistry, 1952, 56, No. 12, 1093-96.
THE CHARACTERISTICS OF THE HYDRATION WATER OF PROTEIN IN RELATION WITH
THE FREEZING AND DRYING DENATURATION
NAOFUMI HANAFUSA
Inst.of Low Temp.Sci., Hokkaido Univ., Sapporo, Japan. 060
ABSTRACT
The alterations of the amount and the extent of the interaction between
the hydration water and the protein molecule of some proteins during
freezing and drying with or without some cryoprotectants to clarify the
mechanisms of freezing and drying denaturation of protein, using lHNMR
and some other methods. The hydration structure of myosin was very
unstable than ovalbumin. The unstability of hydration structure of
protein may be main cause of freezing and drying denaturation.
Cryoprotectants decreased the amount of hydration water and markedly
intensified the interaction between the water and the protein. The
cryoprotectants protect the protein molecule by substituting with part of
hydration water and forming "quasi hydration layer".
THE TEXT
Freeze-thawing and freeze-drying of biological materials has been widely
used for several fields, though some proteins are denatured by the
procedures. But, the mechanism are not clarified. It was reported
rabbit muscle myosin was easy to denatured by freeze-thawing and freeze-
drying of myosin accerelated the denaturation(1-3). This investigation
were carried out to clarify the mechanism, the alteration of the amount
and the extent of the interaction of hydration water and protein using
lHNMR and some other methods. Myosin was used as an example of unstable
protein and ovalbumin was used stable one. The measurement of lHNMR were
carried out by JEOL FX100 NMR apparaous and the details of the mater ids
and methods were described in previous papers(1-3). The hydration water
of ovalbumin aq.soln. was frozen below O°C, because of the interaction
between the hydration water and protein. They gradually decreased to
-20°C. It was retaining similar values between -20 °C to -45°C. Below
-45°C, it decreased gradually again and reached to o. That was an
example of typical pattern of globular protein as reported Kuntz et
al(4). It may be considered the plateau region of water content was
competed and balanced of the binding forces with the protein and around
ice, and rather stable. By cooling of the solution, the freezing
proceeded from weak interacted water layer to strongly restricted
hydration layer in order of the intensity of the interaction. Contrary
to globular protein, hydration water of myosin, shown in Fig 1, decreased
rapidly and it had no plateau region of water content. The amount of
 225
hydration water in the first layer of globular, plateau region, agreed
with the amount of hydration obtained by other physical methods. In
contrary to the globular protein, it seems the hydration structure of
myosin was remarkably unstable and distracted easily by freezing. The
comparison of "c values of myosin and ovalbumin were shown in Fig 2.
Evidently, the values of "c of myosin were smaller than that of
ovalbumin. This indicated the hydration structure of myosin was weaker
and more easy to distructed than ovalbumin. The effect of some
cryoprotectants, such as sugars and amino acids, on the alterations were
examined as shown in Figs 3 and 4. It should be noticed the amount of
the water were extremely decreased than control and the extents of the
interaction were remarkably intensified about 10 or 100 times. In all
cases examined here, non effective additives showed opposed effects. The
amount of the hydration water was rather increased and the extent of the
interaction, " c, was decreased. Further, effect of some non-polar
surfactants were examined, because of it reported non-polar surfactants
were effective as cryoprotectant(5). The extent of the interaction could
be estimated by the value of self diffusion coefficient D, obtained by
the measurement of spin-locking, T1P ' instead of "c. The non-polar
surfactants were effective remarkably as sugars or amino acids. The
amount of hydration water decreased and the value of D decreased from
10- 9 cm/sec to 1/100. The polar surfactants showed opposed effects.
Similar experiments were carried out on the freeze-dried proteins. The
results were similar with freeze-thawing. The amount of residual water
content was extremely smaller than that of freeze-thawing. The extent of
the interaction was weak in myosin than ovalbumin. Cryoprotectants
decreased the residual water content and intensified the extent of the
interaction. In the previous report, the author reported those trends
and the minimum concentration of the protectants to protect protein
denaturation due to freezing and drying were agreed with the mole numbers
of the hydrophylic amino acids residues of the protein, and the mole
ratio was about 1 : 1(3). The residual water content was minimum wales
in this mole ratio and the extent increased gradually. By the results
mentioned above, the role of the hydration water and the stability of the
hydration strucure may play important role on the denaturation. The
hydrophylic cryoprotectants may substitute part of hydration water and
formatted "quasi-hydration layer". Thus, the layer may protect protein
against denaturation.

REFERENCES
1. Hanafusa,N., Denaturation of enzyme protein by freeze-thawing and
freeze-drying. Contr.Low Temp.Sci., 1972, B17, 1-38.
2. Hanafusa,N., Hydration characteristics of myosin. Japanese J. of
Freezing and drying. 1983, 29. 30-31. (in Japanese)
3. Hanafusa.N .• The behavior of hydration water of protein with the
protectants in the view of 1HNMR. Develop.biol.Stand .• 1991 74.
pp241-245.
4. Kuntz,J.D. and Kauzman W.• Hydration of proteins and polypeptides.
In.Adv.Protein Chem., 1974. 28. pp239-245. Acad. Press. N.Y. and
London.
5. Doi.E., The surface denaturation and food denaturation of food
protein. New Food Industry, 1986, 28. 80-87. (in Japanese)
226

Temperalure (- 'C)
o myosin SF o 20 40 60
• myosin RF
OJ myosin FT. denntured
e myosin denatured. (Heat)

Temperature eel

F ig.l The relat ion of the amount of unfrozen - 0 - myosin/O.5M KCI soln.
water and freezi ng temperature. Protei n
...•.. ovalbumin/ Aq. soln.
cone.0.S% in 0.5 M KCI. SF. slow freezing:
RF.rapid freezing; FT denatured by
f reeze-t haw i ng; fleet da natured: denatured by
heat i ng at 60'C 201l in.

3.5 4.0 4.5 5.0

Fig.2. The comparison of molecular

correlation time (,.1 of unfrozen water of
myosin and ovalbumin. Protein cone. 0.5%.

Temperalue (tl
1.5
10" r-~___--=2;:.O_~_-..::4;CO_~_..:-::c6~0,-_~

• 0.57% E A (cont)
1.0 'II
o + O.OIM sucrose
"
';;
(}) + O.OIM glycerol
~
.
25
e + O.OIM DMSO

i!.
!:J 0.5

t 10"
.~
o .§
~
i!
US
10- 10 •+
o
0.57%E A (coni)
O.OIM sucrose
Ell + o.olM glycerol
-0.5 L--'-_~~_~_~-::-:-_~~
o -20 -40 -60 -80 e + O.OIM DMSO
Temperalure ("C) () + O.OIM K·Na P. B.

Fig.S. The alteration of the amount of
unfrozen water of ovalbumin against freezing
temoerature with and without some
cryoorotectants. Prote i n cone. 0.57'/'
aQ.sol n.
10-"'7_ _ _ _ _~----...,._=__----~
3.5 4.5 5
F ig.4. Effects of some cryoorotectants on tha
molecular correlat ion time,. of the unfrozen
water of ova I bum in. The va I ues of , . were
obtained from relaxation time Tt
 NMR STUDY OF THB DYNAMIC BBHAVIOUR OF WATBR IN FOOD
SYSTBMS: GUAR GALACTOMANNAN SOLUTIONS AND PBCTIN GELS

E. BROSIO, A. D'UBALDO, B. VERZEGNASSI

Dep.t of Chemistry - University of Rome "La Sapienza"
P.le A. Moro, 5 - 00185 Roma (Italy)

ABSTRACT

The diffusion coefficient of water in non-gelled galactomannan

systems was found not to vary for a wide range of polymer
concentrations, despite of a strong increase in viscosity. In
the two-component gelling system, pectin/sucrose, a normalized
diffusion coefficient, Dgel/Dsuc, was found to be dependent on
the extent of gelation.

INTRODUCTION

NMR relaxation times of water in guar galactomannan solutions

and highly methylated (HDE) pectin solutions and gels were
found to change regularly with viscosity, either in gelled and
non-gelled systems (1,2). In the present study additional
insights into the description of water translational mobility
in polysaccharide systems are obtained by the Pulsed Field
Gradient Spin-Echo (PFG-SE) technique (3).

MATBRIALS AND MBTHODS

Samples
Guar galactomannan (Multi-Kern Corp., medium mesh) and pectin
(Fluka, DE > 70 %) were used without further purification.
Solutions and gels were prepared as previously reported (1,2).

NMR Measurements
All measurements were performed on a Bruker Minispec PC120
pulsed NMR spectrometer. The standard PFG-SE sequence (4) was
used in our experiments. 'Itle rragnetic field gradient arrplitude was
varied fran 0.6 T/m to 2.4 T/m, the duration of the gradient pulses was 1
ms, and the spacing between pulses varied fran 5 InS to 20 InS.
228
RESULTS AND DISCUSSION

The diffusion coefficients measured in the investigated systems

are reported in table 1.
TABLE 1
Diffusion coefficients measured in guar galactomannan solutions
and in 5 % pectin/sucrose solutions and gels

guar conc. D sucrose conc. D

(% w/w) (10- 9 m2 s- 1 ) (% w/w) (10- 9 m2 s- 1 )

2.5 2.34 0 1.85

3.0 2.34 10 1. 84
3.5 2.42 15 1.56
4.0 2.27 20 1.05
4.5 2.39 25 1.19
5.0 2.23 30 0.51
35 0.73
40 0.27
45 0.17
55 0.06

It can be seen that in guar solutions D values keep constant

with concentration, even though changes in rheological
properties and NMR relaxation times are observed (1).
Furthermore, no restricted diffusion was evidenced within the
used diffusion times. The upper limit explored for the
diffusion time (17.2 ms) yields a lower limit for the
heterogeneity domain size of about 9 ~m. These results show
that in non-gelling polysaccharide systems, despite of strong
thickening of the solution, the fluid medium keeps its motional
properties unaltered.

TABLE 2
Diffusion coefficients measured in pectin solutions and sucrose
solutions at various concentrations
pectin sucrose
concentration D concentration
(% w/w) (10- 9 m2 s- 1 ) (% w/w)

1 2.23 10 2.14
3 1.90 20 1. 28
5 1.85 30 0.99
7 1. 74 40 0.53
9 1. 74 55 0.22
11 1.56

A different behaviour was found in gelling pectin/sucrose

systems. As shown in table 1, a regular decrease of D values
with sucrose concentration is observed. Moreover, no
restriction to diffusion was evidenced in our experiments, and
a lower limit for the domain size was evaluated in about 1.5
 229
~m. In order to split the contribution from the two components
(pectin and sucrose), we investigated the self - diffusion
behaviour of pectin solutions and sucrose solutions. The
results are reported in table 2. It is evident that D values
for pectin solutions keep almost constant in the order of
magnitude for pure water, while in sucrose solutions a large
decrease in water diffusion coefficient is observed. That
suggests to define a normalized diffusion coefficient,
Dgel/Dsuc, which should describe the effective contribution due
to gel formation. The normalized diffusion coefficient is
plotted as a function of sucrose concentration in figure 1.

0.8
~

(.)
;;)
III
0.6
Q
"""""'-
<I>
Ol
0.4
Q
0.2

0
0 10 20 30 40 50 60
sucrose conc. (% w/w)

Figure 1. Normalized diffusion coefficient of water in

pectin/sucrose solutions and gels.

'!he observed linear decrease of Dgel/Dsuc is solely due to the process of

gel setting. In other words, the fornation of physical networks in
polysaccharide system results in a substantial reduction of water
translational rrcibility, even t:.hrugh no restricted diffusion, Le. no true
carp3.rt:nEntation, is evidenced.

REFERENCES

1. Brosio, E., D'Ubaloo, A., carnovale, E., Giusti, A.M., NYIR

investigation of the dynamic behaviour of guar galactarannan water
solutions. cell. Mol. Bioi., 1993, 39(2}, 199-204.

2. Brosio, E., Delfini, M., Di Nola, A., D'Ubaloo, A., Lintas, C., IH
and 23Na NYIR rel.aJ<ation tinEs study of pectin solutions and gels.
Cell. Mol. BioI., 1993, sulInitted.

3. StiThs, P., Farrier tr:ansfonn pulsed-gradient spin-echo studies of

nolecular diffusion. Prog. N1R SQect:rosc., 1987, 19, 1-45.

4. Tcmner, J. E., Stejskal, E. 0., Restricted self-diffusion of protons

in colloidal system by the p.llsed-gradient spin-echo mathod. Jl....
ChErn. Plws., 1968,49, 1768-1777.
 NMR-MEASUREMEMTS TO STUDY BINDING STATES AND MOBILITY OF
WATER IN CELLULOSE PACKAGING MATERIALS

FRANZ LIEBENSPACHER and HORST WEISSER

Institute for Brewery Installations and Food Packaging Technology.
Technical University of Munich, 85350 Freising-Weihenstephan, Federal Republic of Germany

ABSTRACT
Water plays an important role as transport medium, solvent, and reaction partner for many
processes in food packaging interactions. The degree of water availability does not only
depend on the water content but also on its mobility and type of sorption binding. Binding
states and mobility can easily be determined using different NMR-techniques. The relaxation
times Tl and T2 of water were measured for different cellulose materials which can serve as a
model for biopolymers at various water contents. Self-diffusion coefficients were determined
for different water contents and temperatures using low resolution pulsed NMR-spectroscopy.

INTRODUCTION
Cellulose based packaging materials like paper, cardboard, cellophane have a great importance
in food packaging technology. They are used to produce pouches, bags, sacks, envelopes,
sachets, wrappers and labels. The water content has a considerable effect on the production
and processing of these materials and on their use as packaging material. Since these materials
consist ofbiopolymers, they can also be used as a model system for many food products.

MATERIALS AND METHODS

Materials
Cellulose materials like kraft, filter paper and label paper were used. In contrast to plastics
these materials were made of macromolecules having natural polymers as raw material source.
NMR measurements
A Broker minispec pc 120 process analyzer equipped with a permanent magnet kept at 40°C
(proton resonance frequenzy Vo = 20 MHz, magnetic field strength Bo = 0.47 Tesla, air gap
between poles I = 25 mm, sample glass tube diameter d == 10 and 18 mm) was used for
IH NMR measurements. By using a temperature-controlled probehead and various pulse
sequences the influence of temperature on the spin-lattice relaxation time Tl and the spin-spin
relaxation time T2 was determined. Self diffusion coefficients Ds have been measured using a
recently available extra pulsed field-gradient unit together with a temperable special probehead
containing additional gradient coils [1, 2].
 231
RESULTS AND DISCUSSION
Spin-lattice relaxation time
From Figure 1 can be observed that the spin-lattice relaxation times TI of water sorbed on the
packaging materials has minimum values of TI at water contents w in the range of
wmin < 0.10-0.15 g H 20/g dm. Earlier similar results were observed for coffee extract [3]. The
water molecules are very tightly 'bound' and thus show a reduced mobility for very dry
materials. Consequently TI increases as w decreases below wmin . However, it should be
considered that this minimum of TI is not a real physical characteristic of the materials because
wmin depends on the magnetic field strength Do and thus on the resonance frequency Vo of the
NMR spectrometer in use.

---
4
+ Label paper
t Kroft pq>er
3.5
V - - ---
u
o fitter paper
3

.5
Vo///
2
+/ I
I

__1.-_._'-- ...--------- ------------

.5
/
/ 1/' -- I
*
- --..,. -;-- ;--- ---- -- ",~A

o.5 /:
11:
>JJ-'

40+-1-f--+--+--+-+--tf--t--+-+-+-I-+-+--+-+-H
5 7.5 10 12.5 15 17.5 20 22.5 25 27.5 10 15 20 25 30
Wafer confenf (9/100 9 dm) Wafer content (9/100 9 dm)

Figure 1. Influence of the water content w on the spin-lattice relaxation time TI and the spin-
spin relaxation time T2 of papers at 20°C.
Spin-spin relaxation time
The spin-spin relaxation time T2,1 of the cellulose protons with solid state behaviour is in the
range from 5 to 9 ~s. With low resolution NMR no influence of the water content could be found.
With increasing water content the spin-spin relaxation time of the protons in water
increases. Two fractions of different mobility could be found. The T22 of the main fraction is
shown in Figure 1. At very high water contents (w > 0.25 g H 20/g dm) a third T23 can be
detected being about 7-10 times smaller than the T22 at the same water content. This T2 3
corresponds to a very immobile, tightly bound water' phase of w :<;; 0.045-0.05 g H 20/g d~
equal to llw == 0.3-0.35. In some respects it behaves as part of the solids and does not take part
in transport processes and chemical reactions.
Figure 2 shows a semilogarithmic plot of the spin-spin relaxation rate ~2=1/T2 2 of the
slower decaying T2 ,2 as a function of the water content. In the range of w <' 14 g H20/g dm
C3w<0.85-0.9) ~2 decreases almost linearly with increasing w. The reduction of ~2 is
supposed to be d~e to an enlarging polymolecular layer of water molecules at the active
sorption sites or condensation in very small voids. If more molecules are attached to a sorption
site the interaction between the matrix and the water molecules is reduced. The energy of
binding decreases and the mobility is growing resulting in a reduction of ~ 2' For w ~
14 g H 20/g dm a new linear region of ~,2 was measured being a sign of a highe~ mobility of
the water protons due to condensation of the water within the larger capillaries of the cellulose
fibres. This assumption is supported by the observation that at very high water contents ~ 2
decreases only slightly. More and more water condenses within the capillaries. As long as the
capillary diameters are in the same order of magnitude the mobility is more or less the same.
232
5
LOOeI paper 540C 460(; 40 250(;
i
liI. ....
-
Kroft p,,*,
~

~
FRIer MM'r
0.5 ~
,.... J.
'j ~, ..." ~
E
~
o
5 I\~ *,\ 0
...~
~
" ........ _-
•..it.......
.
\' \.. ••• M. ••••••• I
~ 0.5 g O. .... .... "
~
.... .:1-.0 0 8- ____
.... ....
~ 0-- - ....
-Ao 0 0.05
.
-t .... ....
Kraft p~er ....
+w=25.0% a .. 1 EA= 17kJ/moI
'w=16,8% o. '" 0.91 EA = 39 kJ/moi
ow=13,5% o. '" 0,83 E. = 60 kJ/moi
o 0.0
5 10 15 20 25 30 3 3.1 3.2 3.3 3.4
Waler conlenl (g/100 g dm) 1/T (10.3 1/K)

Figure 2. Dependence of the spin-spin relaxation Figure 3. Arrhenius plot of Ds of kraft

rate ~ = lIT2>2 on the water content at 20°C. paper at different water contents.

Self-Diffusion Measurements
With the NMR method of pulsed field gradients [1, 2] the self-diffusion coefficient of water in
sulphate paper was determined in the temperature range 25°C < ~ < 54°C on three samples
with different water contents representing water activities of about 0.83, 0.91 and almost 1.
From Figure 3 it can be seen that the temperature dependency of D. can be described with an
Arrhenius equation. With the calculated activation energies listed in the figure it is possible to
determine D. for temperatures where it is nearly impossible to measure the self-diffusion
coefficients of water in paper since the ratio of the echo amplitudes with and without gradient
pulse is almost the same. If there are so little protons as in our samples, at very low diffusion
coefficients the noise is higher than the difference in the signal amplitudes.
The activation energy EA = 17 kJ/mol of the paper with llw "" 1 is in the same range as pure
water with E A = 19.4 kJ/mo!. That indicates the probability of interchange of sites being similar
to pure water. For the transverse movement of the water molecules of the sample with
llw "" 0.83 the activation energy is similar to the one of ice.
CONCLUSIONS
NMR measurement of the spin-spin relaxation time T2 of paper showed that water exists in at
least three fractions with different mobilities.
The binding state of the water molecules at llw ~ 1 is similar to free water. Their mobility
measured as D. is still smaller. The activation energy for self diffusion of water in paper below
llw ~ 0.85-0.9 is in the same order of magnitude as in ice.
REFERENCES
1. Liebenspacher, F., Untersuchungen uber Wasserdarnpfsorption und Wasserdampftransport in
Packstoffen auf Zellulosebasis, doctoral thesis. Technische Universitat Munchen, Fakultiit fUr
Brauwesen, Lebensrnitteltechnologie und Milchwissenschaft, Freising-Weihenstephan, 1991.
2. Weisser, H. and Liebenspacher, F., Application ofNMR spectroscopy in studying self-diffusion of
water in cellulose packaging materials. In Advances in Food Engineering, ed. R.P. Singh and M.A.
Wirakartakusumah, CRC Press, Boca Raton, 1992, pp. 33-47.
3. Weisser H., NMR-techniques in studying bound water in foods. In Food Process Engineering,
Vol. 1, ed. P. Linko, Applied Science, Barking, 1980, pp. 326-336.
 HEAT CAPACITY MODELING OF FROZEN FOOD GELS
FROM NMR DETERMINATION OF WATER STATES

P. CORNILLON*, J. ANDRIEU*, J-c. DUPLAN**, M. LAURENT***

* LAGEP - URA CNRS 1328 - UCB Lyon 1 Bat. 721 - 43, Bd. du 11 novembre 191869622
Villeurbanne Cedex FRANCE. ** Centre de RMN UCB Lyon 1 - 69622 Villeurbanne Cedex
FRANCE. *** CETHIL INSA Lyon - URA CNRS 1372 - Bat. 502 - 20, avo A. Einstein
69621 Villeurbanne Cedex FRANCE

ABSTRACT

Low resolution NMR was used to measure the different states of water - namely the ice,
freezable and bound water fractions - during the freezing of four types of food gels up to
-40cC. The bound water fraction for each gel type allowed to model the heat capacity of the
same gels. The knowledge of mass fractions of ice and other water fractions as a function of
temperature improves the precision of the heat capacity modeling as compared to a simple
additive model based on RAOULT's law for the determination of the various states of water.

INTRODUCTION

The thermophysical properties of foods in a frozen state are necessary to model the freezing
and thawing processes. It has been proved that these values are very sensitive to the different
states of water below oce, especially the ice fraction [1]. So, the improvement of the
modeling and the prediction of these properties depend on the accurate measurement of the
different water states by physical methods. Among them, we choosed the low resolution
NMR technique to determine the different water states as a function of temperature.

MATERIALS AND METHODS

Five kinds of gels were investigated, namely water+agar, water+sucrose+agar,

water+starch+agar, water+gelatin and water+ovalbumin. NMR measurements were made
with a BRUKER Minispec PC20. After applying a nl2 pulse to the gel spin system, we
measured the FID signal intensity after 70 Ils. This signal is proportional to the concentration
of "liquid" protons ; adopting a standard value of 100% for the signal at positive
temperatures, we deduced by a simple proportionality relationship the percentage of liquid
water at any negative temperature. The heat capacity was measured by differential
calorimetry for the same model food gels. Measurements were proceeded by decreasing
linearly the temperature from 20 cC to -40°C with a home-made differential calorimeter [2].

RESULTS AND DISCUSSION

NMR data
Assuming that the remaining unfrozen water at -40 cC represents the bound water, we
obtained the following mean values of bound water ratio to each solid matrix: 0.66 g!(g of
234

Dry Matter) (DM) for agar, 0.05 g/(g DM) for sucrose, 0.26 g/(g DM) for starch, 0.44 g/(g
DM) for gelatin and 0.31 g/(g DM) for ovalbumin. The knowledge of the FID variation as a
function of temperature gives the absolute values of the different water fractions as mentioned
on figure 1 for the 13.5% ovalbumin-water gel (mass fraction of ovalbumin). Generally, the
NMR values are in perfect agreement with the literature data obtained by calorimetry or
NMR. This physical technique is pretty interesting because it gives absolute values of Xice, Xb
or Xuw without any calibration. So, it is possible to plot the different water fractions as a
function of temperature as indicated on figure 1.

c:
0
.~
0,8

0,6
• • •
•
.. ~ • • Ice

£ C Dry matter
<J) 0,4
<J)

'"
:2 0,2
• Unfrozen water
c c Ii DC
... c c
•
C ~ C

°
.. ~

-50 -3 ° - 10 10 30

Temperature, T (0C)

Figure 1 : Water states as a function of temperature for the 13.5% ovalbumin-water gel

Differential calorimetry data

We measured the heat capacity of all precited gels between 20 0 e and -40o e. Generally, the
evolution is the same as indicated on figure 2.

8 r-----~------r--\r_. .------r-----_r----~
~

I 7t------+------t-----~------~----~----~
P
I 6 t------+------t--.r-~------~----~----~
~
~ 5+------+----<n~~--~------~----~----~
~
g 4 +------+,,~~~r=~~~~~~~~~~"rn

f
Il. 2
~
3 ~~~~1r1t.~..~~~~==~t=====Jt==~~
,-
•
~:""""::=-=--+-----j.------+l
0
10% (mass Iractlon 01 sucrose)
20% (mass fraction 01 sucrose)

i
J:
1 +----+----j.---+l 0
•

•
30% (mass Iractlon 01 sucrose)
40% (mass fraction 01 sucrose)
50% (mass Iractlon 01 sucrose)

O~--~----~--~====~==~==~
-40 -30 -20 -10 o 10 20
Temperature, T ("C)

Figure 2 : Heat capacity as a function of temperature of sucrose-water gels

Experimental results given by Differential Calorimetry show that: for positive temperatures,
the heat capacity increases with the total water fraction and for negative temperatures, the
heat capacity increases with the dry matter content i.e. when the total water fraction
decreases. This variation can be explained by the following: the more the dry matter fraction,
the less the unfrozen water or ice fraction.
Finally, the heat capacity was intepreted with an additive law as follows:
dXi
Cp app = Xs Cp s + Xi Cp i + xuw Cp uw - ~Hf dT ( 1)
 235
the different water fractions values being either determined by NMR or calculated by
application of the RAOUL T's law [31. These predicted values were compared to the
experimental ones' as a function of temperature as shown on the table 1.

TABLE 1
Experimental and calculated values of the heat capacity and the ice fraction
for the 13.5% ovalbumin-water gel.

T xice xice Cp (NMR) Cp Calculated Cp Jkg-1OC-l

(0C) (NMR) (calcul. ) Jk[;r I°C-l Jk~rloC-1 eXE.(decreasing T)
10 0 0 3770 3770 3754
0 0 0 3765 3765 3900
-3.3 0.755 0.714 20000 16000
-12.6 0.811 0.808 2990 2778 3226
-26.0 0.821 0.824 2070 2056 2211
-42.5 0.830 0.829 1915 1867 1900

One can observe that the two models are in pretty good agreement, the values obtained
with NMR fractions being closer to the experimental values. Generally, this tendency was
also observed for the other gels types [21. Nevertheless, the uncertainty increases for the gels
with high solid content; this fact can easily be interpreted by the increasing non ideality of
the solution as the solid concentration increases, the limit of nonideality being around 20% of
DM [2].

CONCLUSION

Generally, the introduction of experimental values of the different states of water improves
the heat capacity modeling. The improvement of the prediction of water fractions depends on
a more advanced knowledge of the water activity as a function of the temperature. This
method could be used efficiently for commercial products as ice cream, vegetables, paste
products.

ACKNOWLEDGEMENT

The authors are grateful to Societe L'AIR LIQUIDE and MRE for their financial support.

NOMENCLATURE: SUBSCRIPTS:
Cp : Heat capaci ty (J. kg- 1°C-l ) app : apparent s : solid
. . .
~H : Latent heat of ice fusion (J .kg- 1) b : bound water lor Ice: Ice
T : Temperature (K or DC) uw : unfrozen water
x : Mass fraction

REFERENCES

[1] - Renaud, T., Briery, P., Andrieu, J., Laurent, M., Thermal properties of model food
gels. L Food Eng., 1992, 15, 83-97.

[2] - Comillon, P., Mesure et modelisation des differentes fractions d'eau et des
proprietes thermophysiques des denrees alimentaires congelees et non congelees. These
de Doctorat, 1993, Universite C. Bernard Lyon 1, FRANCE (to be published).

[3] - Mannapperuma, J.D., Singh, R.P., A computer-aided method for the prediction of
properties and freezing/thawing times of foods. L Food Eng., 1989,9,275-304.
MEASUREMENT OF MOISTURE DIFFUSION IN A SOYBEAN SEED BY
PULSED·FIELD·GRADIENT NMR METHOD

Hisahiko Watanabe, Mika Fukuoka

Department of Food Science and Technology
Tokyo University of Fisheries
Konan 4, Minato, Tokyo 108 Japan

Shinji Shimada
National Agricultural Research Center, MAFF
Kannondai 3, Tsukuba 305 Japan

ABSTRACT

Although there is need for moisture diffusivity data in food materials, only a little diffusivity
data is available in literature. One of the main cause of this gap may be the difficulty in
methods for measuring diffusivity. PFG-NMR may be an alternative which enables to
measure effective diffusivity varying diffusion time, scale of observation. In this
presentation elements of PFG-NMR was explained and the change of moisture diffusivity
in soybean seed during drying (maturation) was reported.

INTRODUCTION

The rate of Mass-transfer (Moisture-transfer) in food during processing/storage is one

of the major concerns of food engineers/technologists. There is need for diffusivity data.
Unfortunately, however, only a small number of diffusivity data is available in literature.
This gap is seemingly due to two causes:complexity of food material and difficulty in
measuring intrinsic diffusivity. Food is a complicated material. It is not easy to identify
the system in which diffusion takes place. This makes it difficult to publish data. Another
cause may be the difficulty in measurement. The method used to measure diffusivity affects
the result. Measured diffusivity is no longer an intrinsic value as a physical property but an
apparent diffusivity which depends on the assumptions (mathematical model) used.
Existing methods for measuring diffusivity in food are by analyzing (a) the change of
concentration profile (optical method, NMR imaging method, mechanical slicing method),
 237
(b) the change of overall weight during sorption/drying (gravimetric method), (c) the
change of tracer signal intensity (tracer method). Pulsed-Field-Gradient NMR provides an
alternative.

ELEMENTS OF PFG-NMR

In the NMR study concerning biological system, proton is one of the major target
nuclei. Dealing with proton signal, you may follow the behavior of water molecule. A
standard setup of an NMR spectroscopy machine has two kinds of coils. A large coil and a
small coil. A large coil is for creating a strong static magnetic field which offers the basic
environment for NMR experiment. A small coil, which is rounded about the sample, is
used for transmitting an rf pulse to stimulate the spins (protons) in the sample and just after
switching the circuit from a transmitter to a receiver, it is used for receiving the response
signal from the protons.
When proton is placed in a static magnetic field, the proton turns itself about the axis
of the applied magnetic field; this is called precession. The frequency of precession (vo) is
proportional to the magnitude of the magnetic field (Ro). When the precession of spin
viewed from a rotating frame which rotates at a frequency vo, the spin looks as if it stands
still. In NMR experiment, the receiver can be regarded as rotating with a frequency vo, and
the situation described above occurs.
For diffusion measurements we use a special device, a gradient coil and its driver. A
gradient coil generates a magnetic field the magnitude of which is proportional to the
distance from the middle of the sample. Thence, the protons have precession frequencies
which are position dependent. When you observe the precessing protons from a frame
rotating with a frequency Vo which is the frequency at the middle of the sample, the protons
having positive X coordinate may look as if it makes a clockwise turn while protons having
negative X coordinate an anticlockwise tum.
Using a gradient coil, you can distinguish the position of the protons, namely the
position of water molecules. This position related information enables you to measure
molecular diffusion. The application of gradient pulse encodes the position of each protons
in the sample onto the phase of the precession of each proton, resulting in dephasing of the
spins, and the signal intensity decays. Application of another appropriate gradient pulse
with inverse magnitude may rephase the spins and possibly results in a complete recovery
of the signal intensity. Any diffusional migration of spins which may takes place during the
interval (~) of the two gradient pulses may cause a failure of recovering the signal.
Attenuation caused by diffusion (R) may be described as:
R = exp{ _y2g2&2(~_&/3)D}
238
The explanation made above is oversimplified: the effect of inhomogeneity of static
magnetic field and the effect of relaxation are neglected. The pulse sequences actually we
used for diffusion measurements are one for spin echoes and another for stimulated echoes.
The obtained echo signals were Fourier Transformed to distinguish water signal, oil signal
and others.

MOISTURE DIFFUSIVITY IN SOYBEAN SEED

The change of moisture diffusivity in soybean seed during soaking was measured by a
PFG-NMR and compared with that measured by a gravimetric method. Moisture
diffusivity measured by PFG-NMR was ten times larger than that by a gravimetric method.
The cause of the discrepancy may be due to the difference in scale of observation. In NMR
method moisture diffusivity during 20 ms was measured. During this short time period,
water molecule might migrate about 6!lm. This is comparable to the size of a cell. On the
other hand, moisture diffusion over 10 hours was observed by a gravimetric method. The
length of molecule migration might be 2.7 mm. This is comparable to the size of a seed.
Thence it is evident that the scale of observation is important.
One of the remarkable feature of PFG-NMR is that you can strictly define the
diffusion time (11) of observation. And when you vary diffusion time, you may see
something different. When water molecules are allowed to move randomly in a free space,
measured diffusivity may not be affected by the diffusion time during which diffusion is
observed. In contrast, when water molecules are confined in a small compartment, the
restricted feature of random movement may be detected by varying diffusion time.
Moisture diffusivity in cotyledon of soybean seed during maturation was measured by
PFG-NMR varying diffusion time (11). As a cotyledon of a soybean seed grows in a pod, it
reaches a maximum size of oval-shape, and then it proceeds maturation; changes its shape
from oval to sphere, reduces its size, and changes color and reduces moisture content. We
measured the change of diffusivity of a cotyledon which follows the course of maturation.
The logarithm of the signal attenuation measured (R) was plotted against diffusion time and
it was compared with a similar plot of calculation based on a simple model of restricted
diffusion confined between parallel leaves of plates. The result clearly showed the
restricted feature of moisture diffusion in soybean cotyledon, and the change of the size of
diffusing space as the progress of maturation was observed.
The attenuation of echo signal caused by the applied field gradient was analyzed
assuming that free diffusion is taking place. Thus obtained diffusivity, called apparent
diffusivity of soybean cotyledon became smaller as drying (maturation) proceeded. The
effect of moisture diffusivity was considerably affected by diffusion time. The longer the
diffusion time, the more diffusion depended on moisture content.
 MOISTURE DIFFUSION IN FRESH AND DEHYDRATED FISH FLESH
MEASURED BY PFG-NMR
Mika Fukuoka, Zhang Wei Wen, Tomoo Mihori, Hisahiko Watanabe
Department of Food Science and Technology
Tokyo University of Fisheries
Konan 4, Minato, Tokyo 108 Japan

ABSTRACT

Diffusion coefficient of moisture in codfish was measured using PFG-NMR. Intact as well
as processed (minced with and without NaCI, fresh and heated, freeze-dried) sample were
examined. The effect of diffusion time ranging from 28 ms to 100 ms was examined.
Intact flesh showed umestricted feature of moisture diffusion while minced flesh showed
restricted moisture diffusion. Dependence of moisture content on diffusion coefficient was
obtained.

INTRODUCTION

Although there is need for moisture diffusion data in food, published data is very much
limited. One of the main causes of this gap is the difficulty in measuring intrinsic diffusion
coefficient. PFG-NMR is an alternative. In this study, diffusion coefficient of moisture in
fish flesh, intact as well as processed ones, was measured using PFG-NMR.

MATERIALS AND METHODS

Codfish, purchased in the market, was used. The muscular tissue of codfish was
excised to 10 mm cube and used as an intact sample. Muscular tissue minced by a food
processor with and without NaCl was used as the minced sample. Freeze-driying was used
to prepare dehydrated fish flesh.
The NMR spectroscope used was a Broker AM200WB(4.7 Tesla) with a micro-
imaging accessary. Diffusion coefficient of moisture was measured at 25°C through a
240
PFG-NMR method using a stimulated echos. It has the application of three 90° pulses and
a pair of field gradient pulses of magnitude g, duration 6, and interval ~. Apparent or free
diffusion coefficient D may be calculated by

where y is the nuclear gyromagnetic ratio, R is the ratio of echo signal amplitudes in the
presence and absence of the gradient.

RESULTS AND DISCUSSION

Echo attenuation due to moisture diffusion in intact and minced codfish flesh was measured
with varied diffusion times ranging from 28ms to lOOms at constant 6g for several values
of g. The plot of echo attenuation (R) for moisture in intact flesh vs. diffusion time for
constant og made a straight line, indicating that unrestricted moisture diffusion was taking
place. On the other hand, the similar plot for moisture in minced flesh showed curvature at
large ~, indicating restricted diffusion. This result is quite unexpected one, because
molecular motion of moisture in muscular tissue is supposedly confined in narrow
compartments to show a restricted feature. And it is supposedly when the muscular tissue is
minced and wall of compartment is broken up, that water molecule obtains more space for
migration. One of the potential explanations for this anomaly may be as follows. There may
exist in fish flesh something like pools in which water molecule can freely migrate. The
signals come from these molecules are dominant in PFG-NMR. Mincing fish flesh may
break up the pool as well as disintegrate protein structure to expose hydrophilic amino acid
residue. The water molecules coming out of the pools may be caught by these side-chains
of protein and can not migrate freely anymore. Once the freely migrating molecules are
quenched, the NMR signals from moisture in cellular compartments becomes detectable.
Moisture diffusion coefficient in intact fish flesh, (Do), was 1.88 X 1O-5cm2s-1. Those
in minced flesh with and without NaCl ( 3% wt/wt ) were 0.79Do and 0.89Do,
respectively. Mincing fish flesh with NaCl facilitates myosin, actomyosine etc. to dissolve
into water forming fish meat sol. The increase of macroscopic viscosity due to network
formation decreased moisture diffusion, but not dramatically. Heating fish meat sol (with
NaCl) increased the jelly strength but did not affect the moisture diffusion.
Moisture diffusion coefficient in dehydrated fish flesh was found to be moisture
content (C) dependent as:

D=Doexp(kC), k=0.28

k value was independent of diffusion time, while Do depended on diffusion time.

 IH-NMR IMAGES AND LOCALIZED SPECTRA OF FOODS

NOBUAKI ISHIDA, HIROMI KANO· and HIDEJIRO OGAWA··

National Food Research Institute, Tsukuba, Ibaraki 30S, Japan, • National Institute of
Agrobiological Resources, Tsukuba, Ibaraki 30S, Japan, ··JEOL Datum Ud., Akishima,
Tokyo 196, Japan

ABSTRACT

Localized IH-nuclear magnetic resonance (NMR) spectra as well as IH-NMR images of

foods were measured by a JEOL GSX-270WB imaging system. Chemical shift of water
varied with tissues in a young green tomato fruit and the differences were reduced upon
maturation. Sugar signals in locular cavities were visualized after coloring of the fruit. IH-
NMR images showed different textures among sausages made of various materials and with
various preparation processes. Localized spectra provided information about relative
amounts and mobilities of water and oils in specified areas of the materials.

INTRODUCTION

IH-nuclear magnetic resonance (NMR) imaging is a useful method to estimate the quality of
fresh and prepared foods and to trace changes in their qualities over time or when affected
by environmental conditions [1-4]. IH-NMR imaging detects water, oils, and soluble
organic compounds to create images of the materials, which are integrated information of
physical states of proton nuclei, but not of the individual compounds present. In this investi-
gation, localized IH-NMR spectra at determined areas on IH-NMR images [S] were meas-
ured to detect soluble compounds in foods separately by the selective excitation method
using sinc-function-modulated pulses together with gradient magnetic fields.

MATERIALS AND METHODS

A spectrometer (JEOL GSX-270WB) with a superconducting magnet operating at 270 MHz

for IH was used. This imaging system has a S.O-em diameter detection coil and another coil
for producing a magnetic gradient (Max. 32 mT/m). A spin-echo imaging technique was
used for measurement of IH-NMR images (Fig. 1, a) with spatial resolution of 0.2 mm x 0.2
mm and a 2.0 mm slice thickness [1]. Localized IH-NMR spectra were also obtained by the
spin-echo 2D-FT method. The vertical position to be measured was determined by apply-
ing a Z-axis gradient magnetic field together with the 900 sinc-function-modulated pulse,
and the horizontal position by a X-axis gradient magnetic field together with the 1800 sinc-
function-modulated pulse, and phase encoding by a Y-axis (Fig. 1, b), which enabled the
measurement of a spectrum in a specified 1.0 mm x 0.2 mm area with a 2.0 mm slice thick-
242
ness. Relaxation time (T1) was measured by the progressive satumtion method using pulse
sequences with three different repetition times (0.5, 1.5, and 5.0 s).
90" 180" echo

RF

(a ~ ___ -JnL-___--' L
y
Z ~ Ur-----:...-____
~

( b)
X ---------',
y----~~=r------------
z ~~--~LJr----------
Fig. 1. Pulse sequences for the imaging (a) and the localized spectml measurement (b).
Radio frequencies were applied with sinc-function-modulated pulses (RF). Echo signals
were acquired with a X-axis gradient for imaging (a) and without an X-axis gradient for
localized spectral measurement (b).

, ,
~~
'1' "l' "l'" '~'~" "1" 'I" 'l"' ,~,;"
Fig. 2. IH-NMR images and spectral images at X-axis positions indicated by vertical
arrows of the young green (a,b) and mature red (c,d) tomato fruits, and localized spectra (e,t)
at Y-axis positions (indicated by horizontal arrows) of the mature red fruit.

RESULTS

Tomato fruits
Large amounts of water were located in the epidermis and around the seeds in a green fruit,
and the increased levels of water were found in a red fruit (Fig. 2, a, c) [1]. Water signals in
a collumela were not strong both in the green and red fruits. A spectral image of the green
fruit (Fig. 2, b) showed that the water signals shifted to the lower side of the magnetic field
in the pericarp, where water may be ordered. On the other hand, sugar signals were observed
beside the water signal, mainly in locular cavities in the red fruit (Fig. 2, d, e). Since sugars
were contained in the 80% ethanol extracts of green fruits as well as red fruits (data not
shown), sugars in the red fruit are assumed to acquire mobility by increasing water and
consuming other stored materials accompanied by physiological changes that accompany
maturation. The approximate amounts of sugars were estimated using water suppressed
spectra (Fig. 2, e, t).
Sausages
IH-NMR images and spectml images before and after boiling a raw sausage are shown in
Fig.3,a-d. Signals of IH-NMR images (a, c) were composed of water (4.8 ppm) and oils
(1.3 ppm) (b, d). Water signals were stronger than those of oils in the raw sausage, which
were broader after boiling except for the places where bulk water had penetrated during
boiling (arrows). Changes in the relative amounts of water and oils along with Y-axis posi-
 243
tions are shown in stacked profiles (Fig. 3, e,f). Further, relaxation times (T1) of water and
oils were obtained separately on the spectral images measured by pulse sequences with
various repetition times. Relaxation times of water varied with position (1.0 s - 1.3 s) while

,
those of oils were around 0.5 s at all positions [5].
,
L L

Fig. 3. IH-NMR images (a,c), spectral images (b,d), and stacked spectra (e,f) at X-axis
positions indicated by vertical arrows of before and after boiling the raw sausage.
Arrows beside spectral image and stacked spectra indicate the direction of Y-axis encoding.

DISCUSSION

This study was conducted to obtain further information about food materials by measuring
localized spectra using the imaging apparatus. The soluble sugars in increased amounts of
water in locular cavities, which may provide tomato fruits softness and sweetness, were
detected in the mature red fruit by localized spectra. The localized spectra are surmised to
show the contents of tissues which are important to decide qualities of fresh foods. The
relative amounts of water and oils, and mobility of water, important parameters which
determine the texture and taste of foods, changed before and after boiling a raw sausage
(Fig. 3), and they varied among sausages made of different materials and by different pro-
cesses (data not Shown), and by position within the sausage [5]. The fact that chemical
compounds soluble in liquid phases of foods and relaxation times of the individual com-
pounds were separately detected may provide useful information to aid in choosing the
proper materials for a given processing method and to trace quality changes of many kinds
of food.

REFERENCES

1. Ishida, N., Kobayashi, T., Koizumi. M. and Kano, H., IH-NMR imaging of tomato fruits.
Agric. BioI. Chern., 1989,53,2363-2367.

2. Ishida, N" Kobayashi, T., Kano, H., Nagai, S. and Ogawa H., 23Na-NMR imaging of
foods. Agric. BioI. Chern., 1991,55,2195-2200.

3. McCarthy, M. J. and Kauten, R. J., Magnetic resonance imaging applications in food

research. Tr. Food Sci. Technol., 1990,1,134-139.

4. Pope, J. M. and Sarafis, V., NMR microscopy. Chern. Aust., 1990, 221-224.

5. Ishida, N., Kano, H. and Ogawa, H., IH-NMR image and the localized spectra of foods.
Bunseki Kagaku, 1992, 41, 561-566.
 INVESTIGATION OF JET AGGLOMERATION PROCESSES

HARALD SCHUCHMANN, HELMAR SCHUBERT

Institute of Food Process Engineering, Karlsruhe University, Kaiserstr. 12,
0-76128 Karlsruhe, Germany

ABSTRACT
Jet agglomeration has been used in the food industry for several years to produce ag-
glomerates with good instant properies from fine powders. The process is suitable for all
foodstuffs that form sticky surfaces when wetted. The short average residence time and
narrow residence time distribution allow the processing of materials with a minimum thermal
degradation. A short overview of the basic principles and mechanisms underlying the process
based on the studies of a pilot plant are given.

INVESTIGA TIONS
In jet agglomeration plants, fme powders are agglomerated to obtain good instant properties
[1,2]. The freely falling raw material is wetted in a spray cone by droplets or in a steam jet by
condensation at the particle surface. Due to these two wetting mechanisms a liquid layer
develops on the surface of the particles. In a subsequent region of high particle volume
concentration, collisions between particles can occur. Agglomerates form, if the forces of
adhesion are strong enough to sufficiently dissipate the kinetic energy of the particles. The
primary particles stick together, joined by viscous liquid bridges with high content of soluble
substances. While drying the crystallization of these substances leads to the formation of
solid bridges.
Due to the condensation of steam on the particle surface, there is a mass flow towards the
particle, which increases with decreasing initial particle temperature. A numerical simulation
of heat and mass transfer from the surrounding steam jet towards the particles on a computer
helps to design new agglomeration plants [3]. An example for changes in particle temperature
and moisture for given operating conditions is shown in Fig. 1.
Primary particles of 90 11m average size, mostly in the form of dry agglomerates of about
900 11m size, were fed into the plant. Air flow through the feed duct was adjusted to allow
 245
the survival of most of these dry agglomerates. Wet agglomeration inside the plant then led to
the production of agglomerates of 1800 Jl1ll average size. The influence of particle size can be
judged by comparing the curves representing the primary particles with the curves of either
dry or wet agglomerates. Note that the moisture content never exceeded 3% wJw.
These calculations have been verified by experiments, where the uptake of surface water was
measured by Karl-Fischer-titration for different foodstuffs (mixtures of sugar, cocoa and
soluble coffee) at different locations between particle feeder and product outlet [2].

II III
80t--1====~,~--~I-i:!~ru~nn~~~;~~~o~c~,,~=~0~.~~-i!~t=.m~p~.~~~ru~~~-----t10
, II :! wetting area' 80 °c , .. 0.9$
•••••• ~. 1
III drying area; eooc, ... 0.10 1 .---==n-=r.:----.
f ',-..-. :i i
:! ~--------~ 8

.
60 ....... ..... "1"..........,
, ,'.:
••••
oo
I ',1. ...........
1a
.
~

I!
I
I
' :
~ .........~ ....... 6
1:
: ' ... : "38'"'' .!
il 40
... 1___ ............._._._._._ ....... _l .. _. ___ .~........ _... _...... _.. __ ... _....... __ ... :..... __ .............. __ .. __ .
I :... :
":~.~_"'!~!'....... ..
5
i ,
',.
!......!
l: ........ - .~-"-lr---
:
________ 4 S
...u
! ,• ' . . . . . ! ! ;
1---/-......... ·····'·~·,·~·t·:-·..·;~~~~·:~·::·~·~=·~=-t~~·~~·~~~~~~~=.......................
,- ' - ·..... 1 1

20 ........ 2
water content "....... 2b! ••••• _-••••
i
-1 .. ------· - - - - ...
_part - - - __ ..

o ... 0
o 500 1000 length I mm 1 500

Figure 1. Calculated temperature and water content of primary particles, dry agglomerates
and wet agglomerates during the agglomeration process for a feeding temperature of 0 dc.

Ground sugar particles and to a certain degree native sugar crystals showed reduced
flowability for relatively small particle diameter. The reason for this was dry agglomeration
caused by electrostatic and van der Waals forces. Since the surfaces of sugar particles are not
curved but consist of numerous plain faces, the contact areas between sugar particles (or any
other crystals) can be quite large. Therefore, the attractive forces can outweigh the separating
forces up to a maximum particle diameter of 200 11m. For example, sugar crystals of 120 Jl1ll
average diameter formed dry agglomerates of about 800 Jl1ll average diameter (Fig. 2).
246
1.0

0.8
~~~~:;:m_j-mmmmm __mmmmm !-_mJ.J ~mm
0.6 -:~~:=.::tl::j-----l
: : , agglomeration
~ ::-
0.4 ......................................................···11....... ... ............................... ······ .. ·r··························
1 . agglomeration after wet .. .

before w e t :
agglomeration ij particle
after vibrating
feeder
0.2 ,
.............................. ........................................... -............. .

0.0
10' 1 02 103 10·
particle diameter I J.l.m
Figure 2. Changes in the particle size distribution during the agglomeration process,
measured after a vibrating feeder, below a sieve, after passing a funnel and below the wet
agglomeration area
These brittle agglomerates could be substantially reduced in size by placing a vibrating sieve
between dispenser and feed inlet. Depending upon the velocity of the air flowing through the
feed inlet the dry agglomerates could be further reduced in size before entering the wet
agglomeration zone. In the wet agglomeration zone, the average size increased again, mainly
through the reduction of the number of rmes [4].

ACKNOWLEDGEMENT
The authors gratefully acknowledge the financial support of this research project by the
Deutsche Forschungsgemeinschaft.
REFERENCES

1. Schuchmann, H.; Hogekamp, S.; Schubert, H.: Jet agglomeration process for instant
foods. Trends in Food Science & Technolo/:y, June 1993.
2. Schuchmann, H., PhD thesis, University of Karlsruhe, Germany, 1992.
3. Szembek, M.; Schuchmann, H.; Schubert, H: EinfluB des Warme- und Stoff-
transportes auf die Agglomerateigenschaften bei der Strahlagglomeration. Wis-
senschaftliche AbschluBberichte, 27. Intemationales Seminar, Juli 1992.
4. Schuchmann, H.; Schubert, H.: In-Line Particle Size Determination for Jet Ag-
glomeration Processes. Preprints 5. Europrusches Symposium PartikelmeBtechnik,
Niimberg, Germany, 24. - 26. Marz 1992, S. 423 - 432.
 The Effect of Interparticle Forces on the Separation
of Fine Powders from Gas-Solid Two Phase Flow

H.O. Kono and T. Hikosaka

Chemical Engineering Department
West Virginia University
Morgantown, WV 26506-6101

ABSTRACT

The effect of fine powder's interparticle forces on the powder classification efficiency in
Turbo-type classifiers was experimentally investigated. The fine calcium carbonate
powders (2- 100 micron) were used as sample powders, which have different particle
interaction force in accordance with its particle properties.
The property of the above bulk powders was characterized by the method proposed by
Kono et. al. recently [1, 2, 3 and 4].

INTRODUCTION

Recently Kono et. al. [1, 2, 3 and 4] developed a fine powder characterization method
assuming that the stabilized aerated powders of the "gas-powder colloid structure" can be
simulated for the bulk powders in a dynamic condition. Rheological properties of fine
powders were defined and measured and fundamental dimensionless structural numbers
were introduced. This powder characterization method is further used to characterize the
powder in the intensive dynamic condition such as in the turbo-classifier. The experimental
results showed that the particle interaction is important even under these intensive flow
condition, so far as fine powders smaller than 5 micron are concerned. In the past classical
research of powder classification by turbo-classifiers, the two major forces were considered;
the drag force acting on the particle (Fd) and the centrifugal force (Fc) were taken into
consideration. However, in this paper, we propose the particle interaction forces should
be taken into account for the powder classification, when dealing with very fine powders.
248
EXPERIMENTAL AND RESULTS

The calcium carbonate powders of Pfizer Inc. with the size range 2 - 100 micron, were
used as the starting sample powders and nine different sample powders were used. Namely
three different sizes of sample powders of 4, 5, 8 and 12 Jl. in mean diameters were used.
Based on the correlation among the rheological parameters by Kono et. aI. [4], the
dimensionless numbers of the rheological structure of the aerated powder were defined and
calculated.
The classifier used was a Turbo-classifier, Model TC-15N of Nisshin Engineering Co.,
Ltd. The flow sheet is illustrated in Figure 1.

_.::0

Figure 1. Flow Diagram of the Classification System

As is clearly seen from Figure 2, interesting characteristic curves of the partial

separation efficiency were found to be very significantly different from each other, when
the fine powder sample used in runs 1-3 was classified in the classifier. On the other hand
there are no much differences for the powder used in runs 7-9. It tells us that the effect
of the particle interaction force on the classifier becomes obvious in this classifier, when
the fraction of calcium carbonate powder under 10 micron becomes large. This. critical
powder particle size will be a function of the surface characteristics of the sample powders
and also be affected by the force prevailing in the classifier which will be determined by
the selection of the operating conditions of the classifier.

CONCLUSION

It was found through the experiment that the classification of fine powders such as calcium
carbonate smaller than 10 micron is very much affected by the interparticle force prevailing
among the fine powders even at the high gas turbulency condition.
Therefore, a new model for the powder classification was proposed for the fine powder
classification in a turbo-classifier.
 249
100
I I I I I I II '~/r1 ~
-
..•. m
~

•
-;-
80
~ C.acl_ C...... I.
.,. 1.1,1 ,F
iJI
=~ 60
MI_ ~r IT ..... ~ ~
IJI
.!• I/!j
~. L/II
I
~.
.. W I...p.... lit:
~C' r-
-4 40
• ~~ [! j
""! .I .

:.
~
20
~~
I Q....(r" I.D

0 lft.
~ ...
I

I""
f"o"
~
D
0.1 0.3 Cl5 I 1 10 30 50 100

'arlld. Dlamater Dp f .... )

Key Run Rotor Air Flow Sample Powder Key Run Rotor Air Flow Sample Powder
# Speed Rate ur(pa) dp(Jl) # Speed Rate ur(pa) dp(Jl)
(rpm) (m3/min) (rpm) (m3/m in)

0 1 6700 2.8 <D

•
1500 4.5 2 5700 2.0 1500 4.5
3 4800 1.4 1500 4.5
A- 5 3200 2.0 600 8.0 '"
A-
4 3750
6 2700
2.8
1.4
600
600
8.0
8.0
0 7 2600 2.8 250 rn
•
12.0 8 2200 2.0 250 12.0
9 1850 1.4 250 12.0

Figure 2. Separation Efficiency of Various Runs

REFERENCES

1. Kono, H. 0., T. Ells, S. Chiba, and M. Suzuki, "Quantitative Criteria for Emulsion
Phase Characteristics and for its Transition between Particulate and Bubbling
Fluidization," Powder Tech., 52, 69-76 (1987).

2. Kono, H. 0., et al., "Segregation and Agglomeration of Type C Powders from

Homogeneously Aerated Type A-C Powder Mixtures During Fluidization," Powder
Tech., 53, 163-168 (1987).

3. Kono, H. 0., et al., "Characterization of Fine Bulk Powders by Rheological Properties

of the "Gas-Powder Structure" at Aerated Conditions," AICHE Symposium Series,
(1988).

4. Kono, H. 0., "Fluidization of Fine Powders," chapter in Transport Processes of

Fluidized Particles, Elsevier Publishing Co., Amsterdam, Netherland (1988).
 AN APPLICATION OF CRYO-SHATTERING FOR SEPARATING
LOW-FAT MEAT FROM FATTY MEAT

YOSHIO HAGURA, KANICHI SUZUKI and KIYOSHI KUBOTA

Department of Food Science, Faculty of Applied Biological Science,
Hiroshima University,
1 - 4 - 4, Kagamiyama, Higashi-Hiroshima, Hiroshima 724, Japan

ABSTRACT

A new method for separating low-fat meat from fatty meat was developed by using physical
characteristics of frozen meat. This method applied a cryo-shattering method followed by sieving
shattered frozen meat particles (SFMP) at low temperature and/or by collecting low-fat meat
particles from SFMP at low temperature. Pork containing 45.8% fat was shattered at selected
low temperatures (-20 to -196'C). At -4O'C, a remarkable reduction of the fat content in smaller
particles was observed. Separating smaller particles obtained low-fat meat from fatty meat by use
of sieving method. The recovery of the separated meat contained 19.8 and 32.8% fat was 21.6
and 29.0%, respectively. Moreover, we could distinguish easily the muscle tissue particles and the
adipose tissue particles with the unaided eye from the meat particles shattered at -3O'C. The muscle
tissue particles were manually collected from SFMP. The fat content in the separated muscle tissue
was 5.1 % and the recovery of the muscle tissue was 90.7%.

INTRODUCTION

Reductions in fat intakes can result in reduction of the risk for heart disease [1]. In fact, the
American Heart Association and other health organizations have promoted lowering intake of
dietary fat and cholesterol as means of preventing cardiovascular heart disease [2].
One of the largest problems facing food companies in manufacturing meat products is
controlling the fat-to-Iean content, because there is individual difference in processing raw meat,
such as difference in fat content and difference in geometrical fat distribution. For maintaining a
consistent fatllean ratio in meat products, it is best to separate lean and fat from fatty meat and then
to mix lean with an adequate amount of fat.
In our previous papers [3,4], low-fat flesh separation from fatty fish (e.g., mackerel and
sardine) by cryo-shattering was reported. The purpose of present study is to develop a new
method for separating low-fat meat from fatty meat by using physical characteristics of frozen
meat in order to control fat content in meat products. This study intends to pave the way for more
utilization offatty meat (e.g., offal and belly) for healthy food.

MATERIALS AND METHODS

 251
Materials
Pork consisting of the muscle tissue (lean) and the adipose tissue (fat) was used as a sample. Fat
content and yield for each tissue was listed in TABLE 1.

TABLE 1. Fat content and yield of samples

Materials Portion Yield (%) Fat content (%; on wet basis)

Raw sample Pork 100.0 45.8
Muscle tissue 48.6 3.4
Adipose tissue 51.4 85.8
Collected Separated muscle tissue 44.1 5.1
sample Separated adipose tissue 38.6 75.2
Others 17.3

Procedure
A hammer mill equipped with 8 hammers and a screen with 5.Omm openings was operated at
2,300 rpm. About 1 kg of the sample was cut into about 30mm x 30mm x 30mm. They were
frozen at a preset temperature between -20'C and -196'C and were fed into the mill, which had
been cooled in advance with liquid nitrogen. The discharge from the mill was loaded on a cooled
stack of the standard test sieves (R20/3 series, JIS Z-8801) with apertures of 2.00, 1.41, 1.00,
0.71, and 0.50mm and a receiver. The stack was mechanically shaken in a chamber kept at -20'C
until all particles fell into the sieves through which they were unable to pass. The collected six
fractions were weighed and stored at -40'C until the fat content was determined. The crude fat
content in samples was measured by use of a Soxhlet extractor with diethyl ether.

RESULTS AND DISCUSSION

Separation of the low-fat meat by sieving method

Pork was shattered at temperatures ranging from -20'C to -196'C and separated into six fractions
by cryo-sieving. The fat content in each fraction was determined. The relation between fat content
and particle size showed various dependence upon cryo-shattering temperatures. At -40'C, a
remarkable reduction of the fat content in smaller particles was observed. This result suggests
that the shattering of pork at -40'C followed by the collecting of smaller particles through cryo-
sieving favorably separates the low-fat meat.
Separating smaller particles obtained low-fat meat from SFMP by use of sieving method. The
fat content of each fraction was plotted against the weight percentage of separated meat (Fig.1). A
dotted line represents an ideal separation of muscle tissue from adipose tissue, and a chain line
represents complete mixing of them. The fat content in the collected meat particles depended upon
their recovery. The recovery of the collected meat containing 19.8% fat (make use of fraction 1
and 2) and 32.8% fat (fraction 1,2 and 3) was 21.6 % and 29.0%, respectively.

Separation of the low-fat meat by collecting method

We could distinguish easily the muscle tissue (lean) particles and the adipose tissue (fat) particles
with the unaided eye from the meat particles shattered at around -3O'C. The muscle tissue particles
and the adipose tissue particles were manually separated from SFMP with a pair of tweezers
(Fig.2). Fat content and yield for each separated tissue was listed in TABLE 1 and shown in
Fig.1. The fat content in the separated muscle tissue was 5.1 % on wet basis and the recovery of
the muscle tissue was 90.7%. These results suggest that applying mechanical separation (e.g.,
252
gravity classification) to the cryo-shattering method enables us to separate industrially the low-fat
meat from fatty meat.

Perticle Size (Dp): 2.83~Dp< 4.00mm

C
<U
§ 20
u

20 40 60 80 l()
Weight Percent or epa rated Meat (%) Perticle Size (Dp) : 4 .00mm~ Dp

FigJ Effect of yield on fat content Fig.2 Photographs of separated adipose tissue (A)
in separated meat. and muscle tissue (B) particles.

CONCLUSIONS

A new method for separating low-fat meat from fatty meat was developed by applying a cryo-
shattering method followed by separating smaller particles from SFMP by cryo-sieving (sieving
method) and/or by collecting the muscle tissue particles from SFMP by hand (collecting method).
The recovery of low-fat meat and its quality by the collecting method were far better than those
by the sieving method. These results suggest that a proper separation method enable us to obtain a
good quality low-fat meat.

ACKNOWLEDGMENT

The authors are indebted to Mr. S. Nakai, Technical Official, Hiroshima University, for his help in
the manufacture of hammer mill. This work was supported in part by a research grant from the Ito
Foundation.

REFERENCES

1. Giese, 1., Developing low-fat meat products. Food Technology, 1992, April, tOO-108.
2. AHA, Dietary guidelines for healthy adult Americans. Am. Heart Assn. Circulation, 1986,
74,1465.
3. Hagura, Y, Watanabe, H., Ishikawa, M. and Sakai, Y, An application of cryo-shattering to
low-fat meat separation from whole fish of mackerel and sardine. Nippon Suisan Gakkaishi,
1989, 55, 2119-2122.
4. Hagura, Y and Watanabe, H., Factors affecting separation of low fat flesh from fatty fish by
cryo-shattering. J. Food Sci., 1991,56, 1567-1571.
 UTILIZATION OF BRITTLE FRACTURE PROPERTIES
OF FROZEN FOODSTUFFS

KIYOSHI OKAMOTO, YOSHIO HAGURA, KANICHI SUZUKI and KIYOSHI KUBOTA

Department of Food Science, Faculty of Applied Biological Science
Hiroshima University
1-4-4, Kagamiyama, Higashi-Hiroshima, 724, Japan

ABSTRACT

We have proposed a new cutting method for frozen foodstuffs named "cryo-cutting". This
method applied a bending force to frozen foodstuffs at appropriate low temperatures to cause
brittle fracture. The effect of temperature on fracture stress and smoothness of cut surface of
frozen tuna flesh was studied in the temperature range from -50'C to -15OC. Three-point
bending test for tuna flesh showed that the brittle fracture temperature of the connective tissue
and the muscle fiber were -100'C and -llO'C, respectively. Tuna flesh was cut smoothly at
temperature that muscle fiber became brittle (-11 O'C).

INTRODUCTION

In the size reduction operation (cutting) of frozen foodstuffs (e.g., frozen meat and fishes ),
band saw is usually used. During the band saw cutting process, large amount of sawdust is
produced from foodstuffs, and the most of them are wasted. Moreover, there is a fear of
getting injured in the cutting process. Therefore, it is requested to develop a cutting method of
frozen foodstuffs without band saw.
Hagura and Watanabe [lJ reported that frozen fish flesh showed marked decrease in fracture
stress and brittle fracture occur at appropriate low temperature. This result implied a new
cutting method of frozen foodstuffs without band saw by bending the foodstuff at brittle
fracture temperature. We expected that the method improves yield and quality compared with
the conventional method, because of no sawdust and low temperature processing, without
thawing.
In this paper, we studied the brittle fracture temperature of tuna flesh and smoothness of the
254
cut surface, and verified the usefulness of the cryo-cutting method.

MATERIALS AND METHODS

Lean flesh of frozen yellow fin tuna (Thunnus albacares ) was used as a sample. The muscle
fiber orientations of test pieces (15xI5x6Omm3 ) were 0 degree (parallel) and 90 degree
(perpendicular to the direction of longitudinal axis).
To determine the brittle fracture temperature, we measured fracture stress of frozen tuna
flesh using the three-point bending machine in a temperature-controlled cabinet. In this cabinet,
test pieces were cooled to a preset temperature from -5O'C to -150'C. A stress-strain curve of
bending test was recorded using a load cell and a strain-meter.
We measured the smoothness of test pieces at each temperature, to verify the usefulness of
cryo-cutting method. The degree of smoothness was expressed as the ratio of the number of
test pieces perpendicularly cut to the longitudinal axis, to all the number of test pieces at each
temperature.

RESULTS AND DISCUSSION

The brittle fracture temperature

Fracture stress of 0 degree and 90 degree test pieces are plotted against temperature in Fig. 1(a).
We defined the brittle fracture temperature as the temperature at which fracture stress dropped
markedly. For 0 degree test pieces, fracture stress increased as decreasing temperature to
-100'C, followed by an abrupt drop at just below -100'C. Therefore, we concluded that the
brittle fracture temperature of 0 degree test pieces was -11 O'C, and -100'C for 90 degree test
pieces.
Since fracture stress of 0 degree test pieces and 90 degree test pieces related to the strength
of muscle fiber and connective tissue, respectively [2], we considered that the brittle fracture
temperatures of the connective tissue and the muscle fiber were -100'C and -110'C,
respectively.

Smoothness of cut surface

The degree of smoothness of cut surface of 0 degree and 90 degree test pieces are plotted
against temperature in Figure 1(b). Both test pieces were cut smoothly at -110'C. This result
suggested that the optimum temperature for cryo-cutting of tuna flesh was -I1O'C and the
temperature depended mainly on the brittle fracture property of the muscle fiber.
 255
~

~ 15 (a) ~ 100 (b)

~ Io....j

~dew7
reces
~10
rI.l
rI.l test
~
rI.l
..d
~
CIJ 0
50
e
0
~
~
rI.l
o degree
CIJ 5 rI.l
test pieces
~
S ~
0
~ CIJ
~ 90 degree test pieces CIJ
~ ~
~ 0- 150
I:f)
CIJ
0
-100 -50 Q -150 -100
Temperature[OC] Temperature[OC]
Fig.l Changes in fracture stress (a) and degree of smoothness of cut surface (b)
of tuna flesh at low temperatures.

CONCLUSION

To verify the usefulness of "cryo-cutting" method, we tested the brittle fracture property of the
frozen tuna flesh using three-point bending test. The brittle fracture temperatures of the
connective tissue and the muscle fiber were -100ce and -1lO'C, respectively. Tuna flesh was
cut smoothly at -11 oce, at that temperature muscle fiber became brittle. We concluded that, the
optimum temperature for cryo-cutting of frozen tuna flesh was -11 oce.

ACKNOWLEDGEMENTS

The authors would like to thank Mr. S. Nakai, Technical official, Hiroshima University for his
help in the manufacturing of the bend testing machine.

REFERENCES

1. Hagura, Y and Watanabe, H., Factors Affecting Separation of Low Fat Flesh from fatty
Fish by cryo-shattering. 1. Food Sci., 1991, 56, 1567-71.
2. Munro, P.A., The Tensile Properties of Frozen and Thawed Lean Beef. Meat Sci., 1983,
8, 43-61.

Address inquiries to Dr. Y. Hagura.

 MECHANICAL DISINTEGRATION OF MICROORGANISMS BY WET MILLING
AND HIGH-PRESSURE HOMOGENIZATION - A COMPARATIVE STUDY

* MARTIN PIITROFF AND HELMAR SCHUBERT *

Institute of Food Process Engineering, Karlsruhe University,
Kaiserstr. 12, D-76128 Karlsruhe, Germany

ABSTRACT
In enzyme technology the release of biologically active proteins from microbial cells by,
e.g., disintegration in high-pressure homogenizers or agitated ball mills is a fundamental
step of the total recovery process. In the present report these two types of equipment are
compared for different sets of parameters. Experimental results concerning the disintegra-
tion characteristics of yeast and several bacteria are presented and discussed with regard to
the degree of disintegration, enzymatic activity, particle size distribution and the specific
energy consumption. The comparison has shown significant differences: in rods, high-
pressure homogenization is superior to wet milling; in cocci, wet milling tends to yield
better results than high-pressure homogenization (especially for gram-positive organisms);
virtually complete disintegration of yeast cells is achieved in both types of equipment.

MATERIALS AND METHODS

Saccharomyces cerevisiae (yeast) was purchased from Deutsche Hefewerke, Nuremberg/D.

Cell cultures of Escherichia coli N 100 ,Acinetobacter calcoaceticus (ATCC 17903),
Micrococcus luteus (ATCC 4698) and Bacillus megaterium were provided by E. Merck,
Darmstadt/D. The fermentations were conducted in fed-batch cultures (complex medium)
under empirically optimized conditions until the stationary growth phase was reached. Cell
suspensions of 2.5 % dw were prepared from yeast and bacteria, respectively, using phos-
phate buffer (50 mM, pH 7.0).
Milling experiments were performed in batch mode using a commercial mill (grinding
chamber of 41) and glass beads of 0.75 mm in diameter at a ball filling ratio of 85 % (v/v).
The agitator shaft was equipped with a torque measuring device and a revolution counter to
determine mechanical energy consumption during comminution (angular velocity w, milling
time t). Disruption in a high-pressure homogenizer was accomplished in single and multiple
passages at a constant flow rate of 50 lIh (operating pressure PH, i.e., the volume related
energy input By per passage). A special two stage knife-edge construction [1] acted as a dis-
charge valve, the second stage remaining set to 50 bar during all experiments.
 257
The degree of disruption (A) was determined by biuret-method for total protein in the
supernatant of disintegrated samples [2]. Maximum protein release, i.e., A = 100 % was
obtained by heating of intact samples at 100 °C in alcaline medium (1.5 n NaOH) for some
minutes - the exact length of time depending on the organism tested. The activity of malate
dehydrogenase was investigated by monitoring (365 nm, 25°C) the oxaloacetate/L-malate
conversion. Particle size distributions were determined using a laser light-scattering instru-
ment. Separability of the cell debris was also examined by empirical filtration and centrifu-
gation tests.

RESULTS AND DISCUSSION

Disruption characteristics of the organisms tested are shown in Figs. 1 and 2, using the
appropriate key parameters of wet milling (wt) and high-pressure homogenization {By) for
correlation [3]. Disintegration is represented in form of a lethality graph with (l-A) being
plotted in decimal reduction intervals. Following these results, the morphological features
of the organisms and different mechanisms of disruption significantly influence the disin-
tegration performance of the two machine types. Same evidence is given by evaluating en-
zymatic activity measurements [3] or specific energy consumption. Thus, high-pressure ho-
mogenization (major mechanism is the magnitude and velocity of the pressure drop inside
the valve) is particularly suited for the treatment of rods. By contrast, wet milling (where
disruption is mainly caused by collisions between the circulating grinding beads) will render
the better performance, provided that (especially gram-positive) cocci have to be processed.

--
1
........ -~
........
......
\
. '\ ..........

~
-----..:

~ ~

1 - A 0,1
6 E. coli "" I~
'" "-.,
~
.........
... B. megat. ........... ...........
r-----.. ~
~ ...............
o A. calcoa. ~
r-.. r~
• M.luteus ~ ~
• S. cerev.
~
0,01
o 20000 40000 60000 80000 100000
wt

Figure 1. Disruption of various microorganisms by batchwise wet milling, w > 60 lis.

258
1

-- '-
......~- -, ) -
"'-.........~
"" --...... ::--
.~""-
"" .~
t-.... -...........
"'-........,~ ~,
-~

r. . . . . . ~

1 - A 0,1 t--
t--
f-
• M.luteus ~~ "-
~ ............
.......
t-- o A. calcoa. ...... "- ........
f- "- .......... ............
f- '-. ........
E. coli ""- ..........
"'"
t--
tc,
I'
t--
B. megat.
~
&
f-

~ ~
• S. cerev.

0,01
o 80 160 240 320

Ev /(J/cm 3 )

Figure 2. Disruption of various microorganisms by homogenization, PH = 800 bar.

According to yeast cell disruption both techniques investigated show approximately

the same protein and enzyme releases, but considerably different physical properties of the
disintegrates. Wet milling gives a broad particle size distribution with two fractions: unbro-
ken or perforated cells at 5 j!m and smaller fragments (cell debris) at 0.5 j!m. Suspensions
after one or two homogenizer passages still show a narrow particle size distribution with on-
ly one peak at 5 j!m. By further passes the size distribution widens, indicating that already
broken cells are reduced to smaller fragments belonging to the second peak (0.5 j!m). These
observations are in good accordance with the results of the empirical tests on solid/liquid-
separation and have been confirmed by evaluating scanning electron micrographs [3].

Acknowledgement: This work was supported by the Deutsche Forschungsgemeinschaft.

REFERENCES

1. H.-G. Biischelberger, Untersuchungen zum mechanischen Zellaufschlu6 in Hochdruck-

Homogenisatoren. Thesis (Ph.D.), Karlsruhe University, 1987.

2. D. Herbert, P.J. Phipps, R.E. Strange, Chemical analysis of microbial cells.

In Methods in Microbiology, eds. J.R. Norris and D.W. Ribbons, vol. 5B, chap. 3,
Academic, London, 1971, pp. 242-265.

3. M. Pittroff, M. Wilk, H. Schubert, Mechanischer Aufschlu6 von Mikroorganismen

im Verfahrensvergleich zwischen NaBvermahlung und Hochdruck-Homogenisation.
Chem.-Ing.-Tech., 1992, 64, 950-953.
 DEVELOPEMENT OF MICRO CUT BLENDER

T. HOSOKAWA, H. OMURA', T. TAKEUCHI'

Iwai Kikai Kogyo Co. Ltd., 3-17-10, Higashi Koujiya, Ohta-ku, Tokyo 144, Japan,
'Nakamura Gakuen College, 5-7-1, Befu, Jyonan-ku, Fukuoka 814-01, Japan

ABSTRACT
The micro cut blender (MCB) was developed for continuous direct manufacture of
paste, jelly, and emulsion of fish meat and soybean by synchronized micron-order
pulverization and blending. Several features of MCB and properties of sardine and
soybean pastes were indicated. MCB controls the generation of heat during pulveri-
zation, and has a unique cutlery mechanism which has cutting function with three-
dimensional motion and a manufacturing system which effectively utilizes the
freezing moisture energy of the processed materials. Synchronized operations are
done at low temperature and 300 to 1000 kg of raw materials are processed per hour.
It is hoped that MCB may be useful for the developement of new processed foods.

INTRODUCTION
Among so-called protein foods, thermally-gelled foods vary in quality, size, and
shape. To produce these foods, it is necessary to chop, grind, and mix the raw
materials. These processes suffer from many problems. For example, since muscular
properties of meat vary depending on cuts, sex, age, breeds, and species of the
slaughter animal, it is difficult to produce foods uniformly. In addition, when the
raw material is subjected to an external force, a frictional heat is generated.
Proteins are generally denaturated inducing deterioration of foods. Accordingly,
it is necessary to suppress the generation of heat as much as possible. However,
no solution has been found to this problem.

RESULT
The MeB was developed to resolve the problems by devising method and apparatus.
MCB is composed of 3 units: degradation, pulverization and finishing, as shown
in Fig. 1. The unit is principally constructed of a cylinder and a rotary drum
mounted in the cylinder. Knives with substantially Goliath crane-shaped cutting
edge are attached on the inner wall of the cylinder and on the surface of the drum
arranged specifically according to the kind of unit. Cylinders and drums are inte-
grally connected by a single shaft. This unique cutlery mechanism which has
cutting function with a three-dimensional motion and manufacturing system utilizes
effectively the freezing moisture energy of materials.
260
...-------+t Vacuum

1..---- Liquid
Frozen Raw
Material ~___~ ~ IPaste I
-15 0~-30"C -200~-10"C -100~-5"C -100~-51:
90 W 11 KW 30 KW 15 KW
(750~1500ppm) (1500~2000ppm) (1500~2000ppm)

Figure 1. Flow of MCB

During operation,it is necessary for the frozen raw material to assume a substan-
tially square-shaped block in order to keep the amount of raw material constant for
processing and it is also necessary that the blocks have the same size in each
cycle of the process.
The frozen raw material of -30~-151: is instantaneously broken down continuously
into minute particles up to about 1 mm at -20~-10"C in the degradation unit.
Then, in the pulverizing unit, small amount of salt and some auxiliary materials
such as egg white are added, if necessary, and chopped at a constant rate in the
same manner as degradation. By uniform mixing therewith, pieces of about 600 ~m in
particle size with -10~-5"C are produced. Finally, in the finishing unit, minute
particles are poured with additional liquefied materials,if proceeds continuously,
mixed and kneaded under vacuum. Thus, within a very quick process from entrance of
frozen block of the raw materials, kneaded paste of fine particle of about 100 pm
is continuously produced with capacity between 300 and 1000 kg per hour at
-10~-5"C (Fig.2).
Properties of ground paste and jelly of sardine meat are shown in TABLES 1 and 2.
By employing soybeans soaked in water, better soy-milk and soybean curd can be
also prepared, as shown in TABLE 3.

TABLE 1
Properties of sardine paste"

Hardness C ohes i veness Adhes i veness Gumminess

(X 10 2 dyn/cni) (X10- 1 ) (X 10 2 dyn/cni) (X 10 2 dyn/cni)
6.89±1.61 7. 32±0. 31 7.12±0.39 5. 18±1. 32
*homogenate (20 g paste+ 15 11& H2 0)

TABLE 2
Properties of jelly' of sardine paste

Rupture load Jelly strength Breaking strain

(g) (X10 5 dyn/cni) (X10- 1 mm/mm)
74.7±3.7 4.3±O.4 4.5±O.3
*homogenate (100 g paste+25 11& H2 0) was heated at 80"C for 40 min, cooled and
kept night over in a refrigerator.
 261
TABLE 3
Soy-milk and soybean curd

Soy-milk Soybean curd

Flavor ( ~pm) Hardness Rupture load
I fiexanaI I -exanoI
MCB 0.15±0.03 O. 25±0. 02 37.04 82.10
Conventional
system 0.26±0.01 O. 21±0. 02 22.14 63.16

DISCUSSION
In chopping by MCB, a number of ice pieces dispersed and broken through cutting
inpact by the chopping knives serve as the knife for further chopping.
During processing, the raw material and apparatus need not be cooled. This is
because the frozen raw material is employed as the starting material and any
temperature rise is restricted by a latent heat of fusion of the ice pieces formed
by chopping or because the required time for the processing is very short to allow
the frozen raw material to be completely pulverized before melting. Thus, denatura-
tion of protein may be retarded. In conclusion, features of MCB are demonstrated as
follows: (1) Paste can be prepared even at temperatures below zero, (2) very small
amounts of additive can be evenly blended, (3) frozen raw materials can be directly
used, (4) production begings as the system starts operating, (5) contamination is
controlled due to air-tight system, and (6) circulation washing is easy. Based on
these features, several advantages of the MCB became apparent. When comparing with
a conventional system in making paste of livestock and fish meat, pulverization
of soaked soybean, and softening butter and cheese: (1) Because of continuous
process,space and labor saving are possible, (2) manufacturing process can be
compressed, (3) power load is greatly reduced, (4) thawing space and process
can be eliminated, (5) keeping a high level of sanitation is easy, and (6) dis-
assembly and assembly of the apparatus is easy.

REFERENCE
Patents:
USA(No.5080922),
UK(No.2237495,2253332)
Germany (No. 3990873)
Sweden (No. 9000255-1)
Chi Ie (No. 38069)
Australia(No.628071)
Korea (No. 54230)
Taiwan(No.44886)

Fig u r e 2. ·Production exit

•
 SIMULATION IN FOOD PLANT DESIGN USING SIMAN

J. Peter Clark, Stephen M. Hyman, Richard Symns, and Robert Kelley

A. Epstein and Sons International, Inc.
600 West Fulton Street
Chicago, Illinois 60661-1199 USA

ABSTRACT

The formulation of products from dry ingredients is a common unit operation in

baking, pet food, and dry mixes. Selection of the proper number and size of scales,
mixer, and transport systems is complex and depends upon the formulas, production
schedule, and process capacity. There also are choices among control strategies.
SIMAN is a high level language which can be used to simulate such a situation,
permitting comparison of flow schemes, control rules and equipment parameters. As
with most such languages, SIMAN requires some speCial training and programming.
The availability of color animation makes communication of results dramatic.

INTRODUCTION

The usual objectives of using simulation of a process are:

• Testing feasibility, for example, of alternative process flows

• Develop and test control strategies
• Communicate the concept and performance to others, for example, operating
personnel

This paper describes the use of simulation in the design and development of a mixing
system. Mixing is a common operation in many food industries including baking, pet
foods, breakfast cereal, and many others. There are many possible solutions to the
general mixing problem. If we take the illustrative case of four mixers receiving
several different ingredients, we can imagine using one scale for each mixer or one
scale for the entire system, or one scale for each ingredient. Each one of these
choices can be found in application somewhere. Thus, the selection among them is
not intuitive and it is difficult to analyze the performance of different choices.

In the particular case discussed here, some of the questions that required resolution
include:
 263
• The ability to share mixers in support of multiple process lines
• The proper number of bulk scales
• The need and proper location of surge vessels
• The time required to change formulas
• The proper way to control the system

The topic of control is interesting in connection with simulation. A computer program

can be written to reproduce the behavior of a physical system relying exclusively on
programming techniques. Alternatively, the program can use the same type of
decision rules that the physical system would use for control. This permits the
demonstration and testing of control algorithms using the computer. Such
development and testing in advance is much more efficient than doing the same thing
in the field with the physical system.

SIMAN

SIMAN l11 is one of many high level simulation programs developed primarily for
industrial engineering modeling of manufacturing systems. Some other similar
programs include: Slam, Batches, Extend, Aspen, and GPSS. The various available
programs differ somewhat in their ease of use, cost, hardware requirements, and
output techniques. SIMAN was chosen and used in the present case because it was
available and because the personnel involved were familiar with its use. There is no
suggestion that it is the only or even the best solution to this simulation exercise.
However, it did perform well.

SIMAN is event oriented which means that the computer program looks for triggers
of specific events and then executes them. This can be contrasted with programs
which are time oriented and operate according to some internal virtual clock.

SIMAN is written in a mixture of Fortran and C which means that users needs to be
capable of writing modifications in these languages. Output can be illustrated in
animated diagrams using a program called Cinema. SIMAN is oriented towards
discrete moves of entities rather than continuous flows of material. This means that
the continuous flowing systems must be approximated by multiple entities. In our
case, an entity was defined as 100 pounds of ingredients.

SIMAN was developed to run on mainframe computers, but also runs on personal
computers (PC's) and mini computers such as Digital Equipment (DEC Vax). It is a
very flexible and powerful programming language. One feature is that the user can
customize an interface, for example, by creating a menu of changeable parameters.

Discoveries

In the present case, some specific circumstances required development of a unique

process flow. In the course of executing many simulations, a number of discoveries
were made. Because the process was to be a retrofit, it was decided to use one scale
to feed four mixers. Because the take away rates of the processes were quite
264
different, it became necessary to include a "cross connection", which permitted one
of the mixers to be supplied from either of two surge bins. As part of the
development of a control strategy, it became obvious that rules were necessary for
prioritizing requests for use of the single scale. Various rules were tried and
successful ones identified. The successful rule relied on a measurement of material
in the system (using load cells in "real life") and a calculation of time remaining before
material was exhausted. This permitted devoting attention to the most needy system.

Because of the intention of sharing mixers among processes, it became necessary to

insert a "check for empty" rule to avoid cross contamination. Various size surge
vessels were tested and the minimum required size determined directly from
simulation runs. Minimum surge is desirable in order to shorten turn-around time and
minimize work-in-process. It also reduces capital cost.

The required rates of transfer among systems, performed by pneumatics in "reallife n ,

were determined by trial and error using the simulator. Finally, the number of
operators required to dump bagged ingredients was determined by observation of
simultaneous activity at the bag dumps. It was determined that no more than two
operators would be required.

Summary

SIMAN was capable of simulating the mixing system described here, but it did require
some additional code written in Fortran. Continuous flow was approximated by
discrete entities with no significant loss in accuracy. As an event oriented program,
SIMAN computes the time at which events occur, thus there is no direct relationship
between the length of a simulation and the length of real time simulated. In fact, the
simulator requires adjustment when executed on different computers because of the
different execution speeds.

The model evolved by trial and error and testing of flow sheets. This is probably a
realistic experience in that it is difficult to conceive a full blown simulation correctly
from the start. Likewise, control strategies were changed and developed with
experience. One unanticipated benefit for the simulation is in planning and scheduling
performance of the plant. Various combinations of formulas can be scheduled in
various quantities and the model computes the time for the beginning and the end of
that portion of production. Potential conflicts can be identified and required resources
anticipated.

The general formulation and mixing problem common to many food industries is a
good candidate for simulation because it is complex and important, but it does not
require modeling of unknown chemical reactions or other non-linear processes.

REFERENCES

(1) System Modeling Corporation, SIMAN IV ver 1.36, Sewickly, PA, USA, 1990
 INTELLIGENT MIXER FOR BREAD DOUGH

YASUO TORIKATA and NOBUO BAN

Rheon Automatic Machinery Co. Ltd
2-3 Nozawa, Ustunomiya-shi, Tochigi-ken, 320 Japan

ABSTRACT

This paper describes how to detect "dough development" through digital signal
processings of the mixer's torque and how to control mixing without an ordinary
mathematical model. Applying those methods, our intelligent mixer (speed
variable farinograph, robots for flour-water addition) worked well. Those digital
signal processings were applicable to commercial mixers whose torque
oscillated highly.

INTRODUCTION

Work input by a mixer changes a flour-water paste into a viscoelastic dough (this
change is called "development"). Mixing has been carried out by experienced
operators because no sensing devices for development have been available. We
tried to automate the mixing process based on the following assumptions: (1) As a
dough reacts differently to the drive of mixing rods, according to dough
development, the degree of dough development can be calculated from the
inverse of the signal processing of the mixer's torque. (2) Shearing work is most
effective for protein network formation and apparent dough development. The
mixing efficiency (= shearing work / total work) of various mixers affect the mixing
time, but the effect on signal processings is minimal.
We studied our intelligent mixer, whose oScillating components of torque are
relatively small and regular, and commercial mixers, whose torques oscillate
highly and irregularly.
266
MATERIALS AND METHODS

We tested our intelligent mixer (speed controllable farinograph, robots for flour-
water addition, computers) and commercial mixers (vertical type: Kanto Kongouki
SS-10, horizontal type: Fujisawa FH-3, spiral type: OASE). As digital signal
processings, we used smoothing on cognitive science supplied by Yamatake
Honeywell Co. Ltd, wave form analysis and FFT (256 points at 60-ms sampling
intervals). We prepared dough samples for SEM through freeze drying, gluten
fixation, enzymatic digestion of starch, and freeze drying(1).

RESULTS

Figure 1 shows a smoothed farinogram (the original also shown) at constant

mixing intensity, which has a linear part and a maximum point.

10'~------------------------~
(maximum point)

(infelction point)
original farinogram

"Torque Increasing Rate" = ~

(start) 300
mixing time [ s ]

FIGURE 1. Torque Signals of Farinograph and its Signal Processing.

Dough at the maximum point was judged to be fully developed by experienced

mixing operators and was confirmed by our baking test. An empirical equation
about toruqe-mixing and fuzzy rules to correct nonlinear effects by three
parameters (water content, temperature, and mixing speed) were obtained.
Applying the equation gave a good statistical relation between "torque increasing
rate" (its definition is shown in Fig. 1) and maximum torque value. This statistical
relation led to regulation of the maximum torque value (a good indicator of dough
properties) through flour (or water) addition by robots. Addition of flour or water
in the early mixing stage caused no apparent influence on developed dough
and baked products. The wave form of oscillating components, though there
seemed a lot of background noises (refer to the original farinogram in Fig.1), had a
good correlation with the degree of dough development. Combining wave form
analysis with step-wise change of mixing intensity, it was easy to find the
optimum mixing intensity at each mixing stage. Applying these results, our
 267
intelligent mixer produced a dough with the desired maximum torque (a good
indicator of dough properties) in a short time.
Oscillating components of mixers' torque (or electric power consumption) in
decreasing order were vertical mixer > horizontal mixer > spiral mixer >
farinograph. The spectrum of the vertical mixer's signals after FFT were
correlated with the degree of dough development. From the vertical mixer's
signals, a curve similar to that shown in Fig.1 was obtained through signal
processings: connecting peaks after removing low waves due to skid of mixing
rods. By combining the two methods, the degree of dough development and the
point where dough was developed could be detected faster and more reliably.
No difference in the gluten structure due to mixing action (farinograph and
vertical mixer) was found by SEM; gluten covered half of the large starch
particles in both developed doughs.

DISCUSSION

Enough information could be derived through signal processings of the mixer's

torque or electric power consumption. And our statistical and fuzzy hybrid model
was effective to describe nonlinear relations. Our conparative study of an
intelligent mixer and a commercial vertical mixer showed the validity of our signal
processing and our hybrid model. They are applicable to other types of mixers for
the following reasons:
Kilborn and Tipples(2) reported that dough was developed by sheeting rolls with
less energy requirement than an ordinary mixer. We also got developed dough
by sole shearing work using a mixture of dry flour and powdery ice. And no
peculiar effect on the gluten structure due to mixing actions was found in this
study. So, the mechanical differences among mixers will affect their mixing
efficiency but not the dough properties, and our signal processing and hybrid
model relate only the dough properties.

REFERENCES

1. Nihei, K., Torikata Y., and Kageyama M., Structural Changes of Flour-water
Dough during Mixing Process. Nippon Shokuhin Kogyo Gakkaishi, 1990, 37,
pp.266-269.

2. Kilborn, R. H. and Tipples K. H., Implication of the Mechanical Development of

Bread Dough by Means of Sheeting Rolls. Cereal Chemistry, 1974, 52, pp.648-
657.
EFFECTS OF L-ASCORBIC ACID AID RELATED COlIIPOUJIDS 01 GLUTEI AID STARCH
DURIIG BREADJllAKIIG I I A HOME BAKIIG MACHIKE

LI ZHANG, NAOFUMI MORITA, and MASANOSUKE TAKAGI*

Department of Agricultural Chemistry, College of Agriculture, University
of Osaka Prefecture, 1-1, Gakuen-cho, Sakai, Osaka 593, Japan
*Department of Bioengineering, Faculty of Engineering, Kansai University,
3-3, Yamate-cho, Suita, Osaka 564, Japan

ABSTRACT

The addition of L-ascorbic acid (L-AsA, 8 ppm) or sodium D-AsA (D-AsA-Na,

20 ppm) increased the loaf volume of bread baked. The gelatinization
temperature of starch in the dough containing D-AsA-Na (50 ppm) was about
81°C. The addition of D-AsA-Na (50 ppm) to the wheat flour increased the
mechanical stability of dough during mixing over that of the control or D-
AsA, measured by farinograph. Addition of D-AsA-Na to the dough gave an
intermediary state between the filamentous and viscous states which
correspond to those of L-AsA and the control, respectively. As the
result, D-AsA-Na oxidized the SH groups of gluten mildly, followed by
formation of SS-cross linkages that changed the mode of wrapping of starch
granules, then causing a decrease in the gelatinization temperature.

IITRODUCTIOI

L-Ascorbic acid (L-AsA), which is known as vitamin C, has an antiscorbutic

effect and is also widely used as an additive for foods and beverages in
food industry. D-Isoascorbic acid (D-AsA) also has a strong reducing
activity like that of L-AsA [1]. D-AsA in an aqueous solution is known
to be easily oxidized in the presence of oxygen, though D-AsA is stable as
a powder. So, D-AsA was tried as a dough improver for breadmaking in the
1960's, but it is thought ineffective as an improver for breadmaking [1,
2]. However, no systematic studies have been done. Here, we tested the
effects of L-AsA, D-AsA, and sodium D-AsA (D-AsA-Na) in a home baker.

JlATERIALS AID JIETIIODS

Methods for breadaaking [3,4]. Wheat flour 'Hermes' is a strong type of

flour that contains 11.8% protein and 0.38% ash with 11.8% moisture. The
breadmaking formula was 280 g of flour, 5 g of NaCl, 17 g of sucrose, 3 g
of dry yeast, and 210 g of water with 14 mg (50 ppm) of AsA or D-AsA-Na,
 269
unless otherwise stated. Bread was made in an automatic breadmaker for
family use (SD-BT-3, Matsushita Electric Co., Ltd.). The breadmaking
process took 4 hours including mixing, proofing and baking. After
baking ended, the bread was weighed i ..ediately, and the loaf volume was
measured by the method of rapeseed displacement. The specific volume of
bread baked was obtained from the volume (ml) divided by the weight (g).
Analytical aethods. Differential scanning calorimetry (DSC) was done
with 20 mg of a dough sample containing a given amount of L-AsA or D-AsA-
Na after mixing for 25 min in the baker [3J. For the reference of DSC, 4
mg of wheat starch (moisture 11.0%) and 10 ~l of pure water were used.
Scanning electron microscopy (SEM) was done with a small portion of
dough after mixing for 25 min. The dough sample was fixed with Os04 (2%).

RESULTS AND DISCUSSION

The amount of L- and D-AsA (100 ~M) remaining in the phosphate buffer (pH
6.8) after incubation of the reaction mixture containing AsA oxidase at
25°C was tested photometrically at 265 nm. Ten percent of L-AsA or 47% of
D-AsA remained after 30 min of incubation.
Remaining amount of L- and D-AsA in the dough during breadmaking were
measured photometrically after addition of additives (100 ppm) as
reported by Roe and Keuther (1943). After mixing for 10 min 17% of the D-
AsA remained; 10% after 30 min; and 8% after 90 min. On the other hand,
for L-AsA, the amount remaining was 10% after mixing for 10 min; 4% after
30 min; and less than 2%, after 90 min. Since D-AsA was reported not to
be oxidized to the dehydro-form during mixing [2J, D-AsA was not effective
as an improver for breadmaking. However, from our experiment described
here, D-AsA can be said effective as an improver for breadmaking.
The effects of D- and L-form of AsA on the specific volume of bread
were tested. Addition 30f 8 ppm L-AsA to the ingredients increased the
loaf volume to 4.54 cm /g from the control value of 4.49. At higher
concentrations of L-AsA, the loaf volume decreased more distinctly,
suggesting the overoxidation of SH groups in the gluten molecules.
Addition of 20 ppm D-AsA-Na increased the loaf volume of bread
significantly (p < 0.05) compared with that of the control, then decreased
it at higher concentrations.
Arrival times of farinograms of dough with D-, L-AsA, or D-AsA-Na were
about 2.5 min from the start of mixing. The development time in D-AsA
was about 5 min and was the longest, and that of the other additives used
were about 4 min after mixing. This suggests that D-AsA has the effect
of lowering the pH of the dough, since D-AsA was not easily oxidized in
the absence of AsA oxidase. However, D-AsA-Na was stable until 20 min of
mixing and had a relatively high consistency like that of L-AsA.
SEM observation was done with dough after mixing for 25 min (Fig. 1).
The control dough without additives seems to have gluten that covers the
surface of starch granules almost uniformly. So, the boundaries of starch
granules could not be seen. This may be caused by the uniform wrapping
of gluten mixed with glutenin and gliadin during mixing for 25 min in a
home baker [5J. With the addition of D-AsA-Na a somewhat filamentous
part of the dough can be seen, but as a whole, primary starch granules
were covered with gluten [6J. This suggests that addition of D-AsA-Na to
the dough causes intermediary changes of dough between those of L-AsA and
the control, and also a part of gluten was oxidized to form a three-
dimensional network of gluten.
The onset temperatures of gelatinization of starch granules in the
270

Conlrol x 2.5 K L- ASA x 2.5K D-AsA-Na x 2.5 K

Figure 1. SEM of dough with 50 ppm L-AsA or D-AsA-Na in breadmaking.

dough with L-AsA or the control were both almost 59.0·e, but the peak
temperature differed clearly: 79.8·C for L-AsA and 82.0oe for the control.
But, for the case of D-AsA-Na, the peak temperature of gelatinization of
starch was 80.7°C, and this temperature was almost halfway between that of
L-AsA and the control.
The amount of 20 ppm D-AsA-Na necessary to increase the loaf volume
was large compared with that of L-AsA. This shows that the oxidation
rate of D-AsA is the rate determining factor for the improving effect on
loaf volume. Anyhow, the dehydro form of D-AsA-Na thus formed oxidizes
SH groups and forms a network of gluten molecules. As a matter of fact,
the mode of wrapping of gluten molecules over the surface of starch
granules can be changed by the addition of D-AsA-Na [3,6]. Hence, it may
be considered that the gelatinization temperature of starch upon the
addition of D-AsA-Na is an intermediate between those of L-AsA and the
control. Thus, the addition of D-AsA-Na can make a good quality of bread
both from internal and external points of view. Therefore, D-AsA-Na is
expected to be used as a practical improver for breadmaking.
ACKNOWLEDGaIENT

This study was supported by the Elizabeth Arnold Fuji Foundation for which
the authors' thanks are due.
REFERENCES

1. Lillard, D.W., Seib, P.A., and Hoseney, R.C., Isomeric ascorbic acid
derivatives of ascorbic acid. Cereal Chem., 1982, 59, 291-96.
2. Johannson, H., and Cooke, A., Role of ascorbic acid in short time
bread processes. Bakers Digest, 1971, 45, 30-35.
3. Zhang, L., Yamaguchi, Y., Morita, N., and Takagi, M., Effects of
ascorbic acid and related compounds on SH groups and gelatinization of
starch in wheat flour dough. Denpun Kagaku, 1991, 38, 351-59.
4. Sharp,C.Q. and Kitchens, K.J., Using rice bran in yeast bread in a
home baker. Cereal Foods World, 1990, 35, 1021-24.
5. Zhang, L., Yamaguchi, Y., Morita, N., and Takagi, M., Effects of
ascorbic acid and related compounds on wheat flour dough and starch in
breadmaking. Denpun Kagaku, 1992, 39, 183-87.
6. Pyler, E.J., Baking Science! Technology, 3rd edition. Sosland
Publishing Co., 1988, pp. 83-131.
 ANALYSIS OF CONTINUOUS WHIPPING OF CREAM

M.KIKUCHI, M.ENDO, N. YANAGIHARA, T.MIYAMOTO,

R.WATANABE, S.MATSUMOTO*. Engineering Research Center,Morinaga
Milk Industry Co. ,Ltd. 4-515 Tateno, Higasiyamato-shi, Tokyo 207,
Japan, *Dept. of Biochemistry and Eng., Tohoku Univ. Sendai 980 Japan

ABSTRACT
There are two types of equipment used for making whipped cream, batch
and continuous type whipping machine. The continuous whipping machine has
several advantages to a batch type mixer. Therefore, recently large cake
making companies have utilized the continuous whipping machines. But there
is a severe problem in the continuous whipping called the cycling
phenomenon. The cycling phenomenon means that the firmness of whipped
cream changes periodically. We researched the cause of the cycling
phenomenon, and found that the whipping rate was strongly dependent on the
dasher pressure. Using these results, we developed a new stable continuous
whipping machine.

INTRODUCTION
Whipped cream is an important food material because of its special texture
and flavor, and it is used in the decorating of cakes and other foods. This
study examines the method of making whipped cream continuously, namely,
continuous whipping machine and its stability.
The continuous whipping machine is constructed of an agitator called a
dasher and an outlet piping, shown in Figure 1. In the continuous whipping
machine, mixture of cream and air are sent to the dasher, and adding agitation
we obtain whipped cream at the outlet pipe. The continuous machine has
several advantages to a batch mixer. For example: it is easier to control the
firmness, suitable for mass production, superior in sanitation, and able to
control the overrun which means air volume fraction in the product.
But there is a severe
~r~ibplpe~gi~al\~~ ~~en~yn~l~~gs [Dist~~~~ Dasher
phenomenon in which the N Piping
firmnes s changes
periodically. The purpose G----1I----~
of our study is to reveal
this cycling phenomenon
and to develop a new
continuous machine. We Figure 1. Continuous whipping machine
272
will discuss the outline of our studies. More details were reported in the
papers 1,2,3).
EXPERIMENTAL METHOD
We used two types of raw cream produced by Morinaga Milk Industry
Co.,Ltd.,for our experiments. Cream A was a fresh cream from pure dairy
fat, and cream B was a compound cream created by a mixture of dairy and
vegetable fat. In order to evaluate the firmnes s, or the holding ability of the
shapes of whipped cream, we used a specially designed cone penetrometer.
The penetration depth was measured in millimeter units, so we assigned a PE
value for firmness. As indicated, the firmer the product, the smaller the
numerical value. The PE for optimum firmness for decorating cakes is 17mm.
We used the batch mixer and the conventional type continuous whipping
machine for the tests of cycling phenomenon and whipping rate. We made a
new capillary viscometer to measure the flow properties of whipped cream.
U sing the whipping rate and the flow properties, we tried to analyze the
cycling phenomenon by computer simulation.
Finally we developed a new continuous whipping machine and examined the
dynamics of the machine.
RESULTS AND DISCUSSION
1. Cycling phenomenon of continuous whipping
The Figure.2 shows that by rapidly increasing the dasher's rotational speed
at a steady state, both dasher pressure and pressure drop in the piping
increased. Then a pattern of increasing and decreasing pressure began to
occur, and the firmness of whipped cream also changed and the cycling
phenomenon took place. Even though the dasher was held at a constant rate,
vibrations occurred. The erratic firmness of the whipped cream prevented us
from using it for decorative shapes on cakes or any other products.
2. Analysis of whipping rate
Next we researched the whipping rate of batch and continuous whipping.
We tried to analyze the whipping rate by the firmness, namely, penetration
depth. We defined the whipping index (x) by using the reciprocal of
penetration depth PE. We applied a kinetic equation called auto catalytic-type
equation to the whipping process. From our experiments, we concluded that
the whipping process could be- well expressed by the auto catalytic-type
equation. And the whipping rate constant (k) rapidly decreased by increasing
the dasher pressure. This result suggests that the air volume in the dasher,
namely the surface area of air bubbles, plays an important role in the
continuous whipping process.
3. Flow properties of whipped cream
Whipped cream is a compressible fluid consisting of half air, and it shows a
complex behavior. We analyzed the flow properties under pressurized
conditions by using a newly designed capillary tube viscometer. As a result,
when the pressure was low as 1.21 atm, the flow properties showed as a
non-Newtonian fluid with yield stress. But when the pressure was raised at
3.17 atm, the yield stress was eliminated and the viscosity decreased. It
behaved closely as a Newtonian fluid. We applied the Herschel-Bulkley
model. The flow properties of whipped cream were well expressed by the
model, with firmness, shear rate, and pressure as parameters. As indicated,
we found that the flow property of whipped cream was strongly dependent on
not only shear rate but also pressure.
4. Simulation of cycling phenomenon
Using the whipping rate and the flow properties, we analyzed the cycling
 273
phenomenon by computer simulation. The cycling phenomenon is caused by
the interaction between the whipping rate in the dasher and the pressure drop
of the outlet piping. We made a model of the continuous whipping system in
which the dasher was connected directly to the outlet piping. In this model,
we assumed piston flow, constant temperature, and progress of whipping
occurs only in the dasher and we were able to treat quantitatively by using
whipping index(x).
From our simulation, we could almost reproduce the cycling phenomenon on
a computer and concluded that the cycling phenomenon was a vibration mainly
caused by pressure change.
5. Development of new continuous whippin~ machine
Then we attached a new pump at the outlet of the dasher, and controlled the
dasher pressure. In this way, we developed a new continuous whipping
machine. The stability of the new machine by changing the firmness is shown
in figure. 3. In this system, the dasher pressure was controlled at a constant
value. While increasing the dasher speed from the optimum firmness, a firm
whipped cream was produced and the pressure drop in the outlet piping
increased. After reaching a certain value, the pressure drop become stable and
an extremely firm whipped cream was obtained consistently. No cycling
phenomenon as shown in figure. 2 of old the machine was observed.
Based on our research, we made several kinds of continuous whipping
machines for practical application which can treat 80 to 600 kg of cream per
hour. The new machine is sanitary by using aseptic air, it has extremely stable
characteristics, and can contribute to the mass production of whipped cream.

180r-----------,18 180 18
P
160 P/.,......" / ...-/\ 16 160 ·-'-'J-·-·_._._.-;._._._._.-16
ro 140 ._._.__./1 .
\../
\
\ /" 14 140f- 14
~ 120- 12 T
~

v !1!. 120 12. T

~ 100- N 10 ~ ;:5 100,. N 9.6 __ 10.55-110 ~
~ 80~-->9.5 5- J 1 8 Z 0.. 80 9.0 -- 9.6 5-d PE=9.9 8
0..:- 60r PE=9.1 - 6 <l 60 ,,-- 6 Z
/ .... '\ ,-_/" \ 4 0..:- PE=14.4//
40r ~/ \llP// \ 40 ~:=_~~~?_~"-----/ 4
20~----p~~19~0 L'--' \, /- 2 20F'---- 2
o I PE=29.4
o 1 2 3 0 00 1 2 3 45 6 7 8 9 11:9
Ti me [min] Ti me [min]

Figure 2. The occurrence of Figure 3. Dynamics of a newly

cycling phenomenon caused by developed machine
a step increase in rotational
speed of dasher
REFERENCES
1 Kikuchi,M., H. Endo, K.Inagaki, M. Kanzaki and S. Matsumoto: Kagaku
Kogaku Ronbunshu(Japan), 16, 867-874(1990)
2 Kikuchi, M., T. Miyamoto, N. Yanagihara, M. Kanzaki and S. Matsumoto:
Kagaku Kogaku Ronbunshu(Japan), 18, 136-138(1992)
3 Kikuchi,M., N.Yanagihara, T.Miyamoto, M.Kanzaki and S.Matsumoto:
Kagaku Kogaku Ronbunshu(Japan), 18, 570-575(1992)
 CONTROLLED BIAXIAL EXTENSION OF FOOD POLYMER GELS

MARVIN A. TUNG, IAN J. BRITT AND JUMING TANG

Department of Food Science and Technology, Faculty of Engineering

Technical University of Nova Scotia, Box 1000, Halifax, NS, B3J 2X4, Canada

ABSTRACT

Aqueous gels of polysaccharides and proteins were formed into cylindrical shapes and
compressed between lubricated teflon platens in a mechanical testing machine. The
samples maintained a cylindrical form while decreasing in height and increasing in
diameter, thereby developing biaxial extensional flow. Viscoelastic properties of the
gels were also probed using stress relaxation tests. With experimental values for these
properties, stress-time relationships during the biaxial extensional flow were examined
using an upper convected Maxwell model.

INTRODUCTION

Compression tests are commonly used to study textural properties of food gels. Similar
to other polymer systems, food gels are viscoelastic; thus, their mechanical response
when deformed at constant crosshead speed depends upon the viscoelastic properties
and loading conditions. A thorough description of the biaxial extensional flow during
lubricated compression tests requires consideration of viscous components in the
constitutive relation. Although extensive research has been performed on biaxial
extensional flow of nonfood polymers, limited literature of similar effort is available on
food systems. Biaxial flow of doughs was studied by Bagley [1] in which an upper
convected Maxwell model was applicable to data from constant crosshead speed
experiments. The objective of this study was two-fold: first, to examine whether the
same constitutive relation applies to gels of polysaccharides and proteins and secondly,
to determine if the parameters obtained from stress relaxation tests can be used to
correlate the biaxial extensional flow in compression between lubricated parallel plates.

METHODS

Sample Preparation
Three different types of gels were prepared: 2% gellan polysaccharide (GP), 1% gellan
combined with 1% locust bean polysaccharide (GLB) and surimi (fish protein, SP).
Gellan (KELCOGEL) and locust bean polymer (Kelco Div. of Merck and Co., San
 275
Diego, CA) were dispersed in distilled water and gently heated to 90°C. Predetermined
amounts of calcium chloride were added to give 8mM Ca++ to GP and 5mM to GLB.
The solutions were then filled into cylindrical stainless steel molds of 21 mm diameter
and 140 mm length. The samples were cooled in running water at 15°C for 15 min and
held overnight at 20° before further testing. Surimi samples were prepared according
to Lanier and others [2].

Stress Relaxation Tests

Gel specimens of 20 mm length were cut from the long cylindrical samples. Stress
relaxation tests at constant compressive strain were conducted with a mechanical testing
machine (Model 1125, Instron Corporation, Canton, MA) equipped with teflon plates
that were lubricated with low viscosity oil in contact with test specimens. Five % strain
was suddenly imposed on the specimens at a crosshead speed of 200 mmlmin, and
maintained while measurement of stress during relaxation was made over 10 min.

Biaxial Extension Flow in Compression Tests

For extensional flow, the geometry of the specimens were the same as previously used.
Samples were compressed between lubricated flat teflon surfaces in the mechanica:
testing machine at three different crosshead speeds: 2, 20 and 200 mm/min.

RESULTS

Stress Relaxation
Stress (cr) as a function of time (t) from the relaxation responses were fitted by three-
term Maxwell models using the method of successive residuals [3]:
3
a eo :E
i - 1
(1)

where Eo is the imposed strain, Ei is the modulus of elasticity and T rel.i is the relaxation
time. The corresponding elongational viscosity (Ai) was calculated as in [3]:
(2)

The results of fitting for three gels are summarized in Table 1. The predictive curves
fitted the data closely.

Uniaxial Compression
In lubricated compression at a crosshead speed of V, the stress-time relation for the gel
specimens may be described by the following Maxwell model [1]:

r d 2a +2T do + (1_2V2).2)a = 311 V (3)

rel dt 2 rel dt h2 h
276
where ho and h are the initial and transient heights of the sample, respectively, t is time,
and h=ho-Vt. The cr represents stress, Tre1 is a relaxation time and YJ is the shear
viscosity. Assuming a Newtonian fluid in the dashpot elements of the Maxwell model,
the shear viscosity is related to the elongational viscosity '" by [4]:
1 (4)
" = -A
3

Eq. 3 was solved numerically using commercial software with the value of the initial
condition, (dcr/dt)t=o' determined from the compression tests and the experimental
values for Tre1 and YJ from the relaxation tests at 5% strain. We found that the best
agreement between the prediction using Eq. 3 and experimental data were obtained with
values of Tre1 and YJ corresponding to the first term ofEq. 1 (Table 1). In this case, Eq.
3 accurately describes compression data up to about 30% strain.

TABLE 1
Values from the relaxation tests at 5% initial strain.

Gel El E2 E3 Trel.1 Tre 1. 2 Tre1.3 A.l "'2 A.3 CV

(kPa) (s) (Pa's x 107) (%)

GP 129.8 280.1 121.4 1980 82 10 26 2.3 0.12 5.2

GLB 91.5 13.2 11.5 2210 83 4 20 0.12 0.006 1.9
SP 139.1 34.7 62.5 1940 87 10 27 0.30 0.063 4.0

ACKNOWLEDGEMENTS

This study was supported by research grants to Marvin A. Tung and Juming Tang from
the Natural Sciences and Engineering Research Council of Canada.

REFERENCES

1. Bagley, E.B., Christianson, D.D. and Martindale, lA. Uniaxial compression of a

hard wheat flour dough: data analysis using the upper convected Maxwell model.
J. Texture Studies, 1988, 19, 289-305.
2. Lanier, T.C., Hart, K. and Martin, R.E. A Manual of Standard Methods for
Measuring and Specifying the Properties of Surimi, Univ. of North Carolina Sea
Grant College Program, Raleigh, NC, 1991.
3. Mohsenin, N.N. Physical Properties of Plant and Animal Materials. Gordon and
Breach Science Publishers, New York, 1986.
4. Petrie, C.lS. Elongational Flows - Aspects of the Behaviour of Model Elasticovis-
cous Models. Pitman, London, 1979.
PROCESS OF FORMATION OF POROUS STRUCTURE OF BAKED FOODS DURING
HEATING

YOKO SHIMIYA and KEIKO SASAKI~

Faculty of Home Economics,
Jissen Women's University
4-1-1 Osakaue, Hino-shi, Tokyo 191, Japan
~ Faculty of Education,
Fukushima University
2 Asagawa Aza Sugumichi, Matsukawa-chou, Fukushima-shi
Fukushima 960-12, Japan

ABSTRACT
The process of the formation of porous structure of the typi-
cal baked goods, bread, cake, cookie and chou, during heating
were studied by microscopic observation. Before baking, small
bubbles, 2000 to 35000 /cm2, were entrained in those of the
mixture of ingredients. The distribution of bubble size fre-
quency was approximately logarithmic normal, and the porosity
of those mixtures ranged from 0.05 to 0.66. After baking,
the number of cells decreased to less than 5% of those bubbles
before baking. The distribution curves of the cell size became
a little disordered. On the other hand, the remaining cells
expanded themselves and the porosity of the baked products
increased in the range 0.54 to 0.95.

INTRODUCTION
A lot of bubble nuclei (small air bubbles) were recognized in
the mixture of ingredients of baked goods before baking (1)-
(5). Because the porous structure of ~aked goods should theo-
retically be formed under the influence of those bubble
nuclei, it is important to understand how the structure of
these bubble nuclei changes with time. We tried to measure
quantitatively the bubble nuclei structure entrained before
baking in typical baked goods such as bread. cake, cookie and
chou by microscopic observation, and their time course during
heating. In this paper, the bubble structure before baking and
the porous structure after baking were compared.
278
MATERIALS AND METHODS
Baked goods
The ingredients of the baked goods studied in this experiment
were as follows:
bread: wheat flour. water. yeast. NaCI. sucrose
cake wheat flour. egg. sucrose. water
cookie: wheat flour. egg. sucrose. fat
chou wheat flour. egg. water. fat
They were mixed and baked according to a standard formula.
Microscopic observation
The mixture of ingredients before baking was prepared for the
light microscope and the stereoscope. The thickness of the
preparations was controlled by the bubble size not to collapse
the bubbles. The sections of 2-3 mm in thickness of the baked
products after baking were prepared for the stereoscope.
Bubble (before baking) and cell (after baking) size distribu-
tion
Distribution of bubble (cell) size was analized by SIZE
ANALYZER. using the data of microscopic observation. and was
expressed as ni/A (number!cm 2 ). where A is the area.
The gross number of bubbles (cells) was calculated as N =
E niiA (number!cm 2 ).
Median diameter of bubbles (cells) was determined by use of
the normal probability paper. and was expressed as Dso (~m).
Standard deviation of bubble (cell) size distribution was
expressed by the normal function. LogO"'.
Bubble (before baking) and cell (after baking) area distribu-
tion
Bubble (cell) area distribution was calculated with the fol-
lowing bubble (cell) size ditribution. (Kri 2 ni)/A (cm 2!cm 2 ).
Porosity was calculated as P = (L1tri 2 ni)!A (cm 2 /cm 2 ).
Peak diameter was represented by that of area distribution.
D*(pm) •

RESULTS
Bubble structure before baking
Before baking. small bubbles were present: Bread. 1.5xl0 4 ;
cake. 2.8xl0 4 ; cookie. 3.5xl0 4 ; and chou. 1.9xl0 3 number!cm 2 ,
were entrained in those mixtures of ingredients. The bubble
size ranged from 2 to 300 pm in diamet~r (D~= 15-45 pm), and
the resultant frequency curves were approximately logarithmic
normal distributions (Log~~ 0.2). The porOSity of those
mixture varied widely: Bread, 0.047; cake, 0.656; cookie,
0.204; and chou, 0.051 (cm 2/cm 2 ).
Porous structure after baking
After baking, the cells in the baked goods decreased in number
to less than 5% of those before baking: Bread, 93; cake,
1.lxl03 ; cookie, 5.9xl0 2 ; and chou, 76 number!cm 2 • While the
 279
remaining cells expanded in the range of 30 pm - 30 mm in
diameter (050= 100-400 ~m) and formed larger products in total
volume. The distribution curves of the cell size tended to
extend long toward right side. The porosity of baked products
increased: Bread, 0.604; cake, 0.734; cookie, 0.545; and
chou, 0.953.

DISCUSSION
In the process of baking, over 95% of the bubbles entrained in
the mixture of ingredients vanished during heating. The
remaining 5% in those mixtures expanded and formed larger
products in total volume. When the change of the porosity in
the baking process was small, as with cake, the growth of the
remaining bubbles was considered to be mainly due to the
coalescence of bubbles. On the other hand, when the change of
the porosity in the baking process was large, as with chou and
bread, the growth was considered to be mainly due to water
vapor or other generated gas. In any case, the porous struc-
ture of baked goods was attributed to the bubble structure
before baking.

CONCLUSIONS
A lot of small bubbles, 2000 to 35000 /cm 2 in these baked
goods were entrained in the mixture of ingredients before
baking. The porous structure of baked goods was attributed to
the structure of small bubbles in the mixture.

REFERENCES
1. Noda, M., Foam and polymer. Kobunshi, 1970, 19, 2-8.
2. Mita, T. and Matsumoto, H., Relationship between internal
pressure and bubble size of fermented dough. Nippon Nogei-
kagaku Kaishi, 1978, 62, 111-116.
3. Shimiya, Y. and Yano, T., Analysis of early stage of ba-
king. New Food IndustrY, 1985, 27, 66-76.
4. Shimiya, Y. and Yano, T., Rates of shrinkage and growth
of air bubbles entrained in wheat flour dough. Agric. BioI.
Chem., 1988, 62, 2879-2883.
5. Baker, J.C. and Mize, M.D., The o~igin of the gas cell in
bread dough. Cereal Foods World, 1990, 36, 497-507.
 REDUCTION IN DAMAGE TO DRIED SULTANA DURING
REMOVAL OF CAP-STEMS.

M.R. MOLLAH·, R.J HAYES·, P.R. FRANZ", LV. GOULD'·'

• Sunraysia Horticultural Centre, P.O. Box 905, Mildura 3502,
•• Biometrics, P. O. Box 500, East Melbourne 3002,
••• Food Research Institute, Princes Highway, Werribee,
Department of Agriculture, Victoria, Australia.

ABSTRACT

In order to improve the quality of Australian dried sultanas by reducing mechanical damage
to berries during cap-stemming, an investigation into important machine settings and design
parameters was conducted on a full size (15 t/h) experimental coning machine. In 1992 cone
speed, cone clearance, fruit feed rate, angle and length of outer cone were investigated at two
levels in trial 1. Eight combinations of rub-bar type and number were also investigated in
trial 1. Trial 2 was designed based on the findings of trial 1. This paper presents details of
1992 trials with results. Further trials will be conducted in 1993 over a wider range of cone
speeds and clearances, fruit temperatures and types.

INTRODUCTION

An important step in processing of Australian dried sultanas is the removal of pedicels (cap-
stems) from the berries. The majority of the cap-stems are currently removed when berries
are fed through a gap between an inner revolving cone, enclosed by an outer stationary cone
[1]. Australian sultanas are dipped in an alkaline oil-in water emulsion at harvest to alter the
cuticular wax structure and speed up drying rates, producing light coloured fruit which are
also susceptible to mechanical damage [1,2]. The coning process causes significant damage
to the berry skin, resulting in sugars spreading throughout the product mass [3]. This impedes
the free flowing of berries and causes problems in subsequent processing or repacking,
resulting in some end point buyer resistance.
Results of 1992 trials are presented which show that selected cone speeds, clearances and
configuration of rub-bars on the inner cone will improve coning performance, but further
work is required.
 281

MATERIALS AND METHODS

The figure 1 shows a view of the experimental coning

(cap-stemming) machine used in the trials, and is similar
to existing coning machines in Australian packing
plants.
A fractional factorial design with 1A replication was
used in trial 1 to identify the important parameters for
efficient cap-stemming with minimal damage on trellis
dried fruit. The parameters of the cone investigated were
cone speed (250 & 400 rpm); cone clearance (15 & 32
mm); 8 configuration of continuous (12 mm & 25 mm
thick), short and offset (12 mm thick), and helical (12
mm thick) rub-bars at two spacings; fruit feed rate ( 5
Figure 1. Experimental coning t/h & 10 t/h); angle of outer cone (parallel and variable
machine to inner cone); and length of outer cone (full & two-
third).
For trial 2, cone setting was fixed at 12 continuous (12 mm thick) rub-bars & full parallel
outer cone, a setting which showed promise in trial 1. Fruit feed rate was 10 t/h as there was
no significant effect of feed rate on cap-stemming and damage. A randomised block design
with 4 replications was used to further investigate the two most important factors, cone speed
and clearance at 4 levels (375, 400, 425 & 450 rpm and 15, 20, 25 & 30 mm) and to compare
rack and trellis dried fruits.
The number of cap-stems present and degree of damage were assessed on samples taken
before and after cap-stemming. The moisture content of the sample before processing was
determined and the fruit temperature for each run was recorded. Damage was assessed using
an industry standard ferrous sulphate damage index test. Cap-stems retained by berries were
manually counted for each sample of 500 berries.

RESULTS

Table 1 and figure 2 show results of trial 1 and 2 respectively. Twelve 760mm x 25mm x
12mm deep continuous rub-bars on the inner cone produced better cap-stemming (87%) with
60' damage index. Significantly better cap-stemming with minimum damage occurred for
both rack and trellis dried fruit at 20 mm clearance with 400-425 rpm cone speed.

TABLE 1
The mean effects for some cone parameters on cap-stemming and damage done to sultanas

Speed(rpm) Clearc.(mm) Feed rate(tlh) Outer cone angs. Cone length

250 400 15 32 5 10 para. vari. full 2/3
C.S. 36.9 14.6 21.7 29.8 24.6 26.8 21.3 30.2 23.5 28.0
OJ. 45.6 50.9 49.6 46.9 49.5 47.04 50.9 45.6 48.4 48.0
C.S.=% of cap-stem present; D.I.=<Iamage mdex; LSD(p=0.05)=4.06 for C.S. & 3.76 for DJ.
* : OJ. of 33-60=excellent; 60-70=very good; 70-80=good; 80-90=poor; 90-100= very poor.
282
DISCUSSION

From both trials it is clear that cap-stem removal and

Figure 2. The mean effects for
cone speed in trial 2.
berry damage increase with increasing cone speed and
decrease with increasing cone clearance. An optimum
combination of cone speed and clearance with particular
rub-bar configuration on the inner cone is required for
efficient cap stemming with acceptable damage.
Trial 2 showed that trellis dried fruit had marginally
more cap-sterns present and was significantly more
damaged after processing than rack dried fruit. This
finding contradicts the popular belief - that trellis drying
produces darker and tough fruit which is less susceptible
to damage. Detailed investigation is needed before
making any conclusions on this phenomenon.
The moisture content and temperature of sultanas, cap-
stems remaining and berry damage before processing
were tested as covariates, and showed no significant
effects. Further work with a wider range of cone speed,
clearance, temperature and fruit type is required before
making any recommendations.
CONCLUSIONS

Trials conducted in 1992 show that selected cone speeds, clearances and configuration of rub-
bars on the inner cone improve cap-stem removal and berry damage on sultanas, but more
work is required before making any recommendations.

ACKNOWLEDGMENTS

The authors wish to thank the following members of Department of Agriculture, Victoria for
their contribution: Mr. B.J. Benham -machine construction and assistance with its design; Mr.
B.G. Morey -help with the trials; Mrs. L. Jacka -Graphic design; and Dried Fruits Research
and Development Corporation, Australia for funding the project.

REFERENCES

1. Simmons, I.D., Brien, c.J. and May, P., Processing of dried sultanas - The effect of
temperature and machine settings. Confructa, 1979, 24, 28-37.

2. Dried Fruit Processing Committee, Grape Drying in Australia - Production of sultanas,

raisins, and currents. Sunraysia Daily, Mildura, Australia, 1982.

3. Tarr, C.R., Investigations into stickiness and compaction of dried vine fruit. Research
Report Series No. 99, Department of Agriculture, Victoria, Australia, Oct. 1989.
 THERMAL PROPERTY MEASUREMENTS OF FRIED FOODS USING
DIFFERENTIAL SCANNING CALORIMETER

A.B.BUHRI AND R.PAUL SINGH

Department of Biological and Agricultural Engineering
University of California, Davis, CA 95616, USA

ABSTRACT

Thermal conductivity and specific heat are two important properties necessary in the
understanding of the heat transfer phenomenon in frying of foods. A food model containing
potato starch was fried in palm oil. A Differential Scanning Calorimeter (DSC) was used to
measure the specific heat of the crust (2.93 kJ/kgK @ 110 .C) and the center of the fried
sample (3.62 kJ/kgK @100 .C). Using a new DSC procedure the thermal conductivity for the
inner core was measured to be 0.644 W/mC@ 90 ·C and for the crust layer 0.13 W/mC @
100 ·C.

INTRODUCTION

In designing and evaluating thermal processes, knowledge of food properties is essential. For
example, properties such as thermal conductivity, specific heat, thermal diffusivity give an
indication of how a food material will heat when exposed to a given environment that is at a
temperature different than that of the food material. Experimental determination of thermal
properties of foods has been a subject of numerous studies [1,2,3,4,5,6]. The overall objective
of the research reported in this paper was to use a newly developed method using a Differential
Scanning Calorimeter (DSC) to measure specific heat and thermal conductivity of foods,
particularly small size samples obtained from food materials that are fried in an immersion oil
fryer.

MATERIALS AND METHODS

A potato model system containing potato starch, salt and a preservative was mixed with water
at 62·C to produce a potato mash. The mash was shaped into 54 mm diameter spheres. The
sphere was then immersed in a 190·C palm oil bath for 10 and 20 minutes. The fried sample
was then cooled on a wire mesh. A cylindrical sample containing the crust, and a sample of
material from near the center of the sphere were taken and prepared for the thermal properties
measurement. A Differential Scanning Calorimeter (Perkin Elmer, DSC-7) was used for these
studies. To measure specific heat, a small sample (10 g) was placed into an aluminum sample
284
holder and sealed. The sealed sample was placed into the DSC heating pan. The samples
obtained from the interior of the sphere were heated from 40 to 100·C, while the crust samples
were heated from 100 to 150·C with a temperature increment of lOT per minute.

The DSC has the capability of providing an accurate measure of heat flow through a sample
placed in the heating pan. In order to use this feature to measure thermal conductivity, an
attachment was constructed [1]. The food sample was placed in a sample holder (6.5 mm
diameter and 12.7 nun long) with an aluminum bottom. The sides of the sample holder were
made of a phenolic resin of low thermal conductivity (k =0.04 WfmC). A thermocouple probe
(40 A WG) located inside a 9mm long 20 AWG stainless steel needle could be inserted into the
sample up to any desired depth. The thermocouple probe was inserted into the sample to a
depth of 7.7 nun for the measurement of the thermal conductivity of the interior samples. For
the crust sample, the probe was inserted to the depth of 11.7 mm such that the recording
thermocouple junction was located in the crust layer. The depth of the insertion was measured
by using a micrometer scale. The DSC heating pan was kept at the desired initial temperature
(either 60, 70, 80, 90, 100 or 1 10· C). After 5 minutes, the initial temperature of the sample
was recorded. The pan temperature was immediately raised by 10·C and the final steady state
temperature at the thermocouple junction was recorded. All property measurements were
conducted in triplicate. The following equation is used to obtain thermal conductivity.

where L is sample length; ~Q is the difference in energy required to maintain the pan
temperature; A is the sample area perpendicular to the heat flow; ~T2 is the final temperature
difference between the DSC heating pan and the sample; ~ T 1 is the initial temperature
difference between the DSC heating pan and the sample.

The moisture content of all samples were measured using a vacuum oven at 70·C for 24 hours.
The density was measured using an air comparison pycnometer.

RESULTS AND DISCUSSION

TABLE!.
The specific heat of the fried potato-mix

Temperature Unfried Potato-mix Fried Interior Contents Crust

rC) (J/kgK) (J/kgK) (J/kgK)
50 3604 3320
60 3588 3377
70 3577 3493
80 3691 3522
90 3729 3565
100 3740 3618
110 2930
120 3032
130 3105
140 3130
150 3136
 285
TABLE II.
Thermal Conductivity of the Fried Potato-Mix

Temperature Sample Location Thermal Conductivity Moisture Content

CC) CW/mC) (%wet basis)
70 Center 0.6554 69.6
80 Center 0.6478 69.6
90 Center 0.6440 69.6
100 Crust 0.1521 4.2
110 Crust 0.1398 4.2
120 Crust 0.1302 4.2

The values of specific heat determined in this study are presented in Table I. The specific heat
of the cooked internal samples are presented with values for the unfried potato mix. These
values are considerably larger than the fried crust samples which had higher oil content and
considerably low moisture content. Using these data, two predictive equations may be obtained
as follows:

cp = 3036.31 + 5.951 T for 50·C<T<100·C

cp = 2403.6 +5.06 T for 100·C<T

The thermal conductivity values (averages of triplicate determinations) are shown in Table II.
The thermal conductivity of the crust layer was found to be much lower than that of the interior
contents of the fried sample. These results indicate that the crust layer has a much lower
thermal conductivity and therefore provides considerable resistance to heat transfer from oil
into a food object during frying.

In summary, a newly developed procedure using a DSC to determine thermal properties of

relatively small size samples was used to measure properties of different components of a fried
food product. Knowledge of these property values is essential when analyzing heat transfer
during frying of foods.

REFERENCES

1. Buhri A.B. and R.P.Singh. 1992. Measurement of food thermal conductivity using the
differential scanning calorimeter (DSC). Paper presented at IFT Annual Meeting (June 20-
24) New Orleans.
2. Califano, AN. and Calvelo, A 1991. Thermal conductivity of potato between 50 and 100
.c. J. Food Sci. 56:586-587.
3. Drouzas, AE. and Saravacos, G.D. 1988. Effective thermal conductivity of granular starch
materials. 1. Food Sci. 53:795
4. Griffith, C.L.1985. Thermal conductivity, density and thermal diffusivity of Mexican
Tortillas Dough. J. Food Sci. 50:1333-1337
5. Wang D.Q. and Kolbe, E. 1990. Thermal conductivity of Surimi- measurement and
modeling. J. Food Sci. 5:1217
6. Wang, D.Q. and Kolbe, E. 1991. Thermal properties of Surimi analyzed using DSC. J.
Food Sci. 56:302-308
 COLORIMETRIC MEASUREMENTS OF MEAT PRODUCTS
COOKED USING DIFFERENT SYSTEMS.

P. Pittia, M. Anese, C. Orlando and A. Sensidoni

Department of Food Science, University of Udine
via Marangoni 97 33100 Udine (Italy)

INTRODUCTION

Cooking is a thermal process which causes chemical and physical changes in the product and which
finally leads to an edible and digestible food. Meat products cooked in different methods present
different organoleptic and rheological properties, which depend on the heating mechanism involved
[1,2]. The most important effects of heat treatments on meat are weight loss, protein denaturation
and colour changes [3]. As regards colour changes, the development of browning of meat and meat
derivatives during cooking is generally attributed to myoglobin denaturation, sugar caramelization
and Maillard reaction [4}.
In this paper, the results of a preliminary study on protein digestibility and colour changes ofmeat
products in relation to three different cooking systems can be found.

MATERIALS AND METHODS

Model meat pasties of about 100 g with a spherical shape (about 3 em in diameter and 1 em in
thickness) were prepared by mixing deboned minced chicken with eggs and lard. Raw samples were
heated using commercial infrared, microwave and stearn ovens. The set temperatures were of
135°C and 330°C for the steaming and infrared heating respectively; in the case of microwave oven
the operating power level was 500 W. At different and prefixed cooking times, samples were taken
from the ovens and tested. The lengths of time were not the same for alI the heating processes,
because oftbe different mechanism of heating involved. Dry matter and lipid content of the samples
before and after cooking were evaluated according to AOAC methods [5]. Colorimetric
measurements were carried out using a tristimulus colorimeter (Chromameter-2 Reflectance,
Minolta, Japan) according to Mastrocola and Lerici [6}. Colour of the sur&ce and of the inner
portion of the meat pasties is expressed in L* (lightness), a* and b* (chromaticity coordinates).
Protein digestibility of samples was evaluated following the multienzymatic method of Hsu et al.
[7}. Data are the average of three repetitions.

RESUL TS AND DISCUSSION

Weight loss, dry matter and lipid content of samples before and after heating are reported in Table
1a. It is possible to note that in the case of steaming the weight loss was very slight.
Examples of changes in hydrogenionic concentrations versus time observed during incubation of
the homogenized raw and heated meat with multienzymatic system are reported in Figure 1.
 287
TABLE Ia. TABLE lb.

Cooking [Heating time Weight loss Dry matter Lipids Kx 10-9 r2

system (min) (%) (%) (% on dry ([H+] I min)
matter)
Raw 0 0 50.8 35.9 13.27 0.965
Infrared 15 20.1 50.8 29.0 1.13 0.971
20 26.7 54.1 30.6 7.36 0.950
25 37.0 65.8 28.3 4.36 0.996
Microwaves 5 23.2 59.5 30.7 3.03 0.994
7 33.3 63.7 30.7 4.45 0.986
10 51.7 83.4 27.6 6.21 0.995
Steam 10 2.6 48.1 37.7 1.88 0.988
15 7.0 46.7 34.7 7.52 0.962
20 8.7 46.1 34.4 9.05 0.956

8
_7
co 6
6, 5
~ 4
:::::3
;j; 2
~ 1
o~----------~------~----------~
o 2 4 6 8 10
Incubation time (min)

Figure I. Changes in bydrogenionic concentration versus incubation time of the homogenized raw
(-) and cooked meat samples (Infrared. 25 min 0; Microwaves, 7 min +;
Steam, 20 min 0).

Zero order kinetic rate constants, calculated from linear regression analysis of H+ concentrations
versus incubation time, and their correlation coefficients (r2) are reported in Table lb.
Heat treatment can improve the digestibility of proteins by opening the protein structure through
denaturation. However, short cooking times could cause the formation of "anomalous" protein
structures responsible for the lower digestibility of samples compared to the raw ones [3].
L * values of the surface and the inner portion of the meat pasties differently processed as a function
of the cooking time are reported in Figure 2.
It is possible to note that, only in the case of infrared heating, a great difference in L * values
between the outer and inner portion can be observed. In Figure 2 it is also possible to note that
samples cooked in the tnicrowave oven presented a darker colour in the inner portion than on the
surface after excessive cooking (10 min). Figure 3 shows the difference of cromaticity index A vs.
kinetic rate constant values of the protein digestibility. A index was calculated as follows:

A = [(a*-a*0)2 + (b*.b*o}2]112

where a* and b* are related to the cooked sample and the a*o and b*o to the raw sample.
288
80~-----------'------------'--------4

70
- 60
d<Zl 50
g; 40
~
~ 30
OD
;.:J 20 Steam
Infrared Microwaves
10
O~-+--~--r-~--+-~--~--~~--+-~

o 15 20 25 0 5 7 10 0 10 15 20
Heating time (min)

Figure 2. Changes in L* values as a function of heating time of the outer (open symbol) and inner
portions (solid symbol) of meat samples.

12
-<
~ 10
Q)
~
.S 8
>-.
-'
·s 6
:::;
""
S 4
a
'-
..c: 2
u
0
0 2 4 6 8 10
K X 1O- 9 ([H+] I min)
Figure 3. Chromaticity index values (A) of meat samples cooked using Infrared (0), Microwaves
(.) and Steam (.) systems for different times versus the correspondent kinetic rate constants.

It is possible to note that both digestibility and colour development depended on the nature of the
heating process. In fact, in the case of the infrared cooking method a significant increase in protein
digestibility corresponded to little change in colour, while different results were observed for the
microwave system.

REFERENCES

l. Unklebay, N. Davis, M.E. and Krause, G., Nutrient retention in pork, turkey breast and corned
beefroasts after infrared and convective heat processing. 1. Food Sci., 1983,48,865-904.
2. Janicki, L.J. and Appledorf, H., Effect of broiling, grill frying and microwave cooking on
moisture, some lipid components and total fatty acids of ground beef.J. Food Sci. 1974, 39, 715-17.
3. Mariotti, R., PageUa A, Effetto dei trattamenti termici sui nutrienti piu significativi della came.
In Monografia N. 16 CNR-IPRA 1987, pp.247-69.
4. Lawrie, R.A, The eating quality of the meat. In Meat Science. Pergamon Press, 1979, pp.300-
06.
5. AO.A.C. Official Methods of Analysis, 1980.
6. Mastrocola, D. and Lerici, C.R., Colorimetric measurements of enzymatic and non-enzymatic
browning in apple purees. Ital. J. Food Sci., 1991,3,219-29.
7. Hsu, H.W., Vavak, D.L., Satterlee, L.D. and Miller, G.A, A multienzyme technique for
estimating protein digestibility. 1. Food Sci., 1979,42 (5), 1269-73.
 INTERACTIONS BETWEEN MAILLARD REACTION PRODUCTS AND LIPID
OXIDATION DURING THE ROASTING PROCESS AND STORAGE OF
HAZEL-NUTS (Corylus ave17ana)
SEVERINI C.*, PITTIA P.**, NICOLI M.C.**, PINNAVAIA G.G.*
* Dipartimento di Protezione e Valorizzazione Agro-
Alimentare. Sez. Chimica e Tecnologia degli Alimenti
Universita' di Bologna - ITALY -
** Dipartimento di Scienze degli Alimenti
Universita' di Udine - ITALY -

ABSTRACT
Raw hazel-nuts and the oi 1 extracted from the raw nuts were
heat treated at 150°C in a laboratory scale plant to simulate
a roasting process. Measurements of colour and HMF content
were performed to evaluate the non enzymatic browning. On the
treated oi 1 and on the oi 1 extracted from the roasted nuts I

determinations of lipid oxidation using chemical and

instrumental methods were carried out to establish the
antioxidative effect of the Maillard reaction products (MRP)
formed during the thermal process. After roasting, the
granulated hazel-nuts and the extracted 011 were stored at
50°C to follow the development of 1 ipid oxidation. Results
obtained in prel iminary trials show how the 1 ipid oxidation
of the oil in the nuts was reduced in comparison to oxidation
in raw nut oil subjected to the same heat treatment.
INTRODUCTION
During thermal_ treatments and processing of foods many
chemical and physico-chemical modifications take place.
The hazel-nuts roasting process, as known, is carried out at
high temperature (120°-140°C); in these conditions the nuts,
containing a high fat content, undergo some alterative
processes such as Maillard reaction and lipid oxidation. The
antioxidant effect of non enzymatic browning products is well
known (1, 2, 3); this effect on a vegetable product such as
hazel-nuts, is studied in this paper.
MATERIALS AND METHODS
The roasting is carried out using a thermostatic stove with
air circulation. (a hot-air circulation furnace)(150 °c,
60').The oil of the nuts was extracted by a cold method with
chloroform. The following analytical determinations were made
on the nuts: Aw, using an electronic hygrometer; total lipids
content, by the Soxtec method; colour (L*,a*,b* scale) using
a Tristimulus Colorimeter; HMF content by the HPLC method.
290
on the extracted oil (2).
RESULTS AND DISCUSSION
The raw hazel-nut characteristics, reported in table 1, show
an intermediate Aw level in which the Mai llard reaction is
increased and lipid oxidation is not inhibited. Moreover the
high content of 1 ipids is evident. Under the roastin
conditions used in this experiment (150'C, 60'), lipid
oxidation was initiated at the same time as the Mai llard
reaction, conferring the typical organoleptic characteristics
to roasted hazel-nuts.
TABLE 1
Same chemical-physical characteristics of raw hazel
-nuts and extracted oil.
HMF L* Aw Lipids% PN TBA
(mg/100g) (mg 02/Kg)
0.0 63.7 3.6 19.4 0.7 42.71 0.0 0.155

In table 2 there is an increase in the HMF value and a

decrease in the L* value, both indicators of the initiation
of the Maillard reaction. In the same way, the lipid
oxidation indicators (PN and TBA) increased in comparison to
the raw material. In fact, the HMF, colour, PN and TBA
values, reported in table 1, show that the Maillard reaction
and lipid oxidation had not yet started.
TABLE 2
Maillard reaction and 1 i pi d oxidation indicators in
roasted hazel-nuts and extracted oi l.
HMF L* a* b* PN TBA
(mg/100g) (mg02/Kg)
0.413 42.42 1.55 30.74 13.57 1. 11

Hazel-nuts oil extracted before roasting and after roasting,

was stored at 50'C for more than 60 days. Figure 1 shows the
progress of oxidation (PN versus time) during the period of
storage.
It is evident that the oil treated at high temperature
without MRP is deteriorated much more compared with oil
treated in the presence of the same products.
 291

---
120

loft. rOGltlnv
---
WfttIaut WRP
i J

"-
g 90
Wlffa ...

~
-
f
eo

I
t~ 30

.tor.
0
10 20 .50 40 50 60 70
I~] 0 time at 50 ....C (dap)

Figure 1: Peroxide number(mg 02/Kg) in oil samples treated

at 150°C for 60' versus storage time at 50°C, with and
without Maillard reaction products (MRP).
In conclusion, we can evaluate the antioxidant effect of
Maillard Reaction Products on a food system; in this case,
the heat treatment produces, simultaneously, favourable
conditions for lipid oxidation and antioxidant products which
prolong shelf-life.
REFERENCES
1) Elizalde B.E., Dalla Rosa M., Lerici C.R., Effect of
Mai llard Volati 1e Products on 1 ipidic oxidation. In JAOCS,
1991, 68, 10, 758-762.
2) Kim N.R., Harris D.H., Antioxidant effect of enzymatic
browning reaction products on linoleic acid. In Trend in Food
Science, 1989, (ed) Ghee A.H., Woo F.C., 19-24, Singapore
Institute of Food Science and Technology, Singapore.
3) Tanaka M., Kuei W.C., Nagashima Y., Taguchi T.,
Application of Antioxidative Maillard Products from Histidine
and Glucose to Sardine Products. In Nippon Suisan Gakkaishi,
1988, 54, 8, 1409-1414.
4) Norme ita1iane per il contro1lo di Grassi e Derivati, NGD
C 35, 1976.
5) Hami lton R.J., Rosselj B., Analysis of oi ls and fat.
Elsevier 1986.
 A NEW METHOD OF MONITORING PARTICLE MOTION AND
TEMPERATURE IN A LIQUIDIPARTICLE SYSTEM

H. Sawada* and R. L. Merson

Department of Food Science and Technology,
University of California, Davis, CA 95616, U. S. A.
* House Food Industrial Co., Ltd.,
1-5-7, Mikuriyasakaemachi, Higashiosaka, Osaka 577, Japan.

ABSTRACT
Spherical gellan gum gel particles in an aqueous medium were heated in a clear acrylic can
under axial rotation. Thermochromic liquid crystals (TLCs) dispersed in a few gel particles
provided a means for remotely monitoring temperature changes in the particles. Because of
matching refractive indices, the particles without incorporated TLCs became virtually
invisible in the aqueous medium and allowed uninterrupted observation of the TLC-
containing particles. The ranges of temperatures converted from colors of digitized images
of the particles as well as the duration of the colors in the particle were in fair agreement
with the limits of mathematically predicted time-temperature curves.

INTRODUCTION
Researchers studying heat transfer during thermal processing of liquid/particle systems
encounter difficulties in measuring particle temperatures without restricting free motion of
the particles. Thermochromic liquid crystals (TLCs) have been used for remote sensing of
particle surface temperature [1, 2, 3]. However, those studies were limited to cases where
only one or a small number of particles were present in the system. This work describes a
new approach in the use of TLCs for observation of temperature change in a particle
surrounded by multiple particles.

MATERIALS AND METHODS

Spherical gel particles of nominal diameters 1/2 inch (12.7 mm) and 1 inch (25.4 mm) were
prepared from gellan gum (Kelco Division of Merck & Co., Inc.). To make 1/2 inch
particles, 1.07 ml each of heated gum solution containing 0.25% gum and 0.1 % CaCh was
dispensed dropwise into a chilled column of vegetable oil. Interfacial tension kept the
shape of the drops spherical until gellation. One inch particles were formed individually in
a mold. Particles for monitoring temperature changes (sensor particles) were prepared by
dispersing microencapsulated TLC slurry (Hallcrest Products, Inc.) in the gum solution at a
concentration of 0.25% before forming particles. Three sets of sensor particles having
different temperature ranges for color change (30 - 35°C, 39 - 41°C and 45 - 47°C) were
 293
prepared.
A cylindrical can (size 303 x 406) made of clear acrylic plastic was filled with 0.1 %
CaCh solution and a calculated number of particles (including two particles from each set
of the sensor particles) to a designated void fraction (e=O.6 or 0.8). The can was heated
under axial rotation (55 rpm or 101 rpm) in a 50°C water bath.
The entire contents of the can were videotaped with a VHS camcorder during heating.
Two slide projectors with a DAK projector lamp (3200 0 K) illuminated the can from two
sides at 45° angles with respect to the direction of observation. Polarizing filters mounted
on the camcorder and the projectors reduced specular reflection of the illumination light at
the can surface and improved the visibility of the sensor particles. When the tape was
played back, fields of images for various heating times were digitized on a personal
computer. Hue values in an HSL color coordinate system were then read from selected
pixels of the digitized images using a program developed by Dr. David C. Slaughter of the
Department of Agricultural Engineering, University of California, Davis. The hue values
were converted to temperatures using equations determined from separate calibration
experiments.
Particle temperatures converted from TLC colors were compared with time-
temperature curves predicted by a mathematical model. The heat transfer coefficients were
estimated from only liquid temperature data taken during the same heating experiment.
Details of the model and the method for estimation of heat transfer coefficients are given in
Sawada [4].

RESULTS AND DISCUSSION

FIGURE 1 shows a digitized image of the can containing 1/2 inch particles at e=O.6. Of the
176 particles in the can, only the 6 sensor particles were clearly visible. Because of the
matching refractive indices, the particles without TLCs became nearly transparent in the
CaCl2 solution. Thus, it was possible to observe movement and color of the sensor
particles without significant interference by the particles around them.

FIGURE 1. A digitized image of a can containing 1/2 inch particles at a void fraction (e) of
0.6, and being heated under axial rotation at 55 rpm.

The transparent gel matrix of the sensor particle allowed the microcapsules of TLC
distributed throughout the particles to reflect the illumination light selectively and to show
colors depending on the temperature of their location. Since the TLCs used in this study
had narrow temperature ranges for color change (2-5°C), we were able to obtain semi-
quantitatively such information as symmetry and relative steepness of the temperature
294
gradient in a particle (from symmetry and width of the colored region, respectively) as well
as uniformity of heating in the can (from comparisons of colors in the two sensor particles
having the same temperature range).
On the other hand, because of the temperature gradients along the radius of the
particle, the colors we saw from the surface were mixtures of different colors reflected at
various radii in the particle. Being unable to determine true colors, and hence temperatures,
of specific locations within the particle, we compared only the maximum and the minimum
temperatures experimentally determined from particle images taken at various heating
times with the surface and the center temperatures predicted mathematically. As shown in
FIGURE 2, both the ranges of experimentally determined temperatures for a given heating
time and the time periods during which the sensor particle showed colors were in fair
agreement with the limits of those expected from the predicted temperature curves. Since
no particle temperature data obtained from TLC colors were used in the mathematical
prediction, the agreement was considered as a good check on the mathematical model and
the heat transfer coefficients estimated from the liquid temperature data.

50 50

~
(a) (b)
~45
()
~
()
45
~ ~
~ 40 ~ 40 _"l~-·
:::::J :::::J
...
Cii ~ ///
2i 35 0 Experimental Max. 2i 35 o Experimental Max.
E E
Q)
I- 30
• Experimental Min .
Predicted Surface
Q)
I- 30
• Experimental Min.
Predicted Surface
Predicted Center Predicted Center
25 25
0 600 1200 0 600 1200
Time (sec) Time (sec)

FIGURE 2. Comparisons of experimental and predicted temperatures for (a) 1/2 inch and
(b) 1 inch gel particles, heated at void fraction (e) of 0.8 and rotation speed of 55 rpm.

From these results we may conclude that the method described here could be a useful
tool for investigation of heat transfer in liquid/particle systems at low void fractions.

REFERENCES

1. Balasubramanium, V. M., Heskitt, B. F. and Sastry, S. K., Determination of fluid to

particle convective heat transfer coefficient in continuous flow using liquid crystals.
Paper No. 462, presented at the 52nd Annual Meeting of Institute of Food
Technologists, Dallas, Texas, June 1991.
2. Stoforos, N. G. and Merson, R. L., Measurement of heat transfer coefficients in rotating
liquid/particulate systems. Biotechnol. ~1991, 7, 267-271.
3. Stoforos, N. G., Park, K. H. and Merson, R. L., Heat transfer in particulate foods during
aseptic processing. Paper No. 545, presented at the 50th Annual Meeting of Institute of
Food Technologists, Chicago, IL, June 1989.
4. Sawada, H., An analytical heat transfer model for liquid/particle systems. Ph.D. Thesis,
University of California, Davis, California, 1992.
 CO-ROTATING DISC SCRAPED-SURFACE BEAT EXCHANGER FOR
FOOD PROCESSING

ALAN FRIIS, JENS ADLER-NISSEN & OLE HASSAGER·

Center for Food Research, Department of Biotechnology, block 221
~epartment of Chemical Engineering, block 229
Technical University of Denmark, DK-2800 Lyngby, Denmark

ABSTRACT

A laboratory scale prototype ofthe "Co-Rotating Disc Scraped-Surface Heat Exchanger"

(CD-HE), which represents a new design in scraped surface heat exchangers, is currently
being investigated. The CD-HE is characterized by having a much larger surface-to-volume
ratio than traditional scraped surface heat exchangers and it is especially applicable for proces-
sing of viscous food products. Results show that the CD-HE has a high heat transfer capacity
and promising Residence-Time-Distributions have been achieved.

INTRODUCTION

The purpose of the present work is to extend basic knowledge about process equipment for
continuous pasteurization and sterilization of viscous food products. This is done by con-
struction of laboratory scale heat exchangers and investigation of their performance for the
purpose of building mathematical models for scale up of equipment. The benefits from work-
ing with laboratory scale equipment are the need for a small amount of product when perform-
ing experiments and having equipment that is easy to redesign and scale up [1]. The objective
of the investigation is the "Co-Rotating Disc Scraped-Surface Heat Exchanger" (CD-HE),
which is a new type of scraped-surface heat exchanger (SS-HE) [2] especially suitable for pro-
cessing of viscous liquid food products.

EQUIPMENT AND METHODS

Design of the CD-BE

The main characteristics that differentiate the CD-HE from the traditional SS-HE are that the
CD-HE has a stationary scraping device, no mixing device and rotating heating surfaces. The
first working prototype of this kind in the world was build by the inventor Zehev Tadmor,
296
Dept. of Chemical Engineering, Technion, Israel Institute of Technology, Haifa, Israel and
Jens Adler-Nissen in co-operation. The CD-HE is intended to be constructed as a multi
chamber model. However, this prototype only has one chamber. This is sufficient since the
purpose is to gain basic knowledge about the flow inside the processing chamber.
The processing chamber has the form of an annular channel with a rectangular cross sec-
tion. The processing chamber is 4.75 nun broad, 24.5 mm in height and has a volume of27 mI.
The surface-to-volume ratio of the CD-HE is more advantageous than it is for traditional SS-
HE in the case of scale up. The ring-shaped channel is blocked by a teflon block to separate
the inlet and the outlet port. This blockade does also serve as a stationary scraping device.

Experimental Conditions
In operation the heating surfaces can be rotated in either direction. This gives two modes of
operation one where the heating surfaces are rotated co-currently to the direction of the flow
of product through the processing chamber and one having counter-current rotation.
Experiments have been carried out using water and a blend of86.5% w/w glycerol and
13.5% water (86.5% glycerol) as model fluids.
When processing food products it is important to have a measure of how homogenous
the heat treatment is. This is done by determining the Residence-Time-Distribution (RTD).
The RID is measured by injecting a concentrated portion of coloured tracer in the inlet port
and analysing the variation in concentration as a function of time in the outlet port by
spectrophotometry.

RESULTS AND DISCUSSION

Flow Properties
We assume that the flow in the processing chamber of the CD-HE is laminar in the processing
of viscous fluids. To illustrate the validity of this assumption we compute the Reynolds
numbers for the flow based on a rectangular cross section.
Due to rotation of the discs there is a significant velocity gradient across the processing
chamber. The velocity near the edge of the discs can be up to 10 times higher than the mean
velocity of the flow. Therefore this has to be taken into account when characterizing the type
of flow. To give an upper limit of Reynolds number initial calculations are based on physical
properties of water. Here Reynolds numbers as high as 10.000 can be obtained near the discs.
When using more viscous fluids as 86.5% glycerol the Reynolds numbers are approxi-
mately a factor 30 lower due to the higher viscosity. For fluids of this type the Reynolds
number does not at any position in the channel exceed 500 even at temperatures as high as
150°C. Hence we conclude that our assumption oflarninar flow is valid.

Heat Transfer
As a measure of evaluating the energy efficiency of the CD-HE calculation of the overall heat
transfer coefficient is used. The overall heat transfer coefficient is calculated on basis of the
energy transferred to the fluid in the processing chamber.
In experiments where water has been used values of the overall heat transfer coefficient
in the range 400 - 700 W/(m2J<) have been obtained for a wide range of working conditions.
When using 86.5% glycerol as model fluid and operating the CD-HE in counter-current
rotation mode with a rotation speed of 100 rpm the values of the overall heat transfer coeffi-
cient in the range of375 - 500 W/(m2J<) has been obtained, as shown in figure 1.
 297
500 I I I I I
~I
L""'-'
475 • • -
•
Q)"'-"" I-
'+-
(/)~
~ Temperature of
c E heating medium:
0
L
'--' 450 - -
I- ..........
0 0 115°C
.....0 ~

Q)
:r:
L..-.J

......
c
425 - 0
0 8
- • 135°C

•
Q)
u 400 -
•C0
I-
0
L
;;:: 0
'+-
Q) Q)
>
0
0
u 375 I- -
350 I I I I I I

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Flow rate of product [kg/min]

Figure 1. Overall heat transfer coefficient when 86.5 % glycerol is used as model fluid. The
heating surfaces of the CD-HE is rotated in counter-current rotation at 100 rpm.

The fact that overall heat transfer coefficients this high can be obtained in laminar flow are re-
markable. In turbulent heat exchangers typical values lie in the range of700 - 1100 W/(m~).

Residence-Time-Distributions
The spread in residence time has to be as small as possible to secure a homogeneous heat
treatment. The RID curves obtained in experiments using 86.5% glycerol are promising con-
cerning the spread in residence time. However, the holding time for the fastest moving par-
ticles are insufficient for pasteurizing of food product. This problem arises from having only
one processing chamber and the RID will most certainly prove to be better in a multi chamber
model.

CONCLUSION

The CD-HE has a very high heat transfer capacity. In addition to this the residence time
conditions are promising and in a multi chamber CD-HE it will be possible to obtain the hold-
ing times needed in pasteurizing offood products. This indicates that the CD-HE can prove
to be a useful substitute for traditional heat exchangers in cases when there is a need for
special care with respect to mechanical processing.

REFERENCES

1. Adler-Nissen, 1.; Apparatus for Down-Scaling ofHTST-Processes. Conference

Proceedinss COST9I-bis Concludina Seminar. Gothenburg, Sweden, October 2-5 1989,
Elsevier, Amsterdam.

2. Tadmor, Z.; Corotating-Disk Scraped-Surface Heat Exchanger. EQQd Technolosy, 1985,

Dec., 67-73.
 HEAT TRANSFER IN BAKERY OVENS

Robert E. Altomare
Kraft General Foods
General Foods USA
555 South Broadway, Tarrytown, NY 10591 USA

ABSTRACT

A wide assortment of ovens are available to the commercial baker. These include rack, reel, deck,
continuous band, direct gas-fired tunnel ovens, tunnel ovens with enhanced convection and
impingement or high velocity convection ovens. To highlight similarities and differences, heat transfer
characteristics and air temperature profiles in common home and commercial ovens are discussed in
this paper. Heat transfer coefficients were measured experimentally by heating blocks of aluminum
in the ovens under test. Ovenable data collection systems recorded pertinent temperature-time data.
A semi-log plot of dimensionless temperature against time was used to evaluate the heat transfer
coefficients. Coated blocks of known emissivity were used to determine the separate radiation and
convection contributions. Home electric and rotary reel ovens had similar heat transfer behavior. For
oven temperatures of 150-200"C, overall heat transfer coefficients (h) of 14.2-17.0 W/m'K were
obtained with blocks of 0.2 emissivity. Blocks of emissivity 0.88 raised h to about 28 W/m'K. Total
h values of 15-30 WIm'K were recorded in low convection, commercial band ovens. Overall h values
of 44 WIm'K were obtained in commercial impingement ovens with nozzle velocities of 11 mls.
Values as high as 170 W/m2K were obtained with nozzle velocities of 36 mls at 15D"C. It has been
shown that a simple reproducible technique that permits ready comparison of ovens and quantification
or radiation heat transfer has been developed. These attributes are valuable to the commercial baker.

INTRODUCTION

The question of scale-up in the preparation of commercial baked goods has historically been an
empirical process. Items prepared in a lab and baked in a test oven are brought to the manufacturing
oven where temperatures, baking time, burners, exhausts, air dampers and the like are adjusted until
the desired product attributes are obtained. There are dozens of oven types on the market which helps
explain some of the empiricism. It is readily apparent that the radiation contribution to heat transfer
in an oven with direct flames cannot be easily matched by simply setting an indirect oven to the same
temperature. One of the goals of scale-up is to use a measure that applies at different scales of
operation. This paper investigates heating reference blocks of aluminum in ovens and reviews the
applicability of the technique for scale-up purposes.
 299
Materials and Methods
If a mass of large thermal conductivity is placed in a heated air steam where the rate of heat transfer
to the mass is relatively low it may be assumed that temperature gradients do not exist in the mass.
When these conditions are met a simplified form of the unsteady state energy equation may be used
to evaluate heat transfer coefficients. In this paper an aluminum block, often insulated on all but one
face is heated in an oven and its center temperature measured with time. The appropriate equation [1]
is:

e . (h AI P CpV)t
Tairinit - Tblockinit

A semi-log plot of dimensionless temperatures vs time should yield a straight line whose slope equals
h Alp CpV.
Alwninwn Block Sensor
Aluminum blocks [2] weighing approximately 1800g were insulated with ceramic insulation on 5 sides
and heated. Thermocouples connected to an ovenable data collection system recorded time-temperature
data. Blocks with different emissivities were used to evaluate radiation heat transfer effects.
Results
A semi-log plot for two blocks of differing emissivities heated in a rotating rack oven maintained at
approximately 470°F or 243°C is shown in Figure 1. The linear relationship indicates the validity of
the technique's assumptions. The different responses for the two emissivities can be used to break out
radiation and convection contributions.

o.ooo~~----------------------------------~
DATA USED TO SEPARATE RADIATION
AND CONVECTION CONTRIBUTIONS

~-0.100r---------~~~~------------------------~
:l BARE BLOCK
!;i: EMISSIVITY -.3 =
ffiQ. lest.)
~ -0.200 1-_ _ _ _ _ _ _ _ _ _~h:-----.:.!!IoII....h-=-36-.-3....:lw-/m-2K"':}~--1
I-
U'J
~
~
Z
h = 48.91W/m 2 K}

~
~'Ir.
,
- • ..."

.~

o ~
_-0.300r---------------------------~~----~~.~~--~
o ~
ffi -7
::!E
Q

5-0.400~--------~======================----~~---
~ PAINTED, AIR = -243 C

~ BARE AI. AIR = -243 C

-0.500~~~~~~~~~~~ww~~~~~~~

o 2 4 6 8 10 12 14 16 18 20
TIME IN OVEN (MINUTES)

Figure 1 Semi-log plot used to determine heat transfer coefficients in a rack oven.

Data collection in two commercial band ovens is provided in Figure 2. This figure illustrates how the
energy absorbed by the block can be integrated. In this case temperature profiles were constructed
to illustrate the concept of using the blocks as calorimeters. Representative data are provided in Table
1.
300
250 -OVEN "A" AIR -OVEN "A" BLOCK· - OVEN ·'B" AIR 100
225 -OVEN "B" BLOCK * OVEN "A" HEAT
iii
w
....I
::l
80 ..,0
TEMP.
C,,) 0
ci175r---~~~----~~~~====~~ ....I
g
ew 150r-~r-----------~~~~~~~L-~ 60 ~
C,,)
w 0
III: ....I
::l III
.~
!;;: 0
III:
w 40 I-
a. ~
~ 75~-'------~~------~~--~__~~"r1 I-
::l
I- a.
~
20
!;;:
w
J:

~~~~-L~--~-L--~~O
o 1 2 3 4 5 6 7 8
TI~ (MINUTES)
Figure 2 Heat input comparison in two commercial band ovens using the aluminum block sensor,

Table 1
Representative heat transfer coefficients

Oven Temperature Block Velocity h

("C) (mls) (W/m2K)

Home, Electric 203 Bare 15.2

Home, Electric 210 Painted 21.9
Reel, Electric 208 Bare 22.3
Rack, Gas 243 Painted 48.9
Rack, Gas 243 Bare 36.3
Impingement 177 Bare 22 92.5
Impingement 177 Bare 35 110

Summary
It has been shown that a simple reproducible technique for evaluating heat transfer in ovens has been
developed. It can be used to separate radiation and convection contributions. In addition, the
technique can be used to monitor oven uniformity.

Reference

1. Eckert, E.R.G. and Drake, R.M. Jr., Analysis of Heat and Mass Transfer, McGraw-Hill, New
York, 1912, pp. 138-142.

2. Rubiolo de Reinick, A. and Schwartzberg, H. , Predicting Temperature vs Time behavior During

the Freezing and Thawing of Rectangular Foods, Biotech, Prog. 1986, 2pp. 164-114.
 MODELLING AND SIMULATION OF FOOD FRYING PROCESSES

G.S. MmAL and P. ATEBA

School of Engineering, University of Guelph, Guelph, Ontario, Canada

ABSTRACT

Heat, moistun: and fat transfers during deep-fat frying of meatballs were modelled and simulated. The
experimenl.Jl data provided fat diffusivity and conductivity. During frying, molecular diffusion
transferred the fat in the early stages of frying, and capillary flow was responsible at the later period.

INTRODUCTION

Deep-fat frying is defined as a cooking and drying process of foods in contact with hot oil and it
involves heat, mass and fat transfer. The fat not only acts as a heat transfer medium, but also enters
or leaves the food product. Heat, mass and fat transfer both within and around the food and the
formation of a crust are the basis of deep-fat frying. The overall objective of this research is to study
fat, heat and moisture transport in a beef meatball during deep-fat frying.

MODELLING

During deep fat frying operations, deep fried foods (those initially containing fat) undergo two fat
transfer periods namely: fat adsorption period, fat diffuses from the surroundings into the product, and
fat desorption period, fal migrates from the product to the surroundings. The value for fat diffusivity
during the first period is determined by applying Fick's second law. A capillary flow mechanism is
applied in the second period to determine the value for fat conductivity. The following assumptions
were made: meatball shrinkage was neglected, and crust thickness and effects of crust formation were
neglected.

.
MOisture transfer -arn = Om [2 ama -
-~ + - m] 2 (1)
at r or ar2
fat transfer

. arnf = 0 f [2 amf aZmf] (2)

Fat absorptlOn --0;:- -- - + - -
at r ar ar2

dmf = -KL(mf - mfe)

Fat desorption __ (3)
dt
302

Heat transfer dTat = (l [2 aT

r ar
+ a2rJ
ar2
(4)

initial conditions
mf(r,O) = mfO (5)

(6)

bOundai-Y conditions
dT Ir=o=O;
Tr am Ir=O=O; amf I =0
ar Tr·r=O
(7)

(8)

ami r=O
ar =0 (9)

where, m = moisture content, db; t = time, s; r = radial distam,c, m; Dm = moisture diffusivity, m2/s;
mf = fat content, db; Df = fat diffusivity, m2/s; KL = fat conductivity, 1/s; mfe = equilibrium fat
content, db; T = temperature, K; R = meatball radius, m; hy = heat transfer coefficient, W/(m 2 .K);
k = thermal conductivity, W/(m.K); Lv = latent heat of vaporization, J/kg; (l = thermal diffusivity,
m2/s; Pdm = density of dry matter, kg/m 3; subscript--a = ambient, c = central, e = equilibrium, 0 =
initial, s = surface, mpt = mid point.

PROCEDURES AND RESULTS

Deep-fat frying experiments were conducted to collect, fat, temperature and moisture profIle
histories. The experimental fat profile data was used to validate the fat transfer model. Minced meat
used in making meatballs for the experiments was prepared by using beef trimmings from the
forequarter (shank and brisket) and hindquarter (flank) obtained from the University of Guelph
Abattoir. The trimmings were stored at 4 0 C during the removing of fatty and connective tissues, and
then ground through a 1 cm dia. orifices. The meatball recipe required 1.67 kg hamburger binder/spice
mix, 27.23 kg coarsely grounded beef, 4.59 kg of beef fat, and 2.51 kg of added water to make a
batch of 36 kg. These ingredients were mixed in a mixer (Butcher Boy 150 Meat Mixer, Laser
Manufacturing Company Inc., Los Angeles, CA, USA). The mixture was further grounded through
0.3 cm dia. orifices to enhance homogeneity and to achieve a mince meat consistency appropriate for
making of meatballs. The composition of this minced meat mixture, from chemical analysis results
(AOAC, 1990), showed 65.30% water, 16.54% protein, 13.71% fat, 2.83% carbohydrate, 1.62% ash.

Three meatballs from each of 5 replications were fried for 60, 120, 180, 300,420 and 540 s,
respectively for fat profile data. For the frying experiment, a deep-fat fryer (Model: 941 hid, Hot
Point, Chicago) was used. A temperature acquisition and control system consisted of a datalogger
(Labmate, model 901, Sciemetric Instruments Inc., Nepean, Ontario) and a portable computer.
Temperature data at meatball centre, midpoint, near surface (0.5 mm from surface), 1.5 mm from the
surface, 2 mm from the surface, and 2.5 mm from the surface were measured using high temperature-
rated copper-constantan thermocouples. Beef shortening was used as the frying medium. The
temperature of the frying medium was maintained at 160°C.

Values of fat diffusivity and fat conductivity were determined by minimizing the root mean
 303
0.405~--~----------------------------------~

0.4

;0.395

~ 0.39
l_ 0.385
1
8 0.38

:2
..8' 0.H5

! 03"
. I

0.365+-----_r------r-----~----~------~----~
a 100 200 300 400 500 600
tlme(s)

Figure 1. Observed and predicted average fat content history, Cl observed; - predicted.

square of de~iations between observed and predicted fat concentrations. KL was represented by the
Eyring's ~bsoluL~ reaction rate theory:
KL = YA*T*exp( -_)
YB (10)
T
where T = temperature, K; YA = f~quency factor, l/(s.K); YB = activation energy factor, K. The
average value of Df was 6.I47E-7 m /s., YA = l.02E-4 (s.Kf I, YB =400 K, and mfe =0.32. Other
transport properties were reported by Huang and Mittal (1993).

Figures 1 and 2 show the predicted and observed fat and moisture contents, respectively. The
models were able to predict these profiles closely.
2.9~----------------------------------------__.

! 2.8
~
>- 2.7
i;
]'2.6
........
]'2.5

1f 2.4
£
E
o
2.3
'02.2

i2.1
• 2+-----~.-----_r------~----_.------_r------~

° 100 200 300

time, s
400 500 600

Figure 2. Observed and predicted changes in meatball weight, Cl observed; - predicted.

REFERENCES
Huang, E. and Mittal, G.S., Meatball cooking--simulation and model validation. J. Food Engg., 1983,
submitted.
 MODELLING PASTA COOKING PROCESSES

L. PIAZZA, M. RIVA AND P. MASI*

DISTAM, Universita' degli Studi di Milano, via Celoria, 2 - 20133 Milano, Italia
* Istituto di Industrie Agrarie, Universita' degli Studi di Napoli "Federico II"
Parco Gussone - 80055 Portici (Napoli), Italia

INTRODUCTION

It is well known that during the cooking process the physical and the chemical
characteristics of pasta change in a sensible manner. Usually the weight, the volume, the
consistence, the color and the porosity of pasta are assessed on the uncooked product and
on the cooked product at the optimum cooking time, i.e. the time required for the white
core in the pasta to disappear. Although this approach provides useful informations which
can be utilized by food technologists to improve the cooking performance of pasta, it does
not fully describe the commercial quality of the product. For this purpose it seems more
appropriate to observe the physical and the chemical changes which occur during the
cooking process determining the kinetics with which they evolve and expanding the
observation at times longer than the optimum cooking time.

MATERIALS AND METHODS

Commercial spaghetti of durum wheat from the same lot, having 2.2 mm diameter and
average composition: 9.8 % water and 15.9 % proteins were used. Cooking tests were
performed by using 25 g of pasta which were cooked in 250 ml of tab water. Periodically
the cooking process was stopped, the pasta drained for 2 min collecting the cooking water,
and the physical and chemical characteristics of the pasta determined according to the
following procedures. The weight increment was evaluated weighting the sample in a closed
container to avoid uncontrolled drying. The sediment amount in the cooking water was
evaluated bringing the water volume at its initial value and determining the residual weight
of sample kept for 24 hr in oven at !05°e. The variations of the overall dimension and of
the uncooked core were determined by looking at the sample cross section by means of an
optical microscope (Leitz, mod. SM LUX). The optimum cooking time was evaluated by
observing the disappearance of the central core by pressing the sample between two sheets
of Plexiglas (D'Egidio, M.G., De Stefanis, E., Fortini, S., Galterio, G., Nardi, S., Sgrulletta
D. and Bozzini, A., Standardization of cooking quality analysis in maccaroni and pasta
 305
products. Cereal Food World, 1982,27,367).
Mechanical properties of pasta during the cooking process were evaluated by performing
tensile tests by means of an lNSTRON UTM using a cross-head speed of 50 mm/min. To
avoid the difficulties connected to the sample clamping the terminal portion of each sample
were previously covered by a water-repellent paint so that at any cooking time the sample
ends were sufficient stiff to allow an adequate clamping in the apparatus clamps.
Color measurements were performed by means of the Hunter Minolta Chroma Meter II
colorimeter.

RESULTS AND DISCUSSION

Figure 1 shows the evolution of the main physical parameters during the pasta cooking
process. Due to water uptake the pasta weight increases uniformly with a rate which
decreases in the earliest stage of the process and that, than, becomes constant.
As the cooking process takes place the spaghetti external diameter progressively
increases, the diameter of the region which delimitates the uncooked core reduces and
becomes equal to zero at a time which is commonly considered to be the optimal cooking
time (OCT). The sediment amount in the cooking water increases rapidly during the earliest
stage of the process becoming constant soon. Separate test, not shown here for sake of
brevity, demonstrates that the porosity of pasta increases markately during cooking,
therefore one can conclude that the substances which are released in the cooking water are
only those which are present on the pasta surface.
In order to select a suitable phenomenological model to predict the pasta cooking
kinetics the above phenomena must be considered. The observation that in the course of the
entire cooking process the cooked and the uncooked regions of the spaghetti are delimited
by a well defined interface allows us to exclude the assumption, which has been in same
case made, that the cooking process of starch based food is governed by the simultaneous
occurrence of Fickian water diffusion and starch gelatinization (Bakshi, A.S. and Sing,
R.P., Kinetics of water diffusion and starch gelitinitazion during rice perboiling. L Food
Sci., 1980,45, 1387).

3......------------------,0,125
~~s ,. Sediment * Ri/Ro a Re/_~o~__ J
2,5
0,1
o
a:
a: 2 Figure I.
~ Evolution of physical
~ 1,5 and chemical parameter
a
N
I ~ _ _--_4o,05 'g during the spaghetti
01
U)
$: cooking process.
z
0,025
0,5

200 400 600 800

TIME (s)

The most convenient hypothesis is that water concentration profile modifies during the
process only in the cooked region and that the starch gelatinization occurs only at the
306

interface between the cooked and the uncooked region, whose position will progressively
change with time, as well as that of the interface between the pasta and the water due to the
swelling of the spaghetti. The solution of the set of partial differential equations which
describe mass transfer, starch gelatinization and boundaries movement during the process
can be made by means of well known numerical methods (Crank, J. The Mathematics of
Diffusion. Claredon Press, Oxford, 1975). It is interesting to note that, although this
approach will allow to predict the evolution on time of the pasta physical properties during
the cooking process, no informations it can provide concerning the pasta cooking
performance, in particular concerning the time when the optimal cooking time occurs.
The cooking performance of spaghetti appears to be more related to the evolution of
pasta rheological properties than to mass transfer and starch gelatinization. Figure 2 shows
the evolution of strain and stress at failure of spaghetti at different cooking stages. In
contrast with other physical parameter discussed above, both rheological parameters show a
dramatic variation during the cooking process and, moreover, their behaviour is
characterized by sharp variation corresponding to a consistence which is commonly
considered by the italian consumers to be optimal and usually defined as "a1 dente".

CONCLUSIONS

The cooking process of spaghetti involves many physical and chemical phenomena which
make the overall cooking process very difficult to model. A simplified mathematical
approach based on the assumption of constant Fickian diffusion and homogeneous chemical
reaction was proved to be inadequate.
However, the experimental findings suggested that a possible way to describe in a more
appropriate way the cooking performance of spaghetti could be based on the prediction of
the evolution of the pasta rheological behaviour during the cooking process.

1000 E~----;:======~--l. 100

I.... STRESS

----
-e-STRAINI

C\I ...
< 100 10
E Figure 2.
(f)
-\
.E. ~ Evolution of rheological
.9
(f) Z parameter during the
(f)
w
a: ~ spaghetti cooking
l- 10
(/) process.

1 '--~~~...l_~~~--'--~_~----L~~ _ 0,1
___I

o 200 400 600 800

TIME (s)

ACKNOWLEDGMENTS

Research supported by National Council ofItaly, Special Project RAISA, Sub-Project N.4,
PaperN.927
 OHMIC THAWING OF SHRIMP BLOCKS

MURAT O. BALABAN, TODD HENDERSON, ART TEIXEIRA, W. STEVEN OTWELL

Food Science and Human Nutrition Dept. University of Florida
Gainesville, FL. 32611, USA.

ABSTRACT
Conventional water immersion thawing of frozen seafood blocks requires large
amounts of water, causes cross-contamination problems, and loss of solids to
leaching. Frozen shrimp blocks and fish were thawed by passing an alternating
current through the blocks. 60 Hz and up to 480 V were used. An automated amps
and voltage data acquisition system was developed. A changing current regime
was used to prevent local overheating. Thawing times for different blocks for
different current-voltage regi mes are presented. A pre 1i mi nary economi c
analysis and sensory tests were performed.

INTRODUCTION
Water availability, conservation and reduction of wastewater generated are
vital to shrimp processors in Florida who use more than 3 billion L/year
water, and their wastewater contains 5 million kg dissolved/suspended solids
[1]. Water immersion thawing constitutes almost half of a plant's water use.
This method thaws the surface of the frozen blocks while the center is still
frozen. Warm surface temperatures support mi crobi a1 growth since thawi ng 1asts
from 2 to 3 hrs in water at about 27 C. Water soluble components are leached
from the product [2]. There may be cross contamination of blocks immersed in
the same vat. Another method for thawing frozen blocks is ohmic heating, where
alternating current passes through the block whose electrical resistance gene-
rates heat. Power input can be read more accurately for better control of the
operation. Ohmic heating occurs throughout the bulk rather than at the surface
[3]. Potential drawbacks of ohmic thawing are runaway heating as ice melts and
conductivity increases [3], lack of sensory data, and no water absorption of
shrimp, causing yield loss. If these problems can be solved, ohmic thawing
could reduce the cost of water used by the plant, and that of wastewater.

MATERIALS AND METHODS

An Lexan box (lx57x34x31 cm) was used to collect thaw water and to isolate the
samples. Two steel plate electrodes (30x24xO.16 cm) were connected to a vol-
tage regulator (Staco, Dayton, OH) with variable output to 480 V AC. Voltage
308
was manually adjusted during runs. An ammeter (Pyramid Instr., Lynbrook, NY),
and a multimeter (Extech, Waltham, MA) were connected to an IBM PC via RS-232
cables for data acquisition. Technical feasibility of ohmic thawing was
determined by:
a. Evaluating electrical requirements and ohmic thawing performance of 13
frozen shrimp blocks by measuring volts, amps and time, and by calculating the
heat generated by ohmic and that needed to thaw the block. For a given time
interval: Voltage (volts) x Current (amps) x time (hrs) = Energy (watt-hrs).
The sum of these values is the experimental energy used. By adding the
sensible (above and below freezing), and latent heats of the shrimp block, the
theoretical energy needed to thaw that block was estimated. The percentage of
electrical energy in the total energy required was calculated as coefficient
of performance (C.O.P.).
b. Determining water absorption rates of water-thawed and ohmic-thawed shrimp.
c. Performing sensory triangle test between water and ohmic-thawed shrimp to
see if different thawing methods lead to palatability differences (24
panelists). After thawing, shrimp were boiled for 1 min, and cooled for 2 mins
before serving.
Prel iminary economic feasibil ity was determined by comparing water-thawing and
ohmic-thawing operating costs. Based on available information and experiments
performed, the following data were used: water costs = $0.237/ 1000 kg water,
wastewater costs = $0.58/1000 kg water, electricity costs = $0.06 /KWh,
electrical demand = 0.092 KWh/ kg shrimp.

RESULTS AND DISCUSSION

Table 1 presents the relationship between C.O.P. and thawing time. The longer
the thawing time, the lower the percentage of ohmically generated heat used
in thawing, the rest being absorbed from the surroundings.
Table 1.
Summary of Ohmic Thawing Electrical Demand and Thawing Performance
Time To Energy
Block Size Type C. O. P. Thaw (KWh/kg
(count) (%) (Min) shrimp)
1 40-50 butterfly 77.57 101.5 0.08273
2 500 + peeled 53.08 179.0 0.05728
3 300-500 peeled 78.35 86.4 0.08352
4 300-500 peeled 92.98 40.8 0.10005
5 300-500 peeled 92.28 50.2 0.09373
6 300-500 peeled 86.85 61.9 0.09447
7 300-500 peeled 87.92 64.3 0.11254
8 300-500 peeled 78.69 134.0 0.08525
9 300-500 peeled 90.16 145.0 0.09666
10 30-40 shell-on 69.82 187.0 0.07524
11 30-40 shell-on 94.04 191.0 0.10468
12 30-40 shell-on 37.45 156.0 0.04012
13 30-40 shell-on 25.72 280.8 0.02751

Table 2 shows the average wet basis moisture content of shrimp thawed by
different methods. Statistical analysis revealed that at a 5% level there were
no significant differences between the methods, eliminating the concern for
loss of yield.
 309
Table 2
Percent Moisture Content (each average is that of 8 samples)
air thawed water thawed ohmically thawed
count (%) (%) (%)
300/500 82.8 84.2 81.7
30/40 80.8 82.5 79.6
averages 81.8 83.4 80.7

Table 3 shows water absorption rates of ohmically thawed shrimp placed in an

ice slush. Since the average stay of thawed shrimp in ice slush is about 1 hr
before processing, its water content will be similar to that of conventionally
thawed shri mp. Sensory tests showed that panel i sts coul d not detect a
significant difference between ohmic and conventionally thawed shrimp.

Table 3
Chill Tank Percent Moisture Content Summary
Time Block 1 Block 2 Block 3 Block 4
min 300-500 count, peeled 30-40 count, shell on
15 83.50 84.28 80.60 80.98
30 84.77 84.86 80.55 82.28
45 84.57 85.53 81.07 80.44
60 85.58 85.37 81.68 80.64
75 85.17 86.18 80.86 80.85
90 85.27 86.15 82.05 81.61

Using ohmic heating as an alternative to the water immersion thawing

system currently being used by shrimp processors has the potential of reducing
the amount of water used by the processors without altering the quality of the
final product. The costs savings realized by a shrimp processor is $13.9 per
1000 kg shrimp. The operating cost savings of this method to a processing
plant producing 26,000 kg/day is more than $95,000 per year.

REFERENCES
1. Bough, W.A. and B.E. Perkins. 1977. "Recovery of By-Products From
Seafood Effluents." In W.R. Hess, Jr., ed. Proceedings of the
Interstate Seafood Seminar, VPI Seafood Extension Unit, Hampton,
Virginia. pp. 201-273.
2. Peplow, A., 1975. "Effects of Iced Storage and Thermal Processing on the
Proximate, Mineral, Extractable Protein, and Thiamin Composition of
Shrimp". Master's thesis, University of Florida.
3. Biss, C.H., S.A. Coombes, and P.J. Skudder. 1989. "The Development and
Application of Ohmic Heating for the Continuous Heating of Particulate
Foodstuff." In: R.W. Field and J.A. Howell, eds. Process Engineering
in the Food Industry, Developments and Opportunities. Elsevier Applied
Science, New York. pp. 17 - 25.
 OHMIC HEATING OF FOOD MATERIALS
EFFECTS OF FREQUENCY ON THE HEATING RATE OF FISH PROTEIN

K. Uemurall, A. Noguchj!), S.J. Park 2) and D.U. Kim 3).

I) National Food Research Institute, 2-1-2, Kannondai, Tsukuba, Iabaraki 305 Japan,
2) Ottogi Foods Co., Ltd. Korea, 3) Cheil Foods & Chemicals Inc. Korea.

ABSTRACT

Fish minced meat (Alaska Pollack) was used to make "Kamaboko" samples with NaCI (0-1
w/w%) and/or starch (0-10 w/w%) and treated with alternate current (0-20Y) at various
frequency (50Hz-10kHz) to heat up to 90°C. The resulting products were examined their
breaking strength, color and fine structure. Heating rate was found dependent mainly on
the frequency and increased at 10kHz to 7.5 times as fast as the boiled sample. The
products gave similar light values (L, a, b) and always higher breaking strength which
reached to the maximum value 1.6 times as hard as that of boiled one. The dielectric loss
was found maximum at around 10kHz which likely contributed to quick heating and better
gel formation. These results suggested that water was not homogeneously distributed but
individually separated in the matrix of protein gel and that the gel had the maximum dielec-
tric constant at specified frequency according to the gel structure and distribution of
components.

INTRODUCTION

The quality deterioration of food product during thermal processing can be minimized by
HTST (High lemperature ~hort lime) process. As long as the food liquid involved is not
too viscous, heat transfer to the center of foods will take place hy convection. However, in
processing particulate foods, the solid phase will only heat by conduction which is a much
slower process than convection, due to the low thermal conductivities of most foods (1).
Ohmic heating generates heat within foods by the passage of alternating electric current and
enables solid phase to heat as fast as liquids, thus making it possible to use HTST tech-
niques on solid foods. Ohmic heating rates are critically dependent on the electrical con-
ductivities of the foods (2). When the solid food such as protein gels has the rigid protein
matrix and tightly holds water and electrolytes, it is likely that the frequency of alternate
current will become one of key parameters for quick heating by reducing the impedance of
food. In this paper, the aim was to find the effects of frequency of alternate current on the
 311
heating rate of solid food and also the corrosion of electrode.

Materials and Methods

Frozen fish minced meat (Alaska Pollack) was kneaded under vacuum, if necessary with
NaCI (0 or 1 w/w%) and wheat flour (0, 5, 10 w/w%) and left at lODC to have "suwari"
pregelation. The resulting gel was punched out as the cylindrical tube (0:48mm, h:26mm)
and put between aluminum electrodes (Fig. 1). The therm couple was inserted into the
center of gel. After ohmic heating, the gel was measured its breaking strength, light color
values with the conventional methods. The electrodes was immersed into NaCl solution at
a distance of 40 mm and examined its weight loss at various NaCl concentration and fre-
quency.
Powe r ..,I.f l er Control Ie'

•••• oc e •
• •• • co [] .
0000000 .

Mon i tor
H. . l l ng j lckO!

Figure 1. A schematic diagram of ohmic heating at various frequency

RESULTS

Fig. 2 showed the heating rate of fish protein gels at various frequency. When the gel
contained 1 % NaCl, its heating rate was similar to that of control (boiled sample) and then
remarkably increased as the frequency increased more than 1 kHz and achieved about 7.5
times as fast as the control. This result clearly suggested that the gel had high impedance
due to protein matrix and the frequency was a key factor for quick ohmic heating of sold
foods. However, all samples showed similar electrical constants such as resistance, imped-
ance and reactance more than 100 Hz. When the gel was replaced with possible equivalent
circuit (Fig.3) and examined its dielectric loss, the high frequency around 10 kHz gave the
maximum generated heat. This dielectric loss likely contributed to the higher heating rate
than that of control. The gels prepared with ohmic heating, always showed similar light
value (L, a, b) to that of control and higher breaking strength, gaving an increase of 1.6
times. No clear relationship was found among the breaking strength, imaginary part of
dielectric constant and heating rate. Fish protein gel is known to have the phenomena
called "Modori" around 600C and reduced its breaking str~ngth after prolonged heating at
this temperature. Ohmic heating probably shortens the heating time at this temperature and
leads to the increase of breaking strength of gel.
312

(1)
+J
<U
L..

0'>
c:
8

6
-.-
- O-
- i::! -
bo iii 09
N: O%
N: 1%
N: 1%.
+J
<U
(1)
- O- N: 1 %.
.s:::: 4
u
'+-
U
(1) 2
c..
en

O~====~~==~L-----L-----~
50 Hz 100 Hz 1 kHz 2 kHz 10 kHz
Frequ~nCY
Figure 2. Specific heating rate of fish protein gel at various frequency
N: NaCI S: Starch

Rp
E'Ie: t rode

C r
Equ ivalent Circu i t
Kp: KeS I stance
r : Resistance caused from dielectric loss
C : Ideal condenser
Figure 3. Possible equivalent circuit of fish protein gel

Aluminum electrode was found to be corroded in NaCl solution at 50 Hz as NaCI concen-

tration increased. When the electrode was immersed into 5 % NaCl and examined its
corrosion by changing the frequency of alternate current, the higher frequency was ob-
served to reduce the corrosion and make it negligible at 50 kHz. This result suggests that
the erode aluminum will be caught again with the electrode due to the rapid change of its
polarity.
There remains many unclear points in Ohmic heating. However, these results indicate that
Ohmic heating is one of promising methods for homogeneous and quick heating of solid
foods and that higher frequency is more suitable to have larger heat generation due to the
dielectric loss and to reduce the electrode corrosion and its contamination of foods.
References
1. Halden, K., De Alwis, A.A.P. and Fryer, P.I., Changes in the electrical conductivity of
foods during ohmic heating. lnt. L Food Techno!., 1990,25,9-25.
2. Parrot, D.L., Use of Ohmic heating for aseptic processing of food particulates. Food
Technology, 1992,46,68-72
 HEAT TRANSFER ANALYSIS IN FOOD
HEATED BY FAR-INFRARED RADIATION

NOBORU SAKAI and TAMOTSU HANZAWA

Department of Food Science and Technology
Tokyo University of Fisheries
Tokyo lOS, Japan

ABSTRACT
A computerized method was developed to predict two dimensional heat transfer in
food heated by far-infrared radiation. A finite element method was applied to solve
the fundamental equation and was used to examine the influence of the heater
temperature and surface heat conductivity on heat transfer in the food. Control of
the surface temperature, to keep the food from overheating, was discussed.
Experiments of far-infrared heating were carried out to verify the method.
Temperature distributions in samples were measured and they corresponded with
theoretical predictions.

INTRODUCTION
Far-infrared radiation have attracted interest recently in the thawing of frozen
materia«l>. Since its quality is greatly influenced by the thawing rate and the final
temperature, investigation of heat transfer in food heated by far-infrared radiation,
is important. In this study heat transfer in an unfrozen sample is analyzed as basic
research for the study of thawing by far-infrared radiation.

MATERIALS AND METHODS

Experimental
Agar and raw tuna were used as samples, and they were wrapped in very thin film
to avoid dehydration. The diameter and the height of the cylindrical samples were 8
cm and 3 cm, respectively. The sample was placed in a constant temperature box,
in which fans were set inside to control an air flow rate, and was heated by a
ceramic heater placed above the box. The temperature distribution was measured
by using thermocouples and was recorded on a floppy disk.
314
Mathematical model
It was assumed that the far-infrared radiation energy was only supplied to the upper
surface of the cylindrical sample. It was also assumed to be absorbed at the surface,
because the absorption coefficient of water is very large(2). Then, energy was
transported by heat conduction in the food. Based on theses assumptions, a
governing equation and boundary conditions for heat transfer were derived from
the heat balance in the food. These equations were solved numerically by using
Galerkin's finite element method. Two dimensional linear elements were used to
discretize the equation. The Crank-Nicolson finite difference method was used to
integrate an ordinary differential equation obtained through the discretization.

RESULTS AND DISCUSSION

Figure 1 shows the comparison of calculated temperature distributions with
experimental ones for the axial direction (Fig. I-a) and for the cross-sectional
direction (Fig.I-b). The sample was a cylindrical agar which had a 99% water
content. Plots are experimental results and lines are simulation results. In Fig.I-b,
the central temperature was lower than the outside temperature in the initial stage.
However, after 60 minutes the inside temperature became higher than the outside
one. This means that heat entered from the sides of the sample in the initial stage,
however, after the temperature of the side became 30 "C, heat was lost from the
sides as a result of convection. Experimental results corresponded with theoretical
results.

50 50 ~---------------------,

- - Calc.

• z=3.0cm • r=3.5 em

• z=2.Scm • r=2.0cm
10 10
.. z=l.Ocm .. r=O.Oem

O~~L-~~~~~~~~~ O~~L-~~~~~~~~~

o 20 40 60 80 100 120 o 20 40 60 80 100 120

Time [min] Time [min]

(a) axial direction(r=Ocm) (b) cross-sectional direction(z=lcm)

Figure 1 Comparison of calculated temperature distributions with experimental ones.
Heater temp.=203 "C • air temp.=30 t, initial sample temp.=1 "C , air flow rate=4.5 mls

The effect of the heater temperature and the air flow rate, on the temperature
distributions, were examined. It was found that the higher the heater temperature,
the higher the surface temperature of the sample. As the heater temperature
 315
increased, the temperature difference between the surface and the inside increased,
because the heating rate at the surface was greater than that in the inside. On the
other hand, the higher the air flow rate, the lower the surface temperature of the
sample, because the heat loss through convection increased with the air flow rate.
So, the temperature difference between the surtace and the inside decreased.
In the heating process of food, it is very important to control the surface
temperature, because the quality of food decreases due to overheating. So as a
sample, we tried to control the surface temperature by repeating irradiation and the
shield of radiation energy. To find the time interval of irradiation we simulated the
temperature distributions under the following restrictions:
1. If the surface temperature was higher than 32"C , radiation was cut off.
2. When the temperature went lower than 28"C , irradiation began again.
3. The above restrictions were judged every 1 minute.
Figure 2 shows the simulation results calculated under the restrictions. Solid
lines represent temperature histories at the surface (z=3cm, r=Ocm) and the inside
(z=O.5cm, r=Ocm) of the sample. It is found that the surface temperature did not
exceed 35 "C . The dashed line represents the time interval, that is, the upper and the
lower part of the line mean the period of irradiation and the radiation shield,
respectively.
According to this time interval, we carried out a heating experiment with
using raw tuna. In the experiment, we put an aluminum plate between the heater
and the sample to cut off the irradiation. In Fig.2 plots represent experimental data,
and coincide with the solid lines. Therefore the surface temperature can be
controlled well, using this method.

o ~~~~~--~~~~~--~~~
o 10 20 30 40 50 60
Time [min]

Figure 2 Control of surface temperature with repeating of irradiation and shield of radiation
energy. Heater temp.=200"C , air temp.=30"C , initial sample temp.=1 "C , air flow rate=2.8 mls.

REFERENCES
1. Ohashi, Y., Refrigerator with partial defrosting system. Refrigeration. 1990.65,487-492.
2. Irvine, W.M. and Pollack J.B., Infrared optical properties of water and ice spheres. Icarus.
1968,8,324-360.
 OPTICAL CHARACTERISTICS OF VEGETABLES
IN FAR-INFRARED IRRADIATION HEATING

MASARU SHIMIZU, ATSUSHI HASHIMOTO* and SEI-ICHI OSHITA*

Department of Chemical Engineering, Faculty of Technology,
Tokyo University of Agriculture & Technology,
Nakamachi, Koganei, Tokyo 184, JAPAN
*Department of Bioproduction & Machinery, Faculty of Bioresources,
Mie University, Kamihama-cho, Tsu, Mie 514, JAPAN

ABSTRACT

Eggplant, potato, radish and sweet-potato were used as the samples. A

vegetable model is constructed in order to evaluate an apparent absorption
coefficient of a vegetable, which is calculated with that of the dry material
and water in consideration of the water content and the void fraction. The
damping curves of radiative power within the vegetable model irradiated by
far-infrared (FIR) radiation are estimated by using the apparent absorption
coefficient of the model and the irradiation power. It is thinkable that the
radiative power absorbed by the vegetable model is damped to 1 % of the
initial value at a depth of several deci-millimeters.

INTRODUCTION

Far-infrared (FIR) heaters are now manufactured, and FIR is widely utilized
in our daily life. In the food industry, the application of FIR to thermal
operations in food processing can be expected to draw considerable interest.
However there have been few fundamental studies on its application.
The present purpose is to determine the absorption coefficients of
vegetables and to estimate the penetration depth of radiative power. The
absorption coefficient is a very important optical property of the irradiated
matter, and we should grasp the penetration of radiative power in order to
design the infrared-radiative operation:;; and the process.

MODEL CALCULATION

Consider a vegetable irradiated by a FIR heater. As the vegetable temperature

is much lower than the surface temperature of the heater, the emission within
the vegetable is negligible. The penetration of radiative power into the
vegetable can be estimated by using the damping function [1].
 317

Ia"0 a-Iv,A) qir,AeXP (-«v,AX) dl

ct> (x) (1)
Jr"
o
(l-I v A)
' qir Adl
,

Here qir'A is the irradiation power, a V'A and r V'A are respectively the
absorption coefficient and the reflectance of the vegetable. x is the depth
from the vegetable surface, and A. is the wavelength. r v A may be assumed to
be approximately equal to the reflectance of water [2], as the water content
of the vegetables are very high. qir'A is experimentally obtained. So a V'A is
needed in order to calculate if> (x).
A vegetable model to estimate a V'A is constructed. The vegetable model
is composed of the dry-vegetable material, water and the void. It is assumed
that infrared radiation is transmitted without absorption in the void. a V'A may
be expressed by using the apparent absorption coefficient of the dry-
vegetable material, a d'A' and the absorption coefficient of water, a W'A' as
follows.
(2 )

X=t.) ~ (3 )
w w Pw

(4)

Here x d' Xw and Xv are volume fractions of the dry-vegetable material, water
and the void in the vegetable model, respectively. Ww is the water content in
wet basis. p v and p ware respectively the bulk density of the vegetable and
the density of water. So ad .. is needed in order to estimate a v'A' as a W'A is
available in the literature [3J.

MATERIALS AND METHODS

Eggplant, potato, radish and sweet-potato were used as the samples. The
vegetable was dried in an oven at 338 K for 48 h, and the dry vegetable was
powdered. 1, 2 and 4 mg of the dry-vegetable powders were mixed with 40 mg
of KBr powder. The mixture was pressed, and the sample tablet for the
infrared-spectrum measurement was prepared. The transmittance was
measured by a Fourier transform infrared spectrophotometer (JASCO, FT/IR-
3), and the thickness of the sample tablet was measured by a micrometer.

RESULTS AND DISCUSSION

It is assumed that Beer's law may be approximately applied to the measured

infrared spectrum of the dry-vegetable material, and a d'A is obtained. Figur"e
1 shows the estimated a V'A by using the calculated a d,l. and a W,A. a V'A has
strong absorption bands near 3 and 6 /J. m of wavelength, and shows the
similar trends to a W'A .
318
The calculation results of the penetration of radiative power, ¢ (x), into
the vegetable models are indicated in Fig.2. The irradiation power is evaluated
at a 773 K surface temperature of a typical FIR heater used in the previous
study [1]. In this study, XO.Ol is defined as the penetration depth. The
radiative power absorbed by the vegetable model is damped to 1 % of the
initial value at the depth of XO.01 • As shown in Fig.2, xO.Ol is from 0.26 mm to
0.36 mm. As penetration of radiative power is affected by the conditions of
irradiation and by the vegetables, XO.Ol may vary more or less. Then it is
thinkable that XO.Ol is several deci-millimeters.

---- SNeet potato ---- Svveet potato

_ . - Potato _ . - Potato
-··-Radish -··-Radish
............. Eggplant ............. Eggplant

10~~--L-~~-~-~~-L--~
o 0.4 0.6
X [mm]

Figure 1. Apparent absorption Figure 2. Damping curves of

coefficient of the vegetable models. radiative power into the vegetable
models.

ACKNOWLEDGEMENT

The authors wish to thank Mr. Yutaka Yamazaki for assisting in the
experiments.

REFERENCE

1. Hashimoto, A., Takahashi, M., Honda, T., Shimizu, M. and Watanabe, A.,
Penetration of Infrared Radiation Energy into Sweet Potato, Nippon
Shokuhin Kogyo Gakkaishi, 1990, 37, 887-893 (in Japanese)
2. Hale, G.M. and Querry, M.R., Optical Constants of Water in the 200-nm to
200-,u m Wavelength Region, Applied Optics, 1973, 12, 555-563
3. Hirota, K., graduation thesis, Tokyo University of Agriculture & Technology,
1988 (in Japanese)
 TEMPERATURE DISTRIBUTION IN MICROWAVE OVEN
HEATING - THE INFLUENCE OF DIFFERENT CAVITY MODES

THOMAS OHLSSON AND PER OLOV RISMAN

SIK, Box 5401, S-402 29 GOteborg, Sweden

ABSTRACT

During microwave heating, te~ture distribution controls the product safety in

pasteurization and sterilization. It also ensures that the desired quality of the food is
reached. Microwave heating uniformity is controlled by a number of interacting factors
such as the composition of the food and the package, the geometry of the food and the
microwave field created inside the food. The type of electrical polarity of the oven
fields is of large importance to the coupling of microwave oven energy fields (TE and
TM volume cavity modes) to the food and the resulting temperature distribution. Diffe-
rent types of cavity modes will give large differences in corner- and edge overheating,
an important problem in microwave heating of foods, as illustrated by advanced three-
dimensional numerical simulations and by infrared thermographic measurements.

INTRODUCTION

The temperature distribution in microwave foods is coupled to the microwave energy

distribution. This is controlled by a number of interacting factors such as food and
packaging composition and geometry, the microwave field distribution in the oven and
the resulting field in the foods. For some years, research on microwave heating at SIK
has concentrated on understanding the interacting factors, (Ohlsson, 1983, Risman et
al.., 1987). The influence on food composition on the microwave energy penetration
depth and the resulting temperature profiles have been demonstrated (Ohlsson and
Bengtsson, 1971, Ohlsson and Bengtsson, 1975). The influence of the food packaging
material was shown to be dramatic both on temperature distribution and required
heating time (Ohlsson and Risman 1991, Risman, 1992). For food with spherical and
cylindrical geometry, strong centre heating effects occur for food diameters between 20
and 50 mm (Ohlsson and Risman, 1978). In recent years, the comer and edge over-
heating effect in wedge-shaped food have been studied (Risman and Ohlsson, 1993).
This overheating effect is strongly related to the type of microwave oven field pattern
320
in the oven, and in particular the type of polarization (TE or TM) of the field in
relation to the food. The polarization will greatly affect the "coupling" of the microwa-
ve energy in the oven to the food (Risman, 1993). This will be further presented
below.

MICROWAVE OVEN FffiLD DISTRIBUTION

The wavelength ("0) of microwaves in air (122 mm) is of the same order of magnitude
as the dimensions of a domestic microwave oven cavity. A few wavelength of micro-
waves will "fit" into the cavity. The microwave energy will set up a standing wave
resonance field pattern with maxima and minima in the empty oven cavity. When a
food is introduced, the balance between the modes is affected depending on the geo-
metry and dielectric properties of the food. The oven is designed to promote fields that
combined give a uniform heating of the food. But, energy transfer is also taking place
by non-resonant modes and by direct radiation from the microwave infeed system,
(Risman, 1991). An important characteristic of the electromagnetic microwave fields in
the oven,is that they have two different polarizations; TE = transversal electrical with
the electrical field vector parallel to the food surface, and TM = transversal magnetic
with the electrical field hitting the food surface at an angle. When a microwave beam
falls in on a food surface, part of it will be reflected, part absorbed and part trans-
mitted. The reflection factor will vary greatly for the dielectric properties of the food
and the angle of incidence. Most strikingly is the variation with the field polarization.
For TE, the reflection factor gradually increase to I. For TM, minimum reflection, is
found for high angles of incidence. This will lead to excellent power utilization, and
much improved possibilities to control the heating uniformity,. The microwave power
will be absorbed, where the oven power distribution system delivers it. It will not be
reflected back to the cavity and be part of the less controlled volume resonance fields.
These findings have contributed to the development of improved designs of microwave
power distribution systems, in both new generations of domestic oven and of industrial
microwave heating applicators. The type of field polarization is also important for the
edge and comer overheating effect. Recently, it has shown that this only takes place
when the electrical field is parallel to the edge of the food, (TE polarization). For TM
polarization, with the electrical field hitting the edge of the food at an angle, the edge
is only slightly heated (Risman, 1993). Thus, in order to avoid comer and edge over-
heating, TM fields should dominate in the oven, at least around the edges of the food.

COMPUTER SIMULATION

For analysis of the dynamic development of the temperature distribution during

microwave heating a three-dimensional computer software package, of the Time
Domain Finite Difference (TDFD)-type. In the calculations, a scenario is built up of
the microwave oven, using small cells of a few mm side-length. Each cell can be "fil-
led" with dielectric properties representing foods, packaging, air, oven walls, etc. The
simulations are then performed in time steps until the computer finds a stable solution.
 321
Electrical field patterns in the oven space are shown using UNIRAS graphics software.
as well as the absorbed microwave power density in the food. It must be realized that
computer simulations provide "complete" pictures in the sense that standing waves,
polarization, diffraction, focusing and shielding effects are combined into one single
result. The limitations of the TDFD-software lie in the nature of the numerical method:
the division of the oven and the food into small cells. For reasonable computer time,
the cubical cell side length needs to be about 5 mm. For some types of oven, the
variation in oven cavity dimensions by changes of one cell length can create noticeable
variations in the oven fields. The computer simulation method is also limited in that
not all types of field excitation system can be modelled.

REFERENCES

1. Ohlsson, T. and Risman, P.O., Temperature distribution of microwave heating.

Spheres and cylinders. J. Microwave Power, 1978, 13(4), 303-310.
2. Ohlsson, T., Fundamentals of microwave cooking. Microwave World, 1983,
4(2), 4-9.
3. Ohlsson, T. and Risman, P.O., Temperature distribution in microwave oven
heating of foods. The influence of packaging material. Microwave and High
Frequency symposium. 8-10 Oct., 1991. Nice.
4. Risman, P.O., Ohlsson; T. and Wass, B., Principles and models of power density
distribution in microwave oven loads. J. Microwave Power, 1987, 22(4), 193-
~8. .
5. Risman, P.O., The role of TMz cavity modes in microwave ovens. Microwave
and High Frequency symposium, 8-10 Oct., 1991. Nice.
6. Risman, P.O., Metals in the microwave oven. Microwave World, 1992, 13(1),
28-33.
7. Risman, P.O., and Ohlsson, T., 1993. Microwave heating of wedge shaped
dielectric loads. Submitted to J. Microwaw Power.
8. Risman, P.O., 1993, Cavity volume modes and power deposition patterns in
rectangular microwave cavities and food loads. Submitted to J. Microwaw
&nYer.
 IN-FLOW MICROWAVE HEATING OF PUMPABLE FOODS

T. OHLSSON
SIK, Box 5401, S-402 29 Goteborg, Sweden

ABSTRACT

With the objective of developing a heating system that uniformly heats food containing
particles with minimal mechanical damage to the particles,computer simulations have
been performed in order to find microwave field patterns that give uniform in-flow
heating in a tubular microwave heater. A pilot plant test unit has been built for a 38
mm tube diameter. The initial experiments showed that temperature variations within
5 % of the overall temperature rise can be obtained. Little or no mechanical damage to
particles was noticed.

INTRODUCTION

If one wants to improve efficiency and process control in the heating of foods, new
methods such as microwave heating are of interest. For high-viscosity foods containing
particles, it is also important to find methods where the mechanical damage to the
particles is less than when using conventional heat exchangers. Microwave heating
offers the possibility of heating the interior of foods directly without the need for
severe mixing or the use of hot heat-transfer surfaces. It is very important to be able to
control the microwave field distribution in the food in such cases. In a previous
publication (Risman & Ohlsson, 1975), the microwave field for small tube diameter
heaters has been presented. It was shown that for small tube diameter, the microwave
fields have similarities with the parabolic velocity profile in laminar flow.

Microwave heating in a tightly controlled microwave field applicator is less

sensitive to variations in dielectric properties, as a consequence of variations in raw
material preparation or composition or over the temperature range of the heating as
compared with lower frequency electrical heating methods; e g high frequency or
ohmic heating. By using advanced computer simulation of the microwave field, heating
 323
patterns that give small field variations across the tubular food cross section can be
selected. If the liquid flow system is. carefully designed, the mechanical damage to
particles during processing can be minimized, thereby using the advantages offered by
microwave heating in terms of direct heat transfer to the particles and liquid.

OBJECTIVE

To pasteurize and sterilize high-viscosity foods containing large particles using

microwave heating with tightly controlled microwave field distribution and minimal
process-induced mechanical damage to the particles.

EXPERIMENTAL PILOT PLANT

Computer simulation programs have been developed, both involving analytical and
numerical solutions to Maxwell's equations for the electromagnetic field distribution in
tubular loads enclosed in cylindrical microwave resonators. The analytical program is
written in FORTRAN and run on a Pc. The numerical program is of the Time Domain
Finite Difference-type (TDFD), and runs on a Vax-station. Computer simulations have
been done on systems for inner tube diameters of 25, 38, 52 and 64 mm. Based on the
simulation results, an experimental pilot plant unit has been constructed for in-flow
microwave heating of foods transported in a microwave transparent glass tube with an
inner diameter of 38 mm. The microwave heating unit consists of 4 magnetrons, each
delivering up to 2 kW into specially-designed applicators. A tubular preheater increases
the product temperature to a suitable starting level for the microwave heating. The
foods are transported by pressure differentials or by a SINE-pump.

Temperature measurements were made in different locations in the cross-section

of the tube with fiber optic temperature sensors and thermocouples and by injecting
thermocouples into particles after the heating section. The flow of the particles was
video-recorded for studies of the uniformity of flow. For analysis of the microwave
heating distribution both a gel, that change colour at a given temperature, and a
pancake batter that solidifies during heating were used. The gel was enclosed in a
plastic film tube, which was inserted into the tube during the heating. After removal,
the coloured areas gave a three-dimensional illustration of the heating pattern of the
applicators. In this way the static microwave field was visualized. By slowly putting
the gel tube through the applicator the field heating the pumped food can be visualized.
For the analysis, it is important to distinguish between the microwave heating field and
the resulting temperature distribution across the tube cross-section. The later will of
course include the tube flow velocity distribution effects.
324

RESULTS AND DISCUSSION

Microwave heating patterns

For small tube diameters, such as 15 mm, single mode applicators give a maximum
field concentration in the centre, according to the analytical calculations. At 25 mm
tube diameter, the centre heating tendency still remains. When the tube diameter
exceeds the penetration depth in the food substantially, such as for tube diameters of 38
mm and larger, the field is primarily heating the quadrant of the tube facing the
applicator. With four applicators at 90 degrees from each other a surprisingly uniform
field is reached for also the larger tube diameter.

Pilot plant experiments

A high level of viscosity in the carrier liquid has been found to be of great help in
controlling the uniformity of flow of the particle containing liquids. The best test
conditions were found for a 1.5% CMC solution as carrier with 10-12 mm particles.
Good qualitative agreement was seen between the computer calculated field distribution
and the static heating patterns of the gel.

The temperature measurements show that variations are within + /- 2 C for an

0

average temperature increase of 50 C. The temperature difference between the liquid

0

carrier and the particles was only 1_3 C. The mechanical damage to the particles due
0

to the processing was judged to be small. For industrial applications, a number of

applicators, each with a microwave generator and power pack, can be applied to a
microwave heating tube, to achieve the required temperature increase. The applicators
are simple and low cost. Power packs can also be simple in design and relatively low
cost. However, as often, it is recommended to use conventional less expensive beat
exchangers for the initial non-critical temperature rise, and concentrate the microwave
input to the final, more crucial to control temperature increase.

REFERENCE

Risman, P.O. & Ohlsson, T., 1975

Theory for and experiments with a TM02n applicator.
Journal of Microwave Power 10(1975):3, p. 271-280.
 MATHEMATICAL MODELING OF BATOI HEATING
OF LIQUIDS IN A MICROWAVE CAVITY

A. K. DATTA, H. M. PROSETYA, and W. HU

Department of Agricultural and Biological Engineering
Cornell University, Riley-Robb Hall, Ithaca, NY 14853, USA

ABSTRACT

Using a heat generation term that exponentially decayed from the surface into the liquid, the
energy and flow equations were solved to successfully predict temperatures inside a cylindrical
container of liquid in a microwave cavity (oven). A recirculation pattern was setup where the
warmer liquid near the surface was rising due to higher rate of heating. Transient temperature
proflles were almost linear. The radial temperature proflles were almost uniform, with slight
increase toward the wall. The axial temperatures increased with height showing stratification.
The uniformity of heating slowly decreased with time due to the spatially varying rate of heat
generation.

INTRODUCTION

Although heating of containerized liquids is a vsry common and important application, studies
so far have used a lumped parameter approach(l . Detailed spatial temperature proflles and flow
patterns have simply not been studied. The heat transfer literature has plenty of studies of
natural convection with internal heat generation. However, almost all of these studies are for
constant volumemc heat generation with cooling at one or more surfaces that set up the convec-
tive flow pattern 2). The driving force for convective flow in microwave heating is the spatial
non-uniformity in the volumetric heat generation, which has not been studied and was the ob-
jective of this study.

METHOOOLOGY

The volumetric heat generation in a cylindrical container was considered exponentially decaying
from surface. Additionally, the microwave penetration was considered axisymmetric, reducing
the problem to a two-dimensional one(3). The container was considered microwave transparent
The model fluid was water, whose temperature dependence of dielectric properties was
included by varying the penetration depth with temperature. The boundary conditions used
were zero heat flux from all sides. At the walls and at the bottom the plastic material of the
container would be fairly insulating. Heat loss through the top surface (covered with aluminum
foil) was ignored. Further details of the governing equations, boundary conditions, and
solutign techniques are in Datta et al.(4) The experimental setup is discussed in Prosetya and
Datta( ).
326
RESULTS
The radial temperature profiles at three different times are shown in Figure 2. This uniformity
in heating is partly attributed to the considerable convective flow in a low viscosity fluid such
as water. The axial temperature profiles at three different times are shown in Figure 3. Strati-
fication leads to the higher temperatures with height The calculated and measured coldest
temperatures, useful for estimating the least bacterial inactivation in biochemical and food pro-
cessing, are shown in Figure 4. One of the primary advantages of microwaves is thought to be
higher heating uniformity as compared to conventional heating from the surface. The spread in
temperature values is therefore one measure of success for microwave heating. Figure 5 shows
the spread of temperature values increase with time due to continuous heat generation at
spatially varying rates. This is in contrast with conventional heating where all temperatures
increase toward the boundary temperature, making the temperature spread decrease with dura-
tion of heating. For further information, the reader is referred to Datta et al. (1992).

Applications
Reheating, in-container pasteurization, and sterilization are the possible uses of batch heating of
liquid in a container in the context of food materials. The radial uniformity of heating would
make it easier to design microwave processes. From a microbiological safety point of view,
due to the radial uniformity and stratification of the liquid, one needs to be concerned only with
temperatures at the bottom of the container. From a nutrient destruction point of view, uni-
formity of heating translates to less destruction of nutrients for the same rise in average temper-
ature, making microwave processes advantageous. However, the uniformity of heating
becomes worse with time. For large containers, that need to be heated longer, substantial non-
uniformity of heating can be expected near the end of heating.

CONCLUSIONS
An exponentially decaying spatial heat generation model successfully predicted liquid temper-
atures inside a cylindrical container in a microwave cavity. Transient temperatures at a location
were almost linear. The uniform radial temperatures and the stratification show the effect of
flow and mixing for a low-viscous fluid. The slowest heating points were located at the bot-
tom of the container.

REFERENCES
1. Lentz, R. R. 1980. On the Microwave Heating of Saline Solutions. Journal of Microwave
Power 15 (2) : 107 -111.
2. Kee, R. J., C. S. Landram, and J. C. Miles. 1976. Natural convection of a heat generat-
ing fluid within closed vertical cylinders and spheres. J. Heat Transfer. 198: 55-61.
3. Swami, S. 1982. Microwave heating characteristics of simulated high moisture foods. M.
S. Thesis, University of Massachusets.
4. Datta, A. K., H. M. Prosetya, and W. Hu. 1992. Mathematical modeling of batch heating
of liquids in a microwave cavity. J. Micro. Power and Elect. Ener. 27(1):38-48.
5. Prosetya, H. M. and A. K. Datta. 1991. Batch microwave heating of liquids: An experi-
mental study. J. Micro. Power and Elect. Ener. 26(3): 215-226.
 Top View

r
10.1 an
,

". "
..I' ~,~. ~ .? ...
...
. ' .. ' . ' .. ' .

Elevation

t' :;
Fig.I. Locations of temperature measurements inside the container.

50 H.6.Scm

~""
10 R-5.1 em

45 a·----·--a-·h .--..... ~" ..... ~ 8 : ''''

<.,) 180Secoods
6 6 ! '
J
; - 40
a ..~~.~..~·~··~···----~·-~~~-~--~4~-~.----
120Secoods
..--~ 4 60 Seconds
···
"'~35
~ 30 1:.~..~.~·~~·~----••-----1.~--~.~------ 2
.
,~
./~
60 seconds
:-0- Experlneotai
_ theofelcal .... Th«IrtIicIII
2S~______, -______r -____-,_--
__~~_
·mM_~r'_ 0'r__~~~~__~~~__~
o 2 4 6 8 25 30 35 40 45 50
Radius (em) Temperature ( C )

Fig. 2. Radial temperature proftles at different Fig. 3. Axial temperature profiles at dif-
times showing an almost uniform radial ferent times during heating showing
heating. a stratification with increasing
temperatures toward the top.

: : l ::: ::
u U

'1 co.•
.2

;.
u U
<II
at I '
-
= .--- -- aJ
I!
~II
.-----
-- E
~ll
0

---- -....-..
E
CD
>
I-
"
.,.,. ... "
... ~ II

II
.. -rune (Sect
110
U
» ~ ..
Temperature ( C )
.~ "
Fig. 4. Transient slowest heating pont temIr Fig. 5. Transient volumetric distributions of
eratUTe values useful for microbial temperature showing the increase in
lethality estimations. spread over time dee to non-unifonn
spatial heating.
 MEASUREMENT OF EVAPORATION COEFFICIENT OF WATER
DURING VACUUM COOLING OF LETTUCE

ARMANSYAH H. TAMBUNAN
Faculty of Agricultural Technology, Bogor Agricultural University, Indonesia
YASUHISA SEO, YASUYUKI SAGARA
HIROSHI MORISHIMA, YOSHINORI KAWAGOE
Department of Agricultural Engineering,
Faculty of Agriculture, The University of Tokyo, Japan

ABSTRACT

Experimental investigations had been carried out to measure the evaporation coefficient of
water from lettuce undergoing vacuum cooling, applying an evaporation boundary-layer
model developed for the free surface of pure liquids. The value of evaporation coefficient
for lettuce was found to be dependent on mass flux from the leaf surface, and about 4 times
lower than that for the free surface of water. The effects of operating pressure to the coef-
ficient were not appeared definitely under the experimental conditions.

INTRODUCTION

Vacuum cooling is one of precooling methods, commonly used for leafy vegetables. The
product is cooled down by the consumption of latent heat of evaporation of water due to the
low pressure condition of the cooling chamber. The advantage of the method is its rapid
cooling rate. However, the cooling rate is limited by heat and mass transfer, i.e., evapora-
tion rate of water from the product's surface and inner tisssues. Detail knowledge on the
the evaporation coefficient is important in analyzing the vacuum cooling mechanism which
is essential for the improvement of the method. Those were not appeared yet in literatures.
Objectives of this study were to measure evaporation coefficient of water from let-
tuce undergoing vacuum cooling by applying evaporation boundary-layer model, and dis-
cuss the effects of processing parameters on it.

THEORETICAL MODEL

Based on the kinetic theory of gas, evaporation rate of water from the surface of a substance
to its vapor region under low pressure condition is equal to\,2),
 329

(1)

where, wis evaporation rate (kg/s), a is evaporation coefficient (-), A is effective evapora-
tion surface area (m 2), p is pressure (Pa), T is temperature ( K), with subscript s and v for
saturation and vapor region respectively. The model is called here as Evaporation Bound-
ary-Layer Model. The driving force of the evaporation process is the lbpor Pressure Dif-
ference (VPD), which is the bracketed part on the right hand side of equation (1).
The model states that evaporation rate is proportional to the VPD with the propor-
tionality constant a, i.e., the Evaporation Coefficient. The evaporation coefficient can be
considered as a correction factor of the apparent evaporation rate obtained by experiments
to the theoretical one.

EXPERIMENTAL APPARATUS AND PROCEDURES

APPARATUS
An experimental vacuum cooler was specially constructed to meet the purpose of the exper-
iment. The vacuum cooler was comprised of a vacuum chamber, coldtrap system and a
vacuum pump, equipped with instruments for data acquisition system.
Mass of the sample during the process was measured with an electronic balance
specially designed for operation under low pressure condition. Temperature and pressure
were detected with thermocouples made of 0.1 mm diameter copper-constantan wires and
capacitance type diaphragm manometers respectively. During the cooling process, coldtrap
temperature and pressure of the vacuum chamber were controlled at a constant level. Re-
trievement of all data and controlling process were done automatically with the help of a
personal computer, once every 5 seconds.

PROCEDURES
According to the model, the evaporation coefficient can be determined provided that evapo-
ration rate, effective evaporation surface area and VPD, are experimentally measurable
under a quasi-steady-state condition. The evaporation rate was calculated from the measured
mass, while the evaporation surface area was measured with an immage analyzer and cor-
rected with a previously determined correction factor to get the effective value.
Temperature of the evaporating surface was measured and assumed to be in satura-
tion, while its pressure was calculated from vapor relation at saturation condition. Temper-
ature and pressure of the vapor region was measured at a point 1 em above the evaporating
surface. The position of measurement was fixed referring to a preliminary experiments on
vapor temperature distribution inside the vacuum chamber. The VPD was calculated using
those temperatures and pressures with equation (1). The evaporation coefficient was calcu-
lated from the experimental data obtained within the period of quasi-steady-state which was
satisfied after pressure of the vacuum chamber was controlled at a constant level.
Experiments were performed on distilled water and lettuce. The distilled water was
considered as free surface of water and used as a standard to justify the eligibility of proce-
dures used.

RESULTS AND DISCUSSION

THE EVAPORATION COEFFICIENT

Evaporation coefficient of free surface of water and lettuce, as well as the related parame-
ters at various chamber pressure are shown in Thble 1. Evaporation coefficient of free
surface of water was found to be about 0.004, which was fairly consistent with those
330
reported in literatures3). The consistency indicated that measurement and calculation proce-
dures used in this study can be considered to be essentially acceptable and provide adequate
value of the evaporation coefficient of lettuce.

TABLE 1
Evaporation coefficient of free surface of water and lettuce
at various chamber pressure

Chamber Pressure (Pa)

Parameters 613 I 800 I 1066
W L W L W L
---------------------------------------------------------------
Surface area (m 2 ) 0.05 0.08 I 0.05 0.08 0.05 0.07
Surface temp. (oC) 0.0 0.6 3.4 4.2 6.9 7.9
Vapor temp. (oC) 0.0 5.5 4.6 6.1 7.9 10.8
w (my/.s) 10.0 3.9 8.9 3.7 9.4 2.3
VPD (Pa/K /2) 2.1 2.1 2.6 2.2 2.4 2.4
a (10- 3 ) 5.2 1.5 3.6 1.3 4.2 0.8

Note : W = free surface of water

L = lettuce

The table shows that evaporation coefficient of lettuce was about 0.001, which was
about 4 times lower than that of free surface of water. The lower value indicated its de-
pendency on mass flux from the product, i. e., water from inner tissues of lettuce, with
higher resistances, was also evaporated after that adhered on the surface was evaporated
out, and its purity was lower than the distilled water. The effects of chamber's minimum
pressure on the evaporation coefficient were not appeared definitely under the experimental
conditions.

CONCLUSIONS

1. The evaporation boundary-layer model was found applicable to be used for determination
of evaporation coefficient of vegetables undergoing vacuum cooling.
2. Evaporation coefficient of lettuce was found to be 4 times lower than that of free surface
of water and dependent on mass flux from the product.
3. The effects of chamber's minimum pressure on the coefficient were not appeared defi-
nitely under the experimental conditions.

REFERENCES

(1) Atkins, P.w., Physical Chemistry (4th ed.), Oxford University Press, Oxford, 1990,
Chapter 24
(2) Maa, lR., Evaporation Coefficient of Liquids, In.Eng.Chem .. Fundamentals, 1967,
6(4),504-18
(3) Hickman, K.C.D., Torpey, W.A., Evaporation of Resting Water, Ind.Eng. Chern ..
Engineering. Design. and Process Development, 1954, 46(7), 1446-50
(4) Nakajima, M., Morishima, H., Seo, Y., and Sagara, Y., Actual Process Control and
Operational Conditions of Commercial Vacuum Cooling Plants and Its Improvement
Instructions (Japanese), Nogyo Shisetsu, 1992, 23(1), 25-32
 HEA T AND MASS TRANSFER MODELLING DURING BEEF CARCASS CHILLING
FOR QUALITY CONTROL

G.S. Mittal and P. Mallikarjunan

School of Engineering, University of Guelph, Ontario, C~'1ada.

ABSTRACT

This paper describes the development of a fmite clement two dimensi,-iOal heat and mass transfel
model for beef carcass chilling. The carcass was divided into 5 zones - round sirloin, loin, rib and
chuck. The model was validated against extensive data on temperature profUes and mass change
history, and is being used to develop accelerated beef processing and quality control criteria.

INTRODUCTION

The objectives of this study were to develop the heat and mass transport models for predicting
the temperature and moisture profUes during beef carcass chilling, and to validate the model using
observed temperature and moisture profUes. .

MODEL DEVELOPMENT

The carcass was divided into 5 zones, round (or hip), sirloin, loin, rib and chuck; and carcass
cross sectional structure was assumed uniform within a zone. The governing equations for heat
and mass transfer in 2-dimensional cartesian coordinates are:
~(pC 1) = ~(k aT)+~(k aT)+q (1)
at P ax J:ax ay Yay

aM = ~(D aM)+~(D aM) (2)

at ax "'ax ay may
To solve equations 1 and 2, the initial Conditions were T =To and M =Mo for t ~ O. At the
surface, the conductive heat transfer is equal to the sum of the heat removed by convection and
-(k aTn +k aTn , = h(T -T\-S,D T aM (3)
J: ax J: Yay or t ~ oJ r tT-'v at
the heat removed due to latent heat of vaporization. At the surface, the mass transfer by diffusion
332
is equal to the convective mass transfer to the surrounding air due to vapour pressure difference
between the carcass surface and the surronding air, and is given by:
-D p (aM n +aM n )
1IId'&% ayr = h III
(P -P\
II aI
(4)

The variational method was used to formulate the problem, and using Green's theorem, the
equation was simplified. The Crank-Nicolson central difference method was employed for
marching through time. The model was solved using Fortran-77 on the Numerical Intensive
Computing (NIC) facility of the University of Guelph. The input to the program are the finite
element layout details for each section, initial temperature and moisture contents for each node
and observed ambient temperatures and air relative humidities (10 min interval). The output
printed by the program are the temperature at the desired nodes (according to placement of the
probes) and total carcass mass for every 10 min interval over the period of chilling.

The cross lIectional details were then traced on an acetate paper showing fat, muscle, and bone
portions. The geographical information system, PC ARC/INFO (Environmental Systems Research
Institute Inc., Redlands, CA, U.S.A) was used to generate the fmite element layout of different
carcass sections. The data for temperature and mass histories during chilling was collected. The
grading information for the carcass was: animal age 15 months, 160 kg carcass side mass and 4
mm fat cover thickness i.e. grade AI. The beef physical propenies such as: thermal conductivity,
specific heat capaci¥, and density were based on composition (1). Surface heat transfer coefficient
(h.) was 20 W/(m .K) at an air velocity of 0.5 mls and air temperature of O°C, which was
calculated by using relationship between Nu, Re and Pr for flow across a vertical plate (2).
Surface mass transfer coefficient was determined by using (~ / (h m Lv» = 64.7 Pa/K for air
velocities between 0.5 and 10 mls (3).

RESULTS AND DISCUSSION

The simulation results are shown in the Figures 1 and 2. The temperature histories at the
geometric centres of each zone are shown. In Fig. 1, the initial increase in simulated temperature
in round from 40°C to 44°C was due
S0r---------------------------------~
to the heat released due to ATPase
Round Sirloin Rib reaction. The temperature in deep
"
o o
Obsefved
Simulaled round was over predicted, whereas in
rib and loin sections, the temperature
was under predicted between 6 and 24
h post monem. The time required to
chill the deep round to 5°C was 49 h.
The temperature at the geometric
centre of the longissimus dorsi muscle
(between rib and loin sections) was
lowered to 0 to 2°C at the end of
-10 '----"""'-------'------'-____' - -__-'--__- ' -__---L_ _--1
chilling. The lowering of temperatures
o 5 10 15 20 25 30 35 40 in rib and loin sections were rapid
Time. h when compared to round, sirloin, and
chuck sections. The temperatures in
Figure 1. Temperature profiles for medium chilling rate. rib and loin sections were lowered
from around 40°C to SoC in 10 to 12
h, by that time, the temperature in
 333
other sections were between 15 and
20°C. This was due to the smaller
cross sectional areas of the loin and
rib compared to others. In Fig. 2, the
simulated total mass closely followed
the observed total mass after 5 h post
mortem. The total mass loss by the
end of chilling was 3.58%

Figure 1. Mass h~~tories for different chilling rates

can be used to develop criteria for ac

celt!rated beef processing after incorporating the models for the meat qUality kinetics. Time
required to chill beef carcass and mass loss during chilling can be estimated for process
development.

LIST OF SYMBOLS

p= density, kg/m 3; Pd = density of dry matter, kg/m 3

Cp= specific heat, J/(kg.K); Dm = moisture diffusivity, m2/s
hm = surface mass transfer coefficient, kg of water/(Pam 2.s)
~= surface heat transfer coefficient, W/(m2.K)
kx'~= thermal conductivities in x and y directions, W/(m.K)
Lv = latent heat of vaporization, J/kg; M = moisture content, in dry basis
nx' ny = direction cosines
Pa = partial vapour pressure of air, Pa; Ps = partial vapour pressure at the surface, Pa
q = heat generation due to ATPase reaction, W/m 3
Sf = shape factor, m
T = temperature, K; Ta = air temperature, K; Ts = surface temperature, K
= time, s; x, y = cartesian coordinates. m

REFERENCES

1. Choi, Y. and Okos, M.R., Effect of temperature and composition on the thermal properties
of foods. In Food Engineering and Process Applications,. Vol. 1, eds. M. Le Maguer and
P. Jelen. Elsevier Applied Science Publishers, London, 1986, pp. 93-101.
2. Kreith, F. and Black, W.Z., Basic Heat Transfer, Harper and Row Publishers, New York,
1980.
3. Daudin, J.D. and Swain, M.V.L., Heat and mass transfer in chilling and storage of meat.
J. Food Engg., 1990, 12, 95-115.
CORN DRYING MODELLING THE QUALITY DEGRADATION

F. Courtois(l), A. Lebert(2), J.C. Lasseran(3)

and J.J. Bimbenet(1)(2)
(1) ENSIA, 1 Avenue des Olympiades, 91305 Massy, France.
Tel: 1 69 93 50 50 - Fax: 1 69 2002 30 - E-mail: ENSIlO@Calvacom.fr
(2) INRA, 1 Avenue des Olympiades, 91305 Massy, France.
(3) ITCF, Station Experimentale, 91720 BoigneviIIe, France.

ABSTRACT
Corn is the second largest agricultural produce in France. Nearly 60% of this production
is exported (primarily towards the EEC). Thus, its wet-milling quality has become an important
criterion since it is strongly affected by the artificial drying which occurs just after harvesting.
To preserve the wet-milling quality of the dried corn as well as the energy and the drying rate, a
model based simulation tool has been elaborated. For convenience, quality degradation and
drying sub-models have been uncoupled. Experimental data were then collected during two
campaigns of experiments and a second order reaction has seemed to fit correctly to the results.
Due to the numerical adjustment, simulated results seem quite in agreement with commercial
trials.

INTRODUCTION

Corn is the second largest agricultural produce in France. Nearly 60% of this production
is exported (primarily towards the EEe). But before being sold, grains have to be artificially
dried to 0.176 standard moisture content. This operation which occurs just after harvesting is
energy costing. But the main problem is to treat harvested corn as soon as possible to avoid
mold. Thus dryer operators have a natural trend to increase the air temperature knowing that it
allows better production capacity.

But high capacity is rarely compatible with high quality. Usual criteria to evaluate corn
quality is the 'Turbidity Test' which is well related to the grain wet-milling ability. This test,
measuring the amount of proteins in dried com, is quite a good indicator of the thermal past of
the dried grains. Lebras [1] has shown clearly that wet-milling quality of dried corn is strongly
affected by high drying temperature.
 335
Our purpose was to model the quality degradation in relation with drying conditions.
We have worked in a global modelling approach with a final goal consisting in a CAD software
for corn dryers including quality prediction.

Few authors have modelled simultaneously drying and quality degradation of food
products. Concerning quality and drying, we can divide studies in 3 categories:
-Explicative approach: experimental studies to understand biological and biochemical
phenomena.
-Semi-Representative approach: experimental studies with quantitative aspects such as
regressions.
-Representative approach: experiments are choosen to identify and validate the model.

Our purpose was more to represent correctly the experimental data than to have a complete
comprehension of the phenomena.

MATERIAL AND METHODS

We took our inspiration from the works of Nellist and Bruce [2] on their study about
germination viability and baking quality. This method allows a thermal stress at constant grain
temperature and moisture content. It involves an important assumption: quality degradation and
drying rate are independant. This point has been verified a posteriori.

It consists in heating, by immersion during a defined time, in a water bath controlled to

±O.l°C, small PET bags of maize (60 g approximatly) sealed under vacuum. We measured
quality before and after the immersion. To obtain a kinetic, we were obliged to use as many
samples as the number of graphical points desired.

We have verified, experimentally, with a thermocouple inserted in the middle of one

grain of a bag, that the difference of temperature between grain kernel and water bath became
neglictible in less than 5 minutes at 90°C. Further more, we have verified that the grain moisture
contents before and after the immersion were equals.

The quality criterion Q we have chosen is given by the "Turbidity Test", which we
converted in absorbance units for the model. This test measures, by spectrophotometry, salino-
solubles proteins still in the grain [1]. These proteins being thermo-sensitive, we obtain a
quantification of the heat damages suffered by the grain. The test description is in Le Bras
(1984). For practical uses, the results are expressed in Transmitance units (%T) :
- 0% T is the best,
- 100%T is the worst.
The test precision, in our case, is about ±5%T.

For the continuation, we have assumed that the grain temperature was equal to the water
bath temperature. Then, adjusting the model to the experimental results has permitted to
determine model the parameters.

RESULTS

First experiments were designed to study independantly the influences of the product moisture
content X, water bath temperature T and time of exposition t on the final wet-milling quality Q.
Figure 1 shows a 3D representation of the influence of X and T on Q after 1 hour.

Numerical treatments of data lead to the quality model:

(1)
336
where the model parameter KQ is related to X and T with a classical Arrhenius law.

1.0

0,8

Qf 0,6

0,4

0,2

100

T (OC)

Figure 1. Influence of corn moisture content X and water bath temperature T on final wet-
milling quality after one hour exposure.

Figure 2 shows that experimental and simulated kinetics in a drying case are in good agreement.

0,7

0,6
T=80°C, X=O,162
a
. 0,5

..
u
c: 0,4

..
.Q
!)
.Q 0,3

'" 0,2
t (s)
0,1
0 1000 2000 o 4000

Figure 2. Comparison between drying experiment (with uncertainties) and model prediction.

CONCLUSION

A methodology has been developped to model corn quality degradation during drying. When
mixing quality and drying models, the model can predict wet-milling quality of dried corn with
an error less of 10% for more than 90% of the industrial data.

REFERENCES

1. Le Bras, A., Influence des conditions de sechage sur la qualite amidonniere du mals.
Perspective AWcoles. 1989, hors-serie, pp. 42-55.

2. Nellist, M.E. and Bruce, D.M., Drying and cereal quality. Aspects of Applied Biology,
Cereal Quality, 1987,15,439-455.
 PUFF DRYING OF FOODS IN A FLUIDISED BED

N.C. SIDLTON AND K. NIRANJAN

Department of Food Science and Technology,University of Reading, Whiteknights,
PO Box 226, Reading, Berkshire RG6 2BG, UK

ABSTRACT

Puffing of potatoes has been found to be dependent on the fonnation of a case hardening layer.
This has been tested on cubical samples using penetration experiments. In the fluidised bed
heat transfer has been found to be very rapid and volume shrinkage - which nonnally occurs -
has been avoided, resulting in improved rehydrational properties.

INTRODUCTION

Puffing is a process involving the release or expansion of gas within a product either to create
an internal structure, or to expand or rupture an existing structure (0. One of the most common
examples of a puffed product is popcorn. The puffmg of vegetables is an intermediate step:
puffing is interposed between 'pre-drying' and 'finish drying' stages. Described here is the
puffing of potatoes at around atmospheric pressure in a fluidised bed salt.
Puffmg of vegetables depends on the formation of a case hardening layer during an initial
drying step. Puffing occurs as a result of forced diffusion across this case hardening layer,
induced by the internal pressure. Puffing is followed by a low temperature finish drying stage.

METHODS

Penetration of the Case Hardening Layer

It follows that the fonnation of a case hardening layer is critical to puffing. The toughness of
the case hardening layer was therefore measured in tenns of penetration force using a Stevens
CR Texture Analyser. This was set up using a penetration depth of 1 mm at a speed of 5 mm
per minute. A cone (of angle 60·) was used to penetrate the surface of the potato. Cubes (10 x
10 x 10 mm) were prepared from the pith of the potato and dried, with three cubes comprising
one sample, and three readings being taken from each cube. Four different drying temperatures
were examined, these being 50·,80·,90· and llO·C. Bulk average moisture content of the
samples at the time of testing was also determined.
338
Puffing in a Fluidised Bed
The fluidised bed consisted of a Quick Fit cylindrical glass column (0.16 m diameter) packed
with salt particles to a depth of 0.16 m. The bed was fluidised by hot air entering at a
temperature of 130·C; the air velocity was maintained at 1.5 ms- I . Ten cubes (10 x 10 x 10
mm) were prepared from the pith of the potato and pre-dried at 90·C for 60 minutes. These
would then be placed in a wire mesh box and puff dried in the fluidised bed for a period of
between 10 seconds and four minutes. No pretreatment of the samples, such as sulphiting, was
carried out.
Heat Transfer: A type T thermocouple was inserted into the centre of a single pre-dried
cube. This cube was then placed in the fluidised bed for a period of four minutes and the
temperature was recorded using a data acquisition system every three seconds.
Moisture Loss and Volume Changes: Ten pre-dried cubes were puff dried in the fluidised
bed for increasing time steps ranging from 10 - 240 s. The volume was measured after cooling
the samples to 20·C by using a displacement method. The volume ratio was obtained by
dividing the volume after puffing by the volume before puffmg.
Rehydration Ratio: Rehydration was carried out by boiling the puffed and unpuffed
samples in water for 8 minutes and comparing the rehydration ratios (the weight after
rehydration divided by the mass of dry matter).

RESULTS AND DISCUSSION

Penetration of the Case Hardening Layer

Figure 1 shows that initially the penetration force required falls with the moisture content. This
reaches a minimum at a moisture content corresponding to 65% (wet basis) after which the
penetration force increases with decreasing moisture content.
It was therefore deduced that puffing should not be attempted until the moisture content
dropped below 60% to ensure the formation of a case hardening layer. Under the experimental
conditions this corresponds to a minimum pre-drying time of 60 minutes at 90·C. It can be
seen from Figure 1 that development of the case hardening layer is independent of temperature.
Development seems to be dependent on the moisture content of the cube of potato.

800.----r======il 1.2-r---------~
140
ot.
• 80°C
'+ + 50°C o
1.1 • 120

~ I. IC. IC

t-
• 80
o
x Volume 60
--Temp.
O~~--~~--~~--~ '~__~==~==~+40
o 1 234 5 6 100 200 300
Bulk Mean Moisture Content Time (s)
Figure 1: Stevens Readings versus Moisture Figure 2: Heat Transfer and Volume Changes
Content in a Cube of Potato during Puffing
 339
Puffing in the Fluidised Bed
The process has been found to be most efficient with puffing being carried out for three
minutes. Residence times in excess of four minutes tend to lead to excessive browning. Other
researchers have used longer residence times with different fluidised bed systems (2).
Moisture Loss: Moisture loss commences approximately 10 to 20 s after the samples have
been introduced. The moisture lost subsequently varies linearly with time. The moisture loss in
the puffing stage lasting 180 s is small in comparison with the total moisture removed during
drying.
Heat Transfer: Heat transfer to the centre of the potato cube is very rapid: from an initial
internal temperature of 48·C just before introduction into the fluidised bed, the bed temperature
is reached after 84 s; see figure 2. This suggests that heat transfer is not the rate limiting step.
Volume Changes: As can be seen in Figure 2, on introduction to the fluidised bed a
substantial loss of the volume ratio occurs. Recovery of this loss does not appear to begin until
90 seconds have elapsed. The increase in the volume ratio occurs after the internal temperature
has risen above 60·C, and continues to rise until the bed temperature is attained. Thus there
would seem to be two distinct regions in the process of puffing: (a) where the temperature
increases and (b) an isothermal region. It is therefore possible to retain the original volume
despite significant moisture loss. Further details can be found in (3).
Reconstitution: The puffed samples have been found to rehydrate faster than unpuffed
controls; Typical rehydration ratios are 2.93 for the puffed samples and 2.16 for the unpuffed
ones.

CONCLUSIONS

The puffing of fruits and vegetables is essentially an intermediate structural modification step
interposed between normal drying stages; little loss of moisture occurs during puffing.
However the volume shrinkage accompanying conventional drying can be avoided. Volume
development takes place after the first 90 seconds of the puffing step. Loss of moisture from
the food piece seems to progress linearly with time; it is not lost in a flash!

REFERENCES

I.Payne FA, Taraba JL, Saputra D, (1989), A review of puffing processes for the expansion
of biological products, J. Food Eng., 10, (3), 183-197.
2.Brown GE, Farkas DF, DeMarchena ES, (1972), Centrifugal Fluidized bed; Blanches, dries
and puffs piece form foods, Food Tech., 26, (12), 23-28.
3.Shilton NC, Niranjan K, (1993), Physical changes induced by intermediate puffing of
potatoes during drying, To be published.
 ESTIMATION OF THE EFFECTIVE MOISTURE DIFPUSIVITY
FROM DRYING DATA. APPLICATION TO SOME VEGETABLES.
C.T.KIRANOUDIS, Z.B.MAROULIS, D.MARINOS-KOURIS, G.D.SARAVACOS

Department of Chemical Engineering, National Technical University,

Zografou Campus, GR-15773 Athens, Greece

ABSTRACT

The effective moisture diffusivity as a function of temperature and material moisture

content can be estimated using a model fitting procedure to drying data. A model is proposed
which considers the basic heat and mass transfer mechanism, incorporating the effective
moisture diffusivity as a function of temperature and material moisture content. The parameters
are estimated by regression analysis of the total model, simultaneously on all experimental data.
The proposed methodology was applied to experimental data of four vegetables: potato, onion,
carrot, and green pepper. The experiments involved three thickness levels, five temperatures,
three relative humidities, and three air velocities.

INTRODUCTION

Moisture diffusivity is an important property involved in modeling of drying, rehydration

and storage offoods (1,2).
No standard method exists for experimental determination of moisture diffusivity. Several
methods have been proposed in the literature, most of which have been developed primarily for
polymeric materials (3). The estimation of moisture diffusivity from drying experiments is a
challenging method. Various methods of analysis have been proposed(4), and a generalized
procedure (numerical solution-regression analysis) has been discussed by Marinos-Kouris and
Maroulis (1).
This method uses an experimental drying apparatus and estimates the heat and mass
transport properties as parameters of a drying model, which is fitted to experimental data (5, 6,
7, 8,9).
An important advantage of this method is the determination of moisture diffusivity as a
function of material moisture content and temperature. This can be obtained by introducing the
relevant relationship to the fitted drying model.
An outline of the above method is presented here and the results of application to some
vegetables are discussed. These results are also compared with the results obtained using a
simplified method (4).

MATERIALS AND METIIODS

An experimental drying apparatus was used, in which the air passes through the drying
material and the air humidity, temperature, and velocity are controlled, while the material
 341
moisture content and temperature are monitored.
A total number of 400 experiments were performed for 4 vegetables (onion, pepper,
potato, carrot), with 3 different particle dimensions (5, 10, and 15mm) at 5 air temperatures
(60, 65, 70, 75, and 80 0 C), 3 air velocities (3,4, and 5m/s) and at air humidities ranging from
6 to 22g/kg d.b.
A mathematical model involving simultaneous heat and mass transfer is considered. Mass
transfer inside the material is assumed to be the controlling phenomenon. Heat conduction is
assumed infinite, while the surface heat and mass transfer coefficients are estimated from the
literature (1). The effective moisture diffusivity is assumed to be a function of material moisture
content and temperature, and it is described by the following equation:

D =Do exp(-XolX) exp(-Tom [1]

in which D is the moisture diffusivity, X the material moisture content (dry basis, d.b.), T
the material temperature and Do, Xo, and To are adjustable constants.
The total model is fitted to material moisture content and temperature simultaneously,
using a non linear regression analysis procedure. Parameters Do, Xo, and To are obtained in
this way (9). The simplified method used for comparison, estimates the drying constant and,
from it, the moisture diffusivity.

RESULTS AND DISCUSSION

The results of the proposed method are presented on Table 1. The standard deviation
between experimental and calculated values of material moisture content, obtained by the fitting
procedure, is low and, consequently, the total model (including Equation [1]) is considered as
satisfactory .
The resulting moisture diffusivity as a function of material moisture content at two
temperatures for the products examined, is presented in Figure 1. It is concluded that the
moisture diffusivity is nearly constant at high moisture content, but it decreases sharply as the
moisture content decreases below lkg/kg db.
The simplified method for the estimation of the moisture diffusivity yielded different
values for each temperature and material moisture content. Some of these values, for two
temperatures, are also presented for comparison in Figure 1. The data are scattered and the
temperature effect is not obvious, but a linear dependence on material moisture content could be
considered.
Comparing the two methods, it can be said that they agree only on the order of magnitude
obtained. The integrated procedure of the 'Numerical Solution - Regression Analysis' method
(1) guarantees a smoothing effect on raw data and proposes a reliable function for the
dependence of material moisture content and temperature on moisture diffusivity.

TABLE 1.
Parameter estimates for the equation D = Do exp(-XolX) exp(-Toff)

Vegetable Do (m2/s) To(K) Xo (kg/kg d.b.)

onion 3.72 7. 11 x 103 8.63xlO-2

pepper 7.33xlO-3 4.71xl03 1.05xlO- 1
potato 2.94xlO-7 1.57x103 6.72xlO-2
carrot 4.37xlO-7 1. 65x 103 8.06xlO-2
 342
10~----,,--~~~~~~----.

~~H~~~~~~~~~~~~~~~~L~~~~~~~~~~~~~~:~~~t~~~~~~~~~~~~~~.r~~f~~~~?~
. . . . . . . . . . . . :. . . . . . . J.::::::::::::: ........... j........
~.......
·········f······················
......... ......................
";'

. . . . . . . . . . . ~., ·~·,.(j·:?:*·~·~·=f······················
/ . tti ~ i

(XlO~m'/:) =:~~-~!j~~rf~~~I i Generalized model

....... ····.····r····················j"""············· -----6O·c
i i ---80·C
0.1 -+--C-.~-i-~~--+--~~r-~""""""'"
0.01 0.1 10 100 0.01 0.1 10 100
X (kg/kg d.b.) X (kg/kg d.b.)

Figure 1. Effective Moisture diffusivity (D) of potato and carrot

REFERENCES

1. Marinos-Kouris, D. and Maroulis, Z.B., Transport properties for the drying of solids. In
Handbook ofIndustrial Dtying, 2nd ed., ed. AS.Mujumdar, Marcel Dekker, New York,
1993, in press.

2. Saravacos, G.D., Mass transfer properties offoods. In Engineering Properties of Foods,

eds. M.ARao and S.S.H.Rizvi, Marcel Dekker, New York, 1986, pp. 89-132

3. Frisch, H.L. and Stern, S.A, Diffusion of small molecules in polymers. In CRC Critical
Reviews in Solid State and Materials Science, 1983,2(2), 123-187

4. Moyne, C., Roques, M. and Wolf, W., A collaborative experiment on drying beds of
glass spheres. In Physical Properties of Foods-2, eds. R Jowitt et aI., Elsevier Applied
Science, New York, 1987, pp. 27-54

5. Bertin, R, Delage, P. and Boverie, S., Estimation offunctions in drying equations.

Dtying TechnoI., 1983,2,45-59

6. Mullet, A, Berna, A., Rossello, C. and Pinaga, F., Drying of carrots. Dtying TechnoI.,
1989, 7, 641-661

7. Karathanos, V.T., Villalobos, G. and Saravacos, G.D., Comparison of two methods of

estimation of the effective moisture diffusivity from drying data. 1. Food Sci., 1990, 55,
218-223

8. Kiranoudis, C.T., Maroulis, Z.B. and Marinos-Kouris, D., Drying kinetics of onion and
green pepper. Dtying Techno!., 1992, 10,995-1011

9. Kiranoudis, C.T., Maroulis, Z.B. and Marinos-Kouris, D., Model selection in air drying
of foods. Dtying Techno!., 1992, 10, 1097-1106
 EFFECT OF FREEZING CONDITIONS ON VAPOR
PERMEABILITY OF FREEZE-DRIED FOODS

KOZO NAKAMURA; KOZO MATAGI, TOSHIAKI ODAWARA

AND TERUHIKO HOSHINO
Department of Biological and Chemical Engineering,
Gunma University, Kiryu, Gunma 376, Japan
* Department of Agricultural Olemistry, Faculty of Agriculture.
University of Tokyo. Yayoi. Bunkyo-ku. Tokyo 113. Japan

ABSTRACT

The soybean proteins solution was freeze-dried by conductive heating the

frozen sample conductively from the bottom of the vessel (back-face heating).
The drying rate was analyzed by the uniformly retreating ice front (URIC)
model to evaluate the vapor permeability in the dried layer. The vapor
permeability was affected by the rate of freezing and the levels of supercooling
since they could control the structure of ice crystals, that is, the channels of
vapor flow after the sublimation of ice crystals. The vapor permeability was
inversely proportional to the square root of the freezing rate when the freezing
was accompanied by the weak supercooling.

INTRODUCTION

In Knudsen flow the vapor permeability in the porous material is proportional

to voidage and pore size, while it is inversely proportional to tortuosity.
Therefore the rate of freeze-drying of liquid foods depends on the solid content
and on the freezing condition which produces a certain structure of the frozen
sample. The process of freezing is important in the freeze-drying of such as
instant coffee and tea, and the freezing phenomena in foods itself is an
interesting and unknown phenomenon of phase transition. In this study the
soybean proteins solutions were freeze-dried and the effect of freezing
conditions on the dtying rate were investigated by measurement of the vapor
permeability.

MATERIALS AND METHODS

The soybean proteins (AJIPRON HS2) with protein content 86.5 % were
dissolved in water at 70 'C in about 30 min, and 10 % protein solution was
344
prepared. This sample solution was filled in the cylindrical vessel (I. D. 48
mm x 20 mm high), and it was about 17 mm in depth. The sample was cooled
by the thermo-module attached to the bottom aluminum plate at the desired
freezing rate to reach the final temperature of -30 'C. The temperature
distribution in the sample was recorded by the six thermocouples (wire
diameter 0.2 mm) embedded vertically in the center of the sample. The
temperature of the sample surface was also kept constant with the upper unit
cooler.
The drying was conducted at the constant vacuum pressure and at the
constant temperature (-10 'C) of the bottom aluminum plate, the heater.

RESULTS AND DISCUSSION

Two typical freezing curves are shown in Figs. l(a) and (b) where the super-
cooling occurs weakly and strongly. As seen in Fig. 1, the growth of ice,
which is initially absent in the solution, occurs after nucleation. The nucleation
depends on the sample and the freezing rate, and the delayed nucleation results
in supercooling. The nucleation becomes rapid after the onset of ice formation
(a few second), and there is a rapid release of latent heat of freezing to warm
the supercooled region towards 0 'C. During the continuation of ice formation,
the temperature remains constant. Therefore the length of the plateau is a
measure of the freezing time probably to determine the size of ice crystals. And
the size of the ice crystals can be distributed in the sample, from the finest
formed above the cooling plate to the large dendritic ice formed near the sample
surface. The rate of freezing and the intensity of supercooling determine the
size of ice crystals, which produces the structure of the dried sample with a
certain vapor permeability.
Figure 2 shows the time course of temperature distribution, drying rate
and water content of the drying sample. In this case the sample was frozen at
the freezing rate of -10 'C/hr with a weak supercooling. The drying rate
decreased gradually with decrease in the water content because the resistance to
vapor permeation increased, and its decline became remarkably after the water
content reached about 2 g-water/g-matter. The drying rate of the gradually
decreasing period was used to calculate the vapor permeability by URIF

5 ~--~--~--~---r-' ~--~--~---r---r---'

~ 0 .-....~~~-..::-"'
~
B
r:
t1)
-5
0..
~ -10
E-
-5 'C/hr
-15~--~--~--~--~~
0 2 0 2
Time [lui Time [hr J
(a) (b)

Figure 1. Typical freezing curves with a weak and a intensive supercooling

 345
1.0 r : : : : - - - - ; - - - , - - - - r - - - - r - - - - r - - - - , 10
'i:'
.....C-~
1:l ~
ci:'
0 "
5 C).
"",00
Q)~

~~.2B
0:::0 0
:::::
Beating plate
~ -10 ~'---3---
~ 6
.....
::l 9
e
~
-20
12
0. height above
E heating plate [ nun 1
~ -30~-~----~----~---~----~~
o 2 4 6 8 10
Drying time [hr]
Figure 2. Drying curve

Figure 3. Correlation of freezing rate and vapor permeability

model [1].
Figure 3 shows the effect of freezing rate on the vapor permeability of
the freeze-dried protein samples. The vapor permeabilities for the samples
frozen with the weak supercooling were inversely proportional to the square
root of the freezing rate as shown by the solid line. The intensive super-
cooling resulted in the permeabilities, scattered under the line, and those for
the samples containing oil (3-5 wt %) did not depend on the freezing rate and
remained low level.

REFERENCE

1. Kumagai, H., Nakamura, K. and Yano, T., Agric. Bioi. Chern., 55, pp.
731-736 and pp.737-742 (1991).
 MOISTURE CONTENT DISTRIBUTION AND
MOISTURE DIFFUSION IN FOODS DURING HEATING

CHRISTINA SKJOLDEBRAND and KARIN THORV ALDSSON

SIK, The Swedish institute for food research, Box 5401, S-402 29 GOTEBORG,
Sweden

Background
"The mechanism of drying has been the subject of scientific study for more than 100
years, but it is still not thoroughly understood. External factors such as air temperature,
pressure, humidity and velocity are governed by relatively simple and well known laws,
but not the internal transfer of moisture." Although Van Arsdel wrote this in 1973, the
internal transfer of moisture is still not fully understood. One of the reasons for this is that
until now there has been no way to measure the l<?cal water content inside a sample during
a drying or heating process. A special fibre optic instrument has now been developed at
SIK to measure the local water content, however. This instrument has been used in this
project, in which the water transport in meat during heating has been studied, particularly
in terms of the influence of meat fibre direction.

Hypothesis
Water that moves in a direction parallel to the meat fibres can move straight forwards
along the fibres, but water that moves in a direction perpendicular to the meat fibres has
to cross or move around the fibres or fibre bundles. If the straight path is 1, the path
through the hexagonal system is then 1,244, which means that transport of water should
be about 20% slower in the case of transport perpendicular to the fibres compared to the
case of transport parallel to the fibres.

MA TERIAL AND METHODS

Varying fibre orientation
The effect of fibre orientation on heat and moisture transport in whole bovine muscles was
studied. To avoid shrinkage of the samples during the experiments the whole bovine
muscles were precooked at 75°C. The approximate water content of the samples after
precooking was 65%.

Cooking procedure
The samples were placed in a conventional oven with good temperature stability and no
forced convection. The temperature of the oven was 175°C and the samples were heated
until they reached a center temperature of 100°C. The samples were placed in the oven
on a loose synthetic net, which allowed the air to come into contact with the sample and
prevented heating of the sample via conduction.
 347
Temperature and moisture measurements
The water content was measured in a small spot in front of the optic fibre and during the
heating the fibre is placed in the centre of the sample.
The temperature in the oven and in the samples was measured with copper-constantan
thermocouples.
RESULTS
Heat and water transport
Figure 1 shows a typical diagram of the changes in water content and temperature over
time during the heating of a sample.

Different fiber orientation

The heating times for the samples with different fibre orientation show clear differences.
In the case of heat transport parallel to the meat fibres the heating time is shorter than in
the case of heat transport perpendicular to the meat fibres. See figure 2 and table 1.

DISCUSSION
Heat and water transport
At the start of the heating process, water moves towards the centre of the sample, due to
temperature gradients. The temperature gradient itself does not give rise to any water
movements, but it affects other physical properties that could start water transport. The
volume of liquid water increases with temperature, by an average factor of0,18*lO-3/oC.
With a temperature gradient of lOOC the volume difference would be 1 to 1,016. This is
almost negligible, but since water is almost incompressible, this would make a small
contribution to the water transport towards the centre. The water moving towards the
center also transports heat, so that the heating is speeded up.

After a while, a concentration gradient is built up, the water transport stops and the
heating is slowed down. When the temperature gradient disappears, nothing keeps the
water in the centre and the concentration gradient starts to even out the concentration
differences. This means that water moves from the centre towards the surface, but the
water that comes from the centre is colder than the surroundings it enters. The heat
transfer is then slowed down for a while, but as the water transport decreases the
temperature rises more quickly, until it reaches 100°C, where all energy is used for
evaporation.

DilTerent fibre orientation

The results show that heating is faster in the case of heat transport parallel to the meat
fibres than in heating perpendicular to the meat fibres. This is probably due to the water
transport. The water is heated up near the surface and moves towards the centre where the
surroundings are colder. Since the water transport is faster in the case of parallel transport,
as can be seen in the results, the heat transport also gets faster in this case. In the
hypothesis, the difference in path length for the water in the two cases was 1 to 1.24,
which means that the water transport in the perpendicular case should be 19% slower.
From the results, the difference in water transport is approximately 1 to 1.27, which
corresponds to a 21 % slower transport. The deviation from the hypothesis could be due
to uncertainty in the measurements, the differences that always exists between different
meat pieces and the assumption that the fibres are arranged according to a hexagonal
system.
348
REFERENCES
1. Van Arsdel, Copley and Morgan, Food dehydration, AVI, 1973.
2. Booton, P.E., Harris, P.U. and Shorthose, W.R., Factors influencing cooking losses
from meat., J. Food Sci., 1976,41:1092.
3. Crane, J., Mathematics of diffusion, Oxford Univ. Press, 1975.
4. Godsalve, E.A., Davies.E.A, Gordon, J., Effect of oven conditions and sample
treatment on water loss of dry cooked bovine muscle. J. Food. Sci., 1977, 42: 1325.
5. Locker, R.H. and Davies, G.J., Cooking loss in beef. The effect of cold shortening,
searing and rate of heating; Time Course and Histology of Changes During Cooking.
J. Sci. Fd. Agric., 1974,25:1411.
6. Skjoldebrand, C., Thorvaldsson, K., Andersson, C.G., Torstensson, P.A., A new fiber
optic sensor for on line measurement of moisture in food. SIK-Rapport, 1991, Nr. 583.

TABLE 1

Transport perpendicular to the meat fibres

Sample Heating time Rate of moisture Rate of temperature

(minutes) increase at the increase at the start
start (%/min) COC/min)

1 85 0.5 2.0
2 75 0.5 2.8
3 80 0.6 2.4
4 75 0.6 2.7

Average 80 0.55 2.5

Transport parallel to the meat fibres

Sample Heating time Rate of moisture Rate of temperature

(minutes) increase at the increase at the start
start (%/min) COC/min)

1 55 0.8 3.1
2 70 0.6 2.6
3 70 0.6 2.8
4 65 0.8 2.9

Average 65 0.7 2.9

 MASS TRANSFER IN OSMorIC PROCESSES

z. YAO, M. Le MAGUER
Department of Food Science
university of Guelph
Guelph, Ontario
Canada N1G 2Wl

ABSTRACT
Processes in which pieces of fruits or vegetable materials are contacted with
a concentrated solution of sugars, polyols and or salts give rise to
simultaneous movement of water and solutes from and into the material. This
paper presents a mathematical model and its solution using a finite element
numerical technique. Taking into account the fact that cells embedded in a
continuous cell wall matrix can act as a source or sink for water within the
structure leads to the appearance at an early stage of a strong bulk flow
velocity which can counteract the diffusion of solutes from the solution.
This result observed experimentally is also demonstrated by the model.

NOTATION
A area
B membrane width of the model
C concentration
D apparent diffusion coefficient
J trans-membrane flux
H half thickness of a tissue slice
K mass transfer coefficient at the surface
m total number of solutes present in the system
N flux referred to a fixed frame
t time
v* partial molar volume
v volume average velocity
z macroscopic direction of transfer
SUBSCRIPT
i species indicator
s solution

INTRODUCTION
Osmotic dehydration of food is a process in which the food contacts a solution
of high osmotic pressure. The process is carried out at low temperature and
does not need latent heat to remove water. Therefore, it results in high
quality products and low energy consumption. However, its application in the
food industry is restricted because the mass transfer mechanism has not been
totally understood. The objective of this study is to build a mathematical
model to investigate the mass transfer phenomena in osmotic processes
350
PHYSICAL AND MATHEMATICAL MODELS

The proposed physical model contains intracellular and extracellular spaces

that are separated by a semipermeable membrane (Fig. 1). The intracellular
space consists of many layers stacked together. Each layer can only exchange
mass with the extra-cellular space through the cell membrane.

Membrane

L
--------- '-----------'

Intracellular space Extracellular space

Figure 1. Physical model including intracellular and extracellular space that

are separated by a semipermeable membrane.

By making mass and volume balances in the extracelluar space, the

following equations were obtained:

(1)

m+l
a{v' A) (2 )
az =~

with the following initial condition:

Cj It = 0,% =% = constant (3)

and boundary conditions at the centre (z = 0)

aCj I - (4)
Tz t = t, % = 0 - 0

az It · t,% = 0
avO
0 (5)

and at the surface

or (6)
 351
RESULT AND DISCUSSION

The system of equations (1) through (6) was solved using a computer simulation
procedure in which a one dimensional finite element method was used. The
parameters used in equation (1) were obtained using the procedure described
by Toupin. (1986). The model was adjusted to the experimental observations
[2] and a good agreement was obtained. Furthermore the parameters obtained
were in the range of obtained literature values.
Fig. 2 shows the simulation results of a slice of red beet immersed in
60% of sucrose solution. The solid line represents the concentration profile
of sucrose when diffusion and bulk flow are both considered. The dashed line
represents the profile when bulk flow is ignored. Shrinkage of the tissue was
considered in both cases. The bulk flow prevents the diffusion of the solutes
into the tissue and sharpens the solute concentration profile near the
surface. There is little solute diffusing into the tissue at an early stage
of osmosis and the solute is located within a thin layer near the surface.
This indicates that osmotic dehydration should be terminated at an early stage
if solutes impregnation is undesirable.

- - diffilsion with bulk flow

0.8 ............. diffilsion without bulk flow

eI
0.6
d
a>:: 0.4

0.2
0.05

0.5 0.6 0.7 0.8 0.9

z/H

Figure 2. Simulated concentration profiles of sucrose in a slice of red

beet tissue. Numbers on curves are values of Dt/H 2

CONCLUSION

Multi-component mass transfer in biological tissues immersed in concentrated

solutions was investigated. A mathematical model was obtained to incorporate
diffusion, bulk-flow, trans-membrane flux and shrinkage of the matrix. The
model was adjusted against experimental data and a good agreement was
obtained. Simulation results showed that the bulk flow component l.S
predominant and controls the penetration of solutes into the tissue at an
early stage.

REFERENCES

1. Toupin, C. J. 1986. Osmotically induced mass transfer in biological

systems: The single cell and tissue behaviour. Ph. D thesis, Food
Science Department, University of Alberta, Edmonton, Alberta, Canada.

2. Kohn, P.g. & J. Dainty. 1966. The measurement of permeability to

water in disks of storage tissues. J. Exp. Bot., 17(53), 809-21.
 THE SORPTION OF WATER IN THE AGRICULTURAL MATERIALS

TAKAHARU KAMEOKA, KAZUO HORIBE

Laboratory of Process Engineering and Machinery,
Faculty of Bioresources, Mie university
Kamihama-cho, Tsu-city, Mie 514, Japan

ABSTRACT

The attempt at a statistical treatment of water sorption by

foods was made, based on an approximate method for deriving the
sorption isotherm from the probability density function of
adsorption energies. The use of weibull distribution leads to
the Dubinin-Astakhov(DA) equation and it was shown that this
isotherm is applied successfully to both the non-porous and
porous foods.

INTRODUCTION

The state of water in foods changes from single layer adsorption

to capillary condensation gradually, and this shows that co-
operative sorption takes place. Therefore adsorption potential
theory seems to be very effective to porous food.[1]
In this work, DA equation [2] was derived based on control
band method. [3] Then adsorption data of glucose, maltose and
sucrose and a corn starch were prepared to confirm that this
equation is also applicable to non-porous food.

DERIVATION OF DA EQUATION BY CONTROL BAND METHOD

Food was modelled to be composed of absorbent and water. Total

number of sites at temperature, T, is expressed as Nas(T), and
the number of sites with adsorption potential A as Ns (A,T).

f:
Average surface coverage at T and A is shown as
8( A, 1) = f( A, 1)dA ( 1)
and
f(A,1) = Ns(A, 1)/ Nas(T) (2)
 353
Now the probability of desorption under the adsorption potential
change, A(A, n,
is def ined as
A(A, T) = a In O(A, nl aA (3)
using eq. (3), net differential heat of adsorption, Mal' and
differential entropy of adsorption, ASal,are expressed as
Mal = A - TASal (4)
l1Sal = (J(T)jA(A, T) (5)
where (J(T) is the coefficient of thermal expansion of water
Assuming a Weibull probability density function and integrating
eq.(l) leads to
O(A, T) = exp{-(A / AeY} (6 )
where n is a distribution order, Ae is a characteristic energy.

EXPERIMENTAL RESULTS AND DISCUSSION

sorption data of crystalline glucose, maltose and sucrose, and

a corn starch were obtained using static method at 10"C.
Each crystalline saccharide sorbed a negligible amount of
water at water activity, aw, below each aw determined from each
saturated saccharide solution at 10"C while at higher than this
aw, each series of data fell on a smooth curve which was
calculated by each saccharide-water solution model. Sucrose data
and amorphous data from ref. [4,5] were shown in Fig .1. A
series of data was on the smooth curve extrapolated by the
sucrose-water solution curve. DA equation showed good agreement
with the amorphous data in the range of aw from 0 to saturation

~
250
• 25°C(Amorphous, Ref.[ 4,5])
.0
"0 200 - - - -6- - - 10°C(Crystalline)
?i
........,
....c 150
---Cl--- 25°C(Crystalline,Ref.[ 4,5])
....c
Q)
~

0 Calculated by Eq.(8)
u 100
Q)

....
~
:::J - - - - - Calculated by Eq.(7)
C/)
'0 50
~

0 -- --
0 0.2 0.4 0.6 0.8 1

Water activity

Figure 1. Equilibrium moisture content of sucrose.

354
point. This DA equation is expressed as
M = , R=O. 997
171.5exp{-(-9.451naw t 5} (7)
For the data over the range of aw from 0 to 1, DA equation is
shown as
M = 948exp{-(-201.31naw 3 } , t
R=0.991 (8 )
Fig.2. shows the adsorption data of corn starch. DA equation
satisfactorily described experimental data over the range of aw
from 0 to 1. The equation is as follows.
M=42.7exp{-(-1.681naw t 36 } R=0.991 (9)
As seen from Fig.l and Fig.2, DA equation is applicable to the
non-porous food with adsorption potential space, such as
amorphous saccharides and corn starch. In the case of food, n
value in DA equation means the porous degree and is usually less
than 1, which is different from the definition of original DA
equation. According as n approaches to 0, the food with such n
has more porous structure.

50
--
..q 40
"'0
o Experimental data

~
......... Calculated by Eq.(9)
+-'
C
30
C1>
.....
C
0
() 20
~
:J
+-'
II) 10
·0
::E
0
0 0.2 0.4 0.6 0.8 1

Water activity

Figure 2. Equilibrium moisture content of corn starch.

REFERENCES
1. Kameoka, T., Morishima Sokhansanj, S. and Jayas, D., Food
Engineering and Process Applications, Vol.l, Elsevier Applied
Science Publishers Ltd., 1986, pp.201-209.
2. Dubinin, M.M. and AStakhov, V.A., Adv. Chern. Ser., 1971, 102,
pp. 69-85.
3. Young, D.M. and Crowell, A.D.,Physical Adsorption of Gases,
Butterworth & Co. Publishers Ltd.,1962, pp. 247-275.
4. Makower, B. and Dye, W.B., Agricultural a,nd Food Chemistry,
1956, 4(1),pp. 72-77.
5. Chinachoti, P. and Steinberg, M.p.,Journa,l of Food Science,
1984, 49, pp. 1604-1608.
 The Diffusion of Seasoning Substances in Various Slurries
-Application of Nondestructive Measurement Using Ultrasonic Technique-

Toru TORII* and Sachiko ODAKE**

* Faculty of Agriculture, University of Tokyo
**Faculty of Living Science, Yamanashi Woman's Junior College

ABSTRACT

An apparatus which applies the ultrasonic technique was constructed to measure the
diffusion coefficients of sodium chloride in corn starch sols. At a frequency of 1 MHz,
the ultrasonic velocity was observed to increase proportionally with the concentration of
sodium chloride and of starch. By employing the ultrasonic technique, it becomes
possible to detect the concentration changes continuously and nondestructively.
Furthermore, the accurate diffusion coefficient of substances in fluids may be obtained.
The diffusion coefficients of sodium chloride in 1%, 3% and 5% corn starch sols were
obtained as 1.12, 1.07 and 0.94 x 10- 5 cm2/s, respectively.

INTRODUCTION

With the recent development and proliferation of processed foods in our lives, basic
information such as water or moisture movement, and the diffusion coefficients of
seasonings or additives is needed in order to manufactured well controlled products. The
behavior of seasoning is especially important as it has a direct influence on taste and
changes the physical properties of materials. For example, taste intensity in the mouth
depending on the flux degree of the seasoning substances to the surface of the tongue,
and the diffusion coefficients were a sufficient parameter to express the phenomena.
In order to obtain the diffusion coefficients, concentration changes must be mea-
sured. Furthermore, measurement is more useful when it is carried out continuously and
nondestructively. In this experiment, the ultrasonic technique was applied to check the
compounds in many kinds of solutions.
An apparatus to measure the sodium chloride concentration changes using ultrasonic
technique was constructed and the diffusion coefficient in hydrocolloid sol was studied.

MATERIALS AND METHODS

Preparations of Samples for Diffusion.

Samples were prepared in pairs, one containing sodium chloride and the other
without. The concentrations of corn starch were 1%, 3% and 5% and the concentration
of sodium chloride was 1.0 M against the water in sols.

Measurement of Diffusion Coefficients

The schematic diagram of the test apparatus for measuring the concentration changes
during diffusion is shown in Figure 1. It comprised an acrylic resin cylinder of 160 mm
356
diameter, and 230 mm height. It was separated into two compartments at the center with
a micro-pore filter (ABS resin). The filter was changed according to the viscosity of the
test fluid and a suitable pore size was selected. The natural frequency of the ultrasonic
sensor was 1 MHz with a diameter of 20 mm, and two sensors were located opposite
each other on the curved wall of the cylinder. The lower component was filled with fluid
containing sodium chloride and the upper with fluid without sodium chloride,
consequently sodium chloride diffused through the filter. The fluid in the compartment
was stirred with a rod connected to variable speed motors (5-8Orpm). The measurement
was carried out in a thermostat controlled environment at 25±0.1 °c. Until both fluid
concentrations equalized, each concentration change in the compartment fluid was
observed continuously.

Moto

Ultrasonic Sensor

~--~
Ir;;'i~;;;'i~a0
... ' .. ecce

Figure 1. Schematic diagram of test apparatus

The following equation was proposed to calculate diffusion coefficients [4]:

D
1
= -Bt
In
_c....
l __c....
l'
c2-c2' (1)

where D is the diffusion coefficient, t is time, c 1, c l' are concentrations of sodium chlo-
ride at initial time, and c2, c2' are concentrations at time t. B is the constant which de-
pends on the apparatus.

RESULTS

Ultrasonic Velocity in the Sodium Chloride·Corn Starch Sols.

The ultrasonic velocity in the sodium chloride-corn starch sols were shown in Figure 2.
It was influenced by both sodium chloride and corn starch. With the increase of concen-
1525r-----------------------------------~

1520

~1515
~1510
8
~1505
1% 3% 5%
1500 ................................................... ~........... ~.~.~~.~.~ ........... ::::::~::::: ........... ..

1495~----~------~------~----~------~
o 0.05 0.1 0.15 0.2 0.25
Concentration of salt (M)
Figure 2. The ultrasonic velocity in the sodium chloride-corn starch sols
 357
tration of sodium chloride, the ultrasonic velocity gains linearly and it was observed that
the presence of com starch increased the ultrasonic velocity.

Diffusion Coefficients of Sodium Chloride in Corn Starch Sols

The diffusion coefficients of sodium chloride in com starch sols are shown in
TABLE 1. As the concentration of com starch increased, the diffusion coefficient
decreased.

TABLE 1
Diffusion coefficients of sodium chloride in com starch sols
-------------------------------------------------------------------------------------------------------
com starch concentration (%) 0* 1 3 5
diffusion coefficient xlO-5 (cm2/s) 1.48* 1.12 1.07 0.94
--------------------------------------------------------------------------------------------------------
* the value in the reference [5]
DISCUSSION

In the primary investigation, the sodium chloride solutions, with a known diffusion
coefficient were checked, and a good agreement was observed between the obtained
value using this apparatus and the reported value [5], consequently, the performance of
this apparatus was verified. The results also indicated the possibility of obtaining
diffusion coefficient with precision by monitoring the difference in the concentrations.
The viscosity of the fluid influenced the diffusion of the substances in it [1]. The re-
sults of this study also indicated the influence of viscosity on the diffusion coefficient.
The apparatus constructed in this study can be applied to the obtaining of diffusion
coefficients of various substances in fluid samples.

CONCLUSION

An apparatus for the measurement of diffusion coefficients of sodium chloride in com

starch sols was constructed applying the phenomenon that ultrasonic velocity is influ-
enced by the concentration of substances dissolved in fluid. By employing the ultrasonic
technique, it becomes possible to obtained diffusion coefficient conveniently.

REFERENCES

1. J. L. Kokini, K. Bistany, M. Poole and E. Stier, Use of mass transfer theory to

predict viscosity-sweetness interactions of fructose and sucrose solutions containing
tomato solids. 1. Texture Studies, 1982, 13, 187-200

2. C. Javanaud, Applications of ultrasound to food systems. Ultrasonics, 1988, 26,

117-23

3. P. T. Dunlop, B.T.Steel and J.E.Lane, Experimental methods for studying diffusion

in liquids, gasses and solids. Physical Methods of Chemistry. Vol. I, PartIV. Wiley-
Inter Science 1972.

4. Ultrasonic technical Handbook, Nikkan CO., Tokyo, 1989, pp.1260

5. R. A. Robinson and R.H.Stokes, Electrolyte Solutions, 2nd Ed., Butterworths,

London, 1959, pp.513-5
 DIFFUSION OF SALT IN DRY-SALTED FETA CHEESE.

S. Yanniotis, J. Zarmpoutis and E. Anifantakis

Dept. of Food Science and Technology, Agric.univ.of Athens,
Iera odos 75, Athens 118 55, Greece.

ABSTRACT

The diffusion of NaCI in dry-salted feta cheese for short times

was studied. The "pseudo-Jiiffusion" coefficient of t~n srlt in
the cheese moisture at 13 C was found equal to 2.3*10 m /s.

INTRODUCTION

Feta cheese is a soft cheese made from pasteurized sheep's milk

or from a mixture of sheep's and goat's milk. It has at least 43%
fat (on dry basis) and 56% maximum moisture content (wet basis).
The traditional method of salting of feta cheese is by adding dry
coarse salt on the surface of the cheese.
The diffusion rate of NaCI plays an important role on the
final quality of cheese. It has been studied mainly in brine-
salted hard cheeses. Geurts et al.(1974) published a detailed
investigation on the diffusion of NaCI in Gouda-type cheese.
Diffusion of NaCI in dry-salted cheeses has been studied to a
limited extend. Guinee and Fox (1983) studied dry and brine
salting in Romano-type cheese. Georgakis (1973) studied the
salting of feta cheese but he did not analyze the process in
terms of diffusion.
In the present work the diffusion of sodium chloride in dry
salting of feta cheese was studied.

MATERIALS AND METHODS

The cheese sample was prepared in our pilot plant from whole
sheep's milk with the following procedure. The milk was
pasteurized by heating it to 69°C and then cooled to 33°C. The
heating cycle lasted 14 min and the cooling cycle 21 min. At this
point starter culture (Str.lactis, Str.thermophilus and Lact.
bulgaricus) was added. After 5 min rennet was added. After about
50 min the curd was cut in pieces (2x2x2 em) and put ~radually
in hoops where it was left to drain for 24 h at 13 C before
salting. Pressure was not exerted on the curd. At this stage the
 359
moisture content of the cheese was 57.7% (wet basis) and the fat
content 50% (dry basis). For salting, coarse salt was put on the
surface of a flat screen in the form of a thick layer. On top
of this salt layer, three rectangular pieces of cheese (10x10x8
cm), were placed and left in a closed container at 13°C. All the
sides of each cheese block were wrapped with a thin plastic film
to prevent moisture evaporation, except the side that was in
contact with salt. wi th this arrangement, diffusion in one
direction was obtained, while the salt concentration at the
surface was constant (the cheese surface was all the time in
conduct with NaCl crystals). After 24 and 76 hours from the
beginning of the salting process, one cheese block at a time was
removed. A layer of 1 em from the side surfaces was discarded and
the remaining block was cut with a thin wire in slices parallel
to the salted surface. The thickness of each slice was 2 mm. Each
slice was analyzed in duplicate for NaCl concentration and
moisture content. The NaCl concentration was determined by
measuring the chloride content with the International Dairy
Federation standard method (IDF f 17A:1972). The moisture content
was determined with the IDF 4:1958 method. The results reported
here are the average of the two measurements.

RESULTS AND DISCUSSION

The diffusion of NaCI into the cheese can be described by Fick's

second law. With the initial condition that the salt
concentration is initially Co throughout and the boundary
condition that the surface is maintained at a constant
concentration C; during the experiment, the solution for a semi-
infinite body, assuming constant diffusivity (D), is given by
Crank (1956) as:

(1)

The coefficient D can be calculated by minimizing the

deviation defined by Eq.(2}, that is the average of the relative
percent difference between the experimental C values and the
theoretical C values that are predicted by equation (1) for
various D values.
.. 100"" abs(Cop -Ctheor) (2)
%DeVUltlOn--~
N Cop

In this way, the "pseudo-diffusion" coefficient of NaCI ~fO

tfe moisture in the cheese was calculated to be equal to 2.3*10
m /s with 8.4% deviation using the experimental concentration
profiles of NaCI which were established after 24 and 76 hours of
sal ting. The equilibrium concentration (C;) used in Eq. 1 was
equal to 19.8 g NaCl/g moisture in the cheese. This value was
obtained experimentally by measuring the NaCI concentration in
the moisture of the surface layer of the cheese after 240 hours
of salting. The initial value (Co) used was 0.007 g NaCI/g
moisture. Turhan and Kaletung (1992) obtained a value for D of
360

about 3*10. 10 m2/s for white cheese salted in br:.~dle2at 12.5°C.

Guinee and Fox (1983) obtained a value of 2.8*10 m /s for dry
salting of Romano-type Cheese at 20°C.
In Figure 1 the experimental points are plotted togeth~fD
w~th the theoretical values predicted with Eq.1 for D=2.3*10
m Is. It can be observed that the agreement between the experi-
mental and the predicted values is good. Thus, for short times
where the semi-infinite body solution is valid, the total salt
uptake can be predicted using the above value for the diffusion
coefficient.
9.2

•
L
::J
)I(

• 9.15
~
.oj
0
E <> 24 h
D * 76 h
"
-f --Theor·
•
0
Z
D 9.1
~

...
I:
0

• 9.95
~

L
~

•
I:
U
I:
0
0

9
9 IS 19 15 29 25 39 35
Dist.nce, mm

Figure 1. Concentration profile of NaCI in cheese.

Acknowledgements: The financial support of the Greek Secretariat

of Research and Technology is gratefully acknowledged.

REFERENCES
1. Crank J. "The Mathematics of Diffusion". Oxford Uni versi ty
Press, London, 1956
2. Georgakis S. A. Studies on the manufacture of Greek Feta
pickled cheese 1. Relation between the amount of salt used
and duration of salting. Milchwissenschaft, 1973, 28(8),
500-502.
3. Geurts T. J., P. Walstra and H. Mulder. Transport of salt and
water during salting of cheese. Part 1: Analysis of the
processes involved. Neth. Milk Dairy J., 1974, 28, 102-B
4. Guinee T. and P. Fox. Sodium chloride and moisture changes
in Romano-type cheese during salting. J. Dairy Res., 1983,
50, 511-518.
5. Turhan M., Kaletung G., Modelling of salt diffusion in white
cheese during long-term brining. J. Food Sci., 1992,
57(5), 1082-1085.
 EXTRACTION RATE FOR SOLUBLE SOLIDS DURING COFFEE BREWING

S. OSHITA, A. HASHIMOTO, T. MIYATA* and T. OKADO*

Fac. of Bioresources, Mie Univ., Kamihama-cho, Tsu-city, Mie 514, Japan
*KONDO UNYUKIKO CO. ,LTD. , Ichiriyama-cho, Kariya-city, Aichi 448, Japan

ABSTRACT

Variations of soluble solids concentration in the coffee extract as a

o 0
function of time were measured over the range from 40 C to 90 C under a
constant grind/water proportion for each of three different particle
sizes. The concentration histories were well expressed by the rate equa-
tion of the 2nd order. The rate constant was then interpreted in relation
to the Arrhenius equation and both frequency factor and activation energy
were determined for each granule size. Results suggested that the de-
crease in granule size could effectively increase the rate constant.

INTRODUCTION

Knowledge of coffee extraction rate is important to the coffee industry

since it would provide a more objective measure of efficient brewing
condition. However, there are not so many reports [1,2] relating to
quantitative treatment of coffee brewing. The present study was designed
in order to determine the rate equation explaining the coffee extraction
process. The influence of the particle size and brewing temperature on
the extraction rate has been studied.

MATERIALS AND METHODS

Brazil Santos coffee was used. It was roasted and ground under a given
condition. Mean diameters of the granule particle were 0.68 mm, 0.78 mm
and 0.86 mm. The grinds of 10 g was added to 500 ml of distilled water
at given temperature and it was stirred for 45 minutes at the rate of 400
r.p.m •• Separation of the extract from grounds was achieved by the use of
disposable syringe to which a filter, whose pore size was 5 microns, was
connected. Filtration time was less than 5 seconds. As volatile elements
occupy only about 1 % of the total weight of soluble solids [3] and the
amount of insoluble solids is negligible, the concentration was expressed
in grams of soluble solids per 100 ml coffee extract. It was measured by
362
drying the samples of extract at 105 °c for 24 hours. The grind/water
proportion was kept to 10 g(grind)/500 ml(water) and the effect of tem-
perature and particle size on the extraction rate was examined.

RESULTS AND DISCUSSION

Variations of the extract concentration as a function of time were meas-
ured at 6 levels of temperature for each of 3 particle sizes. Measure-
ments were repeated three times. Four kinds of rate equations:

C= Ce(l-exp(-kt)) (1)
C= C'+C"(l-exp(-kt)), C'+C"=Ce (2)
C= C'(l-exp(-k't))+C"(l-exp(-k"t)), C'+C"=Ce (3)
C= Ce(1-1/(Cekt+1)) (4)

were examined, where Ce and C are the concentration at equilibrium state

and at time t, respectively and k is the rate constant. The superscript
of C and k means the value of each component. Parameters in each equation
were determined by the non-linear least square method. The extraction
process was well expressed by the rate equation of the 2nd order because
the equation (4) gave the smallest standard deviation. Fig. 1 shows the
typical difference in regression curves drawn by equation (1) and (4).
Table 1 shows the parameters determined by the form of equation (4).

Table 1
Equilibrium concentration Ce and rate constant k for each particle size

Temp. Particle diameter Particle diameter Particle diameter

°c R=0.68 mm R=0.78 mm R=0.86 mm
Ce k Ce k Ce k
-------------------------------------------------------------------------
90 0.530 2.40 0.540 1.00 0.531 0.68
80 0.525 2.33 0.529 0.99 0.527 0.62
70 0.514 1.95 0.500 0.88 0.504 0.58
60 0.503 1.58 0.494 0.80 0.486 0.52
50 0.503 1.55 0.483 0.63 0.479 0.46
40 0.450 1.41 0.464 0.50 0.458 0.40
----------------------:1-:1----------------------------------------------
Ce: [g/100ml], k :[s C ]

The rate constant was then interpreted in relation to the Arrhenius

equation and both frequency factor and activation energy were determined
for each particle size. The arrhenius plot of rate constant was shown in
Fig. 2. The rate constants were given as follows:

k= 93 exp(-11000/RT), Particle size=0.68 mm (5)

k= 46 exp(-11000/RT), Particle size=0.76 mm (6)
k= 18 exp(-10000/RT), Particle size=0.86 mm (7)

where R is the gas constant and T is the absolute temperature. Results

indicated that the brewing temperature contributed largely to determine
the value Ce and the particle size governed mainly the value k.
 363
0.6
90 deg.
E 0.48
0
0 A
A
.....
or-

m 0.36
c
...•...
.2
...c
II)
0.24
Particle size : R=O.78 mm
u
c
0 0.12 Solid line : rate eq. of the 2nd order
0
Dashed line : rate eq. of the 1st order
0
0 10 20 30 40 50

Time [min]
Fig.1 Regression curves of concentration variation

3 r-----~----~----~----~----~
Particle size

o
..... 1.4
en
.....
.....
.)t(.

C)
o 0.6

0.3 L . . - - - ' - - _ - - . l ._ _--.l._ _--.l._ _- - '

2.5 2.7' 2.9 3.1 3.3 3.5

(1/T)x1000 [11K]
Fig. 2 Arrhenius plot of rate constant
REFERENCES

1. MABEL C. MERRITT and BERNARD E. PROCTOR, Extraction rates for selected

components in coffee brew, Food Res., 1959, 24(6), pp.735-743.
2. A. VOILLEY and D. SIMATOS, Modeling the solubilization process during
coffee brewing, J. of Food Process Eng., 1980, 3(4), pp.185-198
3. M. SIVETS, Coffee processing technology, vol.2, The Avi publishing
co., Westport, Conn., 1963, pp.168
364

THE MEASUREMENT OF THE LOSSES OF SOLUBLE SOLIDS IN LEGUME SEEDS (LABLAT

purpureus) DURING THE PROCESS OF SOAKING AT DIFFERENT TEMPERATURES AND
IN THE PROCESS OF COOKING.
A. JADAN, M. Silva, M. Moscoso, Department of FOOD SCIENCE and Technology,
Faculty of Food Engineering, P.O. Box 334, Ambato, Ecuador.
INTRODUCTION
Cruz el al. (1978) point out Noor's statement that, in many regions of
the world, the importance of legumes as excellent protein sources in human
diet has been recognized.
In Latin America, beans have been a highly esteemed food to improve diets
based on cereals, since these and legumes form a nutritional complementa-
tion.
Beans playa key role, as the main protein and calories source, in
Mexico's and other Central America's diet.
Beans production and consumption in Ecuador were more important in the
past, hence the necessity to promote them.
LABLAT PURPUREUS, a legume commonly known as "Sarandaja" is cultivated in
Loja, the southernmost province of Ecuador and represents a variety of
high importance due to its protein content of near 25%.
Traditionally, dry beans are treated with a soaking process. It is
commonly believed that this process reduces the cooking time and improves
the product texture. However, it is questionable any improvement in the
beans nutritional value, due to nutrients losses in the soaking water
(Wang et al. 1979).
Studies have been carried out to determine the feasibility of certain
softening salts to reduce hard beans cooking time (Lopez et al. 1983).
The author's work was carried out to investigate the total hydration
conditions for the Lablat purpureus variety of beans; the soluble solids
losses during the soaking and cooking processes; the effect of hydration
on the cooking time; and the effect of sodium carbonate and bicarbonate on
the "Sarandaja" grains softening.
OBJECfIVES
GENERAL OBJECTIVE
To quantify the soluble solids losses in the Lablat purpureus bean and
to determine the optimal soaking conditions.
SPECIFIC OBJECTIVES
- To determine the protein and total sugars losses in various solvents.
- To investigate the temperature conditions for a minimum beans hydration
time.
- To determine the cooking times of hidrated beans at various temperatures.
METHOD
A design containing two factors (a x b) is applied.
Factor "a" with three levels
level a1: drinking water
level a2: sodium bicarbonate solution at 0.5%
level a3: sodium bicarbonate at 0.5% and sodium carbonate at 0.5% solution.
 365
Factor "b" with five levels:
level bl: 20 degrees Celsius
level b2: 35 degrees Celsius
level b3: 50 degrees Celsius
level b4: 65 degrees Celsius
level b5: 80 degrees Celsius
Beans were soaked in the three types of solvents and investigation took
place for each solvent under five temperatures.
Completed the soaking process for each case, samples, representing 10%
of the total material were dried at 30°C until reaching the seed initial
humidity level. Analyses for soluble solids, proteins, and total sugars
were carried out both in the grain and in the soaking solvent. In addition
pH and total solids were studied in the solvents.
After these analyses and for the total material, cooking of the hydrated
grains was conducted afterwards.
In addition, digestibility tests in vitro and phosphorus quantification
in the phytate acid, at the begining and at the end of the soaking process
were also conducted.
RESULTS
1. The beans hydration time is highly dependent on the type of solvent
utilized. The longest hidration time (35.4 h) corresponds to the
drinking water case, and the shortest (32.4 h) to the sodium bicarbo-
nate solution at 0.5%.
2. Similarly, the beans hydration time is significantly affected by tem-
perature. Increasing temperatures result in decreasing hydration times,
having registered 69.33 h for 20°C and 13.00 h for 80°C.
3. In the axb interaction, statistical difference is produced; that means
that the various solvents, given the experiment temperatures, produce
different results. The solvents salts allow the softening of the beans
in lesser time.
In all soaking cases, the amount of total solids increases with the
process duration and with higher soaking temperatures.
Regarding the "a" factor (solvents), the highest solids losses (18.53g)
correspond to the sodium bicarbonate at 0.5% and sodium carbonate at
0.5% solution; and the lowest (9.60 g), to the drinking water case.
Regarding the "b" factor (temperatures), the highest solids losses
(22.00 g) correspond to 80°C and the lowest (8.02 g), to the 35 °c
case. Significant statistical difference exists with the 80, 65 and
50°C cases, not so with the 20 and 35 °c.
SOAKING WATER pH
1. In the case of the "a" factor, a significant difference is given
between the various solvents, being the highest (8.62) for the sodium
bicarbonate al 0.5% and sodium carbonate at 0.5% solution; and the
lowest (5.86) for the drinking water case.
2. In the case of the "b" factor, the highest value (8.27) is reported
for the 80°C case, and the lowest (6.35), for the 35°C case.
PROTEIN LOSS
1. Regarding the "a" factor it exists a significant difference, reporting
the highest value (8.09 g) for the drinking water case; and the
lowest (3.69 g) for the sodium bicarbonate at 0.5% case.
2. Regarding the "b" factor, the highest protein loss (11.42 g) results
for the 20°C case, while the lowest, (1.09 g), for the 80°C case.
3. A significant difference in protein loss is due to the interacting
effects of "a" and "b" factors. The percentage of protein loss due
to the soaking process varies from 7 to 16%, corresponding to soaking
temperatures between 20 and 35°C, and soaking times between 2 and
366

24 hours, with water.

SUGARS LOSSES
1. The "a" factor records the highest loss value (3.00 g) for the sodium
bicarbonate at 0.5% and sodium carbonate at 0.5% solution' and the
lowest (2.41 g) for the sodium bicarbonate at 0.5%. '
2. Regarding the ''b'' factor, a significant difference exists between the
80°C case and the 35 and 20°C cases. Not SO, between the 65 and 50°C.
The highest loss value (3.28 g) was found for the 65 qc case and the
lowest( 1.94 g), for the 20°C case.
3. When considering the interacting factors "a" and "b" significant
differences become evident, when comparing the various cases.
PHYTIC ACID PHOSPHORUS
The percentage value for phosphorus in the phytic acid, in one case
which was measured, was 0.0955%.
PROTEIN DIGESTIBILITY
The percentage of in vitro digestibility is 67%.
CONCLUSIONS
It is possible to reduce the Lablat purpureus cooking times to normal
values through the use of sodium bicarbonate, in the soaking water, for
the ions exchange between the sodium bicarbonate and the calcium phytate.
The least solids loss occurs in the sodium bicarbonate at 0.5% solvent.
The optimal soaking temperature is 35°C, for which the cooking time
is 33 min.
The optimal temperatures found to activate the enzymes are between
35 and 65 °c.
The highest pH occurring in the process is 9, which causes the highest
loss of total solids.
The calcium phytate, inherent to this legume, constitutes the main
factor for determining the soaking and cooking times.
SUMMARY
1. A proximal analysrs is carried out with selected beans.
2. The beans are subjected to a soaking process in three kinds of solvents:
- drinking water;
- sodium bicarbonate at 0.5% solution;
- sodium bicarbonate at 0.5% and
sodium carbonate at 0.5% solution.
3. The soaking process is carried out at five temperatures:
20, 35, 50, 65, and 80°C. The hidration times for dry beans are
determined until weights are constant.
4. The samples, after soak, are dried in an oven at 30°C until they
reach the initial humidity. At the beginning and at the end of the
soaking process analyses are conducted to determine protein and total
sugars content.
These data serve to quantify the protein and total sugars losses after
soaking the beans at various temperatures.
In addition, analyses are conducted to determine pH and total solids
in soak water.
5. After the soaking process, cooking takes place in distilled water
at 92 °c.
 Evaluation of Steam Distillation of Roasted Coffee for
Quality Improvement of Soluble Coffee

NAOTO IMURA and OSAMU MATSUDA

Ajinomoto General Foods, Inc.
Sphere Tower Tennoz, 2-2-8 Higashi-shinagawa,
Shinagawa-ku, Tokyo 140 JAPAN

ABSTRACT

Steaming of roasted and ground coffee is an effective operation to deliver a regular

brewed coffee flavor to a soluble coffee by adding back the steam condensate to the
extract obtained by the steamed coffee. However, acids generated by the steaming
resulted in increasing sourness of the soluble coffee. This increase of sourness is
known as a factor that deteriorates consumer's preference. A screw conveyer type
continuous steaming apparatus minimized the acid generation and contributed to flavor
improvement. Distillation of the steam condensate and "Partial Steaming", wherein only
the Arabica portion of the blend was stripped to recover superior aromas, were
explored. The products made by the techniques were evaluated by a consumer test and
improvement of consumer preference was confirmed by a multivariate analysis.

INTRODUCTION

Steam distillation of roasted coffee in a percolator column prior to water extraction is

known as a technique to deliver a regular brewed coffee flavor to a soluble coffee and
to improve quality of the products [1]. This technique, however, increase sourness in
the cup of soluble coffee made from the steamed coffee due to the generation of
acids during the steaming process [1]. In this study, a continuous system was evaluated
in recovering volatile aromas from coffee and reducing the acid generation during
steaming for quality improvement. For further improvement of flavor, aroma
rebalancing by distillation of the steam condensate and "Partial steaming" (solely the
Arabica portion of the blend was steamed to recover superior aromas) were examined
as means of manipulating coffee flavor.
368

MATERIALS AND METHODS

Materials
Coffee beans were roasted to a 20° L value by colorimeter measurement and ground to
a 3,000J.1m mean particle diameter for being steamed. Colombian coffee and Ivory
Coast were representatives for Arabica and Robusta respectively.

Equipment
Roasted and ground coffee was steamed in the apparatus illustrated below. The main
vessel has 2,000mm in length and 125mm ID. The screw flight pitch is 100mm and the
shaft diameter is 40mm.

Condenser Coffee Feeder

Rotary Valve

Condensate Tank
D- Residual Coffee

Figure 1. Continuous apparatus for recovery of coffee volatiles

The steam condensate was evaporated with an ordinary continuous evaporator

with a plate type heat exchanger under 625 mm Hg vacuum for distillation of the
steam condensate.

Analyses
Acid titration: The acids in the steam condensate and coffee extract were
titrated by 0.1 N sodium hydroxide solution to an end point of pH 8.1 and described in
milli-equivalent /g dry coffee.
Organoleptic evaluation: Freeze dried samples to which the steam condensate
was added back were provided for flavor evaluation. Expert panelists rated the
intensities of the attributes in 13 degrees; aromatic impact, a "Floral" note (a desirable
and typical note for Colombian coffee), an "Earthy" note (a note like soil), harsh note
and sourness. Eighteen ordinary consumer panelists evaluated flavor of the coffee
samples in eleven attributes for preference and ten for strength with five degree ratings.
The ratings on the attributes by the consumers were analyzed by a multivariate analysis
(factor analysis) to characterize the flavor.

RESULTS AND DISCUSSIONS

Excess acid generation
Steam distillation of roasted and ground coffee in a percolator column resulted in
generating an extra amount of acids because the coffee was exposed to high
temperature for a long time as shown in Table 1. This generation of acids caused
increase of sourness in a cup of coffee made from the steamed coffee. The continuous
system enabled to recover as same amount of volatile aromas as the static bed
steaming in a short time as five minutes. The short residence time of coffee grounds
 369

resulted in minimizing the acid generation and consequently in reducing sourness in the
cup.

Table 1
Excess acid generation via steaming roasted and ground coffee in static bed and
continuous system in a counter current manner
(m equivalent / gm dried coffee )
Acids in Acids in Total Acids Acids in non Excess acids
Condensate Residual coffee steamed coffee Generated
Static bed*1 0.0562 0.136 0.192 0.128 0.064
Continuous*2 0.0414 0.136 0.177 0.179 - 0.002
*1 Steam pressure: 0.35 kglcm2, *2 Residence time: 5min.

Distillation and Partial Steaming (Steaming by Coffee Variety)

The freeze dried soluble coffee samples to which distilled steam condensate was added
showed a less harsh note, an undesirable aromatic, and less intensity of sourness by
expert panelists' evaluation. This correlated with the reduction of acid in the condensate
measured by titration. A process flow called "Partial steaming" was devised to obtain
steam condensate only from Colombian coffee (Arabica) portion of the blend for
superior aroma and to blend the residual coffee with non-steamed Ivory Coast
(Robusta) for good "body" and bitterness. The samples to which the "Partial steaming"
was applied showed a "Floral" note characterizing Colombian coffee, and less an
"Earthy" and harsh notes were perceived in the samples compared with the samples to
which steaming both Colombian and ivory Coast was applied. Six prototype samples of
freeze dried coffee were prepared with the combination of those techniques and
evaluated by consumer panelists. The samples evaluated were positioned on the
coordinate of two factors; "Preference of aroma / Strength of taste" and "Overall
preference" by plotting the factor scores for the samples. The samples to which
distillation and/or "Partial steaming" were applied were plotted at further positive side
on the axis that highly correlated with the "Overall preference". This was attributed to
the reduction of acids by distillation and to the preferable aroma, "Floral", derived
from Colombian coffee.

CONCLUSION

A continuous apparatus for steaming of roasted and ground coffee enabled us to

recover regular brewed coffee flavor in a soluble coffee. Sourness in a cup, which was
a disadvantage for steaming a static coffee bed in a percolator column, was also
reduced with minimizing acids generation during steaming. Both distillation and "Partial
steaming" were able to manipulate aroma and sourness of the coffee steam distillation
was applied, and were effective techniques to improve consumer's acceptance of a
soluble coffee.

REFERENCES
1. Imura, N. and Matsuda, 0., Volatile aroma recovery from roasted coffee by
steaming and its effects on soluble coffee products,. Nippon Sbokllbin Kogyo
Crakkaisbi, 1992,39, 531-535.
STUDIES ON THE RECRYSTALLIZATION OF ICE IN FROZEN FOOD SYSTEMS

MIN, S.G. and SPIEB, W.E.L.

Institute of Process Engineering, Federal Research Centre for Nutrition,
Engesserstr. 20, 76131 Karlsruhe, Germany

ABSTRACT

A simple approach to model recrystallization processes is presented which allows to describe

the recrystallization process quantitatively. In terms of quantity, present concepts of
recrystallization in food-like systems investigated consisted of 40 % sucrose solution plus the
hydrocolloids gelatine and Na-alginate. In this system larger crystals have been shown to
grow, while smaller ones disappeared. Presence of hydrocolloids did not prevent, but only
delayed recrystallization. From the results a mathematical relation, Xeq = k· r, was derived.
Using Arrhenius'approach for the rate controlling parameter k, the activation energy for the
recrystallization of ice crystals, according to the present results, depends on the kind of
hydrocolloid, storage time and temperature. The results have also shown that the viscosity of
the system has no direct influence on the recrystallization process.

INTRODUCTION

The quality of deep frozen food is largely dependent on the size of ice crystals in the frozen
matrix. In modem industrial freezing plants frozen food products are usually manufactured
under conditions which allow to produce relatively finely distributed ice crystals especially in
particulate food. This crystal structure changes during storage, however; larger crystals form
at the expense of smaller ones so that the crystal structure becomes coarser and the number
of crystals decreases [1]. The growth of crystals leads to irreversible damage to structure
elements, resulting in a softening of the products, increased drip losses, and, as a
consequence, reduced sensory quality.

MATERIALS AND METHODS

As matrices of pure ice present experimental problems for analysis, matrices of frozen
hydrocolloids were used in view of the the common practice of hydrocolloids being added to
food (e.g. ice cream) before freezing to inhibit or delay recrystallization. Besides, literature
attempts to explain the effect of hydro colloids on an ice matrix are rare [2], [3]. After some
 371
preliminary tests it was decided to use a 40 % sucrose solution plus hydrocolloids Na-alginate
and gelatine at concentrations of 0.2, 0.5 and 1.0 %. Samples of 8 III were transferred onto
slides with a depression of 200 Ilm, covered by a lid and sealed water-vapour tightly. The
samples were frozen at 243 K (Biot number> 10) and stored at temperatures of267, 263 and
259 K (± 0.1 K). Recrystallization was monitored under isothermal conditions by a
cryo-polarization microscope. To study the influence of viscosity on recrystallization, the
flow behaviour of the samples was measured at temperatures of 279, 276 and 270 K using a
rotation vioscosimeter (HAAKE RV3). Size analysis of the particles was evaluated using an
image analyzer.

RESULTS AND DISCUSSION

Recrystallization was observed during storage at all temperatures studied. Lager crystals
increased in size, while smaller ones disappeared to the effect that the total number of crystals
decreased. To watch the behaviour of two crystals in contact was interesting: at the site of
contact, a neck formed; then the two crystals grew together into one big ClYStal. Changes in
crystal size as a function of storage time, temperature and presence of Na-alginate and
gelatine is shown in figures 1 and 2. During storage time investigated, recrystallization was
not prevented, but only delayed by hydrocolloids. The delaying effect of the hydrocolloids
was stronger at higher temperatures than at lower ones. An explanation for this may be that
recrystallization depends on the amount of non-frozen water in the gel matrix. At higher
temperatures more free water is bound to the hydrocolloid than at lower temperatures.
Increasing ice proportions in the matrix are supposed to reduce the hydrocolloid effect.
Further studies are necessary to explain this phenomenon. Ice crystal size as a function of
storage time and temperature can be described by a relation also used in metallurgy.

(Eq 1)

where X"q = equivalent mean crystal diameter (Ilm), k = kinetic constant (Ilmlh n ), n =
exponent, and t = storage time (h).

From an Arrhenius equation

k = ko . exp -Ea/RT (Eq.2)

for the rate-controlling constant k, the activation energy Ea for recrystallization of ice in a
sucrose matrix with hydrocolloids could be calculated. The exponents n calculated
according to Eq. 1 and 2 the activation energy, resp., are shown in table 1. The experimental
results have shown that storage temperature and kind and concentration of hydrocolloids are
important parameters for stablizing crystal structures. However, above a certain hydrocolloid
concentration the effect on recrystallization is not improved any further. At sufficiently low
temperatures, the crystal structure changes to some insignificant extent only. At higher
temperatures, recrystallization is reduced in the presence of hydrocolloids. This effect is of
essential importance for the manufacture of foods consumed in frozen or semi-frozen
condition, as e.g. ice cream. Results of the rheological studies have shown that viscosity has
no direct influence on recrystallization processes. Different flow behaviours of the
hydrocolloid solutions are not reflected by the recrystallization pattern. To what extent our
372
experimental results apply to real food remains to be investigated. Ice crystals in real food
were measured by some working groups [4], [5]; in these experiments, however, not all
parameters of interest had been taken into account. Using the present approach,
recrystallization in simple systems can be described in a general way.

Table 1
Exponent n and activation energy Ea for recrystallization in a sucrose-hydrocolloid matrix.

Sucrose without Sucrose with hydrocolloids

hydro- Gelatine Na-Alginate
colloids
0.2% I 0.5% I
1.0 % 0.2% I 0.5% I 1.0 %
n 0.3913 ± 0.0421 0.3174 ± 0.0304 0.3246 ± 0.0425
Ea 68.66 ± 86.93 ± 115 .84 ± 69.85 ± 69.30 ± 83.80± 1 62.22 ±
[KJ/moIJ 3.53 4.15 1
5.53 1
3.32 3.60 1
4.05 2.24

200 200

alSO alSO
.:; .:;
.,.
:.or 100 JIOO

SO SO
-267K ~263K -»-2SI)K -267K ~263K -<>-2SI)K
O+-------~~----~------~
200 400 600 o 200 400 600
t[h] t [h]

Fig. 1. Recrystallization in sucrose-matrix Fig.2. Recrystallization in sucrose matrix

with 0.5 % Na-alginate with 0.5 % gelatine

REFERENCES

[1] Fennema, O.R, Powrie, W.D. und Marth, E.H., 1973; Low temperature preservation of
foods and living matter. Marcel Dekker, Inc., Now York
[2] Budiaman, E.R und Fennema, O.R., 1987; Linear rate of water crystallization as
influenced by temperature of hydrocolloid suspensions. J. Dairy Sci. 70: 534-546
[3] Muhr A.H. und Blanshard, J.M.Y., 1986; Effect of polysaccharide stabilizers on the rate
of growth of ice. J. Food Technol. 21: 587-604
[4] Martino, M.N. und Zaritzky, N.E., 1988; Ice crystal size modifications during frozen beef
storage. J. Food Sci. 53: 1631-1649
[5] Sy, S.H. und Fennema, O. R, 1973; Rate of recrystallization in beef liver tissue. Procee-
dings of the International Congress of Refrigeration (13 th, Washington) 3: 199.
 ICE STRUCTURE AND ITS CONTROL IN FROZEN FOOD GELS

OSATO MIY AWAKI AND SEOUNG-KWON BAE

Department of Agricultural Chemistry, The University of Tokyo
1-1-1 Yayoi, Bunkyo-ku, Tokyo 113, JAPAN

ABSTRACT

Ice structure of frozen agar gels was measured with two different methods: photographic
method and mercury porosimetry. Although the two methods gave the different absolute
values of the ice structure size, they are in good agreement in their trends of the dependency
on the conditions at the freezing. From the balance equations of heat and mass transfer, the
surface structure of the freezing front was analyzed and the mean ice structure size was
theoretically related to the mass-transfer rate of water molecule and the moving speed of the
freezing front, which represented the effects of the differences in the physical properties of
the sample to be frozen and in the conditions at the freezing. Ice structure was also affected
by solution structure. The effect of the coexisting materials such as sucrose, sodium
chloride, urea, and a surfactant (Triton X-l00) on the ice structure was examined.

INTRODUCTION

Freezing is one of the most important technique for the food preservation and also strongly
related to the unit operations such as freeze-concentration, freeze-drying, and freeze-
texturization. Freezing is a phenomenon with which phase change from liquid phase to solid
phase takes place. This phase change is only for water in most cases. As a result of
freezing, microstructure of ice crystal is formed. In this paper, the ice structure and its
control will be discussed in terms of its relationship with the conditions at the freezing and
solution structure by using agar gel as a model food system.

MATERIALS AND METHODS

Agar gel (3%) was frozen in a sample vessel composed of a cylindrical acrylic resin and a
copper cooling plate of lmm thick at the bottom. The upper and the side spaces of the
sample vessel were thermally insulated with polystylene foam. The sample in the vessel was
started to one-dimensional freezing by dipping the cooling plate of the vessel into the coolant
kept at a constant temperature. Temperature change in the sample was measured by
thermocouples. When freezing was completed, the sample vessel containing the frozen agar
gel was put into a freeze-dryer quickly to avoid thawing of the gel. The freeze-dried porous
374

gel was cut into slices at various distances from the cooling plate. The pore size distribution
of the sliced sample was analyzed photographically and by mercury porosimetry.
In the progressive freeze-concentration, 0.02% blue dextran (tracer) solution was
submitted to one-dimensional freezing at a controlled moving-speed of freezing front with
stirring, if necessary, and the partition coefficient of the tracer between the solid phase and the
liquid phase was measured.

RESULTS AND DISCUSSION

The mean ice structure size measured from the pore size distribution in the freeze-dried agar
sample was determined by the photographic method and the mercury porosimetry. The
results by the two methods are compared with each other as shown in Fig. I. Although the
absolute values of the observed ice structure size was different, they are in good agreement in
their trends of the dependency on the conditions at the freezing. In the previous paper!), we
proposed that the mean ice structure size (dp) is inversely proportional to the moving speed
of freezing front (u) as described by the following equation.
udplDw~ 1 (1)
where Dw is an apparent diffusion coefficient of water. The relationship between dp and u
shown in Fig.2 agreed well with the expectation from Eq.(1).
The relationship between the ice structure size and the structure of the ice-liquid
interface is investigated. When the freezing proceeds at a moving speed of the freezing front
of u, then the corresponding amount of ice must be formed at the interface, the structure of
which is not necessarily a flat plane but has a complex dendritic structure with a finite
thickness. The total rate of ice formation at the interface will be the product of the mass-
transfer rate of water molecule (k) and the surface roughness factor of the interface (s=real
surface area/apparent surface area). The surface roughness is the index of the dendritic
structure of the interface. From these theoretical considerations, the following equation is
obtained.
u=ks (2)
When Eqs.(1) and (2) are compared with each other, the ice structure after freezing is
related to the surface roughness of the ice-liquid interface. The ice structure size decreases
with increasing the surface roughness and vice versa.
80
E 0

-
~
....
>. 70
Q)
E
'00
0.... 60 0
0
a.
~
::J 50
~
Q)
E
40
~
Co
"'0
30L-~-L~--~~~~--~~~

60 80 100 120 140 160

d p by photographic analysis (j.tm]

Fig.1 Correlation between the mean ice structure size(d p)

formed in frozen agar gel measured by mercury
porosimetry and that by photographic analysis.
 Fig.2 Relationship between the mean ice structure size in frozen
agar gel and the moving speed of the freezing front.
Progressive freeze-concentration was carried out for the verification of Eq.(2) and the
apparent partition coefficient of the tracer (blue dextran) between the solid phase and the
liquid phase was measured. With a small surface roughness, a smooth surface area was
obtained and the freeze-concentration took place effectively. On the contrary, a rough surface
gave a poor freeze-concentration efficiency through a mechanical incorporation of the solute
molecule into the interface. Surface roughness was determined either by u or k through
stirring at the ice-liquid interface. Thus, ice structure could be controlled by u and k.
Ice structure might be also affected by solution structure. Therefore, the effect of the
coexisting mateials such as sucrose (water activity modifier), sodium chloride (water
structure maker), and urea (water structure breaker) was examined. The addition of these
materials at 5% level substantially increased the size of ice structure in the same way. The
addition of surfactant (Triton X-IOO at 5%), however, strongly inhibited the increase in the
ice structure size after the addition of sucrose etc., probably because of the inhibition of the
mass-transfer rate of water at the ice-liquid interface.

CONCLUSION

1. Photographic method and mercury porosimetric method for ice structure determination
corresponded well with each other
2. Mean ice structure size formed in agar gel was inversely proportional to the moving speed
of the freezing front.
3. Ice structure was estimated to be strongly dependent on the structure of the freezing front.
4. In the progressive-freeze concentration, the solute was effectively concentrated under the
condition at which a smooth surface structure was obtained at the freezing front.
5. Effect of the solution structure on the ice structure was investigated experimentally.

REFERENCE

1) Miyawaki, 0., Abe, T. and Yano, T., Freezing and ice structure formed in protein gels.,
Biosci. Biotech. Biochem., 1992,56953-957.
 A COMPARATIVE STUDY OF FREEZING TIME PREDICTION
FOR FOOD PRODUCTS

FRANCISZEK KLUZA & WALTER E.L. SPIESS

Agricultural University
Doswiadczalna 44, 20-236 Lublin, Poland
Federal Centre For Nutrition
Engesserstra13e 20, D-7500 Karlsruhe 1, Germany

ABSTRACT

The predicted freezing time of potato and tylose acc. several calculating models were
subjected to multiple regression analysis against the experimental data. It was proved, that
Pham's and Cleland's were the best and the further statistical analysis confirmed that both
models give comparable, accurate results.

INTRODUCTION

The freezing time of foods should be optimum in each case and regarding economic
effectiveness of the process, the freezing equipment should be built most favorably, well
run and should enable an appropriate control of the process. To meet these requirements, it
is necessary to predict exactly the freezing time. The aim of the paper is analysis of
accuracy of freezing time prediction performed with some analytic-empirical methods.

MATERIALS AND METHODS

The following methods were tested: that of Plank (1963), Mott (1964), Mellor (1976),
Pham (1986), IIR (1986), Ramaswamy (1979), de Michelis (1983), Cleland (1979) and
Cleland (1982) [1,6]. The experimental data were adopted from Ilicali (1990) (potato) and
Cleland [2] (tylose). The predicted freezing time of potato was conditioned to ambient
temperature at fixed values of heat-transfer coefficient as well as to this coefficient at fixed
ambient temperature differentiating the final temperature of product. The freezing time of
tylose was conditioned to heat exchange surface of an object at the fixed values of: ambient
temperatures, final temperature of the product and heat transfer coefficient. The results
were estimated using the multiple regression analysis and stepwise selection procedure.
 377
The regression analysis confirmed a distinct and explicit correlation of experimental
freezing time tex and 1;,. (Cleland [3,4]), ~b (Pham [7]) and predominance of both methods
over the rest. Basing on the experimental freezing time [2,6] (383 experiments, six prod-
ucts, five shapes) both methods underwent further analysis. All the results were statistically
analysed through error characteristics and study on fitting of calculated results distribution
to experimental values distribution, analysis of regressive dependencies of experimental
time in relation to calculated time as well as analysis of variance of prediction errors.

RESULTS AND DISCUSSION

Mean values of prediction error under Cleland's and Pham's methods were equal to -4.06%
and 0.32%, respectively while the standard deviation for this error is 13.60% and
13.98%.

TABLE 1
Statistical estimation of relative errors of freezing time prediction

Method: Pham Cleland

Sample size 383 -383
Mean % 0.32 -4.06
95 % Conf. int. for mean % -1.08+ 1.73 -5.42+-2.69
Standard deviation 13.98 13.60
Minimum % -47.88 -85.87
Maximum % 78.96 44.65
Interquartile range % 17.55 13.57
Standarized kurtosis 12.60 26.82
Coeff. of variation 4350.21 -335.28
Mean of error modules % 10.59 10.13
95 % Conf. int. for mean % 9.69+ 11.49 9.23+ 11.03
Regression equation lex = 0.04+0.995~h 0.048+0.931;,.
Prob. level of intercept 0.1198 0.0347
Correlation coeff. 0.9868 0.9896

Pham's method shows some dozen error variation and even so their mean value is better
than mean for Cleland's method, they are statistically more scattered. The distribution
according to Pham's is less slender (Tab. 1). Half the results is characteristic of error found
in the intervals -8.75% +8.8% (Pham) and -10.47% +3.09% (Cleland). The Kolmogorow-
Smirnow test application confirmed the hypothesis Ho in both cases that provides distribu-
tion compatibility of predicted times and distribution of experimental values. The regression
analysis of results after both methods in relation to experimental time gave comparable
effects at the high values of correlation coefficient (Tab. 1). Nevertheless both methods
provide the results well matching the experimental values in this point we deal with results
matching excess or depletion. Hence the absolute values of errors were examined and
one-way analysis of variance (also the LSD test) showed that the values don't vary signifi-
cantly. That is the predicted results obtained after both methods to equal degree match the
experimental values.
From the variance analysis of real values of error, it follows that 95% confidence
378
interval for error mean of Pham' s method (-1. 08 + 1. 73 %) is more favorable than that of
Cleland's one (-5.42+-2.69%) while the tests of Cochran-Cox, Bartlett-Box and Hartley
confirm Ho hypothesis that provides homogeneity of the studied variances. The tests of
Student-Newman-Keuls, Tukey-B and LSD gave evidence of some significant difference
between both studied methods with advantage to Pham's one. Although Pham's method
shows a lower mean value of error, it does not prove its superior accuracy, whatsoever. In
this case the regression analysis of experimental time in relation to predicted time decides
the problem against Pham's method (Tab. 1).

CONCLUSIONS

From the carried out researches it follows that in situations when frozen products correlate
slab, rectangular parallelepiped, cylinder or sphere models, we can deal with a relative
slight error of freezing time prediction. However, it concerns only the results obtained
under Cleland's (1979) and Pham's (1986) methods.

Acknowledgement

The investigations were performed in the scope of project No 5 5612 91 02 financed by the
Scientific Research Committee (KBN) in the years 1992-94.

REFERENCES

1. Cleland, A.C., Food refrigeration processes. Analysis. design and simulation. Elsevier
Science Publishers LTD, London, 1990, pp.79-136.

2. Cleland, A.C. and Earle, R.L., A comparison of freezing calculations including modifi-
cation to take into account initial superheat. Bull. IIR 1976-1, pp. 369-376.

3. Cleland, A.C. and Earle, R.L., Prediction of freezing times for foods in rectangular
packages. J. Food Sci., 1979, 44,964-970.

4. Cleland, A.C., Earle, R.L., A comparison of methods for predicting the freezing times
of cylindrical and spherical foodstuffs. L. Food Sci., 1979, 958-963.

5. Cleland, A.C., Earle, R.L., Freezing time prediction for foods - a simplified procedure.
Int. J. Refrig., 1982, 5, 134-140.

6. Kluza, F., Wyznaczanie czasu zamraZania produkt6w rolniczych i spozywczych. WAR,

Lublin, 1993.

7. Pham, Q.T., Simplified equation for predicting the freezing time of foodstuffs. J. Food
Techno!., 1986, 21, 209-219.
 VARIATION OF CORRECTION FACTORS IN AN APPROXIMATE EQUATION
FOR FREEZING AND THAWING TIME PREDICTION

AMELIA RURTOLO DE RETNICK

Instituto de Desarrollo Tecnol6gico para la Industria Quimica (UNL-CONICET)
Giiemes 3450 - 3000 Santa Fe, Argentina

ABSTRACT
An equation for predicting average temperatures vs. time behaviour at low Biot
number, B, was used to det.ermine the thermal response during freezing and thawing
of food placed in air freezer. The equation based on the effective heat capacity and
characteristics of biological material during phase chan.ge, for isolated infinite slab was
extended with correction factors based on B values to take in account the different
process conditions and object geometries. Rectangular blocks of chopped beef with
two or four exposed sides with equal or dissimilar heat-transfer coefficients on their
opposite surface, for B ranging between 0 and 3 were used. Predicted times were
compared with numerical solution results of non-linear partial differential equations,
P.D.E. previously experimentally tested.

INTRODUCTION
Heat-transfer coefficients, h, for the various surface of rectangular shaped foods dur-
ing freezing and thawing are frequently dissimilar, then, maximum temperature does
not occur at the center plane of the object and shift with time [1].
Simple algebraic equations were developed using equations for correlating ther-
mal properties and solutions of the heat transfer P.D.E. These equations accurately
predict average temperature, l, during freezing and thawing, agree with numerical
solutions of P.D.E. [2] for infinite slab at low B. l is based on the average enthalpy, [3]
and is the equilibrium temperature that would be obtained if the object was placed
in adiabatic enclosure.
In this work correction factors based on B for nonsymmetric conditions were
obtained to use a short-cut method for l vs. time with reasonable accuracy and two
or four surface exposing to heat transfer. In most cases of commercial interests B is
low, therefore this short-cut solution should satisfy most freezer design and operating
analysis needs.
MATERIALS AND METHODS
A general solution for the infinite slab temperature with a value of hL at one surface,
x = L and a value of ho at the opposite surface, x = 0, is used with constant initial
temperature. Integrating this equation between x = 0 and x = L and dividing by L
the expression for l is obtained. When B, are small the first term of the series can
be used with very small error. For infinite rectangular bars or for finite rectangular
solids of volume, V and heat transfer surface A, product of solutions is possible to
use since under these conditions, P.D.E., the boundary conditiond and the initial
condition are satisfied. Following this procedure the equation is
380

j = 1,r (1)

where (2)

t (Q) = (BL + BO)Q1 (3)

an 1 Q21 - B L B 0

Fo = aOI L2, T = (f - t a )/(t1 - t a ), BL = hLLlk, Bo = hoLik, a, thermal difussivity;

t a , ambient temperature; t1 initial temperature; e, time; k, thermal conductivity
and r is the pair of surfaces exposed to heat transfer 1, 2 or 3. It can be shown
that approximation for Qi can be broken into two components characterizing the
opposing sides, i.e.
Qi = BL F(Bd + Bo F(Bo) (4)
where BLF(BL) and BoF(Bo) are the Qi component characterizing side x = Land
x = O. Eq.(l) can be converted, following symmetric technique [2J into an ordinary
differential equation which applies over most of the time range of interest at low B:

aT
- = --2:
T (Qi
- k)r
(5)
dB pCps L2 j=l j

C - C (n wo - bn s )6. H o(to - ti)

where ps - pf+ (to -t)2 t < ti (6)

p is density, Cpf heat capacity at total frozen state, nwO initial weight fraction of
water, b water binding coefficient for solutes and solids, ns weight fraction of solutes
and solids, f:::,.Ho latent heat of fusion of ice, ti initial freezing point temperature, to
normal freezing point temperature of pure water.
Similarly, Cps in Eq.(5) was substituted by Eq.(6). Moreover Eq.(5) was inte-
grated between 0 and Bfor T2 and T below Ti, with BL and Bo based on a constant
effective value of k, G = II(Gdj, subscript f indicating total frozen state and susti-
tuting the terms BkJl(pCpfL 2 ) = tP to obtain:

v---'P_Cp'-"f_ _ _ _ . E(T)
</> = _r_ _ _ _
t < ti (7)
2: [BfLF(Bd + BfoF(Bo)JjAj
j=l

where: E (T) = lnG(Td'1') + (P/T;;)[ln(D) + MJ (7a)

P = (nwo - bns)f:::,.Ho(To - T;)/[Cpf(t1 - ta)J (7b)
D = G'1'2(To - '1')/['1'(To - '1'2 )J (7c)
M = To(T2 - '1')/(To - '1'2)(To - '1') (7d)
The T calculated with Eqs.(l) and (7) were numerically compared. A finite
difference method predicting local temperature and enthalpy to obtain average en-
thalpy and t vs. 0, described in a previously work [3], was used. Experimental works
were carried out to determine central and local temperature at low B, to verify the
 381

finite difference method. The freezing and thawing conditions of some samples are
tabulated in table 1.
TABLE 1
Run Conditions for the Samples
'1 emperature
Sample Thicknesses Weight Density Heat Transfer Coeff.
mm W/m 2 °c Amb. Tnit.
Number kg kg/m 3 °c °c
L1 L2 (hd1 (hoh (hLh (hoh
1 2,').4 --- 0.270 1020.8 14 6 --- --- -22 1,';.0
2 ,'>0.8 - -- 0.506 96,'>.1 2,') 20 - -- - - - -22 1,'>.2
3 ,'>0.8 101.6 0.493 941.0 30 30 32 22 -2,'> 7.2
4 2,'}.4 -- - 0.280 1069.6 2,'> 1.'> - - - - - - 1.'> -1.,>.8
,> ,'>0.8 101.6 0.490 930.5 30 30 30 2,'> 16 -16.6

Samples properties, composition and thermal properties are: ti = -1,035°Cj

nwo = 0.68 Kg water/Kgj (nwo - bns) = 0.63 Kg freezable water/Kgj ns = 0.20
Kg solids and solutes/Kgj fat = 0.12 Kg fat/Kgj Cpo = 3575 J /KgOCj Cpf = 2050
J/KgOCj ko = 0.45 W/m 2 °Cj k f = 1.50 W/m 2 °C.
RESULTS
Three, non-symmetric freezing runs and two similar thawing runs were used to predict
the t vs. (} values given in table 2.
TABLE 2
Predicted Freezing and Thawing Times
~ample HlOt Number Average 'i'lme Prechcted l{elatJve
Temp. (min) Porcentaje
Number
(BLh (Boh (BLh (Bo12 °c Fin.Diff. Prop.Eq. Difference
1 0.;165 0.1.'>6 - - - - - - -lUO 300.8 301.9 0.,,>
2 1.;103 1.042 - -- - -- -lUO 216.7 224.3 3.,')
3 1.nO U6;1 3.:33.'> 2.293 -lUO 11:1.0 12,'>.,'> 9.9
4 0.6,'>0 0.;190 --- --. -1.035 147.0 1.,>,,>.0 ,').4
,> U63 U63 3.126 2.60.'> -1.03,'> 130.0 144.0 10.7

Band Bo were calculated with k f for freezing and with (k f + ko)/2 for thawing
[3]. Results obtained with non-symmetric heat transfer equations are in agreement
with the values predicted with P.D.E. Numerical Solution.
Relative error for four-sided runs were larger than in two-sided runs, they increase
when BL and Bo increase and when evaluated temperature changes decrease.
CONCLUSIONS
Eq.(7) provides a general effective short-cut method for predicting freezing and thaw-
ing times for two and four-sided heat transfer in rectangular bodies with roughly 5
to 10% accuracy when B is less than 3.
REFERENCES
1. De Michelis, A. and Calvelo, A., Mathematical models for nonsymmetric freezing
of beef. L Food Sci., 1982,47,1211-7.
2. Schwartzberg, H.G., Mathematical analysis of the freezing and thawing offoods.
AIChE Summer National Meeting Paper, Detroit, Michigan, August 1981.
3. Rubiolo de Reinick, A.C. and Schwartzberg, H.G., Predicting temperature vs.
time behavior during freezing and thawing of rectangular foods, Biothedmol.
Progr. 1986, 2, 164-74.
 PROGRESSIVE FREEZE-CONCENTRATION AND ITS RELATIONSHIP WITH
THE ICE STRUCTURE AT F~EEZING FRONT

SEOUNG-KWON BAE AND OSATO MIY AW AKI

Department of Agricultural Chemistry,
The University of Tokyo,
Bunkyo-ku, Tokyo 113, JAPAN

ABSTRACT

The concentration efficiency in the progressive freeze-concentration was related to the ice
structure of the freezing front. A high freeze-concentration efficiency was obtained under the
conditions at which a smooth solid-liquid interface was formed. An apparent distribution
coefficient of a solute between the solid and the liquid phase was useful to describe the
concentration efficiency. The apparent distribution coefficient, reflecting the ice structure at
the freezing front,was related to the operating conditions represented by the moving speed of
the freezing front and the speed of stirring in the solution phase.

INTRODUCTION

Progressive freeze-concentration or normal freezing is an effective technique for the

concentration of dilute solute compounds in a solution. In this method, the freezing
front(solid-liquid interface) moves one-dimensionally from one end to the other end through
the solution. At the appropriately selected operating conditions, the solid phase left after the
passage of the freezing front contains only pure ice and the solute compounds are
consecutively rejected from the freezing front to be concentrated into the liquid phase. In this
paper, the concentration efficiency of the solute compound is investigated in the progressive
freeze-concentration.

MATERIALS AND METHODS

Blue dextran(O.02g), as a tracer, was dissolved in distilled water(lOOml) and was poured into
an upright cylidrical sample vessel, which was dipped at the bottom end in the coolant kept at
a constant temperature. The sample vessel was moved down into the coolant at a constant
speed so that the moving speed of the freezing front was kept constant. The solution was
stirred by a propeller located near the freezing front. The concentration of the blue dextran
was measured at an interval by a spectrophotometer.
 383
RESULTS AND DISCUSSION

To describe the concentration efficiency in the progressive freeze-concentration, an apparent

distribution coefficient of solute between the solid and the liquid phase, K(=Cs/CL), was
introduced. K was determined from the mass balance equation at the liquid-solid interface
assuming complete mixing in the liquid phase as follows[l].

(1-K) In f= In (Col CL) (1)

where Co is initial solute concentration in the solution, CL is solute concentration in the liquid
phase, Cs is solute concentration in the solid phase at the liquid-solid interface and f is the
volume fraction of the liquid phase. K was constant(O<K<l) under the defined operating
conditions and was inversely proportional to the concentration efficiency. The double
logarithmic plot between the concentration ratio(CoICL) and the volume fraction(f) showed a
good correlation. The value ofK was calculated from the slope of the regression line(=I-K).

0.3 r - - - - - - - - - - - - ,
0.0
d....... -0.3 K=l
U -0.6
-
"-'"
I::: -0.9
-1.2
•
K=0.244

-1.5 (l-K) lnf =In (Co tel)

-1.8 '---'----'---'--'--'---'--''---'----'----'
-2.0 -1.6 -1.2 -0.8 -0.4 0.0
Inf
Figure 1. Effect of stirring on the apparent distribution coefficient of blue dextran in the
progressive freeze-concentration.
(0, no stirring; ., stirring speed at 1000 rpm; coolant temperature, -10°C)

0.5

0.4 •
~0.3 •
~ 0.2
•
0.1
0.0 A, I I I

0.0 0.5 1.0 1.5 2.0

U [cm/hr]

Figure 2. Relationship between the moving speed of the freezing front and the apparent
distribution coefficient of blue dextran in the progressive freeze-concentration.
384
Figure 1 shows the effect of stirring in the liquid phase near the freezing front in the
progressive freeze-concentration. With no stirring, the value of K was almost 1.0, which
shows that no freeze-concentration occurred. With a stirring speed at 1000 rpm, the value of
K changed to 0.244 showing that a substantial freeze-concentration occurred.
Effect of the moving speed of the freezing front on the apparent distribution coefficient
is shown in Fig. 2. The apparent distribution coefficient increased much with increase in the
moving speed of the freezing front.
As for the effect of the coolant temperature on the freeze-concentration, the lower
temperature gave the higher value of K(poor concentration efficiency; data not shown).
In one-dimensional freezing of a solution, the initial freezing of water occurs at the
cooling side and the dendritic ice crystal grows in p~alled to the direction of the heat flux.
The mass balance equation of water molecule gives the following equation at the freezing
front:

Moving speed of freezing front

= Mass flux of water molecules in unit area at freezing front
x Surface area of freezing front (2)

because freezing basically forms pure ice microscopically. This equation explains the
dependency of the freeze-concentration efficiency on the stirring speed and the moving speed
of the freezing front. According to Eq. (2), a small surface area of the freezing front will be
obtained under a slow moving speed of the freezing front and/or under a rapid stirring
through a high mass flux of water molecule. A small surface area of the freezing front will
give a smooth surface structure so that the solute molecules are effectively rejected from the
solid-liquid interface to give a high freeze-concentration efficiency. On the contrary, a large
surface area of the freezing front with a complex structure will give a poor freeze-
concentration efficiency through the mechanical incorporation of the solute molecules into
the frozen ice phase.

CONCLUSION

The apparent distribution coefficient was useful to describe the concentration efficiency in the
progressive freeze-concentration. The apparent distribution coefficient showed a small value
at a slow moving speed of the freezing front and/or a high speed of stirring in the liquid
phase. A high concentration efficiency was obtained under the conditions at which the
smooth ice structure was formed at the freezing front. The ice structure at the freezing front
was theoretically related to the moving speed of the freezing front and the mass flux of water
molecules in unit area at the freezing front. The theoretical relatipnship was well
corresponded to the experimental results from the progressive freeze-concentration.

REFERENCE
1. Pfann, W.G., Priciples of zone-melting. J. of Metals, 1952,747-753.
 FREEZE CONCENTRATION OF SOME HEAT SENSITIVE SOLUTIONS

O.FLESLAND, X. SONG* , I. S1R0MMEN

Department of Refrigeration Engineering, The Norwegian Institute of Technology,
University of Trondheim, N-7034 Trondheim, Norway. *SINTEF Refrigeration
Engineering, N-7034 Trondheim, Norway

ABSTRACT

Freeze concentration has been studied as one possible method of concentrating heat sensitive
solutions, prior to low temperature drying. Experiments are carried out to study the influence
on product quality and operational conditions. One medical solution showed no loss of
biological activity during freeze concentration. The freezing point depression influences the
energy consumption and the concentration limit for the process. Freezing points have been
measured for stickwater, orange juice, pepsin and a medical solution. Ice growth was studied
in a batch crystallizer, and the ice were separated from concentrate by a centrifuge. An
economical analysis of two different dewatering lines is carried out.

INTRODUCTION

Drying of heat sensitive solutions has been an important part of the research activities at the
Norwegian Institute of Technology, Department of Refrigeration Engineering (5). Freeze
concentration is studied as one possible concentration method prior to atmospheric freeze
drying. Freeze concentration has been applied with success for products like orange juice,
coffee, beer, wine and other liquids where aroma retention is important (1,2). Heat sensitive
macromolecules in aqueous solution are expected to have lower ice growth rate due to lower
diffusion coefficients, and the viscosity dependence on concentration is often higher than for
low molecular solutions. The separation of ice crystals from concentrate in a wash column
is mainly controlled by the ice crystal size and viscosity of the solution (3). This paper
presents some experimental results from freeze concentration of liquids containing
macromolecules. Freezing points are measured. Ice production in a batch crystallizer is
studied. An economic evaluation of dewatering lines for a given production capacity is carried
out.
386
MA TERIALS AND METHODS

Growth of ice was studied in a batch crystallizer, and the ice was separated by a centrifuge.
The freezing points were measured with a PT-lOO resistance sensor with an accuracy of +/-
O.03°e. The solutions were cooled in a temperature controlled bath to a temperature below
the freezing point The crystallizer was seeded to initiate ice crystallization. The temperature
increased to the eqUilibrium temperature. Orange juice, stickwater, pepsin and one medical
solution were studied with respect to ice growth, loss of product and phase diagrams. The
pepsin and the medical solution contain low molecular components in addition to the
macromolecules.

RESULTS

Freezing points are measured for concentrations up to 30%, wet bases (by mass), Fig.1. The
results are compared with results from eqUilibrium thermodynamics for ideal solutions, Fig.l,
from which an average molecular mass can be calculated. The eqUilibrium temperature
(freezing point) is important for the energy consumption for both freeze concentration and
atmospheric freeze drying. Some experiments on ice growth in pepsin solution was carried
out for varying concentrations, see Fig.2. The initial subcooling was O.3°e, the mean
temperature difference between the solution and the bath was O.4°e and the duration after
seeding was 3 hours. The concentrations is the initial concentration in each experiment, and
the relative ice production is the mass of ice separated by a centrifuge, divided by the total
mass. The ice production decreases as the concentration of pepsin increases, Fig.2. At the
same time the impurities in the ice increases from about 2% to 5%, Fig.2. The viscosities
were measured at ooe, to be 17, 34 and 68cp for the concentrations 7, 15 and 28%
respectively. The separation of ice from concentrate was done without any washing. In one
experiment the ice crystals were washed with pure water, and the dissolved solids in the ice
was 0.9% before and 0.2% after washing. The initial concentration of pepsin in solution was
then 7% by mass and the ice crystals were 1-2mm in size.

",
0. 10 15 20 25 30
.Yo 5
10 20 30 40 50 't;,022 ••••Il '<n
.><t.
0 ~ 0.20 4 [3
.2 "
E
U -1
t-
-1 ~ 0.18 "" 3~
-g... 0.16 ....." " ~
....,...
Q)
5 -2 -2 a. 2 en
Q) 0.14 .91
-3 -3 u .1:
Q)
a. .;; 0.12 1 a.
::J
E
fo- -4
Q)
-4
>
:; 0.10 .s
Q) 0
a: 10 15 20 25 30
-5
10 20 30 40 50 Dissolved solids (% by mass)

Dissolved solids [% by mass)

Figure 1 Freezing points as function of Figure 2 Ice growth and impurities in the
concentration. Concentration in (mass of dry separated ice, as function of concentration of
matterl total mass, times tOO) pepsin.
 387
DISCUSSION

Pure macromolecules in aqueous solution give small freezing point depression, and should be
suited for freeze concentration due to energy consumption. On the other hand the diffusion
coefficients of macromolecules are much lower than for instance for sucrose and glucose,
which has been studied by many authors, se i.e. (4). The slower diffusion and often higher
viscosities of solutions decreases the ice growth, and the separation of pure ice from the
concentrate is more difficult. The present experiments show a tendency towards reduced ice
growth and increased loss of product, as the concentration increases, Fig.2. The variation of
product loss with concentration gives valuable information about the possibility of using
freeze concentration for different products, although the total product loss could possible be
minimized with a wash column.

ECONOMIC EVALUATION RELATED TO QUALITY

A low temperature dewatering line which consists of freeze concentration and atmospheric
freeze drying in a fluidized bed drier with heat pump, is compared with evaporation and spray
drying. Investment costs and operational costs from suppliers are considered. A case study
on a production capacity of 500kg dried product per hour and 7000 hours operation per year,
is carried out. The low temperature line should add a value to the product of about 3.3NOK
or 0.5US$ per kilo more than the traditional dewatering line. The return on investment was
set to 15%. For instance if the added value is related to biological activity, the low
temperature line should keep 3% more of the activity than the traditional line, if the product
price is 100NOK.

REFERENCES

1. Thijssen, H.A.C., Evaluation of concentration alternatives for liquid foods. Int.Zeitschrift

fur Lebensmittel-Technologie und Verfahrenstechnik, 1983,34, pp 586-601.

2. Van Mil, P.J.J.M., and Bouman, S, Freeze concentration of dairy products. Neth. Milk
Dairy J..1990, 44, pp 21-31.

3. Thijssen, H.A.C., Current developments in the freeze concentration of liquid foods. I n

Freeze drying and advanced food technology, ed. S.A. Goldblith, L.Rey and W.W.
Rothmayr. Academic Press, London, 1975, pp. 481-501.

4. Huige, N.J.J., Nucleation and growth of ice crystals from water and sugar solutions in
continous stirred tank crystallizers. Ph.D. Thesis, Eindhoven University of Technology,
Netherlands, 1972.

5. Str~mmen, I., and Kramer,K., New applications of heat pumps in drying processes.
ASME Winter annual meeting. Anaheim, California, USA, 1992.

ACKNOWLEDGEMENT. This work is funded by The Research Council of Norway, Division NFFR
 STUDIES ON THE FREEZE CONCENTRATION OF FOODS.--DETERMINATION OF
EUTECTIC TEMPERATURES OF AMINO ACIDS IN AQUEOUS ETHANOL SOLUTION--

M. Shibata, K. Ishikawa, Y. Fukuta, S. Ishikawa*, M. Shimoyamada*

and K. Watanabe*.
Food Research Institute, Aichi Prefectural Goverment, 2-1-1,
Shinpukuji-cho, Nishi-ku, Nagoya, Aichi 451, Japan, *Faculty of
Agriculture, Gifu University, 1-1 Yanagido, Gifu 501-11, Japan

ABSTRACT
In order to apply a freeze concentration to food constituents,
freezing points, solubilities and eutectic temperatures were
estimated for an aqueous solution of amino acid (glycine and sodium
glutamate) containing various concentrations of ethanol. The
eutectic temperatures of each amino acid lowered with the increase
in ethanol concentration, but the concentrations at the starting
point of precipitation were also decreased. Ethanol addition was
effectiveto precipitate glycine and sodium glutamate at their lower
concentrations.
On the basis of the above results, the freeze concentration of
glycine was attempted on a pilot plant scale. In ethanol-water
solution, glycine was recovered effectively.

INTRODUCTION
When a certain aqueous solution is cooled in a suitable vessel
(crystallizer), ice crystal is first formed and the solute(s) become
gradually concentrated. These are the major steps involved in
freeze concentration and they are advantageous for protection of
heat labile components, suppression of flavor removal, and low-cost
dehydration [1].
As the concentration of each solute exceeds its own saturation
point, the solute is expected to precipitate by further cooling. The
temperature at this point is called the eutectic temperature [2].
In the present study, the eutectic temperatures of two kinds of
amino acids were estimated, considering recovery of precipitation.
Addition of some kinds of alcohol was reported to affect the
solubilities of amino acids [3]. So the effect of ethanol on the
eutectic temperatures of amino acids was also estimated. The
efficacy of freeze concentration was also estimated on a pilot plant
scale on the basis of the eutectic temperatures in amino acid-water-
ethanol.
 389

MATERIALS AND METHODS

Measurement of Freezing Temperatures of Amino Acid Solutions
The amino acid solution (glycine or sodium glutamate) was put in a
glass vessel (180 ml), and stirred with a stainless steel plate
rotating at 93 rpm in a deep freezer. Temperature of solution was
measured by thermocouple, which was Type-K and calibrated with a
quart thermometer.
Measurement of Solubitities of Amino acids in Water or Water-Ethanol
An amino acid was dispersed excessively in deionized water and
stirred with magnetic stirrer overnight at a certain definite
temperature. The supernatant was filtered with glass filter (llG4)
and analyzed with formol titration [4].
Pilot Plant Experiment
A 10 L plastic vessel was filled with about 8 L glycine solution
(0.70 mol/kg) and rotated at 4.5 rpm in deep freezer at a little
lower than the freezing point of the examined solution until ice
crystals were formed for one to three nights. The solution with ice
crystal was separated into ice crystal and supernatant by
centrifugal filtration (2,600 rpm at DoC). The supernatant was put
back into the same vessel and repeatedly concentrated in the same
manner until solute was precipitated.
RESULTS AND DISCUSSION
Freezing Temperatures of Amino Acid Solutions with Various
Proportions of Ethanol
The temperature of each solution was gradually decreased and through

Glycine Sodium glutamate

.f> l- f> ~
,........
/ . I
u
°' - '
. .I
~ J.
Q.)
H I,
.I
;:::I
+oJ
co

. .....
H 0 <> J.
i
Q.)

~ .............
.......
, .....,
I

. . . . . ..•.......
Q.)
-..
.~
E-<

".
-10 ~ .... ~.....'.. '. . '''"t,'
~.,.
:
r '"
~..... ..., .
0 1 2 3 0 1 2 3 4
Molality (mol/kg) Molality (mol/kg)
Figure 1. Solid-Liquid Equilibria of Amino Acids in Water-Ethanol.
e, 0: EtOH 0% ..... D,.: 4.6% ••• 0: 9.2%. ..... \l: 13.9%.
Open: Solubility. Close: Freezing point.
390

the supersaturation state, ice crystal was formed in solution. With

an aqueous solution, the freezing points decreased with the increase
of amino acid concentration (Figure 1). Between the freezing points
and the amino acid concentration (molality), there were good
linearlities both in glycine (r=0.998) and sodium glutamate
(r=0.998), and the molar depression calculated was 1.65 and 3.24
deg/mol, respectively. As sodium glutamate dissociates into sodium
cation and glutamate anion, it reaches 1.62 deg/mol ion, nearly
equal to that of glycine. As the ethanol concentration increased,
the molar depression also decreased in both amino acids, and the
linearlity of the freezing point curve was retained between 0 and
13.9 % ethanol solution.
Solubilities of Amino Acids in Water or Water-Ethanol Solution and
Eutectic Temperatures
The results were shown in Figure 1 with freezing point curves. Each
solubility curve was shifted lower and changed from concave to
convex with the increase of ethanol concentration. In this figure,
the crossing points of freezing point curves and solubitity curves
in various ethanol concentrations showed the eutectic points
(eutectic temperatures). In fact, the precipitation of amino acids
was detected to the right of the eutectic point.
Freeze Concentration of Glycine Solution on a Pilot Plant Scale
From the above data, we obtained the freezing points and eutectic
temperatures of glycine and sodium glutamate in water-ethanol
mixture. Then the freeze concentration of glycine was attempted on
the pilot plant scale. With 3 repetitions, the concentrated glycine
solution (1.9 mol/kg) was obtained, and the next freezing step gave
the precipitation of glycine. Loss of glycine in this procedure was
34.9 %. On the other hand, ethanol-containing solution (4.6 %) also
yielded the precipitate with 4 repetitions, and the loss was 20.2 %,
which was effectively lower than that of the aqueous solution.
The aqueous solution of glycine was completely frozen at a
temperature lower than the eutectic temperature, so a relatively
high temperature was used for the freeze concentration. On the
other hand, the ethanol-water solution was not frozen completely
because of the decrease in the solution freezing point. The loss of
glycine in freezing concentration steps was lower in the ethanol-
water system than in the aqueous system. This may mean that most of
the amino acid in solution can be recovered.
REFERENCES
1) Schwartzberg, H.G., Food Freeze Concentration. In Biotechnology
and Food Process Engineering, ed. H.G. Schwartzberg and M.A.Rao,
Marcel Dekkev, Inc., New York, 1990, pp.127-202.
2) Fennema, O.R., Powrie, W.D. and Marth, E.H., Low-Temperature
Preservation of Foods and Living Matter, Marcel Dekkev, Inc., New
York, 1973, pp. 104-13.
3) Orella, C.J. and Kirwan, D.J., The solubility of amino acids in
mixtures of water and alliphatic alcohols, Biotechnol. Prog.,
1989, 5, 89-91.
4) Kallen, R.G., and Jencks, W.P., Equilibria for the reaction of
amines with formaldehyde and protons in aqueous solution. A re-
examination of the formol titration. J. BioI. Chem., 1966, 241,
5864-5878. - ---
 ICE CRYSTALS AGGLOMERATED IN FREEZE CONCENTRATION

YOSHIHITO SHIRAI, KAZUHIRO NAKANISHI*, RYUlCHI MATSUNO**

Department of Biochemical Engineering and Science,
Faculty of Computer Science and Systems Engineering,
Kyushu Institute of Technology, Iizuka, Fukuoka 820.
*Department of Biotechnology, Faculty of Engineering,
Okayama University, Okayama 700.
**Department of Food Science and Technology,
Faculty of Agriculture, Kyoto University,
Sakyo-ku, Kyoto 606, Japan

ABSTRACT

Freeze concentration is required to produce large ice crystals in order to reduce the loss of solutes
adhering to the surface of the ice crystals removed. We examine here if large agglomerated ice crystals
would be produced from a lactose solution in a common batch crystallizer with an inner heat
exchanger. Large agglomerated ice crystals (ca. 2 mm in diameter) were formed in the crystallizer with
a low initial subcooling (at the start of crystallization) in spite of the cooling rates.

INTRODUCTION

Freeze concentration is one of the best methods to remove water from liquid foods because of its low
energy use, lack of evaporation, and low temperature operation. Large ice crystals are required to
reduce the loss of solutes adhering to the surface of ice crystals. Currently, special interest is being paid
to the freeze concentration of dairy products (1, 2). We have also been investigating ice crystallization
for the purpose of freeze concentration of dairy products (3, 4) and previously reported that large
agglomerated ice crystals are produced from a lactose solution in a batch crystallizer with an external
heat exchanger, and confirmed that no solutes are included in the ice crystals agglomerated (3).
However, the apparatus was rather complicated and a more simple one needed to be developed. Here,
we examine if large agglomerated ice crystals would be produced in a common ice crystallizer with an
inner heat exchanger.

MATERIAL AND METHODS

A 5% lactose solution was used as a sample solution. Figure 1 shows the experimental apparatus used
in this work. We used a common ice crystallizer with an inner heat exchanger with a working volume of
300 m\. A pair of aluminum rods was rotated slowly, scratching the wall to prevent ice crystals from
sticking on the wall and to enhance the heat transfer. Figure 1 also shows an example of the
temperature change measured. The temperature jumped up by the heat generated in ice crystallization
after a seed ice crystal was introduced. The difference between the plateau and the minimum
392

Figure 1 Experimental set-up.

temperature is defined as the initial subcooling. Microscopic photographs of ice crystals formed were
taken at specified intervals to check ice crystal size distributions.

RESULTS AND DISCUSSION

We have previously examined ice crystallization in a lactose solution with a batch crystallizer with an
external heat exchanger and found that uniform sized large agglomerated ice crystals were formed as
shown in Fig. 2 after two hours from crystallization (3). However, a rather complicated apparatus was
needed in this method. Large ice crystals should be formed in a more simple crystallizer.
We confirmed that large agglomerated ice crystals were also found in this crystallizer when ice
crystals are formed in the small size of subcooling (= 0.12 K) as shown in Fig. 3. The size of ice crystals

Figure 2 Large ice crystals agglomerated in Figure 3 Large ice crystals agglomerated in
the previous work. this work after 2 h from the start.
 393
was more than 2 mm in diameter. On the other hand, large agglomerated ice crystals were not formed
with a larger initial subcooling (= 1.31 K). The large ice crystals formed had a wider crystal size
distribution than those obtained in the previous work. Heat removal from the wall of the crystallizer
probably results in the wider distribution because larger subcooling always exists in the thin layer
around the wall.
We then checked the effect of heat removal rate from the wall of crystallizer on the formation of
large agglomerated ice crystals. The threefold heat removal rate was given by decreasing the coolant
temperature around the wall. The same large agglomerated ice crystals shown in Fig. 3 were observed,
indicating that the proposed method to produce large ice crystals is available even when the operation
is carried out with higher heat removal rates.

ACKNOWLEDGEMENT

The authors wish to thank Mrs. T. Ohta, N. Ogawa, M. Yamaguchi, and Miss A. Kobayashi for their
assistance with experiments.

REFERENCES

1. Van Mil, P. J. J. M. and Bouman, S., Freeze concentration of dairy products, Neth. Milk Dairy J.,
1990,44, 21-31.
2. Swientek, R. J., Concentrated milk without heat, Food Proc., 1991, October, 62-64.
3. Shirai, Y., Sugimoto, T., Hashimoto, T., Nakanishi, K., and Matsuno, R., Mechanism of ice
growth in a bath crystallizer with an external cooler for freeze concentration, Agric. Bioi. Chern.,
1987, 51, 2359-2366.
4. Shirai, Y., Nakanishi, K., Matsuno, R., and Kamikubo, T., Effects of polymers on secondary
nucleation of ice crystals, J. Food Sci., 1985, 40, 401-406.
 THERMODYNAMIC MODELING OF ~BE OSMO~IC CONCENTRA~ION PROCESS

R. N. BISWAL 1 and M. Le MAGUER2

luniversity of Tennessee, Knoxville, TN 37901, USA
2university of Guelph, Guelph, Ontario, Canada

ABS~RAC~

Mass transfer in osmotic concentration was modeled using thermodynamics

of irreversible processes (TIP). The model assumed no leaching of soluble
solids, but took into account shrinkage of the tissue. The model further
assumed that the insoluble solids of the tissue did not have any effect on
the physical and thermodynamic properties of the aqueous phase of the
tissue. Flux and force relations were found to be non-linear. Rate
parameters for diced carrots in NaCl-water .olutions were estimated.

I~ODUC~ION

The increase of interest in using osmotic concentration as an intermediate

step in fruit and vegetable processing has created the need for better
understanding of mass transfer in the osmotic process. The objectives of
this research were to develop a model for mass transfer in the osmotic
concentration process based on thermodynamics of irreversible processes and
to apply the model to diced carrots in contact with NaCl-water solutions.
The osmotic concentration of carrot tissue in NaCl-water solution
involves movement of four components: (1) NaCl, the osmotic agent (2)
water, (3) active component of original carrot tissue (soluble solids), and
(4) insoluble solids of carrot tissue (inert). If v 4 is the velocity of
component 4 at the surface of the cubes with respect to a fixed frame of
reference (center of the piece), the diffusional fluxes can be written as
(1)
where OJ is the mass concentration of component i (kg i/m3 ) , and v j is the
velocity of component i with respect to the fixed frame of reference. J j 4
for i = 1 or 2 can be obtained from the experimental data using Eq. 2.
(2)

Where: ITIj is the mass of component i (i 1 or 2), A is the surface area of

the cube, and e is the time.
 395
Since component 4 is thermodynamically inactive, the heterogeneous 4-
component system can be converted to a homogeneous 3-component system,
making any two of the three forces (XI' X2, and X3 ) independent. The linear
laws of TIP (1,2) for components 1 and 2 can be written as

J I4 LII XI + LI2 X2 (3)

J 24 L21 XI + L22 X2 (4)

subject to: LI2 = L 21 ; LIIL22 - Lil > O.

If the flux and force relations are non-linear, a pseudolinear form can
be written expressing the coefficients in polynomials of XI and X2 as
suggest by Sauer (3).
The Aqueous phase can be conceptually separated from the tissue and
placed in a single compartment surrounded by an effective membrane made of
inert materials, cell membranes and extracellular space. The resistance to
mass transfer is therefore "lumped" in the effective memberane and the
driving forces defined by the difference in the average concentrations of
the species in the compartment and in the bath.

MATERIALS AND METHODS

Diced carrots (Gold Pak) were soaked in tap water for 1 h and brought in
contact with NaCl-water solutions at two concentrations of NaCl (15% and
10% by weight) and two temperatures (25· and 10· C) in small beakers with
magnetic stirrers, maintaining an initial solid to liquid ratio of 1:2.
Samples were analyzed at 0, 5, 15, 30 and 60 min for reduction in mass,
moisture content and NaCl content of carrot cubes. The NaCl content and
density of solution were also measured. Assuming that the positive flux is
the one away from the center of the piece, J I4 and J 24 were calculated using
Eg. 2. Activity coefficients of NaCl and water were calculated using the
subroutine of Lee (4) based on UNIQUAC equation.

RESULTS AND DISCUSSION

The plots of J I4 and J 24 versus Xl and X2 exhibited non-linear relations

(Fig. 1). Therefore, rate parameters for mass transfer in the process were
obtained using non-linear forms of Eg. 3-4. Multilinear regressions
independently on Egs. 3-4 suggested that only some terms constituting the
straight coefficients were significant (P<0.05), thus uncoupling Egs. 3 and
4. In terms of their significant terms, Eg. 3 and 4 are rewritten as

(L II O + AI~2 + ~I) XI (5)

(B~2 + B3X 22) X2 (6)

The apparent phenomenological parameters LOll' AI2 and A2 B2 and B3 ,

estimated for the three experimental conditions used in the study (15%
NaCl-water,25·i 10% NaCl-water, 25·C; 15% NaCl-water, 10·C), using
multilinear regression, are listed in Table 1 (r2 value 0.97 or higher).
396

(a)
J1R J2R (b)

IIR= I,'!J,'CMIN) I,. = I,'/J,'(MAX)

1.000 1.000
X,. = X,/X,(MIN) X,. = X/X,(MIN)
X,. = X,IX,(MAX) X,. = X,IX,(MAX)
0.668 0.667

0.336 0.333
1.000 1.000

0.738
X1R . 0.476 X1R

00405 ~-0.:'l>.10{,j)8'''/0.213
0.10~\0.213

Figure 1. Fluxes of (a) NaCl and (b) water versus the thermodynamic
forces for diced carrots in contact with NaCl-water solutions.

TABLE 1
The coefficients for diffusional fluxes of NaCl and water during osmotic
concentration of diced carrots in NaCl-water solutions

Experimental Parameters for diffusional fluxes

Condition Flux of NaCl (J 14 )* Flux of water ( JZ4)**

% NaCl/oC LOll A12 A rZ B2 B r2

x109 x10 10 X10~Z x10 8 xlolt

10/25 7.9464 5.8630 6.0078 0.99 2.7305 4.3081 0.99

15/25 0 2.2473 2.0620 0.97 0.7687 9.8047 0.99
15/10 0 4.4362 5.5928 0.99 -0.8163 2.5438 0.99

* Equation (5); ** Equation ( 6)

REFERENCES

1. Chandrasekaran, S.K. and King, C.J. Multicomponent diffusion and

vapor-liquid equilibria of dilute organic components in aqueous
sugar solutions. AIChE Journal, 1972, 8, 513-519.

2. Wisniewski, s., Staniszewski, B. and Szymanik, R. Thermodynamics of

Non-equilibrium Processes. D. Riedel Publishing Companr, Dordrecnt-
Holland/Boston, 1976.

3. Sauer, F. Nonequilibrium thermodynamics of kidney tubule transport,

In Handbook of Physiology. section 8. Renal physiology (j. Orloff
and R. W. Berliner, eds.) Amer. Physiol. Soc. Washington, 1973.

4. Lee, N., Extended UNIQUAC Subroutine, Master of Science Thesis,

University of Alberta, Edmundton, Canada, 1987.
 RECENT ADVANCES IN DEWATERING AND IMPREGNATION
SOAKING PROCESSES

A.L.RAOULT-WACK 1, R.SAUREL1/2, G.M.RIOS 2 and S.GUILBERT1

1 - CIRAD SAR -Food Technology Division - 34032 Montpellier Cedex -France
2 - University of Montpellier 2 -Food Technology and Engineering Research Center
34095 Montpellier Cedex 2 - France

ABSTRACT
Simultaneous dewatering and impregnation can be achieved through soaking of foodstuff
pieces in highly concentrated solutions .This is the so-called "Dewatering and Impregnation
Soaking Process", or briefly DISP . After a brief recall of general process characteristics,
recent advances in mass transfer analysis are presented that more specially concern model
foods ( agar gel cubes) and fruit pieces (frozen and fresh apple) immersed in sugar juices.
Consequences upon the improvement of traditional techniques ( osmotic dehydration, direct
formulation ... ) are discussed. As a correlation, original technological applications and
new processing ways are hinted at .

PRESENTATION OF DISP
Dewatering of foodstuffs is generally achieved by either mechanical (centrifugation, strai-
ning ... ) or thermal (convection or ebullition) methods. A third technique can be defined
which consists in soaking foods (fruits, meat. .. ), whole or in pieces, in highly concentrated
solutions (sugar, salt) [1, 2].
Basically soaking water rich foods in concentrated solutions induces two main couter-
current mass transfer phenomena:
• an important water outflow from product to solution ( up to 70% of the initial
product weight) ; this "drying effect" is particularly attractive since carried out away from
oxygen, at temperatures as low as 30 or 50°C (no phase change) and within short times
(a few hours);
• a solute transfer from solution to product ; this " formulation effect" makes it
possible to insert a desired amount of preserving agent, solute of nutritional interest, or
sensory quality improver into the product.
Product's own solute leaching may also be observed . Although in low quantity, this
loss is essential regarding the final product's organoleptic (acidity for instance) and
nutritional (mainly vitaminic and mineral) qualities . In most cases, the lower the leakage,
the better the process. Mass tranfers are shown in Figure 1 .
The loss of water is usually imputed to osmosis phenomena through the cell "semi-
permeable" -Le. water permeable / solute repellent- membranes. This is the reason why
such processes have been often referred to as "osmotic" dehydration processes. However,
398
it has been proved recently (by immersing model gel foods and observing identical effects)
that semi-permeable membranes are not a necessary condition for high water loss with only
marginal solute pickup. Therefore, it seems better to gather and refer to such processes as
"Dewatering and Impregnation Soaking processes", or briefly DISP as suggested by
Raoult-Wack et al.[ 3,4] .
Up to now, DIS processes have been mainly empirical and not much worked out as
compared to conventional drying methods. This is the aim of the present paper to give fur-
ther information on the influence of operating variables and mechanisms involved in water
outflow, external solute entrance and own-product solute leakage.

concentrated solution product

WATER

SOLUTE(S)

product·s own solutes

Figure 1 . Schematic drawing of mass transfer in soaking processes

BASIC PRINCIPLES OF MASS TRANSFER

Experiments were conducted with either agar gel cubes or fruit pieces (fresh or frozen
apple) some ten millimeters in size, immersed in sucrose or PEG (200 to 20000) solutions
[3 , 5] . Fructose was chosen for "marking" own solute leakage. Various concentrations of
solutions were tried in the range between 10 and 70 % (w/w of solution» . The temperature
was varied from 30 to 70°C and good mixing conditions prevailed. The influence of opera-
ting variables on water 10ss(WL) ,external solute gain (SG) and own solute leakage( SL) of
solids was inferred from dry matter and weight analysis, as well as HPLC measurements.
In most common operating conditions, mass transfer mainly occurs during the fIrst
two hours . Then, exchange rates plainly decrease: WL stops whereas SG goes on increa-
sing very slowly. Hence, the product tends to gain back weight . A lower temperature
induces an increase in the processing time . As a general rule, the higher the external solute
molecular weight , the larger WL and the lower SG . Regarding SL , a signifIcant increase
with temperature and treatment time may be noted.
Typical performances for WL and SG versus initial interface concentration difference,
~C , are shown in Figure 2 . One can see that WL increases markedly with external solute
concentration. At low ~C, SG is greater than WL ("impregnation" situation) ; for interme-
diate values, a maximum is reached; then, for higher concentrations, SG decreases below
WL ("dewatering" situation) . High WL thus proves to be detrimental to SG . This fact has
been related to the formation at the very beginning of the process, and at the periphery of the
solids, of a higly concentrated layer of external solute due to diffusion unbalance between
water and solute apparent diffusivity inside the ternary gel system .When working with fro-
zen fruits (no membranes), phenomena are quite similar. With fresh tissues, i.e. in the pre-
sence of semi-permeable membranes, and small size external solutes that well diffuse inside
the product (pEG 200 as an example), osmotic effects also take place along with diffusion
and related effects (peripheral barrier, solid deformation) .They favour water loss and own
solute leakage by cell breaking . On the contrary , with worse diffusing solutes (PEG 3000
for instance) , WL and SL rates are lowered.
Various models have been proposed to account for all these results.
 399
70
WL , SG (% w/w of initial product)
60 WL

50
40
30

20 SG

10

0
0 10 20 30 40 50 60 LiC
(%w/w)

Figure 2 . Treatment of agar gel cubes in sucrose solutions

( Dimensions : 9 mm ; agar content: 4% ; temperature: 50°C; time: 3 hours )

MAIN TECHNOLOGICAL ASPECTS

Two major characteristics of DIS processes do make the difference as compared to

conventional drying methods:
• a soaking process achieves a twofold transformation of the food item by both a
dewatering effect, and a fOlmulation effect related to impregnation plus leaching;
• a soaking process does not generally produce stable products; it has to be used
as a preprocessing step before a complementary processing, such as final drying,
pasteurizing or addition of preservative agents.
Moreover , it is worth noting that even items in which tissue structures have been
damaged by ripening, as well as chemical or heat pre-treatments, can be easily treated.
Finally , starting from basic mechanisms previously presented , one can imagine new
ways to improve performance such as the use of solutions mixing low and high molecular
weigth solutes, or succesive soaking steps of the same particles in various type solutions.

REFERENCES
1. Le Maguer , M. , Osmotic dehydration: review and future directions. In Proceedinss of
Symposium Pro~ess in Food Preservation processes, Brussels, 1988, 1 ,pp.283-309
2. Raoult-Wack, A.L. , Guilbert, S. and Lenart, A. , Recent advances in drying through
immersion in concentrated solutions. In Drying of solids, ed. A.S.Mujudar, Internatio-
nal Science Publisher, New York, 1992, pp.21-51
3. Raoult-Wack, A.L. , Guilbert, S. , Le Maguer , M. and Rios , G.M. , Simultaneous
water and solute transport in shrinking media -Part 1 : application to dewatering and
impregnation soaking process analysis. Drying TechnololO', 1991 ,9, 3 , pp.589-612
4. Raoult-Wack, A. L. , Botz , O. , Guilbert, S. and Rios , G.M. , Simultaneous water
and solute transport in shrinking media -Part 3: a tentative analysis of the spatial distri-
bution of the impregnating solute in the model gel. DryinS Technology, 1991 , 9 , 3 ,
pp.630-642
5. Saurel, R. , Contribution a l'etude des transferts de matiere en orr de produits biologi-
ques, PhD thesis, University of Montpellier 2 , Montpellier , Mars 1993
 DEW ATERING THROUGH IMMERSION IN SUGAR/SALT CONCENTRATED
SOLUTIONS (" Osmotic Dehydration")
AN INTERESTING ALTERNATIVE FOR SEAFOOD STABILISATION

A.L.RAOULT WACK and A.COLLIGNAN

CIRAD-SAR - B.P. 5035- 34032 Montpellier Cedex - France.

ABSTRACT

Partial dewatering of fish (cod and trout) and seaweed was achieved through soaking them in
mixed concentrated solutions (sucrose/salt or corn syrup/salt). Influences of concentration
and composition of the solution on water loss, sugar gain and salt gain kinetics were studied.
An experimental design (Doehlert network) was used. This methodology allows to estimate
sugar/salt interactions. Results show how the presence of sugar can enhance water loss and
hinder salt entrance, which can be particularly interesting as an alternative to traditional
processing ( time duration reduction, salt entrance control).

INTRODUCTION

Traditional salting processes (1) consist in furthering solute penetration into the product, and
limit water loss. In the case of seafood, the impregnation stage is often followed by air-
drying and/or smoking, which lead to products such as smoked salmon or trout, dried fish,
Yudoshi-wakame ... Recently, there has been increasing interest in treatments of food items in
concentrated solutions so as to enhance water removal and control solute incorporation.
These techniques have been refered to as "Osmotic Dehydration" (2), or more recently as
"Dewatering and Impregnation Soaking Process" (3), so-called "DIS process". The objective
of the present work was to study the interest of DIS process for seafood using mixed blends
(sugar/salt), with particular interest in salt and sugar interactions.

MATERIALS AND METHODS

Fishes were cut into parallelepipedic filets (150x60x8mm 3 ). Seaweeds (Undaria

pinnatifida) were cut into strips (10 x15xO.5mm 3). Factors studied were salt (Cst) and sugar
concentration (Csu) from 0 to saturation, and time (1 to 24 hour for fish; 0 to 47 min for
seaweed). Temperature was lOoC for cod and 45°C for seaweed. An experimental design -
Doehlert network (4)- was used for each product studied. Samples were soaked in mixed
blends water+sait +sugar (sucrose or corn starch syrup). Samples were analysed for water,
salt and sugar contents. Methods are described elsewhere (5). Water loss (WL), salt gain
(StG) and sugar gain (SuG), expressed in g/IOOg initial product, were inferred.
 401
RESULTS AND DISCUSSION

Figure I presents the evolution of WL, StG and SuG versus time for cod soaked in a
salt/sucrose!. blend for the center point concentration conditions of the experimental design.
Results showed that 54% water loss could be achieved. Sugar uptake in cod was much higher
than with trout (5),which could be related to the difference in initial fat content (0.3 and 4.6
for cod and trout respectively). This sugar uptake may be unacceptable from a sensorial point
of view. Figure 2 presents the evolution of SuG for various com starch syrups, (DE 21, DE
38) which shows how the use of high molecular weight solute instead of sucrose could
strongly reduce sugar uptake, whereas water loss remained unchanged.

Figure 3 shows the influence of Cst and Csu on StG for cod filets at 4h54
(corresponding to center point time duration). For low Csu, StG increased markedly with
Cst. For high Csu, StG was independant from Cst and remained low. For instance, when
Csu = 0 and Cst saturated, StG = 8.7%. When both are at saturation, StG = 0.9%. Therefore,
sugar may hinder salt entrance which can be related to the formation of a concentrated
superficial sucrose layer (3).

Figure 4 shows the influence of salt and sugar concentration on water loss for
seaweed at 3 min 53 sec (corresponding to center point time duration). For high sugar
concentration, water loss increased markedly when salt concentration increased, whereas
without sugar, water loss remained low. This strong dewatering effect can be related to high
concentration levels and interactions implemented with mixed salt/sugar solutions, as
compared to salt solutions.

CONCLUSION

This mass transfer study shows the interest of DIS processes for the treatment of seafood. As
compared to traditional processing, it shows how the traditional salting/smoking/drying
sequence could be reduced thanks to one single dewatering and impregnation soaking
operation. Salt entrance can be better controlled thanks to the sugar/salt interactions, and
sugar uptake can be reduced by the use of high molecular weight compounds. Of course,
microbiological validation of such processes must be carried out.

WL, StG, SuG (% initial product)

60,--------------------------, 6 SuG (% initial product)

50 5
•
40 4
+ Water loss • sucrose
• sugar gain 3 c DE 21
30
• DE38
x salt gain 2
20

10
•
o 0
2 8 10
o~~~~~~~~~ 4 6
time (h)
o 10 20 30 40 50 Figure 2 : Influence of sugar molecular
time (h) weight on SuG for cod filets soaked at
Figure 1 : Mass transfer kinetics of cod filets lOoC in salt/sugar solution (Cst = 350
soaked at lOoC in salt/sucrose solution gil water, Csu=1200gll water)
(Cst = 175g1l water, Csu =950 gil water).
 402

4.75

350

Cst (g/l water)

o

Csu (g/l water)

Figure 3: Response surface StG=f(Cst,Csu) for cod filets at t=4h54

/~------
I ;;:'-1
Y'III
WL (%) ( /

40!
! I
t I II
f
•
II I.
?O ~ ij
l .------7II 350
.,-"
o .7
_ _ .~ ___~~,/r 175 Cst (g/l water)
o
950 -. 1900 0
Csu (g/l water)

Figure 4: Response surface WL =f(Cst,Csu) for seaweed at t=3 min 53

REFERENCES

L Del valle, F.R. and Nickerson, J.T.R. Studies on salting and drying fish. II. Dynamic
aspects of the salting of fish .. J. of Food Sci., 1967, 32, 218-224.
2. Ponting J.D., Walters G.G., Forrey R.R., Jackson R., Stanley W.L. Osmotic dehydration of
fruits. Food Techno!., 1966, 20, 125-128.
3. Raoult-Wack A.L., Guilberts S., Lenart A. Recent advances in drying through immersion
in concentrated solutions. In : Drying of Solids. Ed. A.S. Mujumdar, , 1992, 21-5l.
4. Doehlert D.H., 1970. Uniform shell design. Applied statistics, 19, p.231.
5. Collignan A. and Raoult-Wack A.L. Dewatering through immersion in sugar/salt
concentrated solutions at low temperature. An interesting alternative for animal foodstuff
stabilisation. In : Drying 92, Ed. by A.S. Mujumdar, 1992, 1887-1897
 OSMOTIC DEHYDRATION OF PEAS AND ITS EFFECT ON DRYING

F. KAYMAK AND T. ~AKALOZ

Department of Food Engineering, Faculty of E~gi~eering
Ege University, Bornova, Izmir, 35100, TURKIYE

ABSTRACT
Peas have been partially dehydrated by immersion in pure sucrose and
sodiumcitrate solutions or sucrose/sodiumcitrate mixed solutions prior to
air drying. The influence of concentration and composition of osmosis
solutions, immersion time and agitation on the mass transport data
described in terms of water loss and solid gain and on water activity data
were investigated. A mathematical model for this phenomena was proposed and
the apparent diffusion coefficients of water, Da were determined. The
influence of osmotic dehydration on subsequent air drying behaviour and
product quality, relative to air dried peas was also studied. Osmotic
dehydration resulted a successful method for improving the final product
quality.

INTRODUCTION
Recently, osmotic dehydration has been proposed as a preconcentration step
in drying, freezedrying and freezing fruits and vegetables (1,2). Osmotic
dehydration is based on removing only a part of the natural water content
of fruits and vegetables by immersion in concentrated solutions of soluble
solids, having a higher osmotic pressure and lower water activity. During
the osmotic dehydration, water and low molecular weight substances diffuse
out of the foodstuff in to the solution, while solutes diffuse counter-
currently from the solution into the food. Therefore osmosis is a
simultaneous water and solute diffusion process. Different aspects of
osmotic dehydration have been studied extensively and the effect of several
variables and process parameters on the quantity and the rate of water
removal has been investigated. The parameters studied were the
concentration and composition of osmosis solution and the osmosis time
(1,2,3), temperature and agitation of solution (1) and solution to sample
ratio (3). The present study was carried out to investigate mass transfer
mechanism during osmosis of peas and the effects of some parameters on this
mechanism, and to study on a model characterizing diffusion of water from
the food. The effect of osmotic dehydration on subsequent air drying
behaviour was also studied.
404
MATERIALS AND METHODS
Green peas purchased from a local food supplier were used as a material in
this research. After podding and cleaning, peas were graded for size. The
average diameter of peas was determined as 0.9 cm. The osmotic solutions
were prepared as pure solute or mixed solute systems in the following total
concentrations by weight; sucrose, 30, 40, 50 and 60 %; trisodiumcitrate,
15 and 30 % and sucrose / citrate mixture, 50, 60 and 70 % with a varying
compositions. Osmotic dehydration was carried out in a batch system at 30°C
and samples were immersed in concentrated solutions for 24 hr under
constant agitation or static conditions.In our experimental conditions, the
fixed time used was relatively long in order to reach the mass transfer
equilibrium between the sample and the osmosis solution. A ratio of sample
to solution 1:4 by weight was used. Periodically, osmosed samples were
removed from the solution, drained, blotted with a filter paper and
analysed. The osmosed sample was weighed and its total dry matter content
and water activity values were determined. To analyse the data, the
following parameters were determined for each sample; % water content
(WC); water loss (WL), (g / 100g initial product); solid gain (SG), (g /
100g initial product) ; % weight reduction (WR) ; % total solids (TS).
Samples osmosed for 24 hr in selected solutions at 60 % total solids were
chosen for subsequent air drying. The osmosed peas as well as the untreated
control sample were dried at 65°C, using infrared energy. The change of
weight and moisture of samples were determined.
RESULTS AND DISCUSSION
The changes of mass transfer parameters during osmotic dehydration of peas
in 40% sucrose / 20% citrate mixture solution at 30°C under agitated
condition are shown in Fig. 1.

70

60

W 50 WL
(j
Z
«
I
40
WR
U 30 we
0~ 20

10 SG

0
0 4 8 12 16 20 24 28

TIME (h)
Figure 1. Mass transport data for osmotic dehydration of peas with
40 % sucrose / 20 % citrate solution.
 405
As seen in Fig. 1, the most important mass transfers take place in the
first hours of the process, afterwards the transfer rate decreases and the
dynamic equilibrium between pea and the osmosis solution is reached. It
was shown that pea samples has attained to equilibrium level between 8 and
24 hr of osmosis, depending on osmosis solution type. In all osmotic
solutions tested, water loss increases with increasing concentration of
solution, but solid gain are not significantly changed by concentration.
It was determined that water loss and solid gain increase with agitation of
osmotic solution. When pure citrate and sucrose solutions at 30 % solids
were compared, it was found that the citrate solution is very effective in
osmosis while sucrose does not show a significant effect. At the 60 % total
solids level, mixed sucrose/citrate solutions were determined to give high
amounts of water loss and solid gain and a larger decrease of water
activity than sucrose alone. It may be attributed to the higher molar
concentration of the sucrose/citrate solution and ionization of citrate(3).
In order to clarify the effectiveness of osmosis solutions, kinetics
of mass transfer was analysed, using experimental data. The apparent
diffusion coefficients representing water loss during osmotic dehydration
were determined, assuming unsteady state Fickian diffusion (4). Values of
Da were found to be in the range 0.91-1.27*10- 10 m2/s for 30-60 % sucrose
solutions i 7.07-7.63*10-10 m2/s for 15-30 % citrate solutions and 1.50-
5.20*10- 10 mz /s for 50-70 % mixed sucrose /citrate solutions.
The effect of osmotic dehydration on subsequent air drying behaviour
was investigated for peas osmosed in either 60 % sucrose or 40 % sucrose /
20 % citrate, comparing to the natural samples. Essentially, the osmotic
step does not exert any significant influence on air drying rate. On the
other hand, the time required to dry peas to a moisture content
corresponding to a water activity of 0.45 is reduced for osmosed samples
due to water loss and solid gain in osmosis step and higher, allowable,
final moisture content for finished product.
CONCLUSIONS
Osmotic dehydration appears to be a succesfull method of pre-concentrating
peas prior to air drying. A comparison of various osmosis solutions showed
that mixed sucrose/citrate at 60 % total solids are the most effective
solutions, giving a higher dehydration rate and alarger decrease in water
activity. Therefore, 50-80 % of initial water content of peas were removed
and the water activity values were reduced to about 0.9, depending on the
concentration and composition of osmosis solutions used.It was found that
initial osmosis step shortens by 55-70 % the drying time, depending on the
type of osmosis solution and process conditions.
REFERENCES
1. Hawkes, J. and Flink, J.M., Osmotic concentration of fruit slices prior
to freeze dehydration. ~ Food Process. Preserv., 1978, 2, 265-84.
2. Islam, M.N. and Flink, J.M., Dehydration of potato. II. Osmotic
concentration and its effect on air drying behaviour. iL. Food Technol.,
1982, 17, 387-403.
3. Lenart, A. and Flink, J.M., Osmotic concentration of potato. I. Criteria
for the end-point of the osmosis process. ~ Food Technol., 1984, 19,
45-63.
4. Crank, J., The Mathematics of Diffusion, 5 th ed., Oxford University
Press, London, 1970.
 STUDIES ON THE SORPTION CHARACTERISTICS
OF BULK WHEAT AND CANOLA

Weiguo Lang and Shahab Sokhansanj,

College of Engineering
University of Saskatchewan, Saskatoon, SK S7N OWO Canada

ABSTRACT
Drying characteristics of wheat and canola were studied. Experimental results showed that
the total drying time of an equal grain volume was almost the same. The drying model was
improved by including grain shrinkage and variable physical properties.

INTRODUCTION
Moisture sorption characteristics of a grain are related to its composition and structure.
Wheat contains mostly starch while canola is an oil-rich seed. A wheat kernel is of prolate
shape with a deep crease. The kernel measures 6.0, 2.5 and 2.3 mm in major, intermediate
and minor diameters, respectively. A canola (Tobin variety) seed is spherical with diameter
of 1.8 mm. The differences between the two grains may affect their moisture transport
characteristics.

Significant bulk shrinkage up to 25% (Lang, 1992) affects accuracy of drying model
model prediction. To accurately predict the drying, it is essential to develop a drying model
which is able to trace the changes of the drying bed depth by incorporating a shrinkage
factor. The goal of this work was to improve the drying model based on the new
information obtained. The work included: generating accurate experimental data; improving
Bakker-Arkemadrying model; and verifying the improved model.

DRYING EXPERIMENTS
No.1 hard red spring wheat (Triticum sp.) kenyon variety and No.1 canola CBrassica SQ.)
Tobin cultivar were used in this study. Samples in 30 kg lots were rewetted to uniform
moisture contents. Fig. 1 shows the experimental set-up. The drying column was made of a
305-mm diameter, 800-mm length steal cylinder and was insulated with 70-mm thick
fiberglass. The air from a centrifugal fan was heated by a steam heat exchanger. Grain was
removed from sample ports along to column for moisture measurement. The air and grain
 407
temperature was measured by thermocouples. An L VDT was used to measure the change in
the position of grain surface during drying. The exit air relative humidity was computed
from the measured wet and dry bulb temperatures.

Fig. 1 Schematics of deep-bed drying experimental set-up

DRYING MODEL ANALYSES AND IMPROVEMENT

Simulation using Bakker-Arkema's drying model showed differences from the present
experimental data. The model was modified to include bed bulk shrinkage:

(1)

where L\x is a change in position and L\M is a change in the moisture content. The shrinkage
coefficient, A.m, is a function of the air relative humidity and temperature (Lang, 1992).
From experimental observation, a factor was developed:

(2)

where l; is layer constant which is a temperature dependent variable. The drying constant
in the thin layer drying equation was multiplied by fx . The air conditions in the first layer
was also multiplied by a time factor:

fa = 1 - exp(-91't) (3)

where 't is temperature dependent time constant and 9 is drying time, h.

408
RESULTS AND DISCUSSION
The time to remove 18 percentage points MC for the top layer was 26 h for wheat and 28 h
for canola . It appeared that the times to dry the whole bed for the two grains are similar to
the same proportion. Drying curves for canola bed showed sharper edges than those for
wheat. This indicated that canola lost its moisture more easily than wheat. The temperatures
of all the layers dropped sharply within 10 min of drying. The drop and subsequent rise in
air temperature were quicker in the bottom layer than in the upper layers.

The improved model greatly increased the accuracy for prediction of MC in the
bottom and top layers and perfectly predicted for the location 3. For temperature, except for
the bottom layer where the prediction by the improved and original models was essentially
the same, the accuracy of prediction by the improved model was much higher than that by
the original model for the rest locations.

28 Top (surface) CANOLA

24
::c~
~ 20
C
~
c 16
0
u
!:!
.a 12
.~
0
::E 8

4
() 5 J() 15 20 25 30

Time (h)

Figure 2. Typical moisture content during drying

Compared to the experimental data, it appeared a better prediction by the improved

model. The improvement in prediction accuracy increased as the drying location moved
further from the bottom. The improved model performed better for wheat drying.

CONCLUSIONS
1. At a given location in a bulk:, rates of MC and temperature change of canola were faster
than those of wheat.
2. The improved Bakker-Arkema drying model enhanced the accuracy of drying prediction
for all locations in the drying bed.

REFERENCES

Bakker-Arkema, F.W., W.G. Bickert and R.I. Patterson. 1967. Simultaneous heat and mass
transfer during the cooling of a deep bed of biological products under varying inlet
conditions. I. Agric. Engng. Res. 12(4): 297.
Lang, W. 1992. Drying and rewetting characteristics of canola and wheat. Unpublished Ph.D
Thesis.Dept of Agric. & Biores. Eng., University of Saskatchewan, Canada.
DIFFUSIVI TY OF WATER AND SOLUBLES DURING REHYDRATION OF
OSMO-CONVECTION DRIED PLANT TISSUE

ANDRZEJ LENART, BARBARA IWANIUK

Department of Food Engineering
Warsaw Agricultural University (SGGW)
02-766 Warszawa, ul.Nowoursynowska 166, Poland

ABSTRACT

The research comprises mass transfer in osmotic. convection

and osmo-convection dried apple, carrot and pumpkin during its
rehydration in water. The experiments showed that mass
exchange depends on the method of dewatering and the raw
material properties. Analysis based on the water and the
solubles diffusivities shows that the regarded coefficients
depend on the water content of the samples.

INTRODUCTION

The initial osmotic dehydration preceding convection drying

considerably affects chemical composition as well as physical
properties of the final product [1].
Only few reports are concerned with the rehydration
capability of osmo-convection dried plant tissue. Little
information is available as regards the diffusion during
rehydration of osmotic and osmo-convection dried foods,
properties of obtained product and rehydration phenomena [2].
The aim of this work is to analyze the effect of initial
osmotic dehydration on mass transfer in osmo-convection dried
material during its rehydration in water.

MATERIALS AND METHODS

Osmotic dehydration of raw apple and blanched carrot and

pumpkin was carried out in 61,5% saccharose and 67,5% starch
syrup solutions without mixing, at 1:4 (w/w) material to
solution ratio. The process was carried out at: temperature of
30 to 50°C and time from 1.5 to 4 h. After osmotic dehydration
410
samples were surface washed with water, blotted with filter
paper and divided into two parts. One part was dried by
convection. Drying was done in a convection dryer with a
constant air flow velocity of 1.6 mis, at a temperature: 60°C
- apple and 70°C - carrot and pumpkin until the equilibrium
moisture content was reached. The load of dryer was 7 kglm .
Dewatered by osmosis, convection dried and
osmo-convection treated samples were rehydrated by immersion
in water [3]. The description of diffusion during rehydration
is based on the Second Fick~s Law [2].

RESULTS AND DISCU~-:r ON

Changes o£ Water Content During Rehydration

Changes of water content in samples during rehydration are
shown in Fig.1. Independently on the kind of dewatered raw
material, the uptake of water is higher for convection dried
samples as compared to those dewatered by osmosis.
Mass Transport During Rehydration
A considerable influence of water content during rehydration
on values of diffusivities of water (Dv) and solubles (Ds) was
noticed. Independently on the method of dewatering the
correlation between the osmoactive substance and the water
transfer was observed (Fig.2). For each loss osmotic substance
and the dehydration method used, remarkable influence of the
material on the mass exchange during rehydration appeared.

15
E
"0
01

....
C
~

Q/
~ 5
o<.)

L.

....
Q/

o 0
~ 4-------+------+------4-----~r_-----+--
o 1 2 3 4 5
Ti me. h

FIGURE 1. Changes of water content during rehydration of

dewatered carrot: 1 - osmosed in saccharose solution, 2
osmosed in starch syrup solution, 3 - dried by convection, 4
- dried by osmo-convection in saccharose solution, 5 - dried
by osmo-convection in starch syrup solution
 411

5 ---------0--- __
.c::
4
2 3
(">1-
E
r-..
~ 3
.~

o
2

c>----o
o +-~_,~--._+_.-_.~_.--r_~,__r~--~~._~~~~--

o 5 15 20

FIGURE 2. Effect of the kind of material on mass exchange

during rehydration of osmo-convection dried sample: 1 - apple,
2 - carrot, 3 - pumpkin. Osmoactive substance: --- saccharose,
--- starch syrup

CONCLUSI OMS

Independently on the kind of dewatered raw material, the rate

of water content changes during rehydration is higher for
convection dried samples as compared to that dewatered by
osmosis. The mass exchange during the rehydration of apple,
carrot and pumpkin shows that the water diffusivity is not
strongly influenced by the solubles diffusivity.

REFERENCES

1. Lenart,A., Saccharose .a factor modifying

oamo-conyective drying apples, SGGW-AR Press,
Warszawa, 1988, pp. 1-87.
2. Lenart,A., Lewicki,P.P. and Dziuda,J., Diffusion
properties of apple tissue during technological
processing. In Engineering ~ EQQd, ed. W.Spiess and
H.Schubert, Elsevier Applied Science. London, 1990,
pp. 731-740.
3. Mazza,G., Dehydration of carrots. Effects of predrying
treatments on moisture transport and product quality. ~
EQQd Technol , 1983, 18, pp. 113-123.
 STUDY OF REHYDRATION BEFORE EATING ON DEHYDRATED VEGETABLES *

ZHANG MIN, WANG CHENZHI, MA XIAOYU, LI CHUNLI

Northeast Agricultural College, Harbin 150030, P. R. China

ABSTRACT

Rehydrating tests before eating on dehydrated vegetables have been made. The rehydration ratio (Rf )
and restoration ratio (Kf ) curves under different soaking time and temperature have been obtained.
The tenderness values have also been determined by the tendermeter. Finally, the effects of
rehydration on cellular structure and soaking on restoration have been discussed.

INTRODUCTION

Dehydrated vegetables retain the nourishment and flavour of fresh vegetables for the most part. This is
welcomed in the international market (M. Zhang, 1989). However, quality of rehydration before
eating will have a direct effect on its cuisine. So it is necessary to study the properties of rehydration on
dehydrated vegetables.

METHOD AND EQUIPMENT

The selected samples include: dehydrated pieces of sword bean, cucumber, celery, eggplant, carrot,
green onion, pepper, cabbage, tomato, and horseradish with moisture content of 8% (d.b.).
The objectives of the tests are: (1) rehydration ratio R f = GriGg, where Gp Gg are weights of
samples after and before rehydration, g; (2) restoration ratio Kf = (GrlGx ), 100(%), where G x is the
corresponding weight of fresh sample before G g is dried, g; (3) tenderness Sl = IlL, where L is
shearing destroying force of the sample, kgf/cm2 • The higher the values of three objectives, the better
the properties of rehydration are.
The thermostat for rehydration is one of model HH-Sll-4 with controlling accuracy is ± 1°C.
Tenderness determination is made in the tendermeter for food as shown in Figure 1, which can
measure a maximum shear force of 20 kgf (with accuracy of 0.1 kgf).

RESULTS AND DISCUSSION

1. The rehydration of dehydrated vegetables under different soaking temperatures

The ten kinds of dehydrated vegetables with the moisture of 8% (d. b.) were used for the rehydration
tests. The results are as shown in Table 1. From the table, we know the time necessary to saturate R f
shortens with the increase of soaking temperature. From the data of the samples, at 100°C, we know

• The study is supported financially by China National Natural Science Fund.

 413

Table 1 The times (h) needed under

the 90% saturated rehydration ratio
Temperature tC)
Index 20 60 100 Rf
sword bean 2.0 1.0 0.5 6.0
cucumber 4.5 2.5 1.0 4.2
celery 3.0 2.0 0.8 4.0
eggplant 2.5 1.5 0.4 6.1
carrot 2.5 2.0 0.5 5.5
green onion 2.0 1.5 0.3 7.3
pepper 3.0 2.0 0.5 6.8
cabbage 2.5 1.5 0.4 4.8
tomato 2.5 1.5 0.3 5.3
horse radish 5.0 4.0 2.0 4.2

we know that the time needed in direct cooking is long, so the importance of rehydration is obvious.
The difference of needed time at three soaking temperatures is obvious because of the difference of the
cellular structure among the samples. The soaking temperature 100°C should be used during
rehydration.

2. Tenderness tests of dehydrated vegetables after different soaking time

In order to compare the difference in masticability after rehydration, the samples after rehydration
at different soaking times and using fresh samples have been used for the tenderness measurement.
The results are shown in Table 2. From the table, the longer the soaking time is, the bigger the value of
its tenderness is. The tenderness of fresh samples is the biggest one, while the value of dried sample has
the smallest. Since there is difference among the cellular structure of strains, the tenderness of samples
has different values. The smallest value belongs to horse radish, while the biggest belongs to tomato.
The differences in tenderness value between the dried and the fresh sword bean and eggplant is little.
The mastic ability of the dried is near that of the fresh.

CONCLUSIONS

1. The time needed up to the saturate R f shortens with the increase of the soaking temperature. The
best rehydration temperature is 100°C.

Table 2 The tenderness after different soaking times on dehydrated

vegetables
Time (h)
Index o (d.s.) 0.5 1 5 (R js) fresh samples
sword bean 108.7 347.2 390.6 584.8 1136.3
cucumber 204.4 546.4 680.3 757.6 7692.3
celery 66.7 68.2 71.4 89.0 12.50
eggplant 371.7 432.9 495.0 534.8 1538.4
carrot 125.6 236.4 249.4 258.4 1282.0
green onion 117.6 280.9 495.0 534.8
pepper 242.1 268.8 304.0 332.2 1851.8
cabbage 246.9 327.0 348.4 366.3 3030.3
tomato 369.0 1000.0 1052.6 1369.8 9370.5
horse radish 57.1 65.4 68.0 76.9
414
2. The longer the soaking time is, the bigger the value of tenderness of dehydrated vegetables is. The
biggest one belongs to the fresh sample, and the smallest one belongs to the dried sample. The
difference between the fresh and saturated rehydrated samples of sword bean and eggplant
tenderness is little.
3. It is because the shrunk cellular structure cannot restore fully that rehydration of dried products
cannot make the dried restore fully.
4. The effect of the soaking time and temperature on rehydration has been discussed.

REFERENCES

1. M. Zhang, The study of the far-infrared drying, pretreatment, and rehydration property on several
vegetables, the Master paper of Zhejiang Agricultural University, 1989.
2. L. Z. Pan, Physiology of plants, People's Education Press, 1962.
 EFFECT OF DRYING CONDITIONS ON EFFECTIVE MOISTURE
DIFFUSIVITY IN "CHUFA II TUBERS (Cyperus esculentus L.)

A. ALTARRIBA, A. NUNEz-LEMOS, D. VIDAL

Department of Food Technology. Polytechnic University
P.O. Box 22012. 46071 Valencia, Spain

ABSTRACT
Only falling rate period was found in "chufa" tubers (Cyperus esculentus L.) drying under
different experimental conditions which were held constant during each run. Influence of tubers'
size was also studied. Results show that drying rate is controlled by transfer of moisture from
the inside to the surface. The effective moisture diffusivity was estimated by the method of
slopes, based on the solution of the Fick's equation for unsteady-state diffusion. From the
results obtained, an ANOV A procedure was applied to determine the relationships between
effective moisture diffusivity and the variables studied. An Arrhenius-type relationship was
proposed in order to extrapolate diffusivities for other drying conditions.

NOTATION LIST

De Effective moisture diffusivity (m 2s- 1) T Temperature ("K)

E Energy of activation (kJ kg-I) t Time (s)
G Air flow rate (kg m- 2 h- 1) X Moisture (kg water kg- 1 dry matter)
R Gas constant (8.311 J mol- 1 oK-I) Subscripts e Equilibrium
r Radius (m) o Initial

INTRODUCTION
Tuber named "chufa" or yellow nut (Cyperus esculentus L.) is farmed in Mediterranean
countries for oil production and for the elaboration of a typical Spanish sweet beverage called
"horchata", obtained mixing water, sucrose and the tubers, previously dried by natural ambient
air convection. The time required for drying under variable atmospheric conditions is long and
large extensions are needed. Besides, this extended drying time is favourable for bacterial and
moulds growth. To be able to achieve a product of low moisture and water activity in a shorter
time it is planned to study a drying process of the tubers under warm-air current.
The basic aim of this study is to represent the experimental drying kinetics of this
unstudied product by means of a empirical model in a way that all experimental information
obtained is contained in its parameters. The analysis of Fick's second law regarding its
application for evaluating effectives diffusivities in other controlled processes.
416
MATERIALS AND METHODS
Apparatus
The laboratory drier consists of a fan S&P-CBT60. The air flow rate was measured by two pitot
tubes and a manually-operated valve was used to adjust the air flow rate in the dryer. The
heating system consisted of a electric 1000 W heater placed inside the chamber. The temperature
was measured with a thermometric probe at the entrance to the product base. The sample is
placed in a graduate tube as in a basket and the air flow was perpendicular to the drying bed.

Procedure
The newly-gathered and washed product was sifted two fractions of 50%, each one with an
screen of circular orifices of 10 mm in diameter. Each portion was dried separately in a different
conditions controlled of air flow and temperature, so as to study the influence of the size on the
particle in the drying. The moisture content of the initial and final product was obtained by
means an infra-red oven coupled to a Mettler 400 scale. The experimental design utilized to
determine the drying conditions during each run was a design centred on four levels of
temperature from 35 to 72.5 °C and three levels of air flow from 6000 to 12000 kglm2h.

RESULTS AND DISCUSSION

Drying analysis

Drying curves (X versus time) and drying rate curves (dX/dt versus time) were plotted.
Both representations make clear that only exists one decreasing drying phase which in this case
can be divided into two falling rate periods.
Supposing the coefficient of effective diffusion constant throughout the falling rate
period, the initial moisture distribution uniform and disregarding the effect of shrinkage on the
product, the solution to Fick's second law for a spherical geometry such as that the "chufa"
tuber was used (1). For the first term of the series and taking as values of the Xe the one
described by the authors, a linear regression analysis was employed to calculate the effective
diffusion coefficient from the slope of the straight line (see figure 1). Values of effective
moisture diffusivity De are in consonance with the one obtained by other authors (2&3).
Apparent diffusivities are between 0.55910- 10 m2s- 1(35°C) and 2.62210- 10 m2s- 1(72.5°C).
In (X-Xe/Xo-Xe)
1~-----'------'-----~----~

C 35°C
o 47°C

~~-----+--~--r-~--;------;

o 200 400 600 800

Time (min)

Figure 1. Dimensionless moisture change as a function of time and temperature.

Air flow 6000 kg/m2h. Particle radius 0.011 m.
 417

Influence of the temperature, air rate flow and radius of the particle
The influence in the coefficient De of the three factors has been analysed by means of
ANOV A procedure. The significance of the factors is expressed in the following table:

TABLE 1
A nallYSIS
. 0 fV anance
. f or M'
olsture Eftectlve DifUSSlvlty
Source df Sum of Squares Mean Square F-ratio Prob
r 1 8.8189 8.8189 6.7798 0.0153
G 2 16.3614 8.1807 6.2891 0.0061
T 3 425.5350 141.8450 109.0500 0**
Error 25 32.5192 1.3008
Total 31 486.6520

An Arrhenius type of relationship with the form De= a exp (-eIRT) has been used to
modelise the effect of the temperature on the coefficient of diffusion. Diffusivity values,
calculated at all temperatures by means of the aforementioned method, versus the inverse of the
absolute temperature are plotted (see figure 2).
The values of energy activation are in consonance with those described by other authors
(2&4) and close to those of products of similar composition ("chufa" is rich in fats, starch and
sugar).
In (De (m2/s) 10 10
3

.. ..
'«'~'~""~l'~~'"~''''''' ~ ~~.
I
0~---4-----+----~----~--~
2,8 2,9 3.0 3,1 3,2 3,3
lIT (K) 10 -3

Figure 2. Effective diffusivity valiation with the temperature. Air flow 10000 kg/m.h.
Palticle radius 0.011 m.

CONCLUSION

Only a falling rate period is found in the drying of chufa. The model of diffusion based on the
solution of the second law of Fick adjusts satisfactorily to the expelimental data. However, it
would be convenient to introduce the shrinkage effect in order to achieve a good prediction.

REFERENCES

1. Pakowski, Z.& Mujumdar, A.S., Basic process calculation in drying. In Handbook of

Industrial Drying, ed. A.S. Mujumdal', Mal'cel Dekker, Inc., New York, 1987, pp.115.
2, Yusheng, Z. & Poulsen, K.P., Diffusion in potato drying. J. Food Eng., 1988,7,249-261.
3. Litchfield, J.B.& Okos, M.R., Moisture diffusivity in pasta during drying .1. Food Eng.,
1992,17, 117-142.
4. Mulet, A., Berna, A. & Rosello, C, Drying of carrots. 1. Drying Models. Drying
Technology, 1989,7(3),537-557.
SIMULTANEOUS HEAT AND MOISTURE TRANSFER WITH HYGROSTRAIN-
STRESS FORMATION IN FOOD UNDERGOING DRYING

K. Hayakawa, T. Tsukada", T. Furuta"", and N. Sakai*""

Food Science Department, Cook College, Rutgers University, P. O. Box 231,
New Brunswick, NJ 08903 USA, "Tohoku Univ., Sendai, Japan, ""Tottori Univ.,
Tottori, Japan. """Tokyo Univ. of Fisheries, Tokyo, Japan.

ABSTRACT

A method was developed for simulating heat and moisture transfer and hygrostrain-
stress formation in food undergoing a drying process. For this, an improved Luikov's
model was coupled with the virtual work principle of elastoplastically deforming body.
Sample simulation results are presented.

INTRODUCTION

It has been known that hygrostrains-stresses are formed within many food during
drying. Since this formation could lead to structural damage of food, many researchers
have developed models for simulating heat and moisture transfer and hygrophysical
changes. These researchers contributed greatly to clarifying interaction between heat
and moisture transfer and hygrostrain-stress formation. However, most of the models
are limited to simplified heat and moisture transfer equations; based on computer
intensive, classical transport equations (mass, momentum, and energy balance); and/or
to one dimensional system. Therefore, the present authors have developed recently
a two dimensional simulation method [1]. The paper is to present the key features of
this method and to present selected simulation results.

MATHEMATICAL MODEL

Luikov's model based on irreversible thermodynamics has been used by many

researchers because of reliability and mathematical simplicity. However, it is
questionable to include a static pressure, p, as a dependent variable in the model
since p being a function of two other dependent variables, temperature (T) and
moisture concentration (C), and to estimate thermodynamically interactive flux, Soret
flux, irrespective of locally available moisture. In view of the above, dependent
 419
variable p was eliminated from the model, and new flux correction parameters were
introduced to account for moisture availability [2]. The developed method assumes
elastoplastic and axisymmetric food with an axial cross-section of any contour.
Hygrostrain-stress formation was estimated applying an initial strain increment method
based on virtual work principle of a deforming body.

Figure I shows sample simulation results on the distributions of the following physical
attributes along a radial direction of a bisected face at 3 hrs. of drying of a starch
hydrate sphere (20 mm initial dia.): Two principal stresses (A), shear stresses (B) and
moisture concentration (e). Double and single bars represent compressional and
tensile stresses, acted on a bisected face, respectively, in Figure IA Their magnitudes
and orientations represent those of stresses. The figure shows two principle stresses,
orthogonal to each other, acting at the intersections of two orthogonal bars.

Figure 1. Distributions of two principal stresses (A), shear stresses (B), and
moisture concentration (e) along a radial direction on a bisected face
of a spherical sample (20 mm initial dia.). For each figure set, the
upper and lower ones are for drying with hm and 0.2 and 2 mis,
respectively.

Figure IB shows two shear stresses at the bar intersections. Their orientations and
magnitudes are represented by those of the bars. Figure Ie shows moisture iso-
concentration lines, ratio of current and initial moisture concentrations. Large tensile
stresses (LTS) were formed at a mid-radial location at 3 hrs (A). However, LTS were
420
formed on the exposed surface at an early stage of drying and location of LTS moved
inward with the progress of drying, LTS being critical for structural damage of brittle
material (the starch hydrate). Figures lA and B show the larger hygrostresses for the
larger convective surface mass transfer coefficients, hm• Figure lC shows faster drying
for the larger hm •

Figure 2 shows simulation results at 3 hrs. drying of 20 mm dia x 80 mm long cylinder

when ~ = 2.0 (moisture concentration ratio, A, two principle stress, B, and shear
stress, C, distributions in a region close to the upper comer within axially bisected
face). When hm = 0.2, hygrostresses formed in the body were less than those shown
in Fig. 2.

A B c
, , .
. . , - , -. , . ,. ,.
I-I. • • I . .. , · .. .
I \
~-.
. I · J. •
I I

-- - \ ... \ , I , '. I· . I I I · I· ·

Figure 2. Moisture concentration (A), two principal stresses (B), and shear stress
(C) distributions in upper corner region within axially bisected face of
20 mm dia x 80 mm long cylinder at 3 hrs. of drying, hm = 2.0.

REFERENCES

1. Tsukada, T., Sakai, N., and Hayakawa, K Computerized model for strain-stress
analysis of food undergoing simultaneous heat and mass transfer. L Food Sci.,
1991, 56, 1438-1445.

2. Furuta, T. and Hayakawa, K Heat and moisture transfer with

thermodynamically interactive fluxes. I. Mathematical model development and
numerical solution. Trans ASAE, 1992,35, 1537-1546.

APPENDIX

This paper is based on work supported in part by the Center for Advanced Food
Technology, a New Jersey Commission on Science and Technology Center, Pittsburgh
National Supercomputing Center Supercomputer time grant, New Jersey State Fund,
Hatch Act Fund, and Rutgers University Computer Services computer time. New
Jersey Agricultural Experiment Station Publication Nos. F-10103-1-93 and F-I0535-1-
93.
 DYNAMIC MODELLING OF FRUIT DEHYDRATION

F. GIROUX( 1), S. GUILBERT (2), G. TRYSTRAM (1)

(1) ENSIA, 1 Ave des Olympiades 91305 Massy, France
(2) CIRAD/SAR, Montpellier, France

ABSTRACT

The paper presents the dynamic modelling of the dewatering and impregnation soaking process. A
compartmental model is developed and parameters are identified from experiments. Both solution
(temperature and concentration) and agitation are taken into account. A specific parameter is
introduced as a modification of external water transfer, to represent the influence of the amount of
agitation.

INTRODUCTION

The dewatering and impregnation soaking process is very useful one as it facilitate the achievement
and the preservation of the quality of product. When pieces of fruit are immersed in a concentrated
solution of salt or sugar, coupled mass transfers are performed. Water is removed from the product
and sugar or salt enters the product. These phenomena are used before drying, for dewatering or for
impregnation to formulate specific products. Variables which influence the process are studied at
length (1). Temperature and concentration of the solution appear to be the most significant
variables. They influence the rate of mass transfer. This rate is dependant upon the boundary layer
around the piece of fruit. This boundary layer is influenced by the agitation and just a few papers
are dedicated to the study of the influence of agitation. Comparisons are discussed for dewatering
with or without agitation (2). But there is no available model. Previous work (3) establishes the
availability of compartmental modelling as a tool to build a dynamic model of the dewatering and
impregnation soaking process (DIS). This approach, including agitation, is adapted to study the
control of the DIS process. Validation is performed using experiments.

MATERIALS AND METHODS

Products
The product used for experimentation is apple (type Granny Smith). Apples are cut in cubes (1.3
cm) and sorted in a such a way that the weight of each piece is 1.5 ± 0.1 g. Before dewatering,
pieces of apple are conserved, for a small time, in a sugar concentrated solution, at the same
concentration as apple. The density of apple is around 730 kg/m3. The total mass of apple,used for
one experiment is 500 g.
422
The solution is a sugar syrup solution obtained by dilution of sugar into water. Several
concentrations are prepared. The basic experiment, which is used for identification of model
parameters and the study of repeatability, is 60° brix and 50°C. The density is 1280 kg/m3. The
total volume of syrup, for one experiment is 9 liters. The initial porosity of the fruit bed is 0.227.

Materials
A phase contactor is built for this work. It consists of a simple cylindric tube with two grids used in
a such way that apple pieces are always submerged in the sugar syrup. A pump is used to create the
circulation of syrup. The pump motor is controlled by a programmable timer. Therefore it is
possible to control the cyclic rate of the pump (figure 1). The flowrate of the pump is variable.
There is no instrumentation, but samples are taken every 15 minutes and analysed at the laboratory
(moisture and sugar content). The temperature of the syrup inside the contactor is controlled.

Methods
A range of experiments are performed in which different solution concentrations and temperatures
are studied. Studies of agitation variation are realized. After laboratory analysis, datas are treated
and two criteria calculated: PE=I00.(W(O)-W(t).(m(t)-m(O») and OS= PE-(lOO.«m(t)-m(O»/m(O»
where m is the mass and W the water content. The duration of an experiment is 180 minutes. The
mass total of the samples extracted during the trials is 7.7% of the total mass of the fruit. This
quantity is assumed to be negligible for the agitation influence study. The accuracy of the
measurement is: d (PE) = ± 2 g/100 g of initial matter and d(OS) = ± 1.3 g/100 g of initial matter.

Hydraulic agitation is performed using the pump. Two control parameters are considered. First the
power of agitation is controlled through the cyclic rate of the pump. This permits the control of the
functioning of the pump with the time basis Tm and the cyclic rate itself; the higher is the time Tm
and the higher is the ratio of time for pump on versus pump off, the higher is the agitation of the
fruit bed. The second variable is the pump flow rate which control the renewal of the syrup in the
contactor. Three states are defined (table 1).

State A State B State C

High agitation High agitation Low agitation
High renewal Low renewal Low renewal
Pump Flowrate O/h) 2500 2500 800
Cyclic rate 1, 1/3, 1/7, 1/25, 1/61), 1/3, 1/7, 1/25, 1/61), 1/3, 1/7, 1/25, 1/61,
0 0 0
Time basis (s) 5 1.5 5
Table 1: States tested for study of hydraulic agitation influence during dewatering of apple pieces.

Fifteen experiments were carried out including repeatability. Transient condition experiments were
performed too, for model validation. Thus, for example, state A is defined for 30 minutes, after
which state B is reached.

Model
The model is derived from previous study (3), it is a compartmental model (2 compartments). Four
parameters are identified to describe internal and external mass transfers. Heat tranfer is neglected.
The simplex method is used to calculate the parameters from one experiment and the validation is
provided with other experiments (temperature from 30°C to 70°C, concentration from 400 brix to
70o brix).

RESULTS AND DISCUSSION

Experimental analysis
The interpretation is focused on the dewatering process. The increase of agitation increases the
dewatering rate: 25 % of increase for state A with or without agitation after one hour. No difference
is perceptible between state A and state B, with the same cyclic rate. This implies that a time basis
of 1.5 s is sufficient for the renewal of the boundary layer. It appears that a high agitation (state A)
is better that a low agitation (state C).
 423
Modelling
From the experimental analysis, two factors appear to be significant;
-instantaneous power, which is characterised by the Reynolds number of the fruit: Re=(D.Va)/n
where D: apple equivalent diameter, Va apple piece velocity, n kinematic viscosity.
-the frequency of agitation and duration, which is characterised by the cyclic rate of the pump;
RC.
The variable R=Re.RC is introduced as an average agitation power. R is available if Re<115 which
is the limiting Reynolds number because if higher number are allowed the mass Biot number is
high and internal mass transfers become limiting factors. The R factor is a decision variable which
is introduced to correct the external mass transfer coefficient to take into account the influence of
agitation;
b2= ~6..Ci . Exp(T/25) + ~(20.R) if Re<Re max
b2= C-J6..Ci . Exp(T/25) + \I(2300.RC) if Re>Re max
After identification, the maximum prediction error is 5% of the predicted value. Figure 1 illustrates
a validation which is obtained for a trial with variable conditions.

CONCLUSION
Agitation appears to be a significant decision variable for the control of the DIS process. Even if
the effect is less than temperature and concentration of the solution, then use of agitation is good
during the first hour of the process. A dynamic model is developed and validated. This is a good
tool to build process control strategies in further studies.

25
,

t(h)

Figure 1: Comparison between experimental values and model prediction of dewatering of apples.

REFERENCES
1. Raoult-wack, A.L., Guilbert, S., Rios, G." Drying Techno!., 1992,589-612.
2. Ponting, J.D., Walters, G.G., Forrey, R.R., Jackson, R., Stanley, W.L., Food Technol., 1966,
20,125-128.
3. Raoult-wack, A.L., Petitdemange, F., Giroux, F., Guilbert, S., Rios, G., Lebert, A., Drying
technol., 1992,613-630.
 DETBRMINATION OF S'.l'Rl.JC'lURA PARAMRTBRS FOR 'l1IB DRIBD LAYER
OF COFFEH SOI.AJTION UNDBHOOING SUBLIMATION DDlYDRATION

JUNICHI ICHIBA, YASUYUKI SAGARA, YOSHINORI KAWAGOE, YASUHISA SE~

Department of Agricultural Engineering
Facul ty of Agriculture
The Uni versi ty of Tokyo
1-1-1 Yayoi, Bunkyo-ku, Tokyo 113, JAPAN

.ABSTBACT

The structural parameters for the dried layer of coffee solution undergoing
sublimation dehydration were determined by using both mercury porosimetry and
value of permeability. The permeability and thermal conductivity were measured
with an automatic measurement system. In the model the porous dried layer was
considered to consist of a bundle of capillary tubes, closed pores and solid
solute matrix. The structural parameters determined were found to be useful to
predict the permeability using a theoretical equation derived on the bases of
collision theory.

To estimate the freeze-drying rate, the thermal conductivi ty and permeabili ty for
the dried layer of material are indispensable transport properties to be
determined. Freeze-dried food is generally recognized to become porous body, and
the structural model for freeze-dried materials is useful for predicting
transport properties as well as designing the quality of final products. The
objectives of this study were to develop a structural model and then determine the
values of its geometrical parameters to predict the permeability of coffee
solution, applying the data obtained from both direct measurement of permeabili ty
and mercury porosimetry for the dried samples.

As shown in FIG.1 the samples of 8-37% solutions were freeze dried at various
constant temperatures ranging 6-60 "C, using radiant heating upon sample surface.
During drying process the transport properties were determined wi th a microcom-
 425
puter-based automatic measurement system(1).
Capillary structural ratio cc'(r/f.) is calculated from a measured value of
permeability by assuming the Knudsen flow for water vapor flowing through
capillary tubes. The porosi ty c c' which is dependent on concentration. is assessed
from cumulative compositions by using the specific volume of 0.626 for pure
coffee extract. and the average pore radius r is measured wi th a mercury porosi-
meter. The structural factor is defined as f.= Ct' T • where c t is the through pore
ratio defined by (cm/cc), and Tis the tortuosity factor. Since c. is obtained
from mercury porosimetry, all structural parameters can be determined, indepen-
dently.

Freeze-drying
'Temperature J;---~~--~~-l

.. Dried samples . Mercury porosimetry

·Average pore diameter

,
·Pressure . Total po re area
'Weight ·Total intrusion volume
----
Transport properties; ,
A=a P.12(LlH+S;'Cpcte)
K~/J p.1 2RTf /M.
I r, CII

' IA ~"
I"com~!ivit, tfJ'l
Permeability K
Capillary structural ratio f~-

r
Knudsen flow model r.- c'f. -.j :f. I
•
c, :Porosity (concentration)
l r :Average pore radius
K=A/'i'
A= r c..l.[,8R- t= Ct' T :Structural factor Structural parameters
fa '3 1t'M. E. t= E./ E., : Through pore ratio
T= (L/L)2:Tortuosity factor r, f'., ct,T

FIGURE 1. Experimental procedure to determine the structural parameters

RESULTS AND DISCUSSION

TABLE 1 presents the values of structural parameters for coffee solutions. To

certificate the validity of the data obtained, these values were applied to an
equation derived on the basis of collision theory for the permeability in terms
of the mean free path A, giving

(1)
where 2 -
Dk~-vr, - ~-
v= BRT, KT
- , " -_.c..=._
3 7rM" .f27rr?,.p

This equation has been found to fit the experimental results for freeze-dried
beef samples over wide pressure range. Especially, below a pressure of about 67Pa
it has been showed to provide a better fi t to Knudsen's diffusion coefficient
measured by a steady-state method. (I) Equation (1) embodies a structural
constant k'=2.6 as well as a roughness factor 01=(31t'/16),and the term (3/32k')-
(2r/ A )+0 1 are considered to be appropriate one for freeze-dried products.
FIG.2 shows the relationship between permeability and porosity at various
426
TABLE 1
Values of structural parameters for the dried layer of coffee solutions
Average Porosity Through
~oa1~~~t- COnce-nt- MercurI ry pore Strrtural Tor}uosity
SIlPle r~ml ration porOll,e rati 0 actor actor
No. (I)
c
(.um)
r
H
ec
H
em
H
et
, H
fs U
(-)
1- . "

7.8 19.3 0.94 0.29 0.307 0.540 1. 759

18.9 38.5 0.89 0.44 0.496 0.114 2.248
27.8 20.6 0.81 0.38 0.471 0.860 1. 828
4 35.2 10.1 0.75 0.51 0.672 0.610 0.908
5 35.0 11.4 0.75 0.47 0.621 0.456 0.735
6 35.0 12.2 0.76 0.48 0.638 0.469 0.735
37.4 9. 1 0.72 0.41 0.574 0.318 0.534
• et=e,,(e c " f=e s' T ' " T t= (L/L)2

dried layer temperatures. The plots for the permeabilities includes both data
obtained from direct measurement and prediction based on the values of structural
parameters. Permeability was found to depend mainly on solute concentration or
the porosity of the dried layer and then other factors such as temperature or
pressure of it. Curve was calculated from Eq.1 applying the values of drying
condi tions for sample No.6 and structural parameters obtained. In the calculation
a linearity was assumed between concentration and the porosities E. as well as
E m' respectively. As the curve demonstrates a similar porosi ty-dependent behavior
of the plots, the assumptions made in the structural model and the values
determined appeared to be essentially reasonable to predict the permeability of
coffee solution.

Key ; Temp, e Pressure, p I

! 11-?6 I 20-65

-----------r ---• -- - ----------

HO
'-
"s
-14-3
C;
1.5 Tneoretlw Curve lor Dryrng-COnmron
---------~r

0 of Sample No.6 7r=19.2, p=48.1 " ..

x •
"'"
. . . -----
---------------------------------------------~-----

....
>0

0" • K ---.-
...
-0
~----------!~--------~----------
. . •
....--------------. .
::...'"
0.5
a...
~
.

0.1 . 0.8 O. 9
Poros, ty ,Eo H
FIGURE 2. Porosity versus permeability

(1). Sagara, Y. ; Transport Properties Measurement of Food Sample Undergoing

Sublimation Dehydration. Drying '86, vol.1, pp413-421 (1986)
(2). Mellor,J.D. ;Fundamentals of Freeze-Drying.Academic Press ,London. pp94-
128 (1978)
 QUALITY ASPECTS IN DRYING OF FOOD GRANULAR MATERIALS

M.Sc. Kjartan Kramer

SINTEF Refrigeration Engineering, N-7034 Trondheim-NTH, NORWAY
Dr.ing. Ingvald Str~mmen
The Division of Refrigeration Engineering, The Norwegian Institute of technology,
N-7034 Trondheim-NTH, NORWAY
Dr .ing. Xiaomei Song
SINTEF Refrigeration Engineering, N-7034 Trondheim-NTH, NORWAY

ABSTRACT

During the last 4-5 years, heat pump dryers with fluidized bed, have been investigated at The
Norwegian Institute of Technology. A laboratory plant was constructed for drying at
athmospheric pressures. A series of drying experiments on different type of temperature
sensitive products were carried out. Drying at both temperatures below and above the
freezing point of the product, during the same drying period, was made possible by the heat
pump. Results show that the physical properties of different food products (density,
rehydration ability, color, etc.) can be regulated by using temperature programs.

INTRODUCTION

At the Norwegian Institute of Technology, Division of Refrigeration Engineering, research

concerning the design and dimensioning of heat pump dryers for different purposes has been
a prevailing activity for 15 years. During the last 4-5 years a new, patented, low temperature
drying process has been developed for drying of temperature sensitive materials of different
types. Drying with temperature programs, which are combinations of freeze drying and
warm air drying, may produce special product qualities.

EQUIPMENT AND METHODS

The laboratory dryer is based on the fluidized bed principle with the possibility of using air
428
or inert gases (N2, C02) as drying agents. A closed drying air loop with heat pump
installation ensures good temperature and humidity regulation. With such facilities,
temperature sensitive materials can be dried at freeze drying and non-freeze drying
conditions in air, or inert gases, at athmospheric pressure.
Examples of products dried in the laboratory plant are pre-starter fish feed, shrimps, fish
pieces [1], meat pieces, peas and macaroni. Important quality properties for such products are
rehydration ability, color, taste, fat oxidation, appearance and mechanical strength. The
rehydration ability (ra) isdefmed as :

mw
ra=-xlOO% (1)
mdm
ffiw: water content of the product after rehydration (kg).
m.un: weight of dry matter of the product (kg)

To study the influence of drying conditions on the color change of product, a Minolta Chroma
Meter was used. This instrument digitalizes the color of a product.
The mechanical strength of dried granules was found by measuring the size distribution of
particles after mechanical stress.

RESULTS
Temperatures of -5°C and 30°C and temperature programs of -5°C/30°C were used for the
drying experiments. Figure 1 shows the rehydration ability of raw dried cod pieces as a
function of temperature program and wetting time. The rehydration ability is controlled by
using temperature programs. An increasing length of freeze drying period improves the
rehydration ability. With long freeze drying period (>-20h), the rehydration ability is at the
same level as by vacuum freeze drying. The figure indicates that even a short freele drying
period (e.g. 2,5 h) leads to a significant improvement of the rehydration ability. Figure 2

~ __~~~__----------------x

300

shows the rehydration ability of boiled and dried meat pieces. The figures show the same
tendencies as for fish pieces, although the boiled meat pieces did not rehydrate as quick as
the raw fish pieces.
 429
The pieces dried by vacuum freeze drying showed a relatively high number of broken pieces
(fines), after mechanical stress.
Experiments also indicated that the freeze dried granules had a higher content of the white
component and a lower content of the yellow component. "Yellowish" products are normally
considered as "poor quality".
Further experiments have shown that temperature program can provide favourable drying

160 Initial water content (before drying)

i5
:a\11
c
.2 80
...
Cii
"'C
>-
..c
Q)
ex: 0----0----0------<>- ---------------~
o~------------------------------------------~
o TIme (min) 10

Figure 2 Rehydration of beef pieces

conditions considering biological activity [1]. Bakterias have been dried with near 100%
survival, and enzymes have shown no loss of activity after drying.

DISCUSSION AND CONCLUSIONS

Several types of food granular materials have been dried in the laboratory dryer. Different
quality parameters have been measured. Food granular materials could be dried to a better
quality by applying temperature programs, compared to drying at temperatures above OCC
only. Their rehydration ability and color can achieve a comparable level to the vacuum freeze
dried, and their mechanical strength is more favorable. Further; different kind of "biological
activity" can be maintained at maximum level.
Together with energy and environmental aspects, the above results indicate that low
temperature heat pump fluidized bed drying may show several advantages compared to other
drying methods.

REFERENCES
1. Kramer, K. and Str0mmen, I. : Lavtemperatur vannfjerning av fiskerirelatert rAstoff
(Low temperature water removal from fish products). SINTEF-report, Trondheim,
Norway 1993.
 CONSIDERATIONS ON THE DIFFUSMTIES
OF MOISTURE AND AROMA COMPONENTS

W.J. COUMANS*, AAJ. KETELAARS AND P.J.AM. KERKHOF

Eindhoven University of Technology, Department of Chemical Engineering
Postbox 513, 5600 MB Eindhoven, the Netherlands

ABSTRACT

Diffusivities, derived from experimental drying curves, depend both on the experimental con-
ditions and the drying model used in the evaluation of the experimental data. Diffusivities de-
rived from experimental moisture profiles do not depend significantly on process conditions
and enable an excellent prediction of drying curves at different drying conditions.
The mass transport of aroma components should be considered as a ternary diffusion process,
which means that three diffusivities are required.

DIFFUSIVITY AND SHRINKAGE

In most cases real materials are rather complex in their constitution and are therefore usually
modelled as pseudo-binary systems with the components water (w) and solid (s). The diffusi-
vity is usually defined according to Bird et.al. [1]:

ii =(ii +fl)ro -Jl)·pVro (l)

w wsw w

where the fluxes iiw and fl, (kg/m2 s) are defined with respect to a fixed frame of reference
(Eulerian space). Jl) is the diffusivity tensor (m2/s), p is the total density (kg/m3) and row is the
mass fraction of water. In order to obtain an explicit equation for the water mass flux flw a
second equation is needed to eliminate the solids mass flux fl. and this is the point where
shrinkage comes into play. The volume flux of solid is related to the volume flux of water and
this relation depends on the shrinkage behaviour. In case of non-ideal shrinkage a porous sys-
tem is being built up during drying. For the description of mass transfer in a porous material
the physical reality becomes rather complicated. Therefore equation (1) for defining the diffusi-
vity is extended to porous systems, which are considered as pseudo one-phase systems.
Transformation from Eulerian to Lagrangian space [3] occurs by means of the so-called
 431

deformation tensor, which relates the actual position i (t) of a given infinitesimal small volume
of solid material to its original position z. Actually, the deformation tensor depends nearly
fully on the shrinkage behaviour: no-shrinkage (p=O), uni-directional (p=1), bi-directional
(p=1/2) and isotropic (p=1/3) . By assuming isotropic diffusivity, uni-directional mass transfer
and the above mentioned types of shrinkage behaviour, the water mass flux and the diffusion
equation in Lagrangian space are given by:

D* =D Ps ( Jl+P (2)
PsO

aJ (P 2P

=D P : J
au a ** au
at = ctz[D with D
** (3)
s

where u is water content (kg w/kg solid) and index 0 refers to initial values. In general for the
"new" diffusivities it holds that D· :;t D". In literature one can find procedures to correct dif-
fusivities for shrinkage effects [4,5]. From the foregoing one can state that this is tricky be-
cause the "correction" can be reasoned according to two equations, viz. (2) or (3); except for
p=l the "correction" will lead to an inconsistent diffusion model.
Diffusivities of ideal and isotropic shrinking systems e.g. maltodextrin solutions [2] are derived
from drying curves of slabs. Usually for such systems the shrinkage behaviour is taken to be
uni-directional. In this shrinkage model the surface area remains constant whereas the slab
thickness (=diffusion path) decreases rather strongly. From computer simulations [3] it follows
that a non-shrinking model is a better approximation for isotropic shrinkage behaviour, be-
cause of compensating errors with respect to surface area and slab thickness.

MOISTURE DIFFUSMTIES

Drying curves
By evaluating diffusivities from drying curves a relation ship between diffusivity and moisture
content has to be put forward, for instance:

D-D a au
- 0 u or D --D0 e (4)

The diffusion equation was solved numerically for a wide range of values for parameters a and
Do. By comparing the simulated and experimental drying curve, it appears that a broad range
of values for a and Do gives excellent agreement [3]. Moreover, the two fit parameters appear
to be rather dependent on experimental drying conditions e.g. slab thickness and initial mois-
ture content. Therefore, procedures for deriving diffusivities from drying curves are question-
able and need a thourough reconsideration.

Concentration profiles
Moisture concentration profiles during drying can be measured by using e.g. NMR-imaging,
neutron absorption. These techniques provide microscopic information, whereas drying curves
432
are macroscopic by nature and reveal some averaged behaviour. For non-shrinking materials
(clays) it was found that the concentration dependence of the diffusivity shows a minimum lo-
cated at low moisture contents. This behaviour can be understood from microscopic mass
transfer processes involving liquid capillary flow, vapour diffusion and bound water diffusion.
Diffusivities from moisture profiles appear to be independent on process conditions, they even
enable an excellent prediciton of drying curves measured under many different conditions [3].

DIFFUSIVITIES OF AROMA COMPONENTS

The retention of aroma components during drying can be promoted by applying the concept of
selective diffusion [6]. In order to study this concept an experimental set up has been devel-
oped to measure water and aroma losses during the drying of slabs (e.g. maltodextrin solutions
with a trace of acetone). With respect to mass transfer of aroma components it appears that bi-
nary diffusion models give very rough estimates of aroma losses. It was established from dry-
ing experiments that aroma diffusivities, based on binary diffusion models were strongly de-
pendent on the intial drying rate. By applying the Maxwell-Stefan approach [6] to this selective
diffusion problem, it clearly follows that the aroma flux j; depends on the water flux j: ac-
cording to:

iJp
jS =-D ~+G·p jS with D =D (D ,D )
a adx aw a a aw as (5)
G =G(D ,D ,D ,equilibria)
ws aw as

where D ws, Daw and D as, are diffusivities, depending on the friction forces between the three
species: water, solid and aroma. Moreover the G function depends also both on the equilibrium
relation for water and for aroma component. In future research equation (5) will be important
for establishing diffusivities of aroma components in drying experiments.

REFERENCES

1. Bird, R.B., Stewart, W.E. and Lightfoot, E.N., Transport Phenomena, Wiley, 1963
2. Coumans, W.J., Power Law Diffusion in Drying Processes, Ph.D. Thesis, Eindhoven Uni-
versityof Technology, the Netherlands, 1987
3. Ketelaars, A.A.J., Drying Deformable Media, Ph.D. Thesis, Eindhoven University of Tech-
nology, the Netherlands, 1992
4. Gekas, V., Motarjemi, Y., Lamberg, I. and Hallstrom, B., Evaluation of diffusion coeffi-
cients in a shrinking system, Preconcentration and drying offood materials, Ed. S. Bruin,
Elsevier ScLPubl. Amsterdam, pp.317-31Q
5. Viollaz, P.E. Diffusion in systems with multidirectional shrinkage, , ibid. pp.321-322
6. Coumans, W.J., Kerkhof, P.J.A.M. and Bruin, S., Theoretical and Practical Aspects of
Aroma Retention in Spray Drying and Freeze Drying, submitted to Drying Technology.
 FLAVOUR RETENTION IN DIFFERENT METHODS OF SPRAY DRYING

ADEL SENOUSSI, BESH BHANDARI, ELISABETH DUMOULIN, ZEKI BERK*

Ecole Nationale Superieure des Indutries Alimentaires,
1 avenue des Olympiades, 91305 Massy, France.
*Technion, Haifa, Israel.

ABSTRACT
"Leaflash" and centrifugal atomizers were used to spray dry liquids containing volatile flavour
compounds. The behaviour of several aromatical model substances (vanilline, diacetyl, citral,
linalyl acetate) was observed. Support materials such as maltodextrins, gums, sugars and milk
were tested. The main drying parameters were inlet and outlet air temperatures, dry matter
content and feed composition. The experimental conditions for best aroma retention with
minimal alteration were defined. Particle size, flow ability and density were the physico-
chemical properties used to define the powders. The results can be applied in industrial
preparation of powders in food and pharmaceutical industries.

INTRODUCTION

Most food flavour are instable in the presence of air, light, moisture and high temperature.
Spray drying is a common microencapsulation technique in the food industry to produce stable
powders. It is based on the relative diffusivity of water and aroma in liquid droplets which is
variable with dry matter content and temperature (1). The effects of process conditions, feed
composition, nature of support and aroma materials to protect the aromas have been treated in
many works (2, 3, 4). In this study we will summarize our experience concerning the drying
behaviour of some aroma models into various food supports.

REAGENTS - ANALYSES

Powders were prepared by spray drying emulsion (or solutions) in which the continuous phase
was a solution of the support and the dispersed phase (or soluble constituent) was the core
material, the aroma.
Series I and II, experiments were realized by the differing of the proportion and the
nature of aroma and the nature of the support. The efficiency of the retention during processing
was defined as the ratio of aroma content in the dry product (g aroma/g TDM) to its content in
the feed solution. Fraction of aroma weakly retained on the powder surface was washed with
pentane and analyzed. Gas chromatography was used to study profiles of aroma (I). Properties
of powders were checked: moisture content and water activity; size and distribution of size;
surface studies by microscopy; loose and packed bulk densities; flowability.
434
TABLE 1
Operating conditions (TDM: total dry matter content - DE: dextrose equivalent)

SUPPORT AROMA FEED

-maltodextrin citra! + linalyl acetate emulsion

(DE variable) (ratio 80/20) 30 to 60 % TDM
(I) -malto + gum arabic variable viscosity
(ratio variable) 20 % TDM (or 25)

-maltodextrin (DE 12) solution

vanillin + diacetyl 40 to 50 % TDM
(II) -malto + sucrose (80/20) 40%TDM
0.5 + 0.5 (% of DM)
-whole, skim milk 40%TDM

Dryers
Two types of spray dryers were used:
-a laboratory-scale spray-dryer (Niro Atomizer, Copenhagen, model Minor Lab.) with a
vaned centrifugal atomizer (100 mm diameter) driven by an air turbine of speeds up to
35000 rpm.
-a Leaflash 100 spray dryer (Ste Aoustin, Rouen). Hot air with a pressure of 1.4 bar, is
given an helicoi'dal swirl, then is accelerated in a convergent duct, before impacting the
liquid for atomization as well as for drying. Introduction of cold air at the basis of the conic
part of the dryer produces sudden chilling of the powder before separation from the air.
Feed was metered into the dryers by means of a peristaltic pump. Since in the
Leaflash the liquid is simply flown through the central head, atomization of viscous liquids
(up to 600 mPa.s) is easier with this type of dryer.
In both driers, the drying air was electrically heated and controlled, up to 250°C for
Niro and 450°C for Leaflash. A cyclone air separator/powder recovery system was used. In
our experiments the walls of the drying chamber of the Leaflash were thermostated at 60°C
or 70°C.
Maximal evaporative capacity was 3.5 kg/h for Niro and 7 kg/h for Leaflash, at inlet
air temperatures of 250 and 450°C respectively. Experimental mean residence times defined
as the time elapsed between entrance of the liquid in the atomizer and the first appearance of
dry powder in the product collector was 30 s for Niro and 5 s for Leaflash.
The controlled parameters were the feed rate, the initial feed solids concentration, the
speed of the centrifugal atomizer, the pressure and flow rate of air, the inlet and outlet air
temperatures. The design of the driers is such that the outlet air temperature is controlled by
regulating the feed rate, first with water, then with the product, for a fixed inlet air
temperature and air flow rate. Inlet air temperatures ranged from 180 to 200°C for Niro and
230 to 400°C for Leaflash; outlet temperatures were adjusted between 90 and 100°C or 95
and 110°C respectively. Air flow rates were 64 kg/h for Niro and 83 kg/h for Leaflash.

RESULTS

Physical properties of powders

They are similar for all the powders from Niro and Leaflash spray dryers ; except for the
water content, which is higher for Leaflash (2.5 to 5) than for Niro (2 to 3.5 g water/ 100 g
dry matter). Water activities are between 0.08 and 0.15.
Increasing dry matter content in maltodextrin leads to decreased values of particle
densities. Viscosity and size of drops, then the size of dried particles are increased: more air
is enclosed in particles.
If we replace maltodextrin by sucrose with the same total dry matter content, particle
densities of powders with sucrose (MO/S = 80/20) are higher; the viscosity of the feed is
 435
decreased and the particles are more regular. The powder is more compact, more
hygroscopic, with a better wettability due to properties of sucrose.
Whole milk leads to powders with lower densities than skimmed milk, due to the
presence of fat. Agglomeration of particles and less regular shape are observed for whole
milk (higher packed to loose densities ratio). The flow ability is higher for skim milk, and
for Niro than for Leaflash.

Variation of aroma retention with type of support

(I) - As usual when the outlet air temperature increases, we observed some tendency to a
higher retention corresponding to a faster drying. With Leaflash we used air inlet
temperatures up to 400°C without any observed cracking or bursting of resulting powder
particles. If the outlet air temperature increased, the retention decreased and the percentage
of aroma at the surface increased due to more cracking of particles.
Emulsion size droplet (1 to 5 ~m) did not significantly affect retention with Leaflash
as already mention ned in other works. Retention increased from 72 to 84 % when the dry
matter content increased from 45 to 55 %. Gum arabic giving filmogen properties, we tried
to use a mixture of gum and maltodextrin as a carrier: the maltodextrin gives a high dry
matter content without increasing viscosity against what the gums do. The best proportion
was 2/3 (= G/MD) with a 50 % dry matter content. In all the experiments, the proportion of
flavour on the surface was inferior to 5 %. We did not observe any alteration of citral an~
linalyl acetate (5).

(II) -In our study, whatever the supports and the operating conditions, vanillin is stable
during drying with a retention between 90 and 100 %.
Retention of diacetyl was 60 to 80 %. It decreases (-5 to -15 %) when decreasing the
dry matter content in maltodextrin, and with replacement of maltodextrin by sucrose
(80/20). Retention is 60 to 80 % with milk with a tendency for better retention with whole
milk compared to skimmed milk.
The better stability of vanillin compared to diacetyl may be attributed to the larger
size of the vanillin molecule (low diffusivity in drops), and the higher volatility of diacetyl.
The replacement of the high molecular weight maltodextrin by low molecular weight
sucrose is favourable to the departure of aroma molecules with water during drying.
In whole milk the aromas are soluble in fat phase and the retention may be improved.
Better results on diacetyl retention with the Leaflash spray drying technique (X %
against Y % Niro) may be due to the higher inlet air temperature leading to faster initial
drying. No significative relation may be established between datas on aroma retention and
water content of the powder.
Partial replacement of maltodextrin by arabic gum (MD/gum = 80/20 ; 40 % total
OM) gives a value for retention of diacetyl between those obtained with maltodextrin alone
with a total dry matter content of 40 % and 50 %.

Research has been supported by French Agricultural Ministry and EEC programme FLAIR
REFERENCES

1. Rulkens W.H., Thijssen H.AC. The retention of organic volatiles in spray drying
aqueous carbohydrate solutions. J. Food Techno!., 1972,7,95-105
2. King C.J. Spray drying of food liquids and volatiles retention. in: "Preconcentration
and drying offood materials", S. Bruin ed., Elsevier, Amsterdam, 1988,147-161.
3. Risch S.J., Reineccius G.A. Flavor encapsulation, American Chemical Soc,
Washington DC, 1988.
4. Rosenberg M., Kopelman U., Talman Y. Factors affecting retention in spray drying
microencapsulation of volatile materials. J. Agr. Food Chern., 1990,38, 1288-1294.
5. Bhandari B.R., Dumoulin E.D., Richard H.MJ., Noleau I., Lebert AM., Flavor
encapsulation by spray drying: application to citral and linalyl acetate. J. Food Sci.,
1992, 57(1), 217-221.
 IMPROVEMENT OF I-MENTHOL RETENTION BY CYCLODEXTRlN DURING DRYING

TAKESHI FURUTAl, HIDEFUMI YOSHII 2 AND AKIRA YASUNISHI l

lDepartment of Biotechnology, Tottori University, Tottori 680, Japan,
~epartment of Biochemical Engineering, Toyama National College of Technology, Toyama 939, Japan.

ABSTRACT

Various kinds of liquid foods have been dried in powder form by spray drying. The key technique
in drying is to prevent the loss of volatile components in the food liquid. In this study, a molecular
encapsulation method using cyclodextrins was used to improve the retention of hydrophobic flavor
components during drying of a single droplet. Significant improvement of the I-menthol retention was
observed under various drying conditions. Numerical calculations based on the selective diffusion
hypothesis could estimate well the final retention of I-menthol under various concentrations of cyclodextrin
and maltodextrin.

INTRODUCTION

Good taste and flavor of the spray-dried powder products are the most important quality factors demanded
by consumers. Previously, the retention of a trace amount of ethanol was examined experimentally, by
using the drying process of a single droplet [I]. Furthermore, the selective diffusion theory [1,2] has
successfully been applied to estimate the amount of ethanol retained during drying. In this study, some
hydrophobic flavor substances, present in most food liquid and would easily evaporate on drying, are
encapsulated in the molecular cavity of cyclodextrin to prevent loss on drying. I-Menthol was used as the
model flavor. The effect of maltodextrin dissolved in the sample solution, which exhibited a marked
effect on the retention of hydrophilic flavors such as ethanol, was also examined using this system.

MATERIALS AND METIIODS

Materials
J3-cyclodextrin (J3-CD) was from Ensuiko Sugar Chemical Co. I-Menthol was from Wako Chemical Co.
Chloroform was from Kanto Chemical Co., Inc. Maltodextrin (MD) was from Nippon Starch Chemical
Co. Ltd. These were all of reagent grade.
 437
Drying Experiment of a Single Droplet
In distilled water, I-menthol was dissolved to make 400 ppm. The designated amount of ~-CD and MD
were added and dissolved, then the inclusion complex of I-menthol in ~-CD was formed in the solution.
The experimental equipment for drying the single droplet was the same as that used previously [I]. An
up-stream hot air of uniform temperature, humidity and velocity was realized by the assembly of a packed
bed humidifier, a computer-aided electric heater and a contraction nozzle. Precisely, 5p) of the sample
solution was measured with a micro-syringe, then a small droplet (ca. 1.6 mm dia.) was formed at the tip
of a glass filament and hung in the up-stream hot air. After drying for the designated time, the droplet
was put and dissolved into a mini vial, in which lOOp) of distilled water and 50p) of chloroform were
poured. Almost all the I-menthol in the dried droplet was extracted into the chloroform phase. The
concentration of I-menthol in the chloroform phase was measured by a gas chromatograph (Shimadzu)
with the separation of Thermon 1500 (Shimadzu). The retention of menthol was defined as the ratio of
I-menthol concentration in the droplet before and after drying.

RESULTS AND DISCUSSION

Figure 1 shows the time course of the I-menthol retention in a droplet during drying under various
concentrations of f3-CD in 10% aqueous MD solution. The temperature of the drying air was 70°C. The
retention of menthol decreased markedly for about 2-3 min. in the initial period of drying, followed by a
constant final value of retention. The increase of ~-CD concentration predominately increased the
retention at every drying time, particularly the final retention. However, even if ~-CD was added at an
excess concentration such as 5 times as much as the moles of I-menthol, 100% retention could not be

100
MD Cone. =10 %
=70 C
-..
Air Temp.
~ 80
'0 o
.r:.
c 60
at
:E
'5
c 40
0 1
•.;:3 []
c
I 20
a: ~-CDlMenthol =0 ( alar ratio)
•
0
0 2 4 6 8 10
Drying Time (min)
FIGURE 1 Effect of ~-CD Content on the Retention of I-Menthol on Drying

obtained. It was found that the retention of I-menthol was also increased with the increase in the
concentration of MD when the content of ~-CD was constant. These findings suggest that I-menthol in
the droplet is present in two states; inclusion complex and free state. These two states show an
438
equilibrium each other as:

[Me] + [P-CD] ...;...i==~.~ [Me • P-CD]

As drying proceeds, the complexed I-menthol remains unchanged, but the free I-menthol dissolved in the
solution is partly evaporated and is encapsulated by the dried film of maltodextrin. These postulations can
explain the experimental results in Figure I qualitatively. To estimate the final retention of I-menthol
theoretically, the selective diffusion theory was applied to this system, based on the following hypothesis.
a) The inclusion complex of I-menthol remains unchange in the droplet during drying.
b) Water and the free I-menthol are transferred in the droplet by only the molecular diffusion.
c) The water diffusivity obtained previously [1] can be applied.
d) The diffusivity of I-menthol can be formulated by the equation proposed by Kerkhof [2].

100

-..
~
15 80
~
c
CD
:E 60
'0
c
• MD=O%

0
"01:1 40 <> 10%
C

I
a:
A 20%

ii 20 - - Calculated 0 30%
c
u::: Air Temp" =70 C 0 50%
0
0 2 4 6 8 10
/3-CDlMenthol (molar ratio)

FIGURE 2 Estimation of the Final Retention of I-Menthol on Drying

The theoretical estimation of the final retention of I-menthol is shown by the solid lines in Figure 2, being
compared with the experimental results at 70°C of the drying air temperature. The theoretical results
show good agreement with the experimental results. Similar results were obtained at the drying air
temperature of 90°C, where the parameters used to calculate the diffusivity is the same as that at 70°C
and Arrhenius equation was applied to estimate the temperature effect on the diffusivity. The theoretical
estimations was also in good agreement with the measurements at the drying air temperature of 90°C.

REFERENCES

I. Furuta, T., Tsujimoto, S., Okazaki, M. and Toei, R., Effect of Drying on Retention of Ethanol in
Maltodextrin Solution During Drying of a Single Droplet, Drying Technology, 1984, 2, pp.311-327.
2. Kerkhof, P.J.A.M., A Quantitative study on the Effect of Process Variables on the Retention of
Volatile Trace Component in Drying, 1975, Ph. D. Thesis, Technical University Eindhoven, The
Netherlands.
 SPRAY DRYING LEAFLASH TECHNIQUE: AN INVESTIGATION ON
PULVERIZATION AND FLOW PROFILES

LAURENT CLEMENT, ELISABETH DUMOULIN.

STEPHANE LAGARDE*, CHRISTIAN BOURLIER*
Ecole Nationale Superieure des Indutries Alimentaires,
1 avenue des Olympiades, 91305 Massy, France,
*RhOne Poulenc, Centre de Recherche, 93308 Aubervilliers, France.

ABSTRACT
The atomization system Leaflash 100 is a two-fluid atomizer with a central liquid annular
layer. The principal parameters governing the mean diameter and the size distribution of the
droplets generated by this nozzle are examined, by using aqueous maltodextrin solutions and
a laser particle sizer. Images of the spray are obtained with a CDD video camera and two li~ht
sources: a stroboscope; and a pulsed ruby laser allowing an estimation of the liquid veloclty
in the first 12 mm below the nozzle. Air flow pattern in the dryer chamber are visualized ana
simulated using a numerical code FIDAP. The effect of the ratio of vertical and axial
components of the air velocity at the nozzle outlet, was investigated and discussed regarding
the recirculation zones in the chamber.

INTRODUCTION
The Leaflash spray dryer is investigated by looking at the structure of the spray and the
parameters influencing the size and the distribution of size of the generated drops near the
mjector; and by estimating the velocity and the flow pattern of the air in the dryer chamber.

MATERIALS - METHODS
In the Leaflash atomizer the liquid feed and the air are injected by means of a twin fluid
nozzle with an annular liquid sheet. In the injector the air flow is given an heliCOIdal
trajectory and a high velocity when impacting the liquid sheet, with a ratio of about 200.The
air brings the kinetic energy to tear the liquid film into drops and the heat energy to dry the
drops into powder [1, 2].
The head of the dryer was studied alone to visualize the spray and to measure the drop
size. A transparent pvC chamber is added to observe the air flow dlstribution.
440
AIR LIQUID
flow rate 20 a 100 Nm31h flow rate 4-10 1/h
pressure 1,2-1,5 bar abs (± 0,01) viscosity 20 to 170 mPa.s
distance nozzle-diaphragm (e) 1 to 3 mrr surface tension 30 to 53 mN/m
temperature 25 -300° C dry matter content 20 a 55 % poids
(electrical heating + thermocouples) (maltodextrins, water)
thickness of the film ",1 mm
..
Table 1. Operatmg condltIons

RESULTS
Images of the spray
A video camera is monitored in real time with two different light sources:
- a stroboscopic light (flashes 40 Jls) which shows a non continuous conical water jet with a
shrinking neck followed by an expansion.
- a laser (20 ns) light with two pulses distant from 50 Jls gives two successive pictures leading
to an estimation of the velOCIty of liquid particles (size<lOmm). At 10 mm far from the
injector in the dryer axis the liquid velocity becomes equal to the air velocity.
Drops size
A laser beam (diffractometer Malvern 2600) with a diameter of 8 mm crosses the spray. The
energy corning from the delimited spray volume is expressed with two criteria: a median
volurnique distribution (D(v,0.5), and a mean Sauter dIameter D(3,2). Measurements were
made at two distances from the injector outlet 2.5 and 4.5 cm (Fig. 1).

.. Injector
Radial
position of O(v; 0,5)
Lenght of
measuremen
,, i.
" '--0)
E

...
o (k) the beam volume
It)
,,
I (Ilm) (mm)
N
I
(a) 1 29,4 10
" S
t

t \ •• e 27,0 24
E
o
t
c 2
".
t •

"" t
N

__________~\~b) 3 37,0 22
. i
~+: \
t
0
I

n 4 40,0 24

~8030mm
3 0Smm
(a) 5 27,0 10
1 30,0 20
4
5 (a) 2 22,8 35
S
1 e 3 24,4 40
c
2 t 4 26,6 50
3 i
4- + - - - - - - - - F.-" 50mm 0 5 27,0 50
n
5 6 23,0 40
6 (b)
7 24,0 30
7
8 (b) 8 27,0 10

Figure 1. Exploration of spray with laser granulometer and video images (air: 150°C,
0.20 bar abs.; e 2mm; DM 20%w/w) (j: jet outlet diameter; k: neck)
 441
After screening of the parameters important for a two-fluid nozzle [3] a correlation has
been established between the size of drops and the four major parameters: velocity,
temperature and pressure of air; dry matter content of the feed.
Variations in size are checked using these models:
- larger drops correspond to higher total solid content in the feed and higher air temprature,
- at 4.5 em D(v,0.5) IS rather more homogenous, between 20 and 30~.
Concerning the air pressure P, the size of drops is proportional to 1/P2 (or 1/U2 where U is the
air velocity).
Drop size distribution
We observed a bimodal distribution between 10 and 200 !lm (I:ig. 2). The proportion of large
drops is higher with high dry matter content (55%) or wIth high inlet aIr temperature.
Increasing the air pressure produces a shift of small sizes towards smaller ones. In Leaflash
spray drying we have two simultaneous phenomena, atomization and drying. The velocity
ratio between air and liquid is high before Impact. We may assume that in the first millimeters
from the nozzle the surface of the liquid sheet is easily pealed by air leading to small drops.
And the liquid sheet is torn out in large drops, these ones leading to further secondary
atomization.

18r----------".-----..1oo 18..------------_-----,.100
16 20 % TDM 90 16 55 % TDM 90
eo 14 eo
70
6O~
50~
40ij
3IJ"#-
20
10

~ ~ E
.,
",- ~ ~ ;::
~-+++++-Ht-+ 0

Size (urn) Size (jJ.m)

Figure 2. Drying of aqueous solutions of maltodextrins. Size distribution at 2.5 cm under the
Leaflash nozzle (air: 225°C, 1.35 bar abs.; e 2mm).
Air flow pattern in the drying chamber
The study was made with air at 25°C, without liquid, varying the air pressure from 1.1 to 1.4
bar. We measured axial and tangential components Vo, Vz, of air velocity with a Pitot tube.
With this data near the injector it was possible to simulate the flow and the velocity pattern in
the chamber (code FlDAP, from Fluid Dynamics International). Results were valIdated both
by velocity measurements inside the chamber; and with the movements of wool plies placed
regularly in the dryer chamber, and observed by video camera
With this model it was possible to simulate the influence of the ratio VoN z on the air
recirculation zones. This parameter may be important to predict the possible alteration of
product during drying. Residence time distributIon of the powder in the dryer is one of the
main parameters influencing the design of such dryers. Measurements have to be extended to
the couple air-product to have a more relevant model.

REFERENCES
1. Prudhon, E, Atomiseur LEAFLASH, Informations Chimie, 1979, 189, mai, 10 pp.
2. Bhandari B.R., Dumoulin E.D., Richard H.M.I., Noleau I., Lebert A.M., Flavor
encapsulation by spray drying: application to citral and linalyl acetate. I. Food Sci.,
1992, 57(1), pp. 217-221.
3. Lefebvre, A. R., Twin-fluid atomization : factors influencing mean drop size,
Atomization and Sprays, 1992,2, pp. 101-119.
 Bl!8EABC11 OF FAB-INFRABIID DRYING ON WmTE MUSHROOM

ZHANG MIN XU NAIZIIANG

(Beijing AgrieultunI (Zhejiang AgrieuItunI
EngiDeering UDiv.,PRC) Univenity,lIaDgzhou,PBC)

In this paper,the thin-layer d:ryi:og te&t.8 OD white mU8h:room have heeD made
by the far-iDhved drying test box. The better drying period faeton (box tem-
perature, ndiatiDg distaDee aDd ClODveetive airDow iD the box) aDd material rae-
ton(thickDeaJ of material aud loadiDg ClOeffieieDt) have heeD foUlld aeeol'tliJll to
the compreheD8ive study of dryiDg rate aDd power eoDlUDlptioD. ADd the compa-
riBoD betweeD far-iDfrared aud heat aiI'fIow dryiDg hal heeD made. FiDaIIy, the
iDflueuee of plates for redueiD8 ndiatioD OD heat deD8lty aud their thiD dryiDg
equatioD have heeD tIhJewIBed.

INTBODUurION

White mUBllroom (WM) II a commOD edible fUDgwJ iD ChiDa. The dried IJIIee
of WJI has lOme Dl8I'ket iD iD&ematioDal tnde(y.Doug,l986).It is very DUtritioUS
aud delielous, TIle iDgredieDfB seepiDg from WJI eaD be used iD medieatioD for
treatiDg Uvel' dI8eue(JICAC,1981).
Far-iDfnre drying is a DeW method of fast dehydntioD OD lOme footlll,espe-
eiaIIy the vegetables with the high moisture couteDt (Zhaug JIiD,l989). Far-
iDfrared rays eaD iDUltnte iDto the iD8ide of the material, 80 that the whole
material eaD be heated very evtmly at almOBt the l&ID.e time(Z.Ou,l987). However,
there are few I'eIea1'eh wOl'b OD vegetable dryiDg iD the wOl'ld,1O the aim iD the
paper is to fiDd out the affeetiDg faeton OD dryiDg rate aDd power eommmptioD
for the pnetital produetioD.

EQUIPMENT AND JDmIOD

TIle far-iDfrared and coDvective dryiDg teBt-box is a modified venioD of

 443
Model JT766-1 dryiDg bos:.The rotatioal speed 01 Ole faD II eontrolled witll a
governor.The Kmiperature In Ole 1Ios: II adjwJW wlOl a Type WIIZK-01 Kmipen-
tu:re eontroller. The plate 01 irradiation II made 01 I blend 01 8lC and eIIy. The
proeedure 01 teN .: grading,euttIDg IIlceI 01 1-8mm, prei:reatment, weighing 109,
apeadIng Ole material evenly In Ole layer,dryIng• •. , . Ole apeeIaI note In Ole
paper, Ole values 01 tile drying futon are : ndlating ttiltanee-14e.m; 1Ios: tem.-
penture-80"C ;airDow rate-3.6m/mln. The weighJng proeeII Interval II every
hall hour,untiJ tile moisture eontent II leas tun 8%(w.b). The objeetives 01 Ole
tests are:(1)fune needed lor drying till 88"; and(9) power eoll8UJD.ption.

B1tilJL'l.'S AND ANALYSIS

1.The drying tea&8 under different drying proeeIlI Iaeton in tile bos:(ornitted)
2. The drying teN under diHerent material futon(ornitted)
3.The eomparilon test. between lar-inIrared and airIIowdrying
One and two thin alwnlnlUID plates we:re pIaeed l'tl8p8etively between Ole mate-
rial layer and tile irradiating lOuree. The resulfl 01 tea&8 are shown. The drying
tilDe II longer when tile plate (II) II UI8d while Ole tUffe:renee between one and two
plate(s) II negleetable In the analysis.The drying proee. after placement of plate
ill almost Ole SlIDe III tile airIIow one.So the drying tJm.e 01 lar-Infrared drying ill
about 1.83 tilDes III that 01 airIIow drying while tile power eoll8UJD.ption ill almost
the IllUDe from Ole eurves.

DI8CU88ION

1.Efleet 01 iIolating plate (II) on Ole irradiating level

It ill UIIUDled that 0Ie:re are n plates lor iIolating heat,and the eoUlelent 01
the blaek body 01 the plate,irradiating 101lI'ee and receiving plate are £(a),£(1),
£(9) :respectively. Aeeording to Ole irradiating tIleory'lI eonculllion, the eombind
heat denalty q (1,2) 01 no isolating plate III III lollows.
q(1,2)==4.9£(1,2). ((Tl/lOO)4-(TI/100)4) •••••• (1)
From anaIysiI on tile situation 01 plaelng n plates, tile eomblned heat deBIlity
q(a) ill III lollows.
q(a)-4.9 (1/E(1,2)+(2/E(I)-1).n) -1. ((Tl/lOO)4_(TI/lOO)4) •••••• (2)
Using Ole equipment, £(1,1)- 11 (1/E(l)+1/E(1)-1) -0.419, £(a)-o.08, while
Tl- 498K, TI- aoI. When n equals 0, 1, I, we eo obtain Ole values of q (a)
through eaeulating: q(.,O-O)-992.48 w/m-, q(a,...1)-68.74 w/m-, and q(a,n-I)-
M.68 w/m-. 80 tIleir :relation III q(.,o-o): q(., ...l): q(a,n-I)- 28.7 : 1.9 : 1.
Therefore, it III obtained that tile heat denaIty 01 plaeing tile plate (II) deereues
eompared wlth that of no plate, while 0Ie:re III IiWe differenee between one
plate and two, wbleh III in aeeord wlOi the l'eIUItII of the test..
2.The thin -layer equation
Slnee Ole loading thleknellll In the study ill below 1.3em, It belonp to the OlIn-
444
layer drJing of the qrieultunl produets are eommoD (J.P .Cheu, 1988) a& follows:
MB = A.exp(-rt), or IDMB -InA-it •••••. (3)
lIB - exp(-rt), or IDHB -rt ...... (4)
lIB - exp(-flD), or 1D(-ID1IR)-IDr+D1Dt •••••• (0)
The data from the dlffereut temperatures were UIed for the eoordiDatfB IDlIB
-t,ID(-IDHB)-lDt.The I'eIUlti are moWD.80 the BUitable equatioD II MB-eIp(-rra).

CONCLUSIONS

I.Durl:og far-iDfrared dryiDg,the best temperature ill the box II about 6O"C;the
best irndJatiDg dis&aDee iJ 14em;the mOlt suitable thic1meIIJ of material is 1-8mm;
the best loading eoeffieieDt II 4.okg.m8 •
I.H the lint eoD8ideratioD II productivity, the best eoDVeetiDg air rate iJ
1.74m/miD; if the prefered eoDlideratioD ill power colllUlllptioD, the D8tura1 COD-
veetioD II suitable.
8. The dryiDg rate of far-iDfrared is about as 1.88 times that of airflow dryiDg.
4. The heat deDBlty of plaeiDg the plate (a) deereuea obvioU8ly eompared with
that of DO plate.
o.The mOllt suitable model for tbiD-layer dryiDg II MB-exp(-flD)

NOMENCLATURE

W (t)- moi8ture CODteDt of material at time, % dry buhI; G(t)-weight of

material after dryiDg time t,g; G(g)-boue dry weight,g; E(1,2),E(a)--eoffieieDti
of blaekbody of the system aDd iBolatiDg plate; Tl, T2- a_lute temperatures of
irradiatiDg 80111'00 aDd upper layer iDBide the box,k; E(1), E(2)- eoeftieieDti of
blaekbody of irradiatiDg 80nne aDd upper layer ID8ide the box without plaeiDg
the plate(a);_Dumber of the platea of iBoIatiDg heat; q(1,2),q(a)- combiDed
heat deD8ities of DO plate aDd placiDg D platea; MB- moi8ture ratio; A,m,r-
eoeffieieutl of the dryiDg models; t-dryiDg times,miD.

I.J.p.Cheu,I988,DryiDg fruitl aDd vegetableliil,ChbIe8e FiDaDeiaI EeoDomie PreBl.

I.Middie ChiDa AgrieuIturaI College, 1981, Storage aDd proeeIIiDg of vegetable8,
AgrieuIturai PrB.
8. W .D.Peug,I984,Miero bioehemiBtry experimeutl,ShaDghal 8eieuee aDd Teelmology
PrtwJ.
4. Y.X.Doug,I988, A haDdbook of edible fuDgua ill ChiDa, ChJDa Tourism PreBI.
 CONVECTIVE DRYING OF LACTOBACILLUS PLANTARUM

L.J.M. LINDERS, G. MEERDINK, K. VAN 'T RIET

Wageningen Agricultural University
Food and Bioprocess Engineering Group, Department of Food Science,
P.O. Box 8129,6700 EV Wageningen, The Netherlands

ABSTRACT

Starter cultures of lactic acid bacteria are often used in the production of food and feed.
Convective drying is an economic method of drying large quantities of micro-organisms but
has the disadvantage of causing an important inactivation during drying. Inactivation of L.
plantarum during drying is caused by two mechanisms: thermal inactivation and inactivation
due to dehydration of which the latter is discussed in this paper. An overall drying model is
developed predicting the measured inactivation of L. plantarum during a fluidized-bed drying
process.
Separate experiments with a newly developed DNAIDNase method show that dehydration,
unlike thermal inactivation, causes damage to the cell membrane. Cell membrane disruption
caused by drying can be decreased by the addition of sugars.

INTRODUCTION

Lactic acid bacteria are frequently applied as starter cultures in the food and feed industry.
They are used in fermentation processes in for example the dairy industry, the meat industry
and as silage additive in agriculture. Methods of conserving large quantities of bacteria are
freezing and freeze drying. Because of the high transport and storage costs of frozen products
446
and the complexity and high costs of freeze-dI)'ing of large quantities more economical
methods like convective dI)'ing are required. A disadvantage of convective dI)'ing is that it
causes an important degree of inactivation during dI)'ing.
The inactivation of L. plantarum caused by dI)'ing can be described by two separate
mechanisms: thermal inactivation and inactivation due to dehydration. An overall model
containing both types of inactivation has been established [Lievense, 1991].
The physiological mechanisms for thermal and dehydration inactivation are not yet fully
understood. Cytoplasmatic membrane damage is generally considered as the main mechanism
of dehydration inactivation. Membrane damage can be measured with the DNNDNase
method. This method is based on the assumption that cells with damaged cell membranes are
permeable for the enzyme DNase whereas the intact cells are not. The amount of hydrolysed
DNA diffused into the medium will give an indication of the amount of cells with membrane
damage.
Lievense [1991] showed that the residual activity of low temperature dried cells is related to
the amount of damaged cells, determined with the DNNDNase method. Amelioration of the
resistance of bacteria to dI)'ing could be achieved by protection of the cells against cell wall
damage. In freeze-dI)'ing sugars are known to protect bacterial cells during drying. They
possibly prevent damage of the cell membrane during dehydration.

MATERIALS AND METHODS

Fluidized bed dI)'ing experiments with L.plantarum (P743, NIZO, The Netherlands) have been
carried out to show the protective effect of sucrose. Sugar was mixed with cell pellet before
starch was added to form a paste that was extruded to form granulates (0 = 1 mm, 1 = 1 cm).
The granulates were dried at 30°C to prevent thermal inactivation to occur.

RESULTS AND DISCUSSION

Following the expected protective effect of sugar on bacteria during freeze dI)'ing the effect of
sucrose on the residual activity of fluidized bed dried L.plantarum is investigated. Figure 1
shows that the addition of sucrose achieves an amelioration of the resistance to dI)'ing. The
decrease in water activity caused by the addition of sucrose only explains for a part of the
 447
amelioration. A possible mechanism is the replacement of the water around the polar
membrane phospholipids by sugar thus maintaining the integrity of the membrane.

Residual Activity [-]

~~--------------------------~
+
+* • ••
+ + :1:+* •
**
'-...
+
";;

o.~
••
~
• no sugar
+ sucro'S"e

o~~~~~~~~~~~~~
o 0.2 0.4 0.6 0.8
moisture content [kg H20/kg dry weight]

Figure 1 Residual Activity of fluidized bed dried (30°C) L. plantarum-starch

granulate with maltose, sucrose and no sugar added.

Additional experiments showed that the addition of sorbitol caused an increase in residual
activity after vacuum desiccation drying and a lower cell membrane damage as shown by the
DNAiDNase method. This indicates that addition of sugars or sugar alcohol's might protect
bacteria against dehydration by lowering the cell membrane damage.

CONCLUSIONS

The dehydration inactivation of Lactobacillus plantarun1 during drying can be explained by cell
membrane damage as shown by the DNA-DNase method. This inactivation can be decreased
by the addition of sucrose, maltose or sorbitol who probably prevent the cell membrane from
disruption.

REFERENCES

Lievense, L.e., The inactivation of Lactobacillus piantarum during drying, Ph.D.-thesis,

Agricultural University Wageningen, 1991.
 CONVECTIVE DRYING OF LACTIC ACID STARTER CULTURES
WITH HIGH SURVIVAL RATES

ULRICH KESSLER, WERNERBAUER*

Technical University Munich, Fraunhofer-Institute for Food Process Engineering and
Packaging, D-8000 Munchen, Germany, * NESTEC Ltd., Nestle Research Centre, Vers-
chez-Ies-Blanc, CH-IOOO Lausanne, Switzerland

ABSTRACT

Economic preservation methods as fluidized bed drying are not broadly used for lactic acid
starter cultures because of the high inactivation during drying. It was found that the
concentration of lactic acid during drying is the main reason for the reduction of the viable
number. An inactivation kinetics for Lactococcus diacetyiactis based on the concentration
of lactic acid was developed and transferred to a drying model in which temperature,
moisture and lactic acid concentration are considered. A good correlation between the
calculated and experimental determined survival rates could be stated.

INTRODUCTION

Economic preservation methods as fluidized bed drying are not broadly used for lactic acid
starter cultures because of the high inactivation during drying. The objective was to derive
the drying conditions for a better survival by the mean of a suitable inactivation kinetics.
The microorganism under investigation is Lactococcus diacetyiactis.

In a laboratory sized fluidized bed drier the microorganisms were sprayed onto the carrier
materials Lactose or Saccharose crystals. To find out the extent of damage during the
course of drying the microorganisms were dried convective on glass plates.
 449
RESULTS

The deactivation of microorganisms dried at low product temperatures cannot be

explained only by temperature effects considering thermal inactivation kinetics.
Considering the course of inactivation during drying a neglectible deactivation took
place in the beginning of drying. Only at low moistures the viable number decreased.
Below a certain moisture content no more deactivation could be noticed. A concentration
of the non volatile metabolic product lactic acid was presumed the reason for the
reduction in viable numbers. Therefore inactivation experiments dependent on
temperature and lactic acid concentration were carried out. The activation energy
decreased from 156 kJ/mol to 123 kJ/mol with an increasing lactic acid concentration
from 0.1 mol/l to 1 mol/l (Figure 1). This range of activation energy indicates chemical
reactions as reason for the deactivation.

~ 10- 1"'f---'3..=-"'c..:::::...._----=J--.::¥;,;;::----'-..:...-=+:------j
~
8
.~ 10-3T----II---t--~"-n-~~__-~o____l
(;)
o
Q)
>
1O-S................................--o...+-..................-+---o-...............- - -..........>IU-.........-f
0.003 0.0032 0.0034 1fT [11K] 0.0036

.<f---
. temperature T [0C]
60 50 40 30 20 10

Figure 1. Arrhenius-plot of the velocity constant dependent on the lactic acid concen-
tration for the inactivation of Le. diacetylactis

In figure 2 the relation between the transition of deactivation to stabilization at the end
of drying and water activity is shown. The dependence of the sorption isotherms on the
mixture fermentation broth and microorganisms were taken into consideration. The
transition in the range of a water activity of 0.84 and 0.88 can be explained by a reduced
water mobility. Using these inactivation kinetics to predict the viable numbers during
450

drying, a good correlation between the calculated and experimental determined survival
rates was found. Fluidized bed drying experiments showed distinct differences between
the two carrier materials depending on structural surface characteristics of the carriers.
Saccharose crystals with plane surfaces tends to agglomerate at higher spray rates. With
this carrier material survival rates of 90 % could be achieved at a product temperature of
30 °e. This high survival rate could be related to the fact that a droplet of the
microorganism suspension will spread on the carrier surface which leads to a large
drying surface and a fast drying rate. Lactose is a porous particle which is composed of
small crystals. These carrier material can be loaded with a 10 times higher spray rate
before agglomeration occurs. The survival rate was determined in the range of 10 - 20
%. The reason is that the droplets are soaked into the porous particle which causes a
reduced drying surface and drying rate. Most responsible for the low drying rate is a skin
which is formed during drying and was detected by the mean of an electronic
microscope.

Beside a very fast drying rate and a low product temperature during drying the survival
rate can be improved by reducing the lactic acid concentration. Microorganisms which
were washed twice with a isotonic NaCl-solution showed a 55 % survival compared to a
survival rate of 1,5 % of a untreated sample under the same drying conditions.

1.5 r-;=:=o=~=:c: : I~s=~=~=~:=~=~:=~C:~=:;=~ =j=~=~:r: g: : :;- - '-,rnl:rf;-tiiTl-

75%
50%
25%
~ 100 %fermentation broth
~ (elliker bouillon)
::l
~ 0.5 +----+----j----+--...,..:...,+1IIi.>1--+
E

o
o 0.2 0.4 0.6
activity of water a W[-j
0.8

Figure 2. Influence of the water activity on the inactivation

 DRYING OF ENZYMES:RELATIONSHIPS BETWEEN THERMAL
STABILITY OF ENZYMES AND WATER CONCENTRATION

SHUICHI YAMAMOTO, Yuzo KASUGA AND YUJI SANO

Department of Chemical Engineering,
Yamaguchi University, Tokiwadai, Ube 755, JAPAN

ABSTRACT
Enzyme retention mechanism during drying of a single droplet is analyzed on the basis
of a model that consists of the water diffusion equation, the heat balance equation, the
inactivation rate equation and the mass transfer equation at the droplet surface. Methods
for determining important variables in these equations are briefly reviewed. Effects of dis-
solved solids to the enzyme retention are also discussed.

INTRODUCTION
Drying of enzymes or proteins is an important unit operation both in food processing
and in the downstream processing of protein purification. Drying operations are usually
analyzed on the basis of combined heat and mass transfer models. However, when one
wishes to predict the enzyme retention during drying, inactivation rate equations must be
incorporated into the drying model as the thermal stability of enzymes is known to depend
strongly on both temperature and water concentration.
In this paper. enzyme retention mechanism during drying of a single droplet is an-
alyzed on the basis of a model that consists of the water diffusion equation, the heat
balance equation, the inactivation rate equation and the mass transfer equation at the
droplet surface. Methods for determining important variables in these equations such as
water diffusion coefficients, water adsorption isotherms and inactivation-rate constants are
briefly reviewed. Effects of dissolved solids to the enzyme retention are also discussed.

RESULTS AND DISCUSSION

Figure 1 shows a block diagram of the model for drying of a single droplet of aqueous
solutions containing an enzyme. This model was first presented by researchers in the
Netherlands[1-3]. As to the experimental investigation, we employed an apparatus shown
in Fig.2[4-6]. In addition to the weight and the temperature of the droplet, the enzyme
activity was measured as a function of the drying time[4-6]. Fig.3 shows a typical drying
and enzyme retention behavior. It is clearly seen that the enzyme inactivation occurs at
the period where the drop temperature rises from the wet- bulb temperature to the hot air
temperature and the drying rate becomes very low. This period is characterized as the
"regular regime" [7-9], where the drying rate is governed by the water diffusion inside the
drop due to very low water diffusion coefficients at low water concentrations.
452
Water diffusion coefficient: The water dif- Diffusion equation W( r, t) with Dw(W, T)
fusion coefficient Dw of many sugars and
polymers decreases very sharply with de- Surface mass transfer equation with Aw(W, T)
creasing water concentration W[1-31 The [ Heat balance equation TdCt) I
use of the regular regime isothermal dry-
ing curve is very attractive for determin- Inactivation rate equation PEC r, t) with kd(W, T)
ing such concentration dependent Dw val- ssumptions
ues [7-9j. Based on this method, we de- • Uniform shrinking due to the loss of water.
termined Dw - W relationships for various • No temperature distribution in the droplet.
sugars [4,5j. • Water movement in the droplet by molecular
diffusion.
Water sorption isotherm: The water vapor Fig.l Model of drying of a droplet[l-4). r=radial
pressure at the surface of t.he droplet is re- distance, t=time, T=temperature, PE=enzyme
lat.ed t.o the saturated water vapor pressure activity, Td=drop temperature
via. the water activit.y Aw as a function of Telescope
W. The Aw - W relationships are called ,If I=t I Thermocouple
water sorption isothems and measured for
example by a desiccat.or method. Although
various models have been proposed, the Cantilever n Droplet
Guggenheim- Anderson-de Boer equation beam ba1an~

7HO;_A;~
is claimed to be very valuable for describ- I
ing water sorption isotherms of food mate-
rials[2,5,8,lOj.
Cath,tom""
Inactivation rate constants: The inactiva- (Constant-temperature and uniform velocity field)
tion rate const.ant kd also strongly depends
Fig.2
on WI as well as the t.emperature[1-4,6,8j. paratus Schematical drawing of the experimental ap-
for drying of a single droplet[4-6).
The kd - W relationship varies from enzyme 1.0 .----~~"-T--4l>----o___,_1~--==c-----,--,
to enzyme and is also influenced by the type
of the dissolved solid. Simple met.hods for enzyme:,B-gOllactosidase

determining kd - W relationships must be

developed. --Sr-~====~==~~-'--------'I
:=!
"0
Effect of dissolved solids: Although it is VI 4 u
empirically known that enzymes are highly tillI
o

stabilized at low water concentrations, the ~3 ..,

mechanism is still not fully clarified. We ...ro
l-
II
so -
measured the enzyme retention of {3- ;t 2 ~
galactosidase dissolved in various sugar so- tillI ...'ro"
t;
lutions, and found that the retention varies -""1 Q.

from sugar to sugar[5,6j. This may be as- ~ E

cribable t.o the Aw - W relationships. The o '---_ _ _-.-L_ _ _ _--'-_ _ _ _ 0
~
II
~

o 100 200 300 2

choice of the dissolved solid is very impor- DRYING TIME (5I Cl
tant to obtain high enzyme ret.ention but
must be done by trial-error or empirical Fig.3 Drying behavior of single droplets of
knowledge at present. a sucrose solution[6).XE=enzyme retention
REFERENCES
l.Bruin,S. and Luyben,K.Ch.A.M., Advances in Drying ,eds. A.S.Mujumdar,
Hemisphere 1979. vol:1, p.155. . . .
2.Luyben,K.Ch.A.M'J..LlOu).K. and BrUIn,S., BlOtech.Bweng.,1982, 24 ,533;
Biotech. Bioeng~, hl85h~7 ,109.
3.Kerkhof,P.J.A.M. and ;,choeber,W.J.A.H., Advances in Preconcentration and
Dehydration oj Foods, ~d. A.Spicer, Ap'plied Science,1974,p.349.
4.Yamamoto,S.,A.y;awa,:v}-,Nakano,H. and Sano,Y.prying. 'tf5,
eds. R.ToeJ and-A.S.lVluj!lmdar, Hemisphere, 1905, p.3'JO.
5.Yamamoto,S. and Sano, Y'1 Proc. World Congress III oj Chem.Eng., Tokyo,
MYU Research, 1986,voLI ,}!.564.
6.YamamotQ,..~.and Sano,Y rt, Chem. En.q. Sci. ,1992, 47 ,177
7.Schoeber, vv .J.A.H. and Ihijssen, H~A.C., AIChE Symp.Ser. ,1977, 73 ,12.
8.Liou,J.K, Ph.D.thesis, Wageningen A!{ricultural University, The Netherlands 1982.
9.Coumans" W ..J.",Ph.D.thesls, Technica.l University Eindhoven, The Netherlands ,1987
lO.Yan den nerg, \...-., Engineerina and Food ,Elsevier,1984,voLI, p.311.
 PREDICTION OF KINETIC PARAMETERS FOR FOOD QUALITY CHANGES
USING EQUIVALENT TIME AT STANDARD TEMPERATURE

RYUNG-YONG eRO and YU-RYANG PYUN

Department of Food Engineering, Yonsei University, Seoul 120-749, Korea

ABSTRACT

The effectiveness of three different numerical methods has been assessed for
the estimation of kinetic parameters from literatural and experimental data.
Generalized one step procedure using equivalent time, which provides a
method for converting any variable temperature history to equivalent time at
specified reference temperature, gave the smallest sum of squares for error
and 6-15 times faster computing rate than those estimated by the
conventional one step method.

INTRODUCTION

A common procedure for analyzing and reporting kinetic data consists of

two steps : (1) regressing concentration on time at constant temperature to
determine rate constant k and (2) regression In k on reciprocal temperature
to determine Ea. Generally two step procedure results in a relatively large
standard deviation in Ea. Parameter estimation can be much improved by the
direct estimation method from raw data in one step using a nonlinear
regression[11 The purpose of this paper is to derive and validate a
generalized one step method using equivalent time at specified reference
temperature.

Equivalent time(teq )
In general, for foods, the changes of quality function Q[ACT, t)] can be expressed
as a function of quality index A.

Q[ACT,t)] = - I A
A,
dA
An =
It0
k dt = ko It
0
exp[ - Ea
RT(t)] dt (1)

Consider equivalent product which is held at constant temperature Ts for the time
I:.eq such that Q[A(Ts, I:.eq)] is identical to that which would be achieved by the
variable temperature history given above. Mathematically this can be expressed as:

Q[ACT,t)] = Q[A(T s, t",,)] = ks t"" = ko exp[ - RETs]' t"" (2)

454
Replacing the left-hand side of Eq.O) with the right-hand side of Eq.(2) :
t<Xj -
t -J Ea 1 1
oexp[-T( T(t) - Ts)] dt (3)

which provides a method for converting any variable temperature history to

time at a specified reference temperature[2].

MA TERIALS & METHODS

Data generation
Experiments for ascorbic acid and sucrose degradation were chosen to verify
the prediction method. Liquid model systems were stored under isothermal
and linearly increasing temperature conditions. Literatural data reported by
Singh[3] and Nunes et aI.[4] were also used to analyze different parameter
estimation techniques.
Parameter estimation
Three methods were tested for parameter estimation
1. Least squares linear regression(LSLR)
N 2
(> LSLRl 2:[Aobs - k t ] (4)

#i[
i= 1

(> LSLR2 In(k) - In(k o) + :+]2 (5)

2. Least squares nonlinear regression for first order reaction(n=l, LSNRl)
(>LSNRl=i~[Q(A)ObS - Q(A)cal]2 = i~[(A)ObS-Ao exp(-k o t exp( ~~a »] (6)

3. Least squares nonlinear regression using 1:eq (LSNRE)

Basically, this method is similar to LSNRI except Q (A)cal [= (A I-n -
Ao1-n)/(n-l)] is calculated by Eq.(2).

RESUL TS & DISCUSSIONS

Table 1 presents a summary of kinetic parameters and sum of squares for

error(SSE) for three methods: LSLR, LSNRI and LSNRE. Although both
LSNRI and LSNRE yield almost the same Ea and SSE, LSNRE is preferred
because of its efficiency and its outstanding performance.

TABLE 1
Comparison of different numerical methods for ascorbic acid degradation
under isothermal condition
Literatural data[3] Experimental data
LSLR LSNRI NSNRE LSLR LSNRI LSNRE
n 1 1 1 1 1 1
ko or k.(day-l) 4.73x1011 3.46xlOll ·0.024 1.8x107 3.0x106 ·4.84xlO~3
Ea(kcaVmoI) 18.08 17.94 18.<X:i 13.25 12.00 11.99
SSE 661.96 471.58 470.20 1029.5 485.4 485.49
Unknown parameter ko. Ea n. k.. Ea ko. Ea n. k•• Ea
Iteration number 91 14 104 7
• k. (at 25'C)
 455
Nunes et aI.[4] proposed two integral method, SRAM and EPM, for kinetic
parameter estimation with linearly increasing temperature programs. They
concluded that the SRAM using weight least squares nonlinear regression
(WLSNR) was the best parameter estimation procedure. Table 2 shows that
the method using t.,q with LSNR resulted in better parameter estimation than
the SRAM with WLSNR.

TABLE 2
Comparision of different methods for sucrose hydrolysis under linearly
increasing temperature condition

Standard error
Regression
method (E,JR.) '10-3 In( k) or In(ks )
E,JR. In( k) ·lcf or In(k s ) ·lcf
Second Rational Approximation Method(SRAM)[4]
LSNR 13.88 -1.684 f64.3 27.4
WLSNR 12.34 -1.674 336.9 11.4
Equivalent Point Method(EPM)[4]
LSNR 13.99 -1.699 653.4 29.3
WLSNR 12.59 -1.674 '2B6.7 9.5
Generalized-one step Method Using i.,q
LSNRE[4] 12.00 -4.135 132.3 9.3
LSNRE8 13.00 -7.015 143.8 9.5
a : Experimental data

CONCLUSIONS

The proposed procedure using t.,q provided an accurate and efficient method
for obtaining kinetic parameters(n, kg and Ea) in one step for isothermal and
nonisothermal reaction. This procedure may be useful for predicting more
accurately food quality changes and assessing the distribution system at the
wide temperature range.

REFERENCES

1. Arabshahi, A. and Lund, D.B., Considerations in calculating kinetic

parameters from experimental data. L Food Pro. Eng., 1985, 7, 239-43.
2. Rosenfeld, P. E., Shelf-life testing. In Engineering and Foods, ed. B. M.
Mckenna, Elsevier Applied Science Publishers, London, 1983, pp. 709-30.
3. Singh, R.K., Kinetics and computer simulation of storage stability in
intermediate moisture foods. Ph.D. thesis, Univ. of Wisconsin-Madison,
1983.
4. Nunes, R. V., Rhim, ]. W., and Swartzel, K. R., Kinetic parameter
evaluation with linearly increasing temperature profiles : Integral methods.,
L Food Sci., 1991, 56, 1433-37.
 PREDICfION OF QUALITY CHANGE DURING FOOD STORAGE

Dennis R. Heldman and Roy R. Chao

Food Science/Engineering Unit
University of Missouri
Columbia, MO 65211 U.S.A.

ABSTRACT

Food product quality is influenced by many factors during storage, transportation and
marketing. The objective of this research was to develop mathematical models to
describe the influence of temperature abuse on quality change in a conduction heat
transfer food. The mathematical equations used include first-order rate expressions
for change in the quality parameter, temperature dependence by the Arrhenius
equation and heat conduction in standard geometries. The results of the investigation
are presented in the form of charts with the quality change ratio presented as a
function of Fourier Number at various magnitudes of a dimensionless number
containing the activation energy constant and the initial temperature difference
between the product and the environment.

INTRODUCTION

The quality of foods reaching the consumer is a function of many factors including
the temperature during transportation and marketing. More specifically the exposure
of frozen and refrigerated foods to "abuse" temperatures causes quality losses that are
dependent on the temperature and the length of exposure. The magnitude of these
losses can be predicted from the kinetics of quality change and the heat conduction
within the product.
The objectives of this research were: (1) to investigate the appropriate
mathematical models for heat transfer and kinetics of quality change required to
predict quality change in a solid food; (2) to develop the quality change prediction
equations and charts for determination of quality change as a function of thermal
abuse; and (3) to demonstrate the use of prediction charts for evaluation of quality
loss due to the influence of thermal abuse during storage and distribution of the food.
 457
MATERIALS AND METHODS

The mathematical models used in this research are those most often encountered in
foods when describing changes in quality parameters as a function of temperature.
The change in a quality parameter (Q) was described by a first-order rate equation.
The influence of temperature on the reaction rate constant was described by the
Arrhenius equation.
The temperature influence within the product during exposure to the adverse
temperature was incorporated by using solutions to the heat conduction equations.
These solutions are expressed in terms of a temperature ratio as a function of Fourier
Number for a characteristic dimension and specific location within the product. The
resulting temperatures are dependent on the initial temperature (To) and the
environment temperature (T.), as well as the geometry of the product.
In order to generate information for quality change charts, the following
operating equation was derived:

Ln(i)- -( k.:') f e+ !(:. -:.)ldN F•

(1)

where:
D = characteristic dimension for geometry.
a = thermal diffusivity.
H = 1 - TR [(Ta -To)!fa]
TR = temperature ratio [(T. - T)/(T. - To)]
Q/Qo = quality change ratio
Qo = initial quality
kR = reaction rate constant at reference temperature (TJ
E. = activation energy constant
R = universal gas constant
NFo = Fourier Number = at/D2

RESULTS AND DISCUSSION

The operating equation (1) allows computation of the quality change ratio (Q/Qo) as
a function of Fourier Number (NFo) and for varying magnitudes of the following
dimensionless number: E.I[R(T. - To)]' The operating model was used to create
Figure 1 for quality change ratios (Q/Qo) at the center of an infinite cylinder. For
the conditions presented, the quality retention decreases with increasing Fourier
Number. At higher values of E.I[R(T. - To)] - 1200, the quality retention decreases
slowly for Fourier Numbers up to 2.0. For E./[R(T. - To)] = 300, the quality
retention decreases rapidly before the Fourier Number reaches 1.1. The mass
average quality retention as a function of Fourier Number is presented in Figure 2.
As is evident, the change in quality retention decreases in the same pattern as in
Figure 1. As expected, the mass average quality retention decrease more rapidly
than at the center of the product geometry.
458

I~~~.~~-~~~---------------
::.::
......, .., ............... ....................
"'..
''',
" '" ................. ..

'. "
"
.................
..................
'''\
\ " .•....
'\ \ ........
\ , \
\
\

\\ \\
Eo/(R(T.-To » B
.---------,
Eo/(R(T.-T.» ~
\
\
\\
1,200
gOO -
•••• 1.200
gOO " \
600 --600 \ \
300 _._. 300 , \
\CR"J)2/a; \CR" J)2/a - \
2.02" 10-' 2.02 x 10-' \
\ \
10-1 ~-='=--::'-:-'""-:,":_~_~..I......:-~~_-,-:_-,-........J 10- 1 .7--='=--::'-:---:'-:---'--...,..............,.......-'-....l............
0.0 0.2 0.4 0.6 O.B 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 O.B 1.0 1.2 1.4 1.6
Fourier number, Hr. Fourier number, Hr.
Figure 1. Quality Retention during Transient Heat Figure 2. Quality Retention during Transient Heat
Conduction. Infinite Cylinder - Center Conduction. Infinite Cylinder - Mass Average Quality
Retention.

The quality retention charts (Figures 1 and 2) can be used to determine product
quality loss associated with thermal abuse. From knowledge of the activation energy
constant, gas constant, and a reference reaction rate constant, the quality retention
can be determined for a given thermal abuse (Ta - To) and exposure time as
expressed in the Fourier Number. The results for thermal abuse of a frozen food are
the same as those presented by Scott, et a1. (1986). Similar results may be predicted
for other thermal abuse conditions.

CONCLUSIONS

1. Quality retention has been computed from an expression incorporating the

integration of product temperature as a function of time.
2. Charts presenting quality retention as a function of Fourier Number and various
values of a dimensionless number containing the activation energy constant, gas
constant, and temperature difference (Ta - To)'

REFERENCE

1. Scott, E.P., J.P. Steffe and D.R. Heldman. Frozen food quality improvements
through storage temperature optimization. In Changing Food Technology 2. ed.
by Manfred Kroger and Allan Freed. Technomic Pub. Co, Inc. Lancaster, PA,
1989, pp. 189-208.
 CORRELATIONS BE1WEEN PHYSICOCHEMICAL PROPERTIES AND
KINETIC PARAMETERS OF DIFFERENT CULTIVARS OF RICE

YUNG-HO CHANG
Department of Food and Nutrition, Providence University
Shalu 43301, Taiwan, ROC.

ABSTRACT
Twelve varieties of rice were cooked hydrothermally. The water uptake of rice and
the extent of starch gelatinization were monitored. A mathematical model was used
to calculate the absorption rate at which water was diffused into rice kernels under the
influence of simultaneously occurring gelatinization reaction. Results indicated the
sensitivity of diffusion process to temperature change was similar for the 12 varieties,
however, discrepancy was found in the sensitivity of gelatinization reaction to temper-
ature change. The differences among samples were mainly from the differences in
Do(L) (frequency factor at low temperature range) and Ko(H) (frequency factor at
high temperature range).

INTRODUCTION
This report presents the result of a study on the reaction parameters of water uptake
and gelatinization occurring during hydrothermal-processing of rice, and the relations
between those parameters and the physicochemical properties of rice.

MATERIALS AND METHODS

Rice samples
Milled rices were obtained from Taichung District Agricultural Improvement Station
(TDAIS) , Changhua, Taiwan. All varieties were first season's crop of 1990. Chemical
compositions were determined according to the AACC methods. Pasting behaviors of
rice flour suspensions were determined by a Brabender Viscoamylograph.
Cooking experiments and simulation
Rice kernels were cooked under an excess amount of water. Temperature of cooking
solution was recorded throughout the cooking period, and the samples for water
uptake and gelatinization determinations were collected from the batch at certain
time intervals. The amount of water absorbed by rice at a given cooking time was
calculated as the difference between the moisture content of the sample at that time
and that of the original uncooked sample. The change of enthalpy (Ll.H) of gelatiniza-
460
tion for the raw and cooked rice flours were determined by using of a differential
scanning calorimeter (DSC 92, Setaram, France). The extent of ungelatinization part
of cooked rice was then calculated as the difference of enthalpy change (between
cooked and uncooked samples) divided by the enthalpy change of uncooked sample.
Both the water uptake and starch gelatinizatIOn data were analyzed usmg a
mathematical model developed based on the principles of diffusion and chymical
reaction kinetics (1).

RESULTS AND DISCUSSION

Simulation of kinetic parameters

Figure 1 shows the typical water uptake and starch gelatinization curves which were
the results of simulation based on the estimated parameters (Table 1). The calculated
degree of gelatinization were then compared to the experimentally measured ones.
The high values (2:.0.96) of correlation coefficient (r) indicated that the quantitative
model was capable of predicting the extent of starch gelatinization during hydrother-
mal processing of rice.
o

20 40 60 80 100
TIme (min)
Figure 1. Typical curves of temperature history, water uptake and starch
gelatinization of milled rice under hydrothermal processing.

Rate parameters and physicochemical properties

The calculated value of activation energy for ~elatinization reaction in this study is
within the ran~e as indicated by other investIgator (2). The simulated activation
energy of gelatInization at the high temperature range (E{l K(H», and the activation
energy of diffusion at the low temperature range (E..a D(L») 'snowed less variation than
their correspond.ing values, D (L) and K (H). l'urth~rmC?re, the Ea K(H) .v~lues
a
showed less conSIstent than theDE D(L) va~ues. These ImplIed that the'sensltlVlty of
diffusion to temperature change was similar for the 12 varieties of rice. On the other
hand, discrepancy was found in the sensitivity of gelatinization to temperature change.
Correlation analysis indicated that the Do(L) value was signifIcantly correlated
with ash content, volume value of rice -grain (p < 0.05), amylose content, peak viscosity
(P), setback (SB) values «(><0.01), weight, hot paste viscosity (H)-and cold paste vis-
cosity (C) values (p<0.001). However, the Ko(H) value was barely significantly corre-
lated (p < 0.1) with ash content and SB value. The results imply that the extent of
diffusion of water can be predicted from the physicochemical properties of the rice
grain once the relations between properties and rate parameters are established.
However, a similar application on the prediction of extent of gelatinization needs
further investigations.
 461
TABLE 1
Rate parameters obtained by the simplex pattern search scheme
for rice cooked under hydrothermal conditions

Rice Temperature Do
ranqe
(OC) (m/hr) (cal/q-mol) (l/hr) (cal/q-mol)

TCS17 < 85.5 ab 0.6538E-3 i 3847 a 0.4786E23 35230 b

> 85.5 0.8748E04 15535 0.1066E08 e 9567 a

THu19 < 86.0 b 0.7030E-3 h 3849 a 0.1913E25 d 37885 b

> 86.0 0.8435E04 15475 0.1154E08 9599 a

KSS7 < 84.5 a 0.1000E-2 d 3872 a 0.4076E23 34970 b

> 84.5 0.1493E05 15605 0.1057E08 e 9494

TCS1 < 93.9 f 0.8144E-3 f 3886 a 0.2347E26 39985

> 93.9 0.9075E04 15635 0.8297E07 q 9121 c

TS1 < 86.3 b 0.7279E-3 q 3832 a 0.7626E25 38915 b

> 86.3 0.9335E04 15515 0.1077E08 e 9571 a

TCS10 < 89.0 d 0.8144E-3 f 3849 a 0.1650E25 b 37580 b

> 89.0 0.6914E04 15090 0.1526E08 9474

KS142 < 90.7 e 0.9990E-3 d 3818 a 0.2419E27 d 40645 b

> 90.7 0.8020E04 15240 0.1150E08 9543 a

THu67 < 84.7 a 0.9967E-3 d 3849 a 0.1091E25 d 37255 b

> 84.7 0.1336E05 15505 0.1196E08 9494

TC189 < 86.0 b 0.1027E-2 b 3880 a 0.1403E24 35730

> 86.0 0.1537E05 15660 0.1775E08 a 9703 a

THu70 < 87.6 c 0.1021E-2 c 3882 a 0.5455E27 41880 b

> 87.6 0.2726E05 16055 0.1094E08 e 9446

TCSW1 < 85.8 b 0.9571E-3 e 3829 a 0.2219E28 f 42935 b

> 85.5 0.1483E05 15630 0.1005E08 9536 a

TCW70 < 85.9 b 0.1096E-2 a 3834 a 0.1894E29 44335 b

> 85.9 0.1141E05 15260 0.1295E08 c 9581 a

1 For each column, means with the same letter are not siqnifi-
cantly different at 5% level.

REFERENCES

1. Chang, Y. H., A kinetic studr on the lime-heat treatment of corn for mas a
production.. Ph.D. dissertatIOn, Department of Food Technology, Iowa State
University, Ames, IA., 1987, 102 pages.

2. Suzuki, K., Aki, M., Kubota, K., and Hosaka, H., Studies on the cooking rate
equations of rice. J. Food Sci .. 1977,42, 1545-1548.
 DEVELOPMENT AND VALIDATION OF PREDICTION MODELS FOR TEXTURE
CHANGE UNDER VARYING TEMPERATURE CONDITIONS

BERT E. VERLINDEN and JOSSE DE BAERDEMAEKER

Katholieke Universiteit Leuven, Department of Agricultural Engineering
Kardinaal Mercierlaan 92, 3001 Leuven Belgium

ABSTRACT

A first order kinetic model for texture change during cooking of potatoes, developed for constant cooking
temperature conditions was validated for time-varying temperature situations. The model was expressed as
a differential equation. The varying temperature conditions were incorporated in the numerical solutions. The
calculated predictions of the texture changes during cooking of potatoes compared favourably with the
experimentally assessed texture values.

INTRODUCTION

Models for texture degradation of vegetables during cooking are available in literature [1],[2],[3],[4]. They
mostly are first order kinetic equations in which the temperature dependence of the rate constant can be
described by the Arrhenius law. These models are derived from experiments at constant temperature
conditions. They can be expressed as differential equations with the possibility to use them for process
calculations with varying temperature conditions. However, no validation of the models for these varying
temperature conditions have been reported. In this paper a texture model for potatoes during cooking
derived for constant temperature conditions is validated for real life time-varying cooking temperature
conditions.

MATERIALS AND METHODS

Bintje potatoes stored at 4°C were used. The night (12 hrs.) before each experiment, the potatoes were taken
out of storage and placed at room temperature (20°C). The potatoes were washed and cut into cylindrical
samples with a diameter of 20 mm and a height of 5 mm.
A water circulator with heating as well as cooling capabilities was interfaced with a computer to control
the temperature-time profIles of the cooking water. Distilled water was used to avoid inference of ions with
the texture kinetics. The samples were submitted to various temperature-time profIles. At several time points
in a temperature-time profIle the texture of the potato sample was measured using a universal testing
machine. The rupture force in a uniaxial compression test was used as a measure of texture.

DEVELOPMENT OF THE TEXTURE MODEL

The texture model used is a first order degradation kinetics.

 463
dF (1)
-k.F with F(t=O) = Fo
dt

In which F stands for the texture : the rupture force expressed in newtons, t is the cooking time. The
texture rate constant k is temperature dependent and can be modelled by the Arrhenius law.
k k E" 1 1 (2)
= rereXP(-[/r-T»
ref

In which T stands for the cooking temperature in Kelvin, kref is the reference texture rate constant at the
reference temperature TTlI' Eo is the activation energy and R the universal gas constant.
In order to find the values for the model parameters kref and Eo' experiments were carried out at
different constant cooking temperature conditions between 50 and lODGe. At several time points of the
cooking process the texture of the potato samples was measured. The data were fitted to the analytical
solution of Equation (1) and (2) using the non-linear regression Marquardt algorithm [5). The values for the
model parameters can be found in TABLE 1.

TABLE 1
Texture model parameters obtained from the experiments under constant
temperature conditions

estimate 95% confidence interval

Eo (J/mole) 124000 145000 103000

0.375 0.325 0.425
kref (l/min)

373

VALIDATION OF THE TEXTURE MODEL FOR VARYING TEMPERATURE

CONDITIONS

Four experiments with different cooking temperature-time profIles were carried out to validate the model
for time-varying temperature conditions. First the texture-time profIles for the corresponding temperature-
time profIles were predicted. The model, consisting of Equations (1) and (2) with the parameter values in
TABLE 1 was numerically integrated using a Runge-Kutta 4th order algorithm. Potato samples were

1000 110 1000 110

100 100
J.
90 ""' r-...
z 90
""'
Z 80 '£..!- ' '-"' 80 '""'
£..!- '
'" '"
'-' '" .L

'" '" 10 ... '"

IlJ
...u0
I])
...u 10 I])
...
-'" '"
IlJ

'" 60 ~ ;::l
.B 60 ;::l
:
... 100
'" '" 50 ... ...
I]) 100
'" '" ~
...
50 I])
'" '"
/'

'" '"
I])
;::l

a
;::l
40 8 P-
o.. 40 P-
...
;::l
/' '"
/'
30 B ...
;::l /' '" 30 B 8
20 20
10 10
10 0 10 0
0 5 10 15 20 25 30 35 0 10 20 30 40
cooking time (min) cooking time (min)

Figure 1. a,b : Predicted ( - ) and measured (symbols) texture-time profIle under varying temperature
conditions (- - -)
464
1000 100 1000 100
-, ......
/ ,, 90
,-..,
/
,, 10 ,-.., 90
z
'-'
,,
,, U
70 '0 - '
Z
'-' G'
- 10 0
III til '-'
...
c,)
60 III ...
c,) til

.s so .......3'" .stil 3
til
3
100
.0 III c.. E
...
70
..
""
til
c.. 30 8 c.. 60 c..
8
...
:s
20 ....
til
...
:s
E
SO
10
10 0 100 4O
0 10 20 30 4O SO 0 10 20 30 40 SO 60
cooking time (min) cooking time (min)

Figure 1. c,d : Predicted ( - ) and measured (symbols) texture-time profile under varying temperature
conditions (- - -)

cooked according to the temperature-time profIles used in the prediction calculations. At several time points
in the cooking process, samples were taken and the texture was measured. These measurements were then
compared with the predicted texture-time profIle. The results can be seen in Figure 1.
The texture measurements at each time point are given as the averages of the measurements (the circles)
with their 95% confidence intervals. The predicted texture-time profIles are the full lines and the
temperature-time proftles are the dashed lines. One can see that in most of the cases the confidence interVal
encloses the predicted texture-time profIle. It was concluded that for potatoes the texture model developed
under constant temperature conditions is able to predict the texture changes occurring under the varying
temperature conditions. It can be assumed that this will stand also for other vegetables.

CONCLUSIONS

A texture model developed under constant temperature conditions for cooking of potatoes can be used to
calculate texture changes under time varying temperature regimes. This model facilitates the incorporation
of existing texture models into simulation and optimization algorithms for food processing.

ACKNOWLEDGMENTS

This study has been performed as part of FLAIR (Food linked Agro-Industrial Research) project AGRF-
CT91-0047 (DTEE) with additional support from Flemish IWT and the FKFO project 2.0095.92.

REFERENCES

1. Harada, T., H. Tirtohusodo and K. Paulus, Influence of temperature and time on cooking kinetics of
potatoes. J. Food Sci., 1985, 50, pp.459-462,472.
2. Huang, Y.T. and M.C. Bourne, Kinetics of thermal softening of vegetables. J. Texture Studies, 1983, 14,
pp.I-9.
3. Kozempel, M.F., Modelling the kinetics of cooking and precooking potatoes. J. Food Sci., 1988, 53,
pp.753-755.
4. Verlinden, B.E., K. De Wit, L. Nys and J. De Baerdemaeker, The kinetics of texture development of
potatoes in precooked chilled foods during processing and subsequent storage and reheating. Trends in
Agricultural Engineering, University of Agriculture, Prague, 1992, pp.622-627.
5. SAS/STAT, Users's Guide. SAS Institute Inc., vol.1, Carry, NC, USA, 1989.
 THE EFFECT OF COOK-CHILL PROCESSING ON THE TEXTURE DEVELOPMENT OF
DIFFERENT 1YPES OF VEGETABLES AND FRUITS

BERT E. VERLINDEN, XAVIER K. VANDEWALLE AND JOSSE DE BAERDEMAEKER

Katholieke Universiteit Leuven, Department of Agricultural Engineering
Kardinaal Mercierlaan 92, 3001 Leuven, Belgium

ABSTRACT

The interruption of the cooking process with a refrigerated storage period was studied for its effects on the
texture kinetics of potatoes, green beans, carrots and apples. For potatoes a statistically significant effect on
the texture was found. The inherently high variability of biological material makes that it is very difficult to
use these effects in practical applications. In view of process design for cook-chill foods it is not necessary
to take an interruption effect into account in texture models.

INTRODUCTION

Precooked chilled foods that only need reheating before consumption respond to a growing need for
convenience and time saving in cooking. In view of process design the effect of an interruption, for a cold
storage period, in the cooking process on the texture development may be an important problem. Very little
references were found that document the effect of such process steps on texture. To have more information
for use in process design, experiments were set up in which various vegetables and fruits were cooked at
several constant temperatures in interrupted as well as non-interrupted cooking processes. It was investigated
whether texture models for an interrupted process should be different from those used for conventional
cooking processes [1],[2].

MATERIALS AND METHODS

Materials were purchased at a local vegetable shop and stored in a refrigerator until the start of the test. The
samples of the different vegetables were submitted to cooking processes at several constant temperatures
ranging from 50 to 99°C for several cooking times ranging from 0 to 15 minutes for apples to a range of 0
to 90 minutes for green beans. All the thermal treatments are shown in Table 1. The cooking process was
interrupted at two different time points of the cooking process. Apples after 3 and 6 minutes, potatoes and
carrots after 4 and 8 minutes of cooking and green beans after 6 and 12 minutes. Samples were packed in
aluminum foil and stored overnight (18 h) in a refrigerator at 4°C. The cooking was done in distilled water
to avoid inference of ions with the texture kinetics.
For the potatoes, carrots, and apples the texture was measured with a universal testing machine as the
rupture force in an uniaxial compression test of cylindrical samples with a height of 5 mm and a diameter
of 20 mm. The deformation speed was 50 mm/min. The green beans were tested in back extrusion and
maximum force readings were used as a measure of texture. The stroke of the back extrusion cell was 65
466
mm with a 10 mm bottom clearance. It had an annulus gap of 4 mm. Samples of 50 g fresh weight were back
extruded with a speed of 400 mm/min.

TABLE 1
Cooking treatments used in the experiments

Potatoes Carrots Green Beans Apples

Temp. 50 61 73 85 99 61 73 85 99 62 83 90 99 61 73 85 99
eC)
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Total 5 5 5 4 2 5 5 4 2 5 5 4 3 2 2 1 1
Cook- 10 10 10 8 4 10 10 9 4 10 10 9 7 4 3 2 2
ing 15 15 15 12 6 15 15 15 7 30 30 20 13 8 6 4 4
Time 20 20 20 16 8 22 22 22 14 60 60 40 20 14 10 7 7
(min) 25 25 25 20 10 30 30 30 25 90 90 60 29 20 15 10 10
30 30 30 24 12

RESULTS

An analysis of variance [3] was done to see if the interruption treatment had any effect on the texture
measurements. For potatoes a significant fIrming effect due to the interruption treatment on a 95%
confIdence level was found. This is in agreement with previous experiments [4]. For all the other vegetables
and fruit tested no significant effect was found. Some results were shown in Figures 1 to 4. First order
kinetics is probably not applicable to the data range as shown. The other cooking treatments (see Table 1)
that are not shown give similar results. It has to be noted that the variability of the measurements is high.
This is inherently connected with biological materials. Because of this inherently high variability even the
statistically significant fIrming effect would be difficult to take into account in process design. The effect
found in potatoes is probably due to the high starch content and associated processes of this product. The
green beans appeared to be toughening more in some treatments, but the measurement method was unable
to detect this texture change.

1000 700
900
g 800
700
g 600~ 0
Q
0 0
~ 600 ~500 0 0 v 0
oS 500 oS 8 S

I 400
300
200
100
0
-
-

[)
6 0
i~ 400
300
-
0
- 0
Q c

c 200
0

0 6 10 16 20 26 30 0 6 10 16 20 26 30
total cooking time (min) total cooking time (min)

Figure 1. : Rupture force plotted against cooking
time for carrots cooked at WC. Squares denote a
conventional cooking process, diamonds denote an
interrupted cooking process after 4 minutes and
circles a cooking process interrupted after 8
minutes. Horizontal lines limit 95% confIdence
intervals for the means.
Figure 2. : Rupture force plotted against cooking
time for potatoes cooked at 73°C. Squares denote
a conventional cooking process, diamonds denote
an interrupted cooking process after 4 minutes and
circles a cooking process interrupted after 8
minutes. Horizontal lines limit 95% confIdence
intervals for the means.
 467

----------, --------------

g:f
4000
g
--

1
~ 3000 -
~50 oS _ =
I ,;; - 0 o o
oS 40

i 30
i:!20
~ -00 '"
00 Cl
~ 1000 o
10
o o ~~~~~~~~~~~~~~
o 5 10 15 20 o 10 20 ~ 40 50 00 W ~ 00
total cooking time (min) total cooking time (min)

Figure 3. : Rupture force plotted against cooking
time for apples cooked at 85°C. Squares denote a
conventional cooking process, diamonds denote an
interrupted cooking process after 3 minutes and
circles a cooking process interrupted after 6
minutes. Horizontal lines limit 95% confidence
intervals for the means.
Figure 4. : Rupture force plotted against cooking
time for green beans cooked at 62°C. Squares
denote a conventional cooking process, diamonds
denote an interrupted cooking process after 6
minutes and circles a cooking process interrupted
after 12 minutes. Horizontal lines limit 95%
confidence intervals for the means.

CONCLUSIONS

For the process design of cook-chill foods and especially during process simulation by means of computer
software, one needs kinetic texture models to predict the texture development as a quality parameter for the
process. It seems not necessary to consider special texture models for interrupted cooking processes. The
texture models for conventional cooking processes can be used under time varying process conditions even
for such extremes as an alternating heating and cooling.

ACKNOWLEDGMENTS

This study has been performed as part of EEC - FlAIR (Food linked Agro-Industrial Research) project
AGRF-CT91-oo47 (DTEE) 'Development of Computer-Aided Process Design Procedures to improve Quality
and Safety of Products with a Limited Shelf Life', with additional support from Flemish IWT and the FKFO
project 2.0095.92.

REFERENCES

1. Verlinden, B.E. and J. De Baerdemaeker, Development and validation of prediction models for texture
change under varying temperature conditions. Sixth International Congress on Engineering and Foods,
Chiba, Japan, May 23-27,1993.
2. Kozempel, M.F., Modelling the kinetics of cooking and precooking potatoes. J. Food Sci., 1988, 53,
pp.753-755.
3. SAS/STAT, Users's Guide. SAS Institute Inc., vol.1, Carry, NC, USA, 1989.
4. Verlinden, B.E., K. De Wit, L. Nys and J. De Baerdemaeker, The kinetics of texture development of
potatoes in precooked chilled foods during processing and subsequent storage and reheating. Trends in
Agricultural Engineering, University of Agriculture, Prague, 1992, pp.622-627.
THE EFFECT OF WATER ACTIVITY AND TEMPERATURE ON THE RETROGRADATION RATE
OF GELATINIZED WHEAT FLOUR

C. RHEE AND S.W. LEE

Department of Food Technology, College of Natural Resources
Korea University, 1 Anamdong, Sungbuk-ku, Seoul 136-701, Korea

ABSTRACT

The effect of a y and temperature on retrogradation of gelatinized wheat

flour was investigated at 4°C, 20°C and 30°C. The degree of
retrogradation was determined by enzymic digestibi I i ty method. The
resul ts showed that retrogradation of gelatinized wheat flour was
dependent upon a y and temperature. The effect of a y on retrogradation
was much greater at 4°C than at 30°C. Kinetics of retrogradation was
evaluated by Avrami equation and differential scanning calorimetry.

INTRODUCTION

The availability of water plays an important role in retrogradation,

since the changes of starch from the amorphous state to crystal I ine
form are closely related with moisture content in foods. The reaction
rate is slow at very high and very low moisture contents(I). In
addition, reduction of moisture content of the starch fraction using
hydrophilic materials like humectants competing for water might reduce
the retrogradation of intermediate moisture foods. In this work, we
have investigated the effect of ay and temperature on retrogradation
rate of gelatinized wheat flour.

MATERIALS AND METHODS

In order to prepare dried gelatinized wheat flour samples, a 2%

suspension of commercial wheat flour was autoclaved at 120°C for 1
hour and vacuum-dried at 40°C. Thereafter, dried gelatinized sample
was milled and passed through an 80 mesh sieve. the gelatinized wheat
flour was stored over di fferent ay of 0.52, 0.75, 0.83, 0.88, and
0.93, respectively. The degree of retrogradation was measured by
means of enzymic disgestibility and differential scanning calorimetry.
 469
RESULTS AND DISCUSSION
Figure 1 shows changes of the degree of gelatinization of samples at
different a w against storage time. These results show that at a
higher a w range retrogradation rate is assumed to be directly related
to a w, which is basically in agreement wi th the results of study on
the effect of moisture content of ageing wheat starch gels{I). This
result has also implications that the reduction of a w by adding
hydrophi lie materials like humectants may result in changing the
retrogradation rate.

50r---------------------------------~
45
40 0 - 0 AwO.52 /
35 . - . Aw 0.75 " .-.
6 - 6 Aw 0.83 1It./
30 1It.-1It. Aw 0.88 . //
0 - 0 Aw 0.93 ~/ /

~./""""""::0.
25
20
________IIt./ ~

~::::--- 6~ i=---e:::.---
5

3
.---=
6------e~
O~~c=~--~--~~~--~~--~--~
o 6 9 12 15 18 21 24 27
STORAGE TIME IN DAYS

Figure 1. The effect of a w on retrogradation at 4°C.

The resul ts of a w tests on retrogradation was subjected to an

Avrami analysis(Table 1). It was found that all data from this
experiment fi t ted well the Avrami model. The rate constants for
gelatinized wheat flour at 4°C was approximately three times larger
than those at 20°C and 30°C. This indicates that the retrogradation
rate for gelatinized wheat flour is more closely related wi th the
storage temperature than a w•

The variation of enthalpy(t1H), onset temperature(To), peak

temperature(Tp) and peak height index(PHI) were tabulated in Table 2.
Ouring the storage period of 24 days, there was a gradual increase in
H, and only minor changes in To and Tp were observed. The resul ts
suggest that storage temperature plays a decisive role in reducing
retrogradation of gelatinized wheat flour(2). At the low level of
moisture content, the retrogradation rate decreased sharply with
increasing storage temperature up to 20°C and gradually leveled off
470
TABLE 1
Rate constants(day-l) depending upon temperature and aw.

Aw 0.52 O. 75 0.83 0.88 0.93

Temperature

4°C 0.0023 0.0026 0.0075 0.0135 0.0153

20°C 0.0033 0.0027 0.0037 0.0049 0.0061
30°C 0.0011 0.0023 0.0037 0.0052 0.0057

TABLE 2
DSC thermogram value of gelatinized wheat flour.
Tempera ture( °c ) Aw To(OC) Tp(OC) 4 H(callg) PHI
4 0.83 104.0 117.7 12. 74 1. 66
0.88 102.6 107.3 15.41 3.28
0.93 106.2 117.6 24. 72 2.16

20 0.83 96.0 101. 6 2.37 0.41

0.93 109.5 118.8 5.47 0.59
----------
30 0.83 96.5 100.0 1. 47 0.47
0.93 89.8 101. 6 4.92 0.41

wi th increasing temperature to 30°C. Retrogradation of gelatinized

wheat flour can occur at aw range of 0.5 - 1.0. However, the reaction
rate in retrogradation was more signi ficantly affected by storage
temperature than a".

REFERENCES

1. Longton, J. and LeGrys, G.A., Differential Scanning Calorimetry

Studies on the Crystall ini ty of Ageing Wheat Starch Gels.
Starch/Starke. 1981. 33, 410-414
2. Rhonda, G., McIver, D.W.E., Axford, K.H., Colwell, K.H. and Elton,
E.A., Kinetic Study of the Retrogradation of Gelatinized Starch, J.
Sci. Fd. Agric. 1968. 19, 560-563
 DIFFUSION AND REACTION IN WHEAT GRAINS

A.G.F. STAPLEY, M.P. HOLLEWAND, L.F. GLADDEN and P.I. FRYER

Department of Chemical Engineering,
University of Cambridge, Pembroke St.,
Cambridge, CB2 3RA, United Kingdom.

ABSTRACT

Diffusion of water, followed by reaction with starch, control the processing of whole wheat grains.
Magnetic resonance imaging and gravimetric analysis have been used to study the distribution of
water in wheat grains during cooking in steam and boiling water. Water boiled samples absorbed
water more quickly and to a higher level than did steamed samples. Starch gelatinization was marked
in the images as a well defined "reaction front", which was not observed for the steamed samples.
This can be explained by (i) the higher concentration driving force for mass transfer of liquid water
and that a minimum water content is required to initiate gelatinization. TIle system has been modelled
using a 2D finite element scheme to solve coupled equations for diffusion and reaction.

INTRODUCTION

The cooking of whole wheat grains involves two major processes: diffusion of water into the
grain, followed by conversion of the raw starch within the wheat grain into an edible form due to the
combined presence of heat and moisture. Two types of reaction occur, gelatinization and melting,
whose kinetics are not fully understood [1,2,3]. Understanding the processes that occur during
cooking enables the improvement of industrial operations. The cooking process has been studied in
two different environments of steam and liquid water at 1200c. Magnetic resonance imaging has
been used to probe the spatial distribution of water in the grain. This has been combined with
gravimetric data to obtain semi-quantitative maps of water content. Simultaneous diffusion and
reaction has been modelled using a 2D finite element scheme.
MA TERIALS AND METHODS

Samples of Riband type grains of weight 35 to 40 mg were steam pressure cooked. Distilled water
was brought to boiling before insertion of the samples; steamed samples were placed on a wire mesh,
and water boiled samples were placed in pans containing already boiling water. Nominal cooking
times were measured from the time of insertion of the samples to the time at which depressurisation
commenced. Rapid depressurisation was achieved by venting for boiled samples (60 - 80 sees), but
was allowed to occur by natural cooling for steamed samples (170 - 205 sees). Boiled samples were
then plunged into cold water for 30 seconds to achieve rapid cooling. Moisture contents of the cooked
grains were obtained by weighing before and after cooking; increased mass was attributed to
absorbed water. It was assumed that all raw grains possessed an initial water content of 12%. Cook
times were; 10 - 45 minutes (steamed samples), 5 - 15 minutes (boiled).
Cooked grains were imaged within 3 hours using a Broker MSL200 with a micro-imaging system.
The imaging probe insert had an internal diameter of 5mm. Except where specified the spin-echo
technique was used with an eeho time (1E) of 2.2 ms and a repetition time (TR) of 2 s. Transverse
slices (....().6mm thickness) were taken through the centre of the grains (across the crease) using a slice
472
gradient of - 7 gauss/cm. Gradients in read- and phase-encoding directions were typically 40
gauss/cm. NMR experiments were also performed to determine bulk longitudinal (TO relaxation
times, transverse (1'2) relaxation data, proton intensity (FID), and 1-D imaging profiles perpendicular
to the 2-D imaging slices. These profiles were repeated for a variety of echo and repetition times to
assess to effect of T2 and TI weighting on the images.
RESULTS

The FlD experiments plot showed a good linear relationship between the FlD point height and the
total mass of water in the sample irrespective of the type of cooking involved, showing that the NMR
is responding linearly to water content. Bulk and I-D profile relaxation experiments showed TI (0.39
s to O.44s) weighting on the images to be very small, but some T2 (2rns to 8 ms) weighting does
occur. It is possible to correct make a first order correction using this data. The images obtained at
various stages of cooking of water boiled grains are shown in Figure I(a,b,c). They show water
penetrating evenly into the wheat grain from all points on the boundary including the inside of the
crease. There is also a very well defined "front" where the moisture content drops from a high to a
low concentration over a comparatively short distance. The front moves towards the centre of the
grain as cooking proceeds. The images of steamed grains (Figure 1 d,e,!) are less dramatic, showing
much reduced water contents (as confirmed by gravimetric data). There is no indication of a "front"
and the distribution of water in the grains is much more homogeneous. The level of moisture in the
grains slowly increases with cooking time.
DISCUSSION

It appears that there is a much faster uptake of water into grains during boiling as compared to
steaming. This may be due to a possibly higher thermodynamic driving force for mass transfer in the
former case, or may result from a rapid change in permeability in the outside of the grain during
boiling. The difference in the distribution of water in the grains between steamed and boiled samples
can be attributed to starch gelatinization occurring in the boiled samples. Donovan (I) showed by
DSC, that starch gelatinization only occurs above a minimum water concentration of approximately
30%. This level is not reached in the steamed cooked samples but does occur at the higher water
contents achieved during boiling. One effect of gelatinization appears to be the immobilisation of the
associated water molecules, this process is fast at 1200c resulting in diffusion becoming the rate
limiting step. This causes the sharp profile observed in the boiled samples. The relatively uniform but
slowly rising levels of moisture in the steamed samples indicate that there is a large barrier to mass
transfer at the surface of the grain. There is no obvious effect that any starch is being converted
although some melting of the starch may be occurring.
MODELLING

It is important to be able to model the behaviour of the system. A finite element mesh was
constructed in the shape of a central slice through the wheat grain with 6 noded quadrilateral
elements. The model can incorporate both a mass transfer coefficient boundary condition and a
gelatinization term of the form:
r = k (c-c*) c:>c* and Y > 0
r =0 c<c* or Y = 0.0
where Y - fraction of ungelatinized starch
c - water concentration
c* - minimum water concentration for gelatinization to occur
k - rate constant
Preliminary work was carried out assuming reaction to be "fast" with a Thiele modulus of ten.
Results are shown in figure 2 for the case of gelatinisation using the above equations. Similar
behaviour to the NMR images is seen. Work is continuing to increase the accuracy of the model and
to fit predicted and measured water concentrations.

ACKNOWLEDGEMENTS

Financial support for this work was provided by the Agricultural and Food Research Council
(U.K.) and Weetabix Ltd. The Royal Academy of Engineering and Churchill College, Cambridge are
also gratefully acknowledged for providing travel funding for this conference.
 473

Figure 1. NMR images of wheat grains cooked by boiling and steaming; captions show the time of
cooking

Figure 2. Results of a finite element scheme to simulate the boiling process. Times increase from left
to right in the ratio of 1 : 2 : 4. The darker the figure, the lower the water content.

REFERENCES

1. Donovan, 1. W., Phase Transition of the Starch-Water System. BiQpolymers. 1979, 18, 263-
275.
2. Lund, D.B., Wirakartalrusumah, M., A Model for Starch Gelatinization Phenomena. In
Encineerinc & Food Yol1.- Encineerinc ScienceS in the Food Industry, ed. B.M. McKenna,
Elsevier Applied Science Publishers, London, 1984, pp. 227-69.
3. Reid, D.S., Charoenrein. S. DSC studies of the starch water interaction in the gelatinization
process. 14th North American Thermal Analysis Society Meeting, San Francisco, CA,
September 1985, pp. 335-340.
 STUDIES ON INTERACTION BETWEEN NON ENZYMATIC BROWNING
REACTION AND LIPID OXIDATION IN FOOD PROCESSING.

M.DALLA ROSA., D.BARBANTI*, S.PIZZIRANI** AND F.BRESSA

Dipartimento di Scienze degli Alimenti, Via Marangoni 97, 33100 Udine, Italy.
* Dipartimento di Biotecnologie Agrarie ed Ambientali, Via Brecce Bianche, 60131 Ancona,
Italy
* *Dipartimento di Protezione e Valorizzazione Agro Alimentare, Via S.Giacomo 7, 40126
Bologna, Italy.

ABSTRACT

It is well known that during food processing involving heat treatments (cooking, baking and
flying) non enzymatic browning (NEB) and lipid oxidation can take place. The formation of
NEB products leads to a positive influences on lipid stability during storage of foods. In this
research work, the evaluation of chemical and physical modifications of oil during flying of
different foods in the presence or absence of NEB products formation was carried out.
Furthermore, the influence of the presence of volatiles produced by NEB on reaction lipid
oxidation kinetic of fatty food models was evaluated by means of head space gas
chromatographic analysis. The presence of NEB volatiles was able to slow down the kinetic
rate and to increase the induction time of lipid oxidation. During frying of foods, the
formation of NEB products led to a low level of oil oxidation in comparison with the
oxidation level determined on oil subjected to the same heat treatment without foods.

INTRODUCTION

Maillard Reaction in food brings about some interesting interactions with different alterative
phenomena. Many authors studied the action of Maillard Reaction Products (MRP) to inhibit
the growth of poisoning micro-organisms [1] and the activity on some enzymes such as PPO
and POD [2] and to decrease the level oflipid oxidation [3, 4].
This activity of MRP could be referred to a food subjected to a heat treatment (cooking,
baking, extrusion, drying, sterilisation, etc.) or during storage. In recent years a great number
of MRPs have been proved to be effective as antioxidants in model and food systems [5].
Recently, data on the antioxidant activity of the volatiles produced by Maillard reaction
(MRV), obtained by heating a glucose-glycine solution, were published [6].
Aims of the research work here reported were the evaluation of the effect of MRV on
lowering of oxidation indices (Peroxide value, total volatiles and Hexanal in the head space)
 475
in soybean oil during heat treatment and the assessment of the reduction of oil oxidation
during food frying due to the heat-induced Maillard Reaction.

MATERIALS AND METHODS

Thermoxidation of Model System

Commercial edible soybean oil (SBO) was used as a lipid model system. MRV were obtained
by heating a mother solution of 1.71 M glucose and 2.02 M glycine ( RPE-ACS reagent,
Carlo Erba, Milan, Italy) at 90°C for 24 hr in an air circulating oven, placing 5 mL of
glucose-glycine solution into 20 mL capacity vials hermetically sealed . The pH of the
reaction mixture was adjusted to pH 6.0 with 1 N NaOH (RPE-ACS reagent, Carlo Erba,
Milan, Italy) before starting the reaction. For each sample, 15 mL of heated Maillard solution
head space was withdrawn using a gas-tight syringe and added into a 20 mL capacity vial
containing 5 mL of SBO, after removing the same volume of oil headspace (air). A control
sample (blank) of a 20 mL vial containing 5 mL of SBO alone, equilibrated in the presence
of air, for each oxidative run was also analyzed. Control and MRV treated SBO samples were
heated in a circulating air electric oven at 90°C and tested at regular time intervals for the
determination by Peroxide value (PV) and Head space gas chromatographic (HSGC) analysis
of volatile compounds, as reported in a previous work [6]. The antioxidant activity was
evaluated computing the % of the decrease of peroxide value , total peak area and hexanal
peak area at different times of oxidative heating in respect to the control, and reported as %
of Oxidation Index Reduction (% O.I.R).

Oil Oxidation during Food Frying

Samples of batter (wheat flour with butter, eggs and sugar) were fried in commercial soybean
oil using a thermostatic domestic fryer of about 3 L capacity at 170 ± 5°C for 5 min at
regular intervals of 10 min (4 batches per hr), in order to simulate the frying conditions
normally used in frying operations. An oil/product ratio = 5:1 was used. Frying runs were
repeated for a total time of 8 hr. Oil without any food was heated at the same temperature for
the same time.
The following analyses were performed on oil at 1,2,4,8 hr of heat treatment: Peroxide
value by titration method; the results are expressed as meq of peroxide/k:g sample [7]; Triene
determinations by spectrophotometric measures at 268 om using a Perkin Elmer
Spectrophotometer model 550/SE (Perkin Elmer, Milan, ltaly)[7].

RESULTS AND DISCUSSION

Data reported in Fig. 1 show the effectiveness of the atmosphere modification by MRV in
decreasing the oxidation level of SBO during heat treatment at 90°C. The antioxidant action
was found to be relevant both on primary and secondary oxidation products (peroxides and
volatiles respectively), hence the MRV could be considered active on the initiation as well as
on the termination phase of oxidation, where the formation of rancid flavour occurs.
During frying of batter samples, the presence of the food strongly reduces peroxide and
trienoic group formation in the frying oil in respect to the oil without any food. This action,
referred to in Table 1, is probably due to the development of Maillard Reaction products.
Thus, the antioxidant action of Maillard reaction products is effective to reduce the oxidation
level of frying oil as well as the oxidation of the oil that the food absorbed during the
process. Moreover, the Maillard reaction volatiles outcoming from the food to the package
head space could exert an antioxidant effect during the storage time.
476
70
->t >t_
60

~
50
40
->(-8
~ 30
ci -o---C
~ 20

10
0
0 20 40 60 80 100 120
HEATING TIME (hr)

Figure 1. Percent of reduction of Peroxide Value (A) and Head Space Hexanal (B) and Total
Peaks (C) Area of Soybean Oil heated at 90°C and conditioned with Maillard Reaction
Volatiles.

TABLE 1
Decreasing of Oxidation Index (% O.I.R.) of Soybean Oil during frying of batter samples.

Frying Time (hr) 1 2 4 8

Oxidation Index % Oxidation Index Reduction

Peroxide Value 5.8 38.1 31.2 23.3

O.D.268nm 50.0 58.8 74.1 67.6

REFERENCES
1. Stecchini M.L., Giavedoni P., Sarais 1. and Lerici C.R., Effect of Maillard reaction
products on the growth of selected food-poisoning micro-organisms. Letters in Applied
Microbiology, 1991,13,93-96.
2. Nicoli M.C., Elizalde B.E., Pitotti A. and Lerici C.R., Inhibiting effect of some sugars and
Maillard reaction products on PPO and POD activity. J. Food Biochem., 1991, 15, 169-
184.
3. Lingnert, H and Eriksson C.E., Antioxidative maillard reaction products 1. products from
sugars and free amino acids. J. Food Proc. and Pres., 1980,4, 161-172.
4. Eichner, K., Antioxidative effect of Maillard Reaction lntermediates. Prog. Fd. Nutr. Sci.,
1981,5,441-451.
5. Lingnert H, Eriksson C.E. and Waller G.R., Characterization of antioxidative maillard
reaction products from Histidine and glucose. In The maillard reaction in Foods and
Nutrition, ed.Waller G.R., Feather M.S., American Chemical Soc., Washington, 1983, pp.
125-40.
6. Elizalde B.E., Dalla Rosa M. and Lerici c.R., Effect of Maillard Reaction Volatile
Products on Lipid Oxidation. J. Am. Oil. Chern. Soc., 1991, 68, 758-762.
7. Official and Tentative Methods of the American Oil Chemists' Society, VoU, 3rd edn.,
Champaign, IL, 1973.
 OXIDATION RATE OF POLYUNSATURATED FATTY ACID AND
ANTI OXIDATION EFFECT OF GEL LAYER

KOZO NAKAMURA: KIYOTAKA NAGAI, TOMOMI INOUE AND MASARU HAKODA

Department of Biological and Chemical Engineering, Faculty of Engineering,
Gunma University
Tenjin, Kiryu, Gunma 376, Japan
* Department of Agricultural Chemistry, Faculty of Agriculture,
University of Tokyo, Yayoi, Bunkyo-ku, Tokyo 113, Japan

ABSTRACT

Eicosapentaenoic acid (EPA) and its ethyl ester (EPA-E) were autoxidized in the atmosphere of
different oxygen concentrations, and the oxidation rate (r) could be correlated with the equation,
=
r k [EPA-E) [02t 685 • The oxygen Permeability coefficient in several gels was measured with
an oxygen electrode. Irrespective of the constituents of gels, the permeability coefficient was
almost constant at low water contents, but was increased with increase in moisture contents of
the gels. The retarding effects of the covering gel layer on the autoxidation of EPA-E were
measured and theoretically analyzed in consideration of oxygen mass transfer.

INTRODUCTION

Some polyunsaturated fatty acids (PUFAs), for example eicosapentaenoic acid (EPA) and
docosahexaenoic acid (DHA), have been epidemiologically found to have several physiological
functions such as improvement of serum lipids and prevention against thrombosis. Most
PUFAs are, however, oxidized rapidly, and therefore, development of an effective method for
anti oxidation is required in their utilization. The object of this work is to quantitatively analyze
the anti oxidation effects of gel layer in contact with PUFAs. Firstly, the autoxidation of EPA
ethyl ester (EPA-E) was examined under various conditions of temperature, relative humidity
and oxygen concentration to obtain an equation correlating with the oxidation rate, and then the
oxygen permeability coefficient in several gels was measured using an oxygen electrode.
Finally, the antioxidation effects of the covering gel layer were quantitatively measured and
theoretically analyzed in consideration of oxygen mass transfer.

MATERIALS AND METHODS

478
Autoxidation of EPA-E
Twenty p.1 of EPA-E (99.3 % purity), kindly supplied by Nihon Suisan Corp., Tokyo, Japan
was added in an aluminum vessel (5.3 x 4.7 mm). After standing for the indicated time under
the constant conditions of temperature (298-323 K), relative humidity (90 %) and oxygen
concentration (0.78-8.2 x 10- 2 molll), the residual content of EPA-E was measured by a gas
chromatography.
Oxygen Permeability Coefficient in Gel Layer
The oxygen permeability coefficient ( = (oxygen diffusion coefficient) x (oxygen solubility
coefficient» in several gels was determined by an oxygen electrode, according to the method
for steady-state analysis developed by Ju and Ho [1].
Antioxidation Effects of the Covering Gel Layer
Twenty p.1 of EPA-E in the lower layer of teflon vessel was covered by several kinds of gels
with various moisture contents. After standing under the constant condition (relative humidity,
90 %; temperature, 303 K), the residual content of EPA-E was measured to evaluate the
antioxidation effect of the gels. About 10 % of the water content was found to be reduced
throughout the experiment.

RESULTS AND DISCUSSION

Oxidation Rate for EPA-E

The oxidation rate for EPA-E at lower oxygen concentrations was extensively affected by the
concentration, but the effect was decreased with increase in the concentration. In addition, the
oxidation rate was found to vary with oxidation temperature according to the Arrhenius
equation and the activation energy of oxidation was calculated as about 10 kcallmol. These
results were summarized as follows.
= = =
r k [EPA-E] [02t 685 , k A exp(-EaJRf), A 480 [lo.685/molo.685. sec]
Oxygen Permeability Coefficient in Gel Layer
As shown in Figure 1, the oxygen permeability coefficient in the gel layer with more than 70 %
water content was increased with increase in the water content, irrespective of the constituents.
Antioxidation Effects of the Covering Gel Layer
According to the method proposed by Kishimoto [2], the residual content of EPA-E is shown in
Figure 2 as function of dimensionless time T (= D . tla~ and parameter h (= p. aiD· S . I) in
which D, P, a, I, S are oxygen diffusion coefficient (assumed as 1.0 x 10-5 [cm2/sec]), oxygen
permeability coefficient (our observed values), sample thickness, gel thickness and oxygen
solubility coefficient (= 3.0 x 10- 2 [molll . atm]), respectively. The gels with lower water
contents were found to maintain high antioxidation effects, even though T was longer than
200, while the effects of the gels with higher water contents were steeply reduced with increase
in T , probably because of changes of the gels in structure responsible for rapid permeability of
oxygen. Although Kishimoto has reported that increase of the antioxidation effect was highly
correlated with decrease of h [2], such relationship could not be observed in this experiment
partly because h was as small as 0.001. An analysis in consideration of the oxygen diffusion
rate and the oxidation rate for EPA-E was necessary. The time courses of the oxidation of EPA-
E covered with gel were tentatively simulated by the numerical calculation using the simplified
=
rate equation of oxidation (r k' [EPA-E] [02]) and the equation of mass transfer with the
appropriately assumed rate constants, and the results of simulation were obtained to agree with
the experimental time course of oxidation in several limited case.
 479
.0....., 4.0
:~.: : =
((

IKev Gel
. - J:;:J
~ 3.0 •
•
Sodium Alginate
Relative humidity 90%
Temperature 4O"C

.
Q) • Sodium Alginate,Gelatin
E~ 0 2.5% Dextrin,Gelatin
./
8.. ~ 2.0
II
0
5% Dextrin,Gelatin
10% Dextrin Gelatin •• ..,r
Q (.) ~~ .~
~~ 1.0
-D-'LD-~-Il-"'-fb'- 0

o~~
. . . . 0.0 '--l
o 70 80 90 100
Water concentration [wt%]
Figure 1. Effect of water concentration on oxygen permeability.

--
..
1.0 -.
..... .....
..... ~~'TO
-1,&
',. .
(G.: Gelatin, "-
0.8 S.A.:Sodium Alginate) 'X , "-
,
......... ,
~
~
I I
~ey Gel
Water
h [-J '" "-
~ ~0.6 [%] '"

] g 0.4
-..•
.........
~
~ 0 G.+S.A. 90 0.001
.6 S.A. 95 0.002 'X
0 90 0.0008 "X
"'C:l- G. 60 -
- " ,
~ ~ 0.2
55

•
50 - I
X oxidation without gel >S<:,
'~
0.0
10 0 T [ - ]

Figure 2. Effect of gel on EPA-E oxidation.

CONCLUSIONS

1. The oxidation rate for EPA-E could be correlated with the equation, r = k [EPA-E] [02t· 685•
2. Irrespective of the constituents of gels, the oxygen permeability coefficient was almost
constant at low water contents, but was increased with increase in moisture contents of the gels.
3. Covering with gel layer coud retard the oxidation rate for EPA-E.

REFFERENCES

1. Wang, J. C. T., Ju, L.-K. and Ho, C. S., Oxygen diffusion coefficients in commonly used
topical semisolid creames. Pharma. Res., 1988,5,92-98.
2. Kishimoto, A., Diffusion of oxygen in packed sausage. Bull. Jpn. Soc. Sci. Fishers, 1963,
29, 781-784.
 SIMULATION OF OXIDATION PROCESSES OF LIQUID LIPIDS

SHUJI ADACHI, ruN lMAGI, TATSUJI ISHIGURO, and RYUICHI MATSUNO

Deaprtment of Food Science and Technology, Faculty of Agriculture,
Kyoto University, Sakyo-ku, Kyoto 606-01, Japan

SUMMARY

The autoxidation processes of linoleic acid (LA) and methyl lenoleate (ML) were observed
under various conditions. A kinetic expression which can follow the whole autoxidation
processes of both lA and ML under no diffusional limitation of oxygen has been proposed.
It was found that a kinetic parameter in the expression could be correlated well to the initial
peroxide value (POV) of lipid. Therefore, measurement of the initial POV of a lipid enables
us to predict its oxidation process under no diffusional limitation of oxygen. It has been well
known that the diffusion of oxygen affects largely the overall oxidation progress of lipids.
Simulation of the overall oxidation processes of lipids with different depths suggested that
diffusion of both oxygen and lipid should be considered to predict better the oxidation
processes.

INTRODUCTION

Entrapment of liquid lipids into powdery matrixes of saccharides or proteins [1,2] provides
the lipids some advantages. One of them is a retardation of lipid oxidation [3]. To estimate
how the diffusional limitation of oxygen through the dehydrated matrix contributes to the
retardation and to examine whether the retardation is ascribed to only the diffusional
resistance or not, the knowledges on both the diffusivity of oxygen through the matrix and
the oxidation kinetics of lipid are required.
It has been well known that the specific surface area of lipid affects largely its overall
oxidation process. However, there are a few reports [4] that discuss quantitatively the
oxidation process taking the diffusion of oxygen into consideration.
In this context, we studied the kinetic expression of lipid oxidation under conditions
where the oxidation is a rate-controlling step and the simulation of overall oxidation process
under diffusional limitation of oxygen.

METHODS

Oxidation of lipid: Linoleic acid (lA) or methyl linoleate (ML) was mixed with methyl
palmitate (MP), which was used as an internal standard in a gas-liquid chromatographic
(GLC) analysis, in a ratio of 1:1 in weight. Five microliter of the mixture was put on the
 481
bottom of a small glass vessel. The vessels were
placed in a desiccator in which the humidity was 2
controlled at ca. 75% with saturated solution of
sodium chloride and the partial pressure of oxygen
was also maintained at a given level. The desicca-
tor was held in a temperature-controlled chamber. -2
At appropriate intervals, one of the vessels was
taken out and the amount of unoxidized lA or ML '""""' -4
was measured by a GLC. The oxidation processes ~ 6 LA
of lA and ML with various depths were also .-
measured by the similar way. ~
Diffusion coefficient of oxygen in a lipid: The ~ 2
mixture of IA/MP or MUMP was soaked into a .....
polyfrone membrane filter with an opening of 0.75. ~ 0
The filter was set on a diffusion cell consisting of -2
two chambers. The chambers were filled with N2 'V 343 K
gas, and then air was introduced into one chamber. -4 o 333 K
The change in oxygen pressure of another chamber b. 323 K

was continuously monitored by an oxygen electrode ML o 310K

-6
and recorded. The diffusion coefficient of oxygen
was evaluated from the change.
o 80 120
Time [ h ]
Fig. 1. Applicability of eq. (2) to the
RESULTS AND DISCUSSION oxidation processes of linoleic acid
(LA) and methyl leno1eate (ML) at
Oxidation kinetics: The kinetic expression proposed various temperatures.
by Bolland [5] for the autoxidation of ethyllinoleate
includes the concentration of a-hydroperoxide.
Assuming that the concentration is proportional to the 2.0 LA
decrease in a concentration of lipid, we can obtain the o
following equation. '""""'
dY/dt = -[kmCX/(K + C0]Y(1 - Y) (1) ..",
where Y is the ratio in concentration of unoxidized :10
lipid CL to the total one CLI> Cx is the concentration of ~ .
oxygen in lipid, t is the time, and k m and K are the ~
constants. When Cx is constant, that is, when there is
no diffusional limitation of oxygen, integration of eq.
(1) under an initial condition of Y = Yo at t = 0 gives
the following equation.
In[(1 - Y)!Y] = kt + In[(1 - Yo)lYo] (2) Px [atm]
where Fig. 2. Dependency of the rate
k = kmCy)(K + C0 (3) constant k on the partial pressure of
Oxidation processes of lA and ML were measured oxygen Px for the lA oxidation at
at the partial oxygen pressure of 0.2 and at various 310K.
temperatures, and good applicability of eq.(2) to them
was verified (Fig. 1). The dependency of the rate constant k on the partial pressure of
oxygen Px was expressed by eq.(3) for the oxidation of lA at 310K (Fig. 2).Yo is a
parameter which reflects the initial state of lipid. Correlation between Yo and the initial
peroxide value (POV~, which is commonly used to estimate the deterioration of lipids, was
examined for lA (Fig. 3). From these results, we can predict the oxidation process of lA
under no diffusional limitation of oxygen once its POVo has been measured.
482
Diffusion coefficient of oxygen in lipid: The diffusion
coefficients Dx of oxygen in some lipids were determined LA:MP = l:l(w/w)
at various temperatures. The Dx values in the mixtures of 100
lA/MP and MUMP at 323 K were both 6.7x10- 10 m2/s.
Simulation of oxidation of lipids with different depths:
Figure 4 shows the oxidation processes of lA with various
depths at 323 K, indicating that the specific surface area of
lipid affects largely its oxidation. Supposing no migration
of unoxidized and oxidized lipids, the oxidation processes
can be simulated by solving numerically the following
mass balance equations
vCx/f)t = Dx(f) 2C X/f)x 2) 0.02 0.04
- y[kmCx/(K + Cx)]CL (l - CJC~ (4) 1 - Yo [-]

dCJdt = - [kmCX/(K + Cx)]CL (l - CJC~ (5) Fig. 3. Correlation between the

initial peroxide value (POVcJ
with adequate initial and boundary conditions and by
and the parameter Yo.
integrating the concentration profile CL(x,t) over x = 0 to
1, where y is the stoichio-
metric coefficient, which is
assumed to be unity. Solid
:s
dJo 0
curves in Fig. 4 shows the 00 )
simulated re-sults, which lie CD 11.5 mm
far above the experimental %0
6. ) 0
points. This discrepancy
6. L = 2.6 mm 00 0
may be due to taking no o
Control
account of migration of
oxidized and unoxidized
lipids although the concen-
tration profile CL is steep. 30 40
For better prediction of Time [days]
oxidation process, further Fig. 4. Oxidation processes of linoleic acids with various
investigation is required. depths L at 323K, and their comparison with the calculated
results.

REFERENCES

1) Imagi, J., Kako, N., Nakanishi, K., and Matsuno, R, Entrapment of liquid lipids in
matrixes of saccharides. J. Food Eng., 1990, 12, 207-222.
2) Imagi, J., Yamanouchi, T., Okada, K., Tanimoto, M., and Matsuno, R, Properties of
agents that entrap liquid lipids. Biosci. Biotech. Biochem., 1992, 56, 477-480.
3) Imagi, J., Muraya, K., Yamashita, D., Adachi, S., and Matsuno, R, Retarded
oxidation of liquid lipids entrapped in matrixes of saccharides or proteins. Biosci.
Biotech. Biochem., 1992,56, 1236-1240.
4) Heiss, R, Shrader, V., and Reinelt, G. R, The influence of diffusion and solubility on
the reaction of oxygen in compact food. In Food Process Engineering, Vol. 1, ed. P.
Linko, Y. Malkki, andJ. Olkku, Applied Science Publishers, London, 1980, pp.364-370.
5) Bolland, J. L., Kinetics of olefin oxidation, Quart. Revs., 1949,3,1-21.
 ISOMERIZATION REACTION KINETICS OF CATECHINS IN PACKAGED
TEA DRINKS DURING PROCESSING

YOSHIHIRO KOMATSU, SHINICHI SUEMATSU, YOSHIHIRO HISANOBU,

HIDEAKI SAIGO, RYOKO MATSUDA, AND KYOKO HARA.
Toyo Institute of Food Technology,
4-23-2 Minamihanayashiki, Kawanishi, Hyogo 666, Japan

ABSTRACT
Stability of tea catechins during processing of tea drinks was
examined. A dominant change of tea catechins by heating was
epimerization. Effects of pH and temperature on reaction
kinetics of degradation of -EC, -EGC, -ECg, and -EGCg in green
tea infusion were investigated. A turning point temperature on
the Arrhenius plot of apparent first order reaction rate
constants was observed at 2.82 x 10- 3 (1 /K) ; 82·C. Apparent
energies of activation were temperature-dependent and increased
by 7.3-11.4 times above the turning point temperature. The
reaction rate constant of catechins in slightly acidified green
tea infusion was less than half as much as that in the original
infusion.

INTRODUCTION
Canned or PET bottled drinks of black, oolong, and green tea are
widely supplied for consumer use in Japan. Most packaged tea
drinks are low acid beverages. High temperature heat processing
is required according to Japanese food hygienic regulations. It
is important to minimize changes of constituents in tea drinks
during production and storage. Caffeine is highly stable but
catechins are rather unstable and the dominant change of
catechins seems to be isomerization influenced by pH of the
infusion at the heating(1). The effects of pH of infusion and
heating temperature on the reaction rate and apparent
activation energies of catechins in green tea infusion are
investigated(2).
484
MATERIALS AND METHODS
Standerd reagents of catechins were obtained from Kurita Kogyo
Co. ,Ltd., Japan, and Extrasynthese Co., France. A reaction
apparatus was designed to simulate the extraction, hot filling,
and sterilization. In the apparatus, the solution was heated up
to 121°C within 30 seconds and cooled to ambient temperature
within 5 seconds. Catechins were determined by the HPLC method
described by Terada et al.(3) with a few modifications.

RESULTS AND DISCUSSION

Experiment 1. Reaction kinetics of -EC in citrate-phosphate
buffer. Citrate-phosphate buffer solutions adjusted to pH 4.0,
5.0, 6.0, and 7.0, containing 8 mg/100 m1 of -EC was heat-
processed at 121°C for various times from 1 to 15 min. An
apparent first order reaction proceeded in slightly acidic media
less than pH 5.0, but a different mode of reaction was observed
in media exceeding pH 6.0 (Fig. 1).

~
~
O~~
_Q.5

-\0
-EGC -EGCg

·c'"
01 c: 0
C
·0
C
pH 6.0 E
o
E '"
L..
-1
(I)
a::: '"
~

-2 -EC

10 L-_ _---I_ _ _--'-_ _ _--'----'

o 5 10 15 024 S 16024 S 16
Heating time (min) Heating time (min)
Fig. 1. Apparent First Order Reaction Rate Plot of 025"c. e40"C. "55"C. .SO"C. DS5"C. ~90"C. .95''C.
-EC in Macllvaine Buffer Solution at various pH Fig .. 2. Apparent First Order Reaction Rate plot
under Sterilization Conditions at 121"C. of -EGC, -EGCg,-EC, and -ECg in Green Tea
Infusion at 25. 40, 55, 80, 85, 90, and 95"C ..

Experiment 2. Reaction kinetics of catechins in green tea

infusion. The log of the ratio remaining to the initial
concentration of -EGC, -EGCg, -EC, and -ECg in green tea
infusions for each temperature was plotted against heating time
(Fig. 2). The reaction fitted apparent first order kinetics
for the four kinds of catechins at the temperature examined, but
the reaction rate constants were different for each kind of
catechins. Each Arrhenius plot showed not a straight line but a
concave one consisting of two straight lines which crossed each
other at a specific turning point. The turning point was
commonly observed on each Arrhenius plot at 2.82 X 10- 3 (11K) ;
at 82~ (Fig. 3). The apparent activation energies obtained
were slightly different between the four kinds of catechins.
It was 7.3 to 11.4 times larger above the turning point
temperature. This meant that the actual reaction rate measured
at a temperature higher than 82°C was faster than the reaction
 485
rate predicted by extrapolation from the one measured below 82·C.
Experiment 3. Effects of pH on reaction kine~ics of tea
catechins. A similar experiment was done with or without an
addition of AsA to the green tea infusion. The reaction also
fitted an apparent first order kinetics. 'Total catechins' were
shown as the sum of -EC, -ECg, -EGC, and -EGCg to simplify the
comparison. The turning point temperature on the Arrhenius
plot was also observed in both slightly acidic and original
media at 82~ (Fig. 4). Therefore, a similar mode of reaction
proceeded in slightly acidic infusions.

-3.5 -3.5

-4.0 -4.0

1"
1" -4.5 ~ -4.5
01
~ 0

01 -5.0 -5.0
0

-5.5 -5.5
~--~--~~--~--~
2.6 2.8 3.0 32 3.4
liT (K) x10 3
-6.0 LI-_-'-_---"'--_-'-_-'-'
Fig. 4. Arrhenius plot of Apparent
2.6 2.8 3.0 32 3.4 First Order Reaction Rate Constant
liT (K) xl03 of 'Total Catechins' in Green Tea
Fig. 3. Arrhenius Plot of Apparent First Infusion with or without an Addition
Order Reaction Rate Constants of -ECg, -EC, of L-ascorbic Acid.
-EGC, and -EGCg in Green Tea Infusion.

In summary, a simple time-temperature relationship was not

established in the thermal stability of tea catechins. The
stability of catechins in green tea infusions was subject to
the turning point temperature. A careful consideration of time-
temperature infusing conditions was required to extract natural
catechins from green tea leaves for quality evaluation or
preparation of ingredients for pathological studies.
REFERENCES
1. Suematsu, S., Hisanobu, Y., Saigo, H., Matsuda, R., Hara, K.,
and Komatsu, Y., Effect of pH on stability of constituents in
canned tea drinks. Nippon Shokuhin Kogyo Gakkaishi (in
Japanese), 1992, 39, 178-82.
2. Komatsu, Y., Suematsu, S., Hisanobu, Y., Saigo, H., Matsuda,
R., and Hara, K., Effects of pH and temperature on reaction
kinetics of catechins in green tea infusion. Biosci. Biotech.
Biochem. .• 1993, 57(6), 907-10.
3. Terada, S., Maeda, Y., Masui, T., Suzuki, Y., and Ina, K.,
Comparison of caffeine and catechin components in infusion of
various tea and tea drinks. NiPpon Shokuhin Kogyo Gakk.aishi
(in Japanese), 1987, 34, 20-27.
 CITRUS FRUITS COMBINATION IMPROVING TASTE AND FLAVOR OF
MARMALADE

TOSHIKO MORISHITA and KOZO NISHIO

Department of Food Science" Faculty of
Home Economics, Mukogawa Women's University
Nishinomiya, Hyogo, 663, Japan.

ABSTRACT
The bitterness of citrus fruits increases when they are used to prepare
marmalade. To overcome this problem, the change in bitterness was studied
using objective methods. Marmalade was prepared from a 1:1 mixture of two
kinds of citrus peel. The limonoid and flavonoid levels were determined by
HPLC. Taste evaluation was done by subjects. The amounts of flavonoids
decreased markedly with heating, indicating that they were not responcible
for the bitterness. Limonin also decreased with boiling, but the amount bf
nomilin increased,suggesting that it was respon1sib1e for the bitterness.
Fruits containing naringin were rated as tasting good (5% significance
level). However, fresh fruits containinq larqe amounts of limonin were not
Judged favorably. Mixing two different kinds of citrus fruits is suggested
as a way of improving the flavor of marmalade.

INTRODUCTION

The varieties of citrus fruits commonly grown in Japan have been

changing and thier yields have been increasing. Thus a great demand has
grown ways to use them in processed foods. Their flesh and juice have been
utilized, but there have been little use for the peel. One good way of
utilizing citrus peel is.toprepare marmalade, but the problem arieses of
the bitter taste increasing during the processing.
The bitterness has been attributed to limonoid and flavonoid(1)(2).
In orde~ to find the optimal condition for marmalade preparation,
we examined four kinds of citrus fruits: yuzu, konatsu, kinkan,and navel
orange. The changes in the bitter substances during soaking in hot water,
and the effect on taste evaluation were investigated. How the amounts of
acid, sugar and pectine of these citrus fruits affected the taste of the
marmalade was'examined. We also tried to established the best mixture of
these fruits for producing savory marmalade
 487
MgTERIAL and METHODS
Materials. The peel and juice of yuzu( Citrus junos Sieb, et Tanaka),
konatsu( Citrus taurana Hort. et Tanaka), kinkan( F. crassfolia Swingle)
and navel orange( C. sinensis Osbeck var. brasilliensis Tanaka) were obtain-
ed from i ndustri a 1 processors, " Hamakoh" Kochi prefecture, Japan.
Extraction of limonoid (3). The peels were cut into 2 X 10 mm pieces and
2.0 g was milled with a mortar. Water( 20 ml) was added to the paste, and
the mixture was filterd. The filtrate was shaken 50 times in 10 ml hexane,
and the hexane layer was removed. The residue was shaken 150 times with
chloroform to transfer the limonoid to the chloroform layer. This layer was
washed with 20 ml of water and evaporated on a water bath at 60°C. The
residue was dissolved with 75 %methanol, and passed through a filter of
0.45 um pore size. The filtrate ( 10 ~1) was directly analyzed at 35°C by
HPLC ( Nihon Bunko model, Jasco 880).
Analysis of 1imonoids (4). The 1imonoid content was analyzed by HPLC,
equipped with a solvent delivery system controlled by a model Jasco 880-PU
and a refractometer ( Jasco 875-UV ) at 215 nm. The column was packed with
silica ( 4.6 X 150 mm ), and the eluent was 75 % methyl alcohol used at a
flow rate of ~.2 m1/min.
Extraction of naringin. The peels were cut and 2.0 g was milled with a
mortar. Next 20 ml of water was added to the paste, and the mixture was
sent through a filter of 0.45 um pore size. The filtrate (.1 ~l)was
directly analyzed at 35°C by HPLC.
Analysis of naringin (5). Naringin content was analyzed by HPLC at 283
nm. Using a column packed with silica ( 4.6 X 150 nm) and 25 % acetonitrile
/0.2 % phosphoric acid as the eluent used at a flow rate of 1.0 m1/min.
Preparation of marmalade (6). The peels were cut into 2 X 10 mm pieces,
soaked in nine volumes water for 2 hours, and boiled for 5 minutes. Next,
20 g of peels, 20 g of juice and 24 g of sugar were mixed in a bottle and
heated on the hot plate to a final temperature of 105°C with a sugar
degree of 55-65, pH. 3.0-3.4.
Taste evaluation. Thirteen subjects graded the marmalade for bitterness,
sweetness, flavor and taste. The marmalade were prepared with the peel of
two types of fruits in a 1:1 ratio. The combination were (a) YUlU : konatsu
(b) yuzu : kinkan, (c) yuzu : navel orange, (d) konatsu : navel orange, (f)
kinkan : navel orange.

RESULTS and DISCUSSION

Limonin content in the boiled peels was high in the order of konatsu,
yuzu, kinkan, navel orange, while nomilin content was high in order of
yuzu, konatsu, navel orange, kinkan. Soaking yuzu peel in water foll.owed
by boiling, fost 70 % of its limonin, but increased its nomi1in content
from 5.4 to 15.8 ppm. Water solubility of naringin was increased by
heating and about 70 % was dissolved in water. Its content was lower
than its threshold of 8 mg% . Naringin did not contribute to the bitter-
ness.
488
The intensity of the bitterness of marmalade WaS in order of yuzu, konatsu,
navel orange, kinkan, with the difference between yuzu and kinkan being
significant. The bitterness was correlated to the nomilin content of boiled
peels. Konatsu which contained a large amount of limonin was not favorable
in taste. Also, limonin did not have a good effect on marmalade quality.
A contrastive effect was found between bitterness and sourness, while a
restraining effect was found between bitterness and sweetness and between
sourness and sweetness. The best of balanced flavor and good taste among
the citrus fruits tested was the navel orange. To prepare mor~ tasty
marmalade, various combination of two of these citrus fruits were tried.
Combinilng a strong-tasting fruits with a weaker one did not reduce the
strong taste but rather enhanced it. Mixing fruits with similar bitterness
and sourness helped control the taste. Sourness and sweetness CQuld
alleviate the bitter taste and thus improve the flavor.

CONCLUSION
The limonoid and naringin content of yuzu, konatsu, kinkan and navel
orange were determined by HPLC. Marmalade were prepared with different pair
combination. Taste evaluation was conducted to judge whether the bitter
taste of various fruits influenced the quality.
The results showed that when the peels were soaked in water for 2 hours and
heated for 15 minutes, limonin decreased and nomi1in increased. When the
peels were heated for 15 minutes. naringin became water-soluble and decrease-
ed in content, it fell by 63-76% . The naringin content in the peel and
juice affected the flavor of the marmalade according to taste evaluation.
When a marmalade contained a large amount of limonin. it was considered to
be a favorable taste. Mixing fruits with similar bitter and sour taste,
enabled to control ,the taste. Also, the sour and sweet taste could control
the bitter taste and improve the flavor. The best combination was that of
konatsu and navel orange, while the most disliked was that of kinkan and
yuzu.

.REFIlRHNCES
1. F.Hashinaga and S.Ito, Seasonal changes in limonoi~s in hassaku and
pommelo fruits. J.Japan. Soc .. Hort. .Sci. 1983, 51, (4), pp. 485-492.
2. Hasegawa S. and V.P. Maier, Solutlon to the limoninbitterness
problem of citrus juices, Food Technology, 1983, June, pp. 73-77.
3. T.Morishita, R, iWada and H. Fujii, Behaviors of limonoid in banpeiyu
liquor processing, Bun. Mukogaw'a Women's Univ. Nat. Sci. 1988, 36,
pp. 19-21.
4. T. Morishita, H. Jinnai and H. Nakachi, The isolation of limonoids of
banpeiyu, J.JaRanese Society for Food Sci. and Technology, 1985, 32,
(8) , pp. 44-4b'.
5. T.Mori~hita and R. Wada, Changes of naringin concentration in banpeiyu
during freezing, J. Science of Cookery, 1985, 18, (4), Pp. 55-57.'
6. T.Morishita and R.Wada, Removal of bitter substances in citrus peel
in marmalade processing-application of naringinase, J. Science of
Cookef\~,1986, 19, (4), pp. 96-100.
 KINETICS OF ASPARTAME DEGRADATION IN LIQUID DAIRY BEVERAGES

LEONARD N. BELL
Drug Delivery Research and Development
The Upjohn Company, Kalamazoo, MI 49001, USA

DONALD SHOEMAN, MENEXIA TSOUBELI, AND TIIEODORE P. LABUZA

Department of Food Science and Nutrition
University of Minnesota, St. Paul, MN 55108, USA

ABSTRACT

This research evaluated the stability of aspartame in dairy beverages as influenced by buffer
type, buffer concentration, pH, flavor, and temperature. The pH and storage temperature are
the two most important factors for a successful aspartame-sweetened dairy beverage. The
buffer type and concentration have only a minimal influence on the aspartame stability.
Vanilla does not enhance the degradation of aspartame in dairy beverages. A commercial
chocolate dairy beverage had an acceptable shelf life when held below 4°C. The activation
energy for aspartame degradation in such a beverage was 17.4 kcallmole.

INTRODUCTION

The market for nutnuous, low calorie dairy products continues to grow annually. The
reduced-calorie beverages which are used in weight reduction programs are often powders
which require mixing with skim milk. The few sterile liquid products found in health food
stores are sweetened with saccharin to mask their undesirable cooked flavor. The suggested
carcinogenicity of saccharin makes it desirable to develop such products with another high-
intensity sweetener. Aspartame has been approved for use in numerous dairy products,
including refrigerated dairy-based drinks. However, the major problem with aspartame is that
at the inherent pH of milk (pH 6.6), the rate of aspartame degradation is extremely rapid,
even when refrigerated [1,2]. Other factors which influence the degradation rate of aspartame
include the presence of dairy proteins [2], reducing compounds [3], and buffer salts [1,2].
The degradation of aspartame reduces the sensory shelf life of the product. If aspartame
could be stabilized in a neutral pH liquid, many new reduced-calorie dairy beverages would
be developed. The objective of this research was to evaluate aspartame stability in dairy
beverages of various compositions.
490
MATERIALS AND METHODS

Formulated Dairy Beverage Study

Citrate or phosphate buffer salts and aspartame (600 ppm) were dissolved into 30 L water;
the pH of the systems ranged from 6.38 to 6.66. To this solution, 3 kg nonfat dry milk was
mixed to form a buffered milk beverage. To one half of the mixture, 600 ppm of vanilla was
added while the other half remained unflavored. The beverage was homogenized,
commercially sterilized, and filled into polyethylene bags which were then divided into small
pouches. These pouches were stored at either 4 or 30°C. A pouch from each experimental
condition was removed for analysis at regular time intervals. From each pouch, a 1 mL milk
sample was mixed with 9 mL phosphate buffer at pH 4.4. A portion of this mixture was
placed in a 10,000 molecular weight cut-off ultrafiltration tube which was centrifuged for 30
min at 2500 g. The aspartame in the filtrate was analyzed using a reverse-phase ion-pair
HPLC method [1,2]. Aspartame loss was modeled by pseudo first order kinetics and the rate
constants with 95% confidence intervals (C.I.) were calculated. Half-lives are presented here.

Commercial Chocolate Milk Study

A commercially available aspartame-sweetened (300 ppm) chocolate dairy beverage whose
initial pH was 6.8 was stored at five temperatures between 0 and 30°C. For analysis, samples
were diluted with phosphate buffer at pH 4.4 and centrifuged at 20,000 g for I h. The
aqueous supernatant was placed in a 10,000 molecular weight cut-off ultrafiltration tube which
was centrifuged for 45 min at 1200 RPM. The filtrate was analyzed using HPLC. The
pseudo first order rates constants and activation energy were calculated.

RESULTS AND DISCUSSION

Table 1 lists the half-lives for aspartame degradation in the reduced-calorie dairy beverages
at 30°C and 4°C. Aspartame stability is very poor in the beverage at 30°C. The maximum
stability was found in the beverage at pH 6.38 containing 0.008 M added phosphate buffer;
however the half-life of this beverage was only 3.5 d. To produce and distribute a shelf
stable product, a minimum time of 6 to 9 mo is required. Refrigeration greatly improved the
stability of aspartame in the dairy beverage. The half-life of aspartame in the beverage at pH
6.38 containing 0.008 M added phosphate buffer increased to 53.3 d at 4°C, which is adequate
for the distribution of a refrigerated dairy beverage. The stability was also significantly
improved by reducing the pH which is consistent with previous research [1,2].
The stability of aspartame did not differ significantly at a given pH level as a function
of added buffer concentration as shown in Table 1. At the levels of buffer concentration used
in this study, the ionic groups of the dairy proteins may interact with the buffer so that the
buffer salts are unavailable to catalyze aspartame degradation. Another explanation may be
that the inherent buffer components of the milk are masking the catalyzing effect of the added
buffer. The buffer type also does not significantly influence the rate of aspartame loss in the
reduced-calorie dairy beverages which again indicates the importance of either the inherent
dairy buffer salt types and concentrations or the protective nature of dairy proteins. The
addition of vanilla did not significantly enhance aspartame degradation.
 491
TABLE 1
Half-life of aspartame in formulated dairy beverage

pH Added Buffer Half-life of aspartame in days

6.66 None 1.2 1.6 25.7 23.9

6.51 0.0017 M Citrate 2.6 1.9 40.8 40.8
6.67 0.008 M Citrate 1.3 1.3 23.9 24.8
6.43 0.008 M Citrate 2.2 2.2 57.8 49.5
6.46 0.016 M Citrate 1.4 n.a. 46.2 40.8
6.38 0.008 M Phosphate 3.4 3.2 53.3 57.8
6.39 0.016 M Phosphate 2.6 2.1 49.5 49.5

The half lives of aspartame in the commercial chocolate dairy beverage were 1.3 d at
30°C, 12.5 d at lOoC, 19 d at 4°C, and 33 d at O°C which are consistent with the results at pH
6.7 in Table 1. The activation energy for aspartame degradation in the chocolate beverage
was 17.4 kcallmole which is similar to the 14 to 18 kcallmole found in the model dairy
protein/aspartame solutions [1,2].

CONCLUSIONS

Aspartame can be stabilized in dairy beverages if the pH is decreased by the addition of

buffer salts or acids in low concentrations, the inherent bacterial contamination is eliminated,
and the product is refrigerated. The type of added buffer and concentration are of secondary
importance as compared to the pH. Vanilla extract does not enhance the breakdown of
aspartame in dairy beverages. Aspartame in the chocolate dairy beverage has an acceptable
stability if kept properly refrigerated.

REFERENCES

1. Tsoubeli, M.N. and Labuza, T.P., Accelerated kinetic study of aspartame degradation
in the neutral pH range. J. Food Sci., 1991, 56, 1671-1675.
2. Tsoubeli, M.N. and Labuza, T.P., Influence of dairy proteins on aspartame stability
in the pH 6-7 range. J. Food Sci., 1992, 57, 361-365.
3. Ho, C.T., Huang, T.C., Cha, A.S., and Sotirhos, N., Studies on the stability of aspartame
in the presence of glucose and vanillin. In Frontiers of Flavor. ed. G. Charalambous,
Elsevier Science Publishers, Amsterdam, 1988, pp. 233-240.
 Esterification in Micro Aqueous Organic Phase

Masafumi INOUE, Masanao IMAl and Masaru SHIMIZU

Department of Chemical Engineering, Division of Chemical and Biological Science,
Tokyo University of Agriculture and Technology,
24-16, 2-Chome, Nakamachi, Koganei, Tokyo 184, JAPAN

ABSTRACT

Esterification of oleic acid with lauryl alcohol, i.e. lauryl oleate, with lipase from candida cylindracea,
in micro aqueous phase was demonstrated. Di-2-ethlyehexyl sodium sulfosuccinate (AOT) was used
as amphiphilic molecules in organic solvent. The initial reaction rate changed with the change in the
molar ratio of water to amphiphilic molecules W. The maximum rate of esterification was obtained
at W=12. The reaction kinetic parameters were determined quantitatively according to Michaelis-
Menten analysis based on the sequential mechanism.

INTRODUCTION

Most of the conventional chemical processes need high temperatures, high pressures and special
catalysts. Accordingly, the cost of such productions must be expensive in comparison with the
processes with biocatalysis or enzymes which can be carried out under very moderate operating
conditions.
Micro aqueous phase has been formed by means of self assembled amphiphiIic molecules. In this
system nanometer scale water pools are dispersed in the organic phase, in which enzyme would be
entrapped. So, it is expected for the system to carry out successfully enzyme reaction of lipophilic
substrates. Micro aqueous organic phase can be considered to be simpler than the phase of dispersed
immobilized enzyme particles. Because the former would have only one liquid phase and show high
mass transfer rate. By using the former phase, the reaction can be carried out under moderate
conditions. This is the most advantageous point on industrial applications of biocatalysts.
Previous papers (1, 2) reported conditions for the maximum reaction rate in esterification by
lipase with micro aqueous organic phases. In order to design esterification processes via biocatalytic
routes, quantitative data are necessary for reaction parameters concerning each sub~trate.
In this study kinetic parameters are presented for the esterification of oleic acid and lauryl alcohol
with lipase in micro aqueous organic phase. This system was adopted as a model of the enzyme
reaction of two substrates. Experiments were focused on the change in the reactivity of lipase with the
changes of water content and AOT concentration.
 493
EXPERIMENTAL

Materials and Methods

Esterification of oleic acid (cis-9-octadocenoic acid: OsH340z) with lauryl alcohol (1-dodecanol :
C12H260) was chosen as a model of the enzyme reactions. The amphiphilic component used was di-2-
ethylhexyle sodium sulfosuccininate (C20H3707Na, AOT : Nacalai Tesque, Inc.). Isooctane was used
as organic solvent. The aqueous solution of the lipase (optimal pH: 7.2) from candida cylindracea
(4865units/mg : SIGMA Chemical Company) was used for the esterification. This lipase was used in
this study because it showed high reactivity of esterification for the enzyme reaction of two substrates
(1). The pH in the aqueous phase was adjusted with buffer components of NaOH and KH2P04. The
solution was prepared right before the aqueous enzyme solution was fed into the reactor. All reagents,
except AOT and lipase, were purchased from Wako Pure Chemical Industry, Ltd. and used without
further purification.

Esterification
The lipase of 300mg was used for every experiment. Total liquid volume in the reactor at the initial
state was 400ml. It consisted of the very small amount of a dispersed aqueous phase containing lipase
and an organic phase containing two substrates and AOT. The micro aqueous organic phase was
formed in a glass vessel with four baffle plates at the temperature 303K and agitated by a paddle
impeller at 500 I:p.m. The solutions were transparent for the whole experimental conditions. Samples
were taken out of the reacting solutions, and then by Florisil glass capillary Chromatography, ester
oleate was separated from samples. The amount of produced ester was determined by gas
chromatography. The water content was determined by the Karl-Fischer titration.

RESULTS AND DISCUSSIONS

Effect of pH in the Aqueous Phase

The maximum amount of ester production was obtained at pH 7.2. This pH was the same pH as the
optimum pH value for candida cylindracea. As pointed out by Walde etaZ.'s (3) pH values in micro
water pools tend to change from their initial values due to the presence of impurity components. In
the present study such a pH change was not observed.

Effect of Water Content in Organic Phase

Figure 1 shows the change of initial reaction rate with W. The maximum reaction rate was obtained at
W=12 for all of the AOT concentrations tested in the present experiment. This result was similar to
the Hayes et ai.'s (1) for AOT system and the Chen et aZ.'s (4) for phosphatidylcholine system,using
the same lipase from candida cylindracea.
When W was lower than 12, the initial reaction rate of esterification increased with W. In such
cases the relative amount of water in organic phase seemed to be insufficient for the activity of lipase.
When W was higher than 12, the initial reaction rate decreased with increasing W. Because lipase is
considered to be active for hydrolysis rather than for esterification.

Effect of AOT Concentration in Organic Phase

The maximum reaction rate was obtained at the AOT concentration of 50mM for all of the tested water
contents. The range of relative water content W was from 8 to 20. The ester productivity of lipase
was maximum at the AOT concentration of 50rnM. When AOT concentration lower than 50mM, the
ester production was decreased.It is presumably due to the insufficient entrapment of lipase by AOT
molecules. On the other hand, as AOT concentration is higher than 50mM, too many AOT molecules
around a lipase molecule would inhibit the reaction-diffusion processes.
494

OA=53.5mM
LA = 63.8mM
AOT=50 mM

'"s
i 0.05
a

0.00 '---'---'_-'----'-_"'------'-_'--.....1---'-----1
o 5 10 15 20 25
WI·]

Fig.] Vi vs. W

Kinetic Parameters
Kinetic analysis of the present esterification reactions are shown in terms of Lineweaber-Burk plots,
which clearly satisfy Michaelis-Menten kinetics. Assuming Michaelis-Menten kinetics with sequential
mechanism the following expression can be derived:

in which v is the reaction rate, V = 0.136 [mM/min] is the maximum rate, [SOA] is the concentration
of oleic acid, [SIA] is the concentration oflauryl alcohol, KmSOA = 43.4 [mM] and Km SLA = 4.874
[mM] are Michaelis constant of oleic acid and lauryl alcohol and Ki SOA = 276 [mM] and Ki SLA = 30.9
[mM] are "inhibition constant" of oleic acid and lauryl alcohol, respectively.

CONCLUSIONS

Esterification of oleic acid by lipase in micro aqueous phase formed by amphiphilic molecules was
examined. The results are summarized as follows:

1. Maximum initial reaction rate was achieved at W=12, at the AOT concentration of 50mM.
2. Kinetic parameters were determined by Michaelis-Menten analyses with the sequential mechanism.

REFERENCES

(1) Hayes, D.G., and Gulari, E. Esterification reaction of lipase in reverse micelles. Biotechnology
and Bioengineering, 1990, 35, 793-801.
(2) Rao, A.M., Murray, M.A., and John, Y.T. Characteristics of lipase catalysis during ester
synthesis in reversed micellar system. Biocatalysis, 1991,4,253-64.
(3) Walde, P., Wao, Q., Bru, R., Luisi, P.L., and Kuboi, R. pH artifacts in reverse micellar
enzymology: a warning. Pure & App/. Chern., 1992, 64, 1771-75.
(4) Chen, J., and Pai, H. Hydrolysis of milk fat with lipase in reversed micelles.,Journal of Food
Science, 1991, 56, 234-37.
 THERMAL INACTIVATION OF PECTINESTERASE IN PAPAYA PULP (pH 3.8)

P.R. MASSAGUER, M.A. MAGALHAES & R.M. TOSELLO

Depart. de Ciencia de Alimentos, Fac. de Engenharia de Alimentos, Universi-
dade Estadual de Campinas, C.P. 6121, 13081-970 Campinas, &ID Paulo, Brazil.

ABSTRACT

In order to establish the thermal process required by acidified papaya

pulp (pH 3.8) var. "Formosa", a study was carried out on the kinetics of
thermal inactivation of heat resistant enzymes, present in the pulp. Since
no peroxidase activity was detected, the study was focused on pectinesterase.
The heat inactivation curves of pectinesterase at 75, 77 and 80 0 C
indicated the presence of two different portions of the enzyme, one heat
labile and the other thermostable. The decimal reduction times (D) of
pectinesterase were 0.8, 0.3 and 0.2 min. for the heat labile portion and
16.7, 7.2 and 3.7 min. for the thermostable portion respectively.
The thermal coefficient was 9.2oC for the heat labile portion and 7.8 o C
for the thermostable portion, while the activation energy were 61.7 and
72.7 kcal/mol. Thermal destruction studies conducted with Clostridium
pasteurianum, at the same temperatures, showed that the thermostable
portion of pectinesterase presented greater resistance and should be used
as target for the establishment of the required process.

INTRODUCTION

The use of papaya pulp as a base product for the formulation of mixed
juices, desserts and baby food is presently increasing, even though papaya
is still mainly consumed as fresh fruit. The processing schedule required
for papaya (Carica papaya L.) at its natural pH (4.2-5.6) has a deterio-
rative effect on texture and flavor. In order to obtain a stable product
and eliminate the gelatinization of the pulp caused by enzymes, thermal
treatment associated with acidification is recomended. In such case,
spoilage is caused by gram-positive, non-spore forming bacteria, or by
yeast. For such acid products, Cl. pasteurianum is commonly used as
target organism to establish the process required.
In 1981, Nath and Ranganna (1) reported a 2.5D process for papaya
pieces in syrup (pH 3.8) to ensure a safe acceptable product, which was
based on the thermal inactivation of pectinesterase (PE). There was no
indication of the presence of PE isoenzymes with different thermal
stability, though recently other researchers reported the presence of a
thermolabile (PEPTL) and a thermostable (PEPTR) fraction in other fruits.
In 1978 Ling and Lund (2) described a method for estimating the
kinetic parameters of thermal inactivation of enzymes systems containing
496
no more than two fractions of isoenzymes.
The objective of the present work was to determine the kinetics of
thermal inactivation of heat resistant enzymes in acidified papaya pulp,
pH 3.B (APP) and compare its resistance with that of Cl. pasteurianum to
establish a thermal process.

MATERIAL AND METHOD

1) Preparation of papaya pulp (APP): Papaya var. "Formosa" was used.

Ripeness was visually evaluated. After end-cutting, peeling and removing
of the seed, the fruit was cut and blended for 5 min. Brix, pH peroxidase
and pectinesterase activity (2) were determined at this point, followed
by straining and acidification with citric acid to pH 3.8. The pulp was
stored at -18 o C, and it was thawed and used for experiments.
2) Determination of heat resistance of Cl. pasteurianum in APP: Liver
broth infusion (DIFCO) was used for sporulation of Cl. pasteurianum FTPTA
0203. The spore suspension (8 ml of 10 6 sp/ml) was~noculated in a 3 neck
flask containing 782 ml of APP and heated at programmed intervals at 75,
77 and BOoC. Survivors were counted by NMP method, using liver broth
infusion (DIFCO). D values were calculated from the slope of the
regression curves.
3) Thermal inactivation of pectinesterase in APP: Two ml of APP were
dispensed in each of 12 unsealed TDT tubes and heated in an oil bath
(± 0.5) at 75, 77 and 80 o C. Residual PE was extracted with sodium chloride
(10%) and its activity was determined. Activation energy (Ea) was also
calculated.

RESULTS AND DISCUSSION

No peroxidase activity was found. The residual percent of PE as a

function of heating time was plotted for each temperature (Fig. 1). Since
no first order reaction behaviour was observed, Ling and Lund (2)
procedure was employed to determine the rate constants for the thermo-
labile and thermostable portion. A semilog plot of %PE - A versus time
was made to analyze the PEPTL kinetics (Fig. 2). A is the intercept of
the regression line in Fig. 1. Fig. 3 is the phanton curve of the two
portions of PE.

TABLE 1
Comparison of thermal inactivation parameters for PEPTR and Cl. pasteu-
rianum in acidified papaya pulp (pH 3.B)

Temp. D (min) z (OC) Ea(Kcal/mol)

°C C. Pasteur. PEPTR C. Pasteur. PEPTR C. Pasteur. PEPTR
75 9.7 16.7
77 7.1 7.2
80 2.7 3.7 8.8 7.B 64.7 72.7
 497
2.0
1 - 75°C 1 - 75°C
2 77°C 2 77°C
3 - 80°C ~.,

1.5

1.0
1.0
a::- 1-
Q.
~ ~
~ ~
c:J c:J
0
...J 0
...J 0.0
0.5

0.0

-D.5
~*O+-~~~--r-,--.._........,~~-
0 2 4 6 8 10 12 14 0.0 0.5 1.0 1.5 2.0 2.5 3.0
HEATING TIME (MIN) HEATING TIME (MIN)

Fig. l. Thermal inactivation of Fig. 2. Modified thermal inactivatkn

PE in APP (pH = 3.8). curves for the PEPTL.
2.0
1 PEPTR CONCLUSIONS
PEPTL
Results showed that PEPTR has greater
1.0
D values than C1. pasteurianum at the
temperatures tested (Table 1). Based
0
on the kinetics parameters of PEPTR
c:J
0 an F768 = 4.5 min. may be used to
inac~1vate the enzyme considering
...J

0.0 1. 22 D of PEPTR.

REFERENCES

·1.0+-_.._-...-........,,.......__-_-....:~ 1. Nath, N. and Ranganna, S. Determin

70 72 74 76 78 80 82 8 ation of thermal process schedu1e-
TEMPERATURE (C)
of acidified papaya. J. Fd. Sci.
1981, 46, 201-206. -- ---

Fig. 3. Phanton curve for PE in 2. Ling, A.C. and Lund, D.B. Deter-
APP (pH = 3.8). mining kinetic parameters for
thermal inactivation of heat
labile isoenzymes from thermal
destruction curves. -J. --
Fd. ---
Sci.
1978, 43, 1307-1310.
 HEAT GELATION AND REACTION KINETICS OF WHEY PROTEIN
CONCENTRATES

S.M. TAYLOR. L.F. GLADDEN and PJ. FRYER

Department of Chemical Engineering.
University of Cambridge. Pembroke St..
Cambridge CB2 3RA. United Kingdom.

ABSTRACT
Experimental data for the gelation of whey protein concentrate solutions were monitored
between 63 and 85°C using a Rheometrics dynamic spectrometer. The data were compared
with a model for gelation times based on percolation theory. The technique allows selection of
a fundamental rate law; subsequent gelation is modelled stochastically.
The rate-limiting reaction step in gel formation. and the rate constant of this step were
identified for each experimental condition. The reaction mechanism changes with temperature
and pH:
(i) The rate-limiting step for pH 7.0 solutions for the whole range 63-85 °C was the
addition of denatured protein molecules to an aggregate of denatured molecules. The rate
constant for this step followed Arrhenius temperature kinetics, with a change in the activation
energy at about 73°C.
(ii) The rate-limiting reaction step at pH 5.2 (the isoelectric pH range) changed with
temperature. The two steps which were identified as rate-limiting were the denaturation of
native protein molecules (75-85 0C). and the aggregation of two denatured molecules to form
a 'dimer' (65-73 DC).

INTRODUCTION
Heat gelation of whey protein concentrates allows desirable texture properties to be introduced
to foods. It is important to understand the processes leading to gelation and to be able to
predict the time required for gelation. Here the kinetics of the reactions leading to heat
gelation of whey protein concentrates are studied at different temperatures and solution pH.
Previous models fit integrated rate equations to data [1-2]; this model applies Monte Carlo
techniques to identify a controlling mechanism for gelation and calculate the rate constant of
the rate-limiting reaction.
MA TERIALS AND METHODS
Solutions of a commercial whey protein concentrate (Carbery Milk Products, Ballineen, County
Cork, Eire) containing 75 %(w/w) whey protein were made by dissolving powder in reverse
osmosis water to concentrations of 8-30 %(w/w) protein. Solution pH was adjusted to 5.2 or
7.0 by adding 1 M NaOH or 1 M HCI. The solutions were left stirring for at least 90 minutes to
equilibrate.
Whey protein solutions were then placed in the 25 mm radius cone and plate test fixture
of a Rheometries RDS II dynamic spectrometer (Rheometries Inc., Union, N.J., U.S.A.) and
 499
gelled in situ at temperatures of 63-85 dc. The development of the rheology of the forming
gel was followed usinp, the rheometer operating in dynamic mode; the test conditions used (lyI
= 0.5%, co = 1 rad.s- ) do not interfere with the development of the gel structure [3]. The
gelation time was taken as the time for the storage modulus to start increasing above the noise
level.
Percolation analysis of gelation times
Full details of the mathematical model are given by Steventon et al. [3]. The same approach
was used to analyse the concentration dependence of the gelation time at each temperature and
pH. Heat gelation of globular proteins has been described as two-stage [4]: denaturation of
native protein, followed by aggregation of denatured monomers, eventually leads to gelation.
Three stages of this process have been used as possible reaction mechanisms:

Kinetic step Controlling process

n ~ d denaturation of native protein (l)
2d ~ d2 initiation of aggregate (2)
d + dx ~ dx+l propagation of aggregate (3)
where n is native protein, d is denatured protein, and dx is an aggregate of x denatured protein
molecules.
The best-fitting model was chosen as that giving the smallest difference between
experimental and predicted gelation times.

RESULTS
At pH 7.0, the best fit of the model to the gelation time data over the whole temperatule range
was scheme 3. Figure 1 shows an Arrhenius plot of the rate constant of the rate-limiting step
calculated from the percolation model. Although the mechanism of the rate-controlling step
does not change, a change in the rate-limiting reaction occurs at 73°C, demonstrated by the
substantial change in activation energy.

-3.0
123 kJ/mal
-3.5
-4.0
-4.5
:>i'
:s -5.0
-5.5
-6.0
220kJ/moI
-6.5
-7.0
0.00275 0.00280 0.00285 0.00290 0.00295 0.00300
Irr (11K)
Figure 1. Arrhenius plot of the rate constants (units of (%(w/w).s)-l) for gelation at pH 7.0.

Percolation analysis of the pH 5.2 gelation time data showed more complex behaviour in
that the controlling mechanism changed with temperature. At 63°C the gelation process was
propagation-limited (scheme 3); the initiation step (scheme 2) was rate-limiting in the
tem~rature range 65-73 DC; denaturation (scheme 1) was rate-limiting at temperatures above
73 C. Figure 2 shows Arrhenius behaviour for the pH 5.2 gelation data in both temperature
ranges.
DISCUSSION

Changes in the activation energy of whey proteins have been noted at other temperatures [5-8],
and it has been suggested that this reflects a change from denaturation- to aggregation-limited
reaction. However, the gelation of whey protein at pH 7.0 is propagation-limited over the
whole temperature range. The observed change in activation energy must be due to another
reason. Interactions between denatured protein molecules are known to be complex, involving
500
electrostatic interactions, disulphide bonds, van der Waals forces, dipole-dipole and
hydrophobic interactions [9]; it may be that the change in rate-limiting reaction at 73 QC
results from a change in the relative formation rates of these interactions.

\-
-4.0
-4.5
-5.0
-5.5
:; -6.0
:5
-6.5
• £

185kJ/~
-7.0
-7.5
-8.0 '----'----'-----'-----'-----'
0.00275 0.00280 0.00285 0.00290 0.00295 0.00300
Iff (11K)

Figure 2. Arrhenius plot of the rate constants for gelation at pH 5.2. Rate constants are for
denaturation limited ( • , units of s-1). initiation limited ( Il , units of (%(w/w).s)-1) and
propagation limited ( .. , units of (%(w/w).sy1) gelation.
The percolation analysis of pH 5.2 gelation times demonstrates that gelation is
denaturation-limited at high temperatures (75-85 QC), while it is aggregation-limited at low
temperatures (63-73 QC). The change in slope of the Arrhenius plot (Figure 2) at 80-85 QC is
not accompanied by a change in reaction mechanism (still denaturation-limited). This change
in slope is possibly due to a change in rate-limiting reaction as discussed above for pH 7.0
gelation.
The identification of changes in reaction mechanism with temperature and pH highlights
the power of the application of a Monte Carlo approach to models of gelation.
ACKNOWLEDGEMENTS
Financial support for this work was provided by the Cambridge Commowealth Trust and
Express Foods (Europe). Dr. M.R Mackley is thanked for making the Rheometrics available.
The technical assistance of Mr. RT.J. Marshall and Mrs. H.F. Yao are gratefully acknowledged.

REFERENCES
I. oakenfull , D. and Scott, A., New approaches to the investigation of food gels. In ~
and Stabiliser for the Food IndustIy 3, eds. G.O. Phillips, D.J. Wedlock and P.A. Williams,
Elsevier, London, pp 465-475.
2. Ross-Murphy, S.B., Concentration dependence of gelation time. In Polymer. Gels and
Colloids, ed. E. DiCkinson, RSC Special Publ. no. 82, London, 1991, pp. 357-368.
3. Steventon, A.J., Gladden, L.F. and Fryer, PJ., A percolation analysis of the concentration
dependence of the gelation of whey protein concentrates. J. Texture Studies, 1991, 22,
201-218.
4. Ferry, J.D., Protein Gels. Ady. Protein Chern., 1948,4, 1-78.
5. Lyster, R.LJ., The denaturation of a-lactalbumin and ~-lactoglobulin in heated milk. 1..
Dairy Res., 1970, 37, 233-243.
6. Hillier, RM. and Lyster, RL.J., Whey protein denaturation in skim milk and cheese whey.
J. Dairy Res., 1979,46,95-102.
7. Manji, B. and Kakuda, Y., Thermal denaturation of whey proteins in skim milk. Can. Inst.
Food Sc. Technol. J., 1986, 19, 163-166.
8. Dannenberg, F. and Kessler, H.G., Reaction kinetics of the denaturation of whey proteins
in milk. J. Food Sc., 1988, 53, 258-263.
9. Koning, M.M.G. and Visser, J., Protein interactions. An overview. In Protein Interactions,
ed. J. Visser, VCH, Weinheim, Germany, 1992, pp. 1-24.
 EFFECT OF MICROWAVE HEATING ON NON-ENZYMATIC BROWNING IN
BANANA PRODUCTS

M.PlLAR CANO & VICTORIA LIZARRAGA

Freezing of Fruit and Vegetable Products Department
Instituto del Frio (C.S.I.C.)
Ciudad Universitaria, 28040 Madrid, Spain

ABSTRACT

The effects of thermal pretreatment by microwave on colour

deterioration of banana slices are described. Thermal treatments
are necessary to improve the quality of frozen products due to
enzymic inactivation. Microwave heating produce a high efficient
inactivation of polyphenoloxidase and peroxidase enzymes
(soluble and ionic forms), but it promoves non-enzymatic
reactions that slight modifie the product colour depending on
banana maturity. Selection of fruit maturity and conditions
(power and time) of microwave treatment could improve the
preservation of banana products without the use of any chemical
additives.

INTRODUCTION

Microwave blanching operations have been studied and failed

advantage when compared to conventional steam blanching in the
preservation of several crops as tomato [1] and corn-on-the cob
[2]. Bananas undergo rapid browning as a result of tissue
disruption and exposure to oxigen during the peeling and slicing
operations prior to futher processing. The successful prevention
of browning by enzyme inactivation by sodium bisulphite has been
widely discussed [3], but blanching for enzyme inactivation of
whole fruits or slices has been only reported by Cano et al [4].
This work is an extensive investigation of the effect of
different conditions of microwave treatment along the enzymatic
inactivation and the development of non-enzymatic browning in
processed banana products.

MATERIALS AND METHODS

Fruit ripeninq
Green bananas (Musa cavendishii, var. Enana) from Canary Islands
were stored at 14°C and 85-90% relative humidity for 1 month.
502
Fruits for processing were chosen at 0 (green), 7 (green-
yellow), 10 (yellow-green), 14 (Yellow) and 21 (yellow with
spots) days of storage from among 12 bunches, and were selected
by their similar location in the bunches.
Microwave treatment
Ten fruits were peeled, sliced and heated in a microwave oven at
different conditions: powers 200 W, 475 Wand 750 W for 30 s,
and power 380 W for 15 s, 30 sand 45 s. After treatment the
slices were packed in polyethylene bags and cooled in ice-tap
water for 5 min. Samples were freeze-dried and stored at -24°C
before analysis.

Chemical and biochemical analysis

Estimation of non-enzymatic browning (alcohol soluble color
(ASC) index was carried out by alcoholic extracts and
spectrophotometric readings at 420 nm. Peroxidase (POD) and
polyphenoloxidase (PPO) activities were determined by
spectrophometric method [4). Protein was determined by the Bio-
Rad method described by Bradford (1976) [5).
Data analysis
The study was carried out twice and the mean of the values
obtained for three determinations are reported in Results.
Analysis for variance and statistical significance was made by
t-test.

RESULTS AND DISCUSSION

The effectiviness of microwave heating to inactivate POD and POD

PPO activity (u/min/mg protein)

250

200

150

100

SO
o
Orec. One.-yellow Yellow-lree. Yellow Yellow wltla .pot.

Banana peel color

~ 45 • D 30 • _ 15 • _ Control
Figure 1. Effect of microwave heating on PPO activity of banana.
was depending of fruit maturity. Figure 1 shows the specific PPO
activity towards ripening of banana and the remainded activity
after microwave treatment. Green-yellow fruits (7 days of
storage) were most suitable for processing, attending the
 503

greater thermal inactivation and the smaller non-browning index

obtained (Figure 2).
Non-enzymatic browning (ASe)
8r-----~----------~--~------------------~

7 'Wallow with spots

Yellow-green
8
--. .... _-""'il.
Yellow

3 Green-yellow
2 Green

OL-~------~------~------~-------L------~
30 45 eo 75
Time (8)

Figure 2. Non-enzymatic browning index of microwave treated

banana slices.

ASC index increased directly throught storage time and fruit

maturity due to the greater sugar content available to Maillard
reactions forming non desirable brown pigments.
Enzymatic inactivation by microwave heating at different
powers and same time (30 s) did not show coherent results to the
above exposed. More studies about the mechanisms of microwave
blanching in vegetable tissues must be take account futher.
REFERENCES
1. Porreta, S. and Leoni, C., Preparation of high-quality tomato
products unsing enzyme inactivation by microwave heating. In
Engineering and Food, ed. W.E.L. Spiess and H. Shubert, vol.
2, Elesevier Applied Science Publishers, London, 1989, pp.
251-258.
2. Huxsoll, C.C., Dietrich, W.C. and Morgan, A.I, Jr. Comparison
of microwave with steam or water blanching of corn-on-the-
cob. I. Characteristics of equipment and heat penetration. Food
Technol., 1970, 43(1), 290-292.
3. Garcia, R. Arriola, M.C., Porres, E. and Rolz, C. Process of
banana puree preservation at rural level. Lebesm. wiss.
Technol., 1985, 18, 323-327.
4. Cano, M.P., Marin, M.A and Fuster, C. Freezing of banana
slices. Influence of maturity level prior to freezing. ~
Food Sci., 1990, 55(4), 1070-1072.
5. Bradford, H.A. A rapid and sensitive method for the
quantitation of microgram quantities of protein utilizing the
principle of protein-dye binding. Anal. Biochem., 1976, 72,
248-254.
ACKNOWLEDGMENT
This work was supported by ALI91-0621 CICyT Project.
 EFFECf OF MICROWAVE HEATING RATE ON MALTOSE PRODUCTION
IN SWEETPOTATO

TSUYOSHI NAKAI, JUN SAWAI, ATSUSHI HASHIMOTO*,

TAIJIROU HONDA, and MASARU SHIMIZU
Department of Chemical Engineering, Faculty of Technology, Tokyo University of
Agriculture & Technology, 2-24-16 Nakamachi, Koganei, Tokyo 184, Japan
*Department of Bioproduction & Machinery, Faculty of Bioresources, Mie
University, 1515 Kamihama-cho, Tsu 514, Japan

ABSTRACf

To control maltose production in sweetpotato by the microwave heating,

simulation of maltose production was investigated. Maltose is produced from
starch by the heat treatment and several kinds of enzymes(i.e. J3-amylase) in
sweetpotato. The apparent reaction rate constant k of maltose production was
determined using thin-plate sweetpotato. The frequency factor A and the
activation energy E in the Arrhenius equation were obtained. The experimental
results of maltose production in sweetpotato by the microwave heating were
compared with the calculated results that were obtained by using the k and the
experimental transient temperature. The experimental results were agreed with
the calculated results approximately. In this study, the possibility of
controlling the maltose production by the microwave heating was presented.

INTRODUCTION

Recently, in order to satisfy various needs of consumer, fundamental study of

food processing control has been expected. Thermal operation in the food
process industry is one of important operations. Using the electromagnetic wave
heating, which uses microwave, far infrared and so on, thermal operation may be
able to be controlled easily. Hence, control heating with the electromagnetic
wave radiation have been expected in food processing field.
In the past, simulation of potato softening during non-isothermal and
isothemal processes[l] and prediction of optimum cooking time of vegetables
using the softening rate constant[2] had been studied. However, simulation of
the change of component caused during heat treatment has not been studied
enough.
In this study, we investigated how to control maltose production in
sweetpotato by the microwave heating. Maltose is produced through following
process by heating. Raw starch in sweetpotato, first of all, is gelatinized,
 505
and then the gelatinized starch with a-amylase is changed to maltose, which is
main component of sweet taste. The apparent reaction rate constant of maltose
production from raw starch is examined about sweetpotato. The effect of
microwave heating rate on maltose production in sweetpotato is investigated.
Furthermore, the rate constant is applied to simulation of maltose production
in sweetpotato by the microwave heating.

MATERIALS & MEIHODS

Maltose production in sweetpotato under isot1Jermal condition

Sweetpotato(beniazuma) was prepared as the thin-plate. The thickness of
thin-plate is below 1mm. This sample was heated at a specific temperature(338,
341, 343, 348, 353, 356, 358K) for a specific time(60, 120, 180, 360, 6OOsec)
in a water bath. After heating, the sample was milled and added water.
Centrifugation was carried out, and then maltose content in supernatant was
measured by the Somogyi-Nelson method. Produced maltose content was determined
by subtracting maltose content in unheated sweetpotato from that in heated
sweetpotato.

Maltose production in sweetpotato by microwave Deating

Sweetpotato(beniazuma) was prepared to as rectangular prism (lOem long, 3em
wide and 3cm high). The sample was irradiated by microwave(2.45GHz) from only
one direction at 50W microwave power. Temperatures at the distance of 2.0, 3.5,
5.0, 6.5, 8.0em from irradiated surface were measured everyone minute.
Produced maltose contents were measured, too.

RESULTS & DISCUSSION

7JJe apparent reaction rate constant ofmaltose production

For the maltose production under the isothermal condition, the apparent
reaction rate constant of maltose production is assumed to be expressed
mathematically as the following equation.

d~ =k(1.056S--M) (1)

M is the produced maltose content, t is the heating time and S is the initial
starch content. And, k is the apparent reaction rate constant. Coefficient
1.056 is the conversion factor to convert starch content into maltose content.
Figure 1 shows the experimental results of k. The frequency factor A and the
activation energy E are obtained by using the following equation.

k= Aexp( RET-) (2)

Simulation ofmaltose production in sweetpotato by microwave Deating

Fig.2 shows the experimental transient temperature in sweetpotato at the
distance of 3.5 and 5em from irradiated surface. The apparent reaction rate
constant is obtained by using Eq.(2) at any temperature in the region of
338-358K, because maltose is little produced below 338K and above 358K.
Produced maltose content per unit time is obtained by using the following
equation.
506

(3)

The experimental results of maltose production are compared with the

calculation results in Fig.3. The experimental results are agreed with the
calculation results approximately. Produced maltose content at the distance of
5cm is higher than that at the distance of 3.5cm from irradiated surface. Since
maltose is produced actively in the region of 338-358K, this indicates that the
difference of maltose production depend on temperature course in that region.
In order to increase produced maltose content, it is thinkable that
sweetpotato must be heated more slowly in the region of temperature 338-358K,
particularly near 353K. Furthermore, from simulation results, it may be
concluded that the production of maltose in sweetpotato can be controlled by
adjustment of heating rate.
373 4-4-.---
/-
!.~!
jm!
2

•
-6

•• 273 '-----'-_--L.._-'------"'-----'-_-'
o 3 5 6
~ -7
Time [mi nJ
Fig.2 Transient tempe-rature ill
sweetpotato by the microwave
heatillg
-8
1001r--.-........-.....,---.--,.----,

-o

•
-9

-10 L-_'--_'--~~..........JL..----'--
i' 50
S
..§.
•
--·4------
2.7 2.8 2.9 3.0 4

liT x 10 3 [K . 1] O__ -~~~-L-~-~~

o 3 5 6
Fig.1 Effect of heating temperature
Tim e [m i nJ
on the apparent reaction rate constant
Fig.3 Comparison of produced
maltose cOlltent between experimental
results and calculated results

REFERENCES

1.0kazaki, T., Suzuki, K., Maeshige, S. and Kubota, K., Simulation of potato
softening during non-isothermal and isothermal processes., Nippon Shokuhin
Kogyo Gakkaisi, 1992,39,295-301.
2.Matsuura, Y., Kasai, M., Hatae, K. and Shimada, A., Prediction of optimum
cooking time of vegetables., Nippon Shokuhin Kogyo Gakkaisi, 1989, 36, 97-102.
 STUDIES ON THE MICROWAVE HEATING-RATE EQUATIONS OF FOODS

KIYOSHI KUBOTA, MASAYUKI KUROKAWA, HIDEKI ARAKI,

TAKASHI OKAZAKI, LIANG TONG LU AND YOSHIO HAGURA
Faculty of Applied Biological Science, Hiroshima University
1-4-4, Kagamiyama, Higashi-Hiroshima, 724 JAPAN

ABSTRACT
The microwave heating-rate equations in foods are very
complicated for reason of the microwave heating is affected by
water, component, size, shape, temperature and other factors.
Therefore we must set up the approximating over-all heating-
rate equations for microwave heating. In this study, we
intend to report on the microwave heating-rate equations based
on the simple types for cooking and drying of foods. For the
microwave cooking of potato, the value of the rate constant is
higher than the value obtained for the cooking in hot water.

INTRODUCTION
Microwave techniques can provide several unique advantages
when compared with conventional food processing methods. A
microwave oven is now a common household appliance. In this
study, we report on the microwave cooking- and drying-rate
equations based on the empirical types, and compare the
previous data obtained on the hot water dipped cooking.

EXPERIMENTAL
As samples, we used potato 'and cooked udon wheat flour
noodle). The shapes and sizes are shown in Tables 1 and 2.
These samples were put into an electronic oven. Potato sample
was loosely bundled and cooked in a package made of poly-
vinylidene chloride which prevented to lose moisture vapor.
The degrees of cooking of potatoes were measured by a rheo-
logical method, and the degrees of drying of potatoes and
noodles were calculated by measuring the loss of sample
weight. The temperatures were assumed or measured by using
some thermolabels.
508
MICROWAVE HEATING-RATE EQUATIONS

Normalizin
The cooking as follows.

x = I (Yo - y) / (Yo - y.) (1)

where, y is the degree of cooking or drying, and subscripts 0
and e show the initial and equilibrium states.

Empirical rate equations

For the microwave heating-rate equatiuons, we used the
following empirical rate equations.

n-th order rate equation:

dx/ dO = k (1 - x) n (2)

S-shape rate equation:

dx/ dO = k (1 - x) n (x + p) (3)

where, k, nand P are the rate parameters which can be

obtained from the relationships of a n-th order curve or a S-
shape curve between the normalizing the degree of cooking
/ drying x [-I and the time 0 [minI. For the cooking, k can be
described by an Arrhenius equation.
k = A exp [ -E/ [R g (t + 273.2) I (4)
where, t[Ocl is temperature and Rg =8.314J/(moloK). A and E
are the frequency factor and the activation energy which can
be obtained from the data.

RESULTS AND DISCUSSION

The relationships between the cooking or drying ratio x[-I and

the time 8 [minI were obtained, and the rate parameters in rate
equations were calculated from the data. The values of rate
parameters are shown in Table 1 and 2. The S-shape rate
equation is useful for describing complicated boiling process
of potato, and for the change of temperature during unsteady-
state microwave drying of foods. The values of cooking rate
for microwave cooking (Table 1) are very much higher than
those for hot water cooking.

CONCLUSION

The microwave heating-rate equations based on the empirical

types such as n-th order and S-shape models were useful for
the design of microwave apparatuses, and for the comparison of
cooking rate in different heating mode.
 509

TABLE 1
Rate parameters for cooking-rate equations of potatoes [1]

Heating method n [-I p [-I A [l/s] E [kJ/mol]

Hot water 1.0 O. 1 2.10X10 15 124.3

1.0 1.32X 10 15 124.7
Microwave 1.0 1.52Xl0 9 76.5
1 .5 2.0 9.92Xl0 1O 91.0

Samples for hot water: spherical potato ( diameter= 1 .9cm );

for microwave: cylindrical potatoes ( diameter=3.2cm,
thickness=0.5cm X 3 numbers)

TABLE 2
Rate parameters for drying-rate equations of foods [2,3]

Sample and Heating range n [-I p [-] k [1 /s]

Cooked udon:
Temp.)20°C (Time)Omin) 1.0 0.1 1.68Xl0- 3
Temp.)100°C(Time)assumed) 1 .0 7.85xl0- 4
Potato:
Temp.)20°C (Time)Omin) 1 .0 0.50 3.40Xl0- 3
Temp.)100°C(Time)2min) 1.0 3.68Xl0- 3

Samples: cooked udon ( diameter=O. 2-0. 4cm, length=35cm X 60

numbers ); cylindrical potato ( diameter=3.0cm, thickness
=0.5cm X 3 numbers); Microwave power = 180W.

REFERENCES

1. Kubota, K., Kurokawa, M., Suzuki, K. and Esaka, M., Study on

the Microwave Cooking-rate Equation of Cylindrical Potato.
J. of Japanese Soc. of Food Sci. and Tech., 1988, ~:78-82.
2. Kubota,K., LU,L.T., Yamashita,Y., Okazaki,T., Mochizuki,H.,
Kurokawa,M., Suzuki,K. and Esaka,M., Studies on Simple
Convenient Microwave Heated Drying Instrument and on
Microwave Drying of Potato. J.Fac.Appl.Biol.Sci., Hiroshima
Univ., 1990, 1.2.:51-62.
3. Ji,D.H., Kubota,K., Zhang G. and Hagura,Y., Studies among
the Temperature and Drying-rate on Microwave Heated Drying
of Discus Potato. J.Fac.Appl.Biol.Sci.,Hiroshima Univ.,
1992, 11:121-126.
 STUDIES ON THE COOKING-RATE EQUATIONS OF FOODS

KIYOSHI KUBOTA, YORIMASA MASUMOTO, GE ZHANG AND YOSHIO HAGURA

Faculty of Applied Biological Science, Hiroshima University
1-4-4, Kagamiyama, Higashi-Hiroshima, 724 JAPAN

ABSTRACT
In this study, we report on the cooking-rate equations based
on the empirical types by setting up a convenient cooking
ratio. The degrees of cooking were determined by measuring a
weight or rheological parameters. From the results with
various foods, it was concluded that the empirical overall
cooking-rate equations based on an n-th order and an S-shape
models are very useful for the comparison of the cooking rate
of foodstuffs and for the design of cooking process.

INTRODUCTION
The changes in foods brought about by processing have been
expressed with an n-th order rate equation. However, for many
foodstuffs the changes are so complicated that the changes can
not always be described by a simple n-th order rate equation.
We will report on the cooking-rate equations based on the
empirical types and the soaking-shell model by setting up a
convenient cooking ratio.

COOKING-RATE EQUATIONS
Cooking ratio
The extent (y) of cooking of foodstuffs can be empiricaly
expressed as the change of a simple property such as mass or
rheological property. Most of the cooking have two boundary
values Yo and y., the initial and equilibrium states,
respectively. The cooking ratio x [-) can be expressed with
the following equation.

x = I (Yo - y) / (Yo - Y.) I (1)

Empirical rate equations

Most of the relationships of cooking ratio x [-) vs. cooking
time 8 [min) show a monotonous smooth n-th order curve or an S-
 511
shape curve. The empirical rate equations [1 ,21 are as
follows;

n-th order rate equation:

dx/ dO = k (1 - x) (2 )

S-shape rate equation:

dx/ dO = k (1 x) n (x + P) (3 )

where, k, nand p are rate parameters For the reaction type

transformation such as cooking, the value of k can be
expressed by using the following Arrhenius equation,

k = A exp{ -E/ [R g (t + 273.2) I (4 )

where, t [Ocl is temperature and R g =8.314J/ (mol-K) A and E are

the frequency factor and the activation energy, respectively ..

Soaking-shell model rate equations

As a simple model for the cooking of rice and soybean, we have
proposed a soaking-shell model for spherical, cylindrical, and
infinite materials [21 which has an unsoaking-core and a
soaking-shell zone, and their respective moisture contents are
the initial and equilibrium moisture contents, respectively.
For a spherical material, we have proposed a rate equation as
follows:

dx/d9

where, (1 - x) 1/3 Ro (6 )

R [ (R. /Ro ) 3 -11 r c 3 I 1/3 (7)

where, R [cml is radius, rc [cml is the radius of unsoaking
core, C[g-H20/m31 is the moisture content, and the subscripts
o and e respect the initial and equilibrium states.
k m [m 2 /minl and kr [m/minl are the rate parameters which can be
obtained from the data. These show the parameters with
respect to the diffusion in soaking-shell part and the soaking
at the surface of the soaking core, respectively.

RESULTS AND DISCUSSION

The degrees of cooking for beans and noodles were determined

by measuring the weights, and the degrees for potatoes, radish
and carrot were determined by measuring the rheological
parameters. The empirical rate equations based on the n-th
order and S-shape models were applied to the cooking of
various foodstuffs. The values of rate parameters are shown
in Table 1. These simple equations were useful for the
comparison of the cooking rate of foodstuffs. The cooking-
rate equations based on the soaking-shell model were used for
the cooking of rice and soybean. To simplify the analysis, we
assumed that the shapes are sphere, although in fact the rice
is shaped as a flattened-oval and the soybean is two hemi-
spheres. For many foodstuffs the shapes are so complicated
512
and the soaking phenomena are so complicated that we can not
use this model for the comparison of cooking rate of various
foodstuffs.

CONCLUSION

The cooking-rate equations based on empirical types, sUch as

the n-th order and S-shape models, were useful for the
comparison of the cooking rate of foodstuffs as shown in Table
1. These simple cooking-rate equations will be useful for
the design of cooking process.

TABLE
Rate parameters for cooking-rate equation of foods
(Reference: from papers related Kubota)

Material n [-] p [-] A [l/s] E [kJ/mol]

Raw rice 1.0 5.24X108 79.5
1 .0 5.49X10 2 36.8
75-100, 110-150°C; viscoelastic method )
Steam treated rice 1.0 2.49Xl04 o
( 20-100°C; weighing method
Soybean 1.0 23.2 30.6
Azuki (red bean) 1.0 0.01 1.58X10 3 40.6
( 20-100°C; weighing method
Potato slice 1.0 0.1 1.66Xl0 15 124
Sweet potato slice 1.0 0.1 1.22X10 18 143
Radish slice 2.5 0.02 1.07Xl0 16 128
Carrot slice 1.5 2.82Xl0 16 135
( 70-100°C; impact penetration method )
Spaghetti 2.0 85.2 36.5
Udon (wheat flour noodle) 2.0 6.32x10 4 54.9
Soba (buckwheat noodle) 2.0 4.47X10 3 45.7
( 70-100°C; weighing method

y. for steam treated rice: Expressed as function of

temperature; y. for spaghetti, udon, and soba: Treated
as a rate parameter because it could not be obtained.

REFERENCES
1. Kubota,K., Determination of the Empirical Rate Equation for
the Chemical and Physical Transformations of Foods. J. Fac.
Appl.Biol.Sci.,Hiroshima Univ., 1979, ~:11-30.

2. Kubota,K., Kinetics of Transforming and Cooking of Foods.

In Food Reaction Engineering (Japanese), ed. T.Yano and R.
Toei, Korin, Tokyo, 1990, pp.1-78.
 KINETIC MODEllING OF CInCKEN MUSCLE THERMAL
CONDUCI'IVlTY DURING DEEP-FAT FRYlNG

M. O. NGADI AND L R. CORREIA

Department of Agricultural Engineering
Technical University of Nova Scotia
Halifax, Nova Scotia, Canada B3J 2X4

ABSTRACf

The line heat source method was employed to determine thermal conductivity of
chicken muscle. Product temperature and frying time data were obtained during deep-
fat frying. Nonlinear regression was used to estimate kinetic parameters. Computed
Gibbs activation energy changes were positive, estimated entropy and enthalpy
activation changes were negative, and reaction order estimates were close to zero. Both
frying oil and product temperatures significantly affected Gibbs activation energy
change, entropy and enthalpy activation energy changes, whereas only product
temperature significantly affected reaction order.

INTRODUCTION

Undercooking causes unsafe meat and poultry products. Overcooking uses more energy
and may result in lower nutritional value. Hence, knowledge of thermal conductivity is
required to design suitable frying processes. The objectives of this study were to
estimate kinetic parameters of thermal conductivity changes of chicken muscle during
deep-fat or immersion frying, and to determine the effects of frying oil and product
temperatures on kinetic parameters.

MATERIALS AND MB'IHODS

The line heat source method was used to determine thermal conductivity of chicken
drum muscle slices (1). To avoid heat loss through moisture vaporization, each slice
was placed in an enclosed sample holder and heated in an air oven at room
temperature and at temperatures of 30, 50, 70, 90 and 1100c' Chicken slices were fried
at oil temperatures of 120, 140, 160 and 1800C using a home fryer (Type 3617M, Tefal).
514
Thermal conductivity changes in chicken drum muscle slices were modelled using
the reaction kinetics approach and Eyring's absolute reaction rate theory (2) using the
following equation:

dPr. dT = (Prm - Pr j.KT.exnf AS _ AH).(1 _ X) II (1)

dTdt h "'tR RT

where Pr is thermal conductivity (W/(m.K)), T is product temperature (K), t is frying

time (s), subscripts i and m denote initial and maximum values of Pr, K is the
Boltzmann constant, h is the Planck constant, AS is the entropy activation change
(kJ/(kg.mole.K)), AH is the enthalpy activation change (kJ/(kg.mole)), R is the gas
constant, X is the degree of cooking, and n is reaction order. Nonlinear regression
analysis was used to estimate kinetic parameters AS, AH and n between the product
temperatures of 10 and 90"C, at 100C intervals. Gibbs activation energy changes (AG
= AH - T.AS) were computed at the midpoints of product temperature intervals.
Statistical analysis of data were conducted with SAS software (SAS Institute Inc, Cary,
NC, USA) on an HP 9000 835S mainframe computer.

RESULTS AND DISCUSSION

Observed thermal conductivity of chicken drum muscle increased from 0.527 to 0.558
W/(m.oC) corresponding to the product temperatures 5.33 and 104.78°C. Thermal
conductivity was modelled by regression equation (2).
Pr = O.593(1-exp(-7.74E-3 7) (2)

where asymptotic standard errors were less than 9 % of estimates. Product temperature
of chicken muscle was fitted as a function of frying time by a third-order polynomial
equation (R2 = 0.9993, P > F = 0.0001).

Using observed data of T and t, and computed values of dPr/dT, dT/dt and X,
kinetic parameters were estimated. Computed AG values ranged from 74541 to 98174
kJ/(kg.mole), AS estimates ranged from -393 to -250 kJ/(kg.mole.K), AH estimates
ranged from -45101 to 3160 kJ/(kg.mole) and n ranged from -5.44E-3 to 4.90E-2.
Positive AG values imply that energy was required to increase thermal conductivity.
Negative AS estimates were attributed to protein-protein aggregation, a more ordered
state thereby lowering entropy, and loss of higher entropy free water, contributed
partially by reduced water-holding capacity during denaturation. Almost all AH
estimates were negative signifying exothermic reactions, ascribed to moisture loss
incorporating sensible and latent heat.

An analysis of variance, at the 5 % level, revealed that frying oil and product
temperature significantly affected AG, AS and AH (Table 1), whereas only product
temperature affected reaction order. As product temperature increased from intervals
10 to 20"C, to 80 to 90"C, mean AG increased from 75713 to 96315 kJ/(kg.mole),
whereas mean AS decreased from -283 to -365 kJ/(kg.mole.K) and mean AH decreased
from -5870 to -34410 kJ/(kg.mole). The reaction order (2.42E-2) between 80 and 900C
 515
was significantly higher than its corresponding mean value (4.18E-3) between 10 and
800C product temperature intervals. Estimates of as and aH were significantly
correlated (P < 0.0001, correlation coefficient = 0.993).

TABLE 1

Mean values of kinetic parameters at various frying oil temperatures

Frying oil Kinetic parameters +

temperature

aG as aH n
kJ/(kg.mole) kJ/(kg.mole.K) kJ/(kg.mole)

120 86791 a++ - 331 b - 20710 b 4.77E-3 b

140 85863 b - 306 a - 13625 a 1.03E-2 a
160 85487 b - 310 a - 15240 a 6.62E-3 ab
180 84440 c - 306 a - 14761 a 5.01E-3 b

+ mean of 16 observations
++ means in the same column with identical letters are not significantly different at
the 5 % level by Duncan's multiple range test

CONCLUSIONS

The reaction kinetics approach and Eyring's absolute reaction rate theory suitably
modelled entropy and enthalpy activation changes, and reaction order. Computed
Gibbs activation energy changes were positive, estimated entropy and enthalpy
activation changes were negative, and reaction order estimates were close to zero. Both
frying oil and product temperatures significantly affected Gibbs activation energy change
values, estimated entropy and enthalpy activation energy changes, whereas only product
temperature significantly affected reaction order estimates.

REFERENCES

1. Sweat, V.M., Haugh, e.G. and Stadelman, W.J., Thermal conductivity of chicken
meat at temperatures between -75 and 200e. L Food Sci.. 1973, 38: 158-160.

2. Correia, LR. and Mittal, G.S., Stress relaxation kinetics of meat batters
containing various fillers during smokehouse cooking. J. Text. Stud. 1990, 21: 75-
96.
 COLD EXTRUSION-TRIBOCHEMISTRY

SHAW S. WANG, XIAOGE ZhENG, CHI T. HO and DI QU

Department of Chemical and Biochemical Engineering
Rutgers University, Piscataway, NJ 08855-0909, U. S. A.

ABSTRACT

Extrusion of starch in a single screw extruder and a capillary rheometer were studied at temperatures below
50°C to promote cooking of starch by mechanical (shear) energy input. Results show that at shear stress
higher than 106 dyne/cm2, starch can be cooked, according to DSC (Differential Scanning Calorimeter)
measurements, to a degree close to 80% depending on the residence time of the material in the extruder. We
attribute this to mechanical (shear) energy, instead of thermal energy, initiated cooking. These extrudates
were not puffed, in addition to being cooked, the starch granules in the extrudate were also broken into
smaller particles. When methionine (1 or 2% w/w) was mixed with starch and extruded, methional,
propanethiol, dimethyl disulfide, dimethyl trisulfide and others were found in the extrudate which has a
typical cooked potato flavor. A tribochemical mechanism involving free radical is proposed for the reactions
leading to these compounds from methionine.

INTRODUCTION

Extrusion process is unique in that it is a highly efficient continuos process. The residence time of the
material is usually very short. With limited moisture content, starches are gelatinized or melted [1, 2] when
enough energy is supplied either in thermal and/or mechanical form. Wang et.al. [2] have used a molecular
model based on interactions among anhydroglucose units and water molecules and a computer program to
reach a conclusion that 14 molecules of water per molecule of anhydroglucose is needed for complete
gelatinization of starch. A molecular ratio of 14 to 1 is found in a 61% water (w/w), 39% starch system. In
extrusion processes water contents are much lower than that and so a sizable amount of starch is melted if
the final product is virtually all cooked. Melting, but not gelatinization, is the key phenomenon of starch
conversion in extrusion processes. In extrusion processes the energy inputs to the system are generally both
thermal and mechanical in origin. The importance of mechanical energy input to extrusion processes has
been reported [3, 4]. Using both capillary rheometry and single screw extrusion, Wang et. al. [4]
demonstrated that shear (mechanical) energy alone, in the absence of thermal energy input, can cause
cooking of starch (per DSC measurements). Wang et. al. [4] also suggested that shear energy is more
efficient than thermal energy in causing starch cooking based on activation energy calculations. In
extrusion processes, the effects of thermal and shear induced cooking are additive. At temperatures lower
than the DSC transition peak temperature of a sample [1], the effect of shear energy dominates and at
higher temperatures that of thermal energy prevails.

MATERIALS AND METHODS

The starch used was a waxy corn starch. Methionine used was obtained from Sigma. A Perkin-Elmer DSC-
4 was used to measure and calculate degree of cooking (percent conversion) [1, 2, 4]. Extrusion studies were
 517
carried out in a Brabender single screw extruder with a barrel diameter of 19.1 cm and a barrel length to
diameter ratio of 20:1, and in a capillary rheometer (Model 3211, Instron Co., MA). Starch particle size
was studied using an optical microscope equipped with an imaging analyzer (Olympus Cue-2). Flavor
compounds were analyzed using GC-Mass spectroscopy.

RESULTS AND DISCUSSIONS

Figure I shows the dramatic effect of temperature on viscosity of water-waxy com starch mixture at the
specified water contents and shear rates in a Brabender single screw extruder [5]. At temperature lower than
60°C, viscosity increases exponentially with decreasing temperature. High viscosity allows high shear stress.
So we designed experiments carried out at temperatures lower than 60°C by jacketed cooling and using pre-
chilled samples to capitalize on the effect of high shear stress on starch conversion (cooking), starch
granular size reduction and on shearing of molecules, such as methionine. Figure 2 shows that in a single
screw extruder at a temperature of 40°C, a shear rate of 26 sec· l , a shear stress of 106 dynelcm1, and a
moisture content of 35% (w/w), average granular size of starch can be reduced from 12~ to 7~ in 47
seconds and to 1.35~ in 470 seconds. Simultaneously, 45% of the starch can be converted (per DSC
measurements) to cooked starch in 47 seconds, and 65% in 94 seconds. Similarly, in a capillary rheometer
(at 40°C, 35% moisture content), at a shear rate of 150 sec· 1 and a shear stress of 1.5x107 dynelcml, 491'10 of
the starch can be converted in 3 seconds, and 70% in 12.5 seconds. The rate of conversion in the capillary
rheometer is much higher than that in the single screw extruder because the shear energy used in the former
is fifteen times higher than the latter. These results signify the effect of shear energy in extrusion processes.

200
_ 30% water, 26 sec· 1
\
Q)
160 \ --20% water, 26 sec· 1
\
fI)
--
0't:l
fI) --- 30% water, 104 sec-1
D.c: 120 \ - - 20% water, 104 sec-1
- as \
--
~fI)
;:,
0
80 \.
,.,.,
fI)
Or.
()~
.! .......
> 40 .... .......-..
0 25
..... ......
-- ---"-"
-----
50 75 100 125 150
Temperature, DC
Figure 1. Effect oftemperature, shear rate and moisture content on starch viscosity [5].

20 1.00
D Conversion in single screw
D Conversion in capillary \
16 0.80
c: c:
...
° 0
()

E
12 0.60 ...
fI)

Q)
>
c:
Q) 8 0.40
N
C/) °
()
4 0.20
• Weight average size
00 8 10°·00
2 4 6
Number of passes

Figure 2. The effect of residence time on particle size reduction and starch conversion in a single SCf(m'
extruder (shear stress: 106 dynelcm1, shear rate: 26 sec-I, residence time: 47 seclpass) and a capillary
rheometer (shear stress: 1.5x107, dynelcm1, shear rate: 150 sec-I, residence time: 1.56 seclpass) for waxy
com starch with 35% water (w/w) at 40°C.
518
In the absence of thermal energy input, shear energy alone, i.e. a cold extrusion process, can cause granular
size reduction and starch conversion (cooking). In other words, cooking without heat. One may argue that
there is thermal dissipation in these experiments. We used a thermocouple and a IR pyrometer to measure
the extrudate-temperature and they were -40°C. The cooked starch are unpuffed and has the look of pasta.
Using a proper formulation in the feed, one can design a process to make instant noodles based on the
findings presented here. From a scientific point of view, we can conclude that shear energy is equivalent to
thermal energy in the cooking of starch [4]. Shear energy "shears" and "pressurizes" starch granules and so
caused size reduction and starch-cooking, when methionine was mixed in with the starch, methionine
molecules were also sheared to produce free radicals which reacted further to form more stable compounds
as listed in Table 1. Among these four compounds methional stands out for two reasons: (1) the
concentration of methional is order of magnitude higher than that of the rest, and (2) the concentration of
methional is not dependent on the initial concentration of methionine. The concentrations of dimethyl
disulfide, dimethyl trisulfide and propanethiol are relatively low and increase when methionine
concentration was doubled in the feed. We are working on a mechanism for this interesting reaction system
where only shear energy was available to the system. We prefer to categorize these reactions in the realm of
tribochemistry, the chemistry initiated by friction, shearing, or other mechanical forces. To capitalize on
these physical forces, one should use low temperature environment to get high viscosity for a rheological
mass. If the final product is a flavor compound, low temperature process will certainly reduce the loss of the
compound during processing. This is one way to produce "natural flavor". Other compounds are being
studied.

TABLE I
GC-Mass Spectrum result of concentrations of flavor compound formed by extrusion process.

Concentration {E~m)
Peak Compound 1% Methionine 2% Methionine
1 Propanethiol 2.2 33.8
2 Dimethyl disulfide 28.9 182.7
3 Methional 248.2 239.2
4 Dimethyl trisulfide 9.5 42.1

ACKNOWLEDGMENT
This is publication No. F10544-3-93 of the New Jersey Agricultural Experiment Station supported by State
Funds and the Center for Advanced Food Technology (CAFT).

REFERENCES
l. Wang, S. S., Chiang, W. C. Yeh, A. I., Zhao, B. L. and Kim, I. K., Kinetics of phase transition of waxy
corn starch at extrusion temperatures and moisture contents. L. Food Sci., 1989,54(5), 1298-130l.
2. Wang, S. S., Chiang, W. C., Zhao, B. L., Zheng, X. and Kim, I. K., Experimental analysis and computer
simulation of starch-water interactions during phase transition. L. Food Sci., 1991,56(1), 121-124.

3. Chiang, W. C., Zheng, X. and Wang, S. S., The importance of mechanical energy-induced conversion in
a food extrusion process. Presented at IFf National Meeting, Chicago, IL June, 1989.
4. Wang, S. S., Chiang, W. C., Zheng, X. and Zhao, B. L., Application of an energy equivalent concept to
the study of the kinetics of starch conversion during extrusion. Food Extrusion Science and Technology.,
ed. 1. L. Kokini, C. T. Ho and M. Karwe, Marcel Dekker, Inc., 1992, pp.165-176.

5. Lai, L. S. and Kokini, 1. L., The effect of extrusion operating conditions on the on-line apparent viscosity
Of 98% Amylopectin (Amioca) and 70% Amylose (Rylon 7) corn starches during extrusion. L. Rheol.,
1990,34(8), 1245-1266.
 CHEMICAL AND PHYSICAL CHANGES OF TOFUYO
(FERMENTED TOFU) DURING FERMENTATION

MASAAKI YASUDA and NAOT ADA KOBAMOTO

Professors/Department of Bioscience and Biotechnology,
College of Agriculture, University of the Ryukyus,
1, Senbaru, Nishihara-cho, Okinawa 903-01 Japan

ABSTRACT

The chemical components and physical properties of the tofuyo prepared by Monascus
fungus during fermentation were investigated. While crude protein and fat contents of
tofuyo decreased, reducing sugar contents increased during a ripening period. Values of
the ratio of water-soluble nitrogen or 4%-trichloroacetic-acid-soluble nitrogen to total
nitrogen increased during fermentation. It was found that maturation of tofuyo was
carried out by the phenomenon that soybean protein was hydrolyzed limitedly by protei-
nases in the soak (moromi). The main components which formed the body of tofuyo
consisted of basic subunit in glycinin and other polypeptides (Mr. 55 KDa, 11-15 KDa).
Values of hardness and cohesiveness of the product decreased during a ripening period.

INTRODUCTION

Tofuyo is an excellent vegetable protein food and is made from tofu, by the action of
microorganisms in Okinawa Prefecture, Japan. It is nutritiously rich with protein, fat, and
other nutrients. This food is a creamy cheese type product with a mild flavor, fine texture,
and good taste. In Okinawa, usually we take the product as a relish with the awamori
which is traditional distilled liquor or as the cakes served at tea for ladies. Recently, much
attention has been paid to this food as the vegetable cheese which is one of ideal cholester-
ol free foods. Therefore, it would be also suitable for western people because it could be
utilized almost in the same way as cheese. Studies on processing methods of tofuyo has
been carried out in our laboratory (1). In this paper, the authors report the chemical and
physical changes of tofuyo during a ripening period.

MATERIALS AND METHODS

Tofuyo was prepared as follows: dehydrated tofu was soaked into the soak (moromi)
containing the red koji prepared by growing Monascus on steamed rice, a small amount of
salt, and the awamori and then the soak was maturated at 30 DC for 5 months. The
520
chemical compositions of tofuyo were determined by the conventional analysis method.
Slab SDS-polyacrylamide gel electrophoresis was carried out by the method of Laemmli.
The physical properties of tofuyo were measured by a Rheoner (Model;RE-3305, Yamaden
Co.).

RESULTS AND DISCUSSION

1. Chemical changes of tofuyo during a ripening period

The chemical compositions of the tofuyo prepared by red koji during a ripening period
are shown in Table 1. Crude protein and crude fat contents of tofuyo decreased, but
reducing sugar contents increased during a ripening period. Crude fiber was not detected.
Sodium chloride content of the product was almost constant (around 3%) throughout the
ripening period. Sodium chloride content of tofuyo was lower than that of Chinese Sufu
(2).

Table 1
Changes in Chemical Composition of Tofuyo Prepared by
Monascus Fungus During a Ripening Period
Ripening Crude Crude Crude Crude Reducing NaCI
period protein fat ash fiber sugars
(month) (%) (%) (%) (%) (%) (%)
Raw tofu 49.0 30.6 5.5 0 14.2 0
Dehltdrated 47.3 31.9 6.7 0 13.9 0
to u
Tofulo
O. 32.4 23.1 7.9 0 19.8 3.7
1 30.1 21.8 7.3 0 22.3 3.5
2 29.5 21.5 7.4 0 22.9 3.4
3 29.2 21.2 7.4 0 24.2 3.2
5 28.7 21.7 7.4 0 26.3 3.0

Numerous values described above were shown as oven dried

basis

In order to know the ripening-environment of tofuyo, changes in pH, ethylalcohol, and

proteinase activities in the soak during the fermentation were investigated. The values of
pH in the soak was found to be 6.0 at first and gradually decreased to 5.5 with ripening
time (5 months). Ethyl alcohol concentration in the soak, was found to be about 30% at
first and then was gradually decreased to around 20% with ripening time (3 months). It
was considered that the moisture of tofu cubes was transferred to the soak owing to
osmotic pressure and, as the results, ethyl alcohol concentration in the soak decreased.
Although proteinase activity decreased markedly to 37% for the initial 15-day ripening at
the presence of ethyl alcohol of the awamori, its value was almost maintained at around
30% until the end of 5-month ripening. It was consideted from some of the results
mentioned above that the proteinase activity was controlled by the presence of ethyl
alcohol of the awamori. The protein of a product was degraded slowly during the ripen-
ing. This may result in forming a good texture of tofuyo. It is likely that the hydrolytic
products of proteins and lipids provide the principal constituents of a mild characteristic
flavor of tofu yo. Digestion of soybean protein during fermentation was examined by slab
polyacrylamide gel electrophoresis. Some protein bands of soybean globulin such as a'-,
a- and ~-subunits in ~-conglycinin and acidic subunit in glycinin in the water-insoluble
fraction of tofu yo disappeared after the 3-month ripening but that of basic subunit in
glycinin still remained. It was found that the maturation of tofuyo was carried out by the
 521
phenomenon that soybean protein was hydrolyzed limitedly by proteinases in the soak.
And thus, it was found that the main component which formed the body of tofuyo
consisted of basic subunit in glycinin and other polypeptides (Mr. 55 KDa, 11-15 KDa).
Some polypeptide bands in the water-soluble fraction of tofuyo, which have 31, 25, 23, 21
and 12-10 KDa of molecular weights, were disappeared during the ripening period (0-3
months). The nitrogen components of tofuyo during a ripening period are shown in Fig.
1. The ratio of water-soluble nitrogen to total nitrogen (called protein solubility ratio or
ripening ratio) reached to 36.3% after the 3-month ripening. This value was similar to that
of white miso (39%) but was lower than that of red miso (61 %) (3). The ratio of
4%-trichloroacetic-acid-soluble nitrogen to total nitrogen reached to 35.0% after the
3-month ripening. The soybean proteins are digested by the proteases produced by the
mold into peptides and amino acids. Free amino acids which were good taste constituents
such as glutamic acid, aspartic acid, glycine etc., were found in the water-soluble fraction.

_4 1.0
~ 40 N

~ 3
0.8
'"'"c::
'".,'" ---.,
C1)
o
o 0.6 C1)
...... I'l
2 .~
...
X I'l

Z 20
-0
cd -0
>. 0.4 '"
C1)
~
E-<
Z
:I: V')
0
1 0.2 8
V,)
~O 0.0
~
o ~--~--~--~--~--~~ 0 1 2 3 4 5
o 1 2 3 4 5 Ripening period (month)
Ripening period (month)
Fig.2 Changes in hardness and
Fig.l Changes in the ratio of water - cohesiveness of tofuyo
soluble - nitrogen to total during a ripening period
nitrogen of tofuyo during a
0: Hardness
ripening period
.: Cohesiveness

2. Physical changes of tofuyo during a ripeningperiod

The change of physical properties of tofuyo during the 5-month fermentation which
was carried out by red koji was investigated. As shown in Fig. 2, the values of hardness
and cohesiveness of the product decreased during a ripening period. Deformation of the
product depended upon elasticity, viscosity, and viscoelasticity, respectively.

REFERENCES

1) Yasuda, M., Studies on manufacturing of tofuyo -A scientific analysis for producing-

procedure of tofuyo and its technical developments-, Nippon Shokuhin Kogyo
Gakkaishi, 37(5), 403-409 (1990), (in Japanese).
2) Wai, N., Soybean cheese, Bull. lnst. Chern., Acad. Sinica, No.9, 75-94 (1964), (The
Republic of China).
3) Ebine, H. and Ito, H., Miso, A New Edition of Brewing-Components, edited by The
Editorial Committee of A New Edition of Brewing-Components, Brewing Society of
Japan (Tokyo), p. 337 (1977), (in Japanese).
 STATISTICAL CHARACTERIZATION OF TEMPEH STARTER
FROM THE AROMA COMPONENTS OF SOYBEAN TEMPEH

SUPRIYANTO, YUSAKU FUJIO AND ISAO HAYAKAWA

Department of Food Science and Technology, Faculty of Agriculture,
Kyushu University, Higashiku, Fukuoka 812, Japan

ABSTRACT

Fifty four aroma components of tempeh were detected by head-space dynamic

gas-chromatography. Principal component analysis was carried out among 13
aroma components selected as variables for 8 kinds of starters. The net result
shows that the characteristics of the starter varieties in tempeh aroma may be
segmented and summarized by 1st and 2nd principal component scores with a
cumulative contribution rate of at least 70%. The present methodology components
may be helpful in choosing the fermented food starters accordingly.

INTRODUCTION

The selection of starter (mainly Rhizopus strains) and/or fermentation conditions

could significantly affect the quality of fermented foods. The aroma from various
fermented liquid foods, such as soy sauce (1), coffee (2), etc., have been
evaluated through statistical treatments of the aroma components by gas-
chromatography. However, there are a few reports on the quality evaluation of
fermented solid-foods based on aroma analysis (3,4). Tempeh (Indonesian
fermented soybean) is a nutrition-rich fermented solid-food, known to have a
pleasant and a slightly cheese-like aroma (5). This paper deals on the starter
evaluation through principal component analysis (PCA) of the aroma components
of tempeh as an example of fermented solid-food.

MATERIALS AND METHODS

 523
Tempeh starters
Rhizopus sp. LKN, CJ, WJ, ON and UM isolated from Indonesian ragi (traditional
tempeh starter) No.1, No.2, No.3, oncom and usar, respectively.

Laboratory tempeh preparation

Soybean (200g, Ohio USA) was soaked with tap water for 12 h and then,
autoclaved at 121°C for 15 min to avoid the influence of the other microorganisms
except that of the speci..fied starter. After the soybean was fully drained, it was
packed in a perforated plastic bag and inoculated (0.2%) with the desired starter.
The inoculated soybean was incubated at 30°C for 22-24 h unless otherwise
stated. Matured tempeh was freeze-dried and pulverized finely. The freeze-dried
tempeh powder was used in the aroma component analysis.

Trapping of aroma components and gas-chromatography

Trapping of tempeh aroma component was done by the method of Fujio et al.(4)
i.e. 20g of sample was placed into 300 ml of glass vessel with 100 ml of distilled
water and l/fl of n-octanol spotted on a filter paper (lx2 cm), hung in the head
space of the vessel as an internal standard. Beforehand, the vessel was purged
by nitrogen gassing at 110 ml/min for 2 min. Afterwhich a gas-liquid equilibrium
for volatiles was reached at 40°C under nitrogen gassing at 50 ml/min to a total
volume of 800 ml. Then the volatiles were adsorbed on Tenax-TA (#60-80, 25 mg)
packed in a pre-column by nitrogen gassing at 50 ml/min to a total volume of 800
ml.

Gas-chromatographic analysis
The pre-column was inlaid on the injection block of a gas chromatograph (GC;
Shimadzu GC-14PF, 50m silica capillary column, FID). The injection block was
maintained at 200°C and the oven temperature was maintained at 50°C for 5 min,
and then elevated up to 200°C (3°C/min). Helium was used as the carrier gas and
controlled at an inlet pressure of 0.75 kg/cm 2 • The split ratio was 1:50. The peaks
detected were processed with a data processor.

Principal component analysis (PCA)

Principal component score was evaluated by Fujitsu Analyst Program and by
FACOM M-380S computing system. Correlation coefficients between eigen vectors
and respective variables were referred as factor loading. The information of a
series of variables can be summarized into one or two characteristic composite
variables on the respective event.

RESULTS AND DISCUSSION

Aroma components (volatiles) from tempeh

Fifty four volatile components were detected on a gas-chromatograms. Among
them, 13 volatiles were chosen as PCA variables (see PCA results) for 23 kinds of
tempeh samples (triplicate or more tempeh for a starter) which were made by 5
kinds of starters (LKN,CJ,WJ,ON,UM).
524
PCA results
After 1st PCA pre-trial with 54 volatiles, 13 variables with more than 0.5 of factor
loading were chosen for 2nd trial. The 2nd trial gives a significant result in 21
(1st PC) with 47.1% of contribution rate and 22 (2nd PC) with 25.4%. The sum of
cumulative contribution rate of 21 and 22 was an enough size of 72.5%. Figure 1
shows PCA-scores on a scattergram for 5 Rhizopus isolates (LKN,CJ, WJ,ON, UM).
Z225.4%
2.0
k k

k k

I
I
Z,47.1%
j 2.0
-2.0 j j

h
h
h

h h
h
-2.0
Fig.1. The 1st and 2nd principal component of 23 tempeh samples by 5 kinds of
starters (in Fig.1, h:LKN, i:CJ, j:WJ, k:ON, l:UM) on the scattergram.

The 23 tempeh samples were segmented into 5 groups of starters on the

scattergram (Fig.1). Particularly, UM and ON tempeh groups were significantly
segmented. UM tempeh gave positive scores on the 1st PC while ON tempeh gave
negative scores. It can be concluded from the result that the segmentation of
starter by aroma components could be possible by PCA. The developed statistical
methodology can be beneficial to evaluate the aroma-character of tempeh-starter
and to produce a high-quality tempeh with favourable aroma.

REFERENCES

1. Aishima,T., Agric.Biol.Chem., 45, 2847-2853(1981).

2. Wada,K., Ohgama,S., Sasaki,H., Shimoda,M., Osajima,Y., Agric. Biol.Chem.,51,
1043-1047(1987).
3. Supriyanto, Wada,K., Hayakawa,1. and Fujio,Y., Hakkokogaku, 1991,69,331-336.
4. Fujio,Y., Wada,K., Furuta,H., Hayakawa,l., Kawamura,Y.,Nippon ShokuhinKogyo
Gakkaishi,38, 1137-1142( 1991).
5. Steinkraus,K.H., Yap Bwe Hwa, Van Buren,J.P., Providenti,M.I., Hand,D.B., Food
Res.,25 777-788 (1960)
MANUFACTURING OF SOLID CULTURE MEDIA IN FILM BAGS FOR SHIITAKE CULTIVATION

K. OHSAKI, S. YAMADA, H. FUSE, T. MOTEGI, G. KAWAI, and Y. FUKUSHIMA

Technology Research Center, Production Engineering Division
Kikkoman Corporation, 2470 Imagami, Noda-shi, Chiba 278, Japan

ABSTRACT

A continuous system was developed for manufacturing solid culture

media packed in film bags for shiitake mushroom (Lentinus edodes).(l)
The medium materials, sawdust mixed with bran, are steam sterilized and
cooled, then spawned under agitation. Solid culture media thus
produced continuously are packed in two-ply perforated film bags by a
packaging machine. The semi-commercial plant has a manufacturing capacity
of 430 bags per hour, whi ch contai n 1.8-1. 9kg spawned sawdust medi a and
have cyl i ndri ca 1 shape. The bags are incubated for about 2 weeks until
mycelial coat builds up, then delivered to the shiitake growers.

INTRODUCTION

Japan is the world's second largest producer of shiitake mushroom. In 1991,

78,000 metric tons of fresh shiitake and 10,200 metric tons of dried
shiitake were produced. Recently, cultivation on nutrient-enriched sawdust
media formulated in film bags("synthetic or artificial logs")is popularized
in the shiitake industry because of low labor cost and, shorter cultivation
and harvesting time. Nearly 15% of shiitake for the fresh market is being
produced by these bag-in-media instead of natural hardwood logs.
Among several artificial logs sold to shiitake growers here in Japan,
only our product is manufactured in continuous system of sterilization,
cooling, spawning, and packaging of sawdust media.

MATERIALS AND METHODS

The medium materials, beech sawdust mixed with rice bran, are sterilized by
526
steam in a pressure vessel with a screw conveyer and cooled by water
jacket, then moistened, if necessary. Thus prepared media are mixed under
agitation with the liquid spawn of the strain obtained by our own breeding
program, which was propagated continuously in a submerged culture tank(2).
These spawned media are then packed in two-ply perforated film bags of
polyethylene by a bag making-packaging machine in a clean room. Pore size
and perforation patterns of film were designed to promote the release of
carbon dioxide gas outwards and the supply of oxygen gas, and to control
moisture content of the inoculated media.
The semi-commercial plant(Figure 1.) has a production capacity of 430 bags
per hour, which have a 14 cm diameter, 20 cm height and 1.8-1.9 kg weight.

----- Steam
~--__~L-~1 Sterilizer

t
Mixer I Cooler
Cooling __
water
2.f_._._._._-:=-- =:=---:_._._. _
Packaging
t
machine r--"'"

. _._. _ .........- - - ' -.....

Figure 1. Schematic drawing of solid culture media manufacturing plant.

The bags are incubated for about 2 weeks under temperature-humidity

controlled environments(20C-70%R.H.) until the mycelial coat builds up,
and are delivered to the shiitake growers.

RESULTS

In the bags (artificial sawdust logs), mycelium grows and matures in

three months.(Figure 2.)
The cutting and removal of the film initiates the first flush of fruiting,
and the harvesting time comes soon afterwards.
In a cropping room at the temperature of 10-20C, daily watering treatment
of the logs by spray nozzles or occasional soaking in water bath, causes
 527
several times of fruiting(Figure 3.) for 3-4 months, which gives a
high yield of good quality shiitake.(600-800g per bag)

Figure 2. Mycelium growing and maturing in artificial sawdust logs, from the
left;l hour,2 weeks,4 weeks,1.5 months,3 months after inoculation.

Figure 3. Fruiting shiitake on artificial sawdust logs in a cropping room.

CONCLUSIONS

These i nocul ated and incubated sawdust-bran bags, compared to the

hardwood log system, give growers a shorter cultivation period of 7 months
and higher harvest yield of shiitake by several fold. The continuous system
for manufacturing solid culture media in film bags has enabled reliable
and large scale mass production of artificial sawdust logs for the
shiitake cultivation. It will bring a bright future to the development
of the shiitake industry.

REFERENCES

1. Akao, T., et al., System for manufacturing solid medium, United States
Patent, No.4,922,650 (May 8, 1990)
2. Fukushima, Y., et al., Efficient production mycelium of Lentinus edodes
by a continuous culture, Mushroom Science, 1991, 13, 721-725
 SORGHUM BREWING WITH FACULTATIVE ANAEROBIC Rhizopus
IN PRACTICAL OPERATIONS

HSI-HUA WANG, JUU-CHING TSAO

Laboratory of Applied Microbiology,
Deparnnent of Agricultural Chemistry,
National Taiwan University, Taipei 107, Taiwan, R.O.C.

ABSTRACT

Traditionally, in sorghum brewing (a solidstate fermentation), the mash (600kg of steamed

sorghum grains in a batch) wasputin aclosedcondition to avoid acetate fermentation, but the
isolates (Rhizopus spp.) grown on conventional plates did not grow in the mash. Therefore, a
pure culture of facultative anaerobic Rhizopus was chosen here. The apparent fennentation
heat generation rateswere calculated as 3.3 -12.4xl0-2kcal /kg eh, comparable to those of
conventional process 1.6 -7.4x 10-2 kcal/ kge h. The heat generation rates after turning of the
mashare almost proportional to the alcohol productivities. The supplementary inoculation of
pure cultures of facultative anaerobic Rhizopus enhanced alcohol productivity and was able to
savethe inoculation size and also contribute tominimize thechange of thearoma of sorghum
liquor.

INTRODUCTION

Sorghum brewing (SB) is a typical Chinese traditional process of solid state fennentation
[1]. Sorghum grains aresteamed and inoculated with natural starter, whichis madeofcracked
raw wheat grains and contains undefined, mixed cultures. The unit of SB process in Taiwan is
:600 kg steamed sorghum grains (50% moisture) placed in a stainless steelbox sealed with a
polyethylene sheet [1, 2]., Oxygen concentration in mashes decreases rapidly in the early stage,
finally becoming less than 5%. When the spores of Rhizopus isolated from the conventional
plates are inoculated to the sorghum grains as a starter, there is neither germination nor hyphal
growth in the mash. However, when a small portion of the mash is transferred into a flask or a
jar, saccharification and alcohol fennentation occur. It seems that the developement of low
.oxygen within the natural starter may enrich anaerobic sacchllrolytic molds for brewing.
 529
SUPPLEMENTARY INOCULAnON OF FACULTATIVE ANAEROBIC Rhizopus

Anaerobic growth of 60 isolates of molds isolated from the starters of food fermentation in
Asian countries was indicated by Hesseltine et al . [3]. Recen dy we found that some facultative
Rhizopus cultures isolated from natural starters of SB showed spore germination and
saccharification in full operation (600 kg sorghum/tank) and some of them indicated
characteristic aroma [4,5]. It is evident that the facultative anaerobic Rhizopus RT-I-8,
showed definite growth and fermentation from the temperature increment in the mash and
alcohol production ( Table 1), but it is difficult to find an optimal condition for the use of the

TABLE 1
Brewing data with the use of facultative anaerobic
Rhizopus RT-I-8

Maximal Temp. Fermentation heaf

Increment of LlliJh (kcal/kg.h) Alcohol
eC)
I
(x 10-2)
I
Starter (kg/car)a mash production
Before After Before Afte.r (l/car)
turning turning turning turnmg

l.naturalb (18.0) 4.5 13 7.4 22.0 52.7

2.naturalb (18.0) 4 13 6.6 22.0 57.29
3.naturalb (18.0) 1 15 1.6 25.4 58.4
4.Rhizopus /yeasf (15.0) 3.5 5.5 5.8 9.3 23.02
5.Rhizopus /yeastC (15.0) 2 8. 3.3 13.5 28.1
6.Rhizopus /yeasf (15.0) 5 10.5 8.3 17.8 50.6
7.natural (9.0) 6.5 13 10.8 22.0 62.5
Rhizopus +yeastd (3.0)
8.natural (9.0) 3.5 13 5.8 22.0 73.8
Rhizopus +yeast (3.0)
9.natural (9.0) 7.5 not 12.4 not 89.8
Rhizopus +yeast (3.0) turned turned
1O.natural (18.0) 2 12 3.3 20.3 76.6
a:for one stainless car having 600 kg steamed sorghum grains
b:natural starter
c:Rhizopus RT-I-8 cultured in wheat flour was inoculated at the beginning and the
suspension of pure culture of Saccharomyces S-20[I] was sprayed when the mash was
turned.
d:Both Rhizopus RT-18 and Saccharomyces S-20cultur~d in sterilized wheat flour were
inoculated. The inoculum densities of the Rhizopus and the yeast were 10 spores and 10
cells, respectively. '
e:fermentation heat= heat increcem<?nt x specific heat [6]
mass specific heat: 0.799 kcallkg.oC before turning,
0.815 kcallkg.oC after turning
530
Rhizopus ,because there are many features to be considered, for example, the selection of
yeaststrain, thegrowth stages of the moldand yeast. However it was shown that when the
Rhizopus was used for the reduction of the quantity of natural starters, the alcohol
productivity was enhanced very much. In this case, the aroma of the liquor was not damaged as
experienced when only the pure cultures were used. Since the manufacture of natural starter
requires the spaces and time, the supplementary use of the facultative anaerobic Rhizopus RT-
1-8 can be used for practical operation.

CONCLUSION

Thesupplementary use of facultative anaerobic Rhizopus can save natural starters and also
contribute to minimize the changes of aroma.

REFERENCES

1. Wang, H.H. and Hsieh, T.C., Kao-liang brewing by pure culture. In Ferment. Techno!.
~, ed. G. Terui, Society of Fermentation Technology, Japan, 1972,651-658.

2. Wang, H.H., Huang, S.Y., Lee, C.M.,Lee, C.c. and Ma, T.P., Hyphalmass and CO 2
toxicity in solid mash during Kao-liang brandy brewing. In Proceedings of ~ First
Intersectional Congress of the International Association ofllie Microbiologic Society, Tokyo
1974,5, 146-153.

3. Hesseltine, C. W., Featherston C.L., Lombard, G.L. and Dowell, V .R., Anaerobic growth
of molds isolated from fermentation starters used for foods in Asian countries. Mycolo~ia,
1985, 77, 390-440.

4. Lin, M.S. and Wang, H.H., Anaerobic growth and oxygen toxicityofRhizopus cultures
isolated from starters made by solid state fermentation. Chinese L of Microbiology and
Immunology, 1991,24(2),229-239.

5. Lin, M.S., Wu, C.Y., Lo, S.C. and Wang, H.H., Application of facultatively anaerobic
Rhizopus in solid state fermentation. J. Chinese Agricultural Chemical Society, 1991,29(3),
310-317.

6.Lai, M.N., Chang, F.W. and Wang, H.H., Application of pseudo-steady heat conduction
modelfor solid statefermentation of sorghum mash. NiDDon Shokuhin Kogyo Gakkaishi,
1991, 38(3), 226-234.
 PRODUCTION OF ENZYMES FOR FOOD PROCESSING BY
SOLID STATE FERMENTATION

MASAYUKI TANIGUCHI, KAZUHISA OOSHIMA, and MICHIHIRO FUJII.

Department of Material and Chemical Engineering
Niigata University, Ikarashi 2, Niigata 950-21, Japan

ABSTRACT
When Aspergillus oryzae was cultivated by the solid state fermentation
(SSF) using a fluidized bed fermentor, the oxygen uptake rate per
the weight of the fermented matter was higher than that in the
stationary culture. The dry cell mass per the weight of the
fermented bran in the fluidized bed cuI ture was about 3-fold higher
than that in the stationary culture. The total quantities of
enzymes, such as neutral protease, a-amylase, a-glucosidase, super-
oxide dismutase (SOD), catalase and glutamate dehydrogenase (GDH) ,
per the weight of the fermented matter in the fluidized bed culture
were much greater than those in the static culture.

INTRODUCTION
A number of enzymes for food processing have been produced by solid
state fermentation (SSF) in Japan. Solid culture provides natural
condi tions for life of many microorganisms and is a simple culture
method in which microorganisms grow on a moistened solid or semi-
solid substrate. In general, solid culture has advantage over
submerged culture in the production of highly concentrated enzymes
with high yield. However, the fundamental culture conditions such
as temperature, moisture content, and pH in the static solid
culture are difficult to be regulated stably and homogeneously
during the entire culture period. In order to overcome the
disadvantages of the stationary solid culture, a new fluidized bed
fermentor, in which the solid substrate was fluidized with a
sterile air, is developed. The fluidization of solid substrate
results in maintenance of temperature at a constant level and
homogeneous moisture content of the substrate in the bed. Akao et
al.(l) and Tanaka et al.(2) have reported that the fluidized bed
cuI ture is sui table for enzyme production by filamentous fungi.
In this paper, the productivity of hydrolases (neutral
protease, a-amylase, a-glucosidase) and oxidoreductases
(superoxide dismutase (SOD), catalase and glutamate dehydrogenase
(GDH) produced by SSF using the new fluidized bed fermentor were
investigated as compared with those by static SSF and liquid
532
culture.
MATERIALS AND METHODS
Microorganism and medium
Aspergillus oryzae ATCC 20386 was mainly used for production of
hydrolases and oxidoreductases. Wheat bran powder and a solution
extracted from wheat bran with water were used as a medium for the
SSF and for the liquid culture, respectively.
Fluidized bed fermentor
Figure 1 shows the schematic diagram of the fluidized bed fermentor
wi th total volume of 1.4 liters. The size (mm) of main sections of
the fermentor is also shown at the right side of the figure. The
fluidized bed ferment or was produced by Kikkoman Co., Noda, Japan.
Air from the air pump,l, was filtered with a filter,6, for air
sterilization and blown into the fluidized bed fermentor, 10.
Distilled water, 14, from the pump sterilized through the filter, 16,
was sprayed into the fermentor from the nozzle, 8, with the
sterilized air from the air pump, 1, keeping the moisture content at
a constant level. The moisture content was controlled by measuring
the electrostatic capacity between an electrode,g, and the nozzle.

RESULTS AND DISCUSSION

Enzyme production by fluidized bed culture
Aspergillus oryzae was cultivated by the two solid culture methods.
When Aspergillus oryzae was cultivated by the SSF using the fluidized
bed fermentor, the oxygen uptake rate per the weight of the
fermented matter was higher than that in the stationary culture. At
the early stage of the fluidized bed culture, the glucosamine
content of the fermented matter increased, indicating vigorous
growth of the fungus. The dry cell mass per the weight of the
fermented bran in the fluidized bed cuI ture was about 3-fold higher
than that in the stationary culture (data not shown).
1.air pump
2.condenser
3.separater
4.flow meter I
J-+~I
5.pressure gauge
6.air filter
7.heater I
8.nozzle
9.electrostatic electrode

B~
10.fermentor
11.moter for vibration
12.fan heater 6

--(W
13.accumulator 17 16
'-;---IXI---'

t
14.water tank
15.water pump
16.water filter

~
17.heat-insulating case

(A) (B)
Figure 1. Schematic diagram of the air-solid fluidized bed fermentor (A) and
the size of main sections of the fermentor (B).
 533
Table 1 and 2 show comparisons of the production of hydrolases
(neutral protease, a-amylase, and a-glucosidase) and
oxidoreductases (SOD, catalase and GDH), per the weight of the
fermented matter between the stationary solid culture and the
fluidized bed culture, respectively. The total amounts of three
kinds of oxidoreductase as well as three kinds of hydrolase
obtained by the fluidized bed culture were much greater than those
by the stationary solid cuI ture. The increase in enzyme production
seems to be attributable to enhancement of the cell mass. However,
the ratios of extracellular amounts of a-amylase, a-glucosidase,
and catalase to the total one in the fluidized bed culture were
lower than those in the stationary solid culture.
Extracellular oxidoreductase production by solid culture
More than 50 % of the total amounts of oxidoreductases were
extracellarly produced when several filamentous fungi were grown
in the stationary solid culture. At present, we attempt to apply
the fluidized bed fermentor to extracellular production of some
oxidoreductases.

TABLE 1
Comparison of hydrolase production from Aspergillus oryzae between the stationary solid
culture and the fluidized bed culture.
Enzyme activity [U/g-dry matter]
Cultivation method Neutral protease a-Amylase a -Glucosidase
Stationary culture 21,000(92) 79.8(99) 0.0866(83)
Fluidized bed culture 66,300(93) 458 (67) 0.181 (52)
Data show the maximum value during each fermentation.
Values in parenthesis indicate the percentage of extracellular enzyme amount.

TABLE 2
Comparison of oxidoreductase production from Aspergillus oryzae between the stationary solid
culture and the fluidized bed culture.
Enzyme activity [U/g-dry matter]
Cultivation method SOD Catalase GDH
Stationary culture 86.5 (75) 68.3(93) 0.0886(79)
Fluidized bed culture 482 (82) 595 (44) 3.59 (92)
Data show the maximum value during each fermentation.
Values in parenthesis indicate the percentage of extracellular enzyme amount.

REFERENCES
1. Akao, T. and Okamoto, Y., Cultivation of microorganisms in an
air-solid fluidized bed. Proceedings of the Fourth International
Conference on Fluidization, Kashikojima, Japan, Engineering
Foundation, New York, 1983, p.631-637.
2. Tanaka, M., Kawaide, A., and Matsuno, R. , Cultivation of
Microorganisms in an air-solid fluidized bed fermentor with
agitation. Biotechnol. Bioeng., 1986, 28, 1294-1301.
 CELL GROWTH AND a-AMYLASE PRODUCTION CHARACTERISTICS OF
BACILLUS SUBTILIS ATCC 601

TEREZINHAJ.G. SALVA
Institute of Food Technology
CP 139 CEP 13073-001- Av. Brasil, 2880 Campinas-SP Brazil

IRACEMA O. MORAES
Universidade Estadual Paulista
CP 136 CEP 15054-000 - DETAIUNESP S.J.R.P.-SP Brazil

SUMMARY

The effect of glucose, dextrin, soluble starch, lactose, peptone and yeast extract on the growth
and a-amylase production by Bacillus subtilis ATCC 601 was investigated. The overall specific
enzyme activity was the same in media containing peptone from O. Og/l up to 10. Ogll. A maxim um
overall specific enzyme activity occurred at yeast extract 5.0g/1. Regarding the effect of the
carbon source, the maximum enzyme activity occurred at glucose O.75g/1 and 30.0g/1 or dextrin
1O.Ogll. The lowest enzyme production occurred in lactose media. The overall specific enzyme
activity, however, was higher at low levels of lactose than in glucose and was maximum in
dextrin media.

INTRODUCTION

The effect of the culture medium composition on the enzyme production is one of the major
subjects in the microbial enzyme research. Frequently, the enzyme production increases when
complexes sources of nutrients like peptone, yeast extract corn steep liquor and oil seed cake
are employed instead of pure ones (l, 2). Enzyme substrates like starch dextrin, glycogen and
oligosaccharides containing a-l,4 linkages are thought to be good carbon sources for a-amy lase
synthesis. The enzyme, however, may be produced in medium containing glucose, lactose and
xylose (1). When glucose is the carbon source the enzyme synthesis is highly influenced by its
concentration in the culture medium. A nutrient may influences the amount of enzyme produced
by affecting the enzyme per cell mass unit and/or by affecting the microbial growth. The aim of
the work presented here was to study the effect of the peptone, yeast extract and carbon source
concentrations on the a-amylase production by Bacillus subtilis ATCC 601 trying to correlate
it with microbial growth.
 535
MATERIALS AND METHODS

Enzyme production - Three days old Bacillus subtilis ATCC 601 slant cultures were used
throughout this study. The microorganism was cultured in a basal medium composed of
(g/I):(NH4)2S04 2.5, MgS04 7H20 0.5, KCI 0.5, K2HP04 1.0 and CaCh 0.418. When studying
the effect of the peptone concentration on the enzyme production, yeast extract at 5.0g/l, dextrin
at 4.0g/1 and peptone at some different concentrations were added to the basal medium. When
studying the effect of the yeast extract concentration, peptone at 10. Og/l and dextrin at 4.0g/1
and some different concentrations of that component were employed. The study of the effect of
the carbon source was carried out in basal medium containing yeast extract at 5.0g/1, peptone at
10.0g/1 and some different concentrations of dextrin, glucose, soluble starch or lactose. Peri-
odically a sample of the fermented broth was analysed regarding a-amylase activity and cell
mass concentration.
Enzyme activity and cell mass concent.oation - The dextrinizing activity was assayed through
the method described by MEDDA & CHANDRA (3). The cell mass concentration was
determined through optical density measurements at 660nm. Emax was defined as the maximum
enzyme activity in 72 hours of fermentation. Specific enzyme production rate, ~, was defined
as 11X ~/~t, where X was the maximum cell concentration in a given time interval and ~E/~t
was the mean enzyme production in the same interval. Overall specific enzyme activity was
defined as EmaxlXmax, where Xmax was the maximum cell concentration in 72 hours. Overall
productivity was defined as EmaxlXmax.t, where t was the time elapsed until the maximum
enzyme activity was reached.

RESULTS

The enzyme production was higher at a peptone concentration of 10.0g/1, where the enzyme
produced was twice that in the medium without the nutrient. The profile of the a-amylase
production in the media with several different yeast extract showed that the maximum enzyme
production was 19.9UD/ml and occurred at 5.0g/1. Contrary to peptone there was also a
maximum overall specific enzyme production and overall productivity at this yeast extract
concentration. The experiments with different carbon sources showed that as the initial
carbohydrate concentration increased the I!p values in the first 24 hours of fermentation became
less significant in relation to the ~p values in the time intervals 24-48 hours and 48-72 hours
(Table 1). In media with lactose the microorganism grew slowly and the cell lysis occurred only
in the medium with 5.0g/l. The ~p values were maximum in the interval 48-72 hours regardless
the lactose concentration.

DISCUSSION

Yeast extract seemed to act directly on the regulatory mechanism of the enzyme synthesis
whereas peptone affects mainly the microbial growth. So, the enzyme concentration in the
fermented broth increased as the peptone concentration increased up to 10. Og/l but the overall
specific enzyme activity remained constant. . The results from the experiments with different
carbon sources suggested a catabolite repression and showed that the bulky of the enzyme
production took place after intense microbial growth has ceased. The low enzyme production
per cell mass unit in the culture media with glucose was counter balanced by a high cell mass
concentration resulting in a fermentation broth with high enzyme activity.
536
TABLE 1.
Effect of carbon source on growth and a-amylase production by Bacillus subtilis A TCC 601.

~p E
UD/mg.h UD/ml
carbon carbohydrate ~Pl ~P2 ~p3 24h 48h 72h EmaxlXmax
source conc. gil {0-24}h {24-48}h {48-72}h UD/mS
0.05 0.50 0.21 0.09 6.5 9.2 10.4 19.3
0.75 0.53 0.42 0.03 17.0 30.8 30.9 23.1
glucose 10.0 0.12 0.53 0.13 3.2 17.2 20.6 18.6
20.0 0.14 0.32 0.47 4.1 13.5 21.4 17.7
30.0 0.12 0.49 0.76 2.6 13.3 29.9 32.9
50.0 0.03 0.10 0.00 0.7 3.3 2.3 2.9
0.5 0.43 0.30 0.00 6.8 11.5 11.5 17.7
soluble 10.0 0.41 0.59 0.27 6.4 15.7 19.5 30.0
starch 20.0 0.25 0.75 0.30 4.2 18.6 24.4 30.5
30.0 0.13 0.25 0.62 2.3 6.6 16.0 35.6
2.0 0.67 0.33 0.06 8.8 13.2 13.8 25.1
dextrin 10.0 0.19 0.88 0.71 2.5 16.0 27.0 42.2
20.0 0.09 0.85 1.15 0.5 7.2 18.0 46.2
30.0 0.10 0.69 0.85 0.5 5.8 12.3 38.4
5.0 0.19 0.19 0.62 2.2 4.5 11.8 24.1
lactose 10.0 0.19 0.18 0.70 2.5 5.2 15.6 25.2
20.0 0.26 0.25 0.31 1.9 4.8 9.3 15.5
30.0 0.14 0.14 0.24 1.6 3.4 7.2 10.9

CONCLUSIONS

Bacillus subtilis ATCC 601 was not very sensitive to glucose concentrations. The enzyme is
produced even at high levels of glucose, dextrin, soluble starch and lactose. Yeast extract affects
the enzyme synthesis per cell mass unit whereas peptone affects microbial growth. The bulky
of enzyme synthesis occurs after intense microbial growth has ceased.

REFERENCES

1. CHANDRA, A.K.; MEDDA, S. & BHADRA, A.K. Production of extracellular thermostable

a-amylase by Bacillus Iicheniformis. J. Ferment. Technol., 58(1):1-10,1980.

2. KRINSHNAN, T. & CHANDRA, A.K. Effect of oilseed cakes on a-amylase production by

Bacillus licheniformis CUMC 305 Appl. Enron. Microbiol. 44(2):270-274, 1982.

3. MEDDA, S. & CHANDRA, K. New strains of Bacillus licheniformis and Bacillus coagulans
producing thermostable a-amylase active at alkaline pH. J. Appl. Bacteriol. 48:47-58, 1980.
 FERMENTATION PROCESSES FOR THE PRODUCTION OF NATURAL
COLOUR COMPOUNDS FOR THE FOOD INDUSTRY

A.M. Martin
Food Science Program, Department of Biochemistry,
Memorial University of Newfoundland.
St. John's, Newfoundland, Canada, AlB 3X9

ABSTRACT

In the food and feed industry there is growing demand for natural sources of colour. This
presentation will deal with studies on the development of a technology for growth of the
yeast Phaffia rhodozyma and the production of high yields of carotenoids, specifically
astaxanthin, by this yeast. In aquaculture, it is necessary to incorporate this pigment into the
diets of salmonids. The optimization of the growth and pigment production of P.
rhodozyma, as a function of the agitation speed and fermentation time, will be reported.

INTRODUCTION

The market for natural sources of colours in the feed and food industry is significant, and
new applications are being investigated to keep pace with changes in consumer trends (1).
Carotenoids constitute an important group of color compounds, employed in many food
products, among other applications.

Microbial Production of Colour Compounds

Some microorganisms are known to produce colored pigments under various growth
conditions. P. rhodozyma preparations, when fed to rainbow trout, have been found to
impart a pink color to their flesh (2), and to serve as a supplementary nutrient source in
their diets. Its production could be profitable if a stable market, such as an aquaculture
industry, were found and if waste resources were utilized as substrate.
Liquid extracts from peat produced by acid hydrolysis have been tested as a source
of fermentation substrates (3,4). Potential carotenoid precursors, such as xanthophyll and
carotenes, exist in peat (3). This work reports the study of two growth conditions for the
cultivation of the yeast P. rhodozyma in peat hydrolysates.
538

MATERIALS AND METHODS

Peat Hydrolysate
The hydrolysates used in the research reviewed by this work were produced at a pressure
of 3000 psi and a temperature of 185°C for 2 minutes (5).

Inoculum Culture Preparation

The source of the P. rhodozyma yeast cultures used in the study reviewed here, and their
maintenance and inoculum preparation procedures, were reported by Martin et al. (5).

Culture Conditions
The filtered peat hydrolysates were diluted to adjust the total carbohydrate concentration
(TCH) to the desired level. Portions (50 mL) of the filtrates in 125 mL flasks were adjusted
to the required pH with NaOH, and nutrient supplements added. The media were then
sterilized and inoculated. The cultures were conducted at various agitation speeds and
fermentation times at the initial pH of 5 and a temperature of 20°C in a water bath shaker.
After the experiments, the yeast biomass was harvested by centrifugation.

Analytical Methods
The dry biomass concentrations were calculated on a dry weight basis. The TCH of the
substrates were calculated according to the anthrone method (5). A modified method (6),
and the formula of An et al. (7), were used for the carotenoid determination.

RESULTS AND DISCUSSION

Effect of Agitation Speed

Figure 1 shows that the dry biomass (X), yield (g of dry biomass produced / g of dry
substrate consumed, Y) and efficiency (g of dry biomass produced / g of initial dry
substrate, E) of the yeast attained maximum values at agitation speeds of 150 to 200 rpm.
It appears that the shearing effect of higher agitation speeds could be harmful to this yeast.

:5 4O
t;
z
....
':]
0>
I..J
'" 30
w
0
u:
LL
:3 w
I1J

~
20 c
2 ~
III
C
10 ....J
w
~ 1
E E ;:::
lII!
0 0
150 200 250
AGITATION SPEEO Crpm)

Figure 1. Effect of agitation speed on Phajfia rhodozyma biomass production. Initial

pH = 5, cultivation time = 5 days, and temperature = 20°C. From Martin et al. (5).

Effect of Fermentation Time

In Figure 2, it can be seen that the highest growth parameters were obtained at 5 days of
fermentation. In general, for the latter hydrolysate, there was a gradual increase in growth
 539

from days 2 to 4, which accelerated between days 4 and 5, and a subsequent decrease in
growth after day 5. Statistically, values obtained at all levels of measurement were different
(P > 0.05).

Carotenoid Content
Cultivating the yeast under the optimum growth conditions reported in this work produced
a carotenoid concentration of 1,279.82 p!ig dry yeast biomass (5).

4 35

30 ~
w
~
t;, :3 25
(.)
u u.
u.
w
Bl 20
2
~
aI
15 ~
10 c
...J
>- 1 w
E5 E 5 ;::
~
0 0
2 3 4 5 6
FERvlENTATION TIME (DAYSJ

Figure 2. Effect of time on Phajfia rhodozyma biomass production. Agitation speed

= 250 rpm, initial pH = 5, and temperature = 20°C. From Martin et al. (5).

REFERENCES

1. Dziezak, J.D., Application of food colorants. Food Technol., 1987, 14(4), 78-88.

2. Johnson, E.A., Conklin, D.E. and Lewis, M.J., The yeast Phaffia rhodozyma as a
dietary pigment source for salmonids and crustaceans. J. Fish. Res. Board Can ..
1977,34,2417-21.

3. Fuchsman, C.H., Peat: Industrial Chemistry and Technology, Academic Press, New
York, 1980.

4. Manu-Tawiah, W. and Martin, A.M., Study of operational variables in the

submerged growth of Pleurotus ostreatus mushroom mycelium. .AImL. Biochem.
Biotechnol., 1987, 14,221-9.

5. Martin, A.M., Acheampong, E., Patel, T.R. and Chornet, E. Study of growth
parameters for Phajfia rhodozyma cultivated in peat hydrolysates . .AImL. Biochem.
Biotechnol. (In press, 1993).

6. Gentles, A. and Haard, N., Bioavailability of astaxanthin in Phajfia rhodozyma. In

Advances in Fisheries Technology and Biotechnology for Increased Profitability, eds.
M.N. Voigt and J.R. Botta, Technomic Publishing Co., Inc., Lancaster, Penn., 1990,
pp. 373-386.

7. An, G.H., Schuman, D.B. and Johnson, E.A., Isolation of Phaffia rhodozyma mutants
with increased astaxanthin content. Alm!:. Environ. Microbiol., 1989, 1, 116-124.
 EFFECT OF CASTOR OIL HYDROLYSATE ON THE PRODUCTION OF
,- DECALACTONE AND CIS-6-DODECEN-4-0LIDE BY
SPOROBOLOHYCES ODOR US

SHIOW-LING LEE AND CHENG-CHUN CHOU

Graduate lnst. of Food Sci. &Technol., National Taiwan University,
Taipei, Taiwan, ROC

ABSTRACT
Addition of castor oil hydrolysate to broth at 0 hr of cultivation
significantly increased the yield of I -decalactone while reduced
the production of cis-6-dodecen-4-olide by S. odorus. The highest
yield of, -decalactone of 95.39 mgjL which Is 63 times that in the
control broth, was noted when broth was added with castor oil hydro-
lysate so it contained 0.06% ricinoleic acid.

INTRODUCTION
The consumer perception that nature" is good has led to an increased
demand for natural fragrances. Microbial fermentation is regarded
as a potential means to produce natural flavor substances(l).ln the
present work, we investigated the effect of castor oil hydrolysate on
the production of I -decalactone and cis-6-dodecen-4-01ide, respon-
sible for peach odor and sweet odor of lamb meat, respectively (2).

MATERIALS AND METHODS

Microorganism and cultural conditions
S. odorus AHU3246 was used as the test organism. Fermentation was
carried out in 250 ml screw-capped Erlenmyer flasks. Each flask con-
tained 100 ml modified YNB (Bacto yeast nitrogen base without amino
acids and ammonium sulfate supplemented with 3% mannitol and 0.1%
asparagine). The cultures were incubated at 25 ·c on a rotary shaker
(160 rpm) for 7 days. During fermentation, 0.1 g castor oil or castor
oil hydrolysate in an appropriate amount was added to culture at the
beginning or after various cultivation time.
Chemical and microbial determinations
 541
Castor oil hydrolysate was prepared according to the method described
by Christie (3). Analysis of fatty acids in castor oil was also per-
formed according to Christie (3) with gas chromatography.
Volatile flavor compounds were extracted with diether ether-pen-
tane (1:1) in Likens-Nickerson apparatus from the supernatant fluids
of cultured broth separated by centrifugation. Quantitative analysis
was carried out with gas chromatography. Growth of the yeast was mea-
sured by counting the viable cells on Yeast Meat agar.

RESULTS & DISCUSSION

It was found that the ricinoleic acid was the main dominant component
of castor oil, with a relative concentration of about 86.36%. Other
minor components included palmitic acid, stearic acid, oleic acid and
linoleic acid.
Addition of castor oil (0.1 gilOO ml) to broth increased the yield
of 'l -decalactone. The addition of equivalent amount of hydrolyzed
castor oil resulted in a more profound increase of 'l -decalactone pro-
duction of 88.34 mg/L which is 63 times that in the control broth. Be-
sides, the relative percentage of 'l -decalactone also increased to
98.64% (Fig. lA). Castor oil or castor oil hydrolysate showed no en-
hancing effect on the yields of cis-6-dodecen-4-01ide (Fig. IB). Okui

• 100 A B
0 0
100 • 15 15
80 w~ ~
r-
80
~ ~ 10 10 w
.s 60 60 Cl .s Cl

f
a ~ a...J ~
...J
w
40 40 Wz 5 5
z
w
w
> 20
- >
0 0
20 wa: a:
w
0 0 !l.
0 0 !l.

a b c a b c
FIGURE 1. (A) Y -Decalactone and (8) cis-6-dodecen-4-olide produced by S. odorus in broth. a.
broth containing neither castor oil nor its hydrolysate.; b, broth containing castor oil.; c, broth
containing castor oil hydrolysate.

et al. (4) showed that some Candida spp. oxidized ricinoleic acid
to 'l -decalactone.Castor oil exerted less enhancing effect on the pro-
duction of 'l -decalactone than its hydrolysate as observed by us. It
clearly showed that S. odorus could utilize and metabolize castor oil
hydrolysate more effIciently.
Among the various amounts of castor oil hydrolysate tested, broth
containing 0.06% ricinoleic acid showed the most enhanced 'l -decalactone
production with the greatest yield of 95.39 mg/L which is about 63 times
that (1.51 mg/L) in the control medium after 7-day cultivation (Fig.2).
Besides, the relative percentage of 7 -decalactone was also noted to be
the highest (98.0%).
Time of castor oil hydrolysate addition also affected the produc-
tion of 'l -decalactone (Fig.3).The greatest yield (96.3 mg/L) was ob-
served in the broth with castor oil hydrolysate added at the beginning
of cultivation while the relative percentage of 'l -decalactone were
similar regardless of various time of addition.
542
o

•
100 100 ~

~
o!!.
80 80 -
w
:::J
C, 60 60 ~
S

I
40 I 40 ~
0 w
-'
w 20 I i 20 u
a:
;;:
0 f
i i o
w
a..
o 002 0.04 0.06 0.08 0.1 0.2
CASTOR OIL HYDROLYSATE ('lEo RICINOLEIC ACID)

FIGURE 2. Production of Y -decalactone by S. odorus in broth containing various amounts of castor

oil hydrolysate.

0
100
•
100
~
80 80 w
:::J (!)
CI 60 60 «
I-
.§. 40 z
0
40 w
U
..J
W 20 20 a:
;;: w
0 a..
0
0

TIME OF ADDITION (HR)

FIGURE 3. Production of Y -decalactone by S. odorus in broth added with castor oil hydrolysate
( 0.06% ricinoleic acid) at various cultivation time.

Base on the data we collected in the study, it may concluded that

castor oil hydrolysate can exert more enhancing effect on the produc-
tion of ~ - decalactone than castor oil by S. odorus.

ACKNOWLEDGEMENT
This work was supported by Council of Agriculture, the Republic of China.

REFERENCES

1. Welsh,F.W., Murray, W.O. and Williams, R.E., Microbiological and

enzymatic production of flavor and fragrance chemicals. Crit. Rev.
Biotech., 1989,9, 105-69. - - --
2. Tahara,S., Fujiwara,K. and Mizutani,J., Neutral constituents of
volatiles in cultured broth of Sporobolomyces odorus. Agr. BioI.
Chern., 1973, 37, 2855-61. - --
3. Christie,W.W. 1982. Lipid Analysis. Pergamon Press. N.Y. U.S.A.
4. Okui,S., Uchiyama,M.-ana-Mizugaki,M., Metabolism of hydroxy fatty
acids. II. Intermediates of the oxidative breakdown of ricinoleic
acid by genus Candida. !. Biochem., 1963,54, 536-40.
 MODELLING OF TRANSPORT PHENOMENA DURING THE CULTIVATION OF
Bacillus thuringiensis FOR THE PRODUCTION OF BIOINSECTICIDES.

Perez Galindo, A., Velazquez de la Cruz, G., Medrano-Roldan, B., Robles-

Cardenas, M., Rodriguez-Padilla, C.(*), Tamez-Guerra, R.(*), and J.L. Ga16n-
Wong (*). Instituto Tecno16gico de Durango, Unidad de Biotecnologia
Industrial. Av. Felipe Pescador 1830 ote. 34080 Durango, 090. Mexico. (*)
Fac. de Ciencias Bio16gicas U.A.N.L. Monterrey, N.L. Mexico.

ABSTRACT

This paper reports the preliminary design and application of a

mathematical model which seeks to explain observed growth rates
during propagation of a native strain of Bacillus thuringiensis
var. kumamotoensis for the production of a bioinsecticide. The
model, based on the mass transfer coefficient, has the following
assumptions: 1) Mass transport in the total of cells may be des-
cribed by a single balance equation. 2) Each cell behaves as
the average of the whole population. 3) The substrate used by
the microorganism is homogeneous.

INTRODUCTION

In recent years, there has been considerable interest in the

application of mathematical models to microbial processes (1-3).
Biological models try to predict the growth rate as a function
of substrate, cell and inhibitory products concentrations.
In this work we studied the kinetics of a native strain of
Bacillus thuringiensis characterized as kumamotoensis during its
exponential and stationary phases. In the latter, we are
intere§ted in getting information about production of an
insecticide. Measurements were used to design a mathematical
model, based on mass transfer phenomena, to be applied to the
automatic control of the process. The model could be made more
general by using parameters directly related with process
fermentation variables such as pH, temperature, aeration and
agitation rates, etc.

MATERIALS AND METHODS

The model was tested with experimental data obtained from the
544
literature and from our 20 liter semi-pilot fermentation plant
Lh Engineering series 2000. Equations for substrate consumption
and growth rate were posed with Monod's model and our model.Mod-
el parameters were calculated from the data by means of a least
squares subroutine of EUREKA. They were used in the differen-
tial equations to perform simulations with the aid of ISIM.
Comparisons of models' results against experimental data, by
graphical and numerical means, were effected with QUATTRO PRO.

RESOLTS AND DISCOSSION

The model is based on the general equation of growth which

states that the rate of change of the population is proportional
to the population itself. In this model, an expression for the
specific growth rate was derived by assuming that it depends on
the flow of substrate to the cell, which is a function of the
mass transfer coefficient and microbial physiology.
Solving the mass tranfer equations for the microbe-
substrate system, under assumptions mentioned in the abstract,
the specific growth rate is:

~ = ASX (1-(B(1-e- ct ») ( 1)

where A,B and C are the model constants.

Substitution of Eqn. (1) into the general equations for
microbial concentration, X, and substrate consumption, S, with
a yield coefficient Y results in the following equations:

dX/dt = ASX2 (1-(B(1-e- ct ») dS/dt =- (dX/dt)/Y (2)

Figure 1 shows experimental data (R points), along with predic-

tions from our model (P points) and Monod's model (M points).
As seen there, the growth and substrate consumption curves (on
legend as Log cfu and Subs, respectively) are correctly predict-
ed by both models.
70~--------------,160

·120 ::::-
50 e
-t 40
...
100 l -Log cfu R
+ Subs. R
Ul
80 ..... Log cfu P
~f-<
C)

:3 301 .... 60 V1
"'Subs. P
I .
~
::>
* Log cfu M
20 . V1
+ Subs. M
... 40

20

oL-----------------~~~
o 0.54 0.90 1.23 1.58 1.95 2.33 2.7
TIME (hr)
Figure 1. Comparison of model predictions with yeast kinetics
 545
Data from our experiments are presented in the first two
columns of Table 1. Model predictions are shown in columns third
to sixth of the table, as ratios of predicted to measured
values. A noteworthy fact is the abnormal behavior in the growth
and substrate evolutions, typical of our strain. A visual
comparison of the numbers in columns third and fourth (our
model) against those of Monod's model (columns fifth and sixth),
shows the former to yield better predictions, wi th numbers
always closer to one. This is due to the basic formulation of
Monod's model, which fails to predict these type of kinetics.

TABLE 1
comparison of Model Results with Our Experimental Measurements
Experimental data This Model/Exp. Monod's/Exp.
Time Subst.
(hr) Log (cfu!ml) (mg/ml) fraction fraction
o 6.00 1.72 1. 000 1. 000 1.000 1.000
2 6.68 1.70 0.997 0.826 0.975 0.883
4 7.32 1.20 0.984 0.956 0.955 1.079
6 7.92 0.74 0.962 1.279 0.939 1.494
8 8.48 0.50 0.936 1.589 0.924 1.868
10 8.24 0.50 0.993 1.362 0.994 1.565
12 7.92 0.50 1. 055 1.194 1. 073 1. 302
18 8.12 0.50 1.065 0.912 1.130 0.721
25 8.64 0.51 1. 015 0.784 1.112 0.339
35 8.92 0.52 0.982 0.781 1.108 0.111
CONCLUSIONS

A model was presented which is able to predict the kinetics of

Bacillus thuringiensis var. kumamotoensis. To evaluate the
model, its predictions were compared with experimental mea-
surements of yeast kinetics wi th good results. This model
proved to be better than Monod's model due to the fact that its
formulation is based on mass tranfer. Monod's model, based on
the Michaelis-Menten equation for enzyme kinetics, can not
predict the abnormal behavior of our strain.

REFERENCES

1. Rolz,C., Engineering of biological reactions and processes.

center for Scientific and Technological Studies (ICAITI).
Guatemala. 1989.
2. Korte,G., Rinas,U., Krake-Helm,A. and Schurgel,K., struc-
tured model for cell growth and enzyme production by recombinant
Escherichia coli. Appl. Microbiol. Biotechnol. 1991, 35, 185-
188.
3. Bovee,J.P., strehaiano, P., Goma, G. and Sevely, Y., Alco-
holic fermentation: Modelling based on sole substrate and pro-
duct measurement. Biotechnol. and Bioeng. 1984, 24, 328-334.
 MEASUREMENT OF THE MAIN INDUSTRIAL FERMENTATION PARAMETERS
GOVERNING THE PRODUCTION OF BIOINSECTICIDES BY
Bacillus thurinqiensis yare kumamotoensis

MEDRANO-ROLDAN,H., VELAZQUEZ,C.G., ROBLES,C.M., PEREZ-GALINDO,A., CORREA,C.

P., RODRIGUEZ-PADILLA,C.(*), TAMEZ-GUERRA,R.(*) ABO J.L. GALAH-WONG(*).
Instituto Tecnol6gico de Durango, Centro de Graduados e Investigaci6n,
Secci6n de Biotecnologia. Av. Felipe Pescador 1830 ote. 34080 Durango, 090.
Mexico. (*) Facultad de Ciencias Biol6gicas U.A.N.L. Monterrey N.L. Mexico.

ABSTRACT
We decided to study the independent influences of impeller
dimensions, pumping impellers capacity, impeller speed and bulk
dissolved oxygen concentration on the oxygen uptake kinetics of
Bacillus thuringiensis var. kumamotoensis propagated at 20 liter
semi-pilot fermentation plant level and some aspects on rheology
in situ. The mixing times were shown to quantitatively influence
the substrate transport into cell and the impeller parameters
were shown to have strong and independent influences on the
significant oxygen uptake rate. Important data was found about
differences between rheology in situ and in vitro.

INTRODUCTION
Many industrial fermentations utilize several kinds of sub-
strates similar to those used in bioinsecticides production from
Bacillus thuringiensis strain (1,2,3). When Non-Newtonian
viscous behavior occurs, the local cell environment is. only
indirectly related to conditions in the bulk medium.
A logical starting point for analyzing physical influences on
biological activity in the fermenter would be to define the role
of broth rheology to fermenter performance, specially with
regard to the oxygen supply to the cells. On the other hand,
very little is known about the in situ relationships between
shear, mixing, and biological activity in systems such as this.
It is intended that this research contribute to the ultimate
objective of relating the pertinent independent variables, such
as vessel dimensions, impeller dimensions, impeller speed, etc.

MATERIALS AND METHODS

Microorganism: A native strain of Bacillus thuringiensis
characterized as kumamotoensis isolated from dead larvae of
Spodoptera frugiperda was used in this research.
Growth Medium: For preservation Nutritive Agar (Merck) was used
and for propagation at flask and semi-pilot plant level the
composition of the medium was as follows (gil): Cane Sugar
Molasses, 20; Soybean Meal, 20; Corn Steep Liquor, 30; CaC0 3 ,
0.1; Tap Water, 1000 ml; pH, 7.0 (3). All the nutrients were
 547
industrial grade.
Rheoloqy in vitro and in situ: Rheology in vitro was measured by
using a viscometer Brookfield LVT and in situ was based on
measurements of impeller speed and power consumption (3).
Mass Transfer from the Bulk Liquid to the Cells: using the body
of the data describing the influence of the dissolved oxygen
concentration and the impeller parameters on oxygen uptake
kinetics, our observations were related to more fundamental
principles. One approach is to interpret the results in terms of
impeller shear and flow. Both of these can be estimated in terms
of impeller speed, diameter, and blade height (3).
RESULTS AND DISCUSSION
Process Fermentation for the Production of Bioinsecticides:Table
1 shows the kinetics of Bacillus thuringiensis var. kumamo-
toensis. Important data is the value on oxygen demand ( 0.70
g02/lxHR ) which is low compared with others in the literature
( 1.2-1.8 g02/lxHR ) (3).
Rheological Aspects on the Production of Bioinsecticides: The
design of the impeller which leads to different levels of
pumping capacity is related with the mass transfer coefficient
in Figure 1. Physically, impeller pumping could be related to
broth velocity, circulation time, or to mixing time. The
impeller pumping capacity has been postulated to be an important
parameter in the fluid mechanics of fermentation broths.
Results from experiments in vitro (Brookfield LVT on degassed
samples removed from the fermenter) and in situ (Impeller
Power)are showed in Table 2. This Table contains values for the
flow consistency index, the flow behavior index, and the
calculated apparent viscosity projected from the shear stress
vs. shear rate diagrams using both methods. In all cases, the
flow consistency index, is higher for the in situ measurements.
comparing the values of the flow behavior index, in this Table,
the in situ broth also appears to be much more sensitive to
shear thinning than the in vitro broth.
As a final note, the results in Table 2 show that the apparent
viscosity as calculated using the in vitro technique increases
with cell concentration whereas the in situ measurement shows
the reverse relationship so it is our opinion that the in situ
measurement depicts the rheological characteristics of a broth
more realistically.

CONCLUSIONS
1).- Bacillus thuringiensis var. kumamotoensis shows lower
oxygen demand values than others aerobic microorganisms.
2) . - In situ measurements depict the rheological characteristics
of a broth more realistically.
REFERENCES
1.- Abdel-Hameed, A. Studies on Bacillus thuringiensis H-14
strain isolated in Egypt-II. Ultrastructure Studies. World
Jour. of Microbiol. and Biotechnol. 1990. 6, 302-312.
2.- Blokhina, T.P., Variability in Bacillus thuringiensis under
various growth conditions. Mikrobiologiya 53 (3), 427-431.
548
3.- Medrano-Roldan, H., Estudio sobre los parametros de
fermentaci6n de importancia industrial durante la propaga-
ci6n de Bacillus thuringiensis var. kumamotoensis C-4 para
la producci6n de bioinsecticidas. Tesis Doctoral. Fac. de
Ciencias Bio16gicas U.A.N.L. Monterrey, N.L. Mexico. 1992.
TABLE 1.
Effect of agitation rate on microbial kinetics of Bacillus thuringiensis
var. kumamotoensis at pilot plant level

Agitation X Y(xIs) J.I Na Q02 KLA Hp/

Rate (RPM) (gDC/l) (gDC/gS) (HR-') (g02/ lxHR ) ( g02 / gDCxHR) (HR-' ) 1000 1

200 2.7 0.45 0.20 0.54 0.21 179 0.015

300 2.7 0.45 0.20 0.56 0.22 156 0.015
400 2.7 0.45 0.25 0.60 0.24 199 0.016
500 2.8 0.48 0.25 0.65 0.26 216 0.017
600 2.8 0.48 0.30 0.65 0.27 226 0.019
700 3.0 0.50 0.30 0.70 0.28 223 0.019
800 3.0 0.50 0.30 0.70 0.29 235 0.019

TABLE 2.
comparison of rheological properties of the fermentation broth for Bacillus
thuringiensis var. kumamotoensis at pilot plant level

concentration Flow consistency index Flow behavior Apparent

of biomass (g/cm x sec'-n) index viscosity
(gDC/l) (pseudopoises)
in vitro a in situ b in vitro a in situ b in vitro a in situ b

2.0 2 18 0.90 0.60 0.43 1.8

5.0 3.5 45 0.70 0.40 0.68 1.7
8.0 4.5 62 0.65 0.30 0.77 1.60
a) Determined with the viscosimeter Brookfield on degassed samples
b) Determined by measurement of agitation power consumption
-;; O.20-r------------..,..
~1
IS; 0.18
0.16
I: =::]
8
oil
M ::::
0.\0 Figure 1.Effect of impeller pum-
! ping capacity on the liquid-
!;;
1:111 O.llI
solid mass transfer coefficient
0.06
I~ 0.04 •
:i 0.02 +1~T'2~3-.---".--·1I,..........,8~....7~'T8~,..'---110
Impeller Pump~ Capacity (ND: h)
(l/see)
USE OF SWEET POTATO PROCESSING WASTEWATER AS SUBSTRATE FOR FERMENTATIONS

C. A. A. RIBEIRO; M. E. CAVENAGUI; A. E. R. DE PONTES; G. L. BARROS;

A. L. PESSOA; A. C. BARANA; F. R. GUIDOLIN; V. L. DEL BIANCHI; I. O. MORAES
Departamento de Engenharia e Tecnologia de Alimentos
UNESP - Universidade Estadual Paulista
Rua Cristovao Colombo, 2265 - S. J. Rio Preto, S. P. 15054-000, BR

ABSTRACT

The aim of this work is to use, at laboratory scale, wastewaters resulting from sweet potato processing
to obtain different bioproducts. The wastewater is treated in a bubble column to a COD reduction of
92%. This wastewater, rich in sugars and nitrogen, with different concentrations of total sugars, was
added to enrich a semi solid process with soybean bran to produce mold glucoamylase, in both flask
and column fermentation, and as substrate to produce Bacillus thuringiensis cells by submerged
fermentation. Except in solid state column fermentation, satisfactory results have demonstrated that
these wastewaters have a high potential as a substrate for fermentations.

INTRODUCTION

The food industry, besides its products, is a potential producer of wastes with high organic loads, which
must be treated appropriately to reduce this load to admissible levels, before being discarded into
aquatic bodies, to reduce the environmental impact originated by the wastes.
However, these wastes may be useful. This is the case for molasses, corn steep liquor and
manipueira's wastes in submerged processes (1,2) and manipueira and paper processing wastes in solid
state fermentation (2). In this work, the sweet potato processing wastewater was studied as a substrate
for both solid state and submerged fermentation processes to obtain some bioproducts and for
wastewater treatment.

MATERIALS AND METHODS

For the wastewater treatment a bubble column reactor of 5 liters (9 X 9 X 62 cm) was used,
inoculated with activated sludge from the milk processing industry. After a sludge acclimatization
time, different residence time (18, 24 h), cell concentration (1500, 2600, 3330 mg MLVSS/I) and COD
affluent concentration (1200, 1500, 1850 mg COD/I) were used. The aeration rate was about 1 v.v.m.
The COD reduction was measured by potassium dicromate method.
To the glucoamylase production by flask fermentation, 5 g of soybean meal in 250 ml Erlenmeyers
flasks were used, and then humidified with tap water and sweet potato processing wastewater (100,
500, 1000, 3000 mg COD/l) at 1:1 (w/v), 37°C for 60 h. The enzymatic activity was measured by
the Somogyi method.
550
To the glucoamylase production by column fermentation 16 g of soybean bran in a packed column
of 5 cm diameter and 10 cm height were used and then were humidified with tap water and sweet
potato processing wastewater (1000 mg COD/l) at 1:1 (w/v), ambient temperatures for 48 h. The
enzymatic activity was measured by the Somogyi method.

To the Bacillus thuringiensis production 100 ml of molasses/urea medium and sweet potato
processing wastewater (500, 1000, 2000, 3000 mg COD/I) in 500 ml Erlenmeyer flasks were used at
30°C for 48 h with constant agitation. The cellular growth was measured by optical density.

RESULTS

Table 1 shows that, at same fermentation conditions, the cell concentration and residence time are the
prior factors influencing the COD removal. Over 24 hours, this gives an average value of 92%.
Table 2 shows that sweet potato processing wastewater can be used as a substrate to produce
glucoamylase instead of water at a COD concentration of 1000 mg/1 after 36 h. At this point,
wastewater was 10% better than water as humidifier.
Table 3 shows that, in solid state fermentation column reactor, sweet potato processing wastewater
did not achieve good results. Water humidification was at least 37% better than the wastewater.
Table 4 shows that sweet potato processing wastewater can be used as substrate to produce Bacillus
thuringiensis cells instead of water at a 3000 mg CODII at the end of 60 h of fermentation. At this
point, the wastewater was 68.5% better than molasses/urea as substrate.

Table 1 Average COD affluent, COD effluent and COD

removal data for each cell concentration at 18/24 h
[X] (average value) (mg MLVSS/l)
A verage value 1522 2500 2597 3314
COD affluent (mg/l) 1300 1534 1805 1862
COD effluent (mg/1) 300 217 143 159
COD removal (%) 81 86 91 92
Residence time (hours) 18 18 24 24

Table 2 Enzymatic activities variation at different COD concentrations

Fermentation time (h)
18 24 30 36 48 54 60
Water 0.897 0.541 0.327 0.359 0.669 0.602 0.774
COD 100 0.863 0.607 0.316 0.534 0.692 0.263 0.592
COD 500 0.703 0.722 0.186 0.916 0.617 0.300 0.502
COD 1000 0.802 0.723 0.163 0.991 0.653 0.035 0.215
COD 3000 0.705 0.785 0.028 0.802 0.647 0.143 0.223

Table 3 Enzymatic activities variation with

water and sweet potato process wastewater as
a humidifier
Fermentation time (h)
12 24 36 48
Water 0.273 0.308 0.325 0.285
Wastewater 0.109 0.162 0.179 0.105
 551
Table 4 Cellular growth at different COD concentration
Fermentation time (h)
18 24 30 36 48 54 60
Molasses/urea 0.102 0.213 0.254 0.325 0.368 0.424 0.597
COD 500 0.049 0.110 0.140 0.140 0.145 0.150 0.213
COD 1000 0.117 0.130 0.168 0.188 0.274 0.287 0.319
COD 2000 0.175 0.260 0.287 0.304 0.332 0.340 0.454
COD 3000 0.143 0.451 0.552 0.601 0.725 0.753 0.871

CONCLUSIONS

It was observed that sweet potato processing wastewater has a good potential as a fermentation
substrate in both solid state and submerged processes.

REFERENCES

1. Moraes, I. 0., Studies of submerged fermentation to the production of biological insecticides in lab
scale. PhD thesis, FEEA - UNICAMP, Brazil, 1976.
2. Capalbo, D. M. F.; Moraes, I. 0., A simple culture medium for Btproduction. 8th World Congress
of Food Science and Technology, Toronto, Canada, p. 326, 1991.
 FERMENTATIVE PRODUCTION OF L-LACTATE FROM XYLOSE

Tomoko Ueda, Kenji Tanaka and Ayaaki Ishizaki.

Department of Food Science and Technology,
Faculty of Agriculture, Kyushu University,
Hakozaki 6-10-1, Higashi-ku, Fukuoka-shi 812, Japan

ABSTRACT

L-Iactic acid fermentation employing Lactococcus lactis strain 10-1 (JCM7638) was examined
on xylose, and glucose and xylose media. The yield of lactate on xylose was 0.47g lactate/g
xylose at an initial xylose concentration of 51.2g/1 and the ~max was 1.02 lIh. Xylose culture
was more susceptible to lactate inhibition than were glucose cultures but showed similar kinetic
behavior. The microorganism was capable of complete sugar utilization when grown on a
mixture of20gl1 xylose and 20gl1 glucose and synthesized 0.66g lactate/g sugar.

INTRODUCTION

Biological conversion of xylose into simple chemicals creates an interest in order to utilize
lignocellulosic biomass. L. lactis 10-1 is a homolactic L-lactic acid producing bacterium isolated
and characterized in our laboratory. We have demonstrated that 10-1 is capable of high
conversion rates of glucose to L-lactic acid, with only slight production of volatile fatty acids,
but is subject to end-product uncompetitive inhibition (3). L. lactis 10-1 utilizes xylose as
carbon source too, and so we have studied the fermentative production of L-Iactic acid from
xylose (1,2).

MA TERIALS AND METHODS

Medium
The basal medium consisted of (per litre) 5.0g sodium chloride, 5.0g polypepton (Nihon
Seiyaku, Co. Ltd., Japan) and 5.0g yeast extract (Difco Laboratories, U.S.A.). The medium
was supplemented with up to 51.2gl/ xylose and 20g/l glucose. The same basal medium was
used for inoculum preparation.

Culture conditions
Batch cultures were carried out in a l-L fermenter (working volume; 400cm 3), agitated at
400rpm without gas feed. Temperature was controlled at 37°C by a circulating water bath and
the pH was controlled at 6.0 by automatic addition of2N NaOH.
Analysis
Cell growth was monitored by absorbance at 562nm and converted to dry cell weight. Residual
 553
glucose and L-lactate concentrations in cell-free broth were assayed with a YSI model 23A
glucose analyzer and a YSI model 23L lactate analyzer, respectively (Y. S. I. Co. Ltd.,
U.S.A.). Residual xylose in cell-free broth was assayed by using the method of Somogyi-
Nelson to estimate the total reducing sugar content and compensating for the concentration of
glucose in mixed substrate fermentation.

RESULTS AND DISCUSSION

Figure 1 shows the fennentation profiles of L. lactis 10-1 on 51.2g/1 initial xylose. The yield of
lactate in xylose culture was 0.4 7g lactate/g xylose in comparison with a yield of 0.88g lactate/g
glucose in glucose culture (3).

-- ---
60 1.0

-
t::::: t:::::
bIl 0.0

--
bIl
(J-

+===
(J 0
..... 40
• DCW
-1.0 .J:l
~ A. xylose .....
bIl

--
•
<I)
'Of lactate
==- -2.0 ~
~ ==
<I) <I) < I)
I"'-l (J (J
o0== 20 -3.0

-'"
_
>,(J
~'O ..... -4.0
'0
(J
~
==
0 -5.0
0 20 40 60
Time (b)
Figure 1. Fermentation profile of L lactis 10-1 on 51.2g/1 initial xylose.

12

----
.c: 8
•
-
24 gil
~
A 51 gil
~

4 • 92 gil

o
o 2 4 6 8
Lb (gil)
Figure 2. The effect ofL-lactate concentration on 1/J! in xylose batch culture. These data were
obtained from batch cultures in which the initial xylose concentrations were 24, 51 and 92g/1.
554
L. lac/is 10-1 was grown in batch culture on initial xylose concentrations of 24, 52 and
92g/1 (1). Viable cells were counted to determine the specific growth rates during lactic acid
accumulation phase in the xylose cultures and Figure 2 represents the relationship between the
reciprocal of these specific growth rates and L-Iactate concentration. The straight line shown in
Figure 2 conforms with the relationship predicted by equation [1] with ~max of 1.05 IIh and Ki
of 1.8611g.
~=1.05 1(1+ 1.86 Lb) [1]

30r-----------------------------, 1.0
-..
0.0
c:::::
bll
.....
'-'"

• DCW
-1.0
.....d
bll

-
Q)
A glucose ~
-2.0
.. xylose
Q)
• lactate ()
-3.0
...
>.

-=
~
-4.0

-5.0
10 20 30 SO
Time (h)
Figure 3. The growth, L-Iactate production and substrate consumption of L lac/is 10-1 on
mixed medium of 20gll glucose and 20 gil xylose.

Tyree e/ al. (4) demonstrated that the ability of Lactobacillus xylosus to utilize xylose
was markedly inhibited by only 2g// glucose and was completely inhibited at glucose
concentrations greater than 5g//. The ability of L lactis 10-1 to utilize a mixture of20g/1 xylose
and 20g/1 glucose is illustrated in Figure 3. Although glucose was used at a higher rate than was
xylose in the early stages of the culture, both sugars were utilized simultaneously. Mter the
glucose concentration declined. below 0.49g//, the utilization rate of xylose increased
significantly and there was a suggestion of a diauxic relationship in the data. The overall yield
of lactate was 0.67g lactatelg sugar, intermediate between the values obtained for the single
sugar fermentations.

REFERENCES
1. Ishizaki, A, Ueda, T., Tanaka, K. and P. F. Stanbury, L-Lactate production from
xylose employing Lactococcus lactis 10-1. Biotechnol. Lett., 1992,14,7,599-604.

2. Ishizaki, A, Ueda, T., Tanaka, K. and P. F. Stanbury., The kinetics of end-product

inhibition ofL-lactate production from xylose and glucose by Lactococcus lactis 10-1.
Biotechnol. Lett., 1993,15,5,489-494.

3. Ishizaki, A, Ohta, T. and Kobayashi, G., Batch culture growth model for lactate
fermentation. 1... Ferment. Bioeng., 1989,68,2, 123-139.

4. Tyree, R. W., Clausen, E. C. and Gaddy, 1. L. The fermentative characteristics of

Lactobacillus xylosus on glucose and xylose. Biotechnol. Lett., 1990,12,1,51-56.
 OPTIMIZATION STUDIES OF A LACTIC ACID FERMENTED BEVERAGE
PROCESS

TINA MOE & JENS ADLER-NISSEN

Center for Food Research at DTH, Dept of Biotechnology,
Technical University of Denmark, B-221, 2800 Lyngby, Denmark

ABSTRACT

The present work is concerned with a process for preparation of a lactic acid fennented
beverage based on soyroilk and cereals. Considerations on re-design and optimization of the
process is made using new approaches. The intention is to develop an optimization strategy
as a supplement to traditional design strategies known from the chemical industry. The
principal idea is that the process should be optimized as a whole by optimizing key unit
operations in priority order upstream.

INTRODUCTION

Lactic acid fennentation of cereals, legumes and green vegetables is applied nearly all over
the world in traditional food processes based on experience and good craftsmanship. In the
recent years there has, however, been a strong trend to develop some of these processes into
modern, large-scale food biotechnology processes. Due to the often poor understanding of the
actual chemical changes - or biotransfonnations - taking place and their inherently complex
nature the traditional process design methodology known from the chemical industry is much
too rigid to be sufficient for the optimization of these food biotechnology processes.
The aim is therefore, to develop a tool which can facilitate the decision making on the
optimization and re-design aspects as well as on which further investigations to carry out on
the process. The strategy is based on the proposed concept of a principal process step. The
most significant parameters for e.g. product quality are identified and an order of priority for
optimization of the individual parameters is established. The parameter of most importance
for the product e.g. taste is focused upon at first The optimization is then initiated at the
process step which has the highest impact on this parameter - the principal process step. If
output of critical components, such as organic acids, carbohydrates and soluble proteins, can
be related successfully to input, a cut can be made in the flow sheet and each half of it can
be redesigned separately.
556
OPTIMIZATION STUDIES

The Fermented Beverage Process

The strategy for optimization is currently tested on a non-commercial process for production
of fennented soymilk:. The process begins with a soyrnilk: preparation. Rice flour and barley
malt are thereafter added and a mashing process is carried out to give fennentable
carbohydrates, sweetness and nutritional value. To ensure a high content of soluble proteins
in the beverage after the lactic fermentation an enzymatic proteolysis step is included
concomitant with the mashing. Fmally the lactic fennentation brought about by a selected
Leuconostoc mesenteroides strain [1] gives the beverage a pleasant fresh or apple-juice-like
flavour. Being a combination - or rather addition upon addition - of separate technologies, the
process from the outset presented a challenge with respect to optimization.

The Fermentation as Principal Process Step

Probably the most important parameter in food processing is the palatability of the product.
In this process the palatability depends to a great extent on the fennentation step which was
therefore selected as the first principal step [2]. From studies by Chung [3] it was known that
a balanced fonnation of lactic and acetic acid during fennentation combined with a correct
level of residual sugars was a prerequisite for an acceptable taste. The taste is therefore
influenced by the process steps upstream, especially the mashing step which is essential for
both the supply of fennentable sugars to the fennentation and for the supply of maltose to
give sweetness. The mashing was therefore the natural choice as the next principal step.

The Mashing Step as Principal Process Step

The significant part of the mashing step output is a certain combination of fennentable sugars.
The question was in what way this was related to the mashing step input. At a fIxed mashing
temperature-time profIle the most crucial parameters in the input were found to be the
malt:rice ratio and the initial pH at mashing.
Mashing is a process well known from the brewing industry where the process conditions
malt:unmalted cereal ratio of about 60:40 (% w/w) and an initial pH of 5.3 are typical.
However, one may prefer to keep the amount of malt used low (due to availability, cost etc.)
and a pH of 5.3 could not be achieved naturally in this process.
The consequences by moving away from the well known optimal conditions were
investigated by response surface experiments. The effect of rnalt:rice ratio and initial pH on
the amount of fennentable sugars is shoWn in figure 1. Similar figures for the specific mono-
and disaccharides were achieved (results not shown). The results showed that the
concentration of glucose varied with varying process conditions while the concentration of
maltose did not. The results gave the correlations how to adjust the sugar concentration (and
combination) to the demands of the fermentation by changing pH or malt concentrations. Or
the other way around, what consequences restrictions in pH or malt use will have on the sugar
combination and then further on the residual sweetness of the beverage.
Typical yields of fennentable sugars in the brewers mashing are around 75 %. The figure
shows that the losses derived by mashing at non-conventional process conditions are relatively
low. It can be concluded that these small losses should be without significant influence on
the process optimization especially as the non fennentable carbohydrates are carried through
the process and give nutritional value to the fmal product. The advantage of the partial up-
stream optimization is obvious by a change in starch source. Only a few extra experiments
of mashing will be needed to know the sugar spectrum and the impact upon the product will
then be predictable.
 557

:c
a.
o
c.O

30 50 60 70
Percentage of malt in the cereal-malt mixture of mashing (% w/w)

Figure 1. Response surface plot for yield of fennentable sugars (% of total available
carbohydrates) as function of malt concentration and pH.

The pH of the beverage changes during the mashing step and the change is found to
depend on the initial mashing pH. The correlation between the two pH-values was determined,
since it is of importance for the choice of optimal pH and pH adjustment.
Energy considerations based on the temperature-time proftle of the process led to a
number of considerations of optimization through re-design. The evaluation of the effect of
the change in process flow sheet was much facilitated by the knowledge achieved by the
above drafted approach.

CONCLUSION

The development and test of an up-stream strategy for optimization of a food biotechnology
process by focusing on the. crucial product parameters is still in its innovative phase. The
mashing example is given 'as an illustration of the advantage of the partial up-stream
optimization achieved by the concept of a principal step.

REFERENCES

1. Lee, C.H., Min, K.C., Souane, M., Chung, M.-J., Mathiasen, T. and Adler-Nissen, J.,
Fennentation of prefennented and extruded rice flour by the lactic acid bacteria from
Sikhae. Food Biotechnology, 1992, 6, 239-255.
2. Adler-Nissen, J. and Moe, T., Scale-up studies for the production of high protein content
lactic beverage. In book of proceedings from UNIDO International Workshop on Lactic
Acid Fermentation of Non-Dairy Food and Beverages, Korea Food Research Institute,
Songnam, June 25-27, 1992, pp. 149-161.
3. Chung, M.-J., Flavor formation during lactic acid fennentation of rice-soymilk
mixture. In book of proceedings from UNIDO International Workshop on Lactic
Acid Fennentation of Non-Dairy Food and Beverages, Korea Food Research
Institute, Songnam, June 25-27, 1992, pp. 48-57.
GENERAL METHOD FOR LACTIC ACID BATCH FERMENTATION PROCESSES
MONITORING

ERIC LATRILLE, DANIEL PICQUE and GEORGES CORRIEU

Laboratoire de Genie des Procedes Biotechnologiques Agro-Alimentaires
Centre de Biotechnologies Agro-Industrielles
LN. R. A.
F-78850 Thiverval-Grignon, France

ABSTRACT
The determination of kinetics by measuring pH and electrical conductivity changes in fermented
milk permits to definite feature points. These points are very useful to characterize the
fermentation state and its well behaviour. Predictions of the future pH values over a long time
are obtained with a good accuracy by using neural networks models and geometrical methods.

INTRODUCTION
It is of value to determine the changes in the kinetics of certain physicochemical parameters
during a thermophilic lactic acid fermentation, e.g. pH or the electrical conductivity of the
medium.
Yuguchi et al.[1] established correlations between the rate of pH changes and the viscoelastic
properties of yogurt. The same authors [2] proposed a positive correlation between the mean
particle size of yogurt gel and the fermentation rate, determined by measuring pH changes in
the milk. In the case of lactic acid fermentations using L. plantarum, Lievense et al. [3] showed
a positive linear correlation between inoculation size (or activity related to the physiological
state of the bacteria) and the maximal rate of pH decrease. Using Streptococcus thermophilus,
Spinnler and Corrieu [4] reported a linear relationship between the time at which the maximal
acidification rate appeared and the logarithm of the inoculum size. In addition, the absence of
simple and identifiable models in the case of batch fermentations for accessing biological
parameters (biomass, substrate and metabolite concentrations) requires a different approach
when interpreting the measurements.
The shapes of the pH change and electrical conductivity curves can be compared and classed
by determining the characteristic points of the recorded curves [5]. These characteristics are
most often points of inflection and changes in the curvature of the curve, which correspond to
local minima and maxima of first derivatives vs. time.
Moreover, in batch processes, it is often very important and useful to know and to predict
on-line the evolution of key data characteristics versus time. However, the attainment of this
objective is not always possible due to the lack of reproducibility of the process. During the
production offermented milk, the initial pH ofthe medium is in the vicinity of 6.4 and it decreases
to pH values as low as 4.5. Each type of fermented milk, available on the market, has its own
characteristic final pH. The final pH is a feature of the dairy product and it is determined by the
manufacturer. In a plant where many fermentations are performed simultaneously, the
prediction, in real time, of the final time of each fermentation is very important to schedule the
use of the heat exchangers and of the fermentors.
 559
Some trials to use neural networks for on-line prediction in fermentation were done recently
[6,7].
First, we will describe a method for the real-time determination of the fermentation kinetics
of pH and conductivity [8].
In second part, we propose different methods, based on neural networks and on geometrical
approaches applied to lactic acid batch fermentation. The purpose is to enhance the quality of
the prediction of future pH values and to determine, in real time, the final time at which a
predetermined value of pH is obtained [9].

RESULTS AND DISCUSSION

Kinetics and interpretation of feature points in terms of stages of fermentation

The on-line measurements of pH and electrical conductivity data have made it possible to access
time and rate feature points of thermophilic lactic acid fermentations. The first derivative (dxldt)tn
at instant 1n used the central difference method [10] defined as :

(dx)
dt In
= X'n+l - x'n_}
tn+} - tn_}
where x is the measurement of pH or conductivity, t the time and n an index of time sampling.
Ten feature points characterize curves of acidification and of conductivity changes using the
main points of inflection observed (see figures 1 and 2). These feature points have shown the
excellent reproducibility of fermentations conducted in standard conditions, with coefficients
of variations lower than 5.1 % for nine feature points.
The curve dO/dt = f(t) is always positive, with two local maxima and one local minimum, shown
by the presence of three points of inflection on the curve of conductivity changes with time. The
first maximum tOl reflects the urease activity of S.thermophilus when the second maximum
t03 reflects the increasing production oflactic acid by S.thermophilus and L.bulgaricus. These
feature points can also been used to compare the effect of type of starter, temperature and culture
medium. The temperature effect results in different urease activity and acidification optima.
The presence of fat (32 gil) in the medium does not change any feature point, while the presence
of sucrose (90 gil) results in a decrease in the acidification rate and a longer fermentation time.

I
I: '\ _." -I
7.0 0.000
dpw&'. ...
9.0 r-----Ar-_-_-_-_-_-_-_-
__ ~ 0.025
I:
6.5
'.
... 'iI
''.\ ,., ... -0.005
~
pH '. ",'
.1---- ::.::
6.0
''.
~,*Hldl)2 .e
! - \ ,I"
~
.0.010 0.015
\.r
I

i
: I
~ 7.5 I \. ~
j
\ I
!
I .@'
5.5 : I
: I
f
I
I .. f--I--~~Q2
I I ..
'f
:1 I
:1 I .0.015 ~ O.ol0..a
s.o \1 I
8 1 u ~
\1
:1
'I ,I-
I
I
i c:>
U
G II I
I I
I
..
... ..

!
~

4.5
V
F----+-
I ·0.020 1.. 6.5
I I 0.005
• I I "
I <d!tHldl)1 ::.::a.. dGlclt t/ I I ::I
IpHl
I
I 1pU2 I ./~'J 1G1~211G31 1
4.0 .............,r----r-""-r---+-... -0.025 6.0 L..:£.··-r--r--I-......L.T""".l..1 ....,.--1 0.000 8
o so lOCI ISO 200 2SO o SO lOCI ISO 200 2SO
Fennentation time (min.) Fennentation time (min.)

FIGURE 1 : pH and pH evolution rate (dpWdt) FIGURE 2 : Conductivity G and conductivity evolution rate dG/dt
changes versus fennentation time and detennination of changes versus rennentation time and detennination of
feature points. Skim milk (13% w/v), temperature 44 C, feature points. Skim milk (13% w/v), temperature 44 C,
starter (S.thennopbilus & Lbulgaricus) starter (S.thennopbilus & Lbulgaricus)
560
pH prediction and final fermentation time determination
Lactic acid batch fermentations, with uncontrolled pH on skim milk, can be modelled with a
feedforward neural network to generate the model (pH versus time) of a reference fermentation
(figure 3). In essence, the feedforward neural network is used as a general nonlinear model to
store the information of a series of well-behaved fermentations. This neural network stores the
information of previous fermentations and defines a reference fermentation. The distinct
advantage of neural networks over other class of models is the plasticity of its structure which
allows to easily capture the shape of the fermentation curve. This reference fermentation is used
to perform a comparison with the actual fermentation for the on-line prediction of future pH
values and of the final fermentation time. This time, which occurs at a predetermined pH (pH=4.6
for instance), is predicted with an accuracy of less than 20% at the onset of fermentation and
with a much better accuracy as the fermentation proceeds. The future pH values are predicted
with a mean error of 0.05 pH for a 3 hour prediction horizon. The prediction is obtained with
four geometrical methods by sliding the curve of the reference fermentation along the curve of
the actual fermentation.
mddenlayer
with 4 nodes

temps

bias = 1

FIGURE 3: Example of neural network structure used

to generate a reference fermentation curve.

CONCLUSIONS
During batch lactic acid fermentation, the measures of the conductivity changes permit to clearly
distinguish between the urease activity and the acidification activity of the starter. Feature points
are very useful to characterize the fennentation state and its well behaviour. Predictions of the
future pH values over a long time are obtained with a good accuracy by using neural networks
models and geometrical methods. We could hope to derive a more general fermentation reference
by including the effect of temperature, for instance.

REFERENCES
1. Yuguchi H., Tanai S., Iwatsuki K. and Okonogi S.: The influence of fermentation velocity on the properties
of yogurt. .lim. L Zootecb. &, 1989, 60, 742-746.
2. Yugucbi H., Tanai S., Iwatsuki K. and Okonogi S.: The influence of fermentation velocity on the curd
particle size and properties of yoghurt. lim. L Zootecb. &, 1989, 60, 619-926.
3. Lievense, L.C., Van't Riet, K., and Noomen, A.: Measuring and modelling the glucose-fermenting activity
of Lactobacillus plantarum. Alml. Microbiol. Biotechnol.. 1990,32,669-673.
4. Spinnler, H.E., and Corneu, G.: Automatic method to quantify starter activity based on pH measurement.
L~~, 1989,56,755-764.
5. KonstantinovK.B.,andYoshidaT. : Real-Time Qualitative Analysis of the Temporal Shapes of (Bio)process
Variables:AICbEl!mmal.1992, 38,1703-1715.
6. Cltran, Y., Thibault, J., Cherny, A. and Corneu, G. : Comparison of prediction performances between
models obtained by the Group Method of Data Handling and Neural Networks for the alcoholic fermentation
rate in enology. L Ferm. ~ 1991,71,356-362.
7. Van Breusegem, V., Thibault, J. and Cberny, A. : Adaptive neural models for on-line prediction in
fermentation. Qm.LQfCbem.. Eng .. 1991, 69, 481-487.
8. Latrille, E., Picque, D., Perret, B. and Corneu, G. : Characterizing acidification kinetics by measuring pH
and electrical conductivity in batch thermophilic lactic fermentations. LEmn. Bioeng .. 1992,74 (1), 32-38.
9. Latrille E., Corneu G. and Thibault J. : pH prediction and fmal fermentation time determination in lactic
acid batch fermentations. ~.to Computers &0 Chemical EnBineering. 1993, 17, S423-S428.
to. Siano, S.A., and Mutbarasan, R.: NADH fluorescence and oxygen uptake responses of hybridoma cultures
to substrate pulse and step changes. Biotechnol. ~ 1991,37,141-159.
 ANAEROBIC CONTINUOUS ETHANOL FERMENTATION USING A
COMPUTER-COUPLED MEDIUM FEEDING SYSTEM WHICH
HAS DDC CONTROL pH OF THE CULTURE BROTH

SUDARUT TRIPE~CHKULl, MICHIO TONOKA~Al, AYAAKI ISHIZAKI 1 ,

ZHONGPING SHI AND KAZUYUKI SHIMIZU
1. Department of Food Science and Technology, Faculty of
Agriculture, Kyushu University, Fukuoka, Japan 812
2. Department of Biochemical Engineering and Science,
Kyushu Institute of Technology, Iizuka, Fukuoka, Japar.
820

ABSTRACT
The utilization of pH changes to set the rate of addition of
fresh medium to a continuous ethanol fermentation has been
investigated. The dilution rate and cell mass concentration
depended on the pH set point and the control interval. The
dilution rate, however, is independent of the difference in
pH between the culture broth and feed medium.

INTRODUCTION

The pH-stat control used the pH change caused by growth to

set the rate of addition of medium. This method was intro-
duced for the mass cultivation of bacteria (1,2,3). Larrson
et al.(4) also used the pH-stat as a tool for studying micro-
bial dynamic in continuous fermentation.
This report aims to investigate how pH-stat can be used
to control feed rate of medium. The parameters that affected
automated substrate supply in pH-stat modal feeding technique
in continuous ethanol fermentation using Zymomonas mobilis
have also investigated.

MATERIAL AND METHODS

Microorganism and media

Zymomonas mobilis NRRL-B-14023 was used in this study. The
growth medium was composed of 100 gil glucose, 10 gil yeast
extract, 1 gil KH2P04' 1 gil (NH4)2S04 and 0.5 gil
MgS0 4 .7H 2 0. Composition of feed medium is the same as the
growth medium.
562

Continuous culture
Continuous culture was performed in 1 1 jar fermentor with a
working volume of 500 ml at 30°C and pH 5.50. When the con-
tinuous procedure was switched on, an additional glucose was
intermittently supplied by pH-stat with separate inflows of
alkali and feed medium. The fermentor was connected to a
computer where the medium flow rate and the control interval
were registered.

Analytical methods
Ethanol was estimated with a gas chromatography GC-8 APE .
Glucose concentration was determined with a glucose
analyzer . Biomass was expressed as dry cell weight calibrat-
ed to an optical density of 562 nm using a spectrophotometer.

RESULTS AND DISCUSSION

The pH-stat function by using the pH of medium to maintain

growth at a constant rate and its principle is based on a
growth associated production or consumption of acid or based
(4). The acid production rate can be related to the specific
growth rate under a balanced growth of Z. mobilis in batch
culture of ethanol fermentation (Data iSI10t shown).
Larrson and Enfors(4) reported that the steady-state
biomass concentration could be controlled by the flow quo-
tient between the separated alkali/total flow to the fermen-
tor. Therefore the pH-stat cultivation with separating in-
flows of medium and alkali was operated. By means of this
method, flow rate of medium and alkali is controlled by
metabolic activity of microorganism. Considerable improve-
ments of techniques are need for the system to prove of
practical value. Thus the parameters that affected automated
supply in pH-stat for the control of pH growth rate by regu-
lating the feed rate have been also investigated.
The results suggest that pH set point affected cell mass
concentration and dilution rate (Table 1). The shorter con-
trol interval achieved, the higher medium flow rate and the
bigger dilution rate (Table 2). Increasing in biomass yield
occurred with increasing pH in feed medium. However, the
dilution rate was not effect at all (Table 3).
From all of these results the use of pH mediated computer
in continuous ethanol fermentation can be considered as a
tool to control the feed rate of the medium. To developed
this method much remains in the optimization of the parame-
ters involved in the method such as the buffer capacity of
the inflow medium should be investigated.

REFERENCES

1. Martin, G.A. and Hemfling, W.P., Method for the regulation

of microbial population density during continuous culture at
high growth rate. Arch. Microbial., 1976, 107, 41-47.
2. Stouthamer, A.H. and Bettenhaussen, C.W., Energetic as-
pects of anaerobic growth of Aerobacter aerogens in complex
 563

medium. Arch. Microbiol., 1976, II, 21-23.

3. Oltmanm, L.F., Schoenmaker, G.S., Reijnders, W.N.M. and
Stouthmer, A.H., Modification of the pH-stat culture method
for the mass cultivation of bacteria. Biotechnol. Bioeng.,
1978, 20, 921-925.
4. Larrson, G. and Enfors, S.o., The pH-auxostat as a tool
for studying microbial dynamics in continuous fermentation.
Biotechnol. Bioeng., 1990, 36, 224-239.

TABLE 1
Effect of pH set point for feed medium into the fermentor
(pH 5.50) on the kinetic parameters of the pH-stat cultiva-
tion
pH set point

5.50 5.52 5.55

Specific glucose consumption rate(g/g/h) 7.27 4.21 0.81
Specific ethanol £roduction rate (g/g/h) 3.67 2.06 0.008
Dilution rate (h- ) 0.16 0.08 0.025
Cell concentration (g/l) 2.18 1.93 2.90
Biomass yield (g/g) 0.022 0.019 0.033
Ethanol yield (g/g) 0.49 0.48 0.49
Ethanol productivity (g/l/h) 8.0 4.0 1. 25
TABLE 2
Effect of the control interval of pH value measurement on
the kinetic parameters of the pH-stat cultivation

Control interval
60 s 5 s

Specific glucose consumption rate (g/g/h) 7.27 8.07

Specific ethanol £roduction rate (g/g/h) 3.67 4.27
Dilution rate (h- ) 0.16 0.21
Cell concentration (g/l) 2.18 2.53
Biomass yield (g/g) 0.022 0.026
Ethanol yield (g/g) 0.49 0.49
Ethanol productivity (g/l/h) 7.3 10.8

TABLE 3
Effect of pH difference between the culture and the inflow
medium on the kinetic parameters of continuous ethanol fermen-
tation using ~ mobilis in 100 gil glucose medium

pH of feed medium
5.50 5.55 6.0

Specific glucose consumption rate(g/g/h) 9.61 8.92 8.59

Specific ethanol £roduction rate (g/g/h) 4.49 4.24 4.08
Dilution rate (h- ) 0.25 0.25 0.25
Cell concentration (g/l) 2.27 2.50 2.76
Biomass yield (g/g) 0.026 0.028 0.030
Ethanol yield (g/g) 0.48 0.48 0.49
Ethanol productivity (g/l/h) 10.2 10.6 11.3
 METHOD FOR ON·LINE PREDICTION
OF THE ALCOHOLIC FERMENTATION RATE IN WINE·MAKING

B. PERRET and G. CORRIED

INRA - LGPB A - F-78850- TIUVERV AL-GRIGNON
ABSTRACT
In oenology, the alcoholic fennentation rate is proportional to the outlet CO 2 gas flow rate. In isothennal conditions,
this process is characterized at the onset of the fennentation, by a constant increase (dCOz ! dt = KI.t), and then
by an exponential decrease (dCOz ! dt =e·KZ .t) of the fennentation rate with the time. The intercept of these two
phases detennines the maximum fennentation rate (Vm).
The method for the on-line prediction of the alcoholic fennentation rate consists in :
1at : detennining the acceleration coefficient (K 1) at the beginning of the process (five hours after the CO2 emission).
2nd: using KI in simple mathematical models, to detennine Vm, K2 and tf (end time of fennentation).
This method can be improved if the experimental measurement of Vm is taken into account. In this case, the
accuracy of K2 and tf detennination felt respectively to 4% and 7% instead of 7% and 8%.
I DESCRIPTION OF THE FERMENTATION KINETIC
The alcoholic fennentation rate is computed from the emitted CO2 volume measurement (1,2). In isothennal
conditions, (usual temperatures within 15 and 30T), the fennentation kinetic is characterized by three successive
phases (Figure 1) :
- onset of fennentation : lag phase of cell growth and progressive saturation of the medium with CO 2, The end of
this phase is characterized by the time (td) of onset of gas release.
- a phase of linear increase of the fennentation rate characterized by the acceleration coefficient Kl
(dC02! dt = KI.t). This phase goes on until the maximum fennentation rate Vm (Vm = (dCO z ! dt),tla.) is obtained.
tvrn is the corresponding fennentation time ..
- a phase where the fennentation rate decreases according to an exponential rule. It is characterized by a decrease
constant K2 and it leads to dCOz! dt = Vm.e·K2 .(,.tVm>.

0.4
1 3
v~
0.35

0,.3

=
:os.. 0.25

~ 0."

g 0.16

~
I
0.1

0.06

100 200
FERMENTATION TIME (h)
Figure ~ : typical 'f.rmentation rat_ evolutiCt ......

Phase I : any effective release of carbon dioxide

Phase 2 : linear increase of the rate of carbon dioxide production
Phase 3 : exponential decrease in the rate of carbon dioxide release.
 565
Furthennore, the CO 2 emission measuremelll allow the detennination of the produced ethanol and the
residual sugar concentrations as functions of time (3). The last one, in regard to the initial sugar concentration So,
leads to the final fennentation time (If). For more convenience, we have used a normalized fennentation time: mf
=(tf-td)/S o'
II - MODELLING THE )<'ERMENTATION KINETIC
The real time predictive modelling of the fermentation kinetic consists in the K1 detennination in the frrst hours
following the onset of the alcoholic fermentation. For that purpose, a linear regression of the values of rate of CO 2
production is executed. The K1c computed parameter is then used to predict Vmc' K2c and tfc using relationships
of table 1.
The adaptation of this method to 14 tests realized on a pilot fermentation plant (700 liters and 5000 liters tanks)
allowed the calculation ofK1c and the determination of Ville, K2c and tfc' The mean errors were respectively 7.5%,
6%, 7% and 8% (Figure 2).
A significant improvement is obtained if the K1c determination is replaced by the experimental measurement of
Ym. This maximum fennentation rate is reached early enough regarding the total process time: at about 25 to 30%
of the fennentation progress. The models involving Ym to compute K2c and tfc are linear (figure 3). The mean
errors are 4% for K2c and 7% for tfc' The corresponding relationships are shown in table I.

r
c
0
1.0 C\I
~
0,020
' iii I,'

~ "'~
£
e
0.018

"
1,_

~
'"
0,' G,OI6 Q)
~

(]) iii 0,01. E 1,

§ 0,'
C
0
U
0,012 2 1,'

C 0,010
0 01
:.e g o
0
'E
N~
0,_ C
E 0.0"

'" ::S.
'(jj 0,6

'"
:J- o Expermental dara 0,'" o Experimenlal dara o Expermenral data
E ~ - Model ~ - Model "'0 0,4 - Model
'xtil
E
~
0,2
u~
Q)
0,004

~ 0,002 ~
C <Il
E ',2
::?;~ 0,0 0 00,1)00 (/) ~ o,o~-,--.,.---,---.--..---I
',00 ',02 0,0. ',06 '.00 0,10 0,12 ',00 0,02 0,04 ',06 ',08 0,10 0,12 0,00 0,02 G,04 O,Of> 0,01:1 ,10 1),1.?
COMf'UTEDACGflERA11QN COEFfIQENT K'C(""_1'Ii GQMPUTE04CCUEAATIQNCOEFFIC1ENT K1cllJl.h) COMPUlED ACCB. ERA I ION GQEFF!CIENT _... 11.(&1 hj

Figure 2 : determination of Vm,K2 and tnf from KIe

C
o
0,(120
.~
,,',.----------,
e e
:::~
0,018
Q)

'"
iii
0,016
E
C
0
O,DU

0,012
2
"0
1,'
c c

U
Ol 0,010 ~ _ 0,8 00 0

.~_ 0,008 i;.e> 0 0

o Experimental data
«S'..c 0,006 (ij ~ ',' 0 Experlmenlal data
<3w' CD
- Model ~ - Modet
:::.. 0,004
0,.

<1>N 0,00;' ~ E ',1

a ~ .,ooof--..,.--,---,-,-..,.--,--1 (/)~
O, •• ~..,.--,---,-,.__r-,-___l
0,3 o.~ o.~ 0,6 0,1 D,S D,S 1,0 0,3 0,' O,b 0,6 OJ 0,8 0,9 I,D

MA.XIIIlN FERUENTATICN RATE (IA h) WAXloflAol FEFt.lENTAnrn RAIES (IA h)

Figure 3 : determination of K2 and tn from Vm

Mode~ involving Klc :

=
Ym Klcl(0.0316 + 0.894Klc) r= 0.961
=
K2 K1c/(0.630 + 53.4Klc) r= 0.896
=
lnf 0.636 + 0.103fKlc r= 0.905
Mod,els in~QlviDg Vm :
=
K2 0.004 + 0.0142Vm r= 0.948
tnf 1.93 - 1.39Ym = r =0.910
Table 1 : Empirical relationships used to predict the kinetic fermentation parameters
566
The successive working up of the two methods is of course possible. It leads to reduce the error 011 the further
fennentation kinetic parameters when the fennentation progresses. Figure 4 shows two examples using the models
from on-line KIt and Vm measurements. Those two examples give an account of two different situations with
regard to the quality of experimental data and models fitting (4).

..' A A

..'
0.'

.... Vmck Vms

..'
f\:Vmck
z :;;
~
W
~
... ~

~
0.3

0".
a: .~ It
~
fi ..' li 0.'

!zW ••3 ~ a,ISo Hck

..'
Z
W
::IE 0.' 0.'
a: ~
w
u.
..' ~ 011"
"-

FERMENTATION TIME (h) He

... .00
"'"
FERMENTATION TIME (h)
.."
•.' B B

..• 0.'

.;J!>
Z :;;
..•
•• 7
~ ~
0,3
w
~
a: .~
~
It
.".
z
..' 0.2

~ ....
0
fi
!zw
::IE
." w 0.'
a: .~
~
w w
u. u. Oil"

.. .
0.'

0 2' •• .00 12•

FERMENTATION TIME (h)
..0 .eo
He
... 200
FERMENTAnON TIME (h) He

Figure 4 : comparison between experimental (--) and calculated kinetics (- - - -).

A: Models involving KIt.
B : Models involving Vm.

The application of these methods to non isothennal fermentations (temperature and/or kinetic control) is now being
considered.
REFERENCES
1. Barre P., Chabas J., Corrieu G., Davenel A., EI Haloui N.E., Navarro J.M., Picque D., SablayroUes J.M., Sevila
F et Vannobel C. - Procede de prevision et de contrOie en ligne des fennentations alcooliques et dispositif pour sa
mise en oeuvre. Brevet INRNCEMAGREF, n086 157 79 et n087 402 543 O.
2. EI Haloui N.E., Corrieu G., Cleran Y. and Cherny A. - Method for on-line prediction of kinetics of alcoholic
fennentation in wine making. 1. Fenn. and Bioena., 1989, 68,2,131-135.
3. EI Haloui N.E., Picque D. and Corrieu G .. Alcoholic feonentation in wine making: on-line measurement of
density and carbon dioxyde evolution. J Food Ena., 1988,8,2, 91 108.
4. Cleran Y. - Contribution au developpement d'un procede de suivi et de controle de la fennentation alcooJique,
Thesis, Institut National Polytechnique de Grenoble, 1990.
5. SablayrollesJ.M. etBarre P. - Pilotageautomatique de la tem¢raturede fennentation en conditions oenologiques.
Sci Aliments, 1989,9,239-251.
 MEASUREMENT OF CELL DENSITY IN THE BROTH OF AGGREGATIVE ORGANISM
BY CONTINUOUS-DILUTION-PHOTOMETRIC-ASSA Y

TAKUO YANO. AGUS MASDUKI, YOSHINORI NISHIZAWA. AND HISAO OHTAKE

Department of Fermentation Technology. Faculty of Engineering.
Hiroshima University. 4-1. Kagamiyama 1 chome.
Higashihiroshima 724. Japan

ABSTRACT

A photometric measurement system of cell density was developed and applied

to the culture broth of an aggregative organism. Favolus arcularius. The
system was made up of a photometer with a flow cell. two tubing pumps and a
circulating magnetic pump. The culture broth introduced from a culture flask
was circulated to break up the aggregated cells into small pieces by the
circulating pump for several minutes. The value of cell density measured by
this method gave good agreement with that obtained from the drying method
over the wide range of cell density from 5 to 40 g dry cells/I.

INTRODUCTION

To control and operate a fermentation successfully. the measurement of the

cell density of culture broth is one of the most important operations. In
the broth of an aggregative organism. the cell density was generally
measured by the drying or the photometric method. The drying method takes at
least 1-2 h to get the true value of cell density. The photometric method
with homogenization of broth was troublesome. It is expected to develop a
rapid. simple and precise on-line automatic system for the measurement of
cell density in the culture broth.
In this study. the system developed previously (1) was applied for
measuring the cell density of an aggregative organism. Favolus arcularius.
a kind of mushrooms. and the effects of the operating conditions on the
calculated value of cell density were studied.

MATERIALS AND METHODS

The system was made up of a photometer with a flow cell. two tubing pumps
and a magnetic circulating pump (12 lImin of flow rate). At the first step
of the measurement process. batch dilution was started. i. e.. the drain
valve [No. 4 in Fig. 1] was closed and the pumps 1 and 2 were put in motion.
After a certain time. t b • pump 1 was stopped. If the dilution circuit was
full uP. pump 2 was stopped and pump 3 was put on to mix the culture broth
diluted with water and to break up the aggregated cells into small pieces.
568
The mIxmg was continued until the optical density became approximately
constant. This period was mixing time. t m • When the optical density was
higher than 0.8. pump 2 was put on again and continuous dilution was
started. Surplus diluted broth was washed out from the top of the dilution
circuit. The optical density of diluted broth was measured at 570 nm of
wavelength by the photometer. All the procedures mentioned above were
operated manually in this study.
The cell density of the culture. X i was calculated as follows:
In batch dilution.
(1)
In continuous dilution.
(2)
From Eqs. 1 and 2,
Xi=(VtfVb)X C "exp(D"t2) (3)
Here. V t =F 2 "t,+Vb, D=FdVto Vb=F, "tb. F, and F2 are
flow rates (ml/min) of pump 1 and 2. respectively. t" t2 and tb are
times (min) for batch dilution. continuous dilution and operation of pump 1,
respectively. Vb and V t are volumes (m!) of sample introduced from
fermentor and dilution circuit, respectively. X b and Xc are cell

3 4

Figure 1. Outline of the system developed to measure the cell density in the
culture broth. 1. Sampling pump; 2. Water feeding pump; 3. Circulation pump;
4. Drain valve; 5. Water reservoir; 6. Culture flask; 7. Photometer; 8. Flow
cell; 9. Dilution circuit; 10. Sampling tube.

TABLE 1. Distribution of pellet (number of pellets/ l) during mixing time.

mixing pellet size (mm) (- , not detected)
time
0.25 0.5 1.0 2.0 3.0 4.0 5.0 6.0 7.0
(min) I I I I I I I I I
0.50 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0

0 450 2350 450 200 200 1400 1500 100

2 5xl0 7 1200 400
4 6xl0 7 400
6 4xl0 7 200
9 3xl0 7
13
 569
5

<::' 4
"-.
O'l
'--'" 3
x 2

0
5 10
Culti vati on ti me (d)
Figure 2. CuI ti vation of F. arcularius. Symbols: O. by the drying method;
.... by the system developed in this study under the conditions: F 2 = 50
mIl min. tm = 13 min and Vb = 15 ml.

densities (g/ 1) after batch dilution and contiuous dilution. respectively.

The cell density after dilution was obtained from the predetermined
relationship between optical density and cell concentration.
The cultivation was performed on a rotary shaker (100 rpm) at 28'1: after
transference of 100 ml seed culture into 900 ml of medium in an Erlenmeyer
flask (3 1). To prepare the high cell density suspension. the culture
broth was concentrated by a decantation.

RESUL TS AND DISCUSSION

Distribution of pellet size during mixing time. t. was studied (TABLE. 1).
Eight-days cultured broth containing the pellet of large size was injected
to the dilution circuit, and circulated by the circulating pump. The pellets
were broken up to small pieces immediately and the value of optical density
became stable after 4 min. The breakage of pellets seemed to be caused by
impeller of the circulating pump rotated at high speed.
The effects of operation factors, end point of optical density. OD r•
volume of culture broth introdued from the fermentor. Vb and flow rate of
water. F 2 • on the performance of the system were studied. The value of
OD f in the range of 0.3 to 0.6. 15 ml of V band 50 mllmin of F 2 might
be favored to calculate the cell density.
The developed system was applied to the batch culture of F.
arcularius. Time course of the growth is shown in Fig. 2. The values
measured by this method were in good agreement with those by the drying
method throughout the cultivation.

REFERENCES

1. Yano, T.. Agus Masuduki, and Nishizawa, Y.: Photometric measurement of

high cell density by continuous dilution of broth with a circulating
system. J. Ferment. Bioeng.• 74. 100-103 (1992).
 DYNAMIC MODELLING OF A LARGE SCALE AIR LIFT FERMENTER

G. TRYSTRAM, S. PIGACHE
ENSIA, Food Process Engineering, Massy, France

ABSTRACT

A model of a large scale air lift fermenter used for yeast production from whey is detailed.
This model is a dynamic one, which predicts both biomass and ethanol production, as well as
gas transfer. Parameter identification is performed from industrial experiments. Validation is
characterized with a relative error: 6%.

INTRODUCTION
The Vendome Bel Industries yeast plant produces lactic yeast by continuous aerobic
fermentation of deproteinated wheys in airlift reactors. Considerable research work has been
done for several years to improve the understanding of the process. A previous study reported
elsewhere (1) allowed the determination of the oxygen profile inside the fermenter. In order to
have a dynamic simulation tool able to build control strategies, a physiological model is
developed in addition to the oxygen transfer model.

INDUSTRIAL FERMENTER AND CULTURE

Industrial fermenter
The fermenter used for the experiments is a 120 m3 concentric-tube airlift reactor. Lactose
concentration, substrate and ammonia flow rates are controlled. Inlet and exhaust gases are
analysed. The dilution rate is chosen by adjusting the liquid volume in the fermenter through
substrate flow rate. Temperature of the broth is maintained constant. pH is maintained close to
the set point by adding sulfuric acid. Data logging is supported by a supervision system.

Culture
The population is composed of three complementary species (2): Kluyveromyces Jragilis
(90%), Kfuyveromyces factis (9%) and Torulopsis bovina (1 %). The metabolism of these
strains has been described at length and the authors have shown that the composition of the
flora did not change for low variations of pH, temperature or dilution rate. The culture
behaviour is governed by the predominant strain K. Jragilis which is characterized by a
marked Pasteur effect (3). Even if strict aerobic conditions are maintained, traces of ethanol
are always produced (4) and the biomass production yield Y xis is not at a maximum (5). In
case of oxygen starvation, the phenomenon is accentuated and lactose accumulates in the
 571
medium and/or is converted to ethanol. If aerobic conditions are restored, ethanol will be
consumed by T. bovina (5). K. factis is assumed to grow on lactic acid always present in whey
(5). Therefore, the combination of the three yeast strains allows the production of biomass
whithout loss of yield.

MODELLING

Principle
The principle of the model is to describe the physiological behaviour of the cells in the
fermenter and to take into account the different physiological states of the culture. Each of
these states is reached as a result of various operating conditions or disturbances. This
description coupled with mass balance is able to describe the time evolution of the culture
parameters: biomass, lactose, ethanol. The principle of the model is presented in figure 1.
Metabolism of lactic acid is neglected, because quantities are small (2%). Several approaches
are previously presented but a multi species culture is difficult to model. It appears simpler
more simple to consider the flora as a theoretical one with a specific metabolism which is
chosen as similar to the predominant strain K.fragilis .

Operating conditions
Substrate: Concentration Parameters
and F10wrate Yield ( BIOMASS )
Air flowrate. Liquid Level

Rate of Biomass. C02.

Ethanol production
ate of Lactose. 02.

roduction yield
Productivity
Residual lactose Respiratory Cocfliciem
D Non adequate A6ration

Figure 1 Principle of the dynamic model of the production of yeast from whey

As presented in figure 1, the yeast could reach three states:

state Xl: in case of lactose respiration when oxygen is sufficient in regard to lactose
concentration.
state X2: in case of fermentative evolution of the culture, when oxygen is not sufficient
state X3: which characterize the respiration of Ethanol. This is a transient state.

The total yeast biomass is the sum of Xl, X2 and X3. It is necessary to describe how the
biomass could progress from one state to another. Because of the choice of a theoretical strain,
similar to K. Jragilis , oxygen is the limiting factor. If Oxygen supply is sufficient, the culture
reaches the state Xl and, if ethanol is present in the fermenter, state X3. If Oxygen supply is
not sufficient, an evolution through fermentative state X2 is available. In practice three rates
are determined at each time and comparisons are realized which permit determination of the
evolution of the culture. These are :
-the rate of oxygen supply,
-the rate of oxygen consumption, necessary to oxydate all the lactose in the fermenter,
-the rate of oxygen consumption by yeast.
572
A mass balance is developed and 0 order kinetics are considered for the state changes. Figure
1 presents the relationship between the different parts of the model.

The oxygen transfer must be known. An experimental study allows the development of an
empirical model for overall oxygen transfer versus operating conditions (1):

KLag= Kl +K2.Ln(Qgt)+K3.Ln(VL)

RESULTS
Most of the model parameters are identified from literature. Some adjustments are performed
from experiments using simplex method (6). Figure 2 presents predictive simulations and
comparisons with experiments. It can be seen that both large and small variations are well
followed both for the liquid phase (yeast, ethanol and lactose) and the gas phase (02 and
C02). The uncertainty is for yeast concentration 0.7 gil, 0.2 gil for lactose and 0.1g!l for
ethanol. It gives a relative error of 5%. For a large scale fermenter, this is a good result which
permits the study of control strategies or design of the air lift reactor.

CONCLUSION
Dynamic modelling of the fermenter is validated via industrial experiments and the model is
used as a sensor to predict biomass concentration variations. The interpretation of experiments
is facilitated by the model, and optimization is also performed (6% of increase for the yield).
This work illustrates the ability to model large scale processes.

Biomass (g/l) QR
14,n
...... ~;;!L.1 jM
11
r .
•.. +0. ..~. L.."..;:..:.., •
1., ." . l:~ ~~ ~j:.lIf
.
..'1" • J'

1.0 "~;'.' !.- \ '1~{ rr·~V

0b9~! I I I W I I I
~---:-+-----1'----+--__+--_-_1
I I I I I . 0 100 200 300 400 500 600
100 200 300 400 500 600

Ethanol (g/1 ) C02 (g/l)

1.51
1.4 750T
700~U'~:~ J~:t~rl
lr':
1.2
1
0.8 i~~I-"~'~'\(~':':""::'7 ; "~?
t
0.6
0.4 500 ."~
0.2 450 .. ':'~" •.-'\':\
0 ±----:-+='~__+--_--+--__+.:...::::t__l_I
I I 400 I I I I I I
0 100 200 300 600 0 100 200 300 400 500 600
Time (min) Time (mtn)

Figure 2: Comparisons between modelling and experimental results.

REFERENCES
1. Pigache, S., Trystram, G., Biotechnology and Bioengineering, 1992,39,323-331.
2. Moulin, G., Malige, B., Galzy, P., Le lait, 1981,66,323-332.
3. Alexander, M. A., Jeffries, T.W., Enzime Microb. Techno!., 1990, 12,2-19.
5. Moulin, G., Malige, B., Galzy, P., J. of dairy science, 1983,21-28.
6. Pigache, S., Doctoral thesis, ENSIA, Massy, France, 1993.
 BIOCATAL YTIC PRODUCTION OF CELLOBIOSE
CONTAINING OLIGOSACCHARIDE MIXTURE

MARIANNE ROSSI, YU-YEN LINKO, PEKKA LINKO, TIMO VAARA *,

MARJA TURUNEN*
Laboratory of Biotechnology and Food Engineering,
Department of Chemical Engineering
ljelsinki University of Technology, SF-02150 Espoo, Finland
Biotechnology Unit, Alko Ltd, SF-05200, Rajamaki, Finland

ABSTRACT

The production of cellobiose containing unique and novel oligosaccharide mixture from
starch by cyclomaltodextrin-glucanotransferase (COTase), using cellobiose as the acceptor
sugar, is described. This is believed to be the fIrst time such oligosaccharide production
method has been published. Cellobiose containing oligosaccharides are linear molecules
with one (OS03) or several glucose units (OS04-0S07) adjoined to the acceptor sugar
cellobiose. The effects of dry matter and enzyme concentration, the mass ratio of starch
to acceptor sugar, and reaction time were studied to produce in controlled manner the
oligosaccharide mixtures with a good yield. The best results were obtained in 48 h with
starch of DE value 1, at 30 to 40 % total dry matter, starch/acceptor ratio of 1:1, 60 DC,
and a COTase concentration of 30 to 350 U per g of starch. Up to a 48 % yield of
OS03-5 was obtained.

INTRODUCTION

Oligosaccharides are important new raw materials applied in food, animal feed, pharma-
ceutical and chemical industries because of their low calorie content, mild sweetness, low
cariogenecity, positive physiological effects on health, and good technical properties. A
number of oligosaccharides are already commercially available, e.g. starch-based products
such as maltose, maltotriose, maltotetraose, isomaltose, panose, saccharose-based products,
such as "coupling sugar", fructooligosaccharides, palatinose, lactose-based oligosaccharides
and sugar alcohols such as maltitol, and search for new products is still going on. In this
paper, the production of cellobiose containing oligosaccharide mixture from starch by
CGTase, using cellobiose as the acceptor sugar [1], is described.
574
METHODS

Production of Cellobiose Containing Oligosaccharides

Pretreatment of the starch (barley starch, Alko Ltd) was carried out by adding 30 U/g of
CGTase (activity 7600 U/ml, Alko Ltd) to a slurry containing various quantities of starch
in 50 mM imidazole buffer of pH 6.8, containing 1.5 mM CaClz• The CGTase was
allowed to act at 85°C for 30 minutes with agitation to obtain starch with DE=1.
The cellobiose containing oligosaccharides were produced by adding various amounts
of cellobiose (Sigma) to the pretreated starch. The reaction was started by adding 50 U
CGTase per gram of starch with agitation and carried out for 48 hours at 60 °C.

Characterization of the Oligosaccharide Mixture

The composition of the mixture was determined by Shimadzu HPLC using a ~ondapak:
carbohydrate analysis column (Millipore) at room temperature and 65% acetonitrile as
eluent with a fiowrate of 0.9 ml/min.

RESULTS AND DISCUSSION

The Effect of Substrate Concentration

Table 1 shows the effect of the total substrate concentration on the composiotion of the
oligo saccharides OSG3-0SG7 after 48 hours reaction time. The total quantity of the
oligosaccharides OSG3-0SG5 in the mixture obtained at 10% dry matter content was 4.8
gram per 100 g of the reaction mixture (48.0% of the initial dry matter content). At 30%
dry matter content the respective concentration was 14.6 gll00 g (48.7% of the initial dry
matter).

TABLE 1
The effect of substrate concentration (starch + cellobiose) on composition of the oligosac-
charide mixture produced in 48 hours at ~C. The starch/acceptor mass ratio was 1: 1

Substrate dry matter Product mixture

content in solution Glucose Cellobiose OSG3 OSG4 OSG5 OSG6 OSG7
(%) (giiOO g of the reaction mixture)

10 0.03 2.1 2.0 1.6 1.2 0.8 0.2

30 0.04 6.9 6.0 5.0 3.6 2.9 0.9

Time Course of Oligosaccharide Production

The yield of OSG3-0SG5 oligosaccharides and the cellobiose consumption (30% total
substrate dry matter) during 48 hours incubation is shown in figure 1. Twelve hours was
sufficient to obtain a maximum of about 50% yield.
 575
16
,-... OSG3-0SG5

-8z
QI)

12
bo
'-'

0
E=: 8
Cellobiose
~
Eo-<
OSG3~
~
u 4 OSG
~
u OSG5

10 20 30 40 50
REACTION TIME (h)
Figure 1. Time course of oligosaccharide production with 30% dry matter content

The Effect of Starch/Acceptor Mass Ratio

Starch/acceptor mass ratios of 1:4, 1:2 and 1: 1 in oligosaccharide production with total
substrate dry matter of 30% resulted in total quantities of oligosaccharides OSG3-0SG5
of 13.1 gram per 100 g of the reaction mixture, 14.4 g/100 g and 14.6 g/100 g, respecti-
vely. However, the starch/acceptor mass ratio of 1:2 was superior to the ratio of 1:1
during the fIrst 24 hours of incubation.

Down-Stream Processing

It is possible by free or immobilized glucoamylase treatment in a controlled manner to

increase the yield of OSG3-0SG5 oligosaccharides [2]. Furthermore, removal of free
glucose from the oligosaccharide mixture with free or immobilized yeast reduces the
cariogenicity and calorie content of the mixture. Ethanol (70%) may be used to precipitate
the long chain oligosaccharides and polysaccharides out of the mixture. The product may
be further treated with conventional methods by using active carbon, ion exchanger, spray-
drying or vacuum concentration.

REFERENCES

1. Rossi, M., Linko, Y.-Y., Linko, P., Vaara, T. and Turunen, M., Oligosaccharide
mixture, and procedure for its manufacturing. WO 92/03565, 5.3.1992.

2. Rossi, M., Linko, Y.-Y., Linko, P., Vaara, T. and Turunen, M., Oligosaccharide
mixture, and procedure for its after-treatment. WO 92/07947, 14.5.1992.
 ENZYMA TIC PRODUCTION AND MEMBRANE CONCENTRATION OF
OLIGOSACCHARIDES FROM MILK WHEY.

MIGUEL H. LOPEZ LEIVA and MONICA GUZMAN.

Department of Food Engineering, University of Lund,
P.O. Box 124,221 00 Lund, Sweden

ABSTRACT

We have studied the enzymatic formation of o1igosaccharides when whey UF-permeates

are hydrolysed in an immobilised enzyme reactor.
The immobilised enzyme reactor used was of the "flow-through" type where the enzyme is
immobilised in the pores of a micro porous fIlm and the substrate is forced to pass through
the film. High mass transfer and short residence times can be obtained in such a reactor.
The concentrations of glucose + galactose, lactose and oligosaccharides in the hydrolysates
were determined by HPLC.
It was found that oligosaccharide formation is mainly governed by the concentration of
lactose and the residence time, and in a minor degree by the pH.
An attempt to fractionate the oligosaccharides from a whey permeate hydrolysate by
nanofiltration was also made.

INTRODUCTION

Oligo saccharides are polymeric carbohydrates consisting of two to ten monomer residues
joined through glycosidic bonds. During the enzymatic hydrolysis of the lactose present in
whey and whey UF-permeate, galactosyl - oligo saccharides containing from two to seven
units are formed. The ingestion of these galactosyl-oligosaccharides encourages the
proliferation of Lactobacillus bifidus in the intestine.
These bacteria form the dominant flora of breast infants and are said to improve a number of
body functions [4], among others growth inhibition of harmful entero pathogenic micro-
organisms such as Escherichia coli, Salmonella typhi and Staphylococcus aureus.
Yoghurts with added strains of Bifidobacteria are commercialised world-wide. In 1989
there were already around 250 oligosaccharide-containing products on the Japanese market,
including yoghurts and different kinds of drinks [2].
The aim of this work is to study the influence of the operating conditions in the immobilised
enzyme system, during the enzymatic hydrolysis of whey UF-permeate, and to fmd the
optimal parameters which favour oligosaccharide production.
Since the oligo saccharides are always obtained at rather low concentration, an attempt to
separate or concentrate them by means of nanofiltration was also made.
 577
MA TERIALS AND METHODS

Immobilised Enzyme Reactor.

A porous ftlm with immobilised lactase was purchased from Amerace Corporation (B utler
U.S.A.). The immobilised enzyme is ~-galactosidase from A. oryzae bound via
glutaraldehyde covalent linkages to silica particles entrapped in the microporous of a PVC
sheet. The ftlm is 0.64 mm thick, with a porosity in the range of 70 to 80 %, and pore
diameters between 5 and 24 ~m (8 ~m average).
The reactor is formed (fig. 1) by cut- r-------------------,
ting this ftlm into pieces of 2.5 or 4.7
cm diameter which are then mounted
in Swinex disc fIlter holders Substrate Hydrolysate
(Millipore Corporation) with fIltering out
areas of 3.4 and 13.8 cm 2 respec- in
tively. The substrate is forced to
pass through the disc ftlter holder,
where a good contact substrate -
immobilised enzyme is obtained with
minimal diffusive resistance.
Film with
immobilized enzyme
Figure I.-Immobilized Enzyme Reactor

Whey.
The whey was obtained by rehydration of desalted whey powder. This was bought from the
Swedish Dairies Association (SMR, Falkenberg), having the following composition:
Lactose: 80%, Protein: 13%, Fat: 2%, Salts: 1% and Moisture: 4%.

Ultrafiltration
A batch UF circuit consisting of a Romicon PM 10 hollow fibre cartridge (cut off =10 000)
of 5 sqft area was used for the ultrafiltration of the whey.
Whey powder re-hydrated at 20 % lactose was ultrafiltred at 55°C and the permeate
collected and stored at 4 °C for further utilisation. To avoid microbial growth, 0.1 % sodium
azide was added to the solutions. New solutions were prepared every third week.

Whey hydrolysis.
The set-up depicted in figure 2 was used to study the lactose hydrolysis. Residence time,
substrate concentration, and pH were the parameters studied. The temperature of operation
was always 40°C as suggested by the producers of the immobilised enzyme.
The whey UF-permeate was loaded into a 100 ml syringe (S) and then pumped, by means of
a gear pump, through the immobilised enzyme reactor (IER), which is placed in a bath at
constant temperature.
The residence times were varied either by changing the feed flow or by alternatively using
two immobilised enzyme
S IER films. Oligosaccharide
formation is especially
important at the begin-
ning of the reaction when
the lactose is only
partially hydrolysed
[1,3]. Based on this we
decided to work in the
range of 2.5 to 70 sec-
Water bath onds. Conversions were
lower than 60 %.
Figure 2. Experimental Set-up
578
RESULTS AND DISCUSSION

Effect of the Residence Time and Concentration.

In figure 3 we see the result obtained for the hydrolysis of a UF-penneate containing 20 %
lactose. This figure typifies the results obtained also for other concentrations (6 and 15 %).
The concentration of oligosaccharides reaches a maximum at around 12-seconds and then
decreases, probably when they also begin to be hydrolysed.
The concentration of the
lactose has the largest effect
on oligosaccharide
production; this is especially
!
important for concentrations

~1--1:~==
in the range of 6 to 15 %.
The optimum residence time ~
for maximum production of II:
oligosaccharides depends on ~
the initial lactose ....
concentration, however, it is
always around 10 - 15 sec.
Since it is not possible to
transfonn all the lactose in
oligosaccharides, the use of
purification and/or
concentration steps after the
enzymatic step become
necessary.
o 10 20 30 40 50 60 70
TIME (sec)

Figure 3.- Oligosaccharide fonnation for a

whey UF-penneate containing 20 % lactose

Fractionation ofUF-whey hydrolysate by membrane filtration.

An attempt to concentrate the oligosaccharides present in the whey penneate hydrolysates
was done by fIltrating the solution through a nanofiltration membrane (Nadir NF-PES-5 IPP
100, kindly supplied by Hoechst AB). A stainless steel module for reverse osmosis ("dead
end" type) with 300 ml capacity was used. It was pressurised to up to approximately 20
bars with the help of a nitrogen cylinder.
Even though total separation of the oligo saccharides by nanofiltration was not possible they
were concentrated to a certain extent. In one case the concentration was more than tripled.
The present state of the art in nanofiltration does not allow 100 % rejection for the
oligosaccharides. Therefore the recycling of the penneates from the nanofIltration step to the
hydrolytic step and the extent to which this can be done should be further studied.

REFERENCES

1.- Burvall A, Asp N. G. and . A. 1977. Oligosaccharide fonnation during hydrolysis of

lactose with s. lactis lactase I, II and III. Lund University (Ph. D. thesis).
2.- Com line news service,1989. Comline Biological & Medical. Sagetsu Building, 3-12-7
Kita-Ahoyama, Minato-Ku, Tokyo 107, Japan.
3.- L6pez-Leiva M. H. and zarate S. 1988. Mass transfer in an immobilized enzyme reactor.
Med. Fac. Landbouww. Riksuniv. Gent, 53 (4a): 1639 - 1644
4.- Wijsman R. M. Hereijgers J. L. P. M. and Groote J. M. F. H. 1989. Selective
enumeration of bifidobacteria in fennented dairy products. Neth. Milk Dairy J. 43: 395-405.
 CONSTRUCTION OF INTEGRATED BIOREACTOR SYSTEM FOR
BIOLOGICALLY ACTIVE PEPTIDES FROM ISOLATED SOYBEAN
PROTEIN

KENJI SONOMOTOl,2), YASUNORI OKAMOT02), SHINKICHI KOHN03),

KANEO KAN03)
l)Department of Food Science and Technology, Faculty of Agriculture, Kyushu University,
Hakozaki 6-10-1, Higashi-ku, Fukuoka 812, Japan
2)Department of Biochemical Engineering and Science, Faculty of Computer Science and
Systems Engineering, Kyushu Institute of Technology, Iizuka, Fukuoka 820, Japan
3)Life Science Research Center, Nippon Steel Co., 1618 Ida, Nakahara-ku, Kawasaki 211,
Japan

ABSTRACT
Among the commercially available endopeptidases from different sources tested, mixture of
three kinds of proteases gave an efficient production of di-and/or tri-peptides mixture (average
molecular weight, ca. 400) from a hardly water-soluble isolated soybean protein at 50°C. The
peptides mixture so obtained showed biologically active functions in mammalia, such as
control of obesity and fatigue. Repeated batch system with ultrafiltration membrane (MWCO,
10,000) resulted in the peptides production with high potency during the operation for 20
days. Furthermore, the desired product was continuously obtained for one week by using a
cross-flow membrane reactor over 90 % of the peptides yield.

INTRODUCTION
Functionalities of foods have been greatly attracted in food biotechnology. Especially,
peptides activated by digestion and metabolism are known to have physiological functions
such as control of obesity, adult diseases, immune system, cell differentiation and
proliferation. Moreover, few studies were reported to construct new bioreactors having
continuous production and separation abilities for small peptides. In this paper, we have
attempted biologically active di-and/or tri-peptides production by enzymolysis of isolated
soybean protein ll] and also construction of an integrated bioreactor with down-stream system
for the continuous production by the use of ultrafiltration membrane.

MATERIALS AND METHODS

Isolated soybean protein (protein, 96 %) was obtained from Nishin Oil Co., Japan. Unless
otherwise noted, enzymolysis batch reaction was performed at 50°C in 100 ml of the reaction
mixture containing 5.0 % w/w autoclaved substrate and 0.1 % w/v filtered enzyme mixture.
580
The resulting reaction mixture was centrifuged and used for evaluation of the enzymolysis,
such as average molecular weight of the product, di-and/or tri-peptides yield and nitrogen
yield. The average molecular weight was calculated as (protein concentration in the
supematant)/(molarity of amino group in the supernatant). Di-and/or tri-peptides yield means
its content in the solubilized protein from the substrate and was determined by Sephadex G-10
gel chromatography. Minitan-S ultrafiltration system (Millipore Co., USA) was used as a
cross-flow membrane module in a continuous production bioreactor.

RESULTS AND DISCUSSION

Selection of commercially available endopeptidases at 50°C was done by considering the high
viscosity of substrate suspension and its water-insolubility. As a result, three kinds of
proteases, Biotamilase P-1000 (Nagase Biochemical Industry, Japan), Protease S (Amano
Pharmaceutical Co., Ltd., Japan) and Protease N (Amano) were of greatly high potency.
Furthermore, their mixture (23.1:14.7:70.9 w/w in the above-mentioned order) gave an
efficient production of di-and/or tri-peptide mixture (average molecular weight, ca. 400),
which showed biologically active functions in mammalia, such as control of obesity and
fatigue. Amino acid composition of the peptides was almost same as of the substrate,
containing essential amino acids well-balanced and abundantly. Amino acid mixture of the
same composition as of the peptides as well as the substrate exhibited no biological activity.
Optimum pH and temperature of the enzyme reaction were 7 and 60°C, respectively. Figure 1
shows the typical enzymolysis of isolated soybean protein with the enzyme mixture. The
reaction yield on nitrogen was about 90 % and the yield of the desired peptides was calculated
over 85 % after 24 h of reaction. However, the residual proteolytic activity in the reactor
gradually decreased and was about 40 % of the initial value after 24 h. The enzymolysis
reaction was slightly inhibited by the peptides formed and succcessive addition of the substrate
in the reactor stabilized the enzyme mixture. From this result, it is necessary for the
construction of effectively continuous production system that the product of low molecular
weight should be selectively removed from the reactor and that the enzymes and the substrate
of high molecular weight are kept in a certain space.
Repeated batch production of di-and/or tri-peptides was done with a stirred concentrator
with ultrafiltration membrane (MWCO, 10,000). Each reaction was carried out at 50°C for 22
h in 10 ml of reaction mixture containing 0.1 % enzyme mixture and 2.5 % substrate, followed

/1
..
.....
t1 1000 100 ~
.....
a> II ~
80 '>
-
~ 800
~I
~
ro 'fl~

-e
o:I~

;:j
() 600 60 l~
I/~
a> c'>,
0 £ \:::
a>
400 40 o..g
e~
b!) «l ....
ro ::s \:::
~ 200 20 ~
~
~ 0:::
0 /1 0
0 1 2 3 24
Time (h)

Figure 1. Time courses of enzymolysis of isolated soybean protein. (e) Average molecular
weight, (0) residual proteolytic activity, (&) nitrogen yield.
 581
by filtration of the reaction mixture with nitrogen gas at 0 °C. Enzymes were proved to be kept
in the concentrator by the membrane. Although the residual activity of the enzymes diminished
slowly, this system showed a relatively good operational stability during 21 days. The
average molecular weight of the resulting products after 10 and 18 days of reaction were 435
and 455, respectively. The peptides yield was more than 80 % even after 21 days.
We constructed a continuous production system with a cross-flow ultrafiltration
membrane module to separate the product and to hold the enzymes and the substrate. The
influence of several crucial factors such as enzyme concentration and dilution rate on the
productivity and the stability of the system was investigated. As a result, the desired product
was continuously obtained for one week with a high operational stability (Figure 2). The
average molecular weight of the product was maintained about 400 and the peptide yield was
over 85 % in the course of operation. The recovery of the peptide from the substrate was 67
% w/w. There was no bacterial contamination in the system at the end of reaction. Increase in
the lifetime of the membrane reactor system is now being studied.

500

---. • .. •
'"
(1:$

"'3

• •
(,)
~
0_
~:§ 400
&~
~ 300
100 100 ~
~e
..
:Q
80
• ... 80 .>'
Q)

';;~
~
... ... '"Q)

60 =&
o~
.~
(0"0 60
'"'Q) 8-
~.>'
(,) I::
§~ 40 ...
·c
I

40 .Q.'"
(,) 8 "0
1::-
.Q3 -E I::
..... 20 20 is
(1:$
I

8
~

0 0
0 1 2 3 4 5 6 7
Time (day)

Figure 2. Continuous production of di-and/or tri-peptides with cross-flow membrane reactor.

Substrate 5.0 %, enzyme concentration 0.1 %, dilution rate 0.5 day-I, working volume 110
mI, ultrafiltration membrane molecular weight cut-off 10,000. (e) Average molecular weight
offiltrate, (0) protein concentration in filtrate, (.) nitrogen yield, (A) di-and/or tri-peptides
yield in filtrate.

REFERENCE

1. Sonomoto, K., Okamoto, Y., Kohno, S. and Kano, K., Enzymatic production of
biologically active peptides from isolated soybean protein. 9th International Biotechnology
Symposium, Abstract 460, Crystal City, Virginia, USA, August 1992.
TRYPSIN-CATALYZED SYNTHESIS OF OLIGOPEPTIDES CONTAINING
HYDROPIDLIC AND ESSENTIAL AMINO ACID, L-LYSINE.

YUKITAKA KIMURA, MOTOHIRO SHIMA, TOMOYUKI NOTSU, SHUJI ADACHI,

and RYUICHI MATSUNO
Department of Food Science and Technology, Faculty of Agriculture, Kyoto University,
Sakyo-ku, Kyoto 606-01, Japan

ABSTRACT

The synthesis of peptides from N-(benzyloxycarbonyl)-L-Iysine n-hexyl ester (Z-

K-OHex) and L-Iysine amide (K-NHJ was done in 65% (v/v) acetonitrile. We mainly got
two dipeptides (Z-KK and Z-KK-NH2) and a tripeptide (Z-KKK-NH~. We had Z-KK-
NH2 in the 90% yield but it rapidly decreased as the reaction proceeded. The decrease of
the concentration of trypsin prolonged the period for which the yield was maintained at
90%. The yields of Z-KK and Z-KKK-NH2 increased when the concentration of K-N~
increased. The above results were well simulated by a model of the reaction.

INTRODUCTION

Peptides have some properties superior to proteins and amino acids. Peptides are ab-
sorbed in the hydrolyzed form into the blood stream from intestine faster than amino acids!).
Solution of peptides has lower osmotic pressure than that of amino acids which consist of
the same composition. So, peptides are good as a nutrient after a surgical operation.
Peptides are prepared by chemical, enzymatical, or genetical methods. A good
method of them is enzymatic one because of its mild condition and low consumption of
energy. Many researchers have studied enzymatic synthesis of peptides catalyzed by pro-
teinases 2). However, it is difficult to synthesize peptides consisting of hydrophilic amino
acids without protective groups of their side chains. In this study, we tried to synthesize the
peptides containing a hydrophilic and basic amino acid, L-Iysine, which is one of essential
amino acids, using trypsin as a catalyst.
 583
MATERIALS AND METHODS

Materials
Trypsin from porcine pancreas was purchased from Wako Pure Chemicals (Japan),
and used without further purification. N-(benzyloxycarbonyl)-L-Iysine n-hexyl ester
(Z-K-OHex) was prepared by esterification of a-N-(benzyloxycarbonyl)- E -t-(butyloxy-
carbonyl)-L-Iysine (Kokusan Chemicals, Japan) with n-hexanol and by deblocking with
formic acid. L-Lysine amide hydrochloride (K-NH2·2HCI) was purchased from Kokusan
Chemicals. Other Chemicals were of analytical grade.

Method of synthesis of lysine peptide

The synthesis of peptides from Z-K-OHex and L-lysine amide (K-NHz} was done
in 65% (v/v) acetonitrile. The high contents of acetonitrile gave the high yields of peptides
but K-NHz could not dissolve in more than 65%. The reaction mixture included 50 mM Z-
K-OHex, 100 mM K-NHz·2HCl, 200 mM triethylamine (neutralizing HCI), and 1 mg/ml
trypsin when the reaction started. Its pH was 9.4 and it was kept at 30°C. At appropriate
times, a portion of the reaction mixture was sampled and analyzed by high performance
liquid chromatography with a UV detector (258 nm). The yield shown in this study is based
on the initial concentration of Z-K-OHex.

RESULTS AND DISCUSSION

Figure 1 shows the time course of synthesis. At the initial stage, a dipeptide, Z-KK-
NH2, was obtained with 90% yield but rapidly decreased. Then another dipeptide, Z-KK,
and tripeptide, Z-KKK-NH 2 increased. The hydrolysate, Z-K, also increased constantly.
Considering the order of appearances of the products and the substrate specificity of trypsin,
we assumed a model for synthetic pathways as shown in Fig. 2. The model has ten first-
order reactions. Their reverse reactions can be ignored because of their slow reaction rates.
Factors in the concentration of enzyme was considered into the reaction rate constant, k.
The values of ks were determined to get an agreement between experimental and calculated
values. The calculated yield of products are shown by the solid lines in Fig. 1.

Figure 1. Time course of products in

the synthesis catalyzed by free trypsin.
The keys show the experimental results.
0, Z-KK-NH 2 ; \l, Z-KK; 6., Z-
KKK-NH2; 0, Z-K; 0, Z-K-OHex.
The solid lines show the calculated ones
50 100 150 180

Time [min]
based on the model as shown in Fig.2.
 584

The effects of the concentration of trypsin on the yields of the products were estimat-
ed by the model and the experiments. When the concentration decreased from 1 to 0.01
mg/ml, the period for which the yield of Z-KK-NH2 was maintained at 90% prolonged
from 1 min to more than 60 min. This phenomenon would result from that hydrolysis of the
amide is slower than that of the ester.
Figure 3a shows the time courses of the products when the concentration of K- NH2
increased to 200 mM. The yield of Z-KK-NH2 increased to 95% and its decreasing rate
became slower. The yield of Z-KKK-NH2, one of the objective peptides, also increased to
23%. The undesired product, Z-K,was produced slowly. The calculated values in Fig. 3b
show the same pattern of the experimental values.

k, (K-NH.] k. (K-NH,]

Z-K-OHex
-k' ~1~:J-NH2 -4 k .. ~~~~,:~I:.N[:~O) Figure 2. A model of pathways of the

t k" (K-NH,] t synthesis of lysine peptide.

Z-, N-(benzyloxycarbonyl)-; K, L-
Z-K -.II Z-KK
k, [H,O) lysine; -NH2, amide; -OHex, n-hexyl
ester; k, reaction rate constant.
k, [H,O)

.) b) Figure 3. Time courses of synthesis

when tge concentration of Z-K and K-
NH2 were 50 and 200 mM, respectively.
The left (a) shows the experimental
-- .:-"".-:-:.--- results and the right (b) shows the calcu-
~" .~~--::-.- .. - lated ones. 0, -, Z-KK-NH2; V,
':;-:>"
--Z-KK'
, '"
6 --- Z-KKK-NH' 2'
0,
50
Time [min]
-'-, Z-K; 0, Z-K-OHex.

REFERENCES

1. Ganapathy, V., BurCkhardt, G., Leibach, F.H., Characteristics of Glycylsarcosine

Transport in Rabbit Intestinal Brush-Border Membrane Vesicles. !,. BioI. Chern.,
1984,259,8954-8959

2. Morihara, K., Using proteases in peptide synthesis. TIBTECH, 1987, 5, 164-170

 INTEGRATION OF REACTION AND RECOVERY BY A CONTINUOUS
EMULSION ENZVME REACTOR WITH IN-LINE REMOVAL OF THE OIL AND
WATER PHASE BY MEMBRANE SEPARATION.

C.G.P.H. SCHROEN', A. VAN DER PADT, K. VAN 'T RIET.

Wageningen Agricultural University,
Food and Bioprocess Engineering Group, Department of Food Science,
P.O. Box 8129, 6700 EV Wageningen, The Netherlands.

ABSTRACT

The emulsion/membrane bioreactor consists of an emulsion contained in a stirred

tank and two membrane units. In the emulsion an enzyme catalyzed reaction is carried
out. The oil phase and the water phase are separated by a hydrophobic and a hydrophilic
membrane respectively.
The emulsion/membrane bioreactor can be used for both hydrolysis and
esterification of acylglycerols. A cellulose membrane can be used as the hydrophilic
membrane. The hydrophobic polypropylene membrane has to be modified with block
copolymers in order to prevent protein adsorption and, therewith, to maintain selective
separation.

INTRODUCTION

The enzyme, lipase, can catalyze both hydrolysis and esterification reactions. In
case of esters of glycerol and fatty acids (acylglycerols) the following reactions are
important:

Triglyceride + Water ~ Diglyceride + Fatty Acid

Diglyceride + Water ~ Monoglyceride + Fatty Acid
Monoglyceride + Water ~ Glycerol + Fatty Acid

, To whom correspondence should be addressed

586
Depending on the reaction conditions either mainly esters or mainly fatty acids
are produced. Triglycerides are interesting for the edible oil industry. Monoglycerides can
be used as e.g. emulsifiers. Fatty acids are used, in large quantities, in the soap industry.
The enzyme, lipase, can only be active at an oil/water interface where both
substrates and the enzyme are available. It is important to create a large oil/water
interface to obtain a high volumetric reactor activity. This can be done by using an
emulsion in a stirred tank. The emulsion has to be separated in an oil phase and a water
phase in order to obtain the products. In the emulsion/membrane bioreactor (figure 1)
this separation is carried out by coupling the stirred tank to two membranes. The
membranes are a hydrophilic cellulose membrane (cut-off value 6,000; manufactured by
Enka) to separate the water phase and a hydrophobic polypropylene membrane (pore
size O.llLm, manufactured by Enka) to separate the oil phase. The hydrophilic membrane
is not permeable for the enzyme due to its pore size. The enzyme does not permeate
through the hydrophobic membrane also. Only the oil phase permeates through the
hydrophobic membrane and the enzyme is not soluble in the oil. The retentate of both
membranes contains the enzyme and is pumped back to the emulsion making a
continuous process possible. This reactor concept can be used for both hydrolysis and
esterification reactions.
(E) Fatty acids Glycerol (E)
(H) Triglycerides Water (H)

Hydrophobic Hydrophilic
Membrane Membrane

Lipase containing emulsion

(H, E) Fatty acids Glycerol (H, E)

(H, E) Monoglycerides Water (H, E)
(H, E) Diglycerides
(H, E) Triglycerides

Figure 1. The emulsion/membrane bioreactor; H denotes substrates and products for the
hydrolysis reactor, E denotes substrates and products for the esterification reactor.
 587

RESULTS AND DISCUSSION

Both membranes were tested separately. The hydrophilic membrane was fouled
by lipase but the flux decrease was limited. After an initial decrease in flux of
approximately 10%, the flux value remained constant for seven consecutive days.
Therefore, the hydrophilic cellulose membrane was judged to be appropriate for
application in the emulsion/membrane bioreactor. The hydrophobic membrane, however,
was fouled severely by enzyme adsorption, resulting in permeation of the water phase at
low transmembrane pressure (:::::0.03x1OS Pa). Enzyme adsorption has to be prevented.
This was done by modifying the hydrophobic membrane with an appropriate block
copolymer that gives a steric hindrance to enzyme molecules that come near the
membrane. The block copolymer consists of a hydrophobic, polypropylene oxide, middle
block and two hydrophilic, polyethylene oxide, blocks on both sides of the middle block.
The hydrophobic middle block adsorbs onto the hydrophobic surface, both hydrophilic
groups are extended into the water phase.
The block copolymer-coated membrane was not permeable for the water phase
unless the Laplace pressure was exceeded. The maximum transmembrane pressure at
which selective separation was possible (4Pmax) was 0.Sx1OS Pa. After 10 days of
continuous operation 4Pmax had not decreased. Therefore, it seems reasonable to assume
that no enzyme adsorption took place at the modified membrane otherwise water could
have permeated through the membrane at a lower transmembrane pressure. The
modified membrane was used in continuous emulsion/membrane bioreactors for both
hydrolysis and synthesis of acylglycerols.

CONCLUSIONS

Both hydrolysis and esterification of acylglycerols is carried out in an

emulsion/membrane bioreactor. The flux through the hydrophilic membrane decreases
slightly because of lipase adsorption but the flux level is high enough for continuous
operation. The hydrophobic membrane is heavily fouled by lipase adsorption and has to
be modified with block copolymers in order to prevent lipase adsorption. Through the
modified membrane only the oil phase permeates, selective separation is possible up to
a transmembrane pressure equal to the Laplace pressure.

Acknowledgement

The authors would like to thank the "Programme Committee on Membrane Research"
(PCM) of the Netherlands and the "Dutch Foundation for Technical Sciences" (STW)
for their financial support.
INTERESTERIFICATION OF TRIGLYCERIDES IN ORGANIC SOLVENT USING
MODIFIED LIPASE

KENICHI MOGI**, SOBHI BASHEER*, AVA YAMAOKA*, FRED B PADLEY***,

KAZUHIKO FUJIWARA**, MITSUTOSHI NAKAJIMA*
* National Food Research Institute, MAFF, Kannondai, Tsukuba, Ibaraki 305, Japan
** Nippon Lever B.V., 2-22-3, Shibuya, Tokyo 150, Japan
*** Unilever Research Colworth Laboratory, Colworth House, Bedford, MK44 lLQ, U.K.

ABSTRACT
Interesterification of tripalmitin and stearic acid in hexane was investigated using modified
lipase. The modified lipase is a complex of lipase and surfactant, which has interesterifica-
tion reactivity in organic solvent. Screening tests using various kinds of lipases and surfact-
ants were done to optimize the activity of the modified lipase and its recovery. The modified
lipase from Rhizopus japonicus using sorbitan monostearate as surfactant had the highest
activity in hexane system. At low water concentration, interesterification occurred predomi-
nantly, and the production of diglycerides was very low. Hydrolysis of triglycerides was
increased with the increase in water concentration. In a non-solvent system, interesterifica-
tion occurred similarly as in hexane system under low water concentration.

INTRODUCTION
The physical properties of oil are related to the fatty acid composition of the triglycerides.
By the interesterification reaction, especially using 1,3 specific lipase, the fatty acid compo-
sition of triglycerides can be changed. Several enzymatic interesterification methods are
known; suspension of water in oil, W/O emulsion with surfactant, dispersion of powdered
enzyme, immobilized enzyme, reversed micelles and PEG-modification systems have been
investigated. In these systems, interesterification occurs under low water concentration.
However, it has been reported that 10 to 20% of triglycerides were converted to diglycerides
and monoglycerides by these conventional methods due to addition of water to the systems.
Recently modified lipase, which is a complex of lipase and surfactant, was proposed by
Okahata (1). Figure 1 shows the schematic model of modified lipase. By this modification,
the lipase becomes soluble and/or well dispersed in the organic solvent. Then modified
lipases can have much higher interesterification reactivity than the originallipases.
In this article, the modification process and evaluation of modified lipase in the interes-
terification of tripalmitin and stearic acid in n-hexane and non-solvent systems were inves-
tigated.
 589

Surfactant Lipase Modified Lipase

Figure 1. Schematic model of modified lipase.

MATERIALS AND METHODS

Modified lipases: The modification procedure is shown in Figure 2. Various lipases
from different sources, such as Rhizopus niveus, Rhizopus sp., Aspergillus niger, Mucor
javaniclls, Mucor miehei, Rhizopus delemar, Rhizopus japonicus, Aspergillus oryzae,
Aspergillus sp., Candida rugosa, Candida cylindracea, and Alcaligenes sp. were used.
Sugar esters and sorbitan esters were used as surfactant.

Water 1 L n-Hexane 50 mL
- Lipase 3.0 g - Tripalmitin 500 mg
- Surfactant 1.0 g - Stearic Acid 500 mg
(dispersion in 20 mL ethanol) - Modified Lipase 20-50 mg
Stirring 600 rpm at 5°C for 2 h. Stirring at 500 rpm
I Heating at 40°C in Water Bath

I
Centrifugation at 8000 rpm
at 5°C for 10 min.
Reaction for 3 h.
I
samrling at 30 min. interval
Freezing at -20°C
I Stop Reaction with Chloroform
Freeze Drying I
I GC Analysis
Modified Lipase

Figure 2. Lipase modification procedure. Figure 3. Interesterification experiment.

Substrates: Tripalmitin, stearic acid and n-hexane supplied from Wako Pure Chemical,
Japan, were used for interesterification experiments.
Interesterification reaction: Interesterification experiment is shown in Figure 3. 500 mg
of tripalmitin, 500 mg of stearic acid and 20-50 mg of modified lipase were added in 50
mL n-hexane, in which the water content was decreased by molecular sieves (4A 1/16,
Wako Pure Chemical, Japan). The reaction was carried out for 3 h at 40°C with 500rpm
stirring.
Analysis: Interesterification reactivity was evaluated from the change of triglycerides
composition, which was determined by GC (GC: R-14A with FlD detector, Shimazu, Japan,
Column: HR-TGC1 Capillary column, Shinwa Chemical, Japan). Water concentration in
590

reaction system was determined by Karl Fischer Water Analyzer (684 KF coulometer,
Metrohm, Switzerland). Protein contents of crude and modified lipases were analyzed by
Automatic Nitrogen Analyzer (FP-428, Leco, USA).

RESULTS AND DISCUSSION

Figure 3 shows a typical time course of the interesterification reaction. The concentrations
of tripalmitin and stearic acid decreased, and palmitic acid, PPS and SPS were produced
with time. The reaction system nearly reached equilibrium after 2 h reaction. SSS was not
produced, which means the modified lipase has 1,3 specific activity.
Monoglycerides were not detected at all. The amount of diglycerides produced was less
than 10% of that of triglycerides.
In this system the water concentration was below 20 mglL. The modified lipase had
intercsterification reactivity at such low water concentrations. The original crude lipase did
not have interestcrification reactivity under the same conditions.
From interesterification reaction and protein yield analysis, lipase from Rhizopus japoll-
icus and sorbitan monostearate, HLB 4.7, as surfactant were selected for modified lipase
formation.
Using the modified lipase, interesterification reaction in non-solvent system was also
carried out. Figure 4 shows the comparison of interesterification reaction using hexane and
non-solvent in the reaction systems. The modified lipase also had high reactivity in non-
solvent system.

- 40 .9 100
o::::!.
c 0. in Hexane
ee 30 0.
~
-
0.
E Cf)
0.
c ....
0
(f)
c 20 '+-

-...ccu
:;::: c: U
c:
0

10 ...
'iii
Q)
CIl
(f)
cu > 0.
u c: 0..
C
0 0
U 0 ()

0 30 60 90 120 150 180

Time (h)
Time (min)
Enz/(TG+FFA) =3%
-a- Pa acid -k- PPP -0- SPS o : TG 1wt%, FFA 1wtOlo in Hexane at 40 "C
-e- SI acid -+- PPS -0- SSS ... : TG 50wt%, FFA 50wtOio at 70 "C

Figure 3. Interesterification Figure 4. Comparison of hexane

reaction. and non-solvent systems.

REFERENCE

1. Okahata, Y. and Ijiro, K., A lipid-coated lipase as a new catalyst for triglyceride synthesis
in organic solvents. J. Chern. Soc., Chern. Commun., 1988, 1392-94.
 THE EFFECf OF LIPIDS OXIDATION ON THE ACTIVITY OF INTER-
ESTERIFICATION OF TRIGLYCERIDE BY IMMOBILIZED LIPASE

Toru Nezu,Satoru Kobori and Wataru Mastumoto

Food and Fat Laboratory, Asahi Denka Kogyo KK,
Higashi-ogu 8-4-1, Arakawaku, Tokyo 116, Japan

ABSTRACT

The lipase catalyzed interesterification of high oleic sunflower oil and stearic acid
was carried out on a packed bed reactor, and the effects of lipid oxidation on the
deactivation rate was examined. Oxidation of oils increase the deactivation rate; however
the purification of oils stabilized lipase activity during continuous interesterification,
deceased peroxide value and anisidine value of oils. There was a significant correlation
between the deactivation rate and lipid oxidation parameters mentioned above. Further-
more, the incubation of lipase with 2-unsaturated aldehydes dissolved in hexane caused
inactivation of lipase. It was concluded that the secondary lipid oxidation product strongly
affect the lifetime of lipase activity.

INTRODUCTION

There has been great interest in lipase-catalyzed interesterification as a new method of

industrial fat modification. Using this process selective exchange of acyl groups at 1,3-
position of triglyceride is possible and there is also the potential for the economic synthesis
of cocoa butter equivalents, which cannot be obtained by conventional chemical
interesterification.
To minimize the reaction cost in this process, it is important to extend the lifetime of the
lipase as much as possible. Recently, the poisoning effect of the lipids
hydroperoxide 1),2),3),4) and phospholipids5) on the lipase activity were reported.
In this study, continuous interesterification of high oleic sunflower oil and stearic acid was
carried out, and the effects of lipid oxidation and purification were examined. The signifi-
cant correlation between deactivation rate and lipids oxidation parameter was obtained.
592
MATERIALS AND METHODS

The lipase from Alcarigenes (LIPASE PL) absorbed on diatomaceous earth (RADIOLITE
GK-02, Showa Kagaku Co.,Japan) was supplied from Meito Sangyo Co, Japan. High
oleic sunflower oil (HOS) was oxidized by stirring at 140°C for 60 min in a vacuum of 300
mmHg. This oxidized HOS was purified by neutralization, bleaching and silica gel
chromatography. The reaction mixtures were prepared by mixing HOS, stearic acid and
hexane in a ratio of 1:1:4, and 800ppm of ethanol was added 6 ), This mixture was then
pumped through a column containing 4.0g of immobilized lipase at 45°C. The glyceride
composition of the reaction products was analyzed by reverse phase HPLC, and 2-oleoyl-
distearin (SOS) content was measured.
The interesterification activity was estimated by following equation,

LN (l/I-X) = K/ SV, X = SOS / SOSeq (1)

Where SV and SOSeq are space velocity(hr-1) and SOS content at equilibrium, respectively.
The first-order kinetic model was applied to the decrease in the activity during continuous
interesterification as follows;

LN ( K / KO ) = - KdT (2)

Where KO and T are the initial activity and reaction time, respectively.

RESULTS

The deactivation rates were increased by oxidation of oil. When oxidized and non-
oxidized oils were used as a substrate, the first order constant of deactivation (Kd) were
0.0158 and 0.0285 respectively. The oxidized HOS (OX) was purified by neutralizing-
bleaching (NB), bleaching (B) and silica-gel chromatography (sq, and it was then used as
substrate. The purification of the substrate increased the stability of lipase in the order
B<NB<SC, and reduced phosphorus content, ferrous content, peroxide value and anisidine
value of oils (TABLE-I).
The deactivation rates were dependent on peroxide value and anisidine value, and the
following correlation (P<0.05) was found to be significant:

Kd = 0.00156 (POV) + 0.00076 (AnV) - 0.00412 (3)

TABLE-I. THE EFFECT OF OXIDATION AND PURIFICATION

ON LIPASE DEACTIVATION.

Oil POV AnV Fe P Kd

(meqiKg) (-) (ppm) (ppm) (l/day)
15.5 1.5 1.8 54.4 0.0158
OX 36.9 9.1 1.5 24.5 0.0569
B 0.4 16.2 1.3 11.3 0.0146
NB ND 19.4 0.3 ND 0.0120
SC ND 1.5 0.2 19.0 0.0059
POV : peroxide value AnV : anisidine value
 593
The immobilized lipase was incubated in 100 mM of various aldehydes dissolved in
hexane at 45°C for 3 days, and interesterification activity was measured (TABLE-2). In
saturated aldehyde, only short chain aldehyde caused the deactivation of lipase, but even so
the effects on lipase activity were relatively weak. In monounsaturated aldehyde, regardless
of chain length, 2-unsaturated aldehydes were predominately responsible for the deactiva-
tion of lipase. The formation of 2-unsaturated aldehyde by lipid oxidation was
reported. 7)

TABLE-2. THE EFFECT OF ALDEHYDE ON LIPASE ACTIVITY

Relative activity
Saturated Unsaturated
Double-bond t-2 c-4 t-12
Propionyl 62.6
Valer 54.9
Heptyl 82.5 3.0 60.0
Nonyl 95.0 3.1
Undecyl 124.8 6.8
Tridecyl 107.2 216.2
Control- 100.0 100.0 100.0 100.0
Control: The lipase was incubated with hexane alone.

DISCUSSION

In this study, it was suggested that the content of the lipids hydroperoxide and aldehyde
strongly affected to the lifetime of lipase. While hydroperoxides are easily reduced by
conventional purification, it is not economical to remove the secondary lipid oxidation
product on an industrial scale. Therefore, it is important to use fresh nonoxidized oil as a
substrate. Berger et a1 8) reported that immobilization of Candida cylindracea lipase by
covalent linkage involving the E-amino residue of lysine lead to stabilization against the
deactivation caused by acetaldehyde. The effects of aldehyde observed in this study may
also contribute to the intereaction of E-amino residue of lipase with aldehyde.

REFERENCE

l.Posorske,L.H., G.K.LeFebvre, C.A.Miller, T.T.Hansen and B.L.Grenvig,

1. Am. Oil Chern. Soc.,1988,65,922
2.Wisdom,R.H., P.Dunnil and M.D.Lilly, Biotechnoi. Bioengneer.,1987, 29,1081
3.0hta,Y., T.Yamane and S.Simizu, Agric. BioI. Chem.,1989, 53,1855
4.Wang,Y. and M.H.Gordon, 1. Agric. Food Chem.,1991,39,1693
5.Wang,Y. and M.H.Gordon, 1. Am. Oil Chern. Soc.,1991,68,588
6.Mastumoto, W.,E.Nakai and T.Nezu, USP-5089404,1989
7.Frankel,E.N., Flavor Chemistry of Oils and Fats, ,Ed. D.B.Min and T.H.Smouse,
American Oil Chemists' Society.,1985, pp.1
8.Berger,B. and K.Faberl, 1. Chern. Soc., Chern. Commun., 1991,1991,1198
TRIACYLGLYCEROL SYNTHESIS FROM FATIY ACID AND GLYCEROL USING IMMOBILIZED
LIPASE.

Yoshitsugu Kosugi*, Noboru Tomizuka* and Naoki Azuma**

*National Institute of Bioscience and Human-Technology, Agency of
Industrial Science & Technology, 1-1 Higashi Tsukuba Ibaraki 305, Japan.
**Boso Oil & Fat Co. Ltd., 2-17-1 Hinode-chou Funabashi Chiba 273,
Japan.

ABSTRACT
High free fatty acid (FFA) rice bran oil containing 30-50 % FFA was
converted to an oil containing about 75-90 % triacylglycerol (TG) using
immobilized lipase. Enzymatic refining of the FFA oil was performed
continuously for more than one month using a reactor with two circulation
loops, each connecting a fixed bed reactor and a dehydrator. The
substrate was the stoichiometric amount of the FFA oil and glycerol
required for synthesizing TG. The technology could be applied to
synthesizing TG at more than 95 % yield from polyunsaturated fatty acids
such as docosahexaenoic acid.

INTRODUCTION
Rice bran oil has attracted much attention since it was found to lower
serum cholesterol(I). However, the rice bran oil available in most
Asian countries contains a high proportion of free fatty acid (FFA)
making the oil inedible. Successful enzymatic refining of high FFA rice
bran oil can lead to the use of a potentially large and convenient
domestic edible resource for south Asia (2). Icosapentaenoic acid
(EPA) and docosahexaenoic acid (DHA) are characteristic biochemically
active lipids. There is a demand for the conversion of these
polyunsaturated fatty acids (PUFAs) into natural TG form.
This paper describes continuous enzymatic refining of high FFA rice
bran oil and a high yield enzymatic preparation of TG from PUFA.

MATERIALS
 595

High FFA rice bran oil used was prepared either as described in a previ-
ous paper (3) or mixing oleic acid with refined rice bran oil (1:1). The
latter of a high FFA oil model was easily available in Japan and used
for the continuous refining. Immobilized lipases from Rhizomucor miehei
and Candida antarctica were donated by Novo Nordisk Bioindustry. Their
commercial names are Lipozyme 1M 60 and sp 382, respectively.

RESULTS
TG synthesis conditions using t..obilized lipase
The optimum' amount of substrate for synthesizing TG was the
stoichiometric amount of fatty acid (FA) and glycerol (G) necessary for
synthesizing TG. The reaction was carried out at 60°C with continuous
mixing and dehydrating of reactants. The mixing and the dehydration were
performed by bubbling dry nitrogen. The dehydration rate was faster than
the deacidification rate.
Continuous enZ,YIatic refining of high FFA rice bran oil
During a batch reaction of enzymatic refining, production of monoacyl-
glycerol (MG) was very low. Production of diacylglycerol (DG) increased
rapidly during the first 3-4 hours. After that time production of DG
)4,
r+ M -,...-*+, r-----~ - --,
I ® I
I
I I
I ... I I
I I I
+ tI I
I
I I
7 I
I +
I
N2 ~1 I
15 I I
I I

.
I I
I I

I. I I
I
I L__ +_ --0-
18
__ __ _____ +
~11:r-_-,. r - - - t - - '
____ @!.8____ J
I

L - - - - - - - - - -+- - 16

Figure 1. Schematic Diagram of Continuous Refining Reactor.

1) glycerol vessel; 2) high FFA rice bran oil vessel; 3) and 4) fixed
bed reactor at 60°C (Amounts of the Lipozyme 1M 60 packed into 3) and' 4)
were 1.25 g and 5.0 g respectively.); 5) and 6), dehydrator (Working
volumes for 5) and 6) were about 25 ml and 100 ml respectively.); 7)
molecular sieve column; 8) liquid pump for glycerol{0.228 g/ hr); 9)
liquid pump for high FFA oil (4.2g/hr); 10) and 12) liquid pump for
circulation in the loop (6 ml/min); 11, liquid pump for pumping reactants
from the first loop to the second loop (4.22 g/hr); 13) air pump; 14)
needle valve; 15) nitrogen cylinder; 16) nitrogen damper; 17) outlet for
product (4.22 g/ hr); 18) three way cock. The solid line represents a
path of the reactants and the dotted line, a path of the nitrogen gas.
596
decreased. TG production increased continuously during the reaction for
about 2 days. At the end of the reaction, the TG content reached 85-90 %
and water content was less than 10 ppm. The consecutive reaction for the
synthesis of TG from fatty acid (FA) and glycerol (G) are as follows:
2FA + G = DG + 2H20 ------1
DG + FA = TG + H20 ------2
The rate of reaction-1 was over 10 times faster than that of reaction-2.
Continuous enzymatic refining reactor is shown in Figure 1. In the
first loop which combines the fixed bed reactor 3), pump 10) and dehydra-
tor 5), about 30 % DG was continuously produced from FA and G. In the
second loop, 74.2 % TG was continuously produced for more than one month.
Preparation of TG fro. G and PUFA or PUFA Ethyl Ester
The amount of DHA or EPA in the substrate was 5-10 % higher than the
stoichiometric amount required to compensate for the PUFA loss during the
reaction. In the case of DHA, the immobilized lipase from Candida
antarctica was used because the immobilized lipase from Rhizomucor miehei
did not synthesize TG from the DHA. Other conditions were the same as
those used in the enzymatic refining of the high FFA oil. A TG yield of
more than 95 % was achieved with EPA, DHA or arachidonic acid. Pure TG
was obtained by passing the product through basic aluminum oxide column.

DISCUSSION
Pure TG obtained from PUFA is hopeful for such medical use as the
intravenous infusion for preventing thrombotic disorder (4) as well as
healthy food. The productivity for the continuous refining process is
calculated as follows.
Productivity = (0.00422 x 24) x 0.742 / ( 0.005 + 0.00125)
= 12.0 (Kg of TG)/ day x (kg of Immobilized lipase)
The cost of the enzyme needed to produce 1 kg of TG can be calculated
from the productivity, the life span of the immobilized lipase and the
enzyme price. Enzymatic refining of high FFA rice bran oil can be
industrialized if the enzyme cost is reduced less than 5-10 % the price
of the rice bran oil.

1. Haumann, B. F., Rice bran linked to lower cholesterol. J. Am. Oil

Chem. Soc., 1989, 66, 615-618.
2. Gupta,~p., Rice Bran Offers India an Oil Source, ~ Am. Oil Chem.
Soc., 1989, 66, 620-623.
3. Kosugi, Y., Igusa, H. and Tomizuka, N., Glyceride production of high
free fatty acid rice bran oil using immobilized lipase. ~ Jpn. Oil
Chem. Soc. 1987, 36, 769-776.
4. Hamaza~ T., Fisher,S., Schweer, H., Meese,C.O., Urakase, M.,
Yokoyama, A. and Yano, S., The infusion of trieicosapentaenoylglycer-
01 into humans and the in vivo formation of prostaglandin 13 and
thromboxane A3. Biochem. Biophys. Res. Comm., 1988, 151, 1386-1394.
 NAD(p)H REGENERATION IN YEAST CELLS USING ETHANOL
AS ENERGY SOURCE

TADASHI KOMETANI, HIDEFUMI YOSHII, EITARO KITATSUJI,

RYUICHI MATSUNO'
Department of Chemical and Biochemical Engineering, Toyama National College of
Technology, Hongo 13, Toyama 939, Japan, "Department of Food Science and Technology,
Faculty of Agriculture, Kyoto University, Sakyo-ku, Kyoto 606-01, Japan

ABSTRACT

Reduced nicotinamide cofactors, NAD(P)H, are regenerated from NAD(Pt through the
pathway for ethanol oxidation in baker's yeast, Saccharomyces cerevisiae, that is coupled with
bioreductions of carbonyl compounds. In this paper, the relationship between the ability
using ethanol as energy source for the NAD(P)H regeneration and the culture condition on
oxygen absorption rate was examined. The cells grown under the anaerobic condition could
hardly use ethanol as energy source and those abilities increased with increasing the rate of
oxygen absorption under the culture condition.

INTRODUCTION

Yeast-mediated bioreduction of prochiral ketones has been recognized as a useful technique

for synthetic organic chemists, because many optically active compounds are conveniently
prepared (1,2). In this microbial transformation, the regeneration of NAD(P)H from
NAD(Pt in cells is necessary. The original procedure for the NAD(P)H regeneration in
yeast cells was based on the hexose monophosphate pathway for glucose oxidation (3).
Recently, we reported a new procedure using ethanol in an aerobic condition (4-6). In this
procedure, NAD(P)H were regenerated from NAD(Pt through alcohol oxidation to carbon
dioxide. The new procedure is more efficient and clean than the original one as discussed
in the preceding papers (5,7).
This paper deals with the relationship between the ability using ethanol as energy source
for NAD(P)H regeneration in the yeast cells, Saccharomyces cerevisiae HUT 7099, and the
culture condition on oxygen absorption rate. The reduction reaction of ethyl acetoacetate
(EA) to chiral product, (S)-(+)-ethyl 3-hydroxybutanoate (E 3-HB), an important starting
598
material for many biologically active compounds, was examined to evaluate the ability of the
cells grown under several conditions.
o H ,pH
1I "OOEt----.. NOOEt
CH3~ CH3
Ethyl acetoacetate (S)-(+)-Ethyl
(EA) 3-hydroxybutanoate
(E 3-HB)

MATERIALS AND METHODS

Culture medium A medium was composed of 1 %(w/v) glucose, 0.5% pepton, OJ % yeast
extract, and 0.3 % malt extract. Its pH was adjusted to 5.5 by addition of hydrochloric acid.

Cultivation condition on oxygen absorption rate A 2-i! Sakaguchi culture flask containing
200 ml, 300 ml, 500 ml, or 1 i! medium was shaken at reciprocal shaking of 80 or 100
strokes/min. An aqueous solution of 0.20 M sodium sulfite in the presence of catalytic
cuprous ion was used for a determination of oxygen absorption rate. Changes of
concentration of sodium sulfite in those conditions were measured in the usual manner, and
oxygen absorption rate was calculated.

Cultivation Yeast cells were precultured in a 50-ml Erlenemeyer flask with 10 ml medium
by vigorous reciprocal shaking at 30°C for 24 h. Into a 2-i! Sakaguchi flask containing
a suitable amount of medium, the above culture, 2 %(v/v), was inoculated, and the culture
was incubated at 30°C with reciprocal shaking at a suitable speed for 48 h. The cells were
harvested by suction filtration, washed with 0.9 % saline, and then used as described below.
Weight of dry cells was determined after drying in an oven at 110°C for 8 h

Bioreduction of ethyl acetoacetate A suspension of 1.00 g of wet yeast cells in a 20 ml of

77 mM EA and an energy source, which were dissolved in 0.1 M potassium phosphate buffer
(PH 7.0), was placed in a 100-ml Erlenmeyer flask and shaken on a rotary shaker set at 30
°C and 80 rpm. The concentrations of energy sources were 200, 300, and 200 mM for
ethanol, glucose, and acetate, respectively. As a control experiment, the reaction without
addition of any energy source was also done. After 8 h, the reaction mixture was centrifuged
and the supernatant was allowed to analysis.

Analysis The concentrations of EA, E 3-HB, and ethanol were measured by a gas
chromatography as described in the preceding paper (5).

RESULTS AND DISCUSSION

Reduction rates by the yeast cells grown under various conditions on oxygen absorption rate
were first obtained from concentrations of E 3-HB in reaction mixtures after 8 h per weights
of dry cells. Then, reduction rates by additional energy sources were evaluated by deducting
the rates of control experiments without addition of any energy source. Figure 1 shows the
 599
Fig.1. ~
0.4

CD
u
...>.
•
.... ..,.
'0 ~
0, 0.3
~.
.....
s::
'0
E
~---;;- •
•67.;a
~-
g 0.2
0-

-
fIJ
0
CD ~
111 O
a: 0
c 0.1
o
:;::;
U
::J
'0
CD
a: 0.0 8
~_~_.....L._~ _ _L...-_~_.....L..._~_~
o 20 40 eo 80
Oxygen Absorption Rate under Culture Condition [mmol/h • R]
Reduction rates of ethyl acetoacetate with various energy sources.
Symbols: O. ethanol; •• glucose; A. acetate

effect of oxygen absorption rate under the culture condition on reduction rates of ethyl
acetoacetate with various energy sources. The most important reslut was that the cells grown
under almost anaerovic condition had hardly the ability using ethanol as energy source and
the ability increased with increasing the rate of oxygen absorption under the culture condition.
Then, the reduction rates in the presence of ethanol became nearly constant at the oxygen
absorption rate above 20 mmol!h·.e. In the case of glucose as energy source, NAD(P)H
would be regenerated through both glucose oxidation pathway and ethanol oxidation pathway.
So, the reduction using glucose proceeds much faster than one using ethanol. In the case
of acetate as energy source, the reduction rates with various yeast cells were almost constant.
This result suggests that the ability of NAD(P)H regeneration using acetate is not influenced
by the culture condition on oxygen absorption rate. Thus, we concluded that oxidation
pathway from ethanol into acetate would be influenced by the culture condition on oxygen
absorption rate. The further investigations are now in progress.

REFFERENCES

1. Servi, S., Baker's yeast as a reagent in organic synthesis. Synthesis, 1990, 1-25.
2. Csuk, R. and Glanzer, B., Baker's yeast mediated transformations in organic chemistry.
Chem. Rev., 1991, 91, 49-97.
3. Yamada, H. and Shimizu, S., Microbial and enzymatic processes for the production of
biologically and chemically useful compounds. Angew. Chem. Int. Ed. Engl., 1988, 27,
622-642.
4. Kometani, T., Kitatsuji, E. and Matsuno, R, Baker's yeast mediated bioreduction. A new
procedure using ethanol as an energy source. Chem. Lett., 1989, 1465-1466.
5. Kometani, T., Kitatsuji, E. and Matsuno, R, Baker's yeast mediated bioreduction: Practical
procedure using EtOH as energy source. J. Ferment. Bioeng., 1991, 71, 197-199.
6. Kometani, T., Kitatsuji, E. and Matsuno, R, Bioreduction of ketones mediated by baker's
yeast with acetate as ultimate reducing agent. Agric. BioI. Chem., 1991, 55, 867-868.
7. Kometani, T., Yoshii, H., Kitatsuji, E, Nishimura, H. and Matsuno, R, Large-scale
preparation of (S)-ethyl 3-hydroxybutanoate with a high enantiomeric excess through
baker's yeast-mediated bioreduction. J. Ferment. Bioeng., in press.
LOOSE RO MEMBRANE REACTOR FOR L·AMINO ACID PRODUCTION WITH
COENZYME RECYCLING

Takuya Harada and Osalo Miyawaki

Dept. of Agricultural Chemistry, The Univ. of Tokyo
1-1-1 Yayoi, Bunkyo-ku, Tokyo 113, JAPAN

ABSTRACT

A microscale membrane bioreactor (vol. 2ml) was built up by using loose RO (LRO)
membranes and applied to the production of L-amino acids (L-alanine form pyruvate and L-
serine from hydroxypyruvate) with NADH recycling in a conjugated system of glucose
dehydrogenase (GDH) and alanine dehydrogenase (ALDH). By using UTC-20 membrane, the
reactor with immobilized GDH (695U/ml) and ALDH (206U/ml), operated at the retention
time of 8Omin, produced 0.15M L-alanine at the productivity of 240g/1/day with NAD cycling
number of 150,000 under continuous feed of the substrate solution containing 0.2M pyruvate,
ammonia, glucose and IJlM NAD. This LRO membrane reactor is expected to have a wide
applicability to coenzyme regeneration system for its general rejection characteristics for
dissociable coenzymes.

INTRODUCTION

Immobilization of the dissociable coenzymes, such as NAD, NADP, ATP, are one of the key
technologies for the development of the bioreactor system with multiple enzymes. To this
end, there have been many attempts in which ultrafiltration (UF) membrane reactor was used
combined with the macromolecule-bound coenzyme techniques or affinity chromatographic
reactor [1] . With ultrafiltration membrane, however, the direct immobilization of coenzymes
is difficult because molecular weight of the coenzymes are much smaller than the cut-off
molecular weight of UF membranes. In this study, we employ LRO membranes for direct
immobilization of coenzymes. Three different membranes (negatively charged membrane,
NTR-7410, and two different type LRO membranes, UTC-20 and SSA) are used and
compared with one another for their molecular rejection characteristics. A microscale
membrane bioreactor is built up and applied to the production of L-amino acids with NADH
recycling.

MATERIALS AND METHODS

GDH (Bacillus sp. 69.5U/mg) was gifted by Amano Pharmaceutical. ALDH (Bacillus subtilis
 601
32U/mg) was purchased from Sigma Chemical. NAD, NADP, ATP were purchased from
Bohringer Mannheim. NTR-741O was gifted by Nitto Electric Industrial. UTC-20 and SSA
were gifted by Toray Industries. A microscale membrane bioreactor, 15cm2 in membrane
area and 2cm 3 in volume was built and used for the L-amino acid production. HPLC pump
was used to feed substrates.

RESUL TS AND DISCUSSIONS

TABLE 1 shows the rejection ratio of the membranes for the coenzymes measured at
different concentration of Tris buffer. The negative charged membrane, NTR-741O rejects
native coenzymes through electrostatic repulsion when no buffer coexists. However, the
rejection ratio decreases drastically with increase in the ionic strength by increase in buffer
concentration. SSA shows a higher rejection ratio than NTR-741O, but the rejection ratio
decreases to 50% with increase in buffer concentration. On the other hand, the rejection ratio
of UTC-20 not dependent on ionic strength was almost 90% to all the coenzymes tested.

TABLE 1
Rejection ratio of membranes for coenzymes at various ionic strength.
conc. of Tris buffer
O.OM O.IM 0.5M
NTR-741O NAD 0.850 0.440 0.178
NADP 0.885 0.242 0.131
ATP 0.874 0.338 0.194
SSA NAD 0.925 0.712 0.499
UTC-20 NAD 0.878 0.923 0.963
NADP 0.906 0.904
ATP 0.824 0.890

L-amino acid production with a microscale membrane bioreactor was carried out in the
scheme shown in Fig.l. The system was composed of ALDH as a main reaction enzyme, and
GDH as a regeneration enzyme for NADH. ALDH produces L-alanine from pyruvate and
ammonia with the reducing power of NADH. NAD is oxidized back to NADH by GDH
consuming glucose. In this reactor, the enzymes were completely immobilized in the system.
The coenzyme was fed continuously and partially rejected or immobilized through the
NAD, Substrate

gluQose gIUC~ dehYd~naS\ glucono-o-Iactone

NAD NA H+H+
L-alanine 0( "'==-. .2 pyruvate + NH4 +
L-serine "< hydroxypyruvate + NH4+
alanine dehydrogenase
Enzyme NAD Substrate Product

.Memb;a~........~.•....... ~ .•........ ~ ... ... ... .. NAD, Product

Fig. 1 Membrane bioreactor with coenzyme recycling

602
rejection by the membrane used. Thus, the coenzyme was concentrated in the system
depending on the rejection ratio of the membrane.

TABLE 2 shows the results of continuous L-alanine production at the retention time of
160min with the microscale reactor provided with different membranes. The NAD
concentration with NTR-741O was IOJ.lM, tenfold higher than those with UTC-20 and SSA,
due to its low rejection ratio for the coenzyme at the high buffer concentration. The system
with UTC-20 showed the best result both in NAD cycling number and productivity so that it
is used for further tests.

TABLE 2
Continuous L-alanine production in membrane bioreactors with different membranes.
NTR-741O SSA UTC-20
substrates [M] 0.1 0.2 0.2
NAD [J.lM] 10 1 1
't1/2 [hr] 192 > 800
conversion [%] 60 45 50
productivity [gil/day] 48 72 80
turn over [s-l] 0.86 1.29 1.43
cycling number [1] 6000 90000 100000

The effect of the retention time for UTC-20 system was tested. NAD cycling number was the
highest to be 150,000 at the retention time of 80min, that means that the cost of NAD is
reduced to 1/150,000. The reactor productivity showed the highest value of 260 g/l/day at the
retention time of 53 min. This productivity is high enough compared with the productivity in
fermentation reactors.
Hydroxypyruvate is substrate analog of pyruvate. Therefore, the same system is applicable to
L-serine production when pyruvate is replaced with hydroxypyruvate. With ALDH at
206U/ml and GDH at 695U/ml, the conversion ratio of 20%, NAD cycling number of 6,700,
and the productivity of 57 gil/day was achieved under the continuous feed of 0.1 M substrates
and 3J.lM NAD.

CONCLUSION

A microscale membrane bioreactor with LRO membrane was constructed and applied to L-
alanine production and L-serine production with NAD regeneration. A high productivity was
observed with a high cycling number of NAD. Because of the general rejection characteristics
of the membrane used for dissociable coenzymes, the present system is expected to be widely
applicable to the bioreactor system with coenzyme recycling.

REFERENCE

1. Nakamura, K., Aizawa, M. and Miyawaki, 0., Electoenzymolo~y/Coenzyme Re~eneration,

Biotechnology Monographs, vol.4, Springer-Verlag, Heiderberg, 1988
 STUDIES ON THE BIOLOGICAL REDUCTION OF THE NITRATE CONTENT
OF CARROT JUICE AND OF A MODEL SOLUTION USING
PARA COCCUS DEN/TRIF/CANS DSM 65

MIN, S. H. and SPIEB, W.E.L.

Institute of Process Engineering, Federal Research Centre for Nutrition,
Engsesserstr. 20, 76131 Karlsruhe, Germany

ABSTRACT

The influence of pH, temperature and various electron-donators on the nitrate reductase
activity of Paracoccus denitrificans DSM 65 was investigated. Nitrate reduction rate was
studied in a model solution and in carrot juice. The denitrifying efficiency of P. denitrificans
DSM 65 depended on pH, temperature and mainly on the H donator. Enzyme activity was
highest at 34°C, pH 7.0 and with fructose as H donator. Fructose has been found to have the
greatest influence on nitrate; with fructose as H donator, the specific nitrate degradation rate
was twice to three times as high as with glucose or sucrose. At the higher temperatures and
pH values the nitrate degradation rate increased as well. This applied to both model solution
and carrot juice.

INTRODUCTION

Because of its nutritional quality and its high vitamin content, vegetable food is greatly
appreciated. Some vegetables contain high nitrate concentrations, however. Although nitrate
itself is of little primary toxicity for humans [1], uptake of nitrate-containing food may be
harmful as the nitrate may be reduced to reactive nitrite by the human oral and intestinal flora
[2]. In babies, the nitrite may lead to cyanosis [3]. N-nitrosamine formation represents another
problem. N-nitrosamines form in acid media from nitrous acid and secondary arnines [4].
Therefore elimination of nitrates and nitrite in food is desirable for reasons of health. Because
of previous studies into its metabolism and its applicability as biocatalyst, Paracoccus
denitrificans DSM 65 was selected as the most suitable microorganism for nitrate degradation
in vegetable food [5].
The present work was concerned with factors influencing nitrate reductase activity of
Paracoccus denitrificans DSM 65. We also tried to improve the growth conditions of the
bacterium in order to obtain more biomass of increased nitrate and nitrite reductase activity.
604
MATERIALS AND METHODS

Nutrient medium (modified according to Burnell, 1975)

The medium consists of the following 3 solutions (Table 1) and trace-element solution SL-6.
The three solutions were autoclaved separately and mixed together shortly before use. The
trace element solution was filtrated sterilely and added at a concentration of 2ml1l.

Table 1
Contents of 3 solutions

Solution A Solution B Solution C

KH2P04 xHP 2 g EDTA 0.002 g glucose 2.0 g
K 2HP04 x2HP 5 g MgS04x7HP 0.2 g fructose 8.0 g
Aqua dest 700 mI FeS04x 7~O 0.02 g sucrose 2.0 g
N~S04 0.054 g Aqua dest 150 ml
CaCI2x2~O 0.025 g
NH4 Cl 1.6 g
KN0 3 6.0 g
Aqua dest 150 ml

Conditions of growth and cell harvest

The microorganism was grown in this medium at 30°C under compressed air (6 lImin) for the
first 6 hours. After the supply of compressed air was stopped, the culture was further
incubated until nitrate in the medium was completely degraded. To increase the biomass and
nitrate reductase activity, 6 gil KN03 , 8 gil fructose, and 2 mill trace element solution were
once more added. Growth in the liquid medium was monitored by determination of the optical
density at 578 nm. After complete degradation of the nitrate added, cells were separated from
the medium by centrifugation at 6,000 rpm for 20 min at 4°C. The cells were washed using 0.1
M potassium phosphate buffer of pH 7.0, again centrifuged and frozen at -18°C.

Determination of specific nitrate reduction rate

All experiments to determine the nitrate reduction rate were conducted in a model solution
and in carrot juice at different pHs. The influence of various H donators on the nitrate
reduction rate was also studied in a model solution by enzymatic determination.

Nitrate degradation test:

Model solution: 0.1 g potassium nitrate, resp., and 1 g glucose, 1 g sucrose, 1 g fructose, or 2
g each of glucose and fructose plus 2 g sucrose as H donators were added to 100 mI 0.1 M
potassium phosphate. The pH of the model solution was adjusted to 5.5, 6.0, 6.5 and 7.0.

Carrot juice: Commercial carrot juice was centrifuged to remove solid components (15,000
rpm, 15 min). pH of aliquots was adjusted to 5.5, 6.0, 6.5 and 7.0, resp. 0.1 g potassium
nitrate were added to 100 mI carrot juice.
 605
RESULTS AND DISCUSSION

Fonnation and actIvity of nitrate reductase are influenced by various factors such as
composition of the medium, incubation conditions, kind of H donator, pH and temperature.
The optimal pH and temperature for nitrate reduction using P. denitrificans DSM 65 was
reported to be pH 7.5-8.0 and 30-40°C [5]. As denitrification in this range of pH and
temperature is feasible only to limited extent for hygienic and sensory reasons, growth
conditions and contents of the medium were optimized, so that enzymatic activity was
improved at pH 6.0,6.5 and 7.0 and at temperatures of25 and 30°C.
At the temperatures of 25, 30 and 34°C used for the present experiments P. denitrificans
DSM 65 has been found to show relatively good growth and high enzyme activity. The pH of
freshly pressed carrot juice is between 6.0 and 6.5; the juice contains 2.33 g invert sugar and
2.42 g sucrose per 100 ml [7]. Therefore glucose, fructose, sucrose, or a combination of these
were used as H donators.

The results of these studies have shown that there are virtually no differences in the nitrate
reduction rate between the results obtained from carrot juice and model solution containing 1
g glucose, 1 g fructose and 2 g sucrose. This sugar combination corresponds to that of carrot
juice. When glucose or sucrose alone wes used as H donator in the model solution, a
significantly lower nitrate reduction rate was observed. Fructose had the great influence on
nitrate reduction in the nitrate-degradation test, although the activity determinations had not
shown any major difference among the kinds of sugar. In the presence of fructose, P.
denitrificans degrades nitrate faster than with glucose or sucrose. This will be subject of
further investigations. Because of increased enzyme activity and greater biomass yield under
the improved growth conditions, the applicabilities of P. denitrificans DSM 65 to
denitrification of vegetable foods have clearly improved.

REFERENCES

[1] Corre, W.I; Breimer, T., 1979; Nitrate and nitrite in vegetables. Wageningen:
Centre for Agricultural Publishing and Documentation
[2] Swann, P.F., 1975: The toxicology of nitrate, nitrite and N-nitroso compounds. J.
Sci. Fd. Agric. 26, 1761-1770
[3] Rimpel, R.; Bruchmann, H; Neinass, R., 1981: Zur Problematik der Nitrat- und
Nitritkontamination bei enzymatisch produziertem Mohrensaft.
Lebensmittelindustrie 28 Nr. 6, 263-268
[4] PreuBmann, R., 1982; Nitrosaminbedingte Cancerogenese. VCH, Weinheim, 143-148
[5] Mayer-Miebach, E. und Schubert, H, 1991; Biokatalytische Verringerung der Nitrat-
konzentration in pflanzlichen Lebensmittelrohstoffen. Bio Engineering 2, 34-43
[6] Burnell, IN., 1975; The reversibilty of active sulphate transport in membrane vesicles
of Paracoccus denitrificans. Biochem. I, 150,527-536
[7] Souci, S.W., Fachmann, W., Kraut, H., 1989/1990: Food composition and nutrition
tables. 568, Wissenschaftliches Verlagsgesellschaft mbH, Stuttgart
 DBVBLOPMBN'l' OP A BIORBAC'l'OR !'OR SOl -SOLID FBRMBIITATION
PURPOSBS: BAC'l'BRIAL INSECTICIDB lI'BRMENTATION

DEISE M.F.CAPALBO
Centro Nacional de Pesquisa de Defesa da Agricultura
CP 69 CEP 13820 000 Jaguariuna SP Brazil

lRACEMA O. MORAES
Universidade Estadual Paulista - UNESP / IBILCE-DETA
CP 136 CEP 15054 000 S.J.Rio Preto SP Brazil

REGINA O. MORAES
Universidade Estadua1 de Campinas - UNICAMP / FEAGRI
CP 6011 CEP 13081-970 campinas SP Brazil

ABSTRACT

The development of Bacillus thuringiensis to obtain bacterial insecticide was

well studied in Brazil. In the submerged process it was studied in a
250-liter fermentor, that is semi-pilot size production, and offers the
opportunity scale up to a commercial scale.
When comparing with semi-solid processes, it can be concluded that in
developing countries this is a lower cost process that uses cheaper sources
of carbohydrates and proteins available in those regions. Therefore, it was
decided to study parameters and variables to solve the difficulties
encountered in this process. Heat and mass transfer, and sterility level are
important engineering aspects to be considered. Yield and productivity are
fundamental to cost studies and make the semi-solid process more feasible in
developing countries, using a properly designed bioreactor.

INTRODUCTION

Chemical insecticides proved to be toxic to non-target species, particularly

to man, in addition to the loss of effectiveness with constant use. The
environmental damage observed gave impetus to the search for new means of
insect control.
Biological control is one alternative that has been widely studied and
used in recent years.
The most studied and commercially available biocontrol agent is Bacillus
thuringiensis (Bt). Its toxic activity takes place primarily in the parasporal
crystal formed within the mother cell during sporulation, either in vivo or
in vitro. A mixture of Bt spores and crystals are commercially produced by
fermentation means throughout the world; however, in Brazil, economic reasons
have restricted its wider use.
The costs for Bt production depend to a large extent on the fermentation
 607
conditions (temperature, aeration, etc.), and to the cost of the medium. These
topics were studied and are described elsewhere [1,2,3]. They generated two
patents on Bt production in Brazil [4,5].
Based on the promising results obtained in those laboratory and pilot
scale production by submerged fermentation, the CNPDA, with the cooperation
of UNlCAMP and UNESP initiated studies to employ new methods of fermentation
to avoid the need for large aerated fermentors, by using the semi-solid
fermentation (SSF).
The preliminary studies were developed using small static flasks, and
have been described already [6,7]. The selection of low cost substrate, the
variety of Bt, engineering parameters, temperature, humidification, and
analytical methods were considered.
Those studies showed the adequacy of the SSF for Bt growth,
sporulation, and highly active endotoxin production [7,8]. Some advantages
and problems were also mentioned.
To scale up the process, it was necessary to stablish all the parameters
involved: the enginnering parameters like aeration and agitation rate,
diameter of the bioreactor, loading limit, and the physical characteristics
of the solid substrate.

MATERIAL AIm METHODS

Two types of column bioreactors have been studied: an aerated fixed bed, and
a fluidized bed fermentor.

Fluidized Bed COlumn Reactor

A fluidized bed column reactor was tested for its ability to overcome the
limitation of aeration that occured in static flasks. This kind of
fermentation increased the mixing and the heat transfer rate, and permitted
effective aeration of the mash. The reactor consisted of a jacket glass
column (inside diameter = 7 cm; length = 25 em) with a special internal
support at the bottom. The apparatus was connected to an air-compressed line
with controlled humidity and temperature [FIG.1).

FIGURE 1. Schematic diagram of the fluidized bed column reactor used for
SSF of Bacillus thuringiensis var. kurstaki.

1. Humidified air
2. Internal support
3. water from a water-bath
4. Glass column
5. Glass jacket
6. Fluidized mash
7. water
8. Air
9. Filter

Fixed Bed COlumn Reactor

An aerated, loosely packed fixed bed column will be used. The mash will be
composed of the preliminary selected substrates [7,8] , plus the inoculum of
608
Bt from a liquid fermentation, and tap water [3].
This reactor will be gently packed, aerated with humidified saturated
air, as proposed by Raimbault and Alazard [9]. Previous researchers succeeded
in determining some of the parameters (such as pH and growth kinetics) for
fungi using a similar apparatus [10]. This bioreactor is currently being
assembled to be used with the bacterium Bt.

RESULTS

Experimental data of the preliminary tests developed are being processed to

change or ameliorate the conditions of the columns tested.

BIBLIOGRAPHY

1. Moraes, 1.0.; Santana, M.H.A. and Hokka, C.O., The influence of

oxygen concentration on microbial insecticide production. Advances
in Biotechnology, 1981, 1, 75-79.

2. Martucci, E.T. and Moraes, 1.0., Substrate sterilization for Bacillus

thuringiensis. Process Biochem., 1982, 17(6), 35-36.

3. Moraes, 1.0. and Capalbo, D.M.F., The use of agricultural by-products as

culture media for bioinsecticide production. In Food Engineering and
Process Applications: unit operations, ed. M. LeMauguer and P.Jelen,
Elsevier Publishers, London, 1986, v. 2, pp. 377-81.

4. Moraes, 1.0., Patent BR PI 7608688. Processo de fermentacao submersa para

producao de inseticida bacteriano. lOp. 1984.

5. Moraes, I . 0., Patent BR PI 8500663. Processo de producao de toxina

termoestavel de Bacillus thuringiensis. 8p. 1985.

6. Capalbo, D.M.F. and Moraes, 1.0., Production of proteic protoxin by

Bacillus thuringiensis by semi-solid fermentation. 12th Simposio Anual
da Academia de Ciencias do Estado de Sao Paulo, Campinas, 1988. v. II,
p. 46-55.

7. Capalbo, D.M.F., Desenvolvimento de urn processo de fermentacao

semi-solida para obtencao de Bacillus thuringiensis Berliner. Campinas,
UNICAMP!FEA. 1989, 159p. (PhD thesis).

8. Capalbo, D.M.F.; Moraes, 1.0. and Moraes, R.O., Recent observations on

bacterial insecticide production by semi-solid fermentation technique.
Annales del I Reunion Latinoamericana y del Caribe em Biotecnologia,
Industria y Politicas Publcias para el Control Biologico de Plagas,
Barquisimeto, Venezuela, 1991.

9. Raimbault, M. and Alazard, D., Culture method to study fungal growth

in solid fermentation. European J. Appl. Microbiol. Biotecn., 1980, 9,
199-209.

10. Crooke, P.S.; Hong, K.; Malaney, G.W. and Tanner, R.D. Solid and
semi-solid state bioreactors: static, rotating and fluidized bed
fermentors. J. Biomass Energy Soc. China, 1991, 10(1-2), 1-17.
 COTINUOUS ETHANOL PRODUCTION BY ZYMOMONAS MOBILIS IN A
BIOREACTOR WITH FLAME MAKING CERAMIC CARRIERS

AKIHIKO MORl and TEfSUY A AOKI

Department of Chemical Engineering, Niigata University
Ikarashi 2-nocho, Niigata-shi, Niigata 950-21, JAPAN

ABSTRACT

Continuous ethanol production in a fixed bed bioreactor with flame making ceramic carriers,
APHROCELL, adhesively immobilized by Zymomonas mobilis was studied. In the single
bioreactor, the growth activity and specific conversion rate showed nearly same level at a
higher dilution rate more than 0.6 h- 1. In the multistage bioreactors, final ethanol concent-
rations was nearly same level of 7.9g·[-1 (76% yield) at a higher dilution rate more than 0.6
h-l, while ethanol productivity per total working volume increased as a dilution rate increased.
It is suggested that high productivity and high yield can be obtained simultaneously at a high
dilution rate with fixed bed multi-stage reactors.

INTRODUCTION

Many researchers have studied on ethanol production by Zymomonos mobilis in anaerobic or

aerobic submerged culture, and also in a bioreactor since 1965 [1-3].
While a flame making porous caramic carrier newly developed, APHROCELL, has
superior characteristics for an immobilizing carrier because it has large specific area, many
spaces in the configuration, then a large capacity for gas holding and cell keeping. Many
microorganisms have been used for immobilization to APHROCELL carriers, but genus
Zymomonas have not yet.
In this paper, we studied ethanol production using a fixed bed bioreactor with
APHROCELL carriers adhesively immobilized Zymomonas mobilis. Usually, in continuous
culture, rising productivity and abundant yield of products are anti nomic. But we will like to
propose both high productivity and high yield by the bioreactor, and we found widely
availability of APHROCELL in industrial use.

MATERIALS AND METHODS

Microorganism and medium

The microorganisms used in this study was ZymomonllS mobilis ATCC 10988. RM medium
used in this study consisted of 20 g glucose, 10 g yeast extract, 2 g KH2P04, 0.5 g
L( +)ascorbic acid as reductant and water in 11.
610
Bioreactor system
A bioreactor used in this study was a glass column of total volume of 300 ml packed with a
cylindrical cartridge of APHROCELL No. 13 of working volume of 200 m!. It was continuity
of flame making porous ceramics. It consisted of the hyperbolic paraboloids bodies with 208
pores in 25 mm, and developed by Kirin Brewery Co., Ltd. and Bridgestone Corp. The
outside of cartridge was coated by coating silicon (KE45, Shin-Etsu Chemical Co., Ltd.) then
fixed to the inside of the reactor with the silicon.
Medium was supplied to the lower side face of the column and flowed out from the upper
side face by a peristaltic pump with 2 channels. Nitrogen gas was supplied to the bottom of the
column and exhausted from the top. Sampling was carried out from the outflow tube at the
point after leaving the peristaltic pump. Multi-stage continuous culture was carried out by
connection of a new column with nitrogen gas supply to the outflow of the working one.

Cultivation
The seed culture with a test tube was inoculated by a loop of stock culture, and incubated
statically till foam generated, the preliminary culture with a test tube was inoculated by 3 loops
of seed culture, and incubated statically for 18 h. Batch culture of 150 ml With a Erlenmyer
flask was inoculated by a tube of the preliminary culture and incubated for 18 h under nitrogen
gas supply of 0.2 vvm. Then, the flask was connected with APHROCELL bioreactor and the
culture broth was flowed in circulation for 30 h under the nitrogen gas supply of 0.2 vvm, so
that the bacteria might adhere sufficiently. Thereafter, the bioreactor was separated from the
Erlenmyer flask and washed out with fresh medium for 10 h at a dilution rate of 1.5 h- 1 so that
almost free cells in the bioreactor were flowed out. After that, the bioreactor was operated at
various dilution rates under nitrogen supply of 0.2 vvm till the steady state was obtained. All
cultivations were performed at 30°C.

Analytical methods
Concentration of free cells was measured by optical density at 610 nm and calculated dry cell
weight by calibration curve previously obtained. Cell mass kept in the APHROCELL carriers
was washed out by 1000 ml of 0.1 % Tween 80 solution for 4 h in circulation after finishing
culture, then harvested, centrifuged, dried for 2 h at 105°C and weighed.
Specific growth rate (growth activity) was estimated by measurement of optical density of
a sample which was harvested from a sample point of each reactor in steady state, inoculated
into a shake flask with 100 ml RM medium, incubated by shaking at 140 reciprocation per min
and sampled at 0.5 h intervals for 10 h.
Ethanol concentration was estimated by potassium dichromate oxidation method after
distillation. Glucose was estimated by Somogyi method.
Every dilution rate was represented by flow rate per volume of a single reactor.

RESULTS AND DISCSSION

Kinetic characteristics of Z. mobilis

In RM medium, the growth rate of Z. mobilis was limited by yeast extract concentration, the
maximum specific growth rate, JIm, was 0.625 h- 1 and Ks was 1.86 g yeast extract·Z- 1 broth.
Growth yield was rather limited by glucose concentration than by yeast extract, and YXIS was
0.026 g cell· g-1 glucose.

Continuous culture
In a single stage bioreactor, ethanol and free cell concentrations of outflow had peaks at a
dilution rate of 0.6 h-l. Then ethanol gradually and free cell steeply decreased as dilution
increased, but cells did not be washed out at higher dilution rate. After 1800 h from start of
the culture, total cell mass kept in carriers was 0.5 g, which was calculated to be 2.6 g·Z-l
working volume. This was about lO-fold higher than free ceUs in the outflow at a high
dilution rate.
 611
Measurement of growth activity
In the single stage bioreactor, growth activity measurement was carried out The bacteria
showed synchronized growth at a hifh dilution rate (0 = 1.5 or 5.0 h- 1), but did not show at a
low dilution rate (0 = 0.15 or 0.5 h- ). The plots of specific growth rate of free cells in the
outflow against dilution rate had a peak value of 0.48 h- 1 at 0 = 1.0 h- 1 and showed slightly
down to 0.42 h-l at 0 = 5.0 h- 1. It was suggested that free cells had nearly same growth
activity and a lot of the cells were newly released from the carriers.

Multistage continuous culture

Multistage continuous culture was carried out with 3 bioreactors in series. The concentration
of final outflow was nearly same level of about 7.9 g./-1 (76% yield) at more than 0.6 h- 1 of
dilution rate (Fig. 1). While, ethanol productivity per total working volume of reactors inc-
reased linearly as dilution rate increased, and obtained 7.77 g·/-l.h- 1 at 0 3.0 h- 1 (Fig. 2). =
These results suggest that operation at a high dilution rate with multistage bioreactors can
obtain high productivity and high yield simultaneously in the continuous culture with fixed bed
bioreactors.
From these results APHROCELL bioreactor is expected to be available in the industrial
continuous production by Z. mobilis.

10 10
~~-

8
bL~g
jo~ -------~~
0-0 ~
...
8

O;::;~o/
'" 6
:::-
0
r:
'"
£;
~ o~
fl!
0.
6

W
0 _______
'0 4
U
r:
)w 4
0
U 0
0 Reactor 1
2 6. Reactor 2 2
0 Reactor 3 oy
ts.
0 0
0.0 1.0 2.0 3.0 0.0 1.0 2.0 3.0
D (1/h) o (l/h I
Figure 1. Relationship between dilution Figure 2. Relationship between dilution
rate (0) and ethanol concentration in rate (0) and ethanol productivity in
the multistage production. the multistage production.

ACKNOWLEDGEMENT

We are very grateful to Kirin Brewery Co., Ltd. for their support, and to Mr. A. Suzuki of
Kirin Brewery Co., Ltd. for supplying the carriers and reactors.

REFERENCES

1.Swings, J. and OeLey, J., The biology of Zymomonas. Bacteriol. Rev., 1977, 41, 1-46

2.Sahm, H., Ethanol production by bacteria. In Bioprocess Engineering, The First

Generation, ed. T. Ghose, Ellis Horwood Ltd., Chichester, 1989, pp. 301-315.

3.Amin, G. and Ooelle, H. W., Production of high ethanol concentrations from glucose using
Zymomonas mobilis entrapped in a vertical rotating immobilized cell reactor. Enzyme Microb.
Technil., 1990, 12, 443-446.
 DESIGN AND SCALE-UP OF BIOREACTORS FOR
MICROBIAL POLYSACCHARIDE FERMENTATION

YOSHINORI KAWASE AND MASANARI TSUJIMURA

Biochemical Engineering Research Center
Department of Applied Chemistry
TOYo university, Kawagoe, Saitama 350 ,Japan.

ABSTRACT

We have discussed the effects of rheological properties of

microbial polysaccharides on the design and scale-up of
bioreactors. Oxygen transfer rates and gas holdups are measured
using a bubble column and two external-loop airlift bioreactors.
The volumetric mass transfer coefficients significantly decrease
with increasing viscous non-Newtonian flow behavior of
fermentation media.

INTRODUCTION

Xanthan is an anionic exopolysaccharide produced by bacteria of

the genus Xanthomonas. Xanthan gum has many applications as a
thickener in the food, cosmetics, and pharmaceutical industries.
Pneumatically-agitated fermentors increasingly are being
considered for use in the fermentation industries in place of
the traditional mechanically-agitated bioreactor design because
of the potential savings in both capital and operating cost.
However, the heat and mass transfer rates may not be sufficient.
In this study, we have made an extensive investigation of
the frictional gas holdup in the riser section of external-
circulation-loop airlifts, and of the volumetric liquid-phase
mass transfer coefficient using liquid phases having non-
Newtonian characteristics.

EXPERIMENTAL

The two external-circulation-loop airlifts used in this work

 613
were composed of a riser having a diameter of 0.15 m and a
downcomer having inside diameters of 0.105 m (Airlift I) or 0.07
m (Airlift II). The values of the downcomer-to-riser cross-
sectional area ratio (Ad/Ar) are 0.459 and 0.204, respectively.
The airlifts were equipped with a perforated plate sparger
located at the base of the riser section. The kLa coefficients
were determined by the dynamic method.
Tap water, carboxymethylcellulose (SIGMA Chemical Co.), and
Gum Xanthan (SIGMA Chemical Co.) were used as the liquid.
Aqueous solutions of CMC and Gum Xanthan represented non-
Newtonian flow behavior.

RESULTS and DISCUSSION

Figure 1 shows the kLa data obtained in Airlift I with 0.1 wt%
Xanthan gum aqueous solution (n=O. 402) and 0.15wt% CMC aqueous
solution (n=O. 796). It can be seen that the kLa coefficient
significantly decreases with increasing viscous non-Newtonian
flow behavior or decreasing n. As shown in Fig. 2, the kLa
coefficients for highly viscous non-Newtonian fermentation media
in Airlift II are less than those for water, as well as the
resul ts for Airlift I shown in Fig. 1. In order to improve
oxygen transfer performance, four short draft tubes covered with
perforated plates were inserted in the riser. It is seen in
Fig. 2 that, as expected, the performance of oxygen transfer was
enhanced. In high viscous non-Newtonian media, large bubbles
were formed by coalescence, and a large number of very tiny
bubbles resulted from break-up. For comparison, the
correlations of Bello et al. ( 1) and popovic and Robinson (2 )
are also given in Figs. 1 and 2. The predictions of correlation
for water proposed by Bello et al. (1) agree very well with the
experimental results. On the other hand, the correlation for
non-Newtonian fermentation media (2) predicts considerably
higher kLa coefficients compared with the present data for
Xanthan gum and CMC. A reliable correlation for kLa
coefficients is required.
Nomenclature

kLa volumetric mass transfer coefficient

n flow index in power law model
Usg superficial gas velocity

REFERENCES

1. Bello, R.A., Robinson, C.W. and Moo-YOung, M. Prediction of

the volumetric mass transfer coefficient in pneumatic
contactors. Chem. Eng. Sci., 1985,40, 53-58.

2. Popovic, M. and Robinson, C.W. Mass transfer studies of

external-loop airlifts and a bubble column. AIChE J, 1989,
35, 393-405.
614

5
/
/

'.'
. /
/
/

2 '. / /
....... ': /
("
/ .. '

:::: '\0- 2
'-'
......................................... ,.......;/-:...... :: •....................
: / •. ' X
0: / ... XX
: / .' X
5 / •••.• X 0
/ :.
/. ...: 0
/'" : X 0
,;/
./ :
X:
../ .l
,.' / :
/ :

2 5 2 5
10- 2
US9 (l11/s)
Figure 1. Volumetric mass transfer coefficients in Airlift!!
o water 0 Xanthan gum )<. CMC
Bello et al. ( 1 )
- - - Xanthan gum, ..... - CMC Popovic and Robinson (2 )

2
..('>
--.
-",

;;

2 5 2 5
10- 2
US9 (l11/s)

Figure 2. Volumetric mass transfer coefficients in AirliftII

(symbols as in Fig. 1.)
• draft tubes with perforated plates
 APPLICATION OF A BIOREACTOR SYSTEM TO SOY SAUCE PRODUCTION

TAKASHI HAMADA, YAICHI FUKUSHIMA AND HIROSHI MOTAI

Research Division, Kikkoman Corporation, 399 Noda,
Noda-shi, Chiba, 278, Japan

ABSTRACT

Continuous fermentation of soy sauce with immobilized glutaminase and

immobilized cells of Pediococcus halophil us, Zygosaccharomyces rouxii and
Candida versatilis was investigated. Glutamic acid production and both lactic
acid and alcohol fermentation continued for over 100 days without any
problems. The total time required for the production of soy sauce by the
bioreactor system was much less than that of the conventional method
(about 1/10). The soy sauce made by this system was similar to a conven-
tional one in regard to chemical components and the quality of flavor.

INTRODUCTION

In the conventional method of brewing soy sauce, cooked soybean and

roasted wheat are mixed with seed spores of Aspergillus sojae and/or
Aspergillus oryzae to make koji in 2 days of solid culture. Then koji is
mixed with brine to make moromi. At the first stage of moromi mash, Pedio-
coccus halophilus grows and produces lactic acid which lowers the pH.
Accompanying the decrease in the moromi pH, vigorous alcohol fermentation
by Zygosaccharomyces rouxii occurs. As the result, 2-3% ethanol and many
kinds of aroma components are produced by this yeast. At the same time
phenolic compounds such as 4-ethylguaiacol (4-EG) and 4-ethylphenol,
which add characteristic aroma to soy sauce, are also produced by other
types of yeasts such as Candida versatilis and Candida etchellsii.
It takes over 6 months for the entire fermentation and aging of the
moromi mash. Therefore, shortening this period is significantly important.
In this study, we investigated a bioreactor system for soy sauce produc-
tion, which consisted of reactors containing immobilized glutaminase and
immobilized cells of P. halophilus, Z. rouxii and C. versatilis.

MATERIALS AND METHODS

Immobilization and reactors

Glutaminase from Candida famata was immobilized with Chitopearl followed by
616
cross-linking with glutaraldehyde. P. haJophilus cells were immobilized in AS
gel [1] consisted of alginate and colloidal silica, and Z. rouxii and C. versati-
lis cells were immobilized in alginate gel [2]. Plug flow reactors (total vol.:
1.8 1; H/D=10.8 and total vol.: 7.5 1; H/D=5.4) were used for immobilized
glutaminase and immobilized P. haJophilus cells, respectively. Airlift reactors
with a draft tube (total vol.: 27 1; H/D=6.1 and total vol.: 1 1; H/D=3.6) were
used for immobilized cells of Z. rouxii and C. versatilis, respectively.

Preparation of raw liquid

Cooked soy bean and roasted wheat were mixed with culture broth of A.
oryzae [3] and were enzymatically hydrolyzed in the presence of less than
10% NaCl and at a temperature 45°C for 3 days. This liquid fraction was
used as raw liquid for the bioreactor.

RESULTS AND DISCUSSION

Manufacturbng processes
The processes of soy sauce production using the bioreactor and the conven-
tional methods are quite different as to the periods of lactic acid and alco-
hol fermentation. The entire fermentation with the conventional method takes
several months. On the other hand, only about 2 days are needed for the
bioreactor method using the immobilized whole cells.
In the bioreactor method, the first step is preparation of raw liquid
for bioreactors. Raw liquid was successively passed through, firstly, a
glutaminase reactor to increase glutamic acid, secondly, a P. haJophilus
reactor to carry out lactic acid fermentation and, thirdly, a Z. rouxii reactor
to carry out alcohol fermentation, and a C. versatilis reactor to produce
phenolic compounds such as 4-ethylguaiacol. Two reactors containing immo-
bilized yeast cells were set in parallel and the flow rate of the feed solution
to Z. rouxii and C. versatilis reactors was set in the ratio of ten to one.

Contbnuous fermentation
The fermentation by immobilized cells of P. haJophilus, Z. rouxii and C.
versatilis continued for over 100 days without any troubles such as microb-
ial contamination. A consistently increased level of glutamic acid (the in-
crease being in the range of 0.3-0.4%) was found in the effluent from
glutaminase reactor, with a residence time of 0.7 h. Lactic acid was pro-
duced by immobilized cells of P. haJophilus in quantities of 0.7-1.0% at a
residence time of about 6 h and consequently the pH declined to 4.9-5.0,
similar to that of the conventionally brewed soy sauce. Ethanol, often used
as an index of alcohol fermentation, was produced constantly at a residence
time of about 26 h by immobilized cells of Z. rouxii in quantities of 2.5-2.7%.
This is the standard content in conventional soy sauce. About 10 ppm of 4-
EG was produced by immobilized cells of C. versatilis at a residence time of
about 10 h after passing through the reactor, and the final 4-EG content
resulting after mixing the two fermented liquid issuing from the reactors of
Z. rouxii and C. versatilis was about 1 ppm, which is the optimum concentra-
tion in conventional soy sauce. The total residence time in lactic acid and
alcohol fermentation was about 30 h in this system, compared with the
conventional fermentation periods of 2-3 months required to produce the
same amounts of lactic acid and ethanol.
High numbers of viable cells were present in the gel and in the liquid
for each reactor. The number was 10-100 fold higher than that in moromi
mash. The shortening of the fermentation period in the bioreactor method is
possibly due to the high density of immobilized cells in the gel and free
 617
cells in the liquid. Furthermore, from our previous results [4], free cells of
Z. rouxii and immobilized cells of C. versatilis appear to contribute mainly to
the production of ethanol and 4-EG in each reactor.

Properties
The main chemical components of the fermented liquid after passing through
bioreactors were examined. The content of lactic acid, glucose, ethanol and
formal nitrogen which were important components of soy sauce were kept in
proper proportion. Aroma components present in both the bioreactor-
produced soy sauce and the conventional sauce were not different qualita-
tively. However, the former sauce had more isoamyl alcohol and acetoin, and
less isobutyl alcohol, ethyl lactate, 4-hydroxy-2(or5)-ethyl-5(or2)-methyl-
3(2H)-furanone and 4-hydroxy-5-methyl-3(2H)-furanone than the latter. In
order to evaluate the quality of soy sauce, a sensory test was carried out.
For example, the intensity of the alcoholic, fresh, sweet, acid, and sharp
odors were compared between the bioreactor and conventional soy sauces.
The odors are important for the judgment of the quality of soy sauCe.
Although bioreactor soy sauce was a little weaker in aroma and fresh odor
than conventional soy sauce, the quality of the former was generally judged
to be similar to that of the latter [5].

CONCLUSIONS

Soy sauce production by a bioreactor system was investigated. Both lactic

acid and alcohol fermentation and 4-EG production continued for over 100
days at a satisfactory level. The total time required for the production of
soy sauce by the bioreactor system including enzymatic hydrolysis of raw
materials, fermentation with immobilized whole cells and refining process is
only about 2 weeks. This is considerably shorter than about 6 months in
the conventional method of soy sauce brewing consisted of koji making,
fermentation and aging in moromi mash, and refining process. The bioreactor
soy sauce was considered to be similar to the conventional sauce in regard
to chemical components and the quality of flavor. Therefore, it seems that
this bioreactor system is practical for the production of soy sauce.

REFERENCES

1. Fukushima, Y., Okamura, K., Imai, K. and Motai, H., A new immobilization
technique of whole cells and enzymes with colloidal silica and alginate.
Biotechnol. Bioeng., 1988, 32, 584-594.
2. Hamada, T., Sugishita, M. and Motai, H., Continuous production of 4-
ethylguaiacol by immobilized cells of salt-tolerant Candida versatilis in
an airlift reactor. J. Ferment. Bioeng., 1990, 69, 166-169.
3. Fukushima, Y., !toh, H., Fukase, T. and Motai, H., Continuous protease
production in a carbon-limited chemostat culture by salt tolerant Aspe~
gillus oryzae. illml. Microbiol. Biotechnol., 1989, 30, 604-608.
4. Hamada, T., Sugishita, M. and Motai, H., Contribution of immobilized and
free cells of salt-tolerant Zygosaccharomyces rouxii and Candida versati-
lis to the production of ethanol and 4-ethylguaiacol. A.Iuli. Microbial. Bio-
technol., 1990, 33, 624-628.
5. Hamada, T., Sugishita, M., Fukushima, Y. and Motai, H., Continuous
production of soy sauce by a bioreactor system. Process Biochem., 1991,
26, 39-45.
 SUBSTRATE REMOVAL CHARACTERISTICS IN AN

1M MOBILISED CELL FLUIDIZED BED REACTOR

S M Rao Bhamidimarri
Process and Environmental Technology Department
Massey University, Palmerston North, New Zealand

ABSTRACT

Substrate bioxidation characteristics were studied in an immobilized cell fluidized bed bioreactor. The
reaction-diffusion mechanisms in conjunction with the reactor flow characteristics were modelled to
predict the substrate transport and bioreaction in the reactor. The significant feature of the model is
that it permits evaluation of active biomass in the reactor. A combination of orthogonal collocation and
Runge-Kutta-Gill techniques was used to solve the models. Experimental evidence is presented to
demonstrate the validity of the models.

INTRODUCTION

The application of immobilized cell fluidized bed reactors for substrate conversion, in particular in
wastewater treatment has been attempted by many researchers in recent years. The processes studied
include denitrification (1), nitrification and organic carbon removal (2). In fluidized beds, in which small
particles offer a large surface area. The immobilized cells utilize the substrate in the bulk liquid and
grow into a biofilm resulting in reactor biomass concentrations of an order of magnitude greater
compared to the conventional dispersed growth reactors. However, the volumetric reaction rates in
such reactors are not proportional to the increase in biomass concentration. In order to predict the
substrate removal in these reactors, it is necessary to identify various process mechanisms and to
quantify them. Although, models based on reaction - diffusion mechanisms were developed (3), the
fluidized-bed bioreactors have been continued to be investigated largely on an input-output basis.
Reliable experimental techniques for the determination of dry and wet densities of biofilm and the
effective substrate diffusivity into biofilm have been reported (4).

Predictive models for substrate removal in a fluidized bed reactor with experimental verification for
substrate removal are discussed in this paper.

MODEL EQUATIONS
The biofilms supported on granular media assume spherical geometry in liquid fluidized beds. The
methodology used for chemical reactors containing porous spherical catalyst particles can be adapted
for analysing the substrate conversion in an immobilized cell fluidized-bed reactor.

Diffusion - Reaction
For a biofilm immobilized on support particle and steady-state diffusion of substrate and its conversion
for a pure growth process is described by
 619
ciS (1 )
cPr
where Q is the biofilm dry density. Monod type growth results in the non-linear term on the right hand
side of the above equation, which in the absence of external mass transfer resistance, is subject to the
following boundary conditions:

at r = rbP S = Sb and
at r = r; ds = 0 (2)
dr
S=S'{

rj the inactive radius represents the radius of the core of the particle in which substrate biooxidation
does not occur. The inactive core includes the inert support and the inner layers of biofilm which may
not receive sufficient oxygen to catalyse substrate oxidation.

The substrate concentration at the inactive boundary is calculated based on the relative diffusivities of
oxygen and substrate and the stoichiometry of substrate biodegradation. In a thin section of the
fluidized-bed column reactor, if the bulk substrate concentration is assumed to be uniform Srj and
therefore rj can be evaluated from equations (1) to (4) via numerical integration.

Reactor Model
For a uniform fluidized bed with no significant radial gradients and axial dispersion.
where 11, the effectiveness factor, is defined as the ratio of reaction rate when diffusional resistances
are significant to the rate of reaction in the absence of diffusion effects. Therefore equation (5)
becomes

_.3 ds
dSb rbp x -I~
dr .. (3)
US1 dH +
Ic)
--:,---:=---
(';p - Qf

With the initial condition

(4)

The particle model (equation 1) and the reactor model (equation 3) are solved simultaneously to obtain
the concentration profile through the reactor.

Experimental
The substrate removal was studied in a fluidized bed reactor system as described by Bhamidimarri (4).
The substrate used was phenol and the oxygen was supplied by dissolving pure oxygen in the feed
solution to the fluidized bed column reactor.

RESULTS AND DISCUSSION

Model Analysis
The active biomass concentration varies with the height of the reactor because of varying oxygen and
phenol concentrations and this gives rise to changing inner boundary condition for the particle model
(equation 1). A continuous solution to such a moving boundary problem with non-linear kinetics is not
easy to find. Therefore, a step-wise solution was developed for the analysis of the model equations.
The column reactor was considered to be a series of thin sections. Assuming the substrate
concentration in each thin section to be uniform, the reactor models were analysed for the first section
to evaluate the concentration, which was then used as the inlet concentration for the next section. The
620
diffusion-reaction equation was solved by an orthogonal collocation technique and the Runge-Kutta-Gill
routine was used to solve the reactor flow model. This procedure was repeated to generate
concentration profile in the reactor.

Substrate Removal in the Reactor

The stoichiometry of phenol bio-oxidation is well established and it is reported that 1.45kg of oxygen
are required per kg of phenol (5). Therefore, oxygen concentration is readily evaluated.

The measured oxygen concentration profiles along the height of the reactor were compared with the
predicted values. The parameters used in the model analysis are those of Bhamidimarri (4). Typical
results shown in Figures 1 prove the validity of the methodology adopted for the development of the
models.

1.0
...:1
"01
e.
U
. 0.8

c::
0
u 0.6
N
0
en 0.4
rJj
Q)
-1
c::
0
.r-! 0.2
rJj
c::
Q)
E
.r-! 0
0.2 0.4 0.6 0.8 1.0
'0 0

dimensionless height, H/He

Figurel. Oxygen concentration profile in the reactor for
r c =O.l4mm, biofilm thickness=0.15mm, 0b.=28mgjL,
1
o=observed, line=predicted.

NOMENCLATURE
Ks saturation constant, mg L· 1
r particle radius, cm
rbP radius of particle with biofilm, cm
rc support particle radius, em
Sb substrate concentration, g cm·3
Sbi influent bulk substrate concentration, g cm-3
USi superficial liquid velocity, cm S-1
YI cell yield

REFERENCES
1. Jeris, J.S., Beer, C and Mueller, J.A. (1974). High rate biological denitrification using a
granular fluidized-bed. J. Water Poll. Control Fed., 46, 2118-2128.
2. Jeris, J.S., Owens, R.W., Hickey, R. and Flood, F. (1977). Biological fluidized-bed treatment
for BOD and nitrogen removal. J. Water Poll. Control Fed., 49, 816-831.
3. Mulcahy, L.T. (1978). Mathematical model of the fluidized-bed biofilm reactor. PhD ThesiS,
University of Massachussetts, Amherst, MA.
4. Rao Bhamidimarri, S.M. (1985). Biofilm characteristics and their role in a fluidized-bed
bioreactor. PhD TheSiS, University of Queensland, St Lucia, Qld.
5. Luthy, R.G. (1981). Treatment of coal coking and coal gasification wastewaters. J. Water Poll.
Control Fed. 53, 325-339.
 SOLVENT EXTRACTION OF FLAXSEED

F. Shahidi, P.KJ.P.D. Wanasundara and R. Amarowicz

Department of Biochemistry, Memorial University of Newfoundland
St. John's, NF, Canada, AlB 3X9

ABSTRACT

Extraction of ground flaxseed with hexanes or a two-phase solvent system consisting

of hexanes and a polar solvent mixture was carried out. The polar solvent was alkanol, 95%
(v/v) alkanol or 95% (v/v) alkanol containing 10% (w/v) dissolved ammonia. The latter
solvent extraction system using methanol, produced a meal with high content of proteins and
also retained its amino acid compositional integrity. It also lowered cyanogenic glycosides
and phenolic compounds of the seed. Changing the conditions of parameters of extraction
with methanol-ammonia-waterlhexane resulted in over 90% removal of the major antinutrient
of flaxseed.

INTRODUCTION

Flax or linseed (Linum usitatissimum L.) is the third major oilseed crop in Canada.
The oil of flax (linseed oil) is rich in polyunsaturated fatty acids especially a-linolenic acid
and is used mainly as an industrial oil [1]. Flaxseed meal is currently used as an animal feed.
Presence of antinutrients such as cyanogenic glycosides, phenolic compounds and anti-vitamin
B6 [2] has hindered its utilisation in food formulations [3]. The main objective of this study
was to study the effect of alkanol-ammonia-waterlhexane extraction on the removal of anti-
nutrients of flaxseed meal.

MATERIALS AND METHODS

Flaxseed obtained from Omega Nutrition Company (Vancouver, BC) was extracted by
a two-phase solvent extraction system consisting of alkanol-ammonia-waterlhexanes [4]. The
alcohols used were methanol, ethanol and isopropanol with or without ammonia (10%, w/w)
and water (5%, v/v). The contents of crude proteins [5], cyanogenic glycosides [4], condensed
tannins [6] and total phenolic acids [7] of the extracted meals were determined. The
efficiency of methanol-ammonia-waterlhexane extraction system was studied for the removal
of cyanogenic glycosides by changing the extraction parameters; water 5, 10 and 15% (v/v)
in methanol phase, 15 and 30 min quiescent period and 6.7 and 13.3 polar phase-to-seed ratio
(R). Effect of a multistage extraction (2 and 3 times) was also investigated.
622
RESULTS AND DISCUSSION
The yields of oil, meal and solids following extraction with different solvent systems
are provided in Table 1. Methanol-ammonia-waterlhexane removed more polar components
(solids) and enhanced protein content of the meals (15%). Amino acid profile indicated that
flaxseed meal is high in glutamic acid but low in sulphur-containing amino acids. However,
solvents used for extraction did not change the amino acid composition of the recovered
meals.

TABLE 1
Mass balance of flaxseed as affected by extraction with different solvent systems

Yield, on a dry weight basis (%)

Solvent Loss
Meal Oil Solids (%)
(A) Hexanes only 48.9±1.0 49.2iO.6 0.0 1.9
(B) Methanol!hexanes 46.7±1.0 45.9±1.9 5.2±O.1 2.2
(C) Methanol-water!hexanes 46.7±2.0 48.0iO.5 4.8±O.1 0.5
(D) Methanol-ammonia/hexanes 47.6±2.0 46.6iO.8 4.8iO.7 1.0
(E) Methanol-ammonia-waterlhexanes 46.4±2.0 47.1iO.8 5.7±O.1 0.8
(F) Ethanol-ammonia-waterlhexanes 48.1±1.0 46.8±1.8 4.2±O.1 0.9
(G) Isopropanol-ammonia-water!hexanes 50.0±3.0 48.8±0.5 no phase 1.2
separation

Figures 1 and 2 show the percentage removal of anti-nutrients of flaxseed meal. The
content of condensed tannins of flaxseed was 136±13 mg/l00g, as (+)-catechin equivalents.
The solvent extraction systems employed were able to remove 24-74% (Figure 1) of
condensed tannins present in the meal. The total content of phenolic acids of flaxseed meal
was about 220 ± 13 mg/lOOg as ferolic acid equivalents, and 12-48% (Figure 1) of these were
removed depending on the solvent system employed.

~
>
0
75
E
Q) 60
....
Q)
0> 45
.mc
Q) 30
....
0

a.. 15
Q)

Figure 1. Percentage removal of condensed tannins, _ ; total cyanogenic glycosides, OBI;

and total phenolic acids, 0 ; of flaxseed meal by different solvent extractions (B-G refers
to combinations in Table 1).
 623

ctI 90
>
~ 7fr
0)
....
0)
60
C>
g 45
c:
~ 30
....
~ 15

linustatin
Figure 2. Percentage removal of linustatin and neolinustatin under different extraction
conditions for 15 min at R=6.7 with 10% (w/w) ammonia in methanol containing 5% water,
~; 10% water, _; 15% water, D; 5% water and R=13.3, ~; 5% water and 30 min,
I[JI] ; 5% water, 30 min, R=13.3, Im!; 5% water, twice, _ ; and 5% water, three times, E3.

The predominant cyanogenic glycosides in flaxseed meal were linustatin 4.42 ± 0.08
mg/g) and neolinustatin (1.90 ± 0.03 mg/g). The content of both these glycosides was reduced
by 57% after one extraction with methanol-ammonia-water/hexane. Increasing the water
content in the polar phase removed more cyanogenic glycosides (Figure 2) but gave a sticky
meal. Doubling the volume of the extraction solvent and duration of the extraction period
resulted in the removal of over 80% of cyanogenic glycosides. A three stage extraction gave
a meal with 90% of its cyanogenic glycosides removed. The efficiency of the removal of
condensed tannins, cyanogenic glycosides and total phenolic acids by the solvent mixtures was
in the order of methanol-ammonia-water» isopropanol-ammonia-water > ethanol-ammonia-
water. Removal of the antinutrients of flaxseed meal provides a mean for better utilisation
of meals in feed and possibly for human food formulation.

REFERENCES
1. Anderson, L., Composition of linseed oil and its relation to usage present and future. Flax
Inst. US Proc., 1971,41, 16-20.
2. Klosterman, H.J., Lamoureux, G.L. and Parsons, J.I., Isolation, characterisation and
synthesis of linatine, a vitamin B6 antagonist from flaxseed. Biochem., 1967, 6, 170-77.
3. Singh, N., Linseed as a protein source. Indian Food Packer, 1979,33, 54-7.
4. Wanasundara, P.K.J.P.D., Amarowicz, R., Kara, M.T. and Shahidi, F., Removal of
cyanogenic glycosides of flaxseed meal. Food Chern., 1993, In press.
5. AOAC, Official Methods of Analysis. 151h ed. Association of Official Analytical
Chemists, Arlington, Virginia, 1990.
6. Naczk, M. and Shahidi, F., The effect of methanol-ammonia-water treatment on content
of phenolic acids of canola. Food Chern., 1989,31, 159-64.
7. Krygier, K., Sosulski, F.W. and Hogge, L., Free, esterified and insoluble bound phenolic
acids 2. Composition of phenolic acids in rapeseed flour and hulls. J. Agric. Food Chern.
1982, 30, 334-36.
LARGE SCALE PREPARATION OF HIGHLY-PURIFIED EGG YOLK PHOSPIIATIDYLCIIOLINE
BY HPLC

HIDEAKI KOBAYASHII , HITOSHI NARABEI , MINEO HASEGAWA 1 ,

FUMIYA HARADA2 , KAHOKa NOMURA 2 , KEISHI KITAGAWA 2
lResearch Institute of Q.P.Corp., 5-13-1,Sumiyoshi-cho, Fuchu-shi,
Tokyo 183, Japan, 2 Kumiyama Laboratory of YMC Co. ,LTD., 249,
Morimurahigashi, Kumiyama-cho, Kuse-gun, Kyoto 613, Japan

ABSTRACT
We studied to develop the system which refines egg phosphatidylcholine
(PC) for pharmaceutical and food use. HPLC was applied with silicagel
particles(YMC-GEL S-50) as the column packing and ethanol/water(92:8) as
the safety mobile phase. Six hundred grams of egg lecithin PL-100H (Q.P.
Corp.) were separated into PC and phosphatidylethanolamine(PE) at a time.
The purity of obtained PC was about 100%. And we could easily recover
the mobile phase by distillation and use it repeatedly.
Then we examined oxidative stability. After 35 days in the presence
of the air at 20 \:, the POV of purified PC was less than 2meq/kg.

INTRODUCTION
Phospholipids such as phosphatidylcholine(PC) have the tendency to form
molecular bilayers called liposome. And phospholipids also have some
nutritional functions such as the serum cholesterol reducing effect and
so on. In recent years, purified PC was needed for the drug delivery
system in the pharmaceutical field and for a functional food use. So we
tried to develop the system which refines PC by HPLC with safety organic
solvents.
MATERIALS AND METHODS
Materials
Egg yolk lecithin was obtained from Q.P.Corp. (PL-100H). It contains
about 80% PC, 15% phosphatidylethanolamine(PE) and a slight amount of
cholesterol (CH), sphingomyelin(SPM) and lyso-phosphatidylcholine(LPC). A
commercial purified egg PC for pharmaceutical use was purchased from
NOF Corp. (NC-10S). This purity was more than 98%.
 625
Large scale preparation of egg PC
Six hundred grams of egg yolk lecithin PL-100H were dissolved in ethanol
and purified by preparative HPLC (PLC-200P made by YMC Co.,LTD.)[ YMC-
GEL S-50(1000xlOOmm 1.0.) as the column packing, ethanol/water(92:8)
as the mobile phase, flow rate= 600ml/min, monitoring by UV215nm : This
system is explosion-proof and the maximum flux is about 3l/min. ]
Oxidative stability of egg PC
The oxidative stability of PC was evaluated as follows; The samples were
kept in glass bottles in the presence of the air with no light at 4 t
and 20 t. After 0,7,14,and 35 days, the peroxide-value(POV) was
measured.

RESULTS AND DISCUSSION

Large scale preparation of egg PC
Figure 1 shows the HPLC chromatogram of PL-100H in preparative-scale. By
using preparative column( 1000 x100mm 1.0.), 600g of egg yolk lecithin
PL-100H were separated into PC and PE at a time. The fraction of the
second half of PC peak(2.2h-5.4h) was evaporated to dryness. Then 375u
of purified egg PC was obtained(yield: 62.5%) and the purity was 99.9%
by Iatroscan method. More than 98% of the mobile phase were recovered by
evaporation. After the ratio of ethanol/water were adjusted, it was
able to be used as the mobile phase repeatedly.

PE PC

Time (h)
Figure 1. HPLC chromatogram of PL-100H in preparative-scale

Molecular species of egg PC

There was not much difference between fatty acid composition of HPLC-
purified egg PC and that of a control ( data not shown, a control egg
PC; separated by TLC [chloroform/methanol/water(65:25:4)] from PL-100H,
and the PC was extracted by chloroform/methanol(2:1) from TLC plate and
evaporated to dryness).
Table 1 shows the molecular species of HPLC-purified egg PC. The
molecular species of PC was analyzed by HPLC using the method of Smith,M.
et al. 1) There were some differences between HPLC-purified egg PC and
a control.
626
Table 1
Molecular species of HPLC-purified egg PC
Molecular species
HPLC-purified egg PC Control egg PC
C1-position C2-position (%) (%)
C16-0 / C22-6 3.9 13.6
C16-0 / C20-4 2.2 5.8
C16-0 / C18-2 22.5 24.3
C16-0 / C18-1 44.9 35.9
C18-0 / C20-4 2.2 3.9
C18-0 / C18-2 11.2 6.8
C18-0 / C18-1 11.2 5.8
others 1.9 3.9

Oxidative stability of egg PC

Table 2 shows changes of POV of HPLC-purified egg PC in the presence of
the air. The content of unsaturated fatty acids of egg PC is about 50%
(data not shown). So it is easy for egg PC to be oxidized.
But the POV of HPLC-purified egg PC did not rise after 35 days at
209C. And oxidative stability of HPLC-purified egg PC was significantly
higher than that of a control(a commercial egg PC). This result could
not be explained on the fatty acid composition of PC. So we consider
that the purity and molecular species of PC may have some effects on
the oxidative stability.

Table 2
Changes of POV of HPLC-purified egg PC in the presence of the air

HPLC-purified egg PC (meq/kg) Control" (meq/kg)

days
at 49C at 209C at 20 9C
0 0 0 0
7 0 0 0
14 0 0 2
35 0 0 76
a =a commercial egg PC
REFERENCES
1. Smith,M. and Jungalwala,F.B., J.lipid Res.,1981, 22, 697-704

APPENDIX
This work has been done in part as one of the R&D subjects of 'The
Japanese Research and Development Association for the High Separation
System in Food Industry'.(1988-1992)
 EXTRACTION AND CONCENTRATION OF OMEGA-3
FATTY ACIDS OF SEAL BLUBBER

F. Shahidi,t R.A. Amarowicz,t J. Synowiecki1 and M. Naczk1,l

lDepartment of Biochemistry, Memorial University of Newfoundland
St. John's, Newfoundland, Canada, AlB 3X9
and
lJ>epartment of Nutrition and Consumer Studies, St. Francis Xavier University
Antigonish, Nova Scotia, Canada, B2G ICO

ABSTRACT

Seal blubber lipids consist of approximately 99% neutral triacylglycerols and I % polar
components. Blubber lipids contained 24% long-chain omega-3 fatty acids consisting of
approximately 9% eicosapentaenoic acid (EPA), 5% decosapentaenoic acid (DPA), and
10% docosahexaenoic acid (DHA). The content ofDPA was higher in seal blubber lipids
as compared with other sources of marine oil, except mammalian species. Intramuscular
lipids of seal meat were also rich in their content of DPA and other omega-3 fatty acids.
Omega-3 concentrated blubber lipids containing 76% long-chain omega-3 fatty acids were
prepared by extraction/precipitation of saturated/monounsaturated lipid fatty acids. Stability
of seal blubber lipids was better than that of fish oils.

INTRODUCTION

Marine products play an important role in nutrition and health status of humans.
Polyunsaturated fatty acids have long been recognized as desirable dietary components.
Presence of long-chain omega-3 fatty acids namely eicosapentaenoic acid (EPA),
decosapentaenoic acid (DPA) and docosahexaenoic acid (DHA) in seafoods is responsible
for their health-pro properties. These fatty acids have been shown to be responsible for
biochemical and physiological changes in the body. While DHA is essential for proper
functioning of the eye and may have a structural role in the brain, EPA serves as a
precursor to prostaglandins and thromboxanes; DPA in an intermediary species between
EPA and DHA. Therefore, omega-3 fatty acids have beneficial effects in the prevention
or possible treatment of coronary heart diseases, cancer, diabetes, high blood pressure,
and autoimmune diseases [1]. Seal blubber lipid may have the added advantage of
preferential absorption by the body since its long-chain omega-3 fatty acids are present
628
mainly in 1- and 3-positions of the triacylglycerol molecule. The omega-3 fatty acids of
fish oils are known to be randomly distributed and being present more in the 2-position
of the triacylglycerols.
The present study was undertaken to examine the fatty acid composition of blubber
of harp seal (Phoca groenlandica). Effects ofurea complexation on enrichment of omega-3
fatty acids in seal blubber lipids were also examined.

MATERIALS AND METHODS

Seal oil was prepared from seal blubber by denaturation and precipitation of protein
residues, their subsequent filtration and repeated washing of the oil with water, possibly
in the presence of antioxidantJsequestrant combinations. Enrichment of omega-3 fatty acids
was achieved by a urea complexation procedure which involved hydrolysis, complexation,
fIltration and esterification steps [2]. Fatty acid methyl esters in preparations were
quantified by transmethylation of lipids and were separated on a 30 mm x 25 mm Ld.
fused silica capillary column using a Perkin Elmer 8500 gas chromatograph as described
elsewhere [3].

RESULTS AND DISCUSSION

Blubber from six animals were used in these studies. Proximate composition of blubber
from these animals is summarized in Table 1. The content of neutral lipids accounted for
98.89 ± 0.16% of the tota1lipids. The proportion oflong-chain omega-3 fatty acids in
blubber was in the decreasing order of DHA > EPA > DPA. Presence of DPA, an
intermediary species between DHA and EPA, in relatively large amounts may have added
beneficial effects in human health; studies for better understanding of its role in human
diet is on-going in our laboratory. It is of interest to note that DPA is also found in relative
abundance in human milk [4].
The omega-3 fatty acid esters prepared by urea complexation of hydrolysed blubber
lipids had dominant quantities of EPA, DPA and DHA (see Table 2). However,
depending on process conditions, the enrichment of specific fatty acids varied significantly

TABLE 1
Proximate Composition of Seal Blubber.

Component Content, %
Moisture 2.3 ± 0.1
Protein (N x 6.25) 1.3 ± 0.1
Lipid 95.5 ± 0.7
Minerals 0.8 ± 0.1
 629
TABLE 2
Major Fatty Acids of Seal Blubber and its Concentrates.

Omega-3 Concentrates
Fatty Acid Raw Blubber
Preparation A Preparation B
14:0 5.0 0.7 0.8
16:0 9.5 - -
16:1 13.1 1.4 1.3
18:1 19.1 - -
20:1 11.9 - -
20:5 9.1 27.7 24.6
22:1 6.5 - -
22:5 5.0 9.3 4.8
22:6 10.0 38.9 46.1

(Table 2). Meanwhile the omega-3 enriched fraction retained most of the cholesterol
present in the original oil. Furthermore, oxidative stability of the original seal oil, as
determined by TOTOX values, was better than fish oils, perhaps due to the presence of
natural antioxidants in the former.

ACKNOWLEDGEMENT

We are grateful to the Newfoundland Department of Fisheries, Atlantic Canada

Opportunities Agency, Canadian Sealers Association and Canadian Centre for Fisheries
Innovation for facilitating or providing financial support.

REFERENCES

1. Gordon, D.T. and Ratiff, Y., The implications of omega-3 fatty acids in human
health. In Advances in Seafood Biochemistty: Composition arul Ouality. eds. G.J.
Flick, Jr. and R.E. Martin, Technomic Publishing Company, Inc., Lancaster and
Basel, 1992.
2. Haagsma, N., van Gent, C.M., Luten, J.B., de Jong, R.W. and van Doom, E.,
Preparation of an ",3 fatty acid concentrate from cod liver oil. L. Amer. .Qil Chem.
~, 1982,59, 117-118.

3. Shahidi, F. and Synowiecki, J., Cholesterol and lipid fatty acid composition of
processed seal meat. ~ W1.. food S.d... Technol. L., 1991, 24, 269-272.

4. Koletzko, B., Thiel, I. and ~~Y~tr~" Lipids in human milk: a model for infant
formulae? European L.' , 1992, 46, 545-555.
 SEPARATION OF EPA AND DBA FROM FISH OIL USING SUPERCRITICAL
EXTRACTION WITH Ag COMPLEX PRETREATMENT

KUNIO NAGAHAMA, TATSURU SUZUKI, SATOSHI KIKUCHI, YOSHIHIRA TANAKA,

KAORU NAKANO, HIDETAKA NORITOMI and SATORU KATO
Dept. of Ind. Chern., Tokyo Metropolitan University,
1-1, Minamiohsawa, Hachioji, Tokyo 192-03, JAPAN

ABSTRACT

Supercritical fluid extraction (SFE) of polyunsaturated fatty acid (PUFA) ethyl esters coupled with
pretreatment of Ag+ - :n:-complex formation was studied. PUFA esters such as eicosapentaenoic
acid (EPA) and docosahexaenoic acid (DHA) in fish oil were selectively dissolved into aqueous
AgN03 solution. After that semi batch SFE was carried out to recover PUFA esters from the
pretreated aqueous solution using supercritica1 (SC) CO 2, ethane and ethylene. All the PUFA esters
could be recovered by SFE, and extracted products were consist of PUFA esters and Ag free water
only. The results of SFE showed that the complex stability affected strongly on the SFE behavior.
Selective separation for EPA-Et and DHA-Et was obtained when SC-C02 and ethane were used,
and 90 wt.% of DHA-Et could be produced at the final SFE stage. SC-ethylene has little
selectivity for the PUFA esters, however, the extraction rates of PUFA esters were the fastest. This
is because ethylene makes :n:-complex with Ag+ ion competitively.

INTRODUCTION

Fish oils have attracted wide commercial and academic interests as a rich source of PUFAs,
particularly EPA and DHA, which are reported to possess beneficial physiological activities.
However, the isolation of PUFAs from fish oil is very difficult because it is a complicated mixture
of saturated and unsaturated fatty acids.
To develop a efficient method for the isolation of PUFAs, SFE coupled with pretreatment of Ag+
- complex was studied. PUFA esters originating from sardine oil were preconcentrated in an
aqueous AgN03 solution and it was extracted using supercritical CO 2 [1], ethane and ethylene. The
feasibility of SFE to isolate EPA-Et and DHA-Et form aqueous Ag+ solution was discussed.

EXPERIMENTAL SECTION

Experimental apparatus and procedures have been described in detail elsewhere [1]. Briefly sardine
oil ethyl ester (10 mL) was contacted with 5 M of aqueous AgN03 solution (20 mL) for 12 hours
 631
TABLE 1 Composition of fatty acid ethyl esters [wt.%]

Saturated C18:4 EPA DHA Unsaturated

fatty acids (C20:5) (C22:6) fatty acids
Sardine oil (feed) 31.2 2.4 19.5 8.4 34.2
Aq. AgN03 phase" (5M) 0.0 4.7 55.8 25.8 11.1
Residue in oil phase 46.6 1.4 3.0 0.3 39.6
* water and AgN03 free basis

at 298 K. Then the aqueous phase (5 mL) was separated and used for feed solution as SFE feed
stock. Only PUFA esters having more than 4 double bonds in the oil are soluble in aqueous
AgN03 solution and almost all PUFA esters dissolve in 5 M the solution as listed in Table 1.
Semi batch SFE was carried out with a flow-type system. Extracted samples were fractionated
every 5 - 120 minute intervals and analyzed both composition and weight. We investigated the
presence of Ag in the extracted samples by an atomic flame emission technique, and Ag was not
detected (less than 0.5 mglL). This means that PUFA ester in the aqueous AgN03 solution was
extracted not as Ag complex form but as free PUFA ester.

RESULTS AND DISCUSSION

A typical extraction curve is presented in Fig. 1. Efficient extraction of the PUFA esters occurs
at the early stage of the extraction, thereafter the rate of extraction decreases markedly, while water
was extracted at a constant rate throughout the extraction. Similar curves were obtained for all
experiments. A wide range of operating temperatures and pressures were examined for the CO2
SFE, and we obtained as the results that extraction pressure did not have significant effect on the
extraction rate of the PUFA esters, whereas higher temperature resulted in higher extraction rate
[1]. The results showed that the complex stability affected strongly in the SFE behavior.
The extraction yield curves of the PUFA esters for three supercritical fluid, CO2, ethane and
ethylene under the same pressure (24.5 MPa) and temperature (313 K) were plotted in Fig. 2,
1.2 ...--...----.---r---.----r---.
o Total 100 o 0 1:1. 1:1.
1.0 1:1. Water DO /}./}.I:I.

:Qi o PUFA esters

0
DO 80
ti 0.8 0
{!! 0
x
w 0 ~
L
60
0.6 0 :g
15
E 0 o~oe Q)
;;:: 40
~
0.4 o 0 0 0 0 0 0 1:1.
~ 1:1.1:1. 1:1. C02
« 1:1. 20 o ethane
0.2
• ethylene
0.0 O~-~-~-----~-~-~
0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3
Amount of C02 Used [Nm 3) Quantity of ScF used [Nffi3)
Fig. 1 C02 extraction curve at 313K, 24.5MPa Fig. 2 Extraction yield curve of PUFA esters
at 313 K and 24.5 MPa
632
100
AC18:4
o EPA ••• AC18:4
o
~ 80 •• 80 EPA

•
.DHA .DHA
cf!.
o other esters 'if: o other esters
!.c 60 j 60
c o 0 0 0
.Q
~ o o· • .Q
~
ECD 40 .00 E 40

••••
CD

20
(.) (.)
C
••
20J6 o
o 0 C
o
o o
00
0 o 0
o A ~ g 0....0...0 0 0 O~~~~~~~-=~~~~~
A t:.
0.0 0.1 0.2 0.3 0.00 0.05 0.10 0.15
Amount of C02 Used [Nm 3] Quantity of ethylene used [Nm3]
Fig. 3 Concentration changes of PUFA esters Fig. 4 Concentration changes for ethylene
for C02 extraction at 313 K and 24.5 MPa extraction at 313 K and 24.5 MPa

where yield was defined as the weight of total extracted PUFA esters divided by that of the esters
in feed aqueous solution. The extraction rate of PUFA esters was the order of CO2 < ethane <
ethylene.
The concentration changes of extracted PUFA esters for a SC-C02 extraction are shown in Fig.
3, where the concentration was calculated on water and AgN03 free basis. Selective separation for
EPA and DHA esters was obtained when SC-C02 and SC-ethane were used, and more than 90
wt.% of DHA ester could be produced at the final SFE stage. On the other hand, SC-ethylene
has little selectivity for the PUFA esters as shown in Fig. 4. It is suggested that free PUFA esters
in the aqueous phase are increased because ethylene makes :n-complex with Ag+ ion competitively.

CONCLUSIONS

SFE of PUFA ethyl esters coupled with pretreatment of aqueous AgN03 solution was studied.
PUFA esters in fish oil were selectively dissolved into Ag+ aqueous solution. After that semi batch
SFE was carried out to recover PUFA esters from pretreated aqueous solution using SC-C02 ,
ethane and ethylene. All the PUFA esters in the aqueous solution could be recovered by SFE, and
extracted products were consist of PUFA esters and Ag free water only. The results of SFE
showed that the Ag+ -complex stability affected strongly on the SFE behavior. Selective separation
for EPA and DHA esters was obtained when SC-C02 and ethane were used, and more than 90
wt. % of DHA ester could be produced at the final SFE stage. SC-ethylene has little selectivity
for PUFAs, however, the extraction rates of PUFA esters were the fastest. This is because ethylene
makes :n-complex with Ag+ ion competitively.

REFERENCE

1. Suzuki T., Kikuchi S., Nakano K., Kato S., Nagahama K., Supercritical fluid extraction of
polyunsaturated fatty acid ethyl esters from aqueous silver nitrate solution, Bioseparation,
1993, in press
 SELECTIVE EXTRACTION OF PROTEINS FROM COMPLEX SOLUTIONS
BY REVERSE MICELLES

Kazumitsu NAOE 1 , Masanao IMAI, Masaru SHIMIZU

Department of Chemical Engineering,
Division of Chemical & Biological Science,
Tokyo University of Agriculture & Technology,
24-16 2-Chome Nakamachi Koganei-C. Tokyo, 184 JAPAN

ABSTRACT

Desirable operating conditions for selective extraction of lysozyme from protein mixture were found to
be around minimal Am concentration for dilute feed system (<10-4Mfor lysozyme). Efficient selective
extractions of lysozyme from protein mixtures were successfully carried out by reverse micelles.

INTRODUCTION

Reverse micellar extraction process is expected as a novel protein separation technique for its
ability of large-scale, continuous processing. In practical protein separation, the target protein
is separated from multicomponent system. The labile nature of protein must be considered in
process design and development of protein separation technique. Therefore, recovery
conditions to get high percent extraction and activity in complex system are important for its
industrial applications. In this study, the efficient operating conditions for selective extraction
of protein from protein mixture were investigated.

MATERIALS AND METHODS

In this study, AOT(di-2-ethylhexyl sodium sulfosuccinate}/iso-octane reverse micellar system

was used. Lysozyme, cytochrome c, and ribonuclease A were used as a model protein. The
procedures of extraction experiment and enzymatic activity measurement were described in
previous paper[l]. The protein concentration was determined by UV absorbance. The water
content in the organic phase was measured by Karl-Fischer titration. Analytical separation and
quantification of protein mixture in the aqueous solution was made by HPLC. All experiments

1 Present address: Department of Chemical Engineering, Nara National College of Technology,

Yata, Yamato-Koriyama, Nara, 639-11 JAPAN
634
were carried out at 298K(25°q.

RESULTS AND DISCUSSION

Single Protein Component in Reverse Micellar Phase

Suitable AOT concentration in the organic phase to achieve high recovery amount and
activity of protein: The forward and backward extraction of lysozyme were examined at the
various AOT concentrations. Minimal AOT concentration[2] of lysozyme was found for each
lysozyme feed aqueous concentration.
The relative specific activity (RSA) of lysozyme was measured at 5x IO-4M lysozyme in the
initial aqueous phase. The RSA is the ratio of specific activity of recovered lysozyme to that of
feed lysozyme[l]. In the higher AOT concentration range (>ca.l OOmM), the RSA of lysozyme
decreased. The extraction at lower AOT concentrations is desirable to obtain higher RSA. The
recovered lysozyme activity was preserved at the minimal AOT concentration.
The recovery of activity( = the product of RSA and percent total extraction) took a maximum
value in this experimental range of AOT concentration. Authors propose to call this AOT
concentration "the optimal AOT concentration". The optimal AOT concentration in this system
was around the minimal AOT concentration.
Suitable feed protein concentration to get high recovery of activity: The protein
concentration of feed solution is a significant factor to get a high recovery of activity in
practical extraction, as the concentration of target protein determines the minimal AOT
concentration accordingly. The backward extractions of lysozyme were carried out under the
minimal AOT concentrations. In less than l.OxlO-4M oflysozyme concentration in the reverse
micellar organic phase, the percent backward extraction increased and consequently the
recovery of activity apparently increased up to almost 100%. The recovery of activity was
dominated mainly by the percent backward extraction in this system. As a result, the reverse
micellar extraction processes can be effectively applied to a dilute feed system.

Binary Protein Component in Reverse Micellar Phase

Minimal AOT concentration in binary protein component system: The effect of AOT
concentration on the simultaneous forward extraction of cytochrome c and lysozyme into the
reverse micellar phase was investigated. There was a minimum of initial AOT concentration for
complete forward extraction of the protein mixture. As to the solubilization into reverse micellar
phase, the mixture system behaved like single component system. Hence, this AOT
concentration is recognized as a minimal AOT concentration for the protein mixture of
lysozyme and cytochrome c.
Effect of protein composition in micellar phase: The recovery of lysozyme from the reverse
micellar phase containing binary protein mixture (lysozyme & cytochrome c) was performed
under the minimal AOT concentrations varying the protein composition of mixture. The percent
total extraction of lysozyme was dependent only on the lysozyme concentration in the micellar
phase, not on the total protein concentration.
 635
Selective Extraction of Lysozyme from Protein Mixture
On the basis of findings in above sections, the extraction of lysozyme from protein mixtures
(mixture 1: lysozyme & ribonuclease A, and mixture 2: lysozyme & cytochrome c) was
performed under the condition of minimal AOT concentrations. In the both systems, the
complete recovery of lysozyme was obtained(Figure 1) and the activity of the recovered
lysozyme was well preserved.

(.) (.)

j j
0
9
>.
0
a
0

~ ~
0
..."
~
..."
.0 .0

feed solution recovered aqueous phase 1 recovered aqueous phase 2

Figure 1. HPLC analysis of aqueous solutions obtained in the separation of a binary protein
mixture (lysozyme & cytochrome c) under minimal AOT condition.

CONCLUSIONS

The desirable operating conditions for selective extraction of proteins using reverse micelles
were investigated. Around minimal AOT concentration and dilute feed system (target protein
cone. <104 M for lysozyme) were efficient conditions for high recovery of activity. On the
basis of these findings, the efficient selective extractions of lysozyme from protein mixtures
were successfully carried out and the activity of lysozyme was well preserved.

REFERENCES

1. Naoe, K., M. Imai, and M. Shimizu, In: "Biochemical Engineering for 2001 ", S. Furusaki,
LEndo, R.Matsuno(Eds.), Pp.584-586, Springer-Verlag(1992)
2. Ichikawa, S., M. Imai, and M. Shimizu, Biotechnol. & Bioeng., 39, 20-26(1992)
FRACTIONATION OF ~ -LACTOGLOBULIN IN BOVINE WHEY BY INTERMITTENT
UP FLOW SYSTEM WITH MULTI-STAGE ION EXCHANGE COLUMN

HIDEO OHTOMO AND TAMOTSU KUWATA

Meiji Milk Products Co., Ltd., Central Research Institute,
1-21-3, Sakae-cho, Higashimurayama-shi, Tokyo 189, Japan
ICHIRO KURIHARA, TAKASHI KIKUCHI, AND SABURO FURUSHO
Nippon Rensui Co., Ltd., Laboratory, 1000, Kamoshida, Midori-ku,
Yokohama-shi, Kanagawa 227, Japan

ABSTRACT
We have screened some separating gels for fractionation of bovine
j3-lactoglobulin (j3-LG) in whey, and found that the most suitable
ones are cellulose cation exchangers. Although they are easily
compacted in the conventional single column (eSC) by scaling up,
some problems led from gel compaction could be resolved with "Intermittent
up flow system with multi-stage column (IUFS)".

INTRODUCTION
j3-LG can be fractionated by cellulose cation exchanger (1). However,
compaction of the gel bed is often caused when the CSC packed
wi th such a soft gel is scaled up. Consequently, decrease in the
flow rate, large pressure drop, and channel ing are observed. To
resolve these problems, we have developed IUFS.

MATERIALS AND METHODS

Electrodialyzed (ED) Gouda cheese whey or reconstituted ED whey
was used as starting materials. In the laboratory trials, 5- or
la-stage acrylic column (50mm cjJ x 200mm) was used. For the pilot
scale trials, 5-stage polyvinyl chloride column (400mm cjJ x 200mm)
packed wi th 50 R, of separating gel was installed. Sample was fed
into the multi-stage column with down flow, and the aliquot of
non-adsorbed fraction, i.e. whey passed through the column was
stored in a tank. To loose the pressing gel bed, stored whey was
fed into the column with up flow at intervals just before the gel
bed was compacted. j3-LG adsorbed on the gel was eluted with NaCI
and/or NaOH. The eluent was fed into the column with circulative
 637
down fl ow.

RESULTS
Separating gel
Among several separating gels tested, Indion S3 (Life Technology
Inc.) was one of the most suitable gels. Pressure drop of the CSC
packed with Indion S3 was lower than that with CM-Cellulofine CH
(Chisso Co., Ltd.) which has larger adsorption capaci ty of /3 -LG
than Indion S3.
Laboratory scale IUFS
Several testing condi tions, ending points of chromatographies, and
the pressure drop of the column were shown in Table 1. Inner pressure
of IUFS was lower and more stable than that of CSC.
Pilot scale IUFS
Al though preferential /3 -LG adsorption on Indion S3 was not observed
by using the pilot scale CSC, about 90% of /3 -LG was trapped by
IUFS without the compaction of the gel bed. The amount of adsorbed
/3 -LG was equal to that of laboratory trials (78g1 f2 -gel). The eluate
was concentrated and demineralized by ultrafiltration and diafiltration.
Freeze dried /3-LG isolate showed excellent solubility and gelation
properties. /3-LG accounted for 84% of total protein of the isolate
(Table 2).

TABLE 1
Testing conditions, ending points of chromatographies, and the pressure
drop of laboratory scale IUFS
Test 1 Test 2 Test 3 Test 4 CSC·4
~.~~.~~X... .9.f. ... ~.t~K~ ............... ~.9.................. §.................. ~ ..................~ ..................~................. .
Flow rate(m/h)
Down 0.55 0.55 0.61 0.49 0.49
....... JJ.p. ......................................Q.~.§.~ .......... Q.·.. ~.!? ..........Q:..9.J ................-:.....................-:............... .
Feed time(min)
Down 56 56 54 cont" cont
....... JJ.p. .......................................... 1. .................1..................9...................-:.....................-:............... .
Average space 1. 19 1. 19 1. 22 1. 22 1. 22
Y.~1.9.~.t.j;y..(.!.!.Q.)......................................................................................................................... .
~.~g.~.!!.g .. P.Q.~.!!.t.'O~.Y.>.~~........ ~.(l.: ..? ..........~1.~.Q...........!?~:.. ~ .......... ~.~:..O'~: ........... ~.Q.: ..Q............ .
Pressure drop 0.03 0.06 0.09 0.1 0.10
(kgf/cm2 ) --0.07 --0.09 --0.13 --2.0< --0.30
*1:Sample feeding was stopped when {3 -LG concentration in non-
adsorbed fraction reached to O. 4g/ f2 ({3 -LG concentration in
starting materials was 2. 7g/ f2). (Volume of non-adsorbed fraction)
I(gel volume) was shown as "BV" unit.
*2:Continuous feed with down flow.
*3:Pressure drop exceeded 2kgf/cm2 at 25BV unit.
*4:Conventional single column.
638
DISCUSSION
To elute .B -LG in short time, it was effective to use higher concentration
of NaCI and/or NaOH as the eluent. However, resul tant .B -LG was
partially denaturated, therefore the functional properties were
deteriorated. On the other hand, the lower concentrations of eluent
elongated the elution time. Thus, to prevent protein denaturation
and to shorten elution time, we used 2 step elution. To give buffering
ability to the eluate, small amount of NaCI was fed into the column
at first to elute .B -LG to some degree. At next step, NaOH was
added gradually to raise the pH of circulating eluate to 9 and
to keep it constant. It is very important to consider the interval
of up flow and the concentration of the eluent at the first elution
step to perform .B -LG separation more efficiently.
CONCLUSIONS
When the experiments with pilot scale CSC packed with a soft gel
were carried out, shrinkage and/or cracking of gel bed were observed.
On the other hand, in the case of IUFS, these phenomena did not
occur. Although such chromatographies as a size exclusion mode do
not perform by disorder of gel bed in general, intermittent up flow
did not affect .B -LG separation by ion exchange mode.

TABLE 2
Protein compositions of ,8-LG isolates obtained by pilot scale IUFS
Ig'3 BSA04 a-LA's
Test 1 01 5.9 6.2 3.7 84.3
Test 2'2 5. 8 5. 6 4. 2 84. 5
*1:down flow;SV=1.48, 108min, up flow;SV=1.66, 6min
*2:down flow;SV=1. 98, up flow;SV=1. 66, 6min(Intermittent up flow
was induced when the inner pressure of the column had reached
to O.6kgf/cm2.)
*3: Immunoglobulins, *4:Bovine Serum Albumin
*5: a -Lactalbumin, *6:.B -Lactoglobulin
ACKNOWLEDGEMENT
This study was made by the great help of The Japanese Research
and Development Association for The High Separation System in Food
Industry.
REFERENCE
1. Ohtomo, H., Hamamatsu, K., Hori, E., and Kuwata, T., Studies on
the EI imination of .B -lactoglobulin from Whey Using Carboxymethyl
Cellulose Cation Exchanger. ~ ~ Soc. Food Sci. Technol.,
1988, 35 (11), 755-762.
 RECOVERY OF LYSOZYME AND AVIDIN FROM EGG WHITE BY
ION-EXCHANGE CHROMATOGRAPHY

SHUICHI YAMAMOTO, TOMOYUKI SUEHISA AND YUJI SANO

Department of Chemical Engineering,
Yamaguchi University, Tokiwadai, Ube 755, JAPAN

ABSTRACT
A method for determining the mobile phase composition in stepwise-elution chromatogra-
phy is presented and tested for cation-exchange chromatosraphic separation of lysozyme
and avidin from egg white. he distribution coefficient (K) as a function of the salt con-
centration (1) was determined from linear salt gradient elution experiments at a fixed pH.
Based on the K - I relationships, two purificatIOn schemes were designed and successfully
carried out.

INTRODUCTION
Lysozyme and avidin are two basic proteins contained in egg white. They are important
for therapeutic and diagnostic clinical applications. The process for recovering lysozyme
and avidin from egg white must be such that the properties of the recovered egg white
are comparable to the egg white untreated[1,2].
In this paper, a method for determining the mobile phase composition in stepwise-
elution chromatography is presented and tested for cation-exchange chromatographic sep-
aration of the above-mentioned separation system.

RESULTS AND DISCUSSION

We have shown[3-7] that a large amount of important information can be extracted from
linear salt gradient elution experiments. The distribution coefficient (K) as a function
of the salt concentration (1) can be obtained from the peak salt concentration (1R) -
the normalized gradient slope (GH) relationships (Figs.l and 2). Stepwise elution chro-
matography can be easily designed on the basis of the K - I data [3,4,6,7]. Based on
the K - I data for ovalbumin, lysozyme and avidin at various pH values, two purification
schemes were designed and carried out.
In the first scheme, the I of the elution buffer (1E) of the fixed pH was changed (
Table 1). In the second scheme, both the IE and the pH were varied(Table 2). The
productivity(recovered amount per column volume per process time) of the second scheme
is much higher than that of the first scheme.
The performance of liquid chromatographic separation of proteins is usually gov-
erned by stationary (gel) phase diffusion, and the operating conditions such as flow-rate
or residence time are adjusted so that a desired resolution can be achieved. However,
640
the present study has shown that the productivity of chromatographic separations can
be drastically increased by choosing suitable elutIon buffer conaitions even when such
parameters as flow-rate, column dimension and sample loading are not optimized.
Note: The details of the experimental procedure, the apparatus, the assay methods and the
materials employed in this study are given in [3-6). The purification factor( P F) is defined as
the activity of the recovered fraction per protein content divided by that of the sample. The
recovery ratio (Q R) is the ratio of the total activity recovered to that applied to the column.

TABLE 1 Purification results at a fixed pH=8 a)

Operation V time V IVi 0.1 pH 8.0 I I
sample application bj
[ mL) [min) [ - )
5.0 38 4.0 6
I ( ~
60
washing c)
elution for avidin d )
5.0
1.4
38
11
4.0
1.1 ~
60
I/
elution for lysozyme e ) 6.1 47 4.8
total 18 134 13.9 :::c
I.!' ovalbumin avidin IllYSOIyme
a)V =elution volume, Vi=column volume 60
Column(0.9 em diameter, 2 em height), I J
flow-rate=0.13 mL/min 0.01
b )Egg white was diluted to one-second of the 0.05 0.2
original concentration with buffer A (10mM [M]
phosphate pH8) and used as a sample.
Fig.1 Relationships between CH and IR.
c)buffer A containing 0.03M NaCI
CH =slope of the salt gradient(M/mL) divided
d )buffer A containing 0.22M N aCI by the column gel volume, IR= salt concentra-
QR=79%, PF=26 tion at the peak retention time. These curves
e)buffer A containing 0.50M NaCI are not dependent on the flow-rate, the col-
QR=100%, PF=23 umn size and the particle diameter[6,7).
5 r---~---..,
TABLE 2 Purification results a )
Operation V time V IVt
[ mL) [min) [ - )
sample application bj 3.8 14 3.0
washingc) 3.2 12 2.5
elution for avidin d) 3.2 11 2.5 1
elution for lysozyme e ) 5.9 22 4.6
total 16 60 12.6
0.5
a)Column(0.9 em diameter, 2 em height),
flow-rate=0.27 mL/min avidin pH 8.0
b )Egg white was used as a sample.
c )buffer A containing 0.03M NaCI 0.1 I [M] 0.5
d)buffer B(50 mM carbonate buffer,pH 10)
containi~g 0.05M N aCI Fig.2 Relationships between J( - J(' and I.
QR=89%,P F=87 The curves were determined from the CH -
e)buffer B containing 0.50M NaCI IR curves in Fig.l according to the relation
QR=92%,PF=21 dCH/dI = 1/[J((I) - J('] where J(' is the dis-
tribution coefficient of the salt (gradient su b-
stance or mobile phase modulator)[3,4,6,7].
REFERENCES
1. Chan,E.Li, Nakai,S., Sim,J., Bragg, D.B. and Lo, K.V., J.Food Sci., 1986,51,1032.
2. Ahvenainen,R., Heikonen, M., Kreula, M., Linko, M. and Linko, P., Food Process Engng.,
1980, 2,301.
3. Yamamoto,S., Nomura,M. and Sano,Y.,J.Chromatogr., 1990,512,89.
4. Yamamoto,S., Nomura,M. and Sano,Y., Chem. Eng. Sci., 1992,47,185.
5. Yamamoto,S., Suehisa,T. and Sano,Y., Chem.Eng.Commun., 1993, 119,22l.
6. Yamamoto,S., Nomura,M. and Sano,Y., AIChE J, 1987,33,1426.
7. Yamamoto, S., Nakanishi, K. and Matsuno,R., Ion Exchange Chromatography of Proteins,
Marcel Dekker, New York, 1988.
 DEVELOPMENT AND APPLICATION OF CYCLODEXTRIN POLYMER

HISASHI OKEMOTO, HITOSHI HASHIMOTO

Ensuiko Sugar Refining Co., Ltd.
13-46, Daikoku-cho, Tsurumi-ku, Yokohama, Japan

ABSTRACT

The objective of this study is to establish a technology for

purifying substances which are difficult to separate in food
processing by utilizing the inclusion and/or catalytic
function of cyclodextrin (CD). We examined methods to
synthesize CD polymers by fixing CD to four kinds of organic
matrixes; acrylic polymer, agarose gel, vinyl polymer and
chitosan beads. We applied these CD polymers to separating or
purifing various substances from foods. Chitosan CD polymer
was the most effective in our polymers.

INTRODUCTION

CDs are cyclic oligosaccharides composed of six, seven or

eight glucopyranose units and called a-, 8- and y-CD,
respectively. The internal cavity of the structure of CD is
hydrophobic whereas the external surface is hydrophilic. The
most characteristic or unique function of CDs is the formation
of inclusion complexes by taking various substances into its
hydrophobic cavity. This function can be reffered as one of
lock and key recognition ability. This CD's ability can be
used for separation of optical and structural isomers or
substances which have resembling molecular structure or
similar molecular weight.
Up to now, a considerable number of reports have been
published on the manufacturing methods and properties of CD
polymers, but due to high costs or to problems in physical
strength they have not been actually utilized for industrial
purposes. Accordingly, investigations have been done on
industrial methods of manufacturing novel CD polymers to be
used as separating and isolating materials in food industry.
The present studies were conducted in collaboration with
Japan Organo Co., Ltd.
M2
RESULTS AND DISCUSSION

Firstly, it was found that the fixation of CDs to acrylic

macro-porous polymer, highly hydrophilic agarose gel and vinyl
polymer by using our technique could not increase the amount
of CDs to be fixed and the physical structures of products
were weak.
Then chitosan beads were chosen as a new carrier for the
fixation to be tested. Chitosan is a natural polysaccharide
like agarose, has a relatively high physical strength and has
been utilized as a carrier for immobilized enzymes. Suitable
conditions of synthesizing CD polymer were studied by using
chitosan beads.
The fixation of CD to chitosan beads was performed by
using a spacer. Polyethylene glycol-di-glycidyl ether was
employed as the spacer. After investigating for adequate
reaction time for the introduction of spacer and for adequate
spacer length, a maximum amount of fixed CD, about 110 ~mol/g
dried carrier, was obtained when a spacer of short chain
length was used in a reaction for 5-6 hours.
An attempt to increase the amount of introduced spacer
revealed that increase of introduced spacer did not
necessarily increase the amount of fixed CD, and about 110
~mol/g dried carrier of fixed CD was maximum amount that could
be obtained. This is thought to be due to that the structure
of the carrier inhibits fixation of CD or that plural spacers
combine with a molecule of CD, with the result of low
efficiency of spacer utilization.
Then, for the purpose of examining the standard
performance of chitosan CD polymer, inclusion tests were
conducted by employing nitrophenol as a standard substance.
The results showed that estimation of the amount of adsorbed
nitrophenol in 0.01 M borate buffer (pH 9.18) at 25°C for 30
minutes of reaction time made evaluation of the standard
inclusion capacity of the CD polymer possible. For the
adsorption characteristics of 0-, m- and p-nitrophenol to
chitosan S-CD polymer, there was found a considerable
difference between 0- and m-nitrophenol and p-nitrophenol, and
as the standard substance p-nitrophenol was found to be most
suitable. Existence of the large difference in the adsorbed
amount among isomers suggests that it is possible to isolate
p-isomer from a solution of mixed isomers by means of
adsorption and elution or chromatographic separation.
Then separation of nitrophenol isomers by using a column
of chitosan S-CD polymer was conducted. Application of a mixed
solution of 0-, m- and p-nitrophenol to a column packed with
chitosan S-CD polymer allowed 0- and m-nitrophenol to be
eluted in the earlier half of eluate while p-nitrophenol to be
eluted in the later half; in this way p-nitrophenol could be
isolated by being separated from other isomers. In comparison,
a similar experiment was conducted by using a column packed
with the chitosan beads, the starting material of making the
CD-polymer, under the same conditions to find that all isomers
were not separated at all, being eluted at an identical
position.
 643
In this way utilization of CD polymer in chromatographic
separation was shown possible, and so its application to the
field of food industry was studied.
CD polymer has adsorbing ability of naringin. This is
thought to be due to that benzenediol that forms the terminal
part of the naringin molecule is included by fixed CD. Then
separation of naringin and hesperidin, which have benzenediol-
like structures in common and have close molecular weights,
by column chromatography was studied. Hesperidin is a compound
contained in fruit JUlces and is responsible to JUlce
turbidity like naringin. Chitosan S-CD polymer was packed into
a column and a mixed solution of naringin and hesperidin was
applied onto it to obtain a distinct separation pattern. For
the sake of comparison ,similar chromatographic experiments
were done with a column of chitosan beads and a column of
spacer-introduced chitosan beads to find no separation of
naringin and hesperidin from each other at all; both were
eluted at near void volume. From the molecular structures,
naringin and hesperidin are thought only to be lightly
included by chitosan S-CD polymer, but the slight difference
in the structures is recognized by the polymer and
subsequently they were separated. The fact should provide a
preferable condition to chromatographic separation of
substances.

CONCLUSIONS

Chitosan beads is considered to be excellent material for

making CD polymer in reference to handling easiness and cost.
Chitosan beads is more profitable since the material does not
show hydrophobic adsorption. However, the chitosan CD polymer
available at the present time allows undesirably acids to be
adsorbed by the base material of chitosan and shows somewhat
unsatisfactorily weak physical and chemical strength, and so
more improvements are required in future.
At the start of this development the target was isolation
and purification by adsorption and elution, and so the main
theme of research was to find how to increase the amount of
fixed CD. Unfortunately marked increase of the fixed amount
was not achieved but it was made clear that CD polymer could
be used as a separating material in chromatography. When used
as a separating material in chromatography, its application to
continuous processing in the simulated moving bed method may
be possible, and extensive utilization of CD polymer may be
expected.
More studies for improving strength, durability and
manufacturing cost are needed, but the basic technique of
isolation and fractionation by using CD polymer as
a separating material in chromatography has now been
established. More and more contribution of CD polymer not only
in the field of food industry but also in other fields is
expected.
 TURBULENT HIGH·SCHMIDT NUMBER MASS TRANSFER IN UF

CHRISTER ROSEN, CHRISTIAN TRAGARDH and PETR DEJMEK

Department of Food Engineering,Lund University
P.O.Box 124, S-221 00 Lund, Sweden

ABSTRACT

A computer simulation model have been developed predicting the mass transfer in terms of
turbulent diffusion for high molecular weight substances. The method involves a numerical
solution of the time averaged equations of motion with a spatial resolution down in the very
thin concentration boundary layer appearing in turbulent wall boundary layers.
For the prediction of required Reynolds stresses a so called low Reynolds number k-€
turbulence model has been used. The turbulent diffusion of mass is obtained from the
Reynolds' stresses through a turbulent Schmidt number relationship.

INTRODUCTION

The most common method of predicting mass transfer across the concentration polarization
boundary layer is to use the film theory equation. A basic assumption in the derivation of
the film theory equation is that physical properties are constant in the film. The
macromolecules filtered in UF have physical properties that are highly concentration
dependent, and thus the assumption of constant physical properties is not valid. In the method
presented here, the general conservation equations for species concentration and momentum
are solved throughout the boundary layer with the aid of a computer program based on a
finite volume formulation. This formulation has the important quality that
concentration-dependent physical properties can be calculated throughout the concentration
boundary layer at spatially resolved points.

THEORIES AND METHODS

Governing equations
A statistical approach is taken for the instantaneous velocity,U , and pressure P, separated
into a mean and a fluctuating part; U = U + u' and P = P + p'. In analogy with this is the
 645
instantaneous concentration described as; C == C + c'. The following set of equations, in the
Cartesian coordinate system,iDr a steady, tw.o-dimensillnal boundary layer flows;
ljau +Vau =-1. ap +~(v aU)_~(ulv)
ax ay pax ay ay ay

The averaging process introduces the unknown correlations u'v' and v'c'.

Turbulence models
In Rosen [1] different low Reynolds number k-e turbulence models were tested in respect to
their ability to predict turbulent momentum transport in the viscous sublayer.As a result the
model by Chien [2] was chosen. For predicting u'v', the eddy viscosity concept is used. The
k-e model employs two transport equations, one for the turbulent kinetic energy k and one
for the turbulent kinetic energy dissipation rate e which relate to the eddy viscosity as;

which in turn relates to the u'v'and c'v' correlation as;

The turbulent momentum transport is related to the turbulent transport of species

concentration by a turbulent Schmidt number expression, derived by Rosen [1];

_ Vt _1.153(1-exp(-VCRe-3000)/2000))
a=-------~--~~~----~--~~
t €t 0-0 .112

RESULTS

The simulated results of the membrane concentration, are compared with results calculated
from the film theory equation for two Sherwood correlations in Figure 1; the Deissler [3]
analogy Sh == 0.122 (f/2)0.5 Re 0'0.25 and the correlation of Shaw and Hanratty [4] Sh == 0.0889
(f/2)0.5 Re 0'0.296. The latter model was chosen as the derived turbulent Schmidt number
expression is obtained from their experimental results. A good agreement is found for low
permeate fluxes while going to higher a continuously larger deviation appears. In Figure 2
the spatial resolved near-membrane gradients are shown and in Figure 3 wall concentration
variations over a membrane having a Normal distributed varying permeability. The poly-
saccharide Dextran no at a concentration of 0.142% was the working fluid for which
physical data was taken. Simulated rectangular flow channel has a height of 5.9 rum a width
of 60 mm and a membrane length of 100 mm. Applied pressure was 0.6 MPa. 100 %
retention was assumed.
646

.....
__ _ ...(1\1611
DeIIIIor~ (1971)

• -0- SimuIaIiaa . . .

•
•
•

• • • • • I,OJGI(M)
Figure 1.
__ Membrane concentration versus flux (for corresponding
II-I Re-numbers).
ur-------------------,
--
.. --
Figure 2. Property gradients versus wall distance
,- r:(w~1
ur--------------------,
-
Velocity=1.84 mIs, flux=3.22*1Q-s m/s
.r---~----------------~

."
. -
,. -
" -
Figure 3. The variation of membrane co~~htration with a membrane flux distribution
(Jv=3.22±1. 12)* 10-srn/s.

REFERENCES
1. Rosen, C., Turbulent High-Schmidt-Number Mass Transfer in the Concentration
Polarization Boundary Layer over Ultrafiltration Membranes, Ph.D. thesis, Lund
University, 1992

2. Chien, K.- Y., Prediction of channel and boundary layer flows with a low-Reynolds-
number two-equation model of turbulence, AIAA Paper No 80-0134, 1980

3. Deissler, R., Analysis of turbulent heat transfer, mass transfer and friction in smooth
tubes at high Prandtl and Schmidt numbers. In Advances in Heat and Mass Transfer,
McGraw-Hill, New York, 1961

4. Shaw, D.A., Hanratty, TJ., Turbulent mass transfer rates to a wall for large Schmidt
numbers, AIChE Journal, 1977,23, pp.28-37
 The pore size of ultrafiltration membranes . a novel approach

KENNETH M PERSSON AND GUN TRAGARDH

Dept. of Food Engineering, Chemical Centre, Lund University
PO Box 124, S-221 00 LUND, SWEDEN

ABSTRACT

The pore size of flat sheet membranes of different compositions with nominal
molecular weight cut-off's between 20 kDa and 100 kDa were evaluated in a
cross-flow mode, using mono disperse incompressible silica sols with mean size
from 11 to 45 nm. The observed silica retention, from 97.6 to 98.7 % gave an
indication of the largest pore size of the membranes. The silica content was
low in the permeate at any condition, yet always above the monomeric
equilibrium concentration of about 130 ppm, even for colloidal sizes much
larger than the nominal pore size of the membranes.

INTRODUCTION

mtrafiltration (UF) is used for the separation and concentration of macro-

molecules in e.g. food industry. The characterization of the UF membranes
involves pore size distribution measurements, which can be measured directly
as pores, or indirectly as the retention of well characterized macromolecules of
known sizes (dextran, BSA etc.) [1]. Since the tertiary structure of these
molecules can change, especially when undergoing shear [2], the macromole-
cular permeation in practice is a function of size and actual shape. By using
incompressible silica particles of various sizes, this problem should be limited.
The nominal pore size of a UF membrane is either expressed as a molecular
weight cut-off, nmwco (in kDa) or as a pore diameter (in nm). All the direct
porosimetric methods assume some physical structure of the membrane,
mostly some specific geometric shape of the pores. An advantage with indirect
methods is that they do not preassume any given pore shape.
Silica has been used frequently as a model fouling agent, yet seldom for
retention measurements. Her [3] used UF-membranes with pore sizes from 0.9
nm to 2.0 nm for the fractionation of concentrated silica sols with mean size 1-
2nm.
648
The colloids are in equilibrium with silica monomers in water solution.
About 130 ppm silica is solved as monomer in water, the rest are particles [4].
With Cp as the permeate concentration and Cb the bulk concentration of
silica, the observed retention Robs is defined:

R __ Cp
obs- 1
- (1)
Cb

EXPERIMENTAL

Three polymeric flat sheet (R)NADIR ultrafiltration membranes from Hoechst

AG, Germany were tested: polyamid UF-PA-20 (nmwco 20 kDa with PVP K 30
retention measurements); hydrophilic polysulfone UF-PS-100M (nmwco 100
kDa with dextran T 2000); cellulose acetate UF-CA-100 (nmwco 100 kDa with
dextran T 110).
Colloidal silica sols Bindzil® from EKA-Nobel (Sweden) in different sizes
were used as particle source, see table 1. The particle size distribution was
analyzed with a Malvern photon correlation spectrometer (PCS 100), giving
the hydrodynamic radius. Particles were also studied by SEM and compared
with a flow field fractionation measurement that had been performed
previously [5]. The different methods gave consistent sizes. The silica content
in permeate and retentate was analysed according to Iler [3]. The monomeric
silica was also analysed separately for some cases, without previous decom-
position of the particles.
An automatic UF laboratory plant constructed at the department was used
for the experiments [6]. With controlled temperature, flux, and pressure, and
with tangential velocity measurement devices and computerized permeate
levelling, the total accuracy in the flux measurement was about 5*10-7 mls.
Each experiment followed the same procedure at a circulation velocity of 11
mis, a mean transmembrane pressure of 0.14 MPa and a temperature of
18.5°C. A fresh membrane was mounted and conditionized with water for 30
minutes before permeation. The silica sol was added to a final concentration of
1 %, and permeate was collected for analyses after 1 minute, 10 minutes, 60
minutes and 240 minutes. The plant was thereafter rinsed 3 x 2 minutes with
water. Finally, the pure water flux was measured again for at least one hour.
All water used was MilliQ-filtered.

RESULTS AND DISCUSSION

The silica content in the permeate reached steady state for the particles used
within 30 minutes. Values after 240 minutes are found in table 1. At the high
cross-flow velocity used, the membranes were hardly fouled at all, indicating
no polarization; the membrane fluxes were stable even after hours, except for
a compaction effect with time. The analyse of the monomers gave a theoretical
retention of 98.7 % if maximum monomer concentration was assumed.
A fouled membrane would result in a higher retention, not lower. The lower
actual retention could be explained by the prescence of particles in the perme-
 649
ates. The particles might pass through the small number of large pores, which
have been shown for UF-membranes [7]. The highest silica retention was
found for the polyamid membrane, which could be explained by its lower
nmwco. It is not possible to distinguish between the CAI00 and the PSI00
with the particles used.

TABLE 1.
Particle size and observed retention after 240 minutes
Size data (in nm) Observed retention (in %)
Particle name Size±stand.dev. CAI00 PSI00 PA20
15/500 1l.4±4.5 97.6±0.1 97.7±0.2 97.8±0.1
30/360 18.9±5.6 98.1±0.2 98.0±0.1 98.1±0.1
30/220 21.3±6.0 98.4±0.1 98.3±0.1 98.4±0.1
40/130 26.1±4.5 98.3±0.1 98.4±0.1 98.6±0.2
50/80 45±>10 98.3±0.1 98.3±0.1 98.7±0.1

CONCLUSIONS

Silica particles can be used to investigate the presence of large pores in UF

membranes. Silica is retained to a high extent, yet traces of particles can be
detected in the permeate even for low-nmwco membranes. The results are best
explained by a postulation of a pore size distribution for the membranes. The
method would be improved by the use of smaller particles, about 4 nm in size
or less.
REFERENCES

1. G. Tragardh (1988). Survey of characterization methods for ultrafiltration

membranes, in Characterization of ultrafiltration membranes,
Studentlitteratur, Lund 9-38.
2. KM. Persson and V. Gekas (1993). Factors influencing aggregation of
macromolecules in solution, accepted in Process Biochemistry.
3. Ralph K Her (1982). Colloidal Components in Solutions of Sodium Silicate,
ACS Symposium Series 19496-113.
4. Ralph K Her (1979). The chemistry of silica, chapter 1, John Wiley & Sons,
New York.
5. B. Olsson, K.-G. Wahlund and A. Litzen (1992) Fast size characterization of
silica sols by asymmetrical flow field fractionation (FFF), poster presented at
3:rd into symp. on FFF, Salt Lake City, Utah, 5-7 October 1992.
6. G. Tragardh and K Olund (1986). A method for characterization of
ultrafiltration membranes, Desalination 58 187-198.
7. J. Nilsson (1989). A study of ultrafiltration membrane fouling, thesis, Lund
University, Sweden.
 REVERSE OSIOSIS SYSTEI FOR HIGHLY CONCENTRATED JUICE

HIROSHI NABETANI*. HARUYUKI IGAMI'*. IITSUTOSHI NAKAJIMA*

KIROU HAYAKA'A*'. YASUNORI YAMADA**. YUKIO ISHIGURO**
* National Food Research Institute. Ministry of Agriculture.
Forestry and Fisheries. Tsukuba. Ibaraki. 305 Japan
** Research Institute. Kagome Co.• Ltd.. Nasu. Tochigi. 329-27 Japan

ABSTRACT
To develop the multistage reverse osmosis CRO) system. which is regarded as
useful for highly concentrated fruit juice. the permeate flux value and the
permeate concentration of the system were analyzed in this study according to the
membrane transport equation and the concentration polarization equation. The
pressure drop along the flow channel in the membrane module was also quantified.
Based on these results. a two-stage RO system for clarified apple juice has been
proposed. This continuous system is composed of a tight RO membrane at the first
stage and a loose RO membrane at the second stage. and the membrane modules were
arranged in a tapered fashion. The energy efficiency of the system will be
compared with those of conventional methods.

INTRODUCTION
RO has advantages over other methods of removing water from liquid foods. Its
application in food industries has been developed successfully. However. RO
requires operating pressure higher than the osmotic pressure of the solution and
the resisting pressure of the module is 6-7 MPa. Therefore. RO use is limited to
low concentrations. However. if a loose RO membrane is combined with a
conventional tight RO membrane. higher concentrations of juice can be obtained.
The performance of both tight and loose RO membranes was analyzed in this study
according to the membrane transport equation and the concentration polarization
equation. Based on the results. a two-stage RO system of high energy efficiency
has been proposed.
THEORETICAL
Permeate flux. Jv. and observed rejection. Robs. are predicted by the membrane
transport equation based on nonequlibrium thermodynamics and the concentration
polarization equation. The mass transfer coefficient. k. of the turbulent flow
in the tubular module is calculated using Deissler s correlation.
 651
The membrane transport equation
lv=Lp {2 f..lw/ (f..lp+f..lm)} {.6.P-L (ai.6.II!)} (1)
R=ai (1-F 1 ) / (1-a!F 1 ) (2)
Fi=e xp {- (1-aJ lV/Pi} (3)
The concentration polarization equation
(Cm,i-Cp,i) / (Cb,i-Cp,l) =exp (Jv/k) (4)
Deissler s correlation
N Sh =0.023 N Re o,875 NscO.25 (5)
Real rejection
R!=1-C p,i/C m,i (6)
Observed rejection
Robs, i = 1- C p, i/C b, I (7)

IATERIALS AND mHODS

An RO unit equipped with a tubular module and a high pressure pump with
variable speed motor was used. The tight RO membrane (Type AFC-99) and the loose
RO membrane (Type AFC-30) were supplied by PCI Ltd. The inner diameter, length
and area of the membrane were 1.25 cm, 55.5 cm and 0.0218 m2 , respectively.
Clarified apple juice (weight ratio of monosaccharides(ms) to disaccharide(ds) =
5 : 1) was used as the sample solution.
Total circulation experiments were performed by returning the permeate to the
feed tank to prevent change in concentration of clarified apple juice. The
conditions of the tight RO membrane treatment and the loose RO membrane treatment
were as follows: feed concentration, Cb : 12-26 wt.~; applied pressure, 6P: 2-7
IPa; flow velocity, u: 1.0 and 0.5 m s -1; feed temperature: 25~. The permeate
fluxes and the observed rejections at steady state were measured under each
condition.
RESULTS AND DISCUSSION
1. RO ae.brane treatment of clarified apple juice
The pure water permeability was measured after each treatment. The reflection
coefficient and the solute permeability of loose RO membrane were determined by
the use of the curve fitting method and were as follows: a ms=O. 999, Pms =8.1x10- 7
m S-l, a dS=O. 999, Pds=l. 7x10- 7 m S-l.
The relationship between applied pressure and permeate flux for the tight RO
treatment of clarified apple juice was calculated by solving Eqs. (1) and (4)-(7)
(i = ms, ds). The experimental relationships between applied pressure, permeate
flux and observed rejection for the loose RO treatment of clarified apple juice
were calculated by solving Eqs. (1)-(7) (i = ms, ds). The experimental values of
the tight RO treatment and the loose RO treatment were in good agreement with the
calculated values.
2. Developaent of Two-Stage RO Systea for Highly Concentrated Juice
1) TIO-stage RO system
In order to concentrate clarified apple juice from 10 wt.' (ms 8.33 wt.', ds
1.67 wt.') to 40 wt.~, a two-stage RO system which is composed of a tight RO
membrane (Type AFC-99) at the first stage and a loose RO membrane (Type AFC-30)
at the second stage has been proposed. The first stage of this system has more
membrane units than the second stage, that is the tree type system. In addition,
the permeate of the second stage is returned to the feed tank of the first stage.
The conditions of this system are as follows: applied pressure of the first
and second stage: 7.85 IPa; mass feed rate of clarified apple juice: 1000 kg h- 1
per one membrane unit of the second stage.
652
2) lass balance. concentration of bulk solution and pressure drop
lass balance, concentration of bulk solution and pressure drop in the module
can be expressed by Eqs. (8)-(10).
lass balance
d u / d A = - J v/ S B. C. u=u0 at A = 0 (8)
Concentration of bulk solution
C b = (uoPoCo-f CpJvp p dA) / (up) (9)
Pressure drop
d P d / d A = A. u 2 P / (2 1( d h 2) B. C. Pd= 0 at A = 0 (10)
3) Optimization of the two-stage RO system for highly concentrated juice

:
The first stage area
was calculated by solv-
l00r-------~--------~------_.

1st and 2nd stage

ing Eqs. (1) and (4)-

~ -==-~~===.
(0), and the second
stage area was calculated
by solving Eqs. (1)- (10), L....J
2nd stage
using a personal computer. ~ ~~--------4------~=
The relationship between
the membrane area and the 20 ----·lst stage +----·--··-,~""'-·-+---····--·---·----------I
concentration of clarified
apple juice at the inlet
of the second stage is 20 30 40
shown in Fig. 1.
Fig. 1 Relationship between membrane area
and concentration of clarified apple juice
at the inlet of the 2nd stage
4) Energy consumption
The energy required to remove 1 kg
of water from clarified apple juice Table 1 Energy required to
(For the case of clarified apple juice remove 1 kg of water
concentrated from 10 wt.% to 40 wt.%) f rom c Ianof"Ied app Ie JUIce
is shown Table 1. The energy This system 63 kJ
requirements of this system and freeze Evaporation
concentration were calculated assuming Single effect 2627 kJ
that the efficiency of electric power Triple effect 997 kJ
generation and pump (or compressor) Freeze concentration 272 kJ
are 0.4 and 0.7, respectively.
CONCLUSION
The permeate flux and the observed rejection in the tight RO membrane
treatment and the loose RO membrane treatment of clarified apple juice can be
predicted by the membrane transport equation based on nonequlibrium
thermodynamics and the concentration polarization equation. Based on these
results. a two-stage RO system of high energy efficiency for the concentration of
clarified apple juice has been proposed.
REFERENCE
1) Nabetani, H. , I. Nakaj ima, A. Watanabe, S. Nakao and S. Kimura: J. C.lJelll. EIJg. Japa.ll,
25,575(1992)
 FACTORS AFFECTING THE PERFORMANCE OF CROSSFLOW
FILTRATION OF YEAST CELL SUSPENSION

TAKAAKI TANAKA AND KAZUHIRO NAKANISHI

Department of Biotechnology, Faculty of Engineering, Okayama University, Tsushima-naka,
Okayama 700, Japan

ABSTRACT
Factors affecting the performance of crossnow filtration were investigated in a module with a
thin channel using Saccharomyces cerevisiae cells cultivated in two media. When a
suspension of S. cerevisiae cells cultivated in YPD medium was cross-filtered, the steady-state
nux agreed well with the value predicted by conventional filtration theory using the specific
resistance of the cake of cells and the amount of the cake deposited on the microfiltration
membrane. The nux decreased considerably due to the small particles «1 /-lm) in molasses in
the case of crossflow filtration of a broth of cells cultivated in molasses medium.

INTRODUCTION
In fermentation industry crossflow filtration has been applying to the separation of cells from
broth. However, optimization of crossflow filtration has not been achieved as the mechanism
of the crossflow filtration has not been clarified. In some cases the permeation flux is not
sufficiently high because particles derived from medium such as molasses cause additional
increase in the resistance to permeation. Here, we showed the factors affecting the
performance of crossflow filtration of yeast cell suspension. We also intended to increase the
permeation flux in crossflow filtration of broth of yeast cells cultivated in molasses by
backwashing.

MATER~LS AND METHODS

Yeast Cells
A strain of S. cerevisiae donated by Kanegafuchi Chemical Industry Co., Ltd. was cultivated
with YPD (yeast extract, polypeptone, dextrose) medium or molasses medium containing 2%
of sugar in a 5-dm3 jar fermentor at 30 OC for 24 h. The aeration rate was 0.5 vvm. The YPD
medium consisted of 1% yeast extract, 2% polypeptone, and 2% dextrose. Antifoam was
added at 50 ppm.

Measurement of Specific Resistance

Cells were washed three times with four volumes of saline. Then, the mean specific resistance
was measured at 20°C with a dead"end filtration module by the steady-state method reported
previously [1].
654
Filtration Module and Operations
The module used for crossflow filtration was of thin channel type (24 mm in width, 305 mm in
length, and 2.5 mm in depth) [2]. The membranes used were synthetic membranes from Fuji
Photo Film Co., Ltd: FM45, FM80, FM120, FM300, and FM500 made of cellulose acetate
and with a nominal pore size of 0.45, 0.80, 1.2, 3, and 5 J.tm, respectively, and an SE45
membrane made of polysulfone and with a nominal pore size of 0.45 ~lm. Yeast suspension or
broth was circulated by a rotary pump at 20 oC. The permeate was usually recycled to the
reservoir to maintain the cell concentration. Yeast cell suspensions were prepared by washing
and suspending cells cultivated in YPD medium into saline.
The backwashing was usually done as follows. After 10 min of permeation, the circulation
flow was introduced to a bypass to stop the flow in the filtration module; 7.3 ml of permeate
was allowed to now from the permeate exit into the module by a hydrostatic pressure
difference. Then, the valve for permeate was closed, and the circulation now was again
introduced into the module. Permeation was started 4 min after the start of bypassing.

RESULTS AND DISCUSSION

Crossnow Filtration of Suspension of Yeast Cells Cultivated in YPD medium
The specific resistance of S. cerevisiae cells cultivated in YPD medium was 2.8xlO 11 m/kg at
49 kPa. When a suspension of S. cerevisiae cells cultivated in YPD medium cross-filtered, the
nux was decreased rapidly after the start of filtration and usually reached a steady state within a
few minutes. The steady-state flux was 1.1xlQ4 mls at the circulation flow rate of 30 cm3 /s,
the cell concentration of 80 kg/m3 (wet weight), and the transmembrane pressure of 49 kPa. It
agreed well with the value predicted by conventional filtration theory using the specific
resistance of cake and the amount of the cake deposited on the microfiltration membrane.
As the circulation flow rate increased, the permeation flux increased due to decrease of cell
deposition. The flux considerably decreased at a circulation flow rate of 15 cm3/s or lower. A
part of the module inside was filled up with cells, which caused channeling of a circulation
flow. The channeling was also observed at a high cell concentration. When the transmembrane
pressure difference increased the permeation flux increased but not in proportion to the pressure
difference because of increases of cell deposition and specific resistance.

Crossnow Filtration of Broth of Yeast Cells Cultivated in YPD Medium

When the yeast cells were cultivated in YPD medium, the cell concentration of the broth was 30
kg/m3 and the permeation flux showed a flux similar to that observed with yeast cell suspension
(Figure 1).

Crossnow Filtration of Broth of Yeast Cells Cultivated in Molasses

The molasses used for cultivation contained 1 kg/m 3 of small particles «1 ~lm). The cell
concentration after 24 h of cultivation was 12 kg/m 3. Figure 1 shows the change in flux during
crossflow filtration of broth of cells cultivated in molasses at the circulation flow rate of 30
cm 3 /s and at the transmembrane pressure of 49 kPa. The flux decreased gradually with time.
After 15 min of filtration a steady state was reached. The steady-state flux was nearly 10-5 m/s
for all the membranes tested. This value was much lower than that observed in crossflow
filtration of broth of cells cultivated in YPD medium.

Effect of Membrane Pore Size on Recovery of Permeation Flux by Backwashing

To increase the permeation flux, we tried backwashing. The recovery of flux strongly depended
on pore size of membranes. The flux decreased as in the filtration without backwashing when
the nominal pore size of membrane was 0.8 J.tm or smaller. The nux was, however, almost
recovered when the pore size was 3 to 5 ~lm.
 655
10- 2

SEAS
....... o YPD

]
........
10- 3
• molasses

= 10-
~

---
4 000 o o
•
C
Q

•
Q.I
S
10- 5
• •
L.
Q.I
Q.c
•
10-6~~--~--~~--~--~~

o 20 40 60
Filtration time [min]
Figure 1. Cross flow filtration of broths of yeast cells cultivated in YPD medium and
molasses medium. Broths were cross-filtered at the circulation flow rate of 30 cm3/s and the
transmembrane pressure of 49 kPa.

CONCLUSION

We cross-filtered suspensions of yeast cells, and broths of cells cultivated in YPD medium and
molasses medium. The steady-state flux agreed well with the value predicted by conventional
filtration theory in crossflow filtration of cell suspensions and broth of cells cultivated in YPD
medium. The flux considerably decreased due to the small particles in molasses in the case of
crossflow filtration of a broth of cells cultivated in molasses medium.

REFERENCES

1. Nakanishi, K., Tadokoro, T., Matsuno., R., On the specific resistance of cakes of
microorganisms. Chem. Eng. Commufl., 1987,62: 187-201.

2. Tanaka, T., Kamimura, R., Itoh, K., Nakanishi, K., Matsuno, R., Factors affecting the
performance of crossflow filtration of yeast cell suspension. Biotechllol. Bioeng., 1993,
41: 617-624.
 CROSS-FLOW MEMBRANE FILTRATION OF SOY SAUCE LEES

TOSHIO FURUKAWA. KAORU KATOU. KEIZOU KUMAKURA. AND KATSUMICHI OSAKI

Technology Center. Production Engineering Division.
Kikkoman Corporation
2470 Imagami. Noda-shi. Chiba 278. Japan

ABSTRACT

The amount of soy sauce lees coagulated and precipitated at about 50°C during
3-8 days reaches around 5-15% of the volume of raw soy sauce. Four different
types of polymeric membrane module for cross-flow filtration were applied for
the recovery process of soy sauce from lees. Cross-flow filtration for lees
is discussed from two aspects; energy consumption on permeate and product
properties. Permeate per energy input defined by PIE (L/kW·h) is compared
for tubular. hollow fiber. rotary. and plate & frame-types of modules. The
plate & frame module. mounted with ultra and micro-filtration membranes. was
found to reduce membrane fouling effectively. This particular type of module
provides the least changes in color and taste in soy sauce and the highest
PIE at volume reduction ratio of 5 among the four types of module.

INTRODUCTION

Japanese soy sauce has been used for over 300 years as a food flavoring and
is currently served throughout the world as an all purpose seasoning. The
four major characteristics are aroma. flavor. color. and stability. and these
properties should be considered first when introducing a new process.
Membrane filtration for soy sauce has been researched and applied in the
final refining process since 1978[1] in Japan. The cross-flow ultrafiltration
was first substituted for diatomaceous earth filtration of raw soy sauce and
soy sauce lees about 15 years ago. In view of the developing membrane
technology. the membrane filtration process for lees had to be restructured.

EXPERIMENTAL METHODS

Experimental device and operating conditions: Four different types of module

 657
were tested to measure permeate flux. the possibility of concentration of
soy sauce lees and pumping energy. Each module was operated as a batch
process at 45 ± 2°C. under condi tions described in Table 1.

Analytical procedures: Permeate was analyzed for color change. molecular

weight distribution and sensory evaluation. Ingredients of the permeate were
analyzed by standard methods (SOY SAUCE STANDARD ANALYSIS). Molecular weight
distribution was measured by HPLC (TOSO:PFK-gel G3000SW). Color was measured
by Color Number Meter(OMRON:3WCOL-IO). Sensory evaluation was performed by
10 skilled panels.

TABLE 1
Specifications of membrane modules and operating conditions of each module

Membrane materials Cross-flow Treated volume/

Module configuration /Nominal pore size veloci ty(mis) Area:V /A(Lm')
Trade name
Membrane area Cm") Pumping In 1e t Ou tl e t
power(kW) pressureCkP)

Hollow fiber polyethylene O. 5m / sec 350(L/m 2 )

O. 2 micron
MEMCOR 3.0 m' 1. 0 kW 300/200(kP)

Tubular polyolefin 0.5-1. 5m i sec 200(L i m')

MITSUBISHI RAYON 150 kDa
ENGINEERING 19. 2 m" 7.0 kW 350/100(kP)

Plate & frame a. PTFE b. PIFE

& polysulfone l.0-2.5m sec 170 350(Lm")
NlITO DENKO a. 0.2micron
b. a/50 kDa
a. 3.0 m2 a. 2.0kW 450 50 (kP)
b. 12.0m"(6m'+6m') b. 7 7.5kW

Rotary disk polysulfone 4.0-6.0m!sec 450 (L m' )

HITACHI PLANT ENG. 750 kDa
NlITO DENKO 2. 2 m' 3.0 kW 100200 (kP)

RESULTS AND DISCUSSION

Selection of a suitable module for soy sauce lees: Permeate fluxes were
compared with each other against the volume reduction ratios. The rotary disc
type module indicated a stable flux up to 20 volume reduction ratio(VRR). The
plate and frame module was successful in maintaining a high flux up to 5 VRR.
The tubular and the hollow fiber modules had no significant flux for any VRR.
However. the level of energy consumption for each module were different from
the gradient of flux vs. VRR.
658
Permeate per energy input CP/E) defined by
permeate fluxCL/mloh)
pumping energy per unit membrane area CkW/m2) (1)
is given in Table 2.

TABLE 2
Mean flux. energy input per unit area and permeate per energy input
PIE of each module at 5 VRR at 45°C

Membrane Module Mean fl ux CLMH) E/ACkW/ml) P/ECL/kWoh)

Hollow fiber 8-10 O. 31 25.8-32.3

Tubular 10- 15 0.28-0.36 35.7-41.7

Plate & frame 18-30 0.35-0.63 34.5-51.4

Rotary disk 25-35 1. 55 16.1-22.6

Differences between the UF/MF and the MF/UF- module: The plate and frame
module of 12m 2 with 73 plates is divided into UF membranes C6ml/36p) and MF
membranesC6m 2 /36p). The liquid inlet of the UF/MF module is on the side of
the UF membranes and the PIE value of the UF/MF module is higher between
30-50°C up to 5 VRR than that of the MF/UF module.
Evaluation of permeate properties: The permeability of high molecular
weight materials around 300kDa varied between the UF/MF and the MF/UF-
modules. As high molecular weight materials were extruded into the permeate
by force at a higher operational pressure. the permeability of those
materials was accelerated in the MF/UF module. Although the UF permeate
color in both the UF/MF and the MF/UF-modules underwent a similar change.
the MF permeate in the MF/UF module changed to a lighter color than that of
the UF/MF module. These results are considered as a kind of dynamic
membrane phenomena.

CONCLUSION

The index of P/E. defined by Eq. (I) is useful for the evaluation of various
membrane modules. The membrane module with a high PIE value provides low
energy consumption and the cost reduction of membrane modules and the
membrane replacement. As the PIE value is particularly i~fluenced by the
change of viscosity of the lees which depends mainly on the temperature and
VRR. the P&F membrane filtration should be operated at around 40°C up to
10 VRR. where the change of the permeate properties is relatively small.
Soy sauce lees have been successfully filtrated to 5-10 VRR in the
commercial size of P&F module with UF/MF membranes since 1992.

REFERENCES
1. Tadanobu Nakadai, and Hironaga Hashiba. JAPAN PATENT NO.53-1360Cl978)
 APPLICATIONS OF THE CERAMIC MEMBRANES TO SOY SAUCE

NOBUHIKO KANEKUNI, HISASHI NOGAKI, KENJI TABATA and ATSUO WATANABE.

Research and Development Div .• TOTO Ltd.
2-8-1 Hanson. Chigasaki-shi. Kanagawa 253 Japan

ABSTRACT

We developed two types of cross-flow filtration modules with ceramic

membranes. Each module was applied to recover soy sauce from sediment. It
was confirmed that the permeation flux of the proposed modules were
higher than that of the conventional modules. and soy sauce could be
recovered from sediment at the yield of more than 90%. It seemed that the
proposed modules could work more effectively than the conventional type.

INTRODUCTION

Recently ceramic membranes have attracted special interest because of

their superior characteristics. However. few cross-flow filtration modules
have been used for recovering or retaining any substance with high
concentrated liquid or viscous fluid. due to a difficulty of circulating
viscous fl uid.
In this paper. we propose two novel cross-flow filtration modules to
apply for low fluidity liquid. And the application of the modules for the
recovery of soy sauce from the sediment. which is known ,to be highly
viscous liquid. is discussed.

EXPERIMENTAL

Fig.1 shows the schematic diagrams of two types of cross-flow filtra-

Concentra te
Concentrate

Ceramic Disc
Membrane
ate

Feed Feed

Permeate
a) Stationary Disc Membrane Type b) Rotary Disc Membrane Type
Fig.1 Schematic diagrams of two types of cross-flow filtration modules
with cer am i c mem br anes
660
tion modules with ceramic membrane~):
])Stationarv Disc Membrane TYoe CSDM-typel ; The stationary
membranes are enclosed in a housing with rotating impellers
configurated in the vicinity of membrane surface.
2)Rptary Disc Membrane Type (RDM-type) The membrane is
rotating in the stationary shell. The filtrate is drawn off
through a hollow shaft

The modules were examined by means of two methods: total circulating

filtration and batch filtration. The specifications of membrane modules and
operating conditions are shown in Tables 1 and 2, respectively.
The sediment of soy sauce employed in this study was a blackish fl uid
with viscosity of 5 - 15cP. The particles suspended in the sediment have a
size distributed from 0.3 to 45,um.

Table 1 Characteristics of membrane modules

Mater ial Aluminium-oxide

Pore Diameter O.l,um
Sharp DiscCExternal Pressure Type)
Membrane Area 2.0m 2 CSDM-type),O.3m2 (RDM-type)

Table 2 Operating conditions for cross-flow filtrations

Total Circulation Trans-Membrane Pressure 50 - 200kPa

Filtration Revo I ution 50 - 500rpm

Batch Filtration Trans-membrane Pressure 200kPa

Revolution 500rpm

RESULTS AND DISCUSSION

Fig.2 shows the effects of trans-membrane pressure and revolution on

permeation flux for the SDM-type module. The permeation flux increased
with increasing trans-membrane pressure and with increasing revolution.
The pressure dependence of permeation flux was similar between the two
modules, although the permeation flux for the RDM-type module was 30%
smaller than that for the SDM-type module.
"
~ 50r-----r-----~----~----~----,
..c: I Feed Sedi.ent(So¥ Sauce)
I Viscosity 5.32cp Revolution
Melbrane M8 ter i a I A.1 2 03
Fig.2 Effects of trans- Pore Dialleter
I Modu Ie bpe
O.I/.lI
SDN-t¥pe
500rpm(10m/s)
mem br ane pressure and L....J I Mubrane Area 2.0.?
I Teoperature Z5"C
revolution on permeation
flux for the SDM-type .=; 30
modul e 300rpm(6m/ s)

lOOrpm 2m/s)
r---,,--------~

50 100 150 200 250

Trans-membrane Pressure ~Pm [kPaJ
 661
Fig.3 shows the relationships between permeation flux and volume yield of
permeate. The symbol ..... shows the filtration characteristics of an
organic membrane with hollow fiber2). The permeation flux of all modules
decreased as the yield increased. The permeation flux of the organic
membrane was about as half as that of the SDM-type module. Furthermore.
the upper limit of the yield was 75% in the organic membrane. while the
proposed modules could filtrate to attain the volume yield of 90%.
Fig.4 shows the influence of volume yield of permeate on the viscosity
of sediment. A steep increase in the viscosity was observed at the yield
of 75%. And at the yield of. 90%. the viscosity of the concentrated sedi-
ment became more than 1000 times as much as that of the initial feed
sediment. The conventional hollow fiber module concentrated the sediment
the viscosity of which was less than several poise. whereas the proposed
modules. SDM- and RDM-type modules. could filtrate the sediment with the
viscosity of more than 100 poise.
Concentration Ratio
,..., 50
,-..,
~
.J::
Feed Sedillent(Soy Sauce)
Trans-llelibrane Pressure 200kPa
(IJ
E 40 TeIlPeratuxe__ ~. 25'C 200 Feed
'-../
....... key Module Ty Sedilent(Soy Sauce)
L.J
0 0 SOM-type Viscosity

•
> t:. ROM-type
~
O.05PDise at Yp=O%
'""?
30 Hollow Fiber 150 TelPeature
25'C
x
:::l ~~ . Viscoleter

~~O
~ Brookfield type
c 20
0
......
ro
a;l
••• ~Q
E
~
10 • t::.0:>q,
~t::.
•
a;l
p...
•
0
95 95
Volume Yield Volume Yield
Fig.3 Relationships between permeation Fig.4 Influence of volume yield of
flux and volume yield permeate on viscosity of sediment

CONCLUSION

In order to recover SOy sauce from sediment. we developed two novel

cross-flow filtration modules with ceramic membranes. Compared with the
conventional membrane module. the newly developed modules had the higher
yield as well as the higher permeation flux. The proposed modules seem to
be suitable for the concentration processes in food and medical field.

ACKNOWLEDGMENTS
We would like to express our appreciation to Mr.T.Takatoh of Shizuoka-
kensan Soy Sauce Industrial Co. for his help in the experiments.

REFERENCES
UMatsushita.K .. et al.. Preprints of the 58th Annual Meeting of the Society
of Chem i ca I Engi neer s.Japan.B-206.Kagoshima(1993)
2>Takatoh.T .• et al.. Shoken.19.No.H1993)
 CERAMIC FILTERS FOR FOOD AND BEVERAGE APPLICATIONS

M.FUSHUIMA
Millipore Corporation 80 Ashby Rd. Bedford, MA 01730 USA

Abstract

Ceramic membrane filtration devices address environmental issues and process

improvements in the food and beverage industry. Features and operating modes of ceramic
microfiltration systems are described. Installed production process designs in Europe, USA and
Japan are detailed for juice, soup, beer bottom, wine and vinegar processes to show the
benefits of systems using ceramic filters.

1. Background of improved yield, consistent quality and cost

Diatomaceous earth (DE) filter media
has traditionally been used to clarifY beverage
products such as beer and wine. However,
environmental concerns are forcing users to
seek alternative methods because disposal of
media is becoming more restricted every year.
Membrane microfiltration in a cross flow
mode has been noticed as a potential
replacement because it gives the same or
"more sharp" separation consistently along
with other benefits. Two negative factors
related to conventional polymeric membranes
have prevented a wide usage of cross flow
membrane devices in beverage applications.
They are 1) short membrane life time due to
limited temperature and chemical resistance
and 2) flavor changes caused by the
extraction of polymers, especially adhesive.
Ceramic filters address all of these
issues. Environmental concerns are minimized
by eliminating the use and disposal of filter
media, while offering the additional benefits
savings in both labor and energy. Ceramic
filters can have significantly longer lives
without releasing polymers to product
streams. Thus microfiltration using new
ceramic materials has become not only an
alternative, but a primary technology for
these applications.
Ceramic filters features
Thermal resistant
Chemical/biological resistant
Sanitary design, open channel
Mechanical strength, backpulsable
Steamable, sanitizable
No flavor change
Energy effiCient, 2-6 mm lumen
Minimum pretreatment
Long life 3-5 years

2. Features of ceramic filters
The most significant advantage of
ceramic filter is their extraordinary thennal
resistance, enabling SIP (steam in place)
sterilization and high temperature operation.
The filter module configuration is an open
channel, the best sanitary design. Choices of
lumen diameter of 2-6 mm allow optimizing
energy efficiency and prefilter requirements.
Larger diameters, 4-6 mm are used for
viscous feeds like yeast, while 2-3 mm are
selected for watery streams like wine.
Membranes are available in a variety of
materials; harder ones like a.-alumina have
superior durability compared to y-alumina.
Membrane pore sizes of 0.2J..lm, 0.45J..lm and
1J..lm are used depending on the application.
Semi batch in feed & bleed
811 Continuous operation
Small process tank
lIB
lIB Shortest residence time
Many ceramic filtration systems are
operated in a semi batch mode. Others
include complete batch, feed & bleed
continuous and multi-stage continuous
operating modes. Feed is introduced to a
process tank which is also used as a CIP tank
to minimize tankage requirements. Retentate
is returned to the process tank rather than the
feed tank to avoid property changes. If
necessary, diafiltration water is added after
concentration to achieve higher yields. A
system can be sanitized by steam or chemicals
such as chlorine. Then it can be drained and
stored in a dry mode eliminating storage
steps. A significant advantage comes from the
ceramic filter configuration is the
"backpulse". This is possible because the
membrane is sintered onto the substrate not
adhered. By applying backpulse, higher and
more stable fluxes can be achieved, while
chemical cleaning is reduced.
663
3. Application
Cranberry juice The first application of
ceramic microfiltration was to clarify
cranberry juice. Initially, replacement of a DE
based pressure filter was intended. However
maintaining consistent clarity became more
important, and the ceramic system is now
used to polish output of the pressure filter.
This ceramic microfiltration system in the
USA processes 6 m3/hr of cranberry juice
bringing 10-20 NTU turbidity down to less
than 1 NTU. Other benefits are reduced
usage of DE and extended life of the down
stream evaporator.
Cranberry juice process
Soup The next application commercialized
was soup clarification in Canada. A
traditional method to clarify soup is to absorb
fat and other cloudy material by adding high
quality lean ground beef This process
required a full time operator and had to rely
on his "sense by experience". Replacing this
step with a ceramic filter that can withstand
50-60°C operating temperature, significant
savings were realized by eliminating raw
material while maintaining consistent quality
product.
Beer process
draft beer
664
Beer bottoms Many attempts have been
made to apply cross flow filtration to clarify
beer or to concentrate beer bottoms as many
reports can be cited in the literature. Beer
bottoms concentration is first being
commercialized in Europe and Japan because
it requires relatively small capital investment,
and gains from recovering beer often makes
the pay back period less than one year. Beer
bottoms or yeast are used as additives to
medicine and natural foods. By concentrating
two fold, a saving in transportation and fuel
and recovery of beer are both achieved at the
same time.
Wine process
I Crush I I
In Spain and Italy, ceramic microfiltration has
been introduced to replace these steps. The
benefits are elimination of DE disposal,
reduced labor cost to one fifth, elimination of
wine loss in the DE media or pad filter,
improved wine clarity (fining index 10-20 to
1-5) and extended life time of the final
membrane filter. Another hidden but
important point is that wine quality became
more consistent (separation is done by
absolute pore size rather than depth filter
mechanism). A typical performance and
economic comparison are shown.
Wine economics
us cents I kl
OE+Pe.d Ceramic
capital 36 (DE) '02
I 8 (Pad)
Stabilization I Energy 260 '45

I I DE media 132

I Press I I Blend and store I Pad

""
I I Ceramic 88

I Fermentation I I Ceramic MF I Replaced: Labo< ,00 20

I I ~~~E Total 586 '66

I Racking I I Final filter I ~':'mb P'6
I 1
I Fining I I Filling I Vinegar The last application mentioned here
I is vinegar clarification. After fermentation to
acetic acid, broth can be clarified with either a
Wine A great deal of effort has been made to DE filter or ceramic microfiltration as the
apply membrane technology in the wine process is shown. Another application in the
making process to improve existing vinegar process is stabilization by removing
operations. The applications that have been protein via ultrafiltration. In Italy and Japan,
investigated are juice and must clarification, actual systems have been installed.
clarification of wine and sterilization of wine
Vinegar process
Wine flux data with Ceraflo
Spain 1992 Corn. rice Fruit, alcohol
250 Flux [lmhJ

I__ Whlte-rv::...ot:~~Wn.
200 Yeast
,so .~ f--..- ................. c··········

'00 ...... '\

50 .................. ~ . ........

6 10
I D.E.filter
Time [h,-]

before filling. Also with reverse osmosis,

4. Conclusion
concentration of juice prior to fermentation
Ceramic cross flow filtration has
and dealcoholization of wine have been
found several commercial applications in food
attempted. Among many possibilities, two
and beverage processes. The advantages of
applications in the wine industry are being
both material and configuration fit many
established. Clarification of wine after
needs by food equipment. Therefore, more
fermentation employs a traditional DE filter
and more new applications will be developed
(coarse and fine) followed by a pad filter.
over the next several years.
 Recovery and Functional Properties of Protein from the
Wastewater of Mungbean Starch Processing by Ultrafiltration

W.C. KO, W.J. CHEN, and T.H. LAI

Department of Food Science,
National Chung Hsing University,
250 Kuokuang Road, Taichung, Taiwan, ROC.

ABSTRACT
Wet process, a conventional method for mung bean starch
preparation, may cause about 80% protein loss in wastewater.
In this study, ultrafiltration(UF} was used to recover the
protein from the simulated wastewater of mungbean starch
processing. 40·C and 4 kg/cm 2 were found to be the appropriate
condition for operation. Recovery of protein was 87.8% and
50.3% for mo I ecu I a r we i gh t cu t - 0 f f ( MWCO) 30,000 and 500, 000
membranes respectively. Protein remained in the wastewater
could almost be separated by UF with the MWCO 30,000 membrane.
It seems feasible to recover the protein by UFo This would not
only prevent contamination but also improve the utilization of
mungbean.

INTRODUCTION
Mungbean starch noodle is one kind of tradi tional Chinese food
products. Wet process, the most popular method being ~sed to
prepare mungbean starch in Taiwan' may cause water pollut ion
and loss of a nice protein source(1). It is necessary to
recover the protein to prevent contamination and make the
complete uti I ization of mungbean possible.
There had been some reports about protein recovery from
the wastewater{2}. In spite of over 80% protein being
recovered' those methods were not adapted in actual practice
because protein denaturation might be resulted and large
amounts of coagulants were needed.
Ultrafi Itration could be operated at considerably low
temperatures and could also prevent the protein from denatura-
tion. The objectives were to study the performance and pro-
cess engineering characteristics of the spiral-wound modules
used for isolation of protein from the wastewater of mungbean
starch processing. Functional properties of recovered protein
were also determined.
666
RBSUL TS AND DISCUSSION
The effect of inlet pressure on permeate flux is shown in
Fig.1. Permeate flux increased with the increase of operating
pressure under pressures below 4 and 3kg/cm 2 , and then stabi I ized
at values of 7.0 and 4.3L/m 2/hr for the membranes of MWCO 30,000
and 500, 000 r espec t i ve I y. I t i nd i ca t es t ha t flux in i t i a I I y
increased with pressure and then became independent. The major
resistance to permeate flow in the pressure-independent regions
was due to the concentration polarization-gel layer.

8
0 - 0 YWCD 500.000
.-.YWCD 30,000 ___ 0 - 0
/0
o

/ .-.-.
~>/­
o~
•
1+----r--~~--~---+----r----r--~
o 1 2 3 4 5 6 7
Pressure ( kg/cm 2)

Figure 1. Relationship of operating pressure and permeate flux.

The effect of inlet temperature on permeate flux is shown

in Fig. 2. Inlet temperature enchanced the permeate flux with
a rat io of 3%/oC for both membranes.

11 YWca 30,000 MWCO 500,000

10
9
8
7
6
5
4
3
2
1
O+--r--~~~--+-~
o 10 20 30 40 50 60 0 10 20 30 40 50 60
Temperature ( 0 C )

Figure 2. Relationship of inlet temperature and permeate flux.

0-0 1 kg/cm 2 ; • - . 2 kg/cm 2 ; ~ - ~ 3 kg/cm 2
.. - .. 4 kg/cm 2 ; 0 - 0 5 kg/cm 2 ; • - . 6 kg/cm 2
 667
Rejection coefficient for total sol ids, protein and ash for the
two membranes are illustrated in Table 1. MWCO 30,000 membrane
could recover 90% protein and most total sol ids. On the other
hand, MWCO 500,000 membrane could recover only 60% protein
because 70-80% of protein was agglomerated into protein
bodies(3).
TABLE 1
Recovery of the compositions during UF processing at VCR of 12.
retentate permeate retentate permeate
MWCO 30,000 30,000 500,000 500,000
Total so lids, % 62.5 37.3 55.8 42.3
Ash,% 34.1 59.3 34.9 62.1
Protein,% 92.1 5.2 55.9 43.7
VCR volume concentration ra t i 0
Functional properties, including solubi I ity, emulsify
activity, water-binding capacity and foam expansion, of spray-
and freeze-dried products were determined. Because of the
protein recovered by MWCO 500,000 membrane was mostly existed
in the protein recovered by MWCO 30,000 membrane, functional
properties of the products obtained from two membranes were not
significantly different. Solubilities of the four products were
over 85% and higher than the bubble separation method (2).

Table 2
Functional properties of dr i ded retentates from UF treatment
MWCO 30,000 30,000 500,000 500,000
drying method spray freeze spray freeze
solubi I ity,% 88.71 85.24 86.33 85.15
water-binding 7.23 7.09 8.46 7.89
capacity
foam expansion,% 76.5 80.3 78.9 79.4
emulsify activity 0.640 0.602 0.651 0.615
0.0. ( 500 nm )

REFERENCES
1. Hi I I, J. and Vananuvat, P. Prel iminary investigation of
mungbean protein preparations. Report No.1, Appl ied
Scientific Research Corporation of Thai land, Bangkok, 1968.
2. Peng, W.Y. and Chiang, W.C. Recovery of protein from the
wastewater of mungbean starch noodle factory by absorption
bubble separation. Food Sci.(Taiwan), R.O.C. 1982,8,93-102.
3. Harris, N. and Chrispeels, M.J. Histochemical and biochemical
observations on storage protein metabol ism and protein body
autolysis in cotyledons of germinating mungbeans. Plant
Physiol. 1975,56,292-7.
IMPROVED EXTRACTION AND ULTRAFILTRATION PROCESS FOR PROTEIN
RECOVERY FROM SOY FLOUR

JOHN L. HARRISl, S.K. SAYED RAZAVI l & F. SHERKAT2

1. Department of Chemical & Metallurgical Engineering 2. Food Technology Unit
RMIT, 124 Latrobe Street, Melbourne 3000, AUSTRALIA.

ABSTRACT

By using a feedstock of dry milled enzyme-active full-fat soy flour, median particle size of 50
microns, an extraction process was developed which gave high solids and protein recovery.
Aqueous extraction for 10 min at 80°C, and homogenization at 10 MPa followed by filtration
gave 95.6% solids recovery and 86.9% protein recovery with a nitrogen solubility index of
93%. The native lipoxygenase was inactivated and the activity of trypsin inhibitors was
lowered by 50%. The extract was concentrated to 20 wt% solids by ultrafiltration. The
concentrate changed from dilatant to pseudoplastic behaviour at 14-15 wt% solids.

INTRODPCTION

Soybeans are highly valued for their protein content (about 40%) which can be extracted with
water. The traditional method of extraction by wet-milling the soaked soybeans, heating the
mash for 15-20 min near its boiling point, followed by filtration gives a low solids recovery of
26% and protein recovery of33% (1). Improved yields of solids (58%) and proteins (44%)
were reported for a hot grind process (1). Another method where the soaked soybeans were
blanched at 90°C for 3 mins, ground hot for 5 mins and filtered gave a solids recovery of
73 %, protein recovery of 81.6%, inactivation of lip oxygenase and reduction in trypsin
inhibitor activity of 62% (2). In a recent method, 80% protein solubility was obtained by
extraction at 80-85°C for 2 min (3). A continuous steam infusion cooking system, known as
rapid-hydration hydrothermal cooking (RHHC), made a hot water slurry with full-fat soy
flour at ~80°C followed by cooking at 154°C for 18 s, and resulted in lip oxygenase
inactivation with 100% solids recovery as no separation steps were involved (4).

The heating (temperature-time relationship) required for inactivation of trypsin

inhibitor can have a deleterious effect on product quality in terms of protein solubility. The
desired degree of solubility depends on the end use of the product (5).
 669
Ultrafiltration has been used successfully to remove undesirable factors and produce
specific products from soy extract. 96% removal of oligosaccharides, 90% removal of phytic
acid and 17% reduction of trypsin inhibitor activity have been achieved (6).

MATERIALS AND METHODS

The enzyme active full-fat soy flour was milled from cleaned, dehulled soybeans and had a
proximate composition of 40% protein, 20% crude lipid, 27% carbohydrate, 5% ash and 8%
moisture (supplied by Soy Products of Australia Pty. Ltd.). The particle size specification of
the flour was 95% pasing through 100 mesh sieve. The extraction method involved preheating
the water to the required temperature (30-90°C), mixing the soy flour with water (1: 12 w/w),
stirring at 1200 rpm with a propeller stirrer for 10 min (excessive frothing occurred for longer
extraction times), cooling in a cold water bath to 40°C, homogenizing with a Manton Gaulin
V90 homogeniser and filtering through a 150 micron sieve. The soy extract contained
approximately 6.2% total solids. Ultrafiltration was performed at 50°C and 400 kPa in a DDS
flat plate module with 2.25 m2 area of nominal 50,000 M.W. cut-offpolysulphone membrane.

Analyses performed included total solids [AOAC gravimetric method 925.23 (1990)],
trypsin inhibitor activity [AOCS method Ba 12-75 (1983)], total nitrogen by Kjeldahl method,
total protein as 6.25 x Total Nitrogen and nitrogen solubility index (NSI) [AOCS Method Ba
11-65 (1985)]. Physical properties measured included particle size using Malvern Instruments
Master Sizer X Ver. 1.0 and viscosity using Contraves Rotational Viscometer Rheomat 115.

RESULTS AND DISCUSSION

There is a significant difference between the particle size distribution curves of unfiltered
suspensions of full-fat and defatted flour prepared at 20°C as shown in Fig.I. Both suspensions
exhibited peaks at 50 f..lm, but there is an absence of a 10 f..lm peak for the defatted flour. The
soy extracts prepared at temperatures up to 90°C also exhibited peaks at 50 f..lm as shown in
Fig.2, but the 90°C extract contained more larger particles due to protein denaturation. The
effect of homogenization at 10 MPa was to slightly reduce the median particle size, but
homogenization at 17.5 MPa lead to an increase in frequency of large particles (>200f..lm).

ur-------------------------~ ur-------------------------~
FtJlL.FAT c 60°C []
10
DEFATTED •
10
sooc .t.
90°C 0

\
It

~
\
5 10 20 so 100 200 soo 2 5 10 20 so 100 200 soo
PARTICLE DIAMETER (um) PARTICLE DIAMIn'ER (1l1li)

Figure 1. Particle size distribution of suspen- Figure 2. Effect of extraction temperature

sions offull-fat and defatted soy flour at 20°C. on the particle size distribution.
670
100

9S

SOLIDS RBCOVERY(") 0

70 FROTElN RBCOVERY(") •

NSl(")
A
10
6S

~ ~ ~ ~ 70 ~ 00 10 2:1 30
EXTRACTION TEMPERATIlRE (0C) TIME (miD)
Figure 3. Effect of extraction temperature Figure 4. Heat inactivation of trypsin
on solids recovery, protein recovery and NSI. inhibitor.

Solids recovery exceeded 95% for extraction temperatures up to 80°C, and declined
thereafter (Fig.3). Protein recovery declined from 50DC indicating heat denaturation of protein
molecules. This is supported by the NSI decreasing from 50°C, with sharp decrease above
80DC. Homogenization at 10 MPa gave a 9% increase in NSI at 80D C. Trypsin inhibitor
activity was lowered by 50% for heat treatment of 10 min at 80DC (Fig.4).

The extract was concentrated from 6 to 20 wt% solids by ultrafiltration. A substantial

reduction in the permeate flux rate was observed at 14 wt% solids. Interestingly, the
rheological behaviour of the concentrate measured at 25°C changed from dilatant «14 wt%
solids), to Newtonian for 14-15 wt%, to pseudoplastic (14 - 20 wt% solids).

CONCLUSIONS

Water extraction at 80°C for 10 min with homogenization at 10 MPa resulted in a solids
recovery of95%, protein recovery of 86.9% and nitrogen solubility index of93%. The
residual trypsin inhibitor activity was 50%. The next step of process improvement is to
investigate extraction temperatures above 80°C for extraction times less than 10 min.

REFERENCES

1. Johnson, K.W. and Snyder, H.E., Soymilk: A comparison of processing methods on

yields and composition. J. Food Sci., 1978,43,349.
2. Al-Kishtaini, S.F., Methods of preparation and properties of water extracts of
soybeans. Ph.D. thesis, Univ. of Illinois, Urbana, IL.,1971.
3. Xie, Z.L. and Frelzdorff, B., Optimizing soybean blanching in relation to lip oxygenase
inactivation and protein solubility for soyrnilk production. Zeitschrift fur Lebensmittel-
Untersuchung und-Forschung.1992, 194(1),43-46.
4. Kim, C. Physico-chemical properties of soybean extracts processed by rapid-hydration
hydrothermal cooking. Thesis, Iowa State Univ. of Sci & Tech., Ames, lA., 1989.
5. Snyder, H.E. and Kwon, T.W. Soybean Utilization. AVI Pub!. Co. New York, 1987.
6. Omosaiye, 0., and Cheryan, M. Ultrafiltration of soybean water extracts: Processing
characteristics and yields. J. Food Sci., 1979,44, 1027.
 THE PRODUCTION OF PROTEIN ISOLATES FROM CHINESE RAPESEED

Levente L. Diosady,
Department of Chemical Engineering, University of Toronto,
Toronto, Ontario, Canada, M5S 1A4

A process tailored to the chemical characteristics of rapeseed and canola proteins was
developed, consisting of extraction of oil-free meal at pH 10.5-12.5, isoelectric precipitation to recover
the isoelectric proteins and ultrafiltration followed by diafiltration to concentrate and purify the acid-
soluble proteins remaining in solution. These steps complement one another to give three products with
excellent protein recovery. Using 5 types of canola meal, "isoelectric" and "soluble" protein isolates
containing 87-104% protein (Nx6.25) and a meal residue were obtained. All fractions were free of
glucosinolates. The two isolates were low in phytate, light in color, and mostly bland in taste. The isolate
yield was dependent on the starting meal. The "soluble" protein isolate has a high nitrogen solubility index
throughout the pH range of 2.5 to 11, which makes it suitable for use in carbonated soft-drinks. This
makes it possible to produce safe and nutritious soft drinks for regions of the world that are chronically
short of protein. The process is especially suited to Chinese rapeseed meal, which is now primarily
wasted as a fertilizer, due to its large glucosinolate content.

INTRODUCTION

Canola is the source of 60% of the vegetable oil consumed in Canada, and it is the third largest
oilseed crop world-wide. The meal produced by extracting the oil is an excellent source of proteins, but
unfortunately it contains glucosinolates which are toxic; phytin, which is an anti-nutritional substance;
polyphenols, which produce a dark colour and a bitter flavour; and some 30% hull, which is high in fibre.
These restrict its use in feed, and prevent its use in food products.
We, in the Food Engineering Group at the University of Toronto have developed a novel approach
to the processing of canola. Using a solvent system consisting of methanol containing ammonia, and
hexane a high-quality degummed edible oil and a meal, that is essentially free of glucosinolates and low in
polyphenols, are produced. The meal contains about 50% protein and is light in colour and bland in taste.
It is much superior to conventional canola meal, and its protein content is nutritionally equivalent to
soybean meal, [1].
China is the largest producer of rapeseed, producing mainly varieties that contain very high
levels of glucosinolates. The use of the meal in feed is severely restricted, and it has been primarily
used as fertilizer - a waste of some 1.5 million tons of excellent protein annually. The methanol-
ammonia/hexane extraction process has been successfully applied to Chinese rapeseed, to produce a
meal superior to canola meal, which may be used freely in animal feed. Unfortunately, due to the high
phytate and fibre content of the meal, as well as the residual glucosinolates, the meal is unsuitable for
human consumption.
Canola and rapeseed protein has a somewhat better amino acid distribution than soy protein.
Unfortunately, unlike the protein in soybeans, it is very difficult to extract, and to recover by the
conventional protein isolation techniques. The losses in the process has made its production uneconomical.
By combining membrane concentration and purification techniques with conventional protein
isolation methods we have produced two high-quality protein products:
672

1. an isoelectric protein isolate, (PPI) which contains -95% protein (NxB.25, dry basis),
which has similar functional properties to commercially available soy protein isolates; and
2. an acid soluble protein isolate, (SPI) which contains -95% protein (NxB.25, dry basis),
which is soluble under mildly acidic condttions, down to pH 3.
Both isolates are free of glucosinolates or their breakdown products, and have low phytate levels
(1.0-1.5%). They are white in colour and bland in taste. The goal of our project was to develop a
process for making food-grade protein isolates from canola and conventional (Chinese) rapeseed
varieties.

MATERIALS AND METHODS

Chinese rapeseed (Nin-U 7 from Jiangsu province) was extracted wtth methanol containing 10%
ammonia (w/v) and 5% water (v/v) as described by Rubin et aI., [2]. The resulting meal was extracted
with 30 volumes of aqueous NaOH at pH values between 10.0 and 12.5. The extract was filtered, and
oto 15% CaCI2 was added, and the pH decreased to a predetermined value between 3.5 and 7.0, thus
precipitating a significant fraction of the dissolved protein. The remaining solution was ultrafiltered to a
concentration factor of 10, and diafiltered with a diavolume of 5 to produce a clear solution. The acid-
soluble isolate was recovered by freeze drying of this solution. The procedure closely followed that
published earlier, [3].
The composition and the functional properties of the resu~ing isolates were determined according
to the techniques described earlier [4].

RESULTS AND DISCUSSION

The composition of canola and Chinese rapeseed and the meals produces are shown in Table I.

Rapeseed Oil Protein Glucosinolate Phytic acid

Sample variety (%) (%) (Ilmol/g) (%)
Chinese 38.8±O.2 25.2±O.2 - -
Seed
Canola 44.5±O.3 21.6±O.2 - -
Hexane- Chinese - 48.1±0.4 113.2±5.1 3.12±0.14
defatted
meal Canola - 44.7±0.1 13.0±0.5 3.86±O.05
CH3INH3/H20- Chinese - 54.8±O.3 4.0±0.3 3.86±O.05
hexane-
extracted meal Canola - 48.4-0.2 0.0±0.2 4.39±0.04

At pH 12, the protein extractability of methanol/ammonialwater-hexane-extracted Chinese

rapeseed meal was found to be over 70%, which was some 15% higher than that of canola meal treated
by two-solvent-phase-extraction. Although Chinese rapeseed meal also showed a slightly greater phytic
acid extractability than canola meal, since its initial phytate content was some 30% lower, the amount
of phytic acid dissolved in the alkaline extract was actually somewhat lower than in the case of canola.

The extracted Chinese rapeseed protein exhibited distinctly different solubility and precipitation yields
than typical canola protein. While -55% ofthe dissolved canola protein was precipitated at pH 3.5, more
 673
of the Chinese rapeseed protein remained in solution at this pH, but was insoluble at neutral pH values.
We found that the phytate content of the isolates was significantly lower than that of canola isolates,
and thus CaCI2 treatment to dissociate phytate-protein complexes was unnecessary, thus simplifying
the process.
The pH of a 10% dispersion of CH30H/NH3/H20-hexane-extracted Chinese rapeseed meal
had a pH value in the alkaline range, which was more than one unit higher than its canola counterpart .
The slurries of the isolates were in the neutral pH range, as expected. While soybean protein had a
minimum nitrogen solubility index (NSI) at pH 5, the NSI of Chinese rapeseed PPI reached its lowest point
at pH 6. Over the almost entire pH range, the NSI of Chinese rapeseed SPI remained consistently high.
This will make it possible to incorporate this product in carbonated beverages as atrue solution.
CH30H/NH3/H20-hexane-extracted Chinese rapeseed meal was slightly inferior in most
functional properties to similarly prepared canola meals The water absorption of Chinese rapeseed PPI
was 220%, some 80% lower than that of the canola PPI. All rapeseed protein products investigated
were superior to the soybean protein isolate in fat absorption with Chinese rapeseed SPI giving the
highest value. Both Chinese rapeseed protein isolates showed a excellent whippability. The foaming
properties of the soybean protein isolate were intermediate between those of the rapeseed meals and
isolates. All samples produced foams that were stable for more than 2 hours.

CONCLUSIONS

Aprocess for producing acid-soluble, food grade protein isolates from high-glucosinolate Chinese rapeseed
has been developed. China, and India, the largest rapeseed producers are chronically short of protein. A
soft drink containing acid soluble protein isolate could provide the same protein nutrition as milk, in a safe
source of water. It has the additional advantage to being digestible by many people that cannot tolerate
milk or soy proteins.

REFERENCES

[1] L.J.Rubin, L.L.Diosady and C.R.Phillips - "Solvent Extraction of Oil Bearing Seeds" U.S.
4,460,504 (1984)
[2] L.J.Rubin, L.L.Diosady, M.Naczk and M.Halfani - "The Alkanol-Ammonia-Water/Hexane
Treatment of Canola", Can.lnstFood Sci.TechnoI.J., 19:1, 57-61, (1986)
[3] Y-M. Tzeng, L.L. Diosady and L.J. Rubin - Production of canola protein materials by alkaline
extraction, precipitation, and membrane processing, J. Food Science, 55:41147-56,1990.
[4] M.Naczk, L.L.Diosady, and L.J.Rubin.-'The Functional Properties of Canola Meals Produced by
a Two-Phase Extraction System", J.Food Sci. 50, 1685-1689 (1985)
SEPARATION AND CONCENTRATION OF POLYUNSATURATED FATTY
ACIDS BY A COMBINED SYSTEM OF LIQUID-LIQUID EXTRACTION AND
MEMBRANE SEPARATION

YUKO SAHASHI, HIROTOSHI ISHIZUKA, SEll KOIKE*, and KAZUAKI SUZUKI*

Medical Reseach Lab., Nitto Denko Corp., Shimohozumi, Ibaraki, Osaka 567, Japan
*Fat & Food Lab., Asahi Denka Kogyo K.K., Higashi-ogu, Arakawa-ku, Tokyo 116, Japan

ABSTRACT

PUFAs-rich glycerides were separated from fish oil hydrolysate by membrane separation
system. This system is a combination of extraction using ethanol solution of 75 (v/v)%, and
separation using hydrophilic UF membrane. The results of PUFAs recovery in the system
agreed well with the expected results from analytical data of hydrolysate. It was also found that
our technique was more effective than fractional crystallization in PUFAs recovery from fish
oil. The quality of oil was not damaged throughout these processes.

INTRODUCTION

n-3 Polyunsaturated fatty acids (PUFAs) have recently attracted special interest because of
their physiological activities. It is reported that PUFAs in the form of glycerides are absorbed
to the small intestine better than those in the form of ethylester.
The aim of our work is to develop an economical separation technique, concentration and
purification of PUFAs in glycerides without damaging their qualities. The diagram of our
research, composed of lipase hydrolysis and membrane separation system, is shown in Fig. 1.
Fish oil was first selectively hydrolysed with lipase. The hydrolysate was mainly composed
of PUFAs-rich glycerides and other free fatty acids (FFAs). The glycerides and FFAs were
separated bymembrane separation technique, so that the PUFAs-rich glycerides were
separated. The utilized membrane separation system will further be discussed in this report.
 675

<\' t:8
Fish oil Hydrolysate

~m
Lipase Membrame PUFAs-rich
separation glycerides
hydrolysis
000
0000
000
• PUFAs (EPA, DHA, etc.)
o Others Other free
fatty acids

Figure 1. A combined system of selective hydrolysis with lipase and membrane

separation of PUFAs-rich glycerides.

MATERIALS AND METHODS

A schematic flowsheet of the membrane separation combined with solvent extraction is

shown in Fig. 2. Hydrolysate of fish oil was mixed with extraction solvent to extract only
FFAs. The mixture in form of O/W emulsion was forced by applying pressure to permeate
through a membrane. FFAs permeated through the membrane with solvent. Whearas
glycerides in oil phase were rejected by the membrane. The PUFAs-glycerides were
concentrated in concentrate. The solvent was recycled after passing through an ion-exchange
column to remove FFAs. Cross-flow filtration and rotating disk membrane systems were tested
in a large scale process.

Nz -,---------,
Solvent .__- - . . ,
~----+-(Recycle)

CD
PUFAs-rich
Solvent glycerides
.-'-..................., (Concentrate)
Fish oil hydrolysate
+
Solvent
FFAs
+
Solvent
(Permeate)

CD Reservoir (2) Membrane Q) Batch-type cell @ Ibn exchange column

Figure 2. Schematic flowsheet of a combined system liquid-liquid extraction and

membrane separation.
676
RESULTS AND DISCUSSION

Among various solvents, ethanol solution of 75 (v/v)% was most suitable for separation of
glycerides and FFAs. Glycerides content in raffmate increased to 98% when 20 ml solvent per
g-oil was used. Solvent extraction was combined with membrane separation technique. It was
found that hydrophilic UF membrane (M.W. cut-off=20, 000, Material: Polyimide), was most
suitable for separation of FFAs and glycerides, where almost all FF As permeated through the
membrane, and glycerides composed of mainly triglycerides retained in concentrate.
In hydrolysis ratio of 54%, PUFAs content in concentrate increased from 34% to 52%, and
PUFAs recovery yield from fish oil was 71 %. Relation between PUFAs recovery yield from
fish oil and PUFAs content in glycerides in our membrane separation was compared with that
in fractional crystallization method as shown in Fig.3. The composition of hydrolysate was
analyzed. PUFAs content in glycerides and PUFAs recovery yield were determined. The
closed circles and reverse triangles show the expected results from analytical data of
hydrolysate. After membrane separation was completed, the result was reasonable with a
little losses (shown as open reverse triangles). When PUFAs content was nearly 50%, PUFAs
recovery in our system increased above 2 times as large as that in fractional crystallization.
It was also found that utilizing a rotating disk membrane system in a large scale process was
effective for separation. The quality of the separated oil was good.

~
til
Q)

...
-0
Q)
50
u
>.
tlo
c:
CQ) 40
Co
U
til

~ 30
~ Y:P---~--~~--~--~--~--~
40 60 80 100

PUF As recovery (%)

• Hydrolysis ... Fractional crystallization

." Hydrolysate before membrane separation
v Hydrolysate after membrane separation
Figure 3. Relation between PUFAs recovery yield and PUFAs content in glycerides.

CONCLUSION

It was found that utilizing a combined system of liquid-liquid extraction and membrane
separation was effective for separation for PUFAs-rich glycerides without damage.
 REFINING VEGETABLE OILS BY MEMBRANE TECHNOLOGY

M.CHERYAN, L.P.RAMAN and N.RAJAGOPALAN

University of Illinois, Agricultural Bioprocess Laboratory
1302 W. Pennsylvania Avenue, Urbana IL 61801, USA

ABSTRACT

Membranes can solve some of the major drawbacks of current vegetable oil processing
methods. They can greatly enhance the efficiency of hexane recovery, combine de-
gumming and bleaching into one step, solve the pollution problems of alkali deacidific-
ation, and also be used to treat gas and vapor streams.

INTRODUCTION

Vegetable oil processing is usually conducted in three stages: Pre-processing; Extraction

(where the oil is recovered from the oilseeds); and Refining (where the extracted oil is
purified to remove the undesirable components before being packed for consumer use).
The purification steps include degumming, deacidification, bleaching and deodorization
which are presently accomplished by physical and chemical means.
There are several drawbacks to current oil processing technology: (a) The operations
are energy intensive. The soybean oil industry in the U.S. alone uses 2.2 trillion kilo-
joules of energy annually for the recovery of solvent and oil from the miscella. (b) Sig-
nificant amounts of neutral oil are lost during the purification steps, due to saponification
of tri-glycerides and occlusion of oil in the soapstock. Higher acidity results in higher
refining losses. Hence, it is often uneconomical to refine certain oils like rice bran oil,
which can have up to 40% free fatty acids (FFA) , with current technology. (c) Large
amounts of water and chemicals are utilized, which increases production costs consider-
ably, and (d) Several processing steps result in heavily contaminated discharges; e,g., the
soap stock splitting process results in a heavily polluting effluent.
MEMBRANES IN VEGETABLE OIL PROCESSING

Membrane technology can solve or reduce these problems, by exploiting the differences
in physical and chemical properties at a molecular level between various components of
the feed streams in an oil refinery. The earliest publication in this field is probably the
U.S. patent by Sen Gupta in 1978 [1] for a degumming process. Conceptually,
membranes can be used in almost all the stages of oil production and purification, such
as the recovery of the solvent used in extraction, removal of phospholipids (degumming),
refining (deacidification), bleachinglremoval of color pigments, treatment of effluents
678
and production of nitrogen for packaging (Figure 1). Except for nitrogen production by
gas separations (GS), we are unaware of any commercial membrane installations in the
vegetable oil industry.
Degumming
In the conventional method, oil is first separated from the miscella and then the crude oil
is degummed. In contrast, membrane degumming can be done with the miscella itself,
taking advantage of the amphiphilic nature of the phosphatides, which causes them to
form micelles in the miscella. The phospholipids-micelles act like macromolecules with
molecular weights exceeding 20,000 daltons. Thus, ultrafiltration membranes could be
employed to separate them from oil and hexane [1,2].
Bleaching
"Bleaching" refers primarily to the process of reducing the color of oil by adsorbing the
coloring components such as carotenoids and chlorophyll on to activated clay or carbon.
Some of the color compounds are removed in the membrane degumming process. An
economic analysis of the membrane bleaching process for a plant processing 250 tons of
oil per day indicates potential savings of $(US)730,000 annually [3]. The analysis is for
the bleaching process alone and possible additional savings from the membrane degum-
ming is not included.

Oilseed
Oilseed

~Lecithin
Q)
c:
'"x
Q)
I

•
Crude oil

~
crUdeOil

•t
Acid

Alkali Soap FFA

NF
FFA
•
¢
Effluents

b
Air

Refined Oil
~,
Refined oil Effluent-free
2 Packed Oil
discharge

Figure 1. Left: Conventional vegetable oil processing.

Right: Membrane applications in vegetable oil processing
 679
Solvent recoverv
If membranes could be used to pre-concentrate the miscella before being sent to evapora-
tors, it would reduce the energy needed for this process by 43% [4]. In addition, since
the volume that is being processed is reduced, the size of the evaporators can be re-
duced. This reduction in energy and capital costs is likely to make the alcohol process in
cottonseed extraction economically viable. However, recovery of hexane from oil-hexane
miscella is difficult [5] because very few membranes with the desirable separation char-
acteristics are stable in hexane [4].

Deacidification
Deacidification has the maximum economic impact on oil production and any
inefficiency in this process has a great bearing on any subsequent oil purifying or
modification processes. "Single" membrane systems [6,7], "two-membrane" and "three-
membrane" systems [8] have been proposed in recent years; each has its own advantages
and limitations and may require fundamental changes from current methods.

Other membrane applications

These include dewaxing [9], removal of metals [10], catalyst recovery from hydrogenat-
ed oil [11], production of nitrogen for packaging [12], membrane recovery system for
scrubbing solvent vapors [12] and treatment of frying oils for re-use [13].

REFERENCES

1. Sen Gupta, A.K., Purification process. U.S. Patent, 1978, 4,093,540.

2. Iwama, A. and Y. Kazuse., Process for the purification of crude glyceride oil
compositions. U.S. Patent, 1983,4,414,157.
3. Raman, L.P., M.S. Thesis, University of Illinois, 1993.
4. Cheryan,M., Raman ,L. P. and Rajagopalan,N., Solvent recovery in edible oil manu-
facture. Patent disclosure. University of Illinois, Urbana, 1992.
5. Koseoglu, S.S., Lawhon,lT. and Lusas,E.W., Membrane processing of crude veg-
etable oils: Pilot plant scale removal of solvent from oil miscellas. Journal of Ameri-
can Oil Chemist's Society, 1990, 67, 315-322.
6. Sen Gupta, A.K., Refining. U.S. Patent, 1985, 4,533,501.
7. Cheryan,M., Raman ,L. P. and Rajagopalan,N., Refined vegetable oil. Patent disclo-
sure. University of Illinois, Urbana, 1992.
8. Keurentjes,lT.F., Ph.D. Thesis, Agricultural University ofWageningen, 1991.
9. Mutoh, Y., Matsuda,K., Ohshima,M. and H. Ohuchi,H., Method of dewaxing a
vegetable oil. U.S. Patent, 1985,4,545,940.
10. Keurentjes,J.T.F., Bosklopper,T.G.l, van Drop, L.l and van't Riet,K., The rem-
oval of metals from edible oils by a membrane extraction procedure. Journal of
American Oil Chemist's Society, 1990,67, 28-32.
11. Koseog1u, S.S. and Varva,C.l, Catalyst removal from hydrogenated oil using
membrane processing. INFORM, 1992, 3, 536.
12. Mohr,C., Leeper,S.A. ,Engelgau,D.E. and Charboneau,B.L., Membrane applications
and research in food processing: An assessment,DOE Report DOE/ID 10210, 1988.
13. Koseoglu,S.S., Membrane processing of used frying oils. INFORM, 1991,2,334.

This research was supported by the Illinois Soybean Program Operating Board
and the Illinois Agricultural Experiment Station at Urbana-Champaign.
 MEMBRANE SEPARATION OF SUNFLOWER OIL HYDROLYSATES
IN ORGANIC SOLVENTS

Seiji Koike *, Masakatsu Yokoo, Hiroshi Nabetani, Mitsutoshi Nakajima.

National Food Res. Inst., MAFF; Tsukuba, Ibaraki 305, Japan,
*Asahi Denka Kogyo K.K.; Arakawa-ku, Tokyo 116, Japan

ABSTRACT

Various types of commercial membranes (18) were used to separate the constituents of oils
including triglycerides (TG), diglycerides (DG), monoglycerides (MG) and free fatty acids
(FFA) in organic solvents. Cellulose acetate (CA) membrane in ethanol and gas separation
membrane in hexane gave higher permeate fluxes, and large rejection differences between
solutes. When membrane separation of the oil-ethanol mixture using CA membrane (NTR-
1698) was repeated 16 times in batch mode (diafiltration mode, initial retentate weight 100g,
and total permeate weight 504g), the composition (TG:DG:MG:FFA) of retentate changed
from 31:28:9:32 to 65:29:2:4, respectively.

INTRODUCTION

Solvent recovery in extraction process and fractionation of fats and oils are industrially
energy consuming steps, and more economical processes such as membrane technology
are required. Recently, some researchers have reported that removal of phospholipids
(degumming) was possible by using ultrafiltration (UF) membranes. 1), 2) However, few
reports are available on the use of reverse osmosis (RO) membranes for these processes. In
this study, RO membranes were used to separate the constituents of oils, such as
triglycerides (TG), diglycerides (DG), monoglycerides (MG), and free fatty acids (FFA) in
organic solvents.

MATERIALS AND METHODS

1. Materials
Lipase hydrolysate of high oleic sunflower oil, in which more than 80% of the fatty acid
composition was oleic acid, was used as a lipid mixture for the membrane separation test.
The hydrolysate contained TG, DG, MG and FFA in the weight fraction of 31,28,9 and 32,
respectively.
 681
2. Membrane Separation Test
Membranes (<<1>75 mm, 32 cm 2 area) were evaluated using a flat membrane test cell (Nitto
Denko Corporation, C70-B). The testing unit was operated in batch mode by charging the
cell with 100 cm 3 of oil-solvent mixture and applying pressure as required by nitrogen gas.
The permeated oil-solvent mixture was collected through a port beneath the membrane
support, and the permeate flux was calculated. The compositions of FFA, MG, DG and TG
of initial and final retentates, and permeates were determined by TLC-FID method. 3) The
solvent in the permeate was evaporated, and then solutes weight was measured. From these
determinations the mass-balance was calculated. Each solute weight was corrected by parti-
tioning proportionally. Observed rejections were determined as follows, assuming that
observed rejections are constant during the experiment (Eq.1):

RS,obs =In(Cs / Cs,o) / In(Wo / W) (1)

where, CS,o and Cs are the initial and final concentrations (wt./wt. solvent %) of each solute
in the retentates, and Wo and Ware the initial and final solvent weights in the retentate,
respectively.

RESULTS AND DISCUSSION

1. Membrane screening
Each membrane was tested with ethanol or hexane as a solvent containing 6.67% (wt./wt.
ethanol) or 7.52% (wt./wt. hexane) oil. Results of the flux and the rejection using different
types of membranes are shown in Figures 1 and 2. In the case of oil-ethanol mixture, the
permeate fluxes were in the range of 1-13 g m-2 S-1. Cellulose acetate (CA) membranes
gave higher flux, and large difference of rejection among FFA, MG, DG and TG. Polyvinyl
alcohol, polyamide and polyether membranes yielded a relatively high rejection, but there
was a little difference between solutes.
Lower fluxes and rejections were usually observed in the hexane system compared to the

100 F' PE ~ o 100.-------------__~

-
o o
80 ~ gVA 0
- 80 PA
~

-
~ A
eft. PA '#. A A (
CA1
o <> GS
c: 60 o - 60 El
e c:
o
:;::
CA3 o o GS
<> CA2 :;::
o
CD
40 A u
CD
.~ 'G)
a: 20 o r:x:
~GS
o

Permeate Flux (g m-2 5-1 ) Permeate Flux (g m-2 5- 1 )

6. : TG, 0: DG, 0: MG, 0: FFA, 6.:TG, D:DG, O:MG, O:FFA,

CAl: SC-3000, CA2: NTR-1698, CA3 : DRC-97, CA: NTR-1698, PVAI : NTR-729HF,
A: UTC-70T, PVA: NTR-729HF, PE: PEC-lOOO, PVA2: NTR-7250, A: UTC-70T
PA: NTR-759HR, GS: NTGS-2l00 PA: NTR-759HR, GS: NTGS-2100

Figure 1. Permeate Flux and Rejection Figure 2. Permeate Flux and Rejection
in Ethanol System. in Hexane System.
682
ethanol system. The gas separation membrane gave the highest flux and high rejection (70.2% for
FFA, 94.4% for TG), the differences between substances, however, was not so remarkable
compared to the CA membrane. The CA membrane in ethanol system was therefore used in the
following experiments.

2. Effect of oil concentration on rejection and flux

Figure 3 shows the effect of oil concentration on rejection and permeate flux for the CA membrane.
The oil concentration per weight of
ethanol was changed from 6.7% to 1 00..-tl!r=__
==:;;vr-==......~r--T12 ,:-
84.4%. Permeate flux decreased if DG TG 'en

=:::::::::
according to oil concentration, perhaps 10 ~
80 \

~
due to the effect of osmotic pressure. TG
shows the highest rejection of the 4
~
0 ~ FA 8 ~
substances at all concentrations. 60
c::::
Rejection values of TG were higher than 0 MG 6 ~
:;:: LL
98%. The rejection of DG was 90% at
(.) 40
II)
4 II)

----'7
'Q)
low concentration, which was noticeably a: 20 ~ m
lower than that of TG. At concentrations
greater than 30%, rejection of DG
0 ~~~~~~~~~~~O
2 E..
~
increased and rejection values were
nearly equal to that of TG. Rejections of 0 20 40 60 80 100
MG and FFA were lower than DG and Wt I (Wt Solvent) %
TG, and varied in the range of 50% to Figure 3. The Effect of Oil Concentration on
70%. Permeate Flux and Rejection for CA Membrane.

3. Discontinuous diafiltration
Diafiltration was operated in batch mode by charging the cell with 100g of 10% (wt./wt.
ethanol) oil-solvent mixture. In one batch, about 35g of oil-solvent mixture permeated
through the membrane, and an equal weight of solvent was fed to the cell. This operation
was repeated 16 times (total permeate weight:504g). Permeate flux per batch decreased with
increasing oil concentration of retentate. Initial permeate flux in each batch almost
recovered by feeding solvent to the retentate. However, the flux gradually decreased with
the number of operation, perhaps due to membrane fouling. The fractions of MG and FFA
in the retentate decreased gradually, whereas the fraction of TG increased. Finally, the
composition (TG:DG:MG:FFA) of the retentate changed from 31:28:9:32 to 65:29:2:4,
respectively.

CONCLUSIONS

In this study, 18 types of commercial membranes were tested for their ability to separate the
constituents of oils. A cellulose acetate membrane with ethanol as solvent was found to be
the most suitable system for oil separation. From the results of discontinuous diafiltration,
the fractions of FFA and MG in the retentate decreased, whereas that of TG increased.

REFERENCES

1) AK.S. Gupta, FeU Seifen Anstrichm., 1986,88,79-86

2) A Iwama, b Jpn. Oil Chern. Soc. (Yukagaku), 1985, 10, 852-58
3) M. Tanaka, T. Itoh, and H. Kaneko, b Jpn. Oil Chern. Soc. (Yukagaku), 1976,25, 263-65
 CATALYST REMOVAL FROM HYDROGENATED OILS USING MEMBRANE TECHNOLOGY

CARL VAVRA and S. SEFA KOSEOGLU

Food Protein Research and Development Center
Texas A&M University System
College Station, Texas 77845

ABSTRACT

Nickel catalyst currently is recovered from hydrogenated oil-catalyst mixtures by high capacity filter
presses, but the process is highly labor-intensive. The use of membranes potentially could reduce
labor expenses and the amount of oil lost during catalyst recovery, and can extend the useful life
of catalysts. In the current study, catalyst ranging from 1 to 100 microns in particle size, did not
pass through the membrane. All the nickel catalyst was recovered in the retentate fraction,
resulting in a clear permeate stream. Experiments were performed over various pressure and
temperature ranges, with flow rates ranging from 3.5 to 119 Imh.

INTRODUCTION

Application of membrane technology to edible oil processing is expanding and is expected to make
a considerable impact on profitability and efficiency of the edible oil refineries (Koseoglu 1991;
Iwama, 1988: Gupta, 1986). A summary of the membranes evaluated in edible oil refining during
the last ten years is summarized in Table 1.

TABLE 1 . Examples of membranes evaluated in edible oil refining and their producers

Membrane Membrane Membrane

Producers Name Materials Form

Rhone-Poulenc (F) IRIS 3042 PAN Flat Sheet

PCI (GB) T6B/BX3 PAN/Polysulfone Tubular
SFEC (F) Carbosep Inorganic Tubular
Romicon (USA) PM 1,10, SM10 Polysulfone Hollow Fiber
Bergof (D) 8M 50, 100, 1000 Polyamide Hollow Fiber
Koch (USA) HFM 180 Polyvinylfloride Tubular
Nitto Danko (J) NTU 4220 Polyimide Tubular
Wafilin (NL) WFA 3010 Polyimide Tubular
684
Oil hydrogenation is usually performed in batch reactors and when hydrogenation is complete, the
catalyst is recovered by recirculating the oil through a high capacity filter until the oil is free of
nickel. The filter cake containing the catalyst, is recovered manually and can be reused a number
of times. After final use, the catalyst usually is trucked to another site for disposal. Exhausted
catalyst can be processed to recover its nickel content. However, due to concerns about leaching
of heavy metals into the soil and water, increased disposal costs, and tighter federal and state
regulations, alternative methods are being evaluated. For example, nickel catalyst is being
recovered from wastes by digestion with either organic or inorganic acids. The rate of recovery and
economics of the process are dependent on nickel concentration and oil content of the filter aid-
catalyst mixture. Therefore, its value is directly related to its nickel content.

EXPERIMENTAL

Several membranes, including: 1I ceramic; 21 carbon; and 31 polymeric materials, membranes were
evaluated in this project. The ceramic, PEl and Carbon coated Zirconia membranes were obtained
from Ceramem Corporation, Walton, MA, ROCHEM Separations, Torrence, CA and Rhone-Poulenc
Inc., Cranbury, NJ respectively. The ultrasonicly sealed ROCHEM membranes were packed in plate
and frame MF/UF module. The ceramic membranes were available in honeycomb monolithic
elements with 1.9 mm passage way. The Catalyst evaluation experiments were performed at
pressure range of 25-125 psi and temperature range of 55-100°C. Permeate flow rates were
measured, and samples were analyzed by Atomic Absorption Spectroscopy to determine the nickel
removal selectivity of the each membrane system. Membranes were cleaned according to
manufacturer's recommendations.

RESULTS AND DISCUSSION

All of the membranes, tested removed the catalyst from the hydrogenated soybean oil. Analyses
of the permeate, retentate and feed samples are summarized in Table 1. All of the membranes
were found to be effective for separating catalyst from the oil with reasonable flow rates.
Membrane CM2 produced permeate with 0.9 ppm nickel catalysts. However membrane RC also
performed well and resulted in a residual catalyst concentration of 9.4 ppm, which is negligible
since it can be easily removed during citric acid treatment and polishing filtration. The remaining
few ppm of nickel is usually in the form of nickel soaps that can not be filtered due to their
solubility in fats and oils. The citric acid or any chelating agents can react with nickel and make
them insoluble that could be removed with conventional dead end filters.

Flow rates were largely dependent on membrane pore size, pump speeds, temperature, and
pressure. One of the most important criteria in these experiments was maintaining a high
temperature during processing. Membranes CM2, NCM2, RP8, RP14 gave stable flow rates
ranging from 8.5 Imh to 42.5 Imh. The membrane RC showed a very large flux of 113 Imh.

TABLE 1. Analysis of permeates and retentates obtained from various commercial membrane
systems

Nickel Content, ppm

Membrane Company Membrane Type Feed Permeate Retentate

ROCHEM Separations (RCI PEl 0.01 micron 297 9.4 219

Ceramem Corporation (CM51 Ceramic 0.05 micron 178 2.2 210
Ceramem Corporation (CM21 Ceramic 0.2 micron 275 0.9 883
Ceramem Corporation (NCM21 New Ceramic 0.2 micron 656 8.0 1068
Rhone-Poulenc Inc. (RP81 Carbosep 0.08 micron 178 3.4 229
Rhone-Poulenc Inc. (RP141 Carbosep 0.14 micron 330 2.7 304
 685
Figure 1 shows typical graph of flux and percent rejection rate versus concentration factor. The
volume of the catalyst containing oil was reduced from 32 liters to 2.7 liters at the end of the
evaluations. The initial rejection rate of 82% was increased to 100 percent as the catalyst
concentration increased in the feed.

100 100

.
eo .-
_

80
-- ::IJ
.r; ~.
E 60 60 (I)
(")
::::t.
X 0
.2 .. ::J
....
u.. 40 .. 40
a
(I)

?fl.
20 20
-Flux
+ Aelecllon rate -, ..
0 0
1 2 3 4
Concentration Factor

Figure 1. Flux and percent rejection rate versus concentration factor.

The present study indicated that the hydrogenated oil containing catalyst could be reduced to
1/100 of the original volume. This reduction allows the processor to reuse the undiluted catalyst
for a longer period of time without selectivity changes that usually come from catalyst poisons
adsorbed on the surface of the catalyst and the filter aid. This process also eliminates or
considerably reduces oil loss~s, and could potentially eliminate use of the labor intensive filter
presses. The envisioned process could combine a cross flow membrane system with a very small
filter press, and can be run continuously with minimal labor cost. Further, the used catalyst
mixture would be more valuable to both refiners and to reclaimers of nickel. This work indicates
that membranes are capable of separating catalysts from hydrogenated soybean oils. Preliminary
experiments show that membranes are easy to clean and the original flux can be maintained.
Membrane life, durability, efficiency, and selectivity, and feedback received from the membrane
manufacturer, are being evaluated presently.

REFERENCES

1. Koseoglu. S.S., Membrane Technology in Edible Oil Refining. Oils,& Fats International,
1991,5,16-21.

2. Iwama, A., New Process for Purifying Soybean Oil by Membrane Separation and an
Economical Evaluation of the Process, In Proceedings of World Conference on
Biotechnology for the Fats and Oils Industry, Ed. T. H. Applewhite, Am. Oil Chem. Soc.,
Champaign, IL, 1988, pp. 244-250.

3. Gupta, A.K.S. Neuere Entwicklungen auf dem Gebiet der Raffination der Speiseole. Fette
Seifen Anstrichmittel, 1986, 3, 79-86.
 CONTINUOUS SYNTHESIS OF ASPARTAME PRECURSOR
WITH MEMBRANE ENZYME REACTOR - MEMBRANE EXTRACTOR SYSTEM

YASUYUKI ISONO*, HIROSHI NABETANI and MITSUTOSHI NAKAJIMA

National Food Research Institute, Ministry of Agriculture,
Forestry and Fisheries
* 2-1-2 Kannondai Tsukuba Ibaraki 305 Japan
Dainichiseika Color & Chemicals Mfg. Co., Ltd.
1-9-4 Horinouchi Adachi-ku Tokyo 123 Japan

ABSTRACT

An enzyme membrane reactor and a membrane extractor were used

simultaneously for the synthesis of N-(benzyloxycarbonyl)-
L-aspartyl-L-phenylalanine methyl ester (ZAPM), the precursor
of the artificial sweetener, aspartame. The synthesis of ZAPM
in the biphasic membrane reactor proceeded by an enzymatic
reaction between N-(benzyloxycarbonyl)-L-aspartic acid (ZA)
and L-phenylalanine methyl ester (PM) in the water phase at pH
5-6. The organic phase, where ZAPM and PM are extracted,
comes into contact with the water phase at pH 7 in the
membrane extractor. Because of this pH control, purification
of the ZAPM and recycling of substrates can be achieved. A
simUlation study was performed to confirm the suitability of
the system.

INTRODUCTION

Aspartame is a synthetic sweetener which is about 200 times as

sweet as sucrose. In recent years, several investigations on
enzymatic synthesis of the aspartame precursor (ZAPM) using by
the reverse reaction of hydrolysis have been reported [1-
3]. However, few studies for a purification of ZAPM or
reusing of PM combined with synthesis have been reported. In
order to reduce production costs, reuse of PM is necessary.
In our present study, continuous synthesis of ZAPM with
product purification and substrates recycle using biphasic
membrane enzyme reactor - biphasic membrane extractor system
is proposed. The performance of this system was modeled and
simulated.
 687
SYSTEM

A schematic flow chart of the membrane enzyme reactor -

membrane extractor system is shown in Figure 1. The reactor
consists of a water phase (O.05M acetic buffer saturated with
butyl acetate, pH 5 - 6) containing an enzyme (thermolysin)
and an organic phase (butyl acetate saturated with acetic
buffer). The substrates (ZA and PM) are supplied in the water
phase and the major part of the product of enzymatic reaction
(ZAPM) and some part of PM are extracted into the organic
phase. The organic phase is in contact with a water phase
(O.05M acetic buffer saturated with butyl acetate, pH 7).
About 75% of ZAPM is extracted to the water phase, however
almost PM remains in the organic phase. Therefore, this system
enables us to purify ZAPM and to reuse PM.

Re a c t 0 r Ex t r a c t 0 r
ZA Wa e r P has e (pH5-6)
t Wa t e r P has e (p H 7 )
PM

..}_...]_. .8_ ._...

ZA + PM ~ ZAPM ZA PM ZAPM
~ ~ r+
----j-----------l------------------l-------
~ ZA PM ZAPM ~ ZA PM ZAPM ~

or g ani c P has e Org a n i c Ph a s e

.i

Figure 1. A schematic flow chart of the system

SIMULATION

A simulation of this system was performed on the assumption as

follows: 1. Each phase is completely mixed, 2. Extraction
equilibrium states exist between the water phase and the
organic phase. The following equations represent the system:

Enzyme reaction rate in water phase

k2[Eo] [ZA] [PM]-k-2[W] [Eo] [ZPAM]KmzA.KmPM/KmzAPM
= (1)
(KmZA+ [ZA] ) (KmPM+ [PM] + [ZA] KmPM/KizA) + [ZAPM] KmZA' KmPM/KmzAPM

where k2 and k_2 are the reaction rate constants of the

forward and reverse reactions, Km is the Michaelis constant,
Ki is the inhibi tion constant of ZA, Eo is enzyme
concentration, and [ZA], [PM], [ZAPM] are the concentrations
of ZA, PM, and ZAPM in the water phase, respectively.
688
1+10PH-pK1+102pH-pK1-pK2
Partition coefficient of ZA (2)
Knon
Where the Knon is the partition coefficient of the non-ionized
form of ZA. Partition coefficients of PM and ZAPM are similar.
From the mass balance equations and partitional balance,
the process was simulated.

RESULTS AND DISCUSSION

Figure 2 shows the calculated equilibrium concentrations of

each material where the pH of the reactor is five and the pH
of the extractor is seven. High conversion and similar
productivity was obtained compared to the literature, because
of the PM recycling with extractor. The extractor combined
with reactor is effective for the reaction. Further research
on the enzyme activity and long term stability is to be
investigated.

ZA PM ZAPM ZA PM ZAPM

~ WI 17. 2 42. 9 O. 2
........ -.. -. __ ..... _----_ .... __ ... _--_ ... _------
---") W2 O. 9 O. 9 10. Of--7
........\ _-_._---_. __ .......... _-_ ......... _-_. __ .
~ 01 O. 3 10. 1 7. 1~ O2 O. 0 9. 8 3. 8f-
Reactor Extractor
(rnM)
Conversion = 91. 5 (%)
Productivity = 138.8 (xl0-&ol/(m3,s))

Figure 2. Calculated equilibrium concentrations

REFERENCES

1. Nakanishi, K., Kimura, Y. and Matsuno, R., Kinetics and

equilibrium of enzymatic synthesis of peptides
inaqueousjorganic biphasic system. Eur. J. Biochem., 1986,
161, 541-549

2. Oyama, K., Irino, S. and Hagi, N., Production of aspartame

by immobilized thermoase. Methods in Enzymology, 1987, 136,
503-516
3. Hirata, A., Hirata, M. and Furuzawa, H., Production of
Aspartame Precursor by Semi-Continuous Pulsed Extraction
Column Bioreactor Retaining Free Enzyme in an Aqueous
Phase. Kagaku Kogaku Ronbunshu, 1991, 17(3), 586-588
 ELECTRO-ULTRAFILTRATION BIOREACTOR FOR ENZYMATIC
REACTION IN REVERSE-MICELLES

MASARU HAKODA, YOSHIHIRO OGAWA, TATSUYA AKASHI

AND KOZO NAKAMURA
Department of Biological and Chemical Engineering
Gunma University, Kiryu, Gunma 376, Japan

ABSTRACT

Electro-ultrafiltration (EUF) was applied to separate the Aerosol OT (AOT) reverse micelles
containing lipase from isooctane, and the effect of electric field on the rejection of reverse
micelles and the permeation flux was examined experimentally. It could be said from the
experimental results that EUF was effective for separation of the reverse micelles containing the
enzyme and that the motive forces for the negatively charged reverse micelles were
electrophoretic and dielectrophoretic.

INTRODUCTION

EUF is effective in decreasing the gel layer formation and in increasing the filtration flux,
owing to the electrokinetic phenomena such as electrophoresis and e1ectroosmosis[1,2]. We
applied EUF to the enzymatic reaction of starch hydrolysis and reported the improvement of the
permeation flux with some damage of glucoamylase in the electric field[3]. We reported also the
experimental data of UF and EUF separation of the reverse micelles containing enzyme[4]. In
this work EUF was further applied to separation of the Aerosol OT (AOT) reverse micelles
containing lipase in isooctane, and the effect of electric field on the rejection of reverse micelles
and the permeation flux was examined in detail experimentally.

MATERIALS AND METHODS

The experimental apparatus consisted of a EUF module, a feed tank, a feed pump and a
D.C.electric power supply unit etc. The Carbosep M5 membrane (cutoff M.W. 20,0(0) are
made of carbon with the skin layer of zirconium oxide. A stainless steel rod of 3mm diameter
690
was installed as an electrode in the central axis of the tubular module (inner diameter 6 mm,
thickness 2 mm, Sumitomo Heavy Industries Ltd., Japan). The porous substrate of the
membrane was made of sintered carbon and used as a counter electrode. The surfactant AOT
used was the reagent of Nacalai Tesque, INC., Japan (purity 98.8%) and it was used without
further purification. The lipase used (600U/mg, MW 44,(00) was purchased from Seikagaku
Kogyo Co., Ltd .. All experiments were carried out by the total circulation method under the
condition of temperature 310K and trans-membrane pressure 5OkPa. The rejection, R, is
defined as follows;

R=I-(Cp I Cb) (I)

where Cp and Cb are the concentrations of solute in the permeate solution and in the feed
solution, respectively. It is noted here that the majority of reverse micelles does not contain the
enzyme under the condition of this experiment and their size and separation property could be
different from those of the reverse micelles containing the enzyme.

RESULTS AND DISCUSSION

The EUF of the reversed micelles containing lipase was experimentally investigated with three
different modes of voltage application.
Effect of AOT concentration
Fig. 1 shows the effect of AOf concentration on the permeation flux for three modes of voltage
application. In either mode the flux increased with increase in AOf concentration, probably due
to decrease in the size of reverse micelles. The application of voltage improved the flux, and
this result can be well explained if the approach of the reverse micelles containing enzyme to the
membrane surface is electrically hindered. The flux improvement was obtained irrespective of
the direction of applied voltage, probably because the dielectrophoresis could occur in the
nonuniform electric field.
Effect of Water concentration
Fig.2(a) shows the effect of water concentration on the permeation flux in three modes of
voltage application. The flux decreased in any mode with increase in the water concentration,

Water = O.25M
~ Enzyme = 2.84 " M
"'"e 2.0
Voltage = lOOV
Pressure = 50kPa
:::::' Temp. = 3 \OK
).
~
....
X
~
1;:1 1.0
,g= ~oo
8 IPe~tion SIde

~ 6 0 No voltage
/ 0 Cathode
6 Anode
0.0
0.0 0.1 0.2
Concentration of AOT [M]

Figure 1. Effect of AOT concentration on permeation flux.

 691
and the application of electricity improved the flux. When the anode was set in the permeation
side, the flux was larger than that obtained by the other modes. Fig.2(b) shows the effect of the
water concentration on the rejection of water. The rejection of water increased with increase in
the water concentration in the case of no voltage application, probably due to the increasing size
of reverse micelles. When the anode was set in the permeate side, the rejection of water
decreased with increase in the water concentration. The effect of the water concentration were,
however, reversed by setting the cathode in the permeate side. These results could be well
explained with the assumptions; (1) The increase in the water concentration contributes to
increase the electrophoretic force and to decrease the dielectrophoretic force. (2)The reverse
micelles as well as the free AOT molecules possess the negative charges.

5
100
Permeation Aar=O.~M
side Enzyme =2.84 Jl M
=
~

~ 0 No Voltage Voltage lOOV

! 4 =
Pressure 50kPa 80
..... ~ Cathode
Anode Temp. =3l0K
~
~
.!

-
100

'0 3 .s
x ...~
0
60

~ c0
c 2 '::I 40
c<:) &!
'::I 'a;' Permeation Aar=O.~M
gj ~
o No voltage
.~ =
Enzyme 2.84 Jl M
E 1 20 =
Voltage lOOV
ct o Cathode Pressure = 50kPa
t:. Anode Temp. = 310K

o~~~--~~~--~~~~~ o~~~--~~--~~~--~~~

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Concentration of water [M] Concentration of water [M]
(a) permeation flux (b) rejection
Figure 2. Effects of water concentration on permeation flux and rejection

CONCLUSIONS

1) The application of voltage improved the permeation flux and also the rejection of reversed
micelles in the membrane separation.
2) The voltage gradient set against the direction of permeation improved the flux more than the
reversed gradient.
3) The reverse micelles were supposed to possess the negative charges and to move by the
dielectrophoretic force as well as the electrophoretic force.

REFERENCES

1. Yukawa,H., Obuchi,H., Kobayashi,K., Kagaku Kougaku Ronbunshu, 1980,6, 288-293.

2. Kimura,S. and Nomura,T., Membrane, 1982,7,245-250.
3. Hakoda,M., Chiba,T., Nakamura,K., Kagaku Kougaku ROllbunshu,I991, 17,470-476.
4. Nakamura,K. and Hakoda, M., (eds., Furusaki,S., Endo,l., Matsuno,R.), Biochemical
Engineering/or 2001, Springer-Verlag., 1992,433-436.
 EVALUATION OF THE INTEGRATED-TIME-TEMPERATURE EFFECT OF THERMAL
PROCESSES ON FOODS:
STATE OF THE ART

HENDRICKX, M., MAESMANS, G., DE CORDT, S. AND TOBBACK, P.

Katholieke Universiteit Leuven, Faculty of Agricultural Sciences, Center for Food
Science and Technology, Unit Food Preservation
Kardinaal Mercierlaan 92, 8-3001 Leuven, 8elgium

ABSTRACT

The need for wireless Time-Temperature Integrators as thermal process evaluation tools
to overcome inherent disadvantages of in situ and physical-mathematical process
impact quantification techniques is indicated. A critical overview of existing TTl's is
given. Possibilities of an enzyme-based TTl in heat penetration studies and in monitoring
the impact of processing conditions are illustrated by a case-study.

INTRODUCTION

The widespread use of thermal treatments to eliminate enzymes, (pathogenic)

microorganisms and induce palatability makes proper quality and safety quantification
techniques for this physical food preservation variant of paramount importance for the
nutritional well-being of consumers. The prevalent concept to quantify the integrated
impact of a heat process on a food quality attribute is the processing value F.
Approaches to obtain this processing value can be classified in (i) in situ evaluation, (ii)
physical-mathematical techniques and (iii) by using Time-Temperature-Integrators.
The in situ thermal process evaluation technique relies on direct determination of
the change in a food quality attribute concentration. Disadvantages associated with this
methodology are the large sample size required, isolation and analysis time and
detection limits. Only when information on temperature profile shape and maximum
centre temperature are available, the effect of a previous heat treatment on food safety
can be monitored by the changing concentration profile of innate chemical and
biochemical markers.
The physical-mathematical method is based on determining the temperature
program imposed on a food from direct registration or constructive computation. This
approach is limited by difficulties in measurement of the temperature profile, especially
for heterogeneous foods (rotating retorts, aseptic) and/or lack of accurate input
parameters and models for the constructive computation option.
Time-Temperature-I ntegrators (TTl's) have been suggested as alternative
evaluation technique. A TTl can be defined as 'a small device that shows a time-
 693
temperature dependent, easily and correctly measurable, irreversible change that mimics
the changes of a target quality parameter undergoing the same variable temperature
exposure' [7,2]. Theoretical analysis [8] indicates that TTl and target shall exhibit the
same temperature dependency coefficient (activation energy Ea or z-value).

TTl'S: STATE OF THE ART

Based on the nature of the temperature dependent phenomenon, TTl's can be

subdivided in biological, chemical and physical sensing systems. Examples of each
category are listed in Table 1.
The use of microbiological systems to monitor efficacy of a thermal treatment is
the most endeavoured application of TTl's in food and pharmaceutical industry.
Different tools have been designed to control location and (contamination-free) recovery
of microorganisms or their spores at a specified site in the food product instead of
dispersing it in the food or drug (inoculated pack). With the advent of aseptic processing
technologies for heterogeneous foods, carrier systems for microbiological TTl's have
been miniaturized. To avoid complicated quantitative microbiology, also enzymic TTl's
were proposed.
Temperature dependent diffusion is the basis of physical TTl's described thus far
in literature.
Detection of the concentration change of a chemical compound as a measure for
the impact of a thermal process was advocated two decades ago [5]. Chemical
compounds have been used as single or mUlti-component TTl before, but often fail in
application because of abuse: no generally valid technique is yet available to correlate
the change in status of one or more TTl-component to a change in target quality
attribute when activation energy (or z-value) of TTl and target are different [4].

TABLE 1. Examples of TTl's

Time Temperature Integrator Temperature coefficient Reference

Micro-sized Microbiological TTl's

B. stearothermophilus 10°C (z) [3]
C. sporogenes 12.5-12.7°C(z) [1]
Immobilized peroxidase in dodecane 1 aoc [8]
Chemical TTl's
Thiamine 48°C (z) [5]
Blue#2 degradation 58.2-74.5 kJ/mol (Ea) [6]
Physical TTl's
Coloured compound diffusion 291.2 kJ/mol [9]

THERMAL PROCESS EVALUATION TECHNIQUES BASED ON ENZYMIC TTl'S

Recent work on peroxidase [8] and a-amylase [2] indicates that control of state
(immobilization) and micro-environment (solvent engineering) of the enzyme is a
promising basis for TTl development. Figure 1 illustrates application of an enzyme-based
TTl in determining the coldest particle without restricting its free rotation/translation and
studying the influence of end-over-end rotation for a heterogeneous food model system.
a-amylase (z = 7°C, D 100 c = 13 min) was encapsulated in the central cavity (2.5 mm
0
694
diameter) of a 25 mm diameter teflon sphere. Ten of these TTl-carrier systems were
spatially distributed among 22 others in a 600 ml glass jar. The coldest point shifts from
the lower part of the can (particle 2) in the zero rotation mode to the centre of the
container (particle 8). The TTl response clearly reflects the homogenized and enhanced
heat transfer rates induced by rotation.

F CONCLUSION
35
Time Temperature Integrators can
30
contribute substantially to our
25 understanding of thermal treatment
20 efficacy, especially in processing
conditions where in situ or physical-
15
mathematical evaluation techniques are
10 restricted. The basic requirement of
- 7--- 20
5 proper TTl-functioning, i.e. matching
10
1
hM activation energy of TTl and target, is
5
often overlooked. The search for a non-
3 o microbiological TTl to monitor
PARTICLE sterilization processes continues.

Figure 1 : Spatial processing value distribution (7°CF,00 ocJ in a 600 ml glass jar as a
function of the End-over-End rotational speed, measured with a-amylase as TTl. The
cascading water heat treatment consisted of a C.V. T. of 8'06", 30'05" holding time at
105°C and 20' cooling time to 15°C. Silicone oil (350 CST) was used as brine and
the solid/liquid ratio was 1. 7 .

REFERENCES

Brown, K.L., Ayres, C.A., Gaze, J.E., Newman, M.E. (1984). Thermal destruction of bacterial
spores immobilized in food/Alginate particles. Food Microbiology, 1, pp 187-198.
2 De Cordt, S. ,Hendrickx., M., Maesmans, G., Tobback, P. (1992). Immobilized alfa-amylase
for Bacillus licheniformis: a potential enzymic time-temperature integrator for thermal
processing. International Journal of Food Science and Technology, 27, pp 661-673.
3 Hersom, A.C., Shore, D.T. (1981). Aseptic processing of foods comprising sauce and solids.
Food Technology, 35(4), pp 53-62.
4 Maesmans, G., Hendrickx, M., De Cordt, S. and Tobback, P. (1993) MUlti-component Time
Temperature Integrators in the evaluation of thermal processes. Accepted for publication in
Journal of Food Processing and Preservation.
5 Mulley, A., Stumbo, C., Hunting, W. (1975). Thiamine: a chemical index of the sterilization
efficacy of thermal processing. Journal of Food Science, 40(5), pp 993-996.
6 Sadeghi, F., Swartzel, K.R. (1990). Time-temperature equivalence of discrete particles during
thermal processing. Journal of Food Science, 55(6), pp 1696-1698, 1739.
7 Taoukis, P.S., Labuza, T.P. (1989). Applicability of time-temperature indicators as shelf life
monitors of food products. Journal of Food Science, 54(4), pp 783-788.
8 Tobback, P., Hendrickx, M.E., Weng, Z., Maesmans, G.J., De Cordt, S.V. (1992). The use of
immobilized enzymes as Time-Temperature Indicator system in thermal processing. In
'Advances in Food Engineering', Eds. Singh, R.P. and Wirakartakusumah, M.A., CRC Press,
Boca Raton, Florida, USA, pp 561-574.
9 Wittonsky, R.J. (1977). A new tool for the validation of the sterilization of parenterals.
Bulletin of the Parenteral Drug Association, 11 (6), pp 274-281.
 INACTIVATION KINETICS OF BACTERIAL SPORES IN SIOM MILK,
CONCENTRATES, IN DEPENDENCE ON WATER ACTIVITIES AND UNDER
SEALINGS

H.G. KESSLER, R. BEHRINGER! and 1. PFEIFER2

Institute for Dairy Science and Food Process Engineering
Technical University Munich, D-85350 Freising-Weihenstephan, Germany
INestec, Lausanne; 2Kraft General Foods, Munich

ABSTRACT

The heat resistance of mesophilic spores increased as the concentration of skim milk was raised
up to 40 % total solids. The heat resistance ofthermophilic spores, however, decreased
considerably in concentrated products. This suggests that the product concentration stabilized
the mesophilic mixed flora and destabilized the thermophilic flora. Decimal reduction times of
Bacillus spores were determined in a wide range of relative humidities and air temperatures.
A maximum of heat resistance was found at relative humidities from 20 % to 40 %. Very
interesting results could be found with microorganisms under seaIings, where the heat
resistance was clearly higher.

1 Inactivation of Bacterial Spores in Skim milk (SM) and Skim milk Concentrates (SMC)

Our first objective was to find out which heat treatments are required to destroy the
mesophilic and the thermophilic spores ofthe innate flora in SM and SMC (40 % TS). The
mesophilic flora was by a factor of 1.5 more resistant in SMC than in SM. Surprisingly, the
thermophilic flora was approx. threefold less resistant in the concentrate. Four species of the
Genus Bacillus were isolated and SM and SMC were inoculated with spores of pure cultures
to find out which of these species exhibit a behaviour similar to the innate flora. The
ARRHENIUS plot of the (first order) rate constants (Fig. 1) reveals that B. stearothermo-
philus is the only species which is less resistant in 8MC than in SM, whereas all other spores
were more resistant in the concentrate. Further investigations were done to establish which
factors are responsible for the particular behaviour. The value of each factor like pH-value,
lactose, calcium, phosphate and proteins was separately changed from the 8M level to the
SMC level. It was concluded that in the case of B. stearothermophilus the destabilizing effect
696
of the reduced pH-value and the increased calcium-phosphate concentration predominate,
whereas in the case of all other sporeformers the reduced water activity is the dominant factor.

I
0\
SM \
SMC ~ac.

10- 1 I--_S_M_~ \_\_ch+le_ni_fo_rm_is_ _ _---1

7111

-"
c0
Vi
c
8
.$
S\~c\;; >I B,c '
e 10-'
.r.

\ \ ~ ::,'1'", .\
0QI
"0

\\C\ Bac,
stearothermophilus
\
1O-3 '---'-1_--L.._ _" - - I _ " - - I _ I " - - - - - '
125 120 115 110 105 100
temperature. "C

I i i
I i i i i
I i i
I i

2,50 2,55 2,60 2,65

103/T • K-1

Figure 1. ARRHENIUS plot of death rate constants of the destruction of Bacillus spores in
skim milk and skim milk concentrate.

2 Inactivation of Spores as a Function of the Relative Humidity of Hot Air

Investigations with Bacillus cereus spores were done covering the whole range from 1 % to
100 % r.h.. A special heating device was built up. Hot air of well-defined temperature and
relative humidity was applied as the heating medium. Subsequently the relative humidity of the
hot air was step by step reduced down to 1 % r.h. (Fig. 2). The relative humidity influenced the
frequency factor of the ARRHENIUS equation as well as the activation energy. As a result
minimal rate constants and concurrently a maximal resistance occurs in the range from 20 % to
40 % r.h.. This behaviour is ascribed to the occurence of two separate inactivation
mechanisms.
 697

Bacillus cereus spores \ 0,

,......, 10- I
..
1
8.0
~-?
~ ,'1 0

~
en -':'1
~
~o
. ' '" ~
~ ~
"-.s
t:."-.
~7
A ~
\J
00

K
~jj~-.~"--~~\ti"
.:x.
I- 10-2
+'
c
0
~. 9:>

."• ~0
I" -\'e, . . . '8\.~
~~

r' -:f-9-..
°
+'
en
c
10-3
0
u ~ .~ - - ~-J ~5.fO-~
'1"0 ~1 I e ~~ ~ +
,~ ~~
.....0
Q)
•
L
00
EB ~ 0

10-1. r.h. = 4-0 % _60~

I I 0

2.30 2.40 2.50 2.60 2.70 2.80

1/T [1O-3 K-1 ]
I I I I I
160 140 120 100 80
temperature 19 [OC]

Figure 2. ARRHENIUS plot of the rate constants of the inactivation of Bacillus cereus spores
at various relative humidities

3 Heat Resistance of Spores Under Sealings

Components with sealings are often regarded as critical points. Therefore, we investigated the
heat resistance of bacterial spores, which are enclosed between a sealing surface of stainless
steel and sealings of different materials. The experiments revealed that the resistance of spores
under sealings strongly depends on properties of the sealing material, and that spores under
sealings may withstand a heat treatment much longer than in an aqueous environment. The rate
constants of experiments at various temperatures were compared with the rate constants
depending on the relative humidity. No deviations from the 100 % r.h.-line was observed with
the natural rubber sealings. The experimental points of the silicone rubber sealings, however,
moved towards lower relative humidities at higher temperatures resulting in a 20fold increase
in heat resistance. After these findings the environment between the sealing surface and the
sealing material was characterized by direct measurement of the water activity. The water
activity of the sealing materials, which were placed in a hermetically sealed chamber, was
recorded as a function of the temperature. The water content of the sealings remained almost
constant. It was found that the water activity of the natural rubber was practically independent
of the temperature. By contrast the water activity of the silicone rubber continuously decreased
as the temperature was raised. Consequently the high resistance under the silicone rubber
sealings is due to the reduced water activity.
DETERMINATION OF THE THERMAL DEATH KINETICS OF BACTERIAL
SPORES BY INDIRECT HEATING METHODS

JORGEN HAAS+, RAINER BOLTERMANN*, HELMAR SCHUBERT*

+Oepartment of Biotechnological Production, Dr. Karl Thomae GmbH,
Birkendorfer Str. 65, 0 - 88400 Biberach, Germany
*
Institute of Food Process Engineering, University of Karlsruhe,
Kaiserstr. 12, D - 76128 Karlsruhe, Germany

ABSTRACT

The thermal death kinetics of bacterial spores were investigated by two indirect
heating methods. In the lower temperature range a batchwise capillary tube method
was used. Theoretical and experimental results have shown a negligible lethal
effect during heating up and cooling down (quasi-isothermal conditions) for
temperatures of up to 135 ac. For investigations at higher temperatures a
continuous capillary tube method (turbulent flow conditions in a tube of an inner
diameter of 0.2 mm) was developed. This device allows to realize quasi-isothermal
conditions up to 170 ac. In the temperature range of 135 - 155 ac experiments
show, that for short holding times Bacillus stearothermophilus spores are mainly
activated rather than inactivated. A reaction model to describe this effect is
discussed and compared to the classical First-Order Reaction kinetics.

INTRODUCTION

The quality of food preserved by ultra-high-temperature- (UHT-) treatment might be

increased by applying even higher temperatures for shorter times, so-called
extreme-high-temperature- (EHT-) treatment. Therefore, in a first step, the thermal
death kinetics of bacterial spores must also be investigated in the EHT-range.
For an accurate examination of the thermal death kinetics so called quasi-
isothermal conditions are necessary. This means that the survival-rate 10glO (NINo)
during heating up and cooling down is greater than - 0.1 (N = number of spores,
No = number of spores at time t = 0) and can therefore be neglected.
 699
MATERIALS AND METHODS

In the sub-UHT-range inactivation experiments in a capillary tube of 1.4 mm inner

diameter were carried out batchwise. The tube is immersed into the heating, the
holding and the cooling bath by a computer controlled pneumatic mechanism [1].
To realize quasi-isothermal inactivation conditions in the UHT- and EHT-range
too the spore suspension was heated up and cooled down in tubes of an inner
diameter of 0.2 mm under turbulent flow conditions [1].

RESULTS AND DISCUSSION

Experimental survival curves of Bacillus stearothermophilus spores are shown in

figure 1. In the batch capillary tube the spores were heated up to 135°C.
Experiments up to 155 °C were carried out in the flow-through capillary tube. To
show all results in one figure a logarithmic time scale is used.

temperature/°C 155 150 145 140 135 130 125 121

-5+-~~~~~~~~~~~~--~~~r-~~~

0,01 0,1 1 10 100 1000

time I s

Figure 1. Survival curves of Bacillus stearothermophilus spores (Merck, No. 11499)

in physiological NaCI-solution. Symbols are experimental results, lines are
calculated after evaluation with the Activation-Inactivation model.

For short holding times positive inactivation-rates were obtained. This phenomenon
cannot be described with the classical First-Order-Inactivation model but with the
Activation-Inactivation-model [2]. There can coexist dormant spores (So) and
700
activated spores (8 A)' The dormant spores have to be activated first (by heat
treatment) before they underlie inactivation (inactivated spores 8 1):

8 D -----> 8 A -----> 8 1 (1)

The results of both models are compared in figure 2 for survival-rates < -4. In
addition curves for the undesired degradation of thiamine are shown [3].

i--_+:-'~t--"-::: ~~-C~:i~-~-i---­
N = 10-4No

102 -- - -

-
Q)
E
~O.~ ~_ -:t-··_···_·t··_·· . ~ - -;...= ';'-3 % l.~....._....
- - O. ~ % [- - - T- - /)
1 0° --·~·---·-·-:r··-::·-,,·--t·""··:.;··:··~·:::-~-:·- ::1~' _ ... ..j.. ........ ~ ... .....
=
- - thiamine-degradation
spores-inactivation EHT-range !......... .
-Inact.-model
-Act.-Inac.-model
1 0-2~==r===;:==r==~--+----,i------i---+
120 125 1 30 135 1 40 1 45 1 50 1 55 1 60
temperature / °C

Figure 2. Comparison of the Inactivation and the Activation-Inactivation model.

For survival rates < -1 the Activation-Inact. model predicts shorter times than the
First-Order-Inactivation model to obtain the same inactivation of B. stearothermophi-
Ius spores [1]. With both models destruction of the spores is much more dependant
on temperature than the chemical degradation reactions are. 80 heat treatment in
the EHT-range might provide an even higher food quality than UHT-treatment.
Theoretical considerations have shown that quasi-isothermal conditions are
obtained up to 170°C using the flow-through capillary tube plant [1].

Acknowledgement: The authors thank the Deutsche Forschungsgemeinschaft for

their financial support.

REFERENCES
1. Haas, J., Untersuchungen zur thermischen Abt6tung hitzeresistenter Bakteri-
ensporen. Dissertation, University of Karlsruhe, 1992.
2. Abraham, G., Debray, E., Candau, Y., Piar, G., Mathematic model of thermal
destruction of Bacillus stearothermophilus spores. Applied and Environmental
Microbiology, 6 (1990), 3073 - 3080.
3. Kessler, H.G., Lebensmittel- und Bioverfahrenstechnik - Molkereitechnologie.
A. Kessler, Freising, 1988.
 A FIRST ORDER PROBABILISTIC PERTURBATION ANALYSIS OF THE
GROWTH AND THERMAL INACTIVATION OF LACTOBACILLUS CELLS
DURING COLD STORAGE AND REHEATING OF LASAGNA

BART M. NICOLAI, JAN F. VAN IMPEl & JOSSE DE BAERDEMAEKER

Agricultural Engineering Department, K.U.Leuven, Kardinaal Mercierlaan 92, B-3001 Heverlee,
Belgium

ABSTRACT

A first order probabilistic perturbation method was developed for the analysis of microbial growth and
inactivation as a consequence of conductive heat transfer under random conditions. The algorithm was
used to investigate the effect of random variable refrigerator and oven temperatures on the growth and
inactivation of Lactobacillus cells during cold storage and reheating of lasagna.

INTRODUCTION

Numerical simulation techniques are believed to provide a powerful tool for the design and analysis offood
processes involving several consecutive process steps of heating, cooling, and cold storage. In the case of
solid-like foods such as lasagna, an Italian dish composed of alternating pasta and tomato sauce layers,
this involves the solution of the heat conduction equation together with an appropriate model for microbial
growth and inactivation. The simulations are usually carried out under the assumption of well-known
values of the model parameters. Because of the very nonlinear temperature dependency of the microbial
growth as well as the inactivation, it can be expected that unpredictable disturbances of these parameters
may cause large variabilities on the shelf-life of the food. In this paper, a probabilistic perturbation
method is developed to investigate the effect of random variable model parameter uncertainties on the
microbial load in the product during the consecutive process steps.

METHOD

Heat conduction in solid foods is governed by the Fourier equation with appropriate initial and boundary
conditions and is often computed using the finite element method [1]. In the finite element method, the
continuum is subdivided in (finite) elements of variable size and shape which are interconnected in a
finite number m of nodal points. After spatial discretization and numerical solution of the corresponding
differential system, an approximation for the temperature at every node and time step is found.
Recently, a dynamic model was described for the prediction of both microbial growth and inactivation
in varying temperature conditions [2]. The model consists of two differential equations in two dependent
variables. One variable is the microbial load Y =
log(N), with N the number of organisms per unit
volume; the other variable Yr can be considered as a reference level from which growth restarts after
thermal inactivation. The model equations corresponding to the temperature history T at a specific
place in the food can be written concisely as follows:

y = g(y, T) (1)

with y = [ Y Yr f and g a 2 x 1 matrix function.

The propagation of random variable parameter uncertainties through the heat transfer process and
the associated microbial kinetics can be investigated using Monte Carlo simulation techniques [3]. The
method relies on the generation of a considerable number (e.g. 1000) of pseudo-random samples of the
lSenior research assistant with the N.F.W.O. (Belgian National Fund for Scientific Research)
702
random parameter vector by means of a computer. The model equations are solved for each sample
parameter vector. Finally, the statistical characteristics (mean value, variance, ... ) of the temperature
and microbial load at an arbitrary place in the food can be estimated for each time step by means of the
common statistical procedures. The Monte Carlo method has a very high computational burden.
Probabilistic perturbation methods for heat conduction problems with random variable thermophys-
ical parameters [4] and other process parameters [5] have been shown to be a fast alternative to the
Monte Carlo method and are very well suited for incorporation in computer aided food process design
software [6]. The perturbation method can be applied to the microbial kinetics as follows. It is assumed
that T is random due to random variable parameters of the heat transfer model and that its mean T
and variance V T,T can be computed from the mean values and covariances of the random parameters by
means of the algorithms outlined in ref. [4-5]. Further, the parameter vector PM of the microbial model
and the initial microbial condition Yo are assumed to be subjected to random variable disturbances. It
is assumed that the mean vectors PM and Yo and the covariance matrices VPM,PM and VYo,yo of PM
and Yo, respectively, are known.
First, the dependency ofy on T, PM, and Yo can be formally explicited. A first order Taylor expansion
of y(T, PM, yo) around the solution y(T, PM' Yo) yields

y = y + -~T+
~- {}y
aT
{}y {}y
--~PM + -~Yo
{}PM {}Yo
(2)

in which the partial derivatives must be evaluated using T, PM, and Yo. Similarly, g(y(T, PM, Yo), T, PM)
is expanded around g = g(y, T, PM)

(3)

Note that the partial derivative vectors and matrices are time-dependent in general. Substitution of
equation (2) and (3) in equation (1) and combining appropriate terms in ~T, ~PM' and ~y yields the
following algorithm
y g(y,T) (4)

d
dt
CY)
aT
{}g 8y
ay aT
8g
+ 8T (5)

8g {}y 8g
d ( 8y )
dt apM
---+--
{}y 8PM 8PM
(6)

{}g~
d (a y ) (7)
dt 8yo ay 8yo
=
with 8y/aT and 8y/8PM equal to zero vectors of appropriate dimension at t o. 8y/8yo is equal to a
2 x 2 unity matrix at t = O.
An expression for the covariance matrix Vy,y can then be derived from equation (2)

(8)

NUMERICAL RESULTS AND DISCUSSION

Simulations have been carried out to compare the results for the mean values and variances obtained
by the perturbation method presented above with those obtained by the Monte Carlo method.
The test problem consisted of a lasagna in a ceramic recipient which was stored for 4 days in a
refrigerator and subsequently heated for 60 min in an oven. All parameters except the refrigerator and
oven temperature were assumed to be deterministic. The mean and standard deviation of the refrigerator
temperature were respectively TJridge =
8°C and IJ'Jridge = laC; the surface heat transfer coefficient
was equal to 10 W/m 2o C. For the oven, Toven = 100 0 C , IJ'oven = laC and the surface heat transfer
coefficient was equal to 25 W /m 20 C . Both the refrigerator and oven temperature were considered to
have a Gaussian distribution. The shape, dimensions, thermophysical parameters and finite element grid
were described elsewhere [7]. The equations were solved using a 4th order Runge-Kutta-Gill solver.
In the first Monte Carlo experiment, 100 runs were carried out; in a second experiment, 1000 runs.
From Fig. 1 it is clear that the microbial load variances computed by both Monte Carlo experiments
 703

J:
(b)

., ..
(al

- ,.., "
.~~~
~
~\, ] LO

I
\
\
STORAGE REHEATING 0.5 STORAGE
~~
i 0.0 .......
I I I I I I I I I I

day 0 day 4 24 min 48 min day 0 day 4 24 min 48 min

Tbne 'l1me

FIGURE 1. Mean (a) and variance (b) ofthe microbial load in a heated lasagna with random variable refrigerator
and oven temperature at three different locations during cold storage and subsequent reheating oflasagna. Dashed
lines: perturbation algorithm results at different places on the axis of revolution; * : Monte Carlo solution with
100 runs; 0 : Monte Carlo solution with 1000 runs

agree well with these computed with the variance propagation algorithm. The perturbation algorithm
fails to predict the apparent decrease of the variance at the onset of the thermal inactivation shown by
the Monte Carlo simulations.
The variance of the microbial load increases sharply during the inactivation. Clearly small disturb-
ances in the process parameters cause considerable uncertainties in predictions of the microbial load.

CONCLUSIONS
A first order perturbation algorithm has been derived for the computation of the mean and the variance
of the microbial load during the thermal processing of foods in uncertain conditions. The algorithm has
been applied to the growth and inactivation of Lactobacillus cells during cold storage and subsequent
reheating of lasagna. The simulation results agree with the results obtained using the Monte Carlo
method. The variance of the microbial load increases considerably as the thermal inactivation of the cells
proceeds.

ACKNOWLEDGEMENTS
This study has been performed as a part of EEC-FLAIR (Food Linked Agro-Industrial Research) project
AGRF-CT91-0047 (DTEE) and of FKFO-project (2.0095.92). The authors wish to thank the European
Communities and the Belgian Fund for Collective Fundamental Research for their financial support.

REFERENCES
De Baerdemaeker, J., R.P. Singh and Segerlind L.J., Modelling heat transfer in foods using the finite element
method, Food Process Engineering, 1977, 1, 37-50. .
2 Van Impe, J.F., Nicolai, B.M., Martens, T., De Baerdemaeker J. and Vandewalle, J., Dynamic Mathematical
Model to Predict Microbial Growth and Inactivation during Food Processing. Applied and Environmental
Microbiology, 1992, 58, 2901-2909.
3 Hayakawa, K., De Massaguer, P. and Trout, R.J., Statistical variability of thermal process lethality in
conduction heating food - computerized simulation, Journal of Food Science, 1988, 53, 1887-1893.
4 Nicolai, B.M. and De Baerdemaeker, J., Computation of Heat transfer in Foods with Random Variable
Thermophysical Properties, International Journal for Numerical Methods in Engineering, 1993, 36, 523-
536.
5 Nicolai, B.M. and De Baerdemaeker, J., Probabilistic modeling and sensitivity analysis in heat transfer
computations, In Singh R.P. and M.A. Wirakartakusumah (eds) Advances in Food Engineering. CRC
Press, Boca Raton, Ann Arbor, London, Tokyo, 1992, 193-206.
6 De Baerdemaeker, J. (ed), Second year report on the EEC-FLAIR project 'Development of computer aided
design procedures to improve the quality and safety of products with a limited shelf life', Agricultural
Engineering Dept., K. U .Leu ven, 1993.
7 Van den Broeck, P., Influence of non-stationary heat transport on the microbial quality of foods Thesis
J(. U.Leuven (in dutch), 1991.
 VARIABILITY OF THERMAL PROCESS LETHALITY

K Hayakawa, J. Wang", and P. de Massaguer""

Food Science Department, Cook College Rutgers University, P. O. Box 231, New
Brunswick, NJ 08903 USA. "Process Development Division, Ross Laboratory,
Columbus, OH, USA. ""FEAA, University of Campinas, Campinas, SP, Brazil.

ABSTRACT

The variability of thermal process lethality should be predicted quantitatively for

proper process design. Simple analytical and regression equations were derived for
this prediction. According to sample applications of these equations, a high
temperature-short time process required much tighter control of process parameter
compared a conventional process and heating medium temperature influenced strongly
lethality variability.

INTRODUCTION

Proper thermal process design requires quantitative data on the variability thermal
processes. Therefore, several researches developed predictive methods for this
variability. However, there was a need of a simple and generally applicable method
for predicting the influence of all independent process parameters. Therefore, the
present work was initiated to develop such methods.

PREDICTIVE EQUATIONS DEVELOPMENT

1. Analytical equation for lethality deviation

The following Eq. 1 estimates heating phase lethality (the equation obtained assuming
the negligible value of El{ajIO/z}, valid for virtually all thermal processes).

(1)

Deviation .:1. Fph' Eq. 2, was obtained from dFph when it was expressed in terms of
multidimensional coordinates To, T1, j, f, z, and tb.
 705

(2)

2. Theoretical regression equation

A Monte Carlo simulation based screening and regression analyses were performed,
using a computerized method (1), to develop a new theoretical regression equation for
predicting the coefficient of variations ofFp' Wfp' for a z value of lOC·. The approach
for this development is described in a previous paper (2).

Wfp = 37.7975 + 4.0029 Xfh -6.1851 Xu - 1.5010 ~ + 3.1002 ~

+ 1.3056 ~h + 2.2739 Xrp + 0.9157 ~ ~ - 1.2010 ~1 ~
+ 0.9075 X2u + 0.1 % 70 X3U (3)

where, Xfh = 2.6816 Qn (i;; + 6.4516) -10.293; Xu = 15.77270 Qn (160.3036 - 1'1) -

59.4352; ~ = 11.8880 Qn (27.5 - Wz) - 34.260; ~ = 8.5106 Qn (Wjh + 6.3333) -
22.9744; Wwjh = 3.2688 Qn (Wjh + 2.1667) - 8.6646; Xrp = 17.4793 Qn (18.6889 - Fp ) -
45.7362
(4)

APPLICATION OF PREDICTIVE EQUATIONS AND DISCUSSION

Equation 2 was applied to these parametric means and deviations representative of a

conventional process: F£. = 300s, a tb = 15s, fh = 1800s, afh = 60s, jh = 2.0, ajh = 0.2,
T1 = 120·C, aTl = 0.2C-, To = 70·C, aTo = 2C·, Z = lOC·, and az = 1.0C". The
estimated aFp is 77s or 21.3% of the target Fp. The two most contributors to the Fp
deviation were afh (7.9% of Fp) and ajh (4.8%). The same equation was applied to
a high temperature- short time (HTST) process of a condition heating food with the
following conditions: Fp = 300s, 4tb = 3s, fh = 240s, afh = 7.8s,j = 1.1, aj = 0.1, T1
= 145·C, a T1 = O.2C·, To = 8YC, a To = 0.2 C", z = 1.0 C· and az = O.lC·. The
estimated aFp was 364s (102% of the target Fp). The two most contributors to the
lethality deviation were az(64.2%) and ajh (16.6%). Since it would be extremely
difficult to reduce these parametric variations, a significantly larger target Fp is
required for the HTST process.

Equation 3 was used to perform a sensitivity analysis. For this C!ll parameter except
one were kept at their frequent levels. The frequent values of fh' T1, Wz, Wfh' Wjh, and
Fp were 2400s, 117·C,. 10%, 7.0%, 12~, and 300s, respectively. The result shows that
Wrp was influenced most strongly by T l , followed by Wz and Fp.

CONCLUSIONS

The new equations were obtained for predicting the variability of thermal
process lethality. Sample application of these equatiq,ns showed over 100% deviation
in FPh for an HTST process and strong influence of Tl on Wfp.
706
ACKNOWLEDGMENT

This is publication F-10209-1-93 and F-10S7S-1-93 of the New Jersey

Agricultural Experiment Station supported by State Funds and the Center for
Advanced Food Technology, a New Jersey Commission on Science and Technology
Center; U. S. Army Research Office; U. S. Hatch Act Fund, and N. J. State Fund;
Computer time support by Rutgers University Computing Services; and
supercomputing time grant by Pittsburgh National Supercomputing Center.

REFERENCES

1. Hayakawa, K., de Massaguer, P., and Trout, R J. Statistical variability of

thermal process lethality in conduction heating food--computerized simulation.
1: Food Sci., 1988, 53, 1887-1893.
2. Wang, J., Wolfe, R R, and Hayakawa, K. Thermal process lethality variability
in conduction-heated foods. J. Food Sci., 1991, 56, 1424-1428.

NOMENCLATURE

a = m 10 (-); EI () Exponential integral (-); Fp Sterilizing value (s); f Slope

index of semilogarithmic food temperature history curve (s); 10 = TI-To (Co); j
Intercept constant of semilogarithmic food temperature history curve (-);
Lg = exp[-a(Tg-Tr)/z] (-); T Temperature (DC); Tg & TI Food temperature at the
end of heating phase of process and constant heating medium temperature,
respectively (DC); t Time (s); tb Time at the end of heating phase of process (s);
W Coefficient of variation (standard deviation/mean) x (100) (%); X Statistical
design variable used in Eq. 3. Appended subscript signifies associated process
parametric mean or W (-); z Slope index of phantom thermal death time curve (Co).

Subscript (general): e ... End of process, h ... heating phase, 0 ... initial, r ...
reference

Subscript appended to X and/or W: £h, £p, jh, tI, wfh, wjh, wz, z....X and/or W
related to fa' F p' jh' T I, Wfh , Wjh, Wz, and z, respectively.

Superscript: - average
 MICROWAVE HEATING OF UQUID FOODS

M. KYEREME & R. C. ANANTIIESWARAN

Department of Food Science
The Pennsylvania State University
University Park, PA 16802, USA

ABS1RACf

Time-temperature profiles in model Newtonian and non-Newtonian liquid foods in cylindrical

containers were studied. Dimensional analysis was used to develop prediction models based on
the variables that govern microwave heating of liquid foods. A flow visualization technique was
used to investigate the flow patterns in the heated liquid samples.

INIRODUCTION

Microwave heating and its potential applications have received much attention in the food
industry in recent years (1). Among microwave applications, microwave pasteurization and
sterilization processes for liquid foods hold the greatest promise. Although, it is in the
developmental stages, microwave sterilization has the potential to produce products of superior
overall quality than those processed by conventional methods. The objectives of this study were
to develop mathematical models for predicting time-temperature profiles during microwave
sterilization of Newtonian and non-Newtonian liquid food products.

MATERIALS AND METIlODS

Dimensional analysis was used to develop a model based on experimental data obtained for
different concentrations of sucrose solutions (2). A second model, representing non-Newtonian
liquids, was developed using data for aqueous solutions of sodium carboxymethylcellulose (CMC).
All heating experiments for the model building stage were conducted in a GE (Model JE2800)
household microwave oven (General Electric Co., Louisville, KY) with nominal output power of
700 W at 2450 MHz. A microwave-transparent Sani-Pro PS-300 (Sani-tech, Andover, NJ)
container (7.2 cm dia. x 10.6 cm height) was used to contain the liquid samples. A quantity of 375
mL of each solution was heated and temperature-time data were recorded in intervals of 2 s using
Luxtron (Model 750) fluoroptic probes placed at predetermined locations within the cans. An
improvement of a technique developed by Liu (3) was used to visualize the flow patterns in
heated liquid samples. A pouch containing crystals of potassium permanganate were used to
trace the flow profiles of the liquid. The can was left to stand undisturbed until the crystals
708
dissolved forming a colored solution in the immediate surrounding of the liquid before heating
the can in the oven. A video camera was placed at an appropriate distance from the oven door
and used to record the flow patterns in the heated sample.

RESULTS AND DISCUSSION

Based on the regression analysis of the different dimensionless terms for the Newtonian (sucrose)
data, the Model I (equation 1) was chosen. This model had an R 2 of 0.994:

(e')-O.1025( Z)-2.1689( X)0.0034

R Z (1)

This equation describes the temperature rise as a function of the variables that govern microwave
heating of Newtonian liquids.

A second dimensionless Model II (equation 2) was developed based on non-Newtonian

data obtained for aqueous solutions of different concentration levels of sodium
carboxymethylcellulose (CMC). The R2 for this model was 0.%7:

( z) -2.1467 ( X) 0.03349
(2)
R Z

The apparent viscosity was measured at a shear rate of 0.1 S-l.

Each of the models was verified by comparing predicted temperatures with the
experimentally measured temperatures. For 30% and 60% sucrose solutions, the Model I
predictions and the experimental values were in good agreement. Comparisons of predicted and
observed temperature profiles at selected points in 'Golden Delicious' apple juice were also made
(figure la). The model generally under-predicted temperatures close to the sidewall and along
the central axis. However, predictions at the sample surface were not very good.

Comparisons of temperatures predicted by Model II for 0.5% CMC and 0.8% CMC and
experimentally measured temperature profiles showed close agreement. Model II was validated
using data obtained for 0.65% CMC solution heated in a 2500 W pilot-scale microwave oven for
5 min. The predicted temperature profiles compared reasonably well with the observed values
except at the sample surface (figure Ib). As in the case of Model I, predictions at points close
to the surface deviated from the observed values. Extensive deviations were observed after 200
to 220 s of heating.
 709
14{) 14{)
- - Observed - - Observed
(a) Model 1... / (b) Model II
120 120
,..., .... ,..., .......... /
U
U ...•..... ...
"'"
....
'"
It
100
"0
100 .'

.....
"0 "

It
80 .... 80

"0"
../;../
"0"
.... ........
•...
.... • ••• .1 •••
It It
60 Q. 60
Q.
E E ....., .......
OJ
OJ
I- ",'
I- 4{)
4{)

.'"
20 20

0
0 100 200 300 50 100 150 200 250 300 350

Time (s) Time (s)

Figure 1 Predicted and observed time-temperature profiles along the central axis at a depth
of 23 cm from the surface in (a) apple juice (b) 0,65% CMC.

The following observations were made based on the flow visualization studies. In the case
of water, after 110--120 s, a plume of heated liquid started to form at the center of the can. As
the heating continued, the plume of water gradually moved upwards along the central axis, up to
the sample surface. The fluid at the surface then began to flow down along the sidewalls. Upon
reaching the bottom of the can, the downward flow moved towards the center to join the central
upward flow. The fluid in a zone located halfway between the center core and the sidewall was
virtually stagnant. This was very apparent, especially, close to the bottom of the container.
Complete mixing in the water samples occurred after 200 to 240 s heating. The patterns of flow
in the CMC solutions were generally similar to those described for water.

CONCLUSIONS

Dimensional analysis was used to develop comprehensive models to describe microwave

heating. Each model consisted of dimensionless terms that were significant in accounting for the
variations in the temperature response term. The time-temperature profiles at different locations
(except at the surface) within the liquids heated by microwave energy were predicted with a high
correlation coefficient. An improved technique was developed to visualize the flow patterns
generated in liquid products heated by microwave energy.

REFERENCES

1. Decareau, R.V. Microwaves in the Food Processing Industry. Academic Press, New York, 1985.

2. Komolprasert, V. and Ofoli, R.Y. Mathematical modeling of microwave heating by the method
of dimensional analysis. 1. Food Proc. Pres. 1989, 13: 87-106.

3. Liu, LZ. Effect of viscosity and salt concentration on microwave heating of simulated liquid
foods in cylindrical containers. M.S. thesis. The Pennsylvania State University, University Park,
PA, 1990.
 SELECTION OF BEATING MEDIA FOR THERMAL PROCESSING OF RETORT POUCHES

P.R. MASSAGUER, C.F. CARDELL I & H.G. AGUILERA

Depart. de Ciencia de Alimentos, Fac. de Engenharia de Alimentos, Universi-
dade Estadual de Campinas, C.P. 6121, l308l-970, Campinas, sao
Paulo, Brasil.

ABSTRACT

Heating media for retort pouches processing in a modified still vertical

retort were selected through a heat distribution test at 105, 113, and
l2l o C. Nylon-6 bricks were used as transducer for pouches processed in
the horizontal position. Steam, steam/air (90/10 and 70/30), and water
with overpressure were tested. For all media, l2l oC led to better heat
distribution. Steam mixtures (90/10) and water (126 KPa) had heat
distribution as good as pure steam at l2l o C. No homogeneity was achieved
with mixtures of 70% steam/30% air. No significant differences (a = 0.05)
among heating rates and lag factors were detected for transducers
processed in pure steam or 90% steam/lO% air at different levels of the
rack.

INTRODUCTION

Since temperature distribution is greatly influenced by retort design and

operation procedures, any new retort equipment must take into consideration
the necessity of providing adequate temperature and heat distribution
under fully loaded operation conditions. A steady and predictable
temperature through the retort is important for the delivery of the
desired lethality to all packages.
Steam-heated water with overpressure and steam/air mixtures are
heating media commonly used to sterilize food in pouches but they are less
efficient in the distribution and transfer of heat.
The NFPA (1) National Food Processors Assoc. has recently established
guidelines for temperature distribution verification of retorts. They
suggest that the temperature within the retort after 1 min following come
up should have a maximum range of 30 F (1.7 0 C) and should be within 1.5 0 F
(0.8 0 C) of the reference temperature device, and recommend verification of
heating rate distribution throughout the loading area.
One method to systematically evaluate temperature distribution uses
the mean temperature and standard deviation as a function of processing
holding time and thermocouple location (2). The second method uses the
range (maximum minus minimum) temperature as an expression of temperature
uniformity (3).
The objective of this work was to select heating media, using steam
as pattern, for retort pouches processed in horizontal position in a nxxlified vertical
retort which offer the best tenqJerature distribution and heat penetration.
 711
MATERIAL AND METHODS

A modified vertical retort (Dixie Inc., USA) for retort pouches

processing was used (3). Thirty Nylon-6 bricks (15xllOx150 rr®) were used
to fill a ten-level rack with 20 mm clearance between trays, simulating
commercial size retort pouches filled with a conductive food. The rack
was located 75 mm from the cross steam spreader (bottom).
1) Heating media testing and processing conditions: Pure steam, steam 90%/
10% air, and 70%/30% air and water at the same overpressure as the mixtures
were tested at 105, 113, and l2l o C. For l2l o C gauge pressure were 126 KPa
and 191 KPa for water and steam mixtures. Venting time for pure steam was
7 min. For steam and its mixtures process time was 20 min and for water a
30 min process was set. The heat penetration test was run until the
slowest heating brick reached l20 0 C from ambient temperature.
2) Temperature measurement: For temperature distribution studies, ten
needle type, Ecklund thermocouples were fixed one at each level. Once
secured, the thermocouples formed a descending spiral, in such a way that
in level 10, the thermocouple was close to the retort wall, while in level
one the thermocouple was close to the steam spreader. b) For heat
penetration tests, 5 blocks with the measuring junction of the the~ouple
positioned in the center were used and were located as follows: 2 in level
2, 1 in level 6, and 2 in level 10.
Heating and cooling medium temperatures were also recorded each 0.9
min with a digital temperature indicator (Jonhis Inst. Ltd., Brazil).
The rest of the rack was filled with nylon transducers.

RESULTS AND DISCUSSION

For all heating media tested, setpoint l2l o C led to better distribu-
tions. Results are summarized in Table 1.

TABLE 1
Comparison of heating media during temperature distribution and heat
penetration test with Nylon-6 bricks, setpoint l2l o C

Heating Ti(OC) STD(OC) DGT(OC) Coldest Meana Mean a

mean Rack Heating Lag
Level rate Factor
(min)
100% Steam 121.8 0.11 0.6 1 9.1 0.77
90% Steam 121.4 0.18 0.9 4 9.2 0.71
70% Steam 121.4 0.31 1.8 10
water-LO 121. 7 0.16 1.1 8
water-HO 126.6 0.18 1.4 10

T - Overall mean process temperature excluding come up time.

STD - Average standard deviation.
DGT - Maximum temperature difference.
a - no significant difference (a. == 0.05).
Most efficient media were the mixtures of steam 90% with a DGT of
O.9 0 C close to pure steam DGT ~ 0.6°C. Poor distribution was shown by
steam mixture 70%, with the largest DGT (1.8) and STD ~ 0.3l o C.
The coldest rack level was detected from a plot of the mean tempera-
ture and STn for each location, over the entire temperature maintenance
period. Level 10 was identified as coldest position and was selected to
locate pouches for future heat penetration tests. Mean heating rates and
712
lag factors did not show significant differences (a = 0.05) when compared
by rack position and heating media for pure steam and 90% mixtures. Figs.
1 and 2 show mean temperature and STD as a function of heating time for all
thermocouples for water with 126 KPa (LO) and 191 KPa (RO).
122.5
• MEAN OF 10 CHANNELS AFTER CUT
£
w
0:
:::> 122
I-
«
0:
w
a.
:E
w
I-
:E
«
w
:E
121
0 5 10 15 20 25 30
TIME (MIN)

Figure 1. Temperature distribution for water 126 KPa - 1210C.

122.5.,.-_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _---.
o MEAN OF 10 CHANNELS AFTER CUT
122

121

120~---,---,_--~---r_--_r--~
o 5 10 b 20 25 30
TIME (MIN)

Figure 2. Temperature distribution for water 191 KPa - 1210C.

It is widely recognized that for retort pouches over~ressured processes

are necessary in order to counteract gases expansion withln the packages
during heating and protect their integrity during coolin~. Therefore,
among the heating media tested, we have selected steam/alr mixtures (90/10)
for processing in the modified Dixie retort. Such medium showed a maximum
DGT of 0.9 0 C, not significantly different from those of pure steam in the
transducers.

REFERENCES

1. NFA Guidelines for thermal process development for foods packages in

flexible containers. Nat. Food Process. Ass., Washington, D.C., 1985.
2. Tung, M.A., Britt, I.J. & Ramaswamy, R.S. Food sterilization in steam/
air reports. Food Technol., 1990, 44, 105.
3. Cardelli, C.F., Aguilera, H.G. & Massaguer, P.R. Adaptation of a
vertical retort for thermal processing of retort pouches containing
food: Heat distribution tests. Proceeding of the 4th Brazilian
Thermal Science Meeting, 1992, 149-152.
NUMERICAL ANALYSIS OF HEAT TRANSFER IN THE STERILIZING OF CANNED FOODS BY
VACUUM PACKING

XIANG JUN HUANG, TAMOTSU HANZAWA and NOBORU SAKAI

Department of Food Science and Technology,
Tokyo University of Fisheries,
Tokyo 108, Japan

ABSTRACT

The heat transfer coefficients for steam condensation onto the surface of canned foods by vacuum
packing were obtained by numerical calculation of fundamental equations and experimental results. A
correlative equation was obtained for the heat transfer coefficient in a can under a vacuum pressure.
By using these results, the temperature distributions in vacuum packed cans were calculated
numerically. To check the calculated results, the temperature distributions were measured in a can
under the same conditions as those of the calculations. The experimental temperature distributions
were in close agreement with the calculated values.

INTRODUCTION

Canned foods by vacuum packing were developed in recent years and were packed under high vacuum
conditions with little covering water instead of the traditional packing medium. These canned foods are
not seasoned uniformly due to the lack of packing medium and add a peculiar flavour and a deep
fragrance to the foods. The heat transfer in a sterilizing process for these canned foods is performed by
steam which is converted by evaporation of a little water under conditions of high vacuum and heating.
Therefore, it seems that a degree of vacuum and volume of water are important factors for the heat
transfer in this can. However, there are few works concerning the heat transfer by steam in a can. Thus
the mechanism of heat transfer is not yet established under these operating conditions. In this study, as
basic research on the heat transfer in the can, the effect of degree of vacuum on the heat transfer
coefficient was investigated and a correlative equation was obtained.

EXPERIMENTAL APPARATUS AND PROCEDURE

Figure 1 shows a schematic diagram of the experimental model can. Cans of type No.5 and No.6 were
used. To measure the temperature in the can, Cu-constantan thermocouples were set through a small
hole in the side wall of the can and were hardened in place with an adhesive paste. The sample was a
paste of arum root which was similar to a food with the thermal physical property and surface
condition. The shape and size were cylindrical and 20 mm (diameter) X 20 mm (height). The
experimental data were given for the vacuum pressure of 180 mmHg abs. and for the heating
temperature of 127°C.
714

Figure 1 Model can. 1, Lid; 2, pipe for

pressure measurement; 3, tap of silicon gum;
4, packing gum; 5, vessel of acrylic resin; 6,
thermocouple; 7, bolt.

RESULTS AND CORRELATION

1) Effect of vacuum pressure on heat transfer: Figure 2 shows some typical experimental temperature
variations at the center of the sample (the slowest heating point) in the can with the vacuum pressure,
H G , as a parameter. Can of type No.5 was used and the number of samples as 24 pairs. From Figure 2,
it seems that there is the most suitable vacuum pressure for the heat transfer to the solid food in the
can. In Figure 2, the dotted line shows the temperature variation at the slowest heating point in the
sample in the case of traditional packing medium.

130r-r-~~~~'-'-'-'-1
120

Sample: ro~lte of the arum

80 ( 11.=0.0015 ca~m.s .•c»)
aU
c1=0.00l6 cm7s
Heating temp.: 121 ·c
......u Water: 11.0 g

40
Key (m~ cbs)
0 65
•
t:..
101
160
Jt.. 360

50 Figure 2 Effects of vacuum packing.

Time (minI

2) Correlation of heat transfer coefficient and vacuum pressure: in this study, the heat transfer
coefficient, h, was obtained by comparison of the experimental temperature distributions in a sample
in this can and the calculated ones from fundamental equations under the same conditions as those of
this experiment. Figures 3 and 4 show the same typical case of coincidence with the calculated
 715
temperature distribution on the central axis in the can with the experimental ones for the axial
direction (Figure 3) and for the cross-sectional direction (Figure 4). From these figures, the calculated
values were in sufficiently good agreement with the experimental ones and in this case h were given for
Figure 3 of 0.0027 callcm 2 soC and for Figure 4 of 0.009 cal/cm 2 soC.

140 140

120 120

•
100

80
----, Steam_ _ .~
- : Calc 60 -:Calc
.6.0().: Exp O().: Exp
Sample: Paste of the arum root Sample: Past~ of the arum root
HG = 180 mmHg abs I-t= 80 mmHg abs
h= 0.0027 call'{crrf.s.oC} h= 0.009 cal.l(cm~s.oc
00 00 10 50 60
60
Time (min)

Figure 3 Comparison of calculated temperature Figure 4 Comparison of calculated temperature

profile with experimental ones. profile with experimental ones.

-fr.-: Tung et al for aluminum

stainless steel in forced J)
convection of steam by fUrl
:pUl 0.1 -0-; Badger et al for
'E
~ 0.05
metal in steam8 )
-0-: This work for paste of
'6 arum root
£:!.
.c h =0.087 exp(-0.025 HG)
0.01
0.005

Figure 5 Relation between hand Ha.

Figure 5 shows the relationship between Ha and h. From Figure 5, the following correlative
equation was obtained.
h = 0.087 exp (-0.025Ha)
Figure 5 also shows some previous results for h 1 •2). These results are given for a polished surface and
under forced convection of steam. By using this correlation, a check for h was performed with the
calculation and experiment. The results indicated the validity of correlative equation.

REFERENCES

1. Badger et al. Heat Transfer and Crystallization, p. 24 (1945).

2. Tung, M. A. et al. 1. Food Sci. 49,939 (1984).
 HEAT TRANSFER IN LIQUID-FILLED CONTAINERS
DURING END-OVER-END ROTATION

IAN J. BRIIT, ALLAN T. PAULSON, ROBERT STARK· AND MARVIN A. TUNG

Department of Food Science and Technology, Technical University of Nova Scotia
Box 1000, Halifax, Nova Scotia, B3J 2X4 Canada
·Agriculture Canada Research Station, Kentville, Nova Scotia, B4N US Canada

ABSTRACT

Studies were carried out in a steam/air end-over-end (BOE) rotational batch retort using
liquid-filled cans to assess the importance of radial position and processing conditions on the
uniformity of product heating. Results of these experiments designed to simulate processing
of fluid food materials, indicated a trend toward increased accumulated lethality with
increased radius of rotation. However, differences were small in magnitude and not
significantly different (p > 0.05) for some processes.

INTRODUCTION

Agitation of packages containing fluid foods during thermal processing can increase the
effective heat transfer rate throughout the product during heating and cooling, and thereby
reduce the time required to achieve a safe thermal process. One form of forced convection
may be achieved by rotating entire product cars filled with food containers within a batch-
type retort. In some applications it is necessary or desirable to use overpressure processes
which provide a pressure higher than the steam pressure equivalent to the selected processing
temperature. In such cases, this superimposed pressure can influence headspace gas volume
and hence product agitation during rotation. In this work experiments were conducted in a
1300 mm diameter single-basket forced convection steam/air retort (J. Lagarde Autoclaves,
Montelimar, France) to assess the importance of radial position and processing conditions
on the uniformity of product heating in steam and steam/air mixtures.

MATERIALS AND METHODS

Cylindrical metal cans of size 300x409 (76.2 mm dia. x 115.9 mm high) were filled with
1 % (w/v) aqueous bentonite suspensions, to a headspace equal to 5 % of the can volume, and
closed without vacuum. Test containers were fitted with calibrated needle-type thermocouples
 717
located 13 mm (0.5 in) from the bottom for vertically oriented cans in still processes and at
the geometric center of cans in rotary processes. Temperature responses were assessed at
various positions throughout the fully loaded retort by logging data at 10 s intervals.
Experiments were conducted in triplicate at rotational speeds of 0, 10 and 20 rpm with target
retort temperatures of 115 and 125°C, using "pure" steam as well as mixtures of 75%
steam125 % air to provide overpressure conditions.
Heat penetration temperature history data from test cans in each position were
compared based on accumulated lethality, Fa, (Tref= 121.1°C, z=1O CO) during the heating
phase calculated for each container using the Improved General Method for the first 50 or
20 min of the heating period for retort temperatures of 115 and 125°C, respectively.

RESULTS AND DISCUSSION

Tables 1 and 2 summarize the accumulated lethality results for processes at 115 and 125°C,
respectively. Position #1 was at the center of a vertical section through the load, #2 and #4
were at the edge of the load extending horizontally and vertically, respectively, whereas #5
was at the outside comer of the plane and #3 was midway between #1 and #2. The
thermocouple from Position #1 malfunctioned for some experiments; therefore, those data
were discarded.
Fa-Values from experiments at 115°C showed increased variability for steam/air
compared to steam processes at 0 and 20 rpm; however, there were no significant differences
(p > 0.05) in the mean Fa-values at either steam fraction and there was a small decrease in
variability for steam/air compared to steam processes at 10 rpm.
Process lethalities from experiments at 125°C were more variable for rotational
processes as compared to still cooks (0 rpm) for steam processes and increased in variability
for 0 and 20 rpm processes using steam/air compared to saturated steam. For still processes,
there was a significant (p < 0.05) decrease in mean Fa-Values for steam/air processes which
indicated slower heating, possibly due to decreased surface heat transfer associated with the
steam/air heating medium. While mean Fa-values were not significantly different (p > 0.05)
between the steam and steam/air processes at 20 rpm, variability was nearly twice as large
for steam/air compared to steam processes. However, only small changes in variability were
observed between steam and steam/air processes at 10 rpm.
Overall, there was less variability in processes at 115°C compared to corresponding
processes at 125°C. This was reasonable since low viscosity fluids heat quickly and tend to
follow the retort temperature. Product temperatures approached the retort temperature
asymptotically during heating, and unaccomplished temperature differences (retort minus
product temperature) for the convection heating material approached zero during the heating
period. Since the accumulation of lethality at 115°C calculated by the Improved General
Method is 0.245 of equivalent time at 121.1°C compared with 2.45 times at 125°C, it
follows that less variability would be expected among processes at the lower temperature
where lethality accumulated at a lower rate. It is also noteworthy that while significant
differences (p < 0.05) in mean Fa-values among container positions were not identified for
all process conditions, there was a trend toward increased lethality with increased radius of
rotation at both retort temperatures.

CONCLUSIONS

While rotation resulted in increased heating rates within some containers, the effects were
718
not consistent through the product load; moreover, under these experimental conditions,
some containers did not heat more quickly when agitated than during still processes. These
findings emphasize the need to determine the slowest heating region within a retort for a
specific product and rotational process prior to completing heat penetration experiments, in
order to ensure delivery of a minimum commercial process to all containers in a retort load.

Table I
Individual and Overall Mean Values and Coefficients of Variation (%) for the
Accumulated Lethality (Po, min) During the First 50 min of the Heating Phase at 115°C.

Position Steam Steam/Air

o rpm 10 rpm 20 rpm o rpm 10 rpm 20 rpm

1 8.41 9.0- 7.71

2 8.7b 8.41 9.3b 8.4b 8.81b 8.21
3 8.61b 8.31 9.5ab 8.0- 8.7- 7.81
4 9.2c 8.3- 9.6c 8.9c 9.0b 8.00
5 9.0" 9.3b 9.4ab 8.8c 9.3 c 9.41

Mean 8.8l<Y 8.6l<Y 9.4Y 8.4x 9.Ql<Y 8.4x

CV, % 3.32 5.15 2.25 6.14 2.92 10.39

Values in columns or means with the same letter are not significantly different (p>0.05)
based on 3 runs at each condition.

Table 2
Individual and Overall Mean Values and Coefficients of Variation (%) for the
Accumulated Lethality (Fo• min) During the First 20 min of the Heating Phase at 125°C.

Position Steam Steam/Air

o rpm 10 rpm 20 rpm o rpm 10 rpm 20 rpm

1 11.7- 7.7· 11.21

2 12.71 12.3- 14.4- 11.2c 14.9·b 14.8"
3 12.1- 11.5- 13S 9.QBh 12.31b 12.71
4 13.21 11.81 13.71 10.900 13.5lb 13.4·
5 13.P 15.0- 17.7b 11.4c 16.6b 18.31

Mean 12.& 12.?y 14.8Y 1O.P 13.?Y 14.8Y

CV, % 5.40 15.35 12.64 16.41 17.10 21.39

ACKNOWLEDGEMENTS

The authors acknowledge financial support from the Natural Sciences and Engineering
Research Council of Canada and technical assistance from Mr. Gerry F. Morello.
DEVELOPMENT OF HIGH SPEED BATCHWISE STERILIZER BY STEAM FUSION

KOHJI AONO
Development Department, Iwai Kikai Kogyo Co.,Ltd.
17-10, 3-chome, Higashi-Kohjiya, Ohta-ku, Tokyo 144 ,Japan

ABSTRACT

A new batchwise sterilizer has been developed. Direct

steam mixing and flush cooling are available in the vessel.
As a result, this system achieves a short cycle time
(about 1 hr) and can be operated under UHT as well as plate
heat exchanger and accomplishes an extremely high yield.

INTRODUCTION

This system can be applied to medium-viscous (~3000 cps)

and containing particulate (~10 mm) food products. This
system consists of :
( 1) ve sse I (w i t hag ita tor and sen so r s ), (2) uti lit y s u pp I y
(including steam, air, chilled water) and, (3) control panel
(consists of PLC, displays of sensors,etc.) A flow diagram of
the steam fusion system is showm in Fig. 1.

OPERATING SEQUENCE
First, food is fed into the vessel, and they are gently
mixed by a paddle. Then pure steam is directly injected onto
the surface of the food through nozzles. Food near the nozzles
is strongly agitated and steam condenses Quickly and mixes
with the product.
Thus, they are heated in a few minutes to ultra high
temperature. After holding for the desired time, the pressure
in the vessel is released to atmospheric pressure under
sterile conditions. They are flush cooled in a few minutes to
the boiling point. If further cooling is desired, vacuum
flu s h i n g i s a v a i I a b l e t 0 5 O·C. S c hem a tic d i a g ram 0 f tim e -
temperature in this system is showm in Fig. 2.
Finally, aseptically cooled food is pneumatically fed to
the filler.
720

MATERIAL
o r FEE D - - - [ X I - - - , STEAM
CIP
AIR
VACUUM PUMP~-.....J
It----HEAT I NG
Qr MEDIUM
COuLING

FILLER
L..-_ _:;;.. too r
ASEPTIC TANK

Fig. 1 FLOW DIAGRAM OF STEAM FUS I ON SYSTEM

160
( 'C ) HEATING BY JACKET HOLD ING COOLI NG BY JACKET ~ if lI"

STEAM!MIXING: :FLUSH COOLING:

w ;------i
,,
I IE it' :

p:;
- - - - - __ - - - ___ - ____ - - _____ ~_ - - __ __ I

,,
~ 120
E-<
<:t:
p:;
W
0... 80
~
W
,
E-<
,,
- - - - - - - -- - - -- - - - - - -- -- --- - - - - - - - -1- - - - -- -- -- -- -- - - ---
'
,, '''
, ' ,
4
I I I I
, ,
_____________ .J ________ fo ___ 1_________________ ..j _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

, :,
,
,,
,,
,,,
,
o 4 8 12 16 20 24
TIME (min. )

Fig. 2 SCHEMAT leD I AGRAM OF T I ME-TEMPERATURE

IN STEAM FUSION SYSTEM (IOOL)
 COMPUTER AIDED DESIGN AND OPTIMIZATION
OF STERILIZATION OF CANNED TUNA.

Julio R. Bangal, Antonio A. Alonso2, Jose M. Gallardo2 and Ricardo I. Perez-Martfn2

10ept. Chern. Eng. (Fac. Ciencias), Universidad de Vigo. Apto. 874, 36200 Vigo, Spain.
21nstituto de Investigaciones Marinas (CSIC). Eduardo Cabello, 6. 36208 Vigo, Spain.

ABSTRACT

Methods for optimization of thermal processing have usually been experimentally tested using
isotropic and homogeneous foods or analogs. We have developed a set of computational tools that
allow the design and optimization of the thermal processing of canned foods and applied these tools to
a real and complex food system: canned tuna, which is one of the leading products in the Spanish food
industry. Suitable models were developed for heat transfer and thermal degradation of nutrients and
quality factors.

Several optimization problems were solved and experimentally tested: maximization of the overall
retention of thiamine, maximization of the overall retention of surface lightness and miminization of
process time. The optirhal processes thus calculated have very significant advantages over those used
in the industry: thiamine retention may be increased as much as 150%, surface browning greatly
reduced and process time reduced between 50-80%. Therefore, the use of these computational tools
can significantly improve both the quality and the productivity of the canning industry.

INTRODUCTION

The canned tuna industry is of great economic importance in Spain, where the production was of
116,000 tons in 1991, worth $500 million [1]. The leadership of canned tuna is repeated in the U.S.A.,
where it is the most frequently eaten kind of fish [2]. In spite of all this, the thermal processes currently
in use cause a significant overprocessing, which is detrimental to the final nutritional and organoleptic
quality. Since the pioneer work of Teixeira et al. [3], several authors have used optimal control
techniques to improve thermal processing of canned foods [4, 5, 6], but no experimental validation has
been published. Here we use new advanced computer-aided simulation and optimization methods [7, 8]
to design new processes which, guaranteeing a determined microbiological lethality, give a maximum
retention of nutrients and/or quality factors. The retention of nutrients is evaluated as volume average
retention, because the use of the C cook-value is not recommended [9].
722
METHODS

In order to optimize the process, a good mathematical model of the system is needed, together with
values for all the kinetic and thermophysical parameters involved. We have used the modeling
approach proposed by Banga et al [8], which allows the consideration of anisotropic and non-
homogeneous conduction-heated canned foods. The thermal diffusivity and effective heat transfer
coefficient were simultaneously determined from experimental heat penetration data by multi-variate
least squares fitting [8]. The nutrient considered was thiamine, as this is the most thermolabile vitamin.
With regard to the organoleptic quality factor, surface color (lightness) is the property most valued by
the consumer [10]. The kinetics of thermal degradation of thiamine and surface lightness were
determined using an unsteady-state experimental procedure [10].

The optimization problems (maximization of the final retention of thiamine or surface lightness,
minimization of process time) were solved using a stochastic optimal control algorithm, ICRSIDS
(Integrated Controlled Random Search for Dynamic Systems), which is discussed elsewhere [7].
Optimal constant retort temperature (CRT) and variable retort temperature (VRT) profiles were
calculated. It should be noted that only CRT are currently used in industry. However, VRT profiles
may offer significant advantages in several situations [7, 10], and the use of non-linear control
techniques could ease its implementation [11].

RESULTS

All the optimization problems were successfully solved with small computing effort. Figures 1 and 2
show the retention of thiamine and surface lightness vs. retort temperature for different lethality values
considering the constant retort temperature (CRT) policy. In figure 3 it can be seen that the optimal
CRT that maximizes thiamine retention is always higher than that of surface lightness. These optimal
CRT, with retort temperatures ranging from 125 to 128°C when Fc=7-12 min is considered, are
significantly superior to the processes currently used in the industry, that are typically carried out at
lower temperatures (1IO-1l5°C) and therefore, during longer times. Thiamine retention may be
increased as much as 150%, surface lightness greatly preserved and process time reduced between 50-
80%. In the same way, the use of optimal VRT may reduce process time even more (figure 4).

70~------------------------~ n.---------------------------,
Fc(rrin)

..... 60 \3
~

i
3 5
SO
!40 7
G)
5
c
i 30 7
9

~ 20 9 12
12
10~--~--~----;----+----~ 62 ~--_+----+---___If_--_+----+_....I
100 110 120 130 140 150 100 110 120 130 140 150
Retort temperature rC) Retort temperature ("C)

Fig. 1.- Thiamine retention versus (constant) retort Fig. 2.- Surface L-value versus (constant) retort
temperature for different lethality (Fc) values. temperature for different lethality (Fc) values.
 723
130
140
E Thiamine
; G)

! 125 ~ 100 CRT

8. ! 80
j !.
1:: E 60
.!
i 120
lightness
t::
40
! ~
IX 20
~
0
115 00 10 20 30 40 50
2 4 6 8 10 12 14
time (min)
Critical point lethality, Fe (min)

Fig. 3.- Optimal retort temperatures (max. thiamine Fig. 4.- Optimal (minimum process time) variable
and lightness) versus lethality (Fc). retort temperature (VRT) process compared with
the optimal constant retort temp. (CRT) profile
that assures the same final surface lightness.

REFERENCES

l. Anonymous, Export directory of canned fish, seafood and salted fish. ANFACO, 1992. Vigo, Spain.
2. Vondruska, I.W., W. Steven Otwell and R.E. Martin, Seafood consumption, availability and
quality. Food Technol., 1988, 42(5):168-172.
3. Teixeira, A A, G. E. Zinsmeister and 1. W. Zahradnik, Computer simulation of variable retort
control and container geometry as a possible means of improving thiamine retention in thermally
processed foods. 1. Food Sci., 1975, 40:656
4. Saguy, I. and Karel, M., Optimal retort temperature profile in optimizing thiamine retention in
conduction-type heating of canned foods. 1. Food Sci., 1979, 44:1485.
5. Hildebrand, P., An approach to solving the optimal temperature control problem for sterilization of
conduction-heating foods. 1. Food Proc. Eng., 1980,3:123-142.
6. Nadkarni, M. M. and Hatton, T.A., Optimal nutrient retention during the thermal processing of
conduction-heated canned foods: application of the distributed minimum principle. 1. Food Sci.,
1985, 50:1312-1321.
7. Banga, I.R., R.I. Perez-Martin, 1.M. Gallardo and J.J. Casares, Optimization of the thermal
processing of conduction-heated canned foods: study of several objective functions. 1. Food Eng.,
1991, 14(1):25-51.
8. Banga, 1.R., AA. Alonso, I.M. Gallardo and R.I. Perez-Martin, Mathematical modelling and
simulation of the thermal processing of anisotropic and non-homogeneous conduction-heated canned
foods: application to canned tuna. 1. Food Eng., 1993, il(4)369-387.
9. Silva C., Hendrickx, M. , Oliveira, F. and Tobback P., Critical evaluation of commonly used
objective functions to optimize overall quality and nutrient retention of heat-preserved foods. 1. Food
Im&., 1992, 17(4):241-258.
10. Banga, 1.R., Simulaci6n y optimizaci6n del procesamiento termico de conservas de alimentos.
Ph.D. Thesis, 1991. Dept. ofChem. Eng., Universidad de Santiago de Compostela (Spain)
II. Alonso, AA, R.I. Perez-Martin, N.V. Shukla and P.B. Deshpande, On-line quality control of
(nonlinear) batch systems: application to thermal processing of canned foods. 1. Food Eng., 1993,
19(3):275-289.
 DIFFERENT STRATEGIES FOR CONTROLLING PRESSURE
DURING THE COOLING STAGE IN BATCH RETORTS.

ANTONIO A. ALONSO, JULIO R. BANGA 1 AND RICARDO I. PEREZ-MARTIN

Food Eng. Group. Inst. de Investigacions Marinas (CSIC), Eduardo Cabello,6. 36208 Vigo, Spain.
1 Chern.
Eng. Dept., Universidade de Vigo, Apto 874, 36200 Vigo, Spain.

ABSTRACT

A new strategy for controlling pressure and liquid level during the cooling stage of the sterilization
process in steam retorts is presented. That situation, where valves dynamics are not negligiable is
considered. In the two cases, performace is demonstrated to be superior to the on/off strategies.
Besides this, controllers designed by IMC techniques permit the user to select just one parameter
related to the speed of the response desired making control operation easier.

INTRODUCTION

During the sterilization process, the traditional mode of operation is carried out in several stages
namely venting, heating and cooling. A complete discussion of all these stages is fresented by Lopez 1.
To avoid sudden pressure drop which could result in bursting of cans, Steele proposed a control
strategy, executed on a microprocessor, where the final control elements were on/off valves. As the
author reports, the main problem arises from the fact that quick actions are not possible by on/off
valves and, as a result, sharp fluctuations of pressure occur.
The present work shows a new approach to the control of the cooling stage consisting of the
separation of the process, esentially a multiinput-multioutput (MIMO) system, in a sequence of SISO
(single input single output) subprocesses where pressure is mantained at the desired value using PID-
type controllers designed by IMC techniques 3. This strategy is compared with an on/off algorithm and
its better performance demonstrated on a simulation example.

PROCESS DESCRIPTION

Water is used to cool the product while the pressure is mantained by means of air. To compensate for
excessive pressure the bleeder is employed. Water level must reach a certain value in order to
guarrantee that all the containers are inmersed in water. Then, it has to be mantained by acting on
water flow and/or drain opening. The heat transferred to the containers is also considered, for the case
of conduction-heated products. All these events may be well represented by a multivariable model
 725
consisting on a set of nonlinear ordinary and partial differential equations4 . Parameters needed to solve
the model have been taken from Mulvaney5. The different strategies employed in this work will be
implemented on the basis of this model.

PID-TYPE CONTROLLER DESIGN UNDER IMC.

The 1991 International Process Control Conference states that 90% of industrial control requirements
are still met by proportional-integral-derivative (PID) type controllers. Rivera et al 6 proposed the use
of PID controllers adjusted on the basis of IMC techniques so the number of tunning parameters is
reduced. This fact makes it very attractive from the industrial standpoint because the controller
parameters K, tI and tD are reduced to just one related directly to the form and speed of the closed
loop response we wish (filter parameter).
The IMC design method is resumed in the next two steps:
I-Given the transfer function of the process, calculate its inverse. In case of RHP poles or zeros and/or
dead time, they must not appear in the control law.
2- Apply a low pass filter on the controller with an order at least equal or greater than that of the
inverse.
In our case, plant and model will be conveniently described by first order equations and pure
integrators estimated on the basis of step tests.

RESUL TS AND DISCUSSION

ON/OFF Control and its implementation. An algorithm for the control of pressure during cooling
stage with on/off type of actions has been proposed by Steele 2. Pressure is trying to be mantained
between 2 and 2.04 bar (Pmin and Pmax respectivily). The minimun level (Lm) required is 40 I. Water
flow is 0.2 Kg/s and air flow 0.015 Kg/s. Initially, the system is at 120 QC and 1.95 bar and the retort
is filled halfway by RO-lOO cans (radius=0.0325 m, height=0.0350 m). Figure 1 r;'!presents the
evolution of pressure during the process for a valve action time of 0.5 s. These results agree with those
reported by Steele 2 in the sense that actions of the order of 0.5 s result in poor control during the
cooling cycle

2.50

t v-a.5 s

...
2.25

.A

i::. 2.00
••
...~
1.75

1.50
0 250 500 750 1000
time. _
Figure 1. Pressure responses obtained when ON/OFF control is
implemented.
726
PID type controller designed under IMC for the cooling stage. The proposed strategy consists of
dividing the MIMO system in several SISO structures worldng in serial. Water flow will be mantained
constant at the highest possible value (0.2 Kg/s) in order to guarantee fast cooling. The SISO
subsystems covering the cooling cycle are corresponding to pressure control by acting on air flow,
bleeder and drain opening respectivily.
The results of the simulations, in tenns of pressure, are shown in figure 2 where a filter value
of lOs has been employed to control pressure by acting on the drain opening. Figure 3 represents the
manipulated variables movements. In case that valve dynamics must be considered, the alternative is to
establish a sort of split control involving simultaneous actions on air flow and bleeder opening in
response to the error in pressure. Once the minimun liquid level is achieved, pressure control is carried
out by acting on drain opening.

1.00 . - - - - -........- - - - - . - - - - - . 0 . 0 0 3

rl-0.5 8

2.25 0.75
..
.:
0.002 •
.. air flow )
!:I
•
---''---¥=--------- i
2.00 ......
E:
0.50 ..
a
.
II. ;:
!
IL
a
0.001 'ii
1.75 0.25

1.50 '--_ _ _......L_ _ _ _- ' -_ _ _----I L-....>.....'---'--'-_ _ _---L_ _ _- - - ' 0.000

o 200 400 200 400 lOa
time, • time, I

Figure 2. Pressure response obtained by serial Figure 3. Manipulated variable movements

IMC controllers. obtained with serial IMC controllers.

REFERENCES CITED.

l. Lopez, A. A complete course in canning and related processes. The Canning trade Inc, 1987,
Baltimore.
2. Steele, D.l. Microprocessors and their application to the control of a horizontal batch retort ...lESI
~., 1983, 13(3), 183-193.
3. Morari, M. and Zafiriou, E. Robust process control. Prentice-Hall, 1989.
4. Alonso A.A., R. I. Perez-Martin, N.V. Shukla and P.B. Deshpande. On-line quality control of
(nonlinear) batch systems: Aplication to thennal processing of canned foods. l. Food Eng,
1993,19,275-289.
5. Mulvaney, SJ. Dynamic process modeling and evaluation of computer control of a retortfor thermal
processing. Ph.D. thesis, 1987, Cornell University, Ithaca, NY.
6. Rivera, D.E., M. Morari, S. Skogestad. Internal model control. 4. PID controller design. Ind. Eng.
Chern. Process Des. Dev., 1986,25,252-265.

ACKNOWLEDGEMENTS.- This work was financially supported by a research grant fron the
Comision Interministerial de Ciencia y Tecnologia, Spain (project ALI91-712).
 THE INFLUENCE OF UNCERTAINTIES IN PROCESSING CONDITIONS ON
THERMAL PROCESS CALCULATIONS

BART M. NICOLAI & JOSSE DE BAERDEMAEKER

Agricultural Engineering Department, K.U.Leuven, Kardinaal Mercierlaan 92, B-3001 Heverlee,
Belgium

ABSTRACT

A numerical method for the computation of mean values and variances of the temperature and process
lethality during thermal processing of foods under stochastic retort temperature conditions is derived.
Simulations indicate that the stochastic fluctuations of the retort temperature may cause a considerable
level of uncertainty in the predicted temperature and in the process lethality.

INTRODUCTION
The General Method for the design and evaluation of sterilization processes relies on the knowledge of
the temperature history in the slowest heating zone of the container. During the last three decades there
is a growing interest in the prediction of the temperature course in conduction-type foods. For this
purpose the heat conduction equation is numerically solved under the assumption of accurately known
process conditions such as the retort temperature [1-3]. However, in reality the retort temperature may
fluctuate in an unpredictable way during the course of the thermal treatment. Recently a method has
been proposed for the computation of the mean temperatures and temperature variances inside the food
resulting from stochastic fluctuations of the retort temperatures [4]. In this paper this method is extended
to calculate the mean and variance of the process lethality.

METHOD
Transient linear heat conduction in isotropic media in the absence of heat generation is governed by the
Fourier equation with appropriate initial and boundary conditions. In the finite element method the
continuum is subdivided in (finite) elements of variable size and shape which are interconnected in a
finite number m of nodal points. In every element the unknown temperature can be approximated by a
low order interpolating polynomial in such a way that the temperature is uniquely defined in terms of
the (to be computed) temperatures Uj(t) at the nodes. It is possible to show [5] that the unknown nodal
temperature vector u = [UI ..• umf is the solution of the following linear differential system:

CiI+Ku =f (1)

with C the capacity matrix and K the conductivity matrix, both m x m - matrices and fa m x 1 vector
which is a linear function of Too. In the DOT finite element code [5] equation (1) is solved using an
implicit Euler finite difference method.
At every node i of the finite element grid the process lethality Fi (in D re ! units) can be described by
the following Bigelow model [6]:

Fi = g( Uj) = 2.303/(60Dre !) x exp(2.303( Uj - T re ! )/ z) (2)

subject to the initial condition Fi(t = 0) = 0 and with Dre ! the decimal reduction time (minutes) at
the reference temperature Tre !, while z is the increase in temperature necessary to reduce this time
requirement by a factor of 10.
728
A common formalism for describing time dependent random phenomena is that of a stochastic process
[8]. It is assumed in this paper that the the mean value t:(Too) is constant and equal to Too. Further, the
stochastic fluctuation of the retort temperature, t1Too ~ Too(t) - Too, is assumed to be a stationary first
order Markov process with mean equal to zero, variance O"f~ =
O"~T~ and correlation time aT~. This
means that t1Too can be considered as the output of the following linear system with a zero mean white
noise input w(t) with autocovariance O"~ equal to unity
(3)
The smoothness of the stochastic process is determined by aT~.
A conceptually simple method for the computation of heat transfer and microbial inactivation with
stochastic parameters is the Monte Carlo approach (MC) [7]. In this method a set of stochastic retort
temperature histories is generated and for each element of the set the heat transfer process is solved
and the corresponding process lethality is computed. In the end the mean values and variances of the
temperature for a given place and time are estimated by classical statistical means. A major drawback of
the Monte Carlo method is the large amount of CPU time which is required when the number of Monte
Carlo runs is increased to decrease the uncertainty of the estimates.
An alternative method is based on the stochastic systems theoty [8]. According to this formalism
the the mean temperature vector ii and the mean process lethality F\ at every node i can be found by
solving (1) and (2) using the mean retort temperature. Further, (1) and (2) are linearized around their
mean solutions ii and Fi, respectively.

-C- 1 Kt1u + C-l~t1T. (4)

8Too 00

= (g( ui)2.303/ Z )t1ui (5)

Equation (3), (4) and (5) can be combined into the following linear system
x(t) = Fx(t) + Gw(t) (6)

The variance matrix Vx,x ~ t:(x - x)(x - xf of the solution of (6) then obeys the following Lyapunov
matrix differential equation
Vx,x(t) = FVx,x(t) + Vx,x(t)F
. T
+ GO"w2 G T (7)
In this work the variance propagation algorithm (VP) (7) is solved using a first order explicit Euler
algorithm, after reduction to a more convenient form [5]. The variances of the nodal temperatures and
process values are then easily extracted from Vx,x.
The variance propagation algorithm requires in general only a fraction of the CPU time needed by
the Monte Carlo method. A disadvant~e of the method is that it generally provides only incomplete
stochastic information (mean values, variances, no probability density functions) in contrast to the Monte
Carlo method.

NUMERICAL RESULTS AND DISCUSSION

Simulations have been carried out in order to compare the results for the mean values and variances
obtained by the numerical scheme which was presented above with those obtained by the Monte Carlo
method.
=
The test problem consisted of a AI-can (radius ro 3.41 cm, height L =
10.2 cm) filled with a 30 %
solids content tomato concentrate with k=0.542 W /m.K and pc=3.89 10 6 J /m30 C. The following process
conditions were applied: To = 65°C, Too = 125°C, O"T~ = 1°C, aT~ = 360s, h = 100 W /m2oC. The target
microorganism was Bacillus coagulans for which Dre / =1.15 min at 105°Cand z = 9.2 °C[9]. The target
process lethality at the can center was F=3D. The finite element grid consisted of 100 elements with 4
nodes per element. In a first Monte Carlo experiment 100 runs were carried out, in a second experiment
1000 runs. From Fig. 1 it is clear that variances of the second Monte Carlo experiment agree very well
with these computed with the variance propagation algorithm. However, the variances computed with
the first Monte Carlo experiment are much more scattered indicating that the number of runs was not
sufficient. The process lethality variance at the center of the can is about 0.1 D2, indicating that the
95% confidence interval is equal to [2.4, 3.6]. Clearly small disturbances in the retort temperatures cause
considerable uncertainties in the achieved process value.
 729
(a) (b) 2.0

115 I
I
I
1>
r
* P
*
I

~
.1 10 I~ ~
I t
&0.5 l
I

I )

o 1200 2400
I

3600
i
~
0.0 ~1~11~.f.~~II~I~I~II~I~I!1~II~I~I~II~I~"~"~I~II~"~"'-,-.-.-J
o
I I

1200
I

2400
I

3600
Time (8) Time (8)

FIGURE 1. Mean (a) and variance (b) of the process lethality in a heated AI-can with random variable retort
temperature at three different locations. _ _ : VP at r=3.41 cm; _ _ : VP at r=1.71 cm; __ : VP at
r=O cm; * : MC solution with 100 runs; 0 : MC solution with 1000 runs

CONCLUSIONS

A variance propagation algorithm has been derived for the computation of the mean and the variance
of the process lethality during the thermal processing of foods with uncertain retort temperature. The
mean values and variances of the temperature and process lethality within the container computed by
this algorithm agreed well with those computed with the Monte Carlo method. A relatively large number
of Monte Carlo runs (1000) was required in order to achieve a reasonable accuracy. It was shown that
considerable uncertainties on the achieved process lethality may occur.

ACKNOWLEDGEMENTS

This study has been performed as a part of FLAIR (Food Linked Agro-Industrial Research) project
AGRF-CT91-0047 (DTEE) and of FKFO-project 2.0095.92. The authors wish to thank the European
Communities and the Belgian Fund for Collective Fundamental Research for the financial support.

REFERENCES

Teixeira, A.A. and Shoemaker C.F., Computerized food processing operations, Van Rostrand Reinhold, N.Y.,
1989.
2 Tucker, G.S. and Holdsworth S.D., Mathematical modelling of sterilisation and cooking processes for heat
preserved foods Trans. IChemE, part C, 1991, 69, 5-12.
3 De Baerdemaeker, J., R.P. Singh and Segerlind L.J., Modelling heat transfer in foods using the finite element
method, Food Process Engineering, 1977, 1 37-50.
4 Nicolai, B. M. and De Baerdemaeker J., Simulation of heat transfer in foods with stochastic initial and
boundary conditions, Transactions of the IChemE, part C, 1992, 70, 78-82.
5 Polivka R.M. and Wilson E.L., Report no. UC SESM 76-2. Dept. of Civil Engineering, University of
California, Berkeley, California, 1976.
6 Stumbo, C.R., Thermobacteriology in Food Processing, 2nd Edition, Academic Press, N.Y., 1973.
7 Hayakawa, K., De Massaguer, P. and Trout, R.J., Statistical variability of thermal process lethality in
conduction heating food - computerized simulation, Journal of Food Science, 1988, 53, 1887-1893.
8 Melsa J.L. and Sage A.P., An introduction to probability and stochastic Processes, Prentice-Hall, Englewood
Cliffs, New Jersey, 1973.
9 Mallidis, C.G., Frantzeskakis, P., Balatsouras, G. and C. Katsaboxakis, International Journal of Food
Science and Technology, 1990, 25, 442-448.
 ICRS/DS: A COMPUTER PACKAGE FOR THE OPTIMIZATION OF BATCH
PROCESSES AND ITS APPLICATIONS IN FOOD PROCESSING.

Julio R. Banga1 , Ricardo I. Perez-Martfn2 and R. Paul Singh3 .

1Dept. Chern. Eng. (Fac. Ciencias), Universidad de Vigo. Apto. 874, 36200 Vigo, Spain.
~Instituto de Investigaciones Marinas. CSIC. Eduardo Cabello, 6. 36208 Vigo, Spain.
Dept. Biological and Agric. Eng., Univ. of California, Davis. Davis, CA 95616. USA.

ABSTRACT

As the food industry is largely batch-oriented, many optimization problems in food engineering are
actually optimal control problems. A computer package, ICRS/DS (Integrated £ontrolled Random
S.earch for Dynamic Systems), for the optimal control of batch processes has been developed. It is
based on the reduction of the optimal control problem to a constrained nonlinear optimization problem
via an adequate transformation of the control function. This problem is then solved using an stochastic
optimization algorithm which assures convergence with reasonable computation times. This package
has been successfully used for the optimization of two important batch processes in the food industry:
thermal processing of canned foods and air dehydration of foodstuffs. Different objective functions
(overall nutrient retention, quality factor retention, process time, energy consumption, etc.) and
constraints were considered for these processes. In all cases, ICRS/DS proved to be a reliable, useful
and easy-to-use computational tool. Results show that, in some situations, the optimal non-linear
control profiles have significant advantages over the traditional constant ones.

INTRODUCTION

Most processes in the food industry are batch operations that can be adequately modeled by heat and
mass microscopic balances plus adequate kinetic and thermodynamic relationships. Therefore, these
models usually are distributed parameter systems, where the transport equations are partial differential
equations. Due to the dynamic nature of these systems, its optimization can only be achieved using an
optimal control algorithm. The Maximum Principle of Pontryagin (MPP) is the classical approach to
solve optimal control problems, and has been used to optimize thermal processing [1, 2] and drying of
foods [3, 4, 5]. But the MPP is hard to implement and may not be efficient with highly constrained
problems. It has been suggested in the literature [1, 6, 7, 8] that the so called Control Vector Iteration
(CVI) methods have many benefits over the MPP. The basic idea of CVI is to transform the optimal
control problem in a conventional non-linear programming problem (NLP) by suitable discretization of
the control function(s).
 731
OPTIMIZA nON METHOD

We have developed a new fixed terminal time optimal control algorithm based on the CVI concept.
This method, ICRSIDS lIntegrated hontrolled Random £earch for Qynamic Systems), is based on the
discretization of the control variable(s) using variable-length piecewise-linear polynomials. The
transport phenomena equations are solved using finite differences with orthogonal collocation or finite
elements methods. The resulting constrained NLP problem is then solved using a stochastic direct
search optimization procedure based on the CRS method [9, 10]. ICRSIDS has proved to be a robust
and easy-to-use method, and no convergence problems have been found even in complex problems.

OPTIMIZATION PROBLEMS

Two important food processing operations, thennal processing and air drying, have been optimized
using ICRSIDS. The mathematical models used can be found elsewhere [II, 12, 13]. In order to
validate our models and compare our optimization results with previous works, parameters values
were taken from the literature [1, 3, 14]. In the case of thermal processing, the control variable is the
retort temperature, and we have considered the following objective functions: maximization of the
retention of a nutrient, maximization of the surface retention of a quality factor, minimization of
process time and minimization of energy consumption. Each of these problems has its corresponding
set of constraints, mainly related with the final microbiological lethality and quality retention. In the
case of air drying, the air dry bulb temperature is the main control variable, but in some cases it was
necessary to consider the relative humidity of the air as a second control variable. The objective
functions considered were: maximization of nutrient or enzyme retention, minimization of process
time, maximization of nutrient retention with a constraint on enzyme retention and maximization of
energy efficiency. A more elaborated formulation of these optimization problems is given in Banga et
al. [11] and Banga and Singh [13].

RESULTS AND DISCUSSION

In the case of thermal processing, the results for the maximization of nutrient retention are in
agreement with the literature [1, 2]: the optimal control policy only provides a 5% improvement.

110
130
e 110 V"
~
1
e:
90
f CRT
i 90 !::s 70
CAT Tdb

!. 70 VRT
! -----r;
~1::
i
a:
50

30
,.
!.
E
50

30

10 10
o 50 100 150 0 60 120 180 240
time (min)
Time (min)
Fig. 1: Thermal processing: minimum process time Fig. 2: Drying: optimal variable air temperature (VAT)
variable retort temperature profile (VRT) profile that maximizes energy efficiency compared
compared with the optimal constant temperature with the constant air temperature (CAT) profile of
process (CRT) that assures the same final the same process time. The system temperatures
retention of thiamin. (Ts) are also represented.
732
However, we have shown for the first time that the optimal control present significant advantages over
current processes in the case of maximization of retention of a quality factor (over 20% increase) or
minimization of process time (20-30% reduction, see Figure 1). This new optimal sterilization policies
will soon be experimentally validated [15].

In the case of air drying, we have found that the optimal control profiles give significant increments
of nutrient (13-50%) and enzyme retention (up to 70%). In the same way, process time can be reduced
between 15-20% ,and energy efficiency increased up to 40% (see Figure 2). All these results suggest
that advanced on-line optimal control of food processing operations is a promising area of research.

REFERENCES
1. Saguy, I. and Karel, M., Optimal retort temperature profile in optimizing thiamine retention in
conduction-type heating of canned foods. J. Food Sci., 1979, 44: 1485.
2. Nadkarni, M. M. and Hatton, T.A., Optimal nutrient retention during the thermal processing of
conduction-heated canned foods: application of the distributed minimum principle. J. Food Sci.,
1985, 50:1312-1321.
3. Mishkin, M., I. Saguy and M. Karel, Applications of optimization in food dehydration. Food
Technol.. 1982, 36(7):101-109.
4. Liapis, A.I. and R.I. Litchfiled, Optimal control of a freeze dryer- I. Theoretical development and
quasi steady state analysis. Chern. Eng. Sci., 1979, 34:975-981.
5. Chang, T.N. and Y.H. Ma, Application of optimal control strategy to hybrid microwave and radiant
heat freeze drying system. In Drying '85, eds. R. Toei and A.S. Mujumdar, Hemisphere Pub., New
York, 1985, pp 249-53.
6. Evans, L.B., Optimization theory and its application in food processing. Food Technol., 1982,
36(7):88-96.
7. Edgar, T.F. and D.M. Himmelblau, Optimization of chemical processes. McGraw Hill, New York,
1988, pp. 364-371.
8. Cuthrell, J.E. and Biegler, L.T., Simultaneous optimization and solution methods for batch reactor
control profiles. Comput. and Chern. Eng., 1989, U(l/2):49-62.
9. Goulcher, R and Casares, U., The solution of steady-state chemical engineering optimization
problems using a random search technique. Comput. and Chern. Eng., 1978, 2,:33-36.
10. Casares, U. and J.R Banga, Analysis and evaluation of a wastewater treatment plant model by
stochastic optimization. Appl. Math. Modelling, 1989, 13:420-424.
11. Banga, J.R., R.I. Perez-Martin, J.M. Gallardo and U. Casares, Optimization of the thermal
processing of conduction-heated canned foods: study of several objective functions. J. Food Eng.,
1991, 14(1):25-51.
12. Banga, J.R., A.A. Alonso, J.M. Gallardo and R.I. Perez-Martin, Mathematical modelling and
simulation of the thermal processing of anisotropic and non-homogeneous conduction-heated canned
foods: application to canned tuna. J. Food Eng., 1993, li(4)369-387.
13. Banga, J.R. and RP. Singh, Optimization of air drying offoods. J. Food Eng., 1993, accepted, in
press.
14. Luyben, K.Ch.A.M., J.K. Liou and S. Bruin, Enzyme degradation during drying. Biotechnol. and
Bioeng., 1982, 24: 533-552.
15. Alonso, A.A., RI. Perez-Martin, N.V. Shukla and P.B. Deshpande, On-line quality control of
(nonlinear) batch systems: application to thermal processing of canned foods. J. Food Eng., 1993,
19(3):275-289.
 DISRUPTION OF MICROBIAL CELLS BY FLASH DISCHARGE OF HIGH
PRESSURE GAS

KOZO NAKAMVRl,ATSUSHI ENOMOTO, MASARU HAKODA, HIDEOFUKUSHIMA,

KIYOTAKA NAGAI, TAKAHIRO MUKAEAND YOICHI MASUDA
Department of Biological and Chemical Engineering, Faculty of Engineering,
Gunma University, Tenjin, Kiryu, Gunma 376, Japan
* Department of Agricultural Olemistry, Faculty of Agriculture,
University of Tokyo, Yayoi. Bunkyo-ku. Tokyo 113. Japan

ABSTRACT

To develop a novel sterilization method for heat-sensitive materials, we investigated the burst of
microbial cells by rapid release of gas pressure under the various conditions. The wet cells of S.
cerevisiae were well destroyed, when the organisms were saturated with C02 gas at 40 'C and
40 atm and then the pressure was suddenly released. On the other hand, the dry cells were
poorly disrupted even under the same experimental condition. In particUlar, the gas with low
solubility in water could provide no effects on the survival ratio of the yeast. The survival ratio
for the spores of B. megaterium QMB 1551 was also reduced, when the spores were applied to
our system under the condition of 60 'C and 60 atm. From these findings, the mechanism of cell
disruption is considered to be correlated with the gas absorption and desorption of the cells.

INTRODUCTION

Although heat sterilization is the most traditional and popular method to prevent foods from
microbial spoilage, this process may commonly lead to undesirable changes in the taste and
character of the products because of its high temperature. To resolve the problem, sterilization
methods including no heat process have been widely studied. Among these methods, the use of
ethylene oxide or radial rays is effective, but its application is limited to the special materials by
the law in Japan. A high pressure treatment has been recently proposed as a new sterilization
method, but its extremely high pressure (several thousand atmosphere) requires the expensive
facilities.
In the present paper, we describe the disruption of microbial cells by flash discharge of
high pressure gas under the various conditions of pressure, temperature, duration time and
moisture content of the cells, using Saccharomyces cerevisiae and the spores of Bacillus
megaterium QMB 1551 as test microorganisms. Our new findings may be useful for the
development of a simple, safe and inexpensive sterilization method for heat-sensitive materials.
734
MATERIALS AND METHODS

Microorganisms
S. cerevisiae (baker's yeast) was purchased from Oriental Yeast Co., Tokyo, Japan. The spores
of B. megaterium QMB 1551 were kindly provided by Dr. T. Nagai and Mrs. N. Amaya.
These microorganisms were used in this study without precultivation.

Sterilization of the Microorganisms by Rapid Release of Gas Pressure

A glass cup containing 1.0 x 108 cells of S. cerevisiae or the spores of B. megaterium QMB
1551 was placed in the slurry reservoir for preparative columns (40 ml capacity, GL Sciences,
Tokyo, Japan), and then commercially available gas, C02 or N2, was forced into the reservoir
untill the pressure reached 10-100 atm. After standing for 30-300 min at the indicated
temperature, the gas was expelled as rapidly as possible. The sterilization efficiency of the
procedure was determined by agar plate (viable) counts.

RESULTS AND DISCUSSION

Disruption of S. cerevisiae by Explosive Decompression

As shown in Figure 1, the death of S. cerevisiae by rapid release of C02 gas pressure was
found to be extensively affected by (1) pressure, (2) temperature, (3) duration time and (4)
water content of the cells. Under the optimum condition (pressure, 40 atm; temperature, 40 'C;
duration time, 3 hr), the survival ratio of the wet cells could be reduced to 11108 , while the dry
cells were poorly killed even under the same experimental condition. Nitrogen gas with
relatively low solubility in water, however, could provide no effects on the survival ratio of the
yeast (data not shown). Similar phenomena have also been independently demonstrated by the
other groups [1-2], using several kinds of gas (C02, N2, N20 and Ar) and microorganisms (S.
cerevisiae, E. coli, and B. subtilis). From these results, the reduction of survival ratio is
thought to be highly correlated with the gas absorption and desorption of the cells.

0
r--"l

~
I
,,0 0
0'-2 ~ \
'0
• .-4
\
~
M -4 tg
ta \
> -6
• .-4 0
~
en -8
\
]l 0 20 40 0 20 40 0 2 4 0 50 100
..-I Pressure Temperature Time Water Cone.
[atm] rOC] [hr] [% ]
Figure 1. Effects of pressuring conditions by C02 gas on survival ratio of S. cerevisiae.
(Experimental condition (if not indicated): pressure, 40 atm; temperature, 40 'C; duration time,
5 hr; water content, 80 %)
 735
Although the mechanism underlying the death of microorganisms is still not well
understood, the rapid release of gas pressure seems to be responsible for the disruption of cells.
In fact, the morphological changes of microbial cells such as appearance of "holes" in the cell
surface were recognized by a scanning electron microscope (Figure 2). We are currently
evaluating in detail relationship between the cell disruption and the discharge rate of the gas.

Figure 2. Electron micrographs of S. cerevisiae before (left) and after (right) the treatment.
(pressure, 40 atm; temperature, 40 'C; duration time, 4 he; water content, 80 %)

Application of Our Sterilization Method to Bacterial Spores

Bacterial spores differ markedly from the vegetative cells in shape and structure, and are found
to be extremely resistant to heat, radiation and most chemical disinfectants. The spores of B.
megaterium QMB 1551 were, therefore, treated by rapid release of C02 gas pressure. Within
the limits of this experiment, the survival ratio of the spores could be reduced to 3.0 x 10-2
under the condition of 60 'C and 60 atm. This finding suggests that our system may be useful
for the vegetative cells, and could also be applied to the bacterial spores.

ACKNOWLEDGEMENTS

The authors would like to thank Dr. Tadashi Nagai and Mrs. Noriko Amaya, Technological
Development Center, SUNTORY Ltd., Mishima, Osaka, Japan for their kind gift of the spores
of B. megaterium QMB 1551 and helpful technical advice.

REFERENCES

l.Fraser, D., Bursting bacteria by release of gas pressure. Nature, 1951, 167,33-34.
2. Castor, T. P. and Hong, G. T. Critical fluid disruption of microbial cells. In 2nd International
Symposium on Supercritical Fluids in Boston, 1991, pp. 139-142.
 STERILIZATION OF BEVERAGES UNDER NORMAL TEMPERATURE
BY A HIGH-VOLTAGE, PULSED DISCHARGE

MASAYUKI SATO*,
KIYOSHIKIMURA~,KEIKOIKEDN*, TAKASHI OOIYAMA4* andKOICHIHATA5*
*Department of Biological and Chemical Engineering, Gunma University, Kiryu, Gunma 376,
JAPAN, 2*Yamatake Honeywell Co., 3*Hitachi Co., 4*Fuji Film Co., 5*Honshu Seishi Co.

ABSTRACT

A high-speed, high-voltage pulsed discharge in water could be used to sterilize beverages or other
liquid foods without heating and with no damage of flavor and taste. In the present study, a scaling-
up from several to hundreds ml of sterilization vessel was tried experimentally. Survival ratio of S.
cerevisiae was decreased with increasing electric field strength of the applied pulse. Effective
sterilization was attained by a ciICulation of the sample liquid through the vessel using a peristaltic
pump, in which the survival ratio decreased to about 1()-6. A scaling-up to 345 ml flow system was
tried, and then a beer was treated and bottled. There was found no change in flavor and taste between
the pulse-treated beer and the control.

INTRODUCTION

Recently, sterilization technique using electrical energy is widely recognized as an easily

controllable and energy-saving method. For example, ohmic heating is now practically in use and
has characteristics of direct heating without a heat conduction loss through walls. The application of
a D.C. or A.C. voltage has a strong sterilization effect due to the compounds that are formed
electrolytically on the electrode swface, such as chlorine, phosphorus, superoxide, and other toxic
materials for human. Without formation of such compounds, effective sterilization is possible using
high-speed, high-voltage pulsed discharge under ordinary temperatures without destruction of
temperature-sensitive materials [1]. Some research works have been reported on pulsed discharge
sterilization, where the volume of the sterilization vessel was only a few ml or less [2,3, 4]. In the
present study, the authOlS investigated a scaling-up from a few ml to hundreds ml for the purpose of
practical applications.
 737
EXPERIMENTS AND METHODS

Three kinds of vessels were used to treat the liquid sample containing S. cerevisiae as shown in Fig.
1. Fig. 1 - (a) shows a batch eel) (sample liquid: 17 ml), where liquid is poured in and out through a
hole located on the top of the vessel. A short tube of plexiglass tube (inner diameter. 46 mm; length:
12 mm) is sandwiched between two p1exiglass plates (10 mm thick) with two electrodes glued on the
inner surfaces of the plates. The diameter of the electrodes was 30 mm and the distance between the
electrodes was 6 mm. Pulsed voltage was applied between the two electrodes. Fig. 1 - (b) shows a
circulating system (50 - 100 ml of liquid including pump and flask volume) using the same vessel of
(a), where liquid flows through holes located at the top and the bottom of the vessel by means of the
peristaltic pump. Fig. 1 - (c) is a scaling-up flow system (345 ml), where liquid is supplied from the
bottom and flows out from the top with no circulation.
The pulse power source consists of high voltage transformer, charge-discharge capacitor and
spark gap, which is shown elsewhere [1]. An example of the pulse shape is as follows: pulse height,
5 - 25 kV; rising time, about 50 os; and pulse width, about 500 os.
S. cerevisiae was precultured and then suspended in a sample liquid. After applying the pulse,
the sample liquid was diluted appropriately with water and inoculated onto malt agar medium. After
culturing for 3 days at 30 °C, colonies were counted to calculate the survival ratio.

Plexiglass plates

5m.............

o
Plexiglass tube
... ..
Peristaltic pUmp
... Plexiglass plate
(a) (b) (c)
Fig. 1 Three kinds of vessels for pulsed sterilization.

RESULTS AND DISCUSSION

Using a small sterilization cell volume, it has been reported by some researchers that survival ratio
(SlSo) of the microorganisms decreased with increasing electric field strength of the applied pulse.
In the present study (using 17 mI vessel, shown in Fig. 1 - (a», however, it was difficult to reduce SI
So to less than 1()2 because the electric field strength of each location in the vessel was radically
different between the inner and outer part<; of the electrode gap.
Fig. 2 shows SlSo at several points along the vertical axis of the vessel. As is evident from the
738
figure, the more effective sterilization was carried out between the electrodes, and the rest of the
volume seemed to be a dead space. There are two possible ways to solve this problem: (1) to install
a stirrer in the vessel or (2) to circulate the liquid by pumping.
As shown in Fig. 1 - (b), a modified apparatus was used with a circulation of the sample liquid
by a peristaltic pump. It was found that a great sterilization effect by the circulation was attained, as
shown with the line (c) in Fig. 3, compared to the curve (a) of no circulation, where the horizontal axis
shows applied puIse energy per treated volume of the sample liquid Reduction of S/So by the
circulation (c) was less than 10.5 which means the flow rate of the circulation gives considerable
influence on S/So.
Using a scaled-up flow system for pulse sterilization (Fig. 1 - (c», de-gassed beer was treated
with a flow rate of 80 m1lmin (S/So was around 10-1 by this operating condition), and then bottled
after recarbonation. There were no differences in flavor and taste between the pulse treated beer and
the control.

(a)
::I: 10 -2~_~_ _--+---+------.t
o
~...
1
~
~ 10-4 Circulation
key flow rate
(mllmin)
o 0
A 40
o 160
10 -6l...!i;i'iL::i::&:&Li&:iJ~J...~~~
o 100 200 300
Pulse energy [J/mij
Fig. 2 Radial distribution of survival ratio Fig. 3 Relation between applied pulse energy
using Fig. 1 - (a). and survival ratio with and without
circulation; Capacitor: 4000 pF, Pulse
voltage: 12 kV, Pulse frequency: 50 Hz.

REFERENCES

1. Sato, M., Tokita. K., Sadakata, M., Sakai, T. and Nakanishi, K., Sterilization of microorganisms
by a high-Voltage, pulsed discharge under water., Int. Chem. Eng., 30, 695 - 698 (1990).
2. Sale, A. J. H and Hamilton, W. A., Fffects of high electric fields on microorganisms. I. Killing
of bacteria and yeasts., Biochim. Biophys. Acta, 148, 781 -788 (1967).
3. Htilsheger, H and Niemann. F. -G., Lethal effects of high-voltage pulses on E. coli K12.,
Radial. Environ. Biophys., 18,281- 288 (19&).
4. Jayaram, S., Castle, G. S. P. and Margaritis, A., Kinetics of sterilization of lActobacillus brevis
cells by the application of high voltage pulses., Biotech. Bioeng., 40,1412 - 1420 (1992).
COMPARATIVE EFFECTS OF GAMMA· RAYS AND ELECTRON BEAMS ON FILM
DOSIMETERS AND BACTERIAL SPORES

T.HAYASHI, H.TAKIZAWA*, S.TODORIKI, T.SUZUKI*, M.FURUTA** AND K.TAKAMA*

National Food Research Institute, Ministry of Agriculture, Forestry & Fisheries,
Tsukuba, Ibaraki, 305 Japan
* Faculty of Fisheries, Hokkaido University, Hakodate, Hokkaido, 041 Japan
**Research Institute for Advanced Science, University of Osaka Prefecture,
Sakai, Osaka, 593 Japan

ABSTRACT

The responses of three film dosimeters, CTA, RCF and GAF, to gamma-rays were 30%
larger than those to electron beams. Gamma-irradiation facilitated the germination of
B.pumi/us spores to a slightly higher degree than electron irradiation. Gamma-
irradiation inhibited the outgrowth, growth and the synthesis of protein and RNA to a
greater degree than electron beams. When the spores were irradiated with electron
beams at a dose 30% higher than gamma-rays, the effects of the two types of radiation
were same.

INTRODUCTION

While most of the commercial radiation sterilizations of foods, animal feeds, packaging
materials and medical produces are conducted with gamma-rays from Co-60, electron
beams have been increasingly utilized for treating these commodities. However,
whether there is any difference in the bactericidal effect between the two types of
radiation or not is controversial (1), which can be partly ascribed to inconsistent
results on the dose-rate response of film dosimeters (2).
We comparatively studied the effects of gamma-rays and electron beams on
B.pumilus spores as well as the responses of various film dosimeters to the two types of
radiation.

MATERIALS AND METHODS

Irradiation
Dosimeters and spores were irradiated in a Gamma-cell 220 (2.1x10 2 TBq of Co-60,
4.6x10 3 Gy/hr, AECL) or in a Van de Graaff electron accelerator (2.5MeV, 147J.LA,
740
1.5x10 6 Gy/hr, Nissin High Voltage Co.Ltd.).

Measurement of responses of film dosimeters to radiations

CTA (FTR-125, Fuji Photo Film Co.,Ltd.), RCF (FWT-60-00, Far West Technology
Inc.) and GAF (GAFCHROMIC, GAF Chemicals Co.) were used as film dosimeter in this
study. Several pieces of the film dosimeters were stacked together in layers and
irradiated in the Gamma-cell 220 or the electron accelerator.

Observation of germination, outgrowth and growth of spores

The 00550 of spore suspension in Demain's Medium was measured during incubation at
37C to observe the germination of spores, and the 00550 of spore suspension in G-
Medium was measured to observe the germination, outgrowth and growth.

Measurement of biosynthesis of protein and RNA

Incorporation of C-14 labeled leucine and uridine into TCA precipitates of the spores
during incubation in G-Medium at 37C was used as a measure of the synthesis of protein
and RNA, respectively.

RESULTS AND DISCUSSION

Relationships between OD280 of CTA and OD400 of GAF and between

OD510 of RCF and OD400 of GAF
The curve obtained by plotting 00280 of CTA and 00400 of GAF for gamma-rays was
coincident with that for electron beams, and the curve obtained by plotting 00510 of RCF
and 00400 of GAF for gar:nma~rays was coincident with that for electron beams (Fig.1).
These results indicate that CTA and GAF show the same response to gamma-rays and
electron beams and RCF and GAF respond to the two types of radiation in the same
manner. We have reported that both CTA and RCF show 30% greater responses to
gamma-rays than electron beams (3), which together with the results shown in Fig.1
leads to the conclusion that CTA, RCF and GAF show 30% greater responses to gamma-
rays than responses to electron beams.

Effects of gamma-rays and electron beams on B.pumilus spores

Irradiation facilitated the germination of B.pumilus spores and inhibited their outgrowth
and growth. These effects of gamma-rays were larger than those of electron beams,
when the spores were irradiated with gamma-rays at the same dose as electron beams.
However, the effects of the two types of radiation were same, when the spores were
irradiated with electron beams at a dose 30% higher than gamma-rays. The synthesis
of protein and RNA was also inhibited by gamma-irradiation to a greater degree than
electron irradi.ation, when the spores were irradiated at the same dose. The synthesis of
the macromolecules was inhibited by electron beams to the same degree as gamma-rays,
when the spores were irradiated with electron beams at doses 30% higher than gamma-
rays (Fig.2).
 741
2.5

~ 2
(!)
'0 1.5
81 • CTA-EB • RCF-EB

80.5
~
o CTA-G ¢ RCF-G

o
o 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
00(510) of ReF or 00(280) of CTA
Fig.1 Relationship of responses of film dosimeters

7000ar------------;-----------------------~
- 0 Uridine-G 0 Leucine-G
§. 6000
~ 5000 Uridine-EB • Leucine-EB
~4000
t5 3000
~2000,_ _~

~ 10000 t______====:=~::==::~::~==t=~~~~:::::r
a:
Gamma 0.50 1.00 1.50 2.00 2.50 3.00
EB 0.65 1.30 1.95 2.60 3.25 3.90
Dose (kGy)

Fig.2. Incorporation of radioactivity of C-14 labeled uridine and leucine into

irradiated spores after incubation for 2.5hr (uridine) or 4hr (leucine)

REFERENCES

1. Hayashi,T., Comparative effectiveness of gamma-rays and electron beams

in food irradiation. In Food Irradiation, ed. S,Thorne, Elsevier Applied
Science Publishers, London, 1991, pp.169-206.

2. Hayashi,T., Todoriki,S., Takizawa,H. and Furuta,M., Comparison of the

cellulose triacetate (CTA) dosimeter and radiochromic film (RCF) for
evaluating the bactericidal effects of gamma-rays and electron beams.
Radiat.Phys.Chem., 1992, 40, 593-595.

3. Hayashi,T., Todoriki,S. and Furuta,M., Dose rate dependence of cellulose triacetate

dosimeter and radiochromic film dosimeter. Radiat.Phys.Chem., in press.
 GROWTH-INHIBITORY EFFECT OF CERAMICS POWDER SLURRY ON BACTERIA

JUN SAWAll, HIDEO IGARASHI 2 , ATSUSHI HASHIM0T0 3 , and MASARU SHIMIZU l

IDepartment of Chemical Engineering, Faculty of Technology, Tokyo University

of Agriculture &Technology, 2-24-16 Nakanachi, Koganei, Tokyo 184, JAPAN
2Department of Microbiology, Tokyo Metropolitan Research Laboratory of
Public Helth, 3-24-1 Hyakunin-cho, Tokyo 169, JAPAN
3 Department of Bioproduction & Machinery, Faculty of Bioresources,
Mie University, 1515 Kamihama-cho, Tsu 514, JAPAN

ABSTRACT

Growth-inhibitory effects of ceramics powder slurry on Esherichia coli and

Staphylococcus aureus were determined by measuring the conductance change of
growth medium caused by the bacterial growth (conductance method). MgO, ZnO
and a-A1 2 03 powders were used. It was found that MgO and Zno powders inhibited
the growth of the bacteria. MgO and ZnO powders acted on bacteria in
bactericidal and bacteriostatic manner, respectively. For evaluationg the
growth-inhibitory effect, the conductance method is more useful than
conventional methods (ie. Agar plate method).

INTRODUCTION

With development of the food industry, the relations between food preservation
and food processing are getting closer and becoming more inseparable from each
other, and new food protection technologies are expected. The
growth-inhibitory effect of ceramics on bacteria holds considerable attention.
However, there is no fundamental study on the growth-inhibitory effect of
ceramics, and the method evaluationg the growth-inhibitory effect of ceramics
on bacteria has not been established.
Recently, several automated methods for montoring the growth of
bacteria, such as turbidometric method, conductance method, have been
developed. Since the ceramics powder slurry is muddy, turbidometric method
could not be used. Conductance method can deal with the muddy sample,
 743
therefore, the conductance method may be suitable for evaluating the
growth-inhibitory effect of ceramics on bacteria. The conductance method
relies on the fact that metabolizing microoganisms alters the chemical
composition of the growth medium and that these chemical changes causes a
change in the conductance of the medium [lJ. The present study aims at getting
a grasp of the growth-inhibitory effect of ceramics powder slurry on bacteria
by conductance method.

IlATEHIALS AND I/ETHODS

Test Organism
K coli 745 and S.aureus 9779 stored at Tokyo Metropolitan Research Laboratory
of Public Health were used. The test bacteria were cultured in Brain Heart
Infusion broth (BHI broth; Difco) at 308K for 24 h on a reciprocal shaker. The
culture was suspended in sterile saline to give a final bacterial
concentration of about lOa CFU/ml.

Preparation for Ceramics Powder Slurry

As ceramics powder, we selected three kinds of metallic oxides,MgO, Zno and
a-AI 2 0a(Kishida Chemical Co. LTD. ,extra grade). The ceramics powder was heated
at 453 K for 20 min and was steriled. The powder was suspended in sterile
saline to give a specified concentration.

Measurement of the Conductance Change of Growth Medium

For measuring the conductance change caused by bacterial growth, BACTOMETERR
microbial monitoring system model 64(bio Merieux VITEK) was used. Modified
plate count agar(DIFCO) was poured into the well of module equipped with
paired electrodes until it covered the electrodes. After the agar was
solidified, both of the bacterial suspension and the ceramics powder slurry
were added onto the agar. The modules were set in the incubation room of
BACTOMETER R , and the conductance change of the agar was monitored for 48 h,
at 308 K.

Determination of Bactricidal or Bactariostatic Effect.

If the conductance change was not observed over the incubation, a little
quantity of the slurry was cultured in BHI broth at 308 K for 24 h.
Futhermore, for indentification of bacteria, Eosin Methylene Blue (EKB) agar
and Mannitol Salt agar added 3% Egg Yolk (MSEY agar) were used for E. coli and
~aureus, respectively. From these experiments, it was determined whether
the ceramics powder slurry acted on the test bacteria in bactricidal manner or
bactariostatic one.

RESULTS I DISCUSSION

Figure 1 shows the growth-inhibitory effect of MgO powqer slurry on E.colL

The conductance curve of the control(ceramics powder concentration is 0 mg/ml)
changes markedly at the incubation time of about 6 h. The change is due to
744
the bacterial growth and metabolism. This point is defined as "Detection
Time(DT)". DT becomes higer with increase in the concentration of MgO powder.
DT at the concentration of 2. 7 mg/ml is 8 h, and DT at the concentration of
5.4 mg/ml is 26 h. From these results, it is found that MgO powder inhibits
the growth of E. coli. At the concentration higher than 10.8mg/ml, the DT did
not appeared, indicating no bacterial growth. In this region of MgO powder
concentration, from the results of incubation in BHI broth, MgO powder acted
on E.coli in a bactericidal manner. ZnO powder also showed the
growth-inhibitory effect on E.cali But, in the region where bacterial growth
was not observed, ZnO powder acted on E.coli in bacteriostatic manner.
Growth-inhibitory effect of MgO powder is stronger than that of Zno powder.
a-Al 2 03 powder did not show the antibacterial activity. The results obtained
for S. aureus are simi lar in tendency to those for E. coli. However, for
S. aureus, Zno powder showed the stronger growth-inhibitory effect than MgO
powder.

40 MgO
<lJ
E.col i
u"""
C~ 30
a~ ...
... ... ........ , • • •
...... ...... ...... • • • • • •...
.........................

.• .
...
...
U<lJ2Q •• control
:::JO"l
...
...
"DC

..-.,•
111

§~ 10 5.4mg/ml
Uu
...
.. : 2.7mg/ml lO.8m g /ml", \, ••••

o &11 •••• i1 ...... • • I 1 •

• •• •••••••••••
• . . .'-
• •••
.'lI. . . . . . .

o 6 12 18 24 30
Incubation time [hr]
Fig. 1 Growth inhibition of E. coli by using MgO powder slurry(conductance method)

By using the conductance method, it was found that MgO and ZnO powders
showed the growth-inhibitory effect on E. coli and S. aureus. The conductance
method is very useful for evaluating the growth-inhibitory effect of ceramics
powder slurry on bacteria.

REFERENCE

l)Eden,R. and Eden,G., Impedance MicrobioloKY, Reserch Studies Press,

Hertfordshire, 1984, pp 11-18
 RECENT STUDIES ON ASEPTIC PROCESSING OF PARTICULATE FOODS

NIKOLAOS G. STOFOROS
Thermopilon 5, 35100 Lamia, GREECE

ABSTRACT

Recent studies on mathematical modeling, liquid to particle heat transfer coefficient

calculations, residence time distribution, and microbiological validation of aseptic processes
of particulate foods are briefly presented. The equations coupling particle residence time
distribution with the heat transfer problem are also outlined.

NOMENCLATURE

A system wall heat transfer surface area, m2

Cp specific heat, J/kgK
D time at a constant reference temperature required to achieve a decimal reduction of the
initial concentration of a heat labile substance, s
Fs integrated particle F value, s
fj fraction of exiting solid particles with equal residence times, dimensionless
L total length of system, m
m mass of product in the system, kg
N final concentration of a heat labile substance in the solid product component, number
of microorganisms/ml, g/ml, or any other appropriate unit
No initial concentration of a heat labile substance in the solid product component, number
of microorganisms/ml, g/ml, or any other appropriate unit
n total number of classes of particles with equal residence times, dimensionless
T temperature, °C
t residence time, s
Vo overall heat transfer coefficient, heating medium/system wall/internal liquid, W/m2K
z longitudinal distance measured from system entrance, m
ex thermal diffusivity, m2/s
o time, O=(zJL) tp , s
Subscripts
f liquid product component
1 index indicating a class of particles with the same residence time
m heating medium
p solid product component
Symbols
- appropriately averaged value
746
INTRODUCTION
For proper design of aseptic processes for low-acid particulate foods, the following elements
must be considered: mathematical modeling, liquid to particle heat transfer coefficient,
residence time distribution, and microbiological validation. In this overview, recent studies
associated with the above topics are briefly presented. Space limits not only the discussion,
but also the references included here. A more detailed discussion as well as a rather
complete reference list can be found in [1].
MATHEMATICAL MODELING
Several mathematical models for particle temperature predictions, microbial destruction, and
quality degradation, applicable to aseptic processing of particulate foods, have been reported
in the literature, e.g., [2-8]. The concept of particle residence time distribution, or the fact
that the liquid and solid components might travel through the system with different
velocities, have been incorporated in to the mathematical models by some investigators [2, 4-
6, 8]. However, to our knowledge, the effect of the above on both fluid and particle
temperatures, has not been reported. Equation (1), derived from an overall energy balance, is
proposed here as the governing equation for each component of an aseptic unit (e.g., the
holding tube). There are two main assumptions associated with Eq. (1): first it is assumed
that there is only longitudinal fluid temperature variation, and second that each particle
travels with constant velocity throughout the whole system length.
t dT n dT.
U A(T -T) = ~m C _f +m C ~ f.~ (1)
o m f tf f pf de p pp i=1 t de
Using the same time variable, e, the equation governing the heat transfer, by
conduction, to a solid particle can be written as
2
V T . = _-.E.~
I taT . (2)
pt up ti ae
Remaining microbial or quality factors concentration in each particle can be evaluated
from the known temperature distribution within the particle and appropriate volume averaged
techniques. Average integrated particle sterilization values can be then calculated from Eq.
(3). Presumably, if a continuous probability density function is known, the summation in Eq.
(1) and (3) should be replaced by the appropriate integral.

Fs = DIOglO(No - ~ fi Ni) (3)

1=1
LIQUID TO PARTICLE HEAT TRANSFER COEFFICIENT
The boundary condition at the particle surface requires the use of the liquid-particle film heat
transfer coefficient, hp . There have been basically four methods used to calculate hp . The
traditional one involves the use of thermocouples to monitor particle temperatures e.g., [3,9].
Restriction of particle motion by the use of thermocouples necessitates the knowledge of the
relative fluid to particle velocities for a non conservative use of the results. The second
method is based on calculating hp from the change in concentration of a heat labile substance
(bioindicator) as it goes through the system e.g., [10-11]. Due to inherent difficulties in
using bioindicators, this method should be used only if direct temperature measurements are
impractical [10]. Furthermore, this method requires mathematical models that very
accurately describes the process. This is also one of the main requirement of the third
method which uses only fluid temperature data to calculate hp , e.g., [7]. The fourth method
involves the use of liquid crystals as particle temperature sensors, e.g., [7], thus alleviating
the problem of particle motion restriction. This method is not yet of use under actual
processing conditions. A critical review on hp calculations is presented in [12].
 747
PARTICLE RESIDENCE TIME DISTRmUTION
Particle residence times have been experimentally measured by monitoring tracer particles as
they were passing through the system (visually or by videotaping, or by using photo-sensors
[13], or radioactive or magnetic tracers). Following the traditional approach used to design
in-container processes, aseptic process schedules can be based on center point lethality of the
fastest moving particle. The question to be answered by the thermal process authorities is
what will be the acceptable probability level, for continuous distributions, or the appropriate
sample size, for discrete distributions, for the minimum, critical, residence time to be set.
MICROBIOLOGICAL VALIDATION
Microbiological validation is an essential part of designing thermal processes. The main
problem with aseptic processes is to design an experiment so as to exclude any lethality
accumulated during the cooling portion of the process. P~icle breakage during cooling
might lead to considerably less lethality compared to intact particles. This was verified by
[14] which introduced a homogenizing mixer that immediately disintegrated partially cooled
particles. It must be noted here, that there is always some lethality accumulated in the
particle center during the cooling process, the minimum being (assuming the fluid enters the
cooler at higher than the particle temperature) if the particle breaks when particle center
temperature equals the fluid temperature.
REFERENCES
1. Stoforos, N.G., An overview of aseptic processing of particulate foods. In Developments
in Food Science, Vol. 29, Food Science and Human Nutrition, ed. G. Charalambous,
Elsevier Science Publishers B.V., Amsterdam, The Netherlands, 1992, pp. 665-77.
2. Chandarana, 0.1. and Gavin, A. III, Establishing thermal processes for heterogeneous
foods to be processed aseptically: A theoretical comparison of process development
methods. J. Food Sci.. 1989,54(1), 198-204.
3. Chang, S.Y. and Toledo, R.T., Heat transfer and simulated sterilization of particulate
solids in a continuously flowing system. J. Food Sci.. 1989,54(4), 1017-23,30.
4. Lee, J.H., Singh, R.K., and Larkin, J.W., Determination of lethality and processing time
in a continuous sterilization system containing particulates. J. Food Eng.. 1990, 11,67-
92.
5. Manson, J.E. and Cullen, J.F., Thermal process simulation for aseptic processing of
foods containing discrete particulate matter. J. Food Sci. 1974.39, 1084-89.
6. Sastry, S.K., Mathematical evaluation of process schedules for aseptic processing of
low-acid foods containing discrete particulates. J. Food Sci.. 1986,51(5), 1323-28, 32.
7. Sawada, H., An analytical heat transfer model for liquid/particle systems. Ph.D. thesis,
Dept. of Agricultural Engineering, Univ. of California, Davis, CA, 1992.
8. Yang, B.B., Nunes, R.V., and Swartzel, K.R., Lethality distribution within particles in
the holding section of an aseptic processing system. J. Food Sci.. 1992,57(5), 1258-65.
9. Awuah, G.B., Ramaswamy, H.S., and Simpson, B.K., Surface heat transfer coefficients
associated with heating of food particles in CMC solutions. J. Food Proc. Eng.. 1993,
16,39-57.
10. Hunter, G.M., Continuous sterilization of liquid media containing suspended particles.
Food Technol. in Australia, 1972,4, 158-65.
11. Weng, Z., Hendrickx, M., Maesmans, G., and Tobback, P., Immobilized Peroxidase: A
potential bioindicator for evaluation of thermal processes. J. Food Sci.. 1991, 56(2),
567-70.
12. Maesmans, G., Hendrickx, M., DeCordt, S., Fransis, A., and Tobback, P., Fluid-to-
particle heat transfer coefficient determination of heterogeneous foods: A review. J.
Food Proc. Pres., 1992,16,29-69.
13. Yang, B.B. and Swartzel, K.R., Particle residence time distributions in two-phase flow
in straight round conduit. J. Food Sci.. 1992,57(2),497-502.
14. Unverferth, J.A., Chandarana, 0.1., and Stoforos, N.G., Aseptic processing of particulate
foods. A biological assay of hold tube-only lethality. Paper No. 45, Institute of Food
Technologists Annual Meeting, New Orleans, LA, June 20-24, 1992.
 LAMINAR TUBE FLOW OF A NEWTONIAN FLUID CONTAINING
LARGE SPHERES - APPLICATION TO ASEPTIC PROCESSING

JOHAN FREGERT and CHRISTIAN TRAGARDH

Food Engineering, Lund University
P.O.B 124, S-221 00 Lund, SWEDEN

ABSTRACT

The velocity profiles of a sucrose solution containing large alginate beads have
been measured with hot-wire anemometry for the fluid and with pulsed ultrasonic
velocimetry for the beads. The velocity profiles of two test cases are discussed.
The hot-wire method compares well with expected velocity profiles while the
ultrasonic Doppler methods implementation needs improvements to increase its
precision.

INTRODUCTION

The introduction of continuous aseptic processing lines for low acid liquid foods
containing large solid food particles, e.g. vegetable soup, resulted in increased
interest in the mechanisms of momentum transfer in such a mixture [1].
The fluid residence time in a chemical reactor has been well explored for
such defined vessels as tanks and tubes and also for more rheologically complex
fluids the residence time distributions (RTD) are known [2]. The introduction of
large particles (Dparticki D tube "" 0.1) in the mixture results in a more complicated
distribution and furthermore to different RTDs for the particles and liquid.
The test of a residence time distribution model requires experimental data
of the primitive variables such as local velocities of the phases. One objective of
this study was to implement a set of methods for obtaining them.

MATERIALS AND METHODS

Experimental set up
The velocities of the two phases were measured in a horizontal plexiglass tube
 749
(D t =44 mm). Two measuring sites were situated 1255 mm apart. The particle
velocity at the first site and the fluid velocity at the second site were measured
simultaneously, and vice versa. The particles were made from a 2.0 % alginate
solution (Kelco Manugel DMB) with 0.15 % colloidal iron and set in a 0.10 M
CaCl2 solution. The fluid was a sucrose solution which was deaerated in a vacuum
chamber before being filled into the experimental loop.

Pulsed ultrasonic Doppler velocimetry

The particle velocity was measured with pulsed ultrasonic Doppler velocimetry
developed for measurement in bubble reactors by the Institut rur Technische
Chemie, Universitat Hannover [3]. The probe was traversed at an angle of 45 0
to the flow direction. The measurement duration was 3 sec. 10 measurements
were made and repeated 3 times. With the help of a peak analysis software
(Peakfit, Jande!) the velocity spectra were fitted to a Gaussian distribution. The
mean velocity of all measurements were calculated for each location. The location
of the measurement volume and the velocity were calibrated against a moving
thread.

Hot-wire anemometry
The fluid velocity was measured with hot-wire anemometry (Dantec CTA 56C),
using a conical film probe which was built to withstand mechanical stress. The
velocity measurements were calibrated against a Poiseuillian velocity profile in
a separate loop. The particle interactions were first eliminated and then the mean
velocity over a period of 25 seconds (1 kHz) was calculated.

Physical properties
The fluid density was measured with a calculating digital density meter (Anton
Paar DMA45). The particle density was measured by taking the time for spheres
to free fall in the sucrose solution. The particle diameter was measured on a
photograph.

RESULTS AND DISCUSSION

The physical properties and the flow conditions from the case study are presented
in table 1.

TABLE 1
Physical properties and flow conditions of test case.
Case Pr Pp/Pr J.J Dp Re Vmean Cp
kg/dm3 mPa s mm mls vol. %
I 1.322 1.007 183 4.8 79 0.250 5.4
II 1.202 1.008 9.2 5.1 794 0.138 9.8

In figure 1 and test case I the measurement of the fluid velocity profile is in
accordance with the calculated Poiseuillian flow. In test case II the velocity
750
profiles are different from a Poiseuillian flow due to the different operating
conditions. The precision of the measurement of particle velocity and the location
of the measurement volume was not high. Therefore the particle velocity was
calibrated against the fluid velocity at site II. The precision of the Pulsed
ultrasonic Doppler velocimetry must be increased. The steep gradients of a
laminar velocity profile makes a small error in the measurement of position a
large error in the overall profile and in particular the slip velocity, if any.

o Fluid Poiseuille

FIGURE 1. Velocities of particles and fluid in a horizontal tube of the test cases
I and II. Both profiles at site 2.

CONCLUSIONS

The hot-wire anemometry is a useful method for the study of liquid velocity in
solid-liquid flow. The pulsed ultrasonic Doppler velocimetry implementation needs
further development for sensitive systems with steep velocity gradients.

REFERENCES

1. Sastry S.K and Zuritz C.A., A review of particle behaviour in tube flow:
Applications to aseptic processing. J.Food Process Engng., 1987,10,27-52
2. Wen C.Y. and Fan L.T., Models for flow systems and chemical reactors,
Marcel Dekker, New York, 1975.
3. Liibbert A., Korte T. and Schiigerl L.K, Ultrasonic Doppler measurements
of bubble velocities in bubble columns. In Measuring techniques in gas-
liquid two-phase flows, ed. J.M. Delaye and G. Cognet, Springer Verlag,
Berlin, 1984, 479-494.
 FLOW VELOCITIES OF PARTICLES IN HOLDING TUBE

Hideo Tozuka, Tadashi Fujiwara and Mitsukuni Mori

Research Laboratories, Japan Canners Association

INTRODUCTION
Flow velocity of a particle in a holding tube is one of
the critical factors affecting microbial safety of
continuous thermal processing of foods containing
particles. Some have investigated the flow character-
istics of the particles traveling through the tube
(Taeymans et al., 1985; Dutta and Sastry, 1990; Hang
et al., 1991; Yang and Swartzel, 1992; Bark, 1992).
The objective of this work was to investigate flow rate
of particles as affected by particle size, flow pat-
terns and densities of fluids.

EXPERIMENTAL
Model of particles and carrier fluids
The test particles were plastic balls with a density
of 1160 kg/m 3 and diameters of 10 mm and 15 mm. The
carrier fluids used were tap water (1.0 mPaS) and a
2.0% carboxymethylcellulose (CMC) solution (640 mPaS)j
which showed various flow patterns in this experiment.
The fluid density was adjusted by adding NaCI to study
the influence of density on the flow rates of the
particles.
Setup
Test particles and carrier fluids were circulated in
the experimental circuit consisted of the transparent
plastic tubes and a rotary pump. The tube was composed
of three straight pipes having 1 m length and 34.4 mm
inner diameter connected to 180 U bends. The tube was
0

inclined 2.90
upward to simulate a holding tube as
used in commercial aseptic processing systems. Motion
of the particles was recorded using a video camera.

Operation
All experiments were conducted at room temperature and
atmospheric pressure. Pump speed was adjusted to a
desired flow rate (0.12, 0.16, 0.20, 0.24 m/sec). The
flow rate was measured by an electromagnetic flowmet-
er. Under these conditions, it was assumed, based on
752

the Reynolds numbers, that a flow pattern of the

tap water was a turbulent flow and that of the eMe
solution was a laminar flow. First, to avoid any
interaction wi th particles, a single particle was
introduced into the setup to investigate the effects
of the particle size, flow pat terns and densities of
fluids. Second, a large number of particles was added
(particle to fluid ratio; 1.0% v/v) to examine the
effect of the particle interaction.

RESULTS
Flow behavior of the particle moving in the straight
tube showed a uniform motion. Effects of the flow
patterns and densities of the fluid on the particle
velocity are summarized in Table 1.
In turbulent flow
In water, the veloci ty of the single particle in-
creased linearly as fluid average velocity increased.
The smaller the differences in densities between the
particle and fluid, the greater the particle velocity.
There was no significant difference in the velocity
between particle sizes, 10 mm and 15 mm.
In laminar flow
In the eMe solution with low density (1010 kg/cm 3 ), the
particles flowed slower than the fluid average veloci-
ty. However, they flowed faster than the fluid average
velocity in the eMe solution nearly as dense as parti-
cles. Flow rate of the 15 mm particle was greater than
that of the 10 mm particle when rolling along the
bottom of the tube. This phenomenon indicates that a
greater velocity near the tube centerline under the
laminar flow gave the larger force to the 15 mm parti-
cle than 10 mm particle.
Flow rate distribution of particles
In the eMe solution having a density equal to the
particles, they flow spreads in the tube. Thus, the
flow rates showed a wide distribution. Flow rate
distribution of the particles having a heavier density
than that of fluid was sharp. In fact, most of the
particles moved slowly at the bottom of the tube.
The maximum velocity of the 2.0% eMe solution was
calculated as 1.9 times than average velocity of the
fluid based on the flow behavior index. It was con-
firmed experimentally that no particles flowed faster
than the maximum fluid velocity.
Mixture of various size particles
When the particles having diameters of 10 mm and 15 mm
flowed simultaneously, the 15 mm particle sometimes
collided with the 10 mm particle, hence the flow rate
of the later particle was accelerated.
 753

CONCLUSION
Flow rate of the particle used here was affected by
the density and average velocity of the fluid. The
fluid flow pattern was the critical factor affecting
flow veloci ty of the particles. Since the maximum
velocity of particle was smaller than the maximum rate
of fluid, the process design for continuous thermal
processing should be scheduled in accordance with the
maximum velocity of fluid. However, the majority of
the particles flowing slower than the fluid average
velocity would be over-cooked.

Table 1
Flow velocities of single particle a)
Fluids NaCl Dens i t 3" Average fluid rate(m/s)
content (kg/m ) 0.12 0.16 0.20 0.24
Water b) 0 1000 0.09 0.13 0.17 0.21
20 1140 0.11 0.15 0.19 0.23
23 1160 0.15 0.20 0.24 0.29
25 1180 0.12 0.17 0.21 0.25
2.0%CMC c) 0 1010 0.07 0.10 0.12 0.17
solution 18 1140 0.11 0.18 0.24 0.29
20 1160 0.16 0.21 0.28 0.34
23 1180 0.19 0.26 0.30 0.33
a)10 mm plastic ball (density: 1160 kg/m 3
b)Viscosity: 1.0 mPaS (turbulent flow)
c)Viscosity:620 mPaS (laminar floW)

REFERENCES
1 Taeymans, D., Roelans, E., Lenges, J, Residence time dis-
tribution in a horizontal SSHE used for UHT processing of
liquids containing solids, Presented at the 4th. ICEF, 1985
2 Dutta, B. and Sastry, S. K., Velocity distribution of food
particle suspensions in holding tube flow: Experimental and
modeling studies on average particle velocities, J. Food
Sci, 1990, 1448
3 Hong, C.W., Pan, B. S., Toledo R. T., and Chiou K.M.,
Measurement of residence time distribution of fluid and
particles in turbulent flow, J. Food Sci, 1991, 255
4 Yang, B. B. and Swartzel, K. R., Particle residence time
distributions in two-phase flow in straight round conduit,
J. Food Sci, 1992, 497
5 Bark, G, Fluids and particles, SIK's Annual report
1991/1992, 1992, 11
 FLOW OF SOLID-LIQUID FOOD MIXTURES

sm LIU. I-P PAIN.... PI FRYER.

Dept of Chemical Engineering.
Pembroke St.. Cambridge. UK.
* University of Compiegne. France

ABSTRACT
The design of efficient continuous sterilisation equipment for solid-liquid food mixtures requires that
the temperature and the velocity distributions within the fluid are known. Experiments have been
conducted to measure the variation in velocity in flows of carrots in water, and data given in terms of
flows of single particles.
INTRODUCTION

Some commercial food sterilisation processes involve the transport and heating of solid-liquid food
mixtures of high solids fractions, up to 50% solids of particle diameters up to 25 mm in non-
Newtonian carrier fluids. Fluid velocities are restricted by the need not to damage the mixture. To
design these processes, and confirm the sterility of the final product, it is vital to be able to predict
the temperatures of solid-liquid mixtures during processing. This requires information on the flow
patterns of food mixtures.
Most published work on the conveying of solid-liquid mixtures considers high-density solids in
turbulent water flows [1]. In contrast, food flows can use non-Newtonian carrier fluids and be of
low Reynolds number. In fully developed Newtonian laminar flow, particles can travel twice as fast
as the mean flow; this assumption is commonly used in process plant to ensure sterility. The flow of
single phase food liquids through process plant has been studied [2-4], and work on suspensions
described [5]. Dutta and Sastry [6-7] studied velocity distributions in flows of up to 0.8% by
volume; more realistic concentrations, of up to 40% solids, have been studied by [5] in a scraped
surface heat exchanger. This paper summarises results on the flow of food solids in water carried out
as part of a larger study [8-9].

FLOW OF FOOD SOLIDS

Details of equipment are given in [9]. The test section consisted of a 5 m of perspex pipe of 44 mm
inside diameter. Three coils which detected the passage of metal-doped tracer particles were mounted
along the test section of the pipe at 1m intervals. Tracers were constructed from shaped polythene
wrapped with metal foil. Foods and tracers of similar densities had identical settling rates; tracer
particles were representative of foods of the same density.
The flow in water of 6 mm carrot cubes of up to 35% delivered solids concentration was studied.
Densities of particles ranged between 1010 and 1080 kg/m 3. After the test section the fluid was
returned to the feed tank through a flexible pipe, from which samples could be taken to determine
the flow rate and the delivered solids concentration. Recirculation of particles and tracers allowed
data to be collected rapidly and efficiently.

Single particles. Preliminary experiments studied the flow of single particles [8]. If denser than
the fluid, particles tend to sediment, whilst if lighter they tend to rise. Dimensional analysis suggests
 755
that the particle velocity is a function of a number of variables: particle and fluid densities, particle
and pipe diameters, fluid viscosity and flow velocity. Conventionally, the modified Froude number
vm
Fr, = (1)
P ..J gd(s-I)
where s is the specific gravity of the particle, has been used to correlate particle flows. Several flow
patterns were identified, ranging from particles sitting stationary on the pipe wall; which
corresponds to the stationary deposit bed in pipe flows, which occurs for low flow velocities and/or
high particle density, through sliding flows where the particle moves along the pipe wall with
constant velocity, to fully suspended behaviour at high flow rates.
The ratio of particle to fluid velocity, vr = vrlvm,was used to correlate data. Tube Re does not
correlate the velocity ratio well; for the denser tracers, a Reynolds number of 8 000 was needed to
start the particle moving, whilst for the majority of particles the velocity ratio was already in the
region of unity at this point. An approximate analysis, based on a force balance between fluid drag
and friction for a spherical particle on the wall of a pipe, is used by [9] to suggest that a plot of v r
versus Frp-I will be a straight line. Although this analysis is simplistic, it can be used as a basis for
plotting data. Data for single particles of various shapes and effective diameters between 6.35 mm
and 15.29 mm have been obtained. Variation between data sets is seen, but data are correlated by:

vr = 1.16 - 0.7234/Frp (2)

plotted as the continuous line in Figure 1, which also includes lines of the same slope but with
intercepts 10% greater and less than 1.16 (Le. 1.16 ± 0.116). Most single particle data lie within
these boundaries. At high Froude number the particle becomes suspended, and thus travels faster
than the mean velocity of the fluid. At high Froude numbers flow is turbulent; the fluid centre line
velocity is ca. 1.2 times the mean velocity, explaining the inter,cept at Frp-I = O. Here, it would not
be expected that an analysis based on a friction balance would apply. .

Higher solids fractions. Delivered solids fractions of carrots of up to 35% have been examined.
Characteristically, the flow was stratified; a slow moving bed forms below a region of low solids
fraction. Dye tracer experiments and observation of particles indicated that the liquid velocity was
significantly higher than that of the bed. At high velocity, the flow appeared well mixed with food
particles dispersed over the cross section of the pipe. Some slow moving particles were seen even at
high fluid velocities. Experiments were carried out with cubic tracers of s = 1.01, 1.03, and 1.05,
representative of the heaviest, average and lightest particles. Three types of behaviour were seen:

(i) heavier tracers (s > 1.04) remained trapped among the bed of food particles and sometimes
slid along the bottom of the pipe, giving low velocities,
(ii) medium tracers (s ca. 1.03) could flow in any positions in the pipe, although for the
majority of the time the tracer formed part of the sedimented bed,
(iii) light particles (s < 1.01) tended to flow on top of the bed of food particles or in the upper
portion of the pipe where food particles were fewer, giving high velocities.

Results are given as a function of solids fraction in Figure I, together with equation (2) for
comparison. Figure 1(a) shows data for the heavier particle; the particle velocity is less than that of a
single particle, and the data are not widely scattered. Much greater scatter is seen in Figure 1(b) and
I(c); the particle took up a much more random range of positions in both the bed and the fluid above
it Highest particle velocities are found for the lightest particle, which can travel up to about 1.5 times
the mean flow velocity. Highest velocities are found for solids fractions between 10 and 20%. At
low solids fraction the effect of the particles is minimal, so that the velocity of the particles can be
modelled by the single particle correlation. The bed becomes significant for fractions greater than ca.
10%, increasing the liquid phase velocity. At solids fractions above ca. 20%, the whole pipe is filled
by the high solids fraction phase; the flow can again be predicted by the single particle equation.
Observation suggests that the top of the bed continues to move at higher velocities than the base, but
the difference decreases at higher solids fractions. More complex effects are seen for viscous carrier
fluids; this will be reported later [10].

CONCLUSIONS

It is important to be able to predict velocity and temperature distributions in continuous food

sterilisation plant. Experimental work has been carried out to characterise flows of solid-liquid
mixtures in horizontal tubes. It has been shown that it is possible to obtain significant velocity
756
differences between particles depending on their densities, but that at high solids fractions flows are
more uniform. Worlc is continuing to improve the correlations produced.

Acknowledgements. This work was supported by Sous-Chef Ltd. The authors wish to express
their gratitude to the late Mr Richard Sperring. SL wishes to acknowledge additional financial
support from the British Council. J-P P'S time at Cambridge was supported by the EC .

References
1. WILSON, KC, pp 103-124 in Slurry handling design of solid-liquid systems ed NP Brown and
NI Heywood, Elsevier, 1991
2. RAO, MA, & LONCIN, M, Lebesnm.Wiss. u-Techn., 7, 4-17.1974
3. HEPPEL, NJ, J.Food Eng., 4, 71-84. 1985.
4. SANCHO, MF & RAO, MA. J.FoodEngineering, 15, 1-20, 1992.
5. SINGH, RK, & LEE, JH. pp7-62 in "Advances in Aseptic Processing Technologies", ed RK
Singh and PE Nelson, Elsevier, 1992.
6. DUTrA, B & SASTRY, SKJ. Food Sci., 55,1703-1710.1990.
7. DUTrA, B & SASTRY, SK. J. Food Sci., 55, 1448. 1990
8. Lill, S, PAIN, J-P, & FRYER, PJ. Entropie, 28(170), 50-58, 1992.
9. LIU, S, PAIN, J-P, PROCTOR, JM, DE ALWIS, AAP, & FRYER, PJ. in press, Chem.Eng.
Commun, 1993.
10. LIU,S. PhD thesis, Cambridge University, in preparation.

(a) (b)

1.4 -r----------------, 1.4 -r------------------,

1.2

1.0

0.8
V
r 0.6

0.4

0.2

0.0 -I--.....--~-.--__,,__......-___r---.---l 0.0 -t--..--__,,-----r----r--I--.--..---;

0.00 0.25 0.50 0.75 1.0 0.00 0.25 0.50 0.75 1.00
l/Fr p l/Frp
(c)
1.4 -r-------....-:;;----------,
Figure 1. Plot of the velocity ratio in the flow
of carrots of different concentrations. Heavy 1.2
line shows equation (2); dotted lines show
± 10% limits. 1.0

(a) d = 7.6 mm, s = 1.04 0.8

(b) d = 7.4 mm, s = 1.03 V
(c) d = 8.3 mm, s = 1.01 r 0.6
Delivered solids fractions in all cases:
*
0.4
+
+: 4.3% .. : 8.3% 0.2
0: 11.4% ~ : 11.5%
.: 14.6% a: 16% 0.0 -I--.--__,,__--"T----r---.---,---..-----;
.. : 18% A :26% 0.00 0.25 0.50 0.75 1.00
*:35% l/Frp
 MODELING OF TIME-VARIANT HEAT TRANSFER IN A TWO-PHASE SYSTEM

K. H. PARK and R. L. MERSON

Department of Food Science and Technology
University of California, Davis, CA 95616, U. S. A.

ABSTRACT

A numerical analysis was performed to evaluate the convective heat transfer coefficients at
the particle-liquid interface as a time-dependent function of system properties for liquid foods
containing particulates. The transient energy equations for the sphere and fluid were solved
by an alternating-direction implicit method to find the temperature distribution near the
particle surface for near zero relative velocity, Re = 0.1. The instantaneous convective heat
transfer coefficients varied over the sphere surface and heat flow reversal was observed near
the rear stagnation region, which caused the displacement of the coldest point from the center
to the particle surface. Lethality computations showed that the critical point was shifted from
the particle center toward the rear stagnation point.

INTRODUCTION

The establishment of thermal processes for ensuring bacteriological safety of two-phase

foods in continuous processes is not simple because it is difficult to measure the temperature
of a moving food particle and determine the amount of heat which penetrates to the particle
center. In the absence of experimental data, a mathematical model is one approach to
estimating temperature distributions in the palticles [1, 2, 3, 4, 5]. In these works, designs for
aseptic processes have been developed assuming that the convective heat transfer coefficient
is constant throughout the process. When the magnitude of the relative velocity is small,
however, the coefficient is time dependent [6] and much smaller than usually assumed from
steady state correlations.
Since the hot fluid loses heat as it moves along the particle surface, the heat flux into
particle and the surfacelfluid temperature difference depend on time and the spatial
temperature variation inside the thermal boundary layer. Furthermore, with a variation in the
local heat transfer coefficient on the particle surface, the geometric center of the particle will
not be the coldest point at all times. This is quite important with respect to lethality because
treating the center of the particle as the critical point may be wrong and yield an improper
estimation of the lethality.
Therefore, we conducted an engineering analysis using numerical methods in order to
obtain solutions for transient heat transfer from the fluid to the particle in a holding tube and
explored the migration of the slowest heating point inside the particle as a function of system
properties.
758
NUMERICAL METHOD

A single solid sphere which is suddenly immersed in an unbounded Stokes flow environment
was considered as the idealization of the case of very small relative motion between the fluid
and the particle. Unbounded means that the effects of the tube wall and other particles were
ignored. The non-dimensional velocity components for creeping flow, which have been
found analytically [7], are:
(1)
The velocity distribution satisfies the no-slip conditions; that is, Vr = va = 0 at the surface of
the sphere.
The dimensionless unsteady energy equations for the spherical particle and fluid were
solved simultaneously using the finite difference method of alternating direction to access the
time history of transient heat transfer, with proper boundary conditions [8]. The time-
dependent film coefficient expressed in the form of a dimensionless Nusselt number was
evaluated at specific position and given time from Newton's law of cooling.

NUa =2ahp(e,tp ) =__ 2_( dTf) (2)

kf Ts-1 dr r=1

RESULTS AND DISCUSSION

The surface-averaged Nusselt numbers decreased with time and the values throughout most
of the heating process were found to be significantly below the corresponding steady state
values. The convective heat transfer coefficients decreased from the front of the sphere
toward the rear and the values became negative in the vicinity of the rear stagnation point
toward the end of process [9]. In this region, the heat is transferred not from the fluid to the
sphere, but in the opposite direction. In addition to the local variation of heat transfer, the
thermal wake behind the rear stagnation point caused by the heat flow reversal forced the
coldest point within the sphere to migrate along the axis of symmetry, from the center toward
the rear stagnation point [8].

TABLE 1
Comparison of the accumulated lethality on the axis of symmetry (e = It) of an acrylic sphere
at Pr =1,000. Data used for computation; 1.27 cm for sphere radius, 25°C for initial particle
temperature, 121.1 °C for fluid temperature, z = 10 °C and Fo at 121.1 °C = 4 min.

t Time Lethality
(min.) r=O.O r=0.2 r=O.4 r= 0.5 r=0.6 r=0.8 r = 1.0
0.1 2.26 0.000 0.000 0.000 0.000 0.000 0.000 0.013
0.5 11.28 0.111 0.110 0.120 0.131 0.146 0.196 0.304
0.7 15.79 0.392 0.373 0.373 0.380 0.393 0.437 0.537
0.9 20.31 0.821 0.769 0.740 0.733 0.732 0.747 0.813
1.0 22.56 1.084 1.010 0.961 0.945 0.933 0.926 0.967
1.1 24.82 1.375 1.278 1.206 1.179 1.155 1.123 1.134
 759
The migration of the coldest point significantly affects the lethality computation.
Table 1 shows that the accumulated lethality varies along the axis of symmetry and the value
at r = 0.8 toward the rear stagnation point is about 17 % less compared to the lethality at the
particle center when the center achieves the required lethality of unity. This example shows
that the migration of slowest heating point and the its temperature magnitude could
significantly affect the lethality computation. Since the surface temperature at the rear
stagnation point was high initially so that significant lethality was accumulated in that region
at short time, the critical point having the least lethality was located inside the particle not at
the coldest point.

CONCLUSION

Numerical solution of the transient energy equations showed that the temperature distribution
near the particle surface and the convective heat transfer coefficients were time- and spatially
dependent functions. The coldest point migrated on the axis of symmetry away from the
particle center and at the end of the heating resided at the rear stagnation point. The lethality
comparison showed that the particle center was not the critical point and the lethality at the
critical point was found to be a function of particle and fluid properties.

REFERENCES

1. Chandarana, OJ., Gavin, Ill, A. and Wheaton, F.W., Particle/fluid interface heat transfer
under UHT conditions at low particle/fluid relative velocities. L..EQQd £rQ.Q... ~ 1990,
13, 191-206.
2. Chang, S.Y. and Toledo, R.T., Simultaneous determination of thermal diffusivity and
heat transfer coefficient during sterilization of carrot dices in a packed bed. L EQ.Qd SQ..,
1990,55, 199-205.
3. Sastry, S.K., Heskitt, B.F. and Blaisdell, J.L., Experimental and modeling studies on
convective heat transfer at the particle-liquid interface in aseptic processing systems.
EQ.Qd Technol., 1989,43, 132-36,43.
4. Stoforos, N.G., Park, K.H. and Merson, R.L., Heat transfer in particulate foods during
aseptic processing. Paper No. 545, presented at the Institute of Food Technologists
Annual Meeting, Chicago, IL, June 25-29, 1989.
5. Zuritz, c.A., McCoy, S., and Sastry, S.K., Convective heat transfer coefficients for
irregular particles immersed in non-Newtonian fluid during tube flow. L EQQd~,
1990,11, 159-74.
6. Sun, X., Schmidt, S.J. and Litchfield, J.B., Convective heat transfer coefficient
measurement using magnetic resonance imaging. Paper No. 82-6581, presented at
International winter meeting of ASAE, Nashville, TN, December 15-18, 1992.
7. Bird, R.B., Stewart, W.E. and Lightfoot, E.N., Transport Phenomena, John Wiley and
Sons, New York, 1960, pp. 133.
8. Park, K. H. and Merson, R. L., Coldest point location in convectively heated particles.
Paper No. 92-6850, presented at International winter meeting of ASAE, Nashville, TN,
December 15-18, 1992.
9. Park, K. H. and Merson, R. L., Numerical study of transient heat transfer between a
particle and a fluid in a holding tube. Paper No. 299, presented at the Institute of Food
Technologists Annual Meeting, New Orleans, LA, June 20-24, 1992.
 HEA T GENERATION AND TRANSFER IN
ELECTRIC HEATING OF A LAMINAR FLOW OF FOOD
L. ZHANG and P.I. FRYER
Department of Chemical Engineering,
University of Cambridge, Pembroke St.,
Cambridge, CB2 3RA, United Kingdom.

ABSTRACT
Electric heating allows even heating of food mixtures. Food flows including mixtures are often
characterised in terms of laminar flows of Newtonian or power-law fluids. A model for the
electrical heating of such flows has been developed. High temperatures are possible near the
walls of the heater where food is slowest-moving; however, simulations of power-law flows
where the viscosity is temperature dependent suggests that the region of high temperature is
small.

INTRODUCTION
The continuous electrical (ohmic) heating of foods is a commercial process in which a food
fluid is sterilised in flow due to the passage of electrical current. The sterilisation of the food is
a function of the temperature-time history of the material, which must therefore be predictable.
The electrical processing of food containing particles has been studied thoroughly at
Cambridge [1-3]. The heating rate of a two-phase mixture is a complex function of the
electrical conductivity of the two phases and the shape of the particles. Models for the process
require solution of the field equations around the particles to take account of uneven heating
effects.
Although solid-liquid flows are not homogeneous [4-5], the flow of food mixtures has
often been modelled using continuum approaches such as power-low equations. The
behaviour of such fluids under electric heating is worthy of study because it allows an
estimation of the variation in temperature and flow velocity that will occur in a practical flow
situation. This paper outlines a model for heat transfer and generation in laminar flows in the
electrical heater, and demonstrates temperature and flow variations that might occur in
practice.

FLOW AND HEAT GENERATION

Solution of the equations for the electrical heating of a two-phase mixture is described
elsewhere [2-3]. The voltage distribution must be calculated by solving Laplace's equation for
voltage V:
V(KVV) =0 (1)
from which the electric field intensity, E = grad V, and the heat generation Q = KIEI2 can be
calculated at each point. The Navier-Stokes equation for flow and the energy equation must be
solved simultaneously with (1):
 761
Do
Ii:=: - vp + /.1(V2 U) + pF
DT
PCpDt:=: V(AVT) +Q (2-3)
In a pipe flow, approximations can be made; 2-D axial symmetric solution of (1) and a
laminar Newtonian or power-law equation for the flow of axial velocity u.
Newtonian: II du = - rdp.
T:=: Power-law: T = K (dU)n (4)
,... dr 2dx' dr
The temperature field can be obtained by solving
oT _ o(roT/or) G (5)
u ox - ex rOr +
where ex is the thermal diffusivity and G is the inherent heating rate G = 1<IEI2/pCp [3]. If
physical properties are temperature independent, and a parabolic velocity profile is 'llssumed,
a and G are the only parameters needed. Manipulation of (5) then gives:
U 06 :=: A 0(11 0610r1) + r (6)
o~ llOll
u. .
where U = u'
UO IS mean velocIty,
o e=[,x T] = 'rR> e =~T~~'
T-TO
0
A =~
aL
R uO
and
GL
r = ~.
0 0
In the linear case, A and r can be used in design.
Models for flow which include temperature-dependent physical properties have been
developed. Equation (1) is solved to find the Q field, and an iterative procedure [8] used to
find the resulting temperature and flow fields. Fluid viscosity 11 will be a strong function of
temperature, and will change with x and r, a problem analogous to that in a tubular
polymeriser [6-7]. The solution method used is similar to that of [6]. The velocity profile for a
known radial variation in 11 can be found by integration of equation (4) to give
R R R
R2 fr//.1 dr R2 fr//.1 dr R2 f<r/k) lin dr
u(r)
~ :=: r :=: r u(r)
~ =
r clor power 1aw 1·lqUl·d (7)
UO RR R ' Uo R
2JrJr//.1 drdr Jr 3//.1 dr Jr 2 (r/k)lIn dr
In this I-D model, uo is not a function of x.
Figures I and 2 show velocity and temperature profiles generated for a power law fluid and
for a Newtonian liquid. The effect of the wall is to slow the flow near it; since the temperature
of the material depend on how long it has spent in the heater, slow velocities result in high
temperatures in the wall region. The reduction in viscosity which occurs as a result of the
increase in temperature can flatten the velocity profile of the fluid; the effect of this is that the
system approaches the ideal of plug flow assumed in many models of electrical heating. It may
be that the food fluid contains components whose viscosity increases at high temperature,
however, such as starches which undergo gelation; under these conditions different velocity
profiles may result. Very high temperatures may lead to the formation of fouling deposit on
the walls of the heater or, if the temperature exceeds the boiling point of the fluid at the
pressure in the heater, to nucleate boiling.
Convective effects have not been simulated here, but have been seen even in stationary
fluids [9]; it would thus be expected that the peaks in the temperature profile will be less
pronounced than seen here.
REFERENCES
1. de Alwis, AAP, Halden, K and Fryer, PJ. Chem.Eng. Res. Des. , 67, 159-168, (1989).
2. de Alwis, AAP, Zhang, L, and Fryer, PJ. pp 103-142 in "Advances in Aseptic Processing
Technologies", ed RK Singh and PE Nelson, Elsevier, London. (1992)
3. Zhang L and Fryer, PJ.Chem. Eng. Sci. 48(4)633-642, (1993)
4. Liu, S, Pain, J-P, and Fryer, PJ. Entropie,28(170), 50-58. 1992.
5. Liu, S, Zhang, L, Pain, J-P and Fryer, PJ.IChemE Symposium Ser. 126,79-88. (1992).
6. McLaughlin, HS, Mallikarjun, R and Nauman, EB, AlChE J, 32,(3)419-425, (1986)
7. Lynn, S and Hoff, JE, AIChE J., 17,475 (1971)
8. Zhang L and Fryer, PJ. submitted to Chem. Eng. Sci. (1993)
9. Fryer, PJ de Alwis, AAP, Koury, E, Stapley, AGF and Zhang, L, J. of Fd Eng, 18,101-125
(1993)
762
ACKNOWLEDGEMENTS
Original financial support for work on electrical heating at Cambridge was provided by APV Baker. This work
was supported by an AFRC grant to lZ.
NOMENCLATURE
cp Specific heat capacity (J kglK) V Voltage (V)
E Voltage gradient (VIm) x distance (m)
F Force (N) a Thermal diffusivity (m2/s)
G Inherent heating rate (K/s) r Dimensionless variable (-)
K Coefficient in power-law (-) A Dimensionless variable (-)
L Length (m) K Electrical conductivity (S/m)
n power in power-law (-) A Thermal conductivity (W/mK)
p Pressure (N/m2) 11 viscosity (N/m s2)
Q Heat generation rate (W/m3 ) p Density (kg/m3)
R Radius of heater (m) 6 Dimensionless temperature (-)
r Radius variable (m) T Shear stress (N/m2)
T
t
temperature
time
(K)
(s)
e
Dimensionless distance (-)
Dimensionless radial variable (-)
II
To Initial temperature (K)
U Dimensionless velocity (-) Subscript and superscript
u Velocity (mls) 0 initial or mean .
1.2
power-law -
parabolic ----

---------- ''-......
....

0.8 F"-----_ ' ......"

0.6

0.4

C.2

o~~--~--~--L-~--~--~--~--~~
o 0.01 0.02 0.0) 0.04 0.05 0.06 0.01 0.08 0.09 0.1

r(m)

Figures 1 Velocity profile for the heating of tomato juice using a power law fluid with K =
0.728 exp(-O.058 T) and n = .453 exp (7.13xlO-3 T) for the case where r = 5, A = 0.02,
compared with parabolic flow, in a 0.2m diameter tube.
240

/
power-law -
parabolic - _
220
,-.,
U

t
0 200
'-'

180

P
160

~ 140

120

100

80
0
--
0.01 0.02 0.01
.-'
O.OC 0.05

r(m)
0.06 0.07 0.08 0.09 0.1

Figure 2 Exit temperature distribution for the same systems as figure 1.

 A NEW NONDESTRUCTIVE SYSTEM TO EVALUATE QUANTITATIVELY
THE EFFICIENCY OF FOOD ANTISEPTICS

Katsutada Takahashi
Laboratory of Biophysical Chemistry, College of Agriculture
University of Osaka Prefecture, Sakai, Osaka 593, Japan

ABSTRACT

A system to quantitatively measure the efficiency of food antiseptics based on the detection
of metabolic heat evolved by microbial cells has been designed. The apparatus has 24
measuring units and the growth activities of microbes at various concentrations of an antiseptic
are determined from the heat evolution processes during culture. When the measurement is
completed, a drug potency curve is drawn so that the effects of the antiseptic can be
quantitatively described. The method is nondestructive so that it is also very useful for the
study of food putrefaction.

INTRODUCTION

Although biological activities of microbial cells are often compared by observation of their
metabolic heat[1,2], little effort has been put forth to adapt this property to the quantitative
study of food putrefaction where microbial propagation is essentially responsible. The author
has developed a highly sensitive multiplex batch calorimeter useful for the detection of small
heat effects arising from cellular metabolism, and used it to analyze the growth behavior of
microbes in different culture media as well as for the putrefaction process of foods in order to
obtain the efficiency of the antiseptic on them.
In this paper, the method that also includes the design of the apparatus is described together
with some results obtained from the putrefaction experiment conducted on boiled soybeans
containing different amounts of an antiseptic.

MATERIALS AND METHOD

Apparatus
The apparatus, a type of multiplex calorimeter having 24 calorimetric units , is schematically
shown in Figure 1. (The design is now manufactured by Nippon Medical & Chemical
764

2
8
16 (Zg 15
3

6
00
17
00
5 12
1
4 9 10
1 Whole Assembly of Calorimetric Unit 2 Top Cover
3 Insulation Box 4 Water-circulating pipe 5 Water supply
6 Water outlet 7 Heatsink 8 Lid 9 Thermopile
10 Sample cell holder 11 Sample cell ( vial) 12 DC-amplifier
13 Interface 14 Microcomputer 15 Floppy disk unit
16 CRT 17 X - Y plotter 18 Printer

Figure 1. Schematic illustration of Bio Thermo Analyzer H-201.

Instruments Co. Ltd., Osaka, Japan (Fax: 81-6-445-7641) and is commercially available under
the name "Bio Thermo Analyzer H-20l").
The measurement is made by setting up the sample vial (30 cm3 )(lI) containing samples in
the calorimetric units of the heatsink(7) that is maintained at a constant temperature. The
thermopile plate (9) placed at the bottom of sample cell holder(lO) detects any temperature
differences between the sample cell and the heatsink. The differential voltage between the
thermopile plates of the sample and the reference units is led out, then digitalized by an A/D-
converter, and stored on floppy disks(l5) through a microcomputer(14) for further
computational analysis.

RESULTS AND DISCUSSION

In Figure 2, the heat evolution processes (j(t) curves) obtained for the putrefaction of boiled
soybeans containing various amounts of sodium benzoate are shown. As the concentration of
the antiseptic increased, the J (t) curve shifted toward a longer incubation period with a decrease
in the initial slope. Thus, the change in putrefying ability was correlated with the concentration
of sodium benzoate.
In order to characterize the above effect more quantitatively, the extent of the delay in
putrefaction in the presence of the antiseptic was analyzed by a simple mathematical model. If,
for a given sample containing the antiseptic at concentration i, an incubation time to attain the
putrefaction to a definite level is defined as to.(l), the extent of delay in putrefaction will
conveniently be expressed by to.(O)lto.(z), where to.(O) is the incubation time required for
attaining the same level of putrefaction when no antiseptic is added. In Figure 3 the drug
 765
>,
.';; 1. 01-----
""'
.... .-i
·M
....
'-" ,CJ
co 0.8
00
; ~
·M
0
·M
+J
J:i' 0.6
~
.-i
...
<I)

+J
0
>
<I)
5. 0.4
U
+J
CO
<I)
....
·M
.j 0.2
.<::
<I)

""
CIl 0 L.-_-L-_ _L.-_~_ _L.-_~-=::=01

0 7 14 21 28 0.1 1 10 100 1000

time, t / h
Figure 2. Putrefaction of boiled soy-
beans containing various amounts of
sodium benzoate at 30°C as measured by
Bio Thermo Analyzer.
[Na benzoate] / mM
Figure 3. Drug potency curve for the
action of sodium benzoate on boiled
soybean as determined from the
experiment shown in Figure 2.

potency curve of sodium benzoate against the putrefaction of boiled soybeans obtained from
the above experiment is shown in the normalized form of to.(O)/to.(l) versus i. By a regression
analysis a parameter characterizing the efficiency of the antiseptic was obtained to be Ki =
13.3 mM, the concentration of sodium benzoate at which the putrefying ability of the boiled
soybeans is repressed by 50%.

CONCLUS IONS

From the result given above together with those obtained from the same series of
experiments conducted on the other foods and the antiseptics, it may be concluded that the
method described here is very useful and that it will contribute to further advances in the
quantitative evaluation of food antiseptics.

REFERENCES

1. Takahashi, K. Application of calorimetric methods to cellular processes. Thermochimica

Acta, 1990,163,71-80.
2. Kimura, t., and Takahashi, K., Calorimetric studies on soil microbes. 1. Gen. Microbiol.
(London), 1985,131,3083-89.
DEVELOPMENT AND APPLICATION OF ASEPTIC NEW MATERIALS(ASEPLA)

S.KUNISAKI, K.NODA, T.SAEKI,and T.AMACHI

Institute for Fundamental Research,SUNTORY Ltd.
1023-1, Yamazaki, Shimamoto-cho, Mishima-gun, Osaka 618, Japan.

ABSTRACT
ASEPLA(Aseptic Plastic) is new material with high antimicrobial activity,
prepared by a mixing resin with a powdered antimicrobial zeolite contain-
ing Ag. It was confirmed that ASEPLA has higher antimicrobial activity in
raw water containing mineral components than in the deionized water. Thus,
it was expected that mineral components in raw water would play an impor-
tant role for its antimicrobial activity.It was recognized that ASEPLA
can maintain high antimicrobial activity even in the presence of Cl-.

INTRODUCTION

~n a water-purifying device, residual chlorine is eliminated prior to the

demineralizing process.However,the elimination of chlorine necessarily
results in undesirable propagation of microorganisms in the water-trans-
porting pipes or water-reserving tanks.In order to prevent the micro-
organism contamination, we have developed and applied a system using an
antimicrobial plastic, ASEPLA. ASEPLA is a new resin material with a supe-
rior antimicrobial activity, prepared by mixing an antimicrobial zeolite
containing Ag (ZEOMIC) made by Shinanen Co. Ltd. with plastic materials.
In this paper, fundamental characteristics of ASEPLA will be discussed.

MATERIALS AND METHODS

Antimicrobial Activity of ASEPLA
Microorganisms isolated from raw water was suspended in sterilized water to
at the concentration of about 10 1 to 10 5 cells/ml.The ASEPLA sample(5x5cm)
was soaked in 150 ml of the microbial suspension, and allowed to stand at
28~ for 96 hrs with stirring.One ml portion was taken from the suspension
every 24 hours to count the living cell numbers at each sampling times.

RESULTS
Effect of Initial Microbial Concentration on the
Antimicrobial Activity
 767
Three types of microbial suspension at the initial concentration of
about 10 1 , 10 3 and lOS cells/ml were prepared to test the antimicrobial
activity of ASEPLA.As shown in Figure 1,the antimicrobial activity is
dependent on the initial cell number, that is, the increase of its
activity by decrease of the cell concentration. Concentration of Ag in
each suspension was about 3 ppb.

,·. · · . ·. ·. l. . . ·. . . . . I. . ·. · . . · . I. . . ·. . . . ·L·. _. e:...

i ! ~j~,-.~ .... 'O ~~OA~~~ILA
: ··············r················~ .: .......... ~ ....,;: .............. . - -8. - control
I..
-b . - !
I .-,
J....!
I
- ..... - + ASEPLA
_..l?r-.......!.................!...............!................./............... . ····C···· co nJ[QI
:\·ji···;:::.tt··::;··,,···::;:;
I ... ~...'....::;:;··I.:::;I ..::;:;···4···;:::.
! ..·=1· ....•.... + ASt::tJLA

.'·0·........',··..·....·....·..··,!......·....·....·,I..............·..I,··..·..·..·....·
.. I : i i
1 0° Woj\..,....'~........1!..........-!~............! ...........
o 5 10 15 20 2Ehr
Figure 1. Effect of Initial Microbial Concentration on
the Antimicrobial Activity
Effect of Mineral Components of Water
Antimicrobial activity of ASEPLA was examined both in raw water with
mineral components (total hardness:98) and in deionized water.In the
presence of ASEPLA,living cell in raw water could hardly be detected
within 2 hrs,while in deionized water living cells were observed for 5
hrs but not after 24 hrs(Figure 2).The concentration of silver in raw
water used for this experiment reached to 25 pg/L, which was almost
similar to that of deionized water being used,28 ~g/L.Thus,the effect
of some ionic components can be expected.

·······i················i···· ·······t················t···········
ii i I i! :
__ ~ ~L- ________________________ ~

.. ···········!················l················1········.........
; , I o DW(D~ionized Water)
............!.................l................ j ................. :::g:= ~\%t~a~p~ter)
! j j
··············l·· ............}................{................ • RW+ASEPLA
! ! i I~--~~------------------------...
·..•···· ..•.. ·f .. ·· ....·.. ·· ··:····..·••••....·T ..···....·.... ··i·· ....··.. ··....

1 0° ....&..I..'....",.....",.1.1.1.........11................' .............
o 10 20 30 40 50hr
Figure 2. Effect of Mineral Components of Water

Effect of Silver Concentration Released from ASEPLA

Test solution containing 20 ~g/L of silver was prepared by dipping of
ASEPLA samples in deionized water for 1 week. Diluted solution was used
to investigate the minimum concentration of silver required for the
antimicrobial activity. Living cell in microbial suspension containing
more than 10 pg/L of silver could hardly be detected after 2 hrs, while
in the suspension containing only 1 pg/L of silver propagation of the
microorganisms were observed as shown in Figure 3.
768

100~~~~~__~____~
o 25 50 75 100 125 hr
Figure 3. Effect of Silver Concentration Released from ASEPLA

Comparison of Antimicrobial Activity between ASEPLA and

AgN0 3 in the Presence of NaCl
The antimicrobial activity of ASEPLA was not affected by the presence
of NaCl at 0.8 w/w% concentration(Figure 4) . However, in the case of
AgN0 3 (100 ppb),the effect of silver,which can terminate microorgan-
isms within 2 hrs,was extremely weakened by the co-existence of NaCl
(0.8%). This would be attributed to the formation of insoluble Agel.

o NaCI
---c- NaCI+ASEPLA
• kNJ3
--0- NaCI+AgN0 3

1 01

100 ~~~--~~--~~~
o 10 20 30
Figure 4. Comparison of Antimicrobial Activity between ASEPLA
and AgNOJ in the Presence of NaCl

DISCUSSION
ASEPLA has the antimicrobial property when soaked into aqueous solution.
It was confirmed that ASEPLA shows higher antimicrobial activity in raw
water containing mineral components than in the deionized water.Thus,it
was expected that mineral component in raw water would play an important
role for the antimicrobial activity. It was suggested that concentration
of Ag was correlated with the ASEPLA activity, and the inactivation of
microbes was caused by silver released into liquid from ASEPLA.It is well
known that AgNO J has high antimicrobial activity. However, AgN0 1 cannot
be used for the water containing Cl- such as sea water because of forma-
tion of insoluble salt.On the other hand,ASEPLA can maintain its activity
even in the presence of Cl-. Consequently, it was suggested that ASEPLA
differed somewhat from AgN03 in existing its antimicrobial activity.
 CONTINUOUS STERILIZATION OF PARTICULATE FOODS BY OHMIC
BEA TING: CRITICAL PROCESS DESIGN CONSIDERATIONS

SUDHIR K. SASTRY
The Ohio State University
Department of Agricultural Engineering
590 Woody Hayes Drive
Columbus, OH 43210, USA

ABSTRACT

Research at this laboratory for the past several years has involved determination of
electrical conductivities of foods, microbial death kinetics, process modeling and experimental
verification. Finite element models developed for heating of solid-liquid mixtures in a
continuous flow ohmic heater indicate that if all particles and liquid are of equal electrical
conductivities, the particle cold spots heat slightly faster than the liquid. For high
concentration mixtures, if all particles are of low electrical conductivity, the mixture heats
slowly due to high effective resistance, but the particles still heat faster than the fluid.
However, if a single particle of unusually low electrical conductivity enters the heater, it will
thermally lag the fluid since the current has alternate low-resistance pathways around it.
Under these conditions, the potential for underprocessing exists. Particle concentration has
been found to be important in determining whether or not particles heat faster than fluids.
Under low concentrations, particles will typically lag fluids, while high concentrations favor
faster particle heating. These findings have been verified experimentally in a static ohmic
heater. Conditions involving a radial velocity profile are discussed.

INTRODUCTION

Recent industry interest in continuous sterilization of particulate foods has focused much
attention on the technology of ohmic heating, in which liquid-particle mixtures are heated by
passing an electrical current through them. The resulting internal generation has been
770
reported to cause rapid and uniform heating of foods, and the technology holds promise for
continuous sterilization. However, fundamental understanding of the process is limited.

MA THEMATICAL MODEL FOR OHMIC HEATING

The energy generation rate during ohmic heating depends on the field strength (vV) and a
temperature-dependent electrical conductivity.

where <10 is the electrical conductivity at QOC, m a temperature coefficient, and T the
temperature. When a solid-liquid mixture is heated, the relative rates of heating depend on
the electrical conductivities of the respective phases. Fluid temperatures depend on the
energy generation rate, extent of fluid mixing and heat transfer from solids. Solid
temperatures can be calculated from the conduction heat transfer equation for a solid with
internal energy generation. The voltage field is subject to two types of variations: large scale,
due to the medium being heated along the heater length, and smaller scale variations due to
differences in phase conductivities. Large scale effects result in the major temperature
changes over the heater geometry. Small-scale effects depend on the extent of fluid mixing
in the particle vicinity, and are typically transient in character, depending on the local
particle/fluid structure, temperatures and relative movement. Formulations for prediction of
fluid and solid temperatures in continuous flow heaters in plug flow have been published [1].
A more recent model analyzes the situation where a radial velocity profile exists in the flow.

IMPORTANT RESULTS
Results of simulations have shown that the fluid and particle heating rates depend on the
electrical conductivities (0') and volume fractions of the phases. If particles are less
electrically conductive than the fluid, they can thermally lag the fluid, unless the volume
fractions are sufficiently high. Under this condition, the particles can heat faster than the
fluid. One critical condition involves the presence of an isolated low electrical conductivity
particle in an environment of high electrical conductivity. Under these conditions, substantial
thermal lags can occur. In addition, (within a static heater) the isolated low-conductivity
particle can alter the electrical field around it to create localized hot and cold zones [2]. The
 771
thermal nonuniformity is minimal if sufficient fluid motion occurs.
From the standpoint of process safety, it is necessary to consider the case of the particle of
the lowest possible <1, within a medium of the highest possible (J for a given mixture. If a
residence time distribution exists, the worst-case might be that of the isolated low-
conductivity particle that is also the fastest in the mixture. Orientation effects will of course
have effects, with long-thin particles being significantly affected by being oriented either
parallel to or perpendicular to the field.

Studies within static and continuous flow heaters indicate that some key differences exist
between the two types of heaters, particularly from the standpoint of fluid motions.
Consequently results obtained from one type of heater should not be extended into another
type without adequate accounting of these differences. Additionally, it is important to note
that the behavior of multiple particles is substantially different from those of single particles,
hence care needs to be exercised in extending work with limited particle populations to
conditions of high solids concentrations.

CONCLUSIONS
Critical factors associated with process design involve the condition of the lowest electrical
conductivity phase being surrounded by that of highest electrical conductivity. Additional
safety considerations arise if this particle is also the fastest moving. Particle volume fraction
has significant effects both from the standpoint of internal energy generation and flow
patterns about the particles.

REFERENCES
1. Sastry, S.K. A model for heating of liquid-particle mixtures in a continuous flow
ohmic heater. J. Food Proc. Engr., 1992. 15, 263-278.

2. Fryer, P.J., de Alwis, A.A.P., Koury, E., Stapley, A.G.F., and Zhang, L. Ohmic
processing of solid-liquid mixtures: heat generation and convection effects. J. Food
Engr., 1993, 18, 101-125.
QUALITY CHANGES OF ASEPTICALLY PACKED APPLE JUICE DURING
STORAGE AND THE PREDICTION OF ITS SHELF -LIFE

Shin-Hwei Yuo, Sun-San Lin, Sue-Ywe Chen, Ching-Chuang Chen, and Chu-Chin Chen
Food Industry R&D Institute
P.O. BOX 246, Hsinchu, Taiwan, Republic of China

ABSTRACT
The effects of storage time and temperature on the quality of aseptically-packed single-strength
apple juice were investigated by estimating the changes of sensory characters and chemical
components. The concentrations of 5-hydroxymethyl-furfural and furfural increased with
storage temperature and time, the rate of fonnation followed zero order kinetics with activation
energies of 25.49 and 14.94 KcaVmol, respectively. The Hunter L, a, b values did not change
significantly when stored at low temperatures, i.e.. 4, 20 ·C. The Hunter L value decreased
while the Hunter a value and absorbance at 420nm increased along with storage time when
stored at 37 ·C. The concentration of sucrose decreased along with storage time when stored
at 37 ·C, while the concentrations of glucose and fructose increased. Storage time exhibited a
significant effect on the reduction of malic acid; however, tartaric acid did not change. Sensory
scores decreased during storage and changes in the scores of samples stored at 37 ·C were
greater than those under other storage temperatures. The QI0 values of flavor scores of
samples stored at 4-14 ·C, 14-24 ·C, and 25-35 ·C were 1.652, 1.597, and 1.549
accordingly. The predicted shelf-life of aseptically packed apple juice stored at 37 ·C, 25 ·C,
20 ·C, and 4 ·C was 30.8, 52.0, 68.2, and 161.4 weeks, respectively.

INTRODUCTION
Recently, the consumption of fruit juice in Taiwan has been growing rapidly and most of the
juices were diluted products with 30% juice contents or less. Fruit juices supplied by local
manufacturers include guava, carambola, and passion fruit, and the popular imported juice
consists of orange, apple, grape, and peach concentrates. Non-enzymatic browning (NEB)
reaction is one of the primary cause of quality change in juice during thermal processing and
storage, which is a combined effect of the Maillard browning reactions, degradation of
ascorbic acid, and caramelization of sugars. Furfural and 5-hydroxy-methyl furfural (HMF)
are two known intermediates of NEB reactions. As a consequence, the accumulation of
furfural or HMF is a good indicator for the extent of NEB reactions in juice (4). The
concentrations of these compounds can also be established for estimating the shelf-life of juice
products. Therefore, the purpose of the present study was to analyze the quality changes in
aseptically packed apple juice during storage, coupled with the adequateness of various
chemical and physical parameters to estimate the quality and the shelf-life of aseptically packed
apple juice.
 773
MATERIALS AND METHODS

All chemicals were purchased from E. Merck. All solvents for HPLC were HPLC grade and
were purchased from Fisher Scientific. Apple juice concentrate (70· Brix) was obtained from a
local supplier (Chou Chin Industrial Corp., Changhwa, Taiwan). Aseptic packages (ca.
400mL) were acquired from PKL Corp. Apple juice (13.3 Brix) was diluted from concentrate
with suitable amount of added apple flavor (0.03%, Haannann & Reimer Ltd., Holzminden,
Gennany). Condition for aseptic processing (85·C, 45s) and packaging (ca. 350mL per
package) of apple juices were cited elsewhere (2). Aseptically packed apple juice was divided
into four groups and stored immediately at 4 ·C, 20 ·C, 25 ·C, and 37 ·C for 24 weeks.
Absorbance of apple juice at 420nm was determined on a Hitachi U-200 UV-Visible
Sepectrophotometer. The Hunter L, a, and b values were determined with transmittance
measurement on a Color and Color Difference Meter ("Color Ace" model TC-l). HPLC
analyses were conducted on a Jasco 800 HPLC system equipped with an 830-RI detector and
an 875-UV detector. The procedures for the analysis for furfural and HMF were as described
(4). Mobile phases used for sugars and organic acid were 80% acetonitrile and 1%
(NH4)H2P03 solution (pH = 2.4), respectively.
Sensory evaluations using 9 point hedonic scale were conducted by 12 panelists.
Duplicate juice samples were evaluated at each time.

RESUL TS AND DISCUSSION

The concentrations of HMF and furfural in aseptically packed apple juice were gradually
increased during storage (Fig. 1), which agreed with a previous report (1). Kinetic studies
indicated that the formation of both HMF and furfural followed a zero order reaction, with
activation energy (Ea) of 25.49 and 14.94 kcallmole, respectively. The result agreed with a
previous study (5). According to the result of statistical analysis, the concentrations of both
HMF and furfural were inversely related to the sensory scores of apple juice. The Hunter L
value of apple juice stored at 37 ·C was significantly different from those stored at lower
temperatures (Table. 1). For apple juice stored at 37 ·C, sucrose concentration decreased
gradually with concomitant increase of fructose and glucose. The result agreed well with a
previous report (1). It was also found that the concentration of malic acid decreased gradually
during storage; however, tartaric acid did not change significantly.

60 4.0
FUR
,-.50 4·C
e 3.0
S40 20·C
c::
0
.~ 30 25·C 2.0
l:1
820
§ 1.0
UlO

0 0.0
0 5 10 15 20 25 0 5 10 15 20 25
Weeks Weeks

Figure 1. The change of 5-hydroxymethyl furfural (HMF) and furfural (FUR) concentrations
in aseptically packed apple juice during storage at various temperatures.
774
After 24 weeks of storage, the sensory scores of samples stored at 37 ·C were closer to
the acceptable level (Le. 4). Juice stored at lower temperatures after 24 weeks of storage
appeared to be more acceptable to panelists than those stored at 37·C. Flavor scores were
selected as the parameters for the estimation of the shelf-life of aseptically packed apple juice.
With the Arrhenius equation, the activation energy was estimated to be 7.169 kca1/mole and the
QlO values from 4 ·C to 37 ·C ranged from 1.652 to 1.549.
A sensory flavor score of 4.0 w~s chosen as the minimum score of acceptability, thus the
estimated shelf-life of aseptically packed apple juice stored at 37 ·C was 30.8 weeks. The
shelf-life of other samples stored at lower temperatures was calculated using the same shelf-life
value of those stored at 37"C and applied to the previously described equation (3). The
estimated shelf-life of aseptically packed apple juice stored at 25·C, 20·C, and 4"C was 52,
68.2 and 161.4 weeks, respectively.

TABLE 1
Effects of storage temperatures on Hunter L, a, and b value and absorbance value at 420nm of
aseptically packed apple juice

Temperature L a b A420
4·C 81.232a -1.572a 29.033a 0.487a
20·C 80.817a -1.187ab 28.834a 0.461a
25·C 80. 159a -0.940b 28.954a 0.470a
37·C 77.984b O.lOOc 29.414a 0.519b

NOTE: Duncan's multiple range test, data within the same column with
the same letter are not significant different from each other for P < 0.05.

ACKNOWLEDGMENTS

This research was supported by Ministry of Economy (No. 91T8112B3), Republic of China.

REFERENCES

1. Babsky, N.E., Toribio, J.L. and Lozano, J.E. Influence of storage on the composition of
clarified apple juice concentrate. J.Food Sci., 1986,51,564-567.

2. Chen, C.C., Lin, S.J., Chen, S.Y., Chen, C.C., Zen, S.M. and Jeng, J.G., Development
and evaluation of a laboratory scale aseptic processing and packaging system. Research
Re.port 633-3, Food Industry R&D Institute, Hsinchu, Taiwan, ROC., 1991.

3. Labuza, T.P. and Schmidl, M.K. Accelerated shelf-life testing offood. Food Tech., 1985,
39,57 - 64.

4. Nagy, S. and Dinsmore, H.L., Relationship of furfural to temperature abuse and flavor
change in commercial canned single strength orange juice. J. Food ScL, 1974,39,1116-
1119.

5. Resnik,S.and Chirife). Effect of moisture content and temperature on some aspects of

nonenzymic browning in dehydrated apple. J. Food ScL, 1979,44,601 - 605.
COMPARISON OF mE IMMERSION BIOTEST AND TIlE SPORE-TEST METHOD FOR
EVALUATING THE INTEGRITY OF SEALS OF FLEXIBLE RETORT PACKAGES

GUN WIRTANEN, EERO HURME, RAUA AHVENAINEN,

LENA AXELSON-LARSSON· AND roNA MATTILA-SANDHOLM
VTT Food Research Laboratory, P.O. Box 203, SF-021S1 Espoo, Finland
• Packforsk, Swedish Packaging Research Institute, P.O. Box 9, S-I6493 Kista, Sweden

ABSTRACT

Penetration of vegetative cells of Enterobacter aerogenes and spores of Bacillus subtilis was
investigated through microholes in the seal area of retort packages. The packages were filled with
mashed potatoes. Defects in the packages used in the biotest evaluation were made before sealing by
pushing a wolfram thread of either 50 pm or 100 pm diameter through the seal. In the biotest with E.
aerogenes the packages were immersed for 1 h and in the spore test with B. subtiUs the solution was
left on the seal area for 30 min. The packages were then dried and stored at 30 °C for 6 and 20 days.
The results demonstrated that the storage time did not influence the outcome of the positive results.
The effective diameter of the microholes in the seal was measured by an electrolytic test, Microhole
Tester. When the microhole size was less than 100 pm the spore test gave more positive samples than
the test with vegetative cells and when the size was greater than 100 pm all samples were positive
with both test methods.

INTRODUCTION

The destructive or non-destructive testing of containers during and after processing is one way of
demonstrating the integrity of the packages. In order to select and develop non-destructive leak-testing
methods for flexible and semi-flexible packages it is essential to obtain information, e.g. by biotesting,
concerning the critical microhole size for bacterial penetration [1]. The aim of this study was to
compare the spore test and the immersion test in order to determine which biotest method gives more
accurate information about the integrity of packages and which is less time consuming in practice.

MATERIALS AND METHODS

Packaging of the Model Foodstuff

Mashed potatoes were packed in 200 m1 polypropylene-based containers with ethylene vinyl alcohol
as a barrier layer (Bebo Plastik, Germany). Tungsten thread of either 50 pm or 100 pm in diameter
was placed on the seal area before sealing and then the filled containers were sealed in a commercial
776
processing line (Lieder Maschienenbau, Gennany) and autoclaved. After autoc1aving the threads were
withdrawn from the seals and the containers were tested using either the immersion test or the spore
test. Some of the containers were tested with blocked microholes, the blocking being perfonned with
agar. Intact containers with no holes in the seal area were used as reference packages.

Immersion Biotest with Enterobacter oerogenes

A modification of the biotest method proposed by the National Food Processors Association and
ASTM in the USA was used. In the test the containers were placed in a bath of fresh Enterobacter
aerogenes (ATCC 13048) suspension containing approximately 107 cfu/ml and incubated at 24 ± I
°C for 60 min. Thereafter the containers were removed from the bath, wiped and cleaned and
transferred to the incubator, where they were incubated upside down at 30°C. The samples were
withdrawn after incubation of either 6 or 20 days for investigation of bacterial growth. The containers
were emptied and washed before seal defect testing with the Microhole tester (Packforsk, Sweden).

Spore Biotest with BaciUus subtilis

In the spore test the containers were placed with the longer seal facing up. A suspension containing
spores of Bacillus subtilis (Merck 10649) was pipetted into the seal area and left there for 30 min at
24 ± 1°C. A solution containing about l(f spores/ml sterile distilled water was used. The containers
were further treated as described in the section on the immersion test above.

Confirmation of Bacterial Growth

The bacterial growth of E. aerogenes was confinned in tubes containing Brilliant Green Bile (Difco,
US) solution by gas production in Durham tubes. Positive samples were plated on Levine EMB agar
(Difco, US), where the test organism fonns colonies with grey-brown centres. E. aerogenes is a Gram
negative rod, which was confinned by Gram staining.
The samples taken from containers tested in the spore test with B. subtilis were cultivated on
blood agar plates (Orion Diagnostica, Finland). Hemolysis of the blood was observed around the
colonies on the plates from positive samples. Cells of B. subtilis were also tested by Gram staining
and identified as Gram positive rods.

Measurement and Confirmation of Microholes

The effective diameter of microholes was measured by an electrolytic test (Microhole tester). The
method is based on the fact that the electrical conductance of a package of insulating material is
drastically altered by a small hole [2]. The seals were also tested with a dye test (Ageless Seal check).

RESULTS AND DISCUSSION

The results obtained with the samples tested with the two methods and the two incubation times are
presented in Table 1. Prolongation of the incubation time from 6 to 20 days had no effect on the
outcome of the case of positive samples. When the effective diameter of the microhole measured by
the Microhole tester was less than 100 pm the spore test gave more positive results. A comparison of
the two biotest methods with open and blocked microholes is presented in Table 2. When the size of
the microholes was less than 100 pm the number of positive samples was higher in the spore test than
in the immersion test. Blocking of the microholes had no effect on the number of positive samples
obtained with the two methods. In fact, it is possible that. the blocking agar enabled the motile E.
aerogenes to enter the package. Containers with very small microholes were not tested with E.
aerogenes. When the effective diameter was greater than 100 pm all samples were positive with both
test methods. Microholes were also detected in the control containers with "intact" seals. The spore
test appears to be very sensitive, because some packages with no microholes detected in the
measurement using the Microhole tester showed positive reactions in this test. According to this study,
testing of the integrity of packages should be based on more than one method in order to confinn the
product quality depending on seal integrity.
 777
TABLE 1
The occurrence of bacterial growth in autoclaved plastic containers filled with mashed potatoes and
biotested using two methods. the spore test with spores of Bacillus subtilis and the immersion test
with Enterobacter aerogenes. The microholes in containers both with intact seals and leaking seals
made using tungsten thread of diameter 50 pm were measured with the Microhole tester.

Effective Number of contaminated I biotested packages in

diameter of biotest with B. subtilis biotest with E. aerogenes
microhole (pm)
Storage after biotesting Storage after biotesting
6 days 20 days 6 days 20 days

0 6(25 2(22 1(26 0(25

1-25 6/8 3/6 2/4 2n
26-50 515 2/5 1/3 6n
51-100 8/9 12/12 316 6n
101-150 2(2 1/1 2(2 1/1
151-200 1/1 nd 4/4 nd
>201 nd nd 6/6 515
total 28/50 20/46 19/51 20/52

TABLE 2
As in Table 1 except that the microholes were made using tungsten thread of diameter 100 pm and
the microholes in half of the samples were blocked with agar. The incubation time was 6 days.

Effective Number of contaminated I biotested packages in

diameter of biotest with B. subtilis biotest with E. aerogenes
microhole (pm)
Open Blocked Open Blocked
microholes microholes microholes microholes

0 1(2 nd 0/11 nd
1-25 9/14 nd 0/11 nd
26-50 1/9 4/4 0(2 nd
51-100 3/4 4/4 1/3 nd
101-200 6/6 7n 1/1 nd
201-300 8/8 515 12/12 515
301-400 9/9 4/4 6/6 7n
>400 nd 1/1 6/6 13/13
total 37/52 25(25 26/52 25(25

REFERENCES

1. Ahvenainen. R.. Mattila-Sandholm. T .• Axelson. L. and Wirtanen. G .• The effect of microhole size
and foodstuff on the microbial integrity of aseptic plastic cups. Pack. Techno!. Sci.. 1992. 5. 101-
107.
778
2. Axelson, L., Cavlin, S. and Nordstr(jm, J., Aseptic integrity and microhole detennination
of packages by electrolytic conductance measurement. Pack. Techno!. Sci., 1990, 3, 141-162.
 FOOD PACKAGING AND SHELF-LIFE

IV AN VARSANYI
Canning Technology Department,
University of Horticulture and Food Industry,
Menesi ut 45, H-ll18 Budapest, Hungary

ABSTRACT

The results of our research call attention to the relationship between the shelf-life of packed food items
and packaging. Mathematical models of deterioration allow prediction of the shelf-life of foods when
the packaging and the circumstances of storage do not change. The packaging can protect the product
from outside effects but the modification of consequences on the hygienic state and on the reactivity
of the food matrix is limited by packaging techniques.

INTRODUCTION

One of the most important tasks of food packaging is to ensure the quality - nutritive and sensory
values - and quantity of packed food items from production (harvesting) until consumption. Quality
assurance is a very complex task because the foods are mainly bio-products and consequently they
are unstable, 'living' and changing materials.
To inhibit or reduce the quality changing rate of packed food, with reference to the shelf-life
declared, we have to choose the most appropriate packaging material and technique (e.g. modified
atmosphere packaging). We investigated various products of the food industry to determine the main
factors that change quality and the effect of packaging on shelf-life.
The selection of the most appropriate packaging material depends on the food matrix. Deterioration
of a chemical nature may be caused by internal and external factors, for example oxygen molecule
uptake, double bond changing, or molecule chain breaking.
Food deterioration of physical origin i~ mainly caused by temperature and by micro- and
macroclimate differences between the package and storage room. This causes water adsorption or
water desorption; rheological changing, or modification of physical state.
Very frequently food deterioration is of biological origin. The quality change is caused by
mUltiplication of microorganisms, by toxin production or by enzymes or enzyme systems. It is
important however, to emphasize that the activity of enzymes and microorganisms depends on the
water activity of the food and the temperature of the storage room, which belong to the physical
effects.

MATERIALS AND METHODS

The shelf-life is defined as the time during which the most rapidly changing (deteriorating) important
property or properties of the food change until the well defined value limits of the characteristic
780
property, which are described in the standards in given package under certain storage conditions, are
exceeded.
A continuous mathematical model was set up to follow the quality change of foods and in this way to
determine the shelf-life of packed products. Various physical, chemical and microbiological methods
were used to select the most rapidly changing property(ies) of packed food. For that selection we used
comprehensive mathematical statistics, and two-way variance analyses were applied to determine the
effects of packaging, storage time, temperature, etc. Of course the homogeneity of variances was also
investigated. The number of measurements was 3-5 parallel for 2-3 repetitions.

RESULTS

Analysing the reaction kinetics and mechanism of food deterioration, we found that they may be
followed with five different mathematical models. The value of regression constant (a) is identical with
the critical characteristic value measured at the start of storage and the regression coefficient (b)
defines the rate of deterioration.
A first order polynomial (linear type function) described the rate of deterioration, where the
changing of the critical property (y) is constant in the function of storage time (x). A second order
polynomial (quadratic type function) is suitable for following the quality change of foods stored mainly
between -lOoC and +5°C. The rate of deterioration, described by exponential function, changes
rapidly as a function of storage time. The origin of deterioration is chemical and/or biochemical. The
sigmoid type deterioration is characteristic in about one third of processed foods. The quality generally
changes as a combined effect of several factors (physical, chemical, microbiological). A relatively small
group of food items change as a function of storage time using a hyperbolic function.

CONCLUSIONS

Quality protection of foods needs a suitable and marketable packaging. Quality protection of foods
and the barrier properties of packaging have a tight correlation, with special regard to the structure of
polymers. Therefore an organic part of our research programme was to investigate the correlations
between the kinetics and mechanisms of food deterioration and the packaging quality. Surveying the
results of our investigations it may be stated that the mechanism of deterioration and the packaging
quality. Surveying the results of our investigations it may be stated that the mechanism of deterioration
does not change significantly as a function of packaging but the rate of deterioration is variable. This
means that we can modify the shelf-life of foods (reduce or extend) by packaging methods and by the
quality of packaging materials and by the auxiliaries. The value of permeability constant (P) and the
shelf-life of packed foods have a stochastic relationship which depends on the food matrix and the
barrier properties of packaging.
 MODIFIED ATMOSPHERE AND MODIf1ED HUMIDITY PACKAGING TO EXTEND
TIlE SHELF LIFE OF FRESH MUSHROOMS (Agaricus bisporus)

S. ROY, R. C. ANANTHESWARAN & R. B. BEELMAN

Department of Food Science
The Pennsylvania State University
University Park, PA 16802, U.S.A

ABSTRACf

Modified atmosphere packaging (MAP) increased the shelf life of fresh mushrooms up to 9 days
at 12°C. Sorbitol, a food grade moisture absorber, was used to control the relative humidity
within packages. Less sorbitol was required in MAP than in conventional packages. Optimum
O 2 concentration and optimum relative humidity in MAP was found to be 6% and 90%,
respectively.

INIRODUCTION

The shelf life of fresh mushrooms (Agaricus bisporus) is limited to 1-3 days at ambient
temperature. Retardation of metabolic processes using modified atmosphere packaging (MAP)
will increase the shelf life of mushrooms. The mushrooms stored in lower oxygen concentration
was found to have better color, lower maturity index, and lower weight loss and disease incidence
than those stored in higher oxygen concentration (1). Condensation of water in MAP was
successfully avoided by using moisture absorbents to lower the In-package relative humidity
(IPRH) of packages containing mature green tomatoes (2). In this present study, optimum O 2
concentration in MAP was determined and the effect of sorbitol as a moisture absorber on the
shelf life of mushrooms in packages was evaluated.

MATERIALS AND METIIODS

Mushrooms (Agaricus bisporus) were packaged in conventional packages (CP) and MAP. In CP,
the mushrooms were placed in 600 ml polystyrene trays overwrapped with PVC film. The film was
punctured with two 3 mm holes. In MAP, mushrooms were placed in 1000 mt trays. A 6O-gauge
polyethylene film (Cryovac Inc., Duncan, SC) was used to make a.pouch. The polystyrene trays
containing mushrooms were inserted into the pouch and heat sealed. The moisture absorbers
were sealed in a Tyvak paper pouch which in turn were placed in the tray underneath the
mushrooms. The packaged mushrooms were stored at 12°C chamber with 70% RH and were
analyzed for quality after 3, 6 and 9 days of storage. The maturity index was determined
according to a 7 point scale (1= least mature and 7= most mature) (3). Surface color of
782
mushroom caps was measured using a Minolta Chroma Meter (Minolta CR-200). The surface
moisture content of mushrooms were measured using a near infrared (NIR) spectroscope
(NIRSystems model 6500) (4). The oxygen and carbon dioxide concentrations in the packages
were monitored every day, for a period of 7 days, using a Gas Chromatograph (Hewlett Packard
5890 Series 11). A Vaishala HMP 23 UT RH probe (Campbell Scientific, Logan, U1) was
inserted in packages containing 5, 10 and 15 g of sorbitol and IPRH was monitored at 12°C for
a period of 7 days.

RESULTS AND DISCUSSION

The oxygen concentration of MAP containing different amount of mushrooms are shown in figure
la. The packages containing 100 and 120 g of mushrooms (which reached a steady state O2
concentration of approximately 6% and 2% respectively) did not mature during ~orage (p>0.1)
(figure 1b).

( a) 4 b
25
- 50 g mushroom
'0 - 50 g mushroom --0- 80 g mushroom
~ --0- 80 g mushroom --0- 100 g mushroom
0
> 20 --0- 100 g mushroom --tr- 120 g mushroom
~ --tr- 120 g mushroom

r:::

..
~
ai
C
15

CD
u 10
r:::
0
u
r:::
CD 5
CI
>-
)(
0
1+-----~------~----~----__,
0
0 2 3 4 5 6 7 3 6 9
Days of storage Days of storage

Figure 1. (a) Oz concentration during MAP (b) Effect of days of storage (Oz
concentration) on maturity

Mushrooms in steady state O 2 concentrations of 15% and 10% had higher maturity after 6 and
9 days of storage, respectively. Mushrooms at 6% steady state O 2 concentration had the best
color values at the end of 6 days of storage.

Mushrooms in CP containing 15 g of sorbitol had best color throughout the storage

period. There were no differences in maturity index or color between mushrooms with or without
sorbitol during MAP. Mushrooms packaged without sorbitol and with 5 g of sorbitol had higher
surface moisture content after 9 days of storage than after 3 days of storage. The surface
moisture content of mushrooms packaged with 10 and 15 g of sorbitol remained constant during
storage. The IPRH with different amount of moisture absorbers is shown in figure 2a.
The packages containing 10 g of sorbitol had an IPRH of 88-90% which was therefore considered
optimum to store mushrooms in MAP.

Shelf life of mushrooms in MAP was compared with that in CP, both with and without
sorbitol in the' packages. MAP mushrooms, both with and without sorbitol, had significantly higher
 783
(a) (b)
100
0-5 g sorbitol
+ 10 g sorbitol 110
0-15 9 sorbitol

l:
~

~ 90
....,
:s
80

..,
J: iii
>
~
..,
-
...J
.!!
a: 70
pVC-no Sorbitol
- 0 - PVC-15g Sorbitol
- 0 - MAP·15g Sorbitol
~ MAP·no Sorbitol

80 60
0 2 3 4 5 6 3 6 9
Days of storage
Days of storage

Figure 2(a) Effect of sorbitol on IPRH (b) Effect of packaging and

in MAP sorbitol on L value

L values (figure 2b) than those in CP (p<O.I). After 6 days of storage, the maturity index of
mushrooms in MAP was significantly lower than those in CP (P<O.I). MAP mushrooms with 15
g sorbitol had higher L values and significantly lower AE values than the rest of the treatments
after 6 days of storage (p<O.I).

CONCLUSIONS

Modified atmosphere and modified humidity packaging improved the shelf life of mushrooms to
9 days at 12°C. The color of mushrooms were better when sorbitol was used in conventional PVC
overwrap packages. Lowering the O 2 concentration to 6% in MAP mushrooms retarded the rate
of maturation, but lowering the O 2 concentration to 2% increased the rate of browning. Less
amount of sorbitol was necessary in MAP than in conventional packages.

REFERENCES

1. Burton, K. S. Frost, C E., and Nichols, R. A combination of plastic permeable film system for
controlling post-harvest mushroom quality. Biotechnology Letters. 1987. 9(8):529-534.

2. Shirazi, A Modified Humidity Packaging of Fresh Produce. Ph.D. Dissertation. Michigan State
University, East Lansing, ML 1989.

3. Guthrie, B. D. Studies on the control of bacterial deterioration of fresh, washed mushrooms

(Agaricus bisporuslbrunescens). M.S. Thesis, The Pennsylvania State University, University park,
PA 1984.

4. Roy, S., Anantheswaran, R. C, Shenk, J. S., Westerhaus, M. 0 and Beeiman, R. B.

Determination of moisture content of mushrooms using VIS-NIR spectroscopy. L Sci. Food
Agric. 1993 (in press).
DESIGN OF PERFORATED POLYMERIC PACKAGES FOR THE MODIFIED
ATMOSPHERE STORAGE OF BROCCOLI

JATAL D. MANNAPPERUMA AND R. PAUL SINGH

Department of Biological and Agricultural Engineering
University of California, Davis, CA, USA

ABSTRACT
The optimum modified atmospheres required for the extension of shelf life of fresh produce
span a wide range of C02 and 02 compositions. However, the atmospheres that can be
established inside intact polymeric packages are limited by the narrow range of the permeability
ratio of polymeric films. This limitation can be overcome by using perforations. A
mathematical model to aid in the design of perforated polymeric packages was developed. The
predictions of the model were verified by an experimental study using broccoli.

INTRODUCTION
The commercially available polymeric films span a wide range of gas permeabilities. However,
the ratio of permeability of C02 and 02 for most of the polymers fall within a narrow range
around 3 to 6. This limits the atmospheric compositions obtainable through the use of intact
polymeric packages also to a narrow range. This range covers the complete range of 02 but
does not allow C02 contents above 5%.

It is possible to establish high C02 - low 02 atmospheres through the introduction of

perforation in the polymeric packages. Although this possibility was known for sometime [4],
package design methodology incorporating this concept were never fully developed. The
objective of this study was to develop a method for the design of perforated polymeric
packages to establish high C02 - low 02 atmospheres.

Literature Review
Tomkins [4] investigated the effect of several design parameters on the atmosphere inside a
polymeric package of fresh produce. These parameters included, weight of fruits, type of film,
temperature and number of perforations. The perforations lowered the equilibrium C02 level
and raised the 02 level relative to intact packages. The drop in C02 level was approximately
equal to the rise in 02 level. Use of two films with widely different permeability ratios has
been suggested [5]. A simple mathematical model based on steady state mass balances was
used to evaluate package design parameters. Polyethylene and vegetable parchment films
(permeability ratios, 3.9 and 0.9) were used to construct experimental packages for Macintosh
apples with target package atmosphere 7.6% 02 and 14.0% C02. The experiment verified the
design through close agreement of the steady state package atmosphere [5]. A combination of
 785
relatively impermeable oriented polypropylene film and a microporous film has been used to
package mushrooms [1]

The basic principles involved in the design of modified atmosphere packages have been
previously illustrated by Mannapperuma and Singh [2] using a plot of the recommended
modified atmospheric windows on a two dimensional chart with 02 concentration as the
abscissa and C02 concentration as the ordinate. A simple mathematical model was used in
conjunction with this plot to illustrate the effect of permeability ratio, respiratory quotient,
temperature and perforations on the package atmosphere. The present study is a logical
extension of this work.

Mathematical Model
When fresh produce is stored in a package, the respiration activity lowers the 02 concentration
and elevates the C02 concentration in the package relative to the ambient atmosphere,
establishing a concentration difference. This concentration difference creates a flux of oxygen
into the package and a flux of carbon dioxide out of the package. In a properly designed
package under steady state conditions, these two fluxes should be equal to the 02 consumption
and the CO2 generation by the produce in the package, respectively. The design parameters of
the package are determined using these equalities. An appropriate mathematical model was
developed using perforations and fluxes of gases through the perforations. The solution of this
mathematical model provided following solutions for the weight of fresh produce and area of
perforations required to create the known optimum atmosphere in polymeric package using a
film with known area and thickness.
P A f3-f3
W =....2.LL f p (cI - Xl ) 1
Rlbf f3-f3p
Ap = Af Plf f3 f - f3 p 2
PIp f3-f3p

Where: A is Area of the film (m2); Pis permeability ratio (ratio of permeability of CO2 to 02,
dimensionless); b is thickness of the film ()lm); c is gas concentrations in ambient (% atm); P
is permeability (ml-)lm!m2-h-atm); R is respiration rate (ml/kg-h);W is weight of produce in the
package (kg); x is gas concentrations in package atmosphere (% atm); suffixes 1 and 2 denote
oxygen and carbon dioxide; suffixes f and p denote film and perforations.

EXPERIMENT AL STUDY

The experimental study involved the design of a perforated package for broccoli. The
recommended modified atmosphere for storage of broccoli [3] provided values for xl, x2 and
p. xl = 2.0%; x2 = 9.0%; ~nd /3 = 2.11. The respiration rate of broccoli near optimum
conditions from the literature was used to calculate values for Rl and R2. Rl = 9.5 ml/kg-h;
and R2 = 9.5 ml/kg-h
A low density polyethylene film of 100 )lm thickness, and a packa~e size of 200 mm x 200
mm was selected resulting in the values for Af and bf. Af = 0.08 m ; and bf = 100 )lm. The
permeability of this film at 2.5 C was determined experimentally. Plf = 2120 ml-llm/m2-h-
atm, P2f = 10470 ml-)lm!m 2-h-atm, and Pr= 4.94
786
Substitution of these values in equations I rd 2 results in a design weight of 120 grams of
broccoli and a perforation area of 6700 11m . The experimental study was planned based on
these calculations.

A metal wire heated by discharging a capacitor was used to make the perforations with an
average diameter of 75 Jlm (Area=4500 11m2). Replicate packages of 120g of broccoli with 1,
2 and 3 holes were prepared and stored in an incubator controlled at 2-3 C temperature. The
package atmosphere was monitored at weekly intervals until the steady conditions were
reached.

RESULTS AND DISCUSSION

The equilibrium atmospheres inside packages with 11 2 and 3 holes were as follows. 1 hole:
4500 11m2, Xl =1.2%, x2=5.6%; 2 holes: 9000 11m ' Xl =3.3%, x2=8.8%; 3 holes: 13500
11m2; Xl =4.9%, X2=8.2%

The atmospheres inside packages with 1 and 2 holes were close to the design target atmosphere
ofx1=2.0% and x2=9.0%. The deviation from the design values and the high variation among
experimental values were attributed to non uniformity of perforations.
Improved techniques for making perforations should result in better agreement with the
predictions by the model. The method described in this study can be used successfully to
achieve optimum modified atmospheres of a wide range of fresh fruits and vegetables.

REFERENCES

1. Burton, K. F., Frost, C. E. and Nichols. R. 1987. A combination plastic permeable film
system for controlling post-harvest mushroom quality. Biotechnology Letters, 9(8):529

2. Mannapperuma, J. D. and Singh, R. P. 1993. Modeling of gas exchange in polymeric

packages of fresh fruits and vegetables. Technical Report, University of California, Davis,
CA.

3. Saltveit, M. E. 1989. A summary of requirements and recommendations for the controlled

and modified atmosphere storage of harvested vegetables. Fifth Proceedings of The
International Controlled Atmosphere Research Conference. June 14-16, Wenatchee, WA.

4. Tomkins, R. G. 1962. The conditions produced in film packages by fresh fruits and
vegetables and the effect of these conditions on storage life. Journal of Applied
Bacteriology. 25(2):290.

5. Veeraju, P. and Karel, M. 1966. Controlling atmosphere in a fresh fruit package. Modem
Packaging, 40(2); 166.
 PROTECTION AGAINST PERISHABLENESS -
NEW PACKAGING MATERIALS AND PRINCIPALS OF
PERMEATION MEASUREMENT

JOCHEN HERTLEIN and HORST WEISSER

Institute for Brewery Installations and Food Packaging Technology,
Technical University of Munich, 85350 Freising-Weihenstephan, Fed. Rep. of Germany

ABSTRACT
To ensure food quality and food safety, a large number of plastic barrier materials has been
developed. Good barrier properties are reached by combining properties of different plastic
films and aluminium. Recently a new generation of packaging material is available where a
plastic layer is vacuum coated with a thin layer of either metal or silicium oxide.
The measurement of gas permeation is usually carried out under steady state conditions.
The transient state, before equilibrium is reached, can last for more than 24 hours for high
barrier films. Therefore a numerical model describing the transient state was used to predict
steady state permeability. The duration of experiments can be shortened and more profound
knowledge of the permeation process is won.

INTRODUCTION
One important property of plastics is their permeability to gases and vapors. Because many
food products are sensitive to attack by oxygen and water vapor, a lot of work is done to improve
the barrier properties of plastic films using different techniques: coextrusion of different plastics,
lamination techniques, and vacuum deposition of metals and oxides. Especially the vacuum
deposition gaines importance in food packaging technology as a replacement of aluminium-
plastic laminates [1]. Beside the very good barrier properties, vacuum coated films have several
advantages: They are decorative and the coating is exceedingly thin (around 50 nm) compared
to a 7-12 J.Lm aluminium layer in laminates. The base for the deposited material is PET, BOPP,
OPA, PVC, PE, cellulose, and paper. Some metals that can be used for vacuum coating are
aluminium, nickel, chromium, copper, their alloys, and oxides. Silicium oxide is used in different
ratios of SiO and Si0 2 and is often denoted as SiO x ' Good barrier properties are achieved for x =
1.4-1.8. The wide range of possible polymer-metal combinations and the growing importance of
vacuum coated films in food packaging technology require extensive permeation measurements.
The feasibility of the differential pulse method in combination with the oxygen permeability
test system Mocon Oxtran 100 (ASTM D 3985-81) for plastic films and barrier plastic films was
examined to shorten the duration of permeation measurements.

PERMEATION MEASUREMENT
All test arrangements for permeation measurements can be classified in one of three large groups
of measurement techniques: integral permeation methods, differential permeation methods, and
sorption/ desorption methods. Typical response curves with a transient and steady state portion
are recorded. The permeability is usually determined from the steady state section of these
788
response curves. One interesting alternative is the differential pulse method. Instead of measu-
ring the steady state gas flow through a polymer membrane, the permeability is determined in
a non-steady state experiment [2].

Differential Pulse Method

The most important difference between the traditional techniques and the differential pulse
method is that instead of maintaining a defined partial pressure difference during the entire
experiment, a rectangular partial pressure pulse is sent along one side of the sample. The
pulse is so short that steady state permeation is not reached. Instead, a typical response curve
(marked difference in fig. 1) is recorded. The permeation parameters (diffusion, solubility, and
permeability coefficients) can be calculated by evaluating the peak height F max and half width
to.5' The response curve to the partial pressure pulse can be interpreted as the mathematical
sum of two response curves. The first response curve is the result of a change in the partial
pressure difference from PI to P2, the second of a change from P2 to PI (P2 > pt) (fig. 1). The
latter one is a virtual curve needed to explain the resulting response curve. The time delay
between both curves is identical to the pulse length {}. By introducing dimensionless quantities,
the response signal Y can be written as:
00

Y = 2 I)-It (exp(-n 2 '/r 2 t q) - exp(-n 2 '/r 2 (t q - {}q») (1)

n=I
.h Dt
WIt tq = L2 an d (2)

The maximum response signal Y max and the half width t q ,O.5 can be calculated using equa-
tion (1). These theoretical values are compared with results from experiments (F max, to.5, and
{}). The permeation coefficient P and the diffusion coefficient D can be calculated with:

d
D = L 2 --
tq,O.5
to.5
an (3)

A denotes the sample area and 'T/ the sensor's sensitivity.

Flux (ml/(m 2 bard»

1.0,---~~----~-----------------------------------.

O.S
0.6 response to first pressure change
0.4
0.2 difference
0.0 -~--.:."""""""""""""""""""""""'"'''' .....

-0.2
-0.4
-0.6
-O.S
-1.0
0 500 1000 1500 2000
Time (s)

Figure 1. Response signal versus time for a PETmet!PE (12 p.m/SO Ilm) sample with
pulse time!? = 60 s, D = 2.53 10- 12 m 2 /s, P = S.92 10- 15 ml m/(m 2 Pa s).
 789
RESULTS AND CONCLUSIONS
The most important postulate for the feasibility of the differential pulse method is, that a
rectangular partial pressure pulse can built up at one side of the tested plastic film. It could be
shown in experiments that the Mocon Oxtran 100 A fullfills this important requirement when
the pulse lasts for 60 s and more. The maximum oxygen concentrations measured at the test
chamber entry were 85 to 90% for shorter pulses and the curve was bell-shaped. When the
measured pulse area was evaluated it showed that the area was larger than that for the ideal
pulse by 14 s . bar, regardless of the pulse duration. This means that 14 s had to be added
to all pulses. Other time delays such as time delays in tubes or the response time of the
Sensor did not influence the response curves. Response curves of films with a permeability
<0.5 mIl (m 2 d bar) were too flat for interpretation. Table 1 lists the permeation, diffusion, and
solubility coefficients for several plastic films.
Table 1 shows good agreement for the permeability coefficients determined with the pulse
method and under steady state conditions for the monolayer films. This difference is larger for
the barrier films due to the flatness of the response curve and difficulties in evaluating peak
height and half width. The transport parameters D and P could not be calculated for the
silicium oxide coated PET sample. Still, for a lot of applications the calculated parameters are
precise enough and helpful - especially when a large number of samples has to be compared.
The results shown here were all determined with a 60 s pulse. Longer pulses did not influence
the calculated parameters significantly, shorter pulses should not be used because of the above
mentioned behavior of the test equipment.
The Mocon Oxtran 100 can be used to determine the transport parameters and the oxygen
flux through barriers with the differential pulse method. The time needed to perform traditional
permeation experiments can be reduced at least by a factor of two.

TABLE 1
Values of D, S, and P for different monolayers and vacuum coated films calculated for a
74 s pulse. Test conditions: 23°C and 50% r. h., 10 samples.

pulse method steady state

sample D m' P ( P ( mlm )
• S (m';'~J mlm )
m 2 sPa m ,Pa
2

PVC (4.5 ± 0.1) 10- 13 (8.1 ± 0.5) 10- 1 (3.6 ± 0.3) 10- 13 2.5 10- 13

PA 6 (1.1 ± 0.1) 10- 11 (7.7 ± 0.7) 10- 2 (8.7 ± 0.5) 10- 13 6.8 10- 13

OPP (9.5 ± 0.3) 10- 13 (2.4 ± 0.2) (2.3 ± 0.1) 10- 12 2.6 10- 12

OPAmet/PE (2.6 ± 0.6) 10- 12 (3.0 ± 2.0) 10- 3 (7.0 ± 3.5) 10- 15 2.2 10- 14

PETmet/PE (2.5 ± 0.5) 10- 12 (3.9 ± 2.4) 10- 3 (8.9 ± 4.2) 10- 15 1.1 10- 14

PETsiOx/PE 5.1 10- 15

REFERENCES

1. Jamieson, E. H. H. and Windle, A. H., Structure and oxygen-barrier properties of metal-

lized polymer film. Journal of Materials Science, 1983, 18, 64-80. Die Verpackung, 1992,
33, No.5, 197-200.
2. Pcilmai, G. and 01<ih, K., New differential permeation rate method for determination of
membrane transport parameters of gases. Journal of Membmne Science, 1984,21, No.2,
161-83.
 EDIBLE WHEAT GLUTEN FILMS: OPTIMIZATION OF THE MAIN PROCESS
VARIABLES AND IMPROVEMENT OF WATER VAPOR BARRIER PROPERTIES BY
COMBINING GLUTEN PROTEINS WITH LIPIDS

Nathalie GONTARD, Stephane GUILBERT, Sylvie MARCHESSEAU and Jean·,Louis CUQ

Laboratory of Genie et Technologie Agro-Alimentaire, ClRAD-SAR, BP 5035, 73 rue J. F.
Breton, 34032 Montpellier, France.
and LGBSA, Universite de Montpellier II, place E. Bataillon, 34095 Montpellier, France.

ABSTRACT

An edible wheat gluten film was developed and relationships between film-formation
conditions and properties were studied using Response Surface Methodology. pH and ethanol
concentration of film-forming solution had strong interactive effects on film water solubility
and moisture permeability. Mechanical properties were affected by gluten concentration and
pH. Various formulations and methods of fabricating films consisting of gluten and lipids
were investigated.

INTRODUCTION

Edible films and coatings have been used to protect pharmaceuticals, improve shelf-life and
properties of food products (3, 4). Proteins, especially wheat gluten, as edible film-forming
agents have been studied less extensively than lipids or polysaccharides. The objective of the
presented work was to develop an edible wheat gluten film, to gain a better understanding of
relationships between gluten film-formation conditions and film properties, and to improve
moisture resistance by combining wheat gluten with lipid materials.

MATERIALS AND METHODS

The film was prepared from a film-forming solution which was obtained by dispersing gluten
proteins in ethanol, acetic acid (to adjust the pH) and water. Glycerol (20 % pip gluten) was
added as plasticizer (2). The film-forming solution was used for casting film and allowed to
dry. Puncture test, opacity and water vapor permeability measurements (standard procedures)
were made on films with a controlled constant thickness (0,05mm) and equilibrated at 56%
relative humidity at 25°C (1).

RESULTS AND DISCUSSION

Response surface methodology was used to determine the influence of some film-formation
conditions on film properties. From the statistical results (1), pH and ethanol concentration
(ET) of the film-forming solution had strong interactive effects on film opacity, water
solubility and water vapor permeability (WVP). Mechanical properties seemed to be
particularly affected by the concentration of gluten (GL) and pH.
 791
Figure 1 shows that high GL 02,5%) and pH
above 5 induced high puncture strength which
involve a high number and/or a better
localization of bonds between proteins chains.
During the drying of the film solution, ethanol
and acetic acid, which are responsible for
proteins dispersion, were first evaporated,
allowing the formation of bonds between 5
protein chains. During this stage, the
proximity of protein chains induced by high 4
GL could facilitate and improve the formation :s
of such cross-bond. ~ 3
The film heterogeneity induced by high ET ~
led to complete and rapid disintegration and : g
dispersion in water (figure 2). All of the E 2
effects and interactions resulted in the lowest ~
film solubilities with a minimum value of ~
about 40%, with ET and pH simultaneously
varying between ET= 40% - pH 2 and ET=
5.0
20% - pH 5. Such conditions allowed 7.5
sufficient unfolding of the gluten protein Glut
ell COllce
10.0
chains in the solution and resulted in a water (g/IOO lltratioll
lttL)
resistant film.
The shape of the WVP response surface Figure I. Response surface for the effect of
(figure 3) is characteristic of the strong gluten concentration and pH of the film-forming
interaction between ET and pH. At high ET, solution on film puncture strength at a constant
heterogeneity of the film could explain the ethanol concentration of 45 mL!IOO mL.
ease of water transport. Films formed under
low pH and low ET conditions had very high
WVP. Low pH was certainly responsible for
the unfolding of proteins with exposure of
hydrophilic residues on the protein surface.
The individual negative effects of ethanol and
acetic acid on WVP appeared to be limited by
the simultaneous variation of these two factors
(diagonal). The effect of hydrophilic groups at
low pH seemed to be counterbalanced by the
effect of ethanol, thus improving the exposure
of less polar groups,consequently limiting the
WVP increase. ~ 120

Rather than determining the optimum for all E

responses, it would be preferable to choose ~ 90
particular film-formation combinations based ~~
on specific uses of film. ..
.. 60

To improve the water vapor barrier properties ~ 30 70·?0~

57.5 ~..
of wheat gluten film, gluten proteins have
been combined with lipids materials. Edible o ~y.
45'0c-tlJ ~
composite films constituting of wheat gluten 3
~C
32,5 ,,0-$5
4
as structural matrix and lipids as moisture pl:f
5
~ ~
6 20,0 ~o ~
barrier were developped using emulsion ~ ~
technique. The effect of lipids on the physical ~
properties of composite films depended on the Figure 2. Response surface for the effect of pH
lipids characteristics and on the interactions and ethanol concentration of the film·forming
between the lipid and the protein structural solution on film water solubility at a constant
matrix. gluten concentration of 7.5 g/IOOmL.
Beeswax was the most effective lipid for
decreasing WVP but these films were opaque,
weak and disintegrated easily in water.
 792

Combining wheat gluten proteins with a

diacetyl tartaric ester of monoglycerides
increased resistance and maintained
transparency. Using the coating technique, E_ 1.5
solid lipids such as beeswax or paraffin wax, :B:tf
deposited in a molten state onto the gluten ~ ~ 1.3
base film, were effective water vapor barriers. S ~
A film consisting of wheat gluten, glycerol ~~ 1.1
and diacetyl tartaric ester of monoglyceride as &'8
one layer, and beeswax as the other yielded a ~ a 0.9
WVP of 0,0048 g.mm/m2.mmHg.24h. (table ~ ~
1) which was less than that obtained with low ~!:9
density polyethylene.
3 4
Pli 5

Figure 3. Response surface for the effect of pH

and ethanol concentration of the film-forming
solution on film water vapor permeability at a
TABLE 1 constant gluten concentration of 7.5 g/IOOmL.
Water vapor permeability of edible and non edible films

Film Temperature 6p Thickness Water vapor

permeability
(mmHol (mm) (qrrm' rn2.rrni-tJ.241T.)
Starch 25 19.2-7.3 0.036 29.3
Casein-gelatin 30 28.9-18.5 0.25 7.1
Gluten, Glycerol (16.6%) 30 32.2-0 0.05 1.05
Gluten, OAT EM (20%), Glycerol (13%) 30 32.2-0 0.05 0.55
Plain cellophane 37.8 46.7 0.83-0.166
Polyethylene (low density) 37.7 44.3-0 0.025 0.010
C16-C18 MC/HPMC, Beeswax (4m9/cm2) 25 23,0-0 0.056 0.0076
Gluten, Beeswax (5.3 m9/Cm2) 30 32.2-0 0.09 0.0048
Waxed paper 37.8 46.7-0 0.0016-0.125
Aluminium foil 37.7 44.3-0 0.025 0.00006

CONCLUSION

The feasability of using wheat gluten as a film-forming agent appeared promising considering
the overall film properties and the possible improvement of this properties, especially
moisture barrier property, through introduction of lipids materials.

REFERENCES

1. Gontard N., Guilbert S., Cuq J.L., Edible wheat gluten films: influence of the main
process variables on film properties using response surface methodology. ,1992, J. Food
Sci., 57 (1), 190-195.
2. Gontard N., Guilbert S., Cuq J.L. . Water and glycerol as plasticizers affect mechanical
and water vapor barrier properties of an edible wheat gluten film., 1. Food Sci., 1993,58 (1),
206-211.
3. Guilbert S .. Technology and application of edible protective films. In: Food packaging
and preservation, 1986, Edr. M. Mathlouthi, Elsevier Applied Science Publishers, New
York, 371-394.
4. Kester J.l., Fennema O.R. Edible films and coatings: a review., 1986, Food Techno!.,
40, (12), 47-59.
 SELECTION OF LAMINATED FILM FOR A FOOD PACKAGING

MITSUY A SHIMODA AND YUTAKA OSAflMA

Department of Food Science and Technology, Faculty of Agriculture, Kyushu University,
Hakozaki, Higashi-ku, Fukuoka 812, Japan

ABSTRACT

A method of selecting a laminated film for pouch was proposed by showing the influence of
oxygen permeability [a:cm3jcm2 day atm], storage period [day], the ratio of surface area
[S:cm2] to volume [V:cm3] of pouch. In order to correlate flavor deterioration to the amount
of oxygen permeated, the following equation was introduced; <X=p a T SN, where p is a
partial pressure of oxygen in atmosphere and <X [dimensionless] was defined as a specific
volume of oxygen permeated. As a result, the flavor of packaged soup significantly
deteriorated above <x=0.0034.

INTRODUCTION

Flavor deterioration of food packaged in a pouch can be attributed to sorption of flavor

compounds into a liner 1,2), and an oxidation induced by oxygen permeated 3). Flavor
compounds in Japanese style soup packaged in a plastic pouch lined with linear low density
polyethylene (LLDPE) film were not sorbed into the liner, but the flavor of soup in the
pouch made of nylon(l5f.lm)ILLDPE(60f.lm) film deteriorated during storage. However, the
flavor scarcely deteriorated in the pouch laminated with aluminum foil. The influence of the
oxygen on flavor deterioration was investigated quantitatively using pouches made of films
with different oxygen permeabilities.

MATERIALS AND METHODS

Table 1 shows six kinds of laminated films. The oxygen permeability was ranged from 0.3 to
98 x 10-4 [cm 3jcm 2 day atm], and the film laminated with aluminum foil was used as a
control pouch. The liner of every pouch was LLDPE film. The soup, which was
794
manufactured from soy sauce, dried bonito, and seaweed by Ichiban Food CO.,LTD, was
packaged in the pouches (9xI5cm, 200ml) made ofthe films in Table 1. The sterilization
was done by heating at 90'C for 20 min after packaging and the samples were stored under
dark at 25 'C. Preparation of odor concentrate from the soup was carried out by adsorptive
method using the column packed with Porapack Q. Sample was passed through the column
(2xIOcm) and adsorbed volatile compounds were eluted with 50ml of diethylether after the
column was washed with 50ml of water. Cyclohexanol as an internal standard was added to
the ether solution, and then ether was evaporated under a nitrogen stream. A Shimadzu GC-
14A gas chromatograph equipped with
Table I Laminated films and oxygen penneabilities
a 60m x 0.25mm i.d. DB-WAX column
was used. The oven temperature was Film contruction I) Oxygen penneability2)
programmed at 60 to 230 C at 3 'C/min.
For identification of GC components, 1. PETI2JLLDPE60 98 x 10-4
2. NyI5/LLDPE6o 28 x 10-4
a Shimadzu GCMS-9020DF gas
3. KNyI5/LLDPE60 4.5 x 10-4
chromatograph-mass spectrometer
4. "'KNyI5/LLDPE60 1.5 x 10-4
was used. Sensory test on the soup
5. EVOHI5/LLDPE60 0.3 X 10-4
flavor was done with the samples_
stored for 3 weeks at 25 'C. Sample 1
6. NYI5/AL7/LLDPE60 o
(pouch I) and sample 6 were presented I) PET;polyethylene terephthalate (thickness, 12 J.l.Ill),
to panelists. The remaining 4 samples NY;nylon, KNY;nylon coated with vinylidene chloride,
"'KNY;nylon coated with high barrier vinylidene
were arranged in order of the flavor chloride, EVOH;ethylene vinylalcohol copolymer,
deterioration. AL;aluminum foil, LLDPE;linear low density polyethylene
2) [cm3/cm2 day atm]

RESULTS

Forty seven volatile compounds in the soup were determined and 36 of them were identified.
The similarity of GC pattern before and after storage was used as an index of the flavor
deterioration, since the flavor deterioration could not be attributed to any specific
components. Fig. 1 showed sample 6 was maintained higher than 0.98 even after 4 weeks.
On the other hand sample 1 decreased significantly after a week. The decreasing rate in the
similarity of other samples were compatible with their oxygen permeabilities. As shown in
Table 2, 5 persons of the panelists arranged samples 2 and 3 correctly, while samples 4 and 5
were confused with each other. This showed that flavor deterioration progressed with an

1.0 6 Table 2 Results of sensory test

5
4 Panelist Sample
0
·cCtI 3
2 PI 2 3 5 4 6
] .95
P2 2 3 5 4 6
iZi P3 2 3 4 5 6
P4 2 3 4 5 6
P5 2 3 4 5 6
.90 P6 2 5 3 4 6
P7 2 4 5 3 6

"
0 2 3
I )Samples I and 6 were presented as
Storage period [week]
a pouch of maximum or minimum
Fig. 1 Change in the simirality of gas oxygen permeability and samples
chromatogram during storage 2-5 were arranged between 1 and 6.
Symbols refer to the samples packaged
in the pouch (1-6) in Table 1.
 795
increase in oxygen permeability in the case of oxygen permeability being higher than
4.5xlO- 4 [cm 3/cm 2 day atm], but was not appreciable with the permeability lower than
1.5xlO-4 [cm 3/cm 2 day atm]. Every panelist mentioned that sample 6 preserved well its
flavor, and sample 1 smelled stable flavor. As a result, the soups in the pouches with oxygen
permeability higher than 4.5xlO-4 [cm3/cm 2 day atm] deteriorated its flavor within 3 weeks.

DISCUSSION

In order to select the most appropriate film for the flavor of Japanese style soup packaged in
a pouch, the following consideration was given with variables of oxygen permeability
[a:cm3/cm 2 day atm], storage peJjod [T:day], pouch volume [V:cm3], pouch surface area
[S:cm2]. a=p a T SN, where a [cm3/cm 3; dimensionless] was defined as a specific volume
of oxygen permeated, and p is a partial pressure of atmosphere. Fig. 2 shows the plots of
a vs. TSN with the pouch 1 - 6.-
The limiting a value was estimated .03
from p=O.2, 1.5xlO-4::::;:a::::;:4.5xl0-4,
T=21 days, S=270 cm2, and V=200
ml, and shown as a region hatched.
Fig. 2 shows the flavor deteriorates 3
within 4 days in pouch 2, etc., and
it enables to select a most appropriate 4

film. 5
Above consideration was given under o 50 100 150
the following assumptions: 1) flavor
deterioration is attributed to oxygen TSN
permeated, 2) rate of reactions including
Fig. 2 Relationship between flavor deterioration
oxygen molecule is much faster than and a specific volume of oxygen permeated.
the rate of oxygen permeation. a=p aT SN where a is a specific volume of
oxygen permeated [dimensionless], p;partial
pressure of oxygen in atmosphere, a;oxygen
permeability [cm 3/cm2 day atm], T;storage
period [day], S;surface area of pouch [cm 2],
V;volume of pouch [cm 3].

REFERENCES

1. M. Shimoda, T. Ikegami and Y. Osajima, Sorption of flavor compounds in aqueous

solution into polyethylene film. J. Sci. Food Agric., 1988, 42, 157-163.
2. T. Ikegami, K. Nagashima, M. Shimoda, Y. Tanaka, and Y. Osajima, Sorption of volatile
compounds in aqueous solution by ethylene-vinyl alcohol copolymer films. 1. Food Sci.,
1991,56,500-503.
3. J. R. Mclellan, L. R. Lind, and R. W. Kime, A shelflife evaluation of an oriented
Polyethylene Terephthalate package for use with hot filled apple juice. J. Food Sci., 1987,
52,365- 369.
 CONTINUOUS MICROWAVE DRYING OF POLYE'I1IYLENE TEREPHTIIALATE (PET)

C.A.R. ANJOS, J.A.F. FARIA, and A. MARSAIOLI, JR.

Dept. of Food Engineering (DEA) , Faculty of Food Eng. (FEA) , UN I CAMP
C.P.6121 - 13081-970, Campinas, SP, BRAZIL

ABSTRACT

Microwave heating techniques have been applied to the drying of

polyethylene terephthalate (PET) resin, aiming to improve the critical
operational conditions prevailing in the conventional air drying of that
polymer and which may affect the final quality of molded beverage
bottles if not carefully controlled. Moisture adversely affects
intrinsic viscosity of PET, a high value of it being required to produce
quality preforms. PET is very hygroscopic, which makes drying a
mandatory operation prior to the production of preforms. Considering the
critical variables dew point, air temperature, air volume and product
holding time, it became evident from this study that the last two might
be substantially reduced. That would cause certain advantages to be
gained as concerns to the speed of drying, quality control, and the
overall capitalized cost, which may render microwave heating feasible,
in spite of the high cost of microwave equipment per unit of capacity.

INTRODUCTION

Polyethylene Terephthalate (PET) is a polyester resin that has been

applied to food packaging, particularly into soft drinks bottles by way
of the inJection-stretch-blow-molded process. PET as a raw material in
the solid form absorbs moisture from the atmosphere, i.e. it acts like a
desiccant. Moisture is absorbed during storage to a value as high as 0.6
% (w/w). In practice moisture level of commercial PET is likely to range
from 0.1 to 0.3 % after storage under cover for relatively short
periods. Prior to injection molding of preforms, PET has to be carefully
controlled dried to less than 0.004 % (40 ppm), as an essential
prerequisite to attain maximum product performance in the final
processing. The conventional drying method for PET has been one of using
hot air with desiccation. This can be done on either a batch or
continuous basis under a normal holding time of 4 to 6 hours, carried
out in the range of 170-180oC. Similarly, many other industrial plastics
 797
have to be dried to some acceptable level of residual water content,
usually less than 1.0 % (w/w) , prior to injection molding, using hot air
with desiccation as well. All these methods have in common the use of
relatively high temperatures, ranging from 80 to 120°C depending on the
type of plastic, with a normal dwell time of 4 to 5 hours [1]. This time
can be significantly reduced by introducing a microwavelRF drying cycle
of from 10 to 15 minutes before the hot air cycle, rendering the total
drying time now reduced to 30-60 minutes [2]. It was thought in this
study that the application of microwave heating techniques to PET could
also be less rigorous than in the conventional heating, briI1&ing the
benefit of time or energy saving as well as increasing product quality.
As information on this type of process has been scarqEl as far as the
industrial applications are concerned, the reported preliminary
experiments were conducted either in a laboratory oven or in a pilot
plant, by various exposures to dielectric heating at a frequency of 2.45
GHz, taking advantage of one already existing continuous microwave
drying installation at the Dept. of Food Engineering of State University
of Campinas in the area of microwave processing [3].

MATERIALS AND HETIlons

Commercial PET from ICI (Melinar) was used as raw material for the
experiments, with an average water content of 3500 ppm, a little higher
than the originally delivered product due to a long period of storage.
Microwave heating on a laboratory scale was carried out at 2.45 GHz in a
modified domestic oven (40 liters x 850 watts). Due to the lack of a
continuous microwave power control, a procedure was adopted based on an
empirical model developed by Mudgett [4] applied to the energy coupled
inside the oven cavity as a function of load volume : a spirally wound
flexible polyethylene tube was introduced into the oven, provided with a
suitable hole through which the tube ends crossed the top wall and were
outside connected to the tap water and drain pipe. Water volume was
controlled by the length of the spiralled tube, whereas the flow was
monitored by a precision flowmeter of scale 30-245 liters/hour. Water
inlet and outlet temperatures were also monitored by the use of suitable
thermocouples and a digital indicator. For each test, 50 g PET samples
were weighed into a single spiralled glass tube, which was introduced
into the center of the oven cavity, supported by a special plastic round
tripod and having its extremities also crossing the top oven wall. A
double stage system for filtering and desiccating air, resulting from a
combination of a compressed air dryer by refrigeration in series with a
molecular sieve, was used to supply cleaned air at a -30 to -35°C dew
point, which was measured by means of a precision flowmeter of scale 40-
350 liters/hour and made to flow inside the glass tube packed with the
PET sample. Air moisture was controlled by the use of an automatic Shaw
dew point meter, provided with a precision analog scale from 0 to -80°C.
As an alternative for certain tests, an electrical heater was also used
to heat the air before passing through the PET sample. The temperature
of the PET pellets was recorded as an average taken by inserting a
thermocouple midway inside the glass tube immediately after the
microwave treatment. For tests which approximate industrial conditions,
a novel microwave rotary dryer as described in [3] was adapted through
the removal of the original blower which was substituted by the same
double stage system to generate the fil tered and desiccated air as
798
applied to the laboratory experiments, but operating at a flow rate in
the range of from 4000 to 8500 liters/hour. After adjusting the desired
PET mass flow rate at an average 5.5 kg/hour, experiments were carried
out to establish the residence time distribution of the pellets inside
the rotating cavity, as a function of rotation speed and inclination of
the drum. The moisture contents of PET samples for both laboratory and
pilot plant tests were determined in a Aquastar moisture titrator.

RESULTS AND CONCLUSIONS

Tests were first run in the laboratory oven in order ·to establish the
best combination of energy level and exposure time for the moisture
removal of samples. Processing times ranged from 30 to 60 minutes. Some
difficul ties arose because of lack of homogeneity of the electrical
field distribution, causing deviations of the samples temperatures as
°
great as 10 C. Also energy level evaluations based on calorimetric
measurements produced errors on the order of 20-30 %. Even so samples
were partially dried, ranging from 21.9 % moisture removal for 230 W in
30 min. up to 38.6 % removal for 700 W in 40 min. The pilot scale
treatments were carried out under the conditions : 4250 liters/hour of
air at -35°C dew point, with and without preheating, energy levels from
120 to 360 Wh/kg and average PET mass flow rate of 5.5 kg/hour. Under
the maximum energy level moisture removal was 81.1 % of total water
content, not sufficient yet in relation to the recommended moisture of
less than 40 ppm. From the observed results it can be concluded that the
microwave treatment of PET has a good potential as compared to
conventional drying method, even though a considerable improvement of
the techniques related to the energy modulation of microwave is still
necessary in order to have a better control of the new process.

REFERENCES

1. Hasan, M. and Mujumdar, A.S., Drying of Polymers, in Handbook of

Industrial Drying. Edited by Arun S. Mujumdar, Marcel Dekker, Inc.,
N.York, 1987.

2. Lightsey, G.R. and Russell, L.D., TVA Report (Low Temperature

Processing of Plastics Utilizing Microwave Heating Techniques)
November, 1985, (Unpublished), cited in R.W. Bruce and M.W.Mccurdy,
Dielectric Measurements of Particulate Organic Polymers, in Symposium
Proceedings of the Materials Research Society : Microwave Processing
of Materials, volume 124, Pittsburgh, PA, 1988.

3. Marsaioli Jr., A., Conforti, E. and Kieckbush, T.G., A Prototype of a

Combined Hot Air and Microwave Rotary Cylindrical Oven for Continuous
Drying of Granular Products, Proceedings of the Fifth International
Congress on Engineering and Food, vol. 2, edited by W.E.L. Spiess and
H. Shubert, Elsevier Applied Science, 1990.

4. Mudgett, R.E., Electrical Properties of Foods, in Properties of

Foods, edited by M.A. Rao and S.S.H. Rizvi, Marcel Dekker, New York,
1986, pp. 329-390.
 STERILIZATION OF PACKAGING CONTAINERS
BY GASIFIED HYDROGEN PEROXIDE AND THE
APPLICATION
TO FLEXIBLE ASEPTIC PACKAGING MACHINE
YSHIBAUCHI, T.TANAKA, K.HATANAKA
Technical Research Institute, Snow Brand Milk Products Co. ,Ltd.
Minami-dai 1-1-2, Kawagoe, Saitama 350, Japan

ABSTRACT
Gasified hydrogen peroxide (H 2 0 2 ) was studied to sterilize preformed packaging con-
tainers in the combination of irradiation of ultraviolet rays (UV), and it was successfully
applied to a flexible aseptic packaging machine that is applicable to a variety of preformed
containers with different shapes and sizes. An effective and efficient H2 0 2 vaporizer was
developed based on the study of its vaporization and decomposition characteristics.
The sterilization test with inoculated containers showed a synergistic effect of the H2 0 2
gas and the subsequent UV-rays treatment, inactivating more than 106 spores of B.subtilis
and B.stearothermophilus per a 100ml-container. Residual H2 0 2 of the containers after
drying treatment by heated sterile air was confirmed to be low enough, less than 0.1 ppm
under the drying condition of 80°C for five seconds.

INTRODUCTION
The liquid H2 0 2 has been studied as means of sterilization of packaging materi-
als such as plastic films and sheets since it has a strong inactivation effect for most
microorganisms 1),2),3),4),5),6). However the application to preformed containers has been
limited because of the difficulty in reducing residual H2 0 2 on the surface of containers
after the sterilization treatment. Usually the more complex in the shape of containers are,
the more serious the difficulty in reducing the H2 0 2 residue. Accordingly, in our develop-
ment of the versatile aseptic packaging machine, an effective sterilization method applica-
ble to the containers with complicated surface shapes was required. In addition, since the
softening temperature of common plastics used for food containers such as polyethylene
or polystyrene is about 80 to 90°C, it should be noted that the higher temperature beyond
that cannot be applied to the containers.
800
EXPERIMENTAL RESULTS AND DISCUSSION
In advance of the development of H2 0 2 vaporizer, vaporizing characteristics were
studied. The H2 0 2 droplet of about O.Olg was gently put on the surface of a heated
stainless steel plate to measure the vaporizing time. Fig.l shows the vaporizing time
in comparison with water. In case of water, the vaporizing time decreased as the plate
temperature was increased and came to the minimum point at about 140°C, then increased
again due to the so-called spheroidal phenomenon that declines heat transfer from the
plate to the droplet because of a vapor film formed between them.
The H2 0 2 droplet showed, however, a bit different trend. After reaching the minimum
point at about 170°C, the vaporizing time did not increase so much because it became
unstable due 1.0 oxygen bubbles formed in the droplet by the thermal decomposition of
H2 0 2 • An additional test showed that the decomposition rate became smallest when the
vaporizing time was shortest.

Fig.2 shows the residual H2 0 2 characteristics per lOOml-container when the drying
air temperature was changed. The drying time was fixed to five seconds for all measure-
ments. The amount of residual H2 0 2 decreased exponentially as the drying temperature
increased. The figure indicates that the drying treatment at about 80°C may reduce the
residual H2 0 2 to O.lppm or lower.

Tab.l shows a part of results of the sterilization experiments using H 20 2 gas, UV-
rays and their combination for the lOOml-containers inoculated with spores of A.niger,
B.subtilis and B.stearothermophilus. The concentration of liquid H2 0 2 submitted to gasi-
fication was varied from 3% to 35%, and the intensity of applied UV-rays was about
160mW·s/cm2 for each container. Independent use of H2 0 2 gas or UV-rays resulted in
some residual spores, but their combination sterilized all samples completely even when
3%-H 2 02 was used.

Fig.3 shows the schematic picture of the developed H2 0 2 vaporizer. The H2 0 2 gas
is generated in the heated vaporizing pocket with the help of the heated air, and passes
through the entrainment strainer, in turn get to the gas-outlet after the temperature be-
ing raised properly.

FigA shows the preformed container sterilizer comprising the H2 0 2 gas generator, a
UV-Iamp and an H2 0 2 dryer employing heated air. Each container is suspended by its
flange between two parallel rails and conveyed forward by a shoving plate. The shoving
plate and the walls of the both sides and the top and bottom form a closed chamber
which helps to sterilize whole surface of the container. The distance between the rails is
adjustable so that it would be applicable to different containers in shapes and sizes.
A conformation similar to Fig.3 is applied to the lid sterilization (Fig.5) except its
rotat\onal conveying system. Each lid is held by vacuum heads and submitted to the
similar sterilization process. This lid sterilizer is applicable to any shapes of lids as far as
their sizes are in the range from 60x60mm to 120x120mm.
These container and lid sterilizing system were successfully applied to the flexible
aseptic packaging machine, and the aseptic filling test showed the enough commercial
sterility of products and the stability in operation.
 801

70 90 110 130 150 170 190 210

60 80 100
Temperature [0C]
Air Temperature [0C]
Fig.l Vaporizing Characteristics Fig.2 H 2 0 2-removal Characteristics
of A Droplet by Heated Air

o : Water, 6.: H 2 0 2

ur
Cup Supply
Drying Heated Ai r
HzOz Gas
Heated Carrier Air
UV Lamp
r'

Ent ra inmen t
Strainer

Ai rl,-,._ _ _ _ _ _ _ _ _ _ _ _.=;...--t
Out et
Fig.4 Cup Sterilizer
Drop Iat
Nozzle -""&..z:~~
Tab. 1 Results of Sterilization Experiments
Vaporizing --""",.I"',,.,
Packet
H,Q, Oil UV A. n:i.ger B. subti.1.is B. stearoth.
Heat
Conductor
0 1700 120 172
Heater 3 0 2800 12000
5 0 1886 3920
10 0 81 368
35 0 12 4
3 0 0
Fig.3 H2 0 2 Vaporizer 5 0 0 0
10 0 0 0
35 0 0 0
UV Lamp
H,Q, [~J:Concentration of hydrogen peroxide-water solution
to be vapo r i zed.
UV :Ultraviolet rays "ere applied(O),not applied(-).

REFERENCES
l)Smith,Q.J. and Brown,K.L., Food Technol.,15,169(1980)
2)Stevenson,K.E. and Shafer,B.D., Food Technol.,1l,1l1(1983)
3)Baldry,M.G.C., J .Appl.Bacteriol.,54,417( 1983)
II 4)Stannard,C. and others, J.Food Protect.,46(12),1060{1983)
\+J
I
Cup 5)Waites,W.M. and others, Appl.Microbiol.,7,139(1988)
Fig.S Lid Steri I izer 6)Wang,J. and Toledo,R.T., Food Technol.,1l,60(1986)
 ASEPTIC FILLING OF BEVERAGES IN GLASS BOTTLES -
NEW DEVELOPED PROCESSES IN GERMANY

HORST WEISSER and MICHAEL ZUBER

Institute for Brewery Installations and Food Packaging Technology.
Technical University of Munich, 85350 Freising-Weihenstephan, Federal Republic of Germany

ABSTRACT
The aseptic filling of beverages in glass bottles to achieve commercial sterility requests a high standard
of the packaging material, the filling line, and on the microbiological hygiene and sanitation.
As the consumer prefers food products from sterilization processes using physical rather than
chemical treatment the use of saturated steam is an interesting method to sterilize the glass and the
filling line. Three examples for recently built continuous filling lines of the German companies Bosch
(for milk), KHS (for juices) and Krones (for beer) are shown. The different filling systems are compared
with respect to constructive details, operating conditions, and examples of installation.

INTRODUCTION AND DEFINITIONS

As the consumer prefers food products from sterilization processes using physical rather than chemical
treatment the use of saturated steam is an interesting method to sterilize the glass and the filling system.
Since such concepts as aseptic, ultra-clean, sterile, beverage-sterile cannot be used unambiguously
in the field of sterile packaging, a definition of aseptic packaging is useful as an introduction to the
following examples.
Aseptic packaging represents the packaging of a sterilized product into sterile packages in a sterile
environment by sterile means whereby neither the product nor the packaging material nor the internal
atmosphere is exposed to a further sterilization process after the packaging stage, and the product does
not suffer microbial growth in the context of the environment and storage conditions.
As with sterilized canned food, no absolute sterility is achievable in all cases with aseptically filled
foods. However, the microbial contamination must be reduced drastically, food-poisoning by pathogenic
microorganisms must be avoided with reliability.
With beverage-sterile filling commercial sterility is achieved. During the filling process the entering
of product-spoilage microorganisms through filler, empty bottles, closures, and environment must be
prevented. The product has to be free of beverage-spoilage microorganisms when it is fed to the filler.
The decision whether to use sterile filters or a flash pasteurizer is not so much a question of food
engineering but of marketing (draft beer). It is also determined by the capital investment and operating
costs involved.
Conventional aseptic filling systems
Aseptic filling systems for the filling of soft beverages into
- carton-laminate systems, e.g. Tetra Pak, Pure-Pak and PKL Combibloc, and
- Composite can systems, e.g. Bosch Hypa-S.
now exist for several years [1].
 803
RECENTLY DEVELOPED GLASS ASEPTIC SYSTEMS
In the last years, several companies developed aseptic or at least beverage-sterile packaging processes to
fill premium quality beverages like milk, juices, and beer into glass bottles.
Bosch
A pioneer in this field was the Bosch company which developed the GlaSeptik®-process in co-operation
with a glass manufacturer (Oberland Glas) and the manufacturer of the Twist~ closures
(Continental Can Europe, Schmalbach Lubeca). The filling line for non-acid and low acid products
consists of a rinser, a bottle sterilizer, an aseptic filling machine, and a sealing machine designed
especially for the aseptic closing of glass jars [2]. The different process steps are:
- Rinsing and pre-sterilizing of the jars with steam.
- Final sterilisation with hydrogen peroxide (H202) and drying with sterile air in a bottle sterilizer.
- Transportation of the sterile jars to the filler via a sterile connection tunnel.
- Filling with a longitudinal filler with a underlevel filling system (long filling pipe) to avoid or reduce
the formation of foam or the intrusion of oxygen. If necessary, the oxygen content in the headspace
can be reduced either by steam-exhaustion or by flushing the headspace with an inert gas.
- Sealing the jar with Twist-OfI® or PT- caps in a two-lane linear closing machine with 10 closing
heads. Saturated steam is continuously fed into the top of the machine. This guarantees aseptic
conditions during machine operation and generates a partial vacuum in the package after sealing.
The entire filling line is hermetically closed and a positive pressure of several mbar is applied to prevent
microbial infections during the operation. Whenever it comes to machine standstills during the filling
process, operators can use attached rubber hand gloves at critical points. No direct contact is necessary
to remedy the interruption. The exhausted ~02-vapor is reduced by using a catalyst. The whole line
can be cleaned with CIP and sterilized using the SIP process.
KHS
In 1989 a new aseptic filling system, Aseptronic GRA , was presented at the Drinktec-fair by KHS
(Kl6ckner Holstein Seitz, machinery for the production of beverages) and Granini (juice producer) as a
joint venture project. This aseptic filling line consists of a combined sterilizer/filler and a modified
closing machine [3]. The filling process can be separated into the following steps:
- Warm bottles (60 - 70°C) are supplied from a bottle washer.
- The bottles are pressed against the filling valve by the bottle platform. Bottle platform, bottle, and
filling valve form a unit. They are enclosed into the so-called valve-bell (Figure 1).
- Each bottle is rinsed with steam continuously
emerging from within the bell. A reduced steam
flow enters the bottle via a long filling tube,
flushes the internal bottle surface and the bottle
mouth and after the bottle is full of steam also
the external bottle walls. A uniform bottle tem-
perature is reached.
- The bell is sealed hermetically.
- Bell and bottle are sterilized for 4 s at about
135°C.
- Sterile air or inert gases replace the steam in the
bottle. The bottle is counter-pressurized either
with sterile air or an inert gas.
- A low turbulence filling process using tradi-
tional techniques follows.
- The filling tube is emptied and the bottles are
removed from the filling valves.
- The bottles are transported to a TWist-Offbottle
closing machine in a steam atmosphere. Figure 1. Aseptronic GRA filling system.
804
KroDes
A sterile filling and closing system for beer was developed by Krones (Hermann Kronseder Maschinen-
fabrik) under the name BSF-System (Beverage-Sterile-Filling). The filling line consists of a combined
sterilisation and filling and a conventional, CIP cleanable cap crowning machine [4]. The process steps
are:
- Warm bottles (50-60°C) are
supplied from a bottle washer or a
rinser.
- The bottle's inner surface is steril-
ized with 11 O°C hot saturated
steam (Figure 2). The bottle inter-
nal wall temperature rises to about
105°C. The oxygen content in this
steam atmosphere is very low.
- The bottle is now C02-flushed to c......1
remove steam and air from the
bottle.
- The bottle is pressurized with
carbon dioxide. The bottle is
brought up to filling pressure.
- The low turbulence filling process
starts, when the product and the
return gas valves open. The
product streams along the internal
Pred""Ui. ·
walls into the bottle. The carbon clla,., wiIIIlwill .......... P..-iu<I... prvftcl i....
dioxide is collected in the return 1..,11

gas/snift channel (Figure 2).

- The filling process ends as soon as C. .(,i"lllloll
-t'5=~l- Se,1
the pre-set filling height is
reached. The level probe sends an
electric impulse to close the Figure 2. Krones pneumatic filler for beverage-sterile filling.
product and return gas valve.
- After an adequate settling time the bottles are snifted and fed to a CIP cleanable crowner under
aseptic conditions.

CONCLUSIONS
The Bosch GlaSeptik process is an interesting example of aseptic filling being used for microbiologi-
cally very sensitive low-acid products like milk, milk products, and dietetic products.
Until today the KHS Aseptronic GRA system, as well as the newly developed KHS DDF-filler
(Dampfdruckfiillverfahren = vapor pressure filling process) with long filling tubes, and the Krones
BSF-filler can only be used for low-acid beverages like fruit juices, nectars, and beer. With this type of
beverage-sterile rotary only a commercial sterilization can be achieved.
Aseptic packaging into glass is as well a challenge to the food industry as it is an opportunity for
cutting costs and improving the quality of the end products.

REFERENCES
1. Weisser, H., Fortschritte beim keimarmen und aseptischen Verpacken von Lebensmitteln. Der
Weihenstephaner, 1991,59, No.1, 48-52.
2. Buchner, N., Aseptisches Fullen von Behaltern aus Glas und Kunststoffen. ZFL - Intern. J. Food.
Technol., Marketing, Packaging, and AnalYSiS, 1990,41, No.5, 295-300.
3. Aseptische Abfiillung auch fur Bier. Neue Verpackung 1992,45, No. 10,42-46.
4. Kronseder, Hermann, Getrii.nkesteriles Fullen. Getranketechnik, 1992, No.5, 173-176.
 A KINETIC MODEL FOR FOULING IN MILK PROCESSING

P. J. R. SCHREIER, I. TOYODA *, M. T. BELMAR-BEINY, and P. J. FRYER,

Department of Chemical Engineering,
University of Cambridge, Pembroke Street,
Cambridge CB2 3RA, United Kingdom.
*: Meiji Milk Company, Japan.
ABSTRACT
Fouling from milk proteins at pasteurisation temperatures has been examined experimentally
in a simple counter-current heat exchanger, and a model based on the method of characteristics used
to model the results. Protein reactions both at the surface and in the bulk are necessary to model the
process.
INTRODUCTION

Fouling from milk at temperatures below 100°C results from a complex mixture of reactions
on the heat exchange surface and in the bulk of the fluid [I]. When heated, the structure of the native
milk protein B-Iactoglobulin is irreversibly altered by denaturation (intramolecular) and aggregation
(intermolecular) reactions. Early work on fouling models concentrated on surface reactions [2],
however [3] has shown that bulk reactions can be important in milk protein fouling. Simple models
have been produced in which the amount of fouling was considered proportional to the volume of
fluid which is hot enough to produce denatured and aggregated protein [3,4]. Recently, a
mathematical model has been produced [5] which considers both surface and bulk reactions. The
denaturation reaction is modelled as first order in concentration of native protein with the aggregation
reaction as second order in concentration of denatured protein. The model of [5] describes a complex
plate heat exchanger, in which a range of velocities and temperatures is found. Experiments in tubes,
where the temperatures and flows are more easily known, are easier to intetpret. This paper describes
the basis of a theoretical study of a tubular pasteuriser to examine a model for fouling. The model is
compared with experimental data on fouling produced in a laboratory-scale apparatus.
MODELLING EQUATIONS

Heat transfer
Experiments have been carried out in a simple shell-and-tube configuration [6]. This system
can be modelled [7,8] using enthalpy balances on differential volume elements of fluid on the shell
and tube sides of the generalised exchanger configuration. These lead to the following equations for
fluid temperature distributions:
(a) Tube side fluid temperature, Tf(t,x): -dTf + vf::C-- aTf _ -Uat- (Ts- Tf) (1)
at ax (pCp)r

(b) Shell side fluid temperature, Ts(t,x): (2)

where a is the heat transfer area per unit volume, v the fluid velocity and pCp the thermal capacity.
Equations (1) and (2) assume that the fluids pass along the shell and tube sides in plug flow and that
806
entrance and exit effects may be neglected, with the deposit sufficiently thin for its enthalpy to be
neglected and for the tube side Reynolds number to remain unchanged. To study fouling, equations
(1) and (2) must be solved simultaneously with any fouling model. Together with the fouling model,
equations (1) and (2) may be transfonned from partial differential equations into ordinary differential
equations defined along characteristic lines in the x-t plane using the method of characteristics [9]. In
effect this method allows the equations to follow a fluid element down the tube; it is thus intuitively
sensible as well as mathematically rigorous.
Fouling

Two reactions are assumed, N ~kl 0 ~k2 A, by which native protein, N, is transfonned
to aggregates, A. Reaction (1) is taken as first order and reaction (2) as second order. Fig. I gives a
schematic diagram of the system. The fluid mechanics are simplified by assuming that there are two
regions in the flow, one of unifonn bulk temperature, Tb, and a thennal boundary layer of thickness
B, at the deposit-fluid interface temperature, T wd. Protein concentration in the bulk and the thennal
boundary layer are modelled separately. Mass transfer between the two regions is given by some
mass transfer coefficient:

Rate of transfer of aggregates between bulk and wall layer =km (C~ -CA) (3)

where the concentration of aggregate in the wall region is C~.

The rate of fouling is modelled as an adhesion reaction =kw C~ (4)

Figs. 2 and 3:
EN = 261 kJ/mol, AN = 3.37xl03 1: 70°C-900C
Bulk EO = 312 kJ/mol, AO = 1.36xl043: 70°C-90°C
EO = 56 kJ/mol, AO = 1.86xl06: 90°C-1500C
Boundary Protein concentration = 3.5 wt %
layer C~ T'Ad
Fig. 2:
kw =5xlO-7 mls

Fig. 3:
kw =3x 10-7 mls
Initial denaturation = 30 %

Fig. 1: Thermal boundary layer for TABLE I: Activation energies and

material balance of equations (3) and (4). reaction constants [5].

RESULTS

The model demonstrates the effect that bulk processes can have on fouling. Clearly, in any
heating process, the temperature of the wall region will exceed that of the bulk, and so reaction will be
faster in the wall region. If the concentration of aggregates in the bulk is low, mass transfer into the
bulk will lower the aggregate concentration in the wall layer and the fouling rate; however, if the bulk
is hot enough for protein reactions to take place then the mass transfer rate will fall, and cA and the
fouling rate will increase.
Two types of simulation have been carried out. The sensitivity of the model to variations in a
range of process parameters has been examined. Two important factors are the concentration of native
protein in the milk and the proportion of protein in the milk that has been denatured before it enters the
heat exchanger tube. The latter is important since much experimental work has used reconstituted
whey protein concentrate powders which will be denatured to some degree. Fig. 2 shows the effect of
varying the degree of protein denaturation in the inlet protein, for the constants in Table I; clear
variation in the inlet region is seen.
 807
Experiments have been perfonned [6] in a heat exchanger tube using whey protein
concentrate. The simple geometry of the single pass tubular heat exchanger may be modelled using
equations (1) to (4). Fig. 3 shows a comparison of experimental data with a curve produced by the
model, which shows good agreement with experimental data. This model demonstrates the
importance of bulk reactions; further development is underway to fit a wider range of experimental
data.

~O~------------~~------------------------~

Ne 200
~
.'1160

.
1
....
120
80

!4O
~

o+k==~~~~--~--~---r--~--II~
o 50 100 150 200
Distance (em)

Fig. 2: Effect on fouling of varying degree of inlet protein denaturation.

140 iii
iii
-. 120
Ne
~loo
.'1
§ 80
III

....~
60
EI
.~
0
40
fir 20
0
0
0 50 100 150 200
Distance (cm)

Fig. 3: Comparison of experimental data with the model.

REFERENCES
1. Fryer, P.J., Belmar-Beiny, M.T. and Schreier, P.J.R. (1993), presented at this conf.
2. Lalande, M., Tissier, J.-P. and Corrieu, G. (1984), l. Dairy Res., 51, 557-568.
3. Belmar-Beiny, M.T., Gotham, S.M., Fryer, P.J. and Pritchard, A.M. (1993), l. Food
Eng., 19, 119-139.
4. Paterson, W.R. and Fryer, P.J. (1988), Chern. Eng. Sci., 43, 1714-1717.
5. de Jong, P., Bouman, S., van der Linden, H.J.L.J. (1992), l. Soc. Dairy Tech., 45, 3-8.
6. Gotham., S.M. (1990), Ph.D. Thesis, University of Cambridge.
7. Fryer, P.J. and Slater, N.K.H. (1985), Chern. Eng. l., 31,97-107.
8. Fryer, P.J. and Slater, N.K.H. (1986), Chern. Eng. Sci., 41, 2365-2372.
9. Acrivos, A. (1956), Ind. Eng. Chern., 48, 703-709.
 INITIAL EVENTS IN SURFACE FOULING

MT BELMAR-BEINY, PJR SCHREIER and PJ FRYER

Dept. o/Chemical Engineering,
Pembroke Street, Cambridge, UK, CB2 3RA

ABSTRACT

Deposit formation (fouling) from dairy fluids is a severe industrial problem. Experiments were
conducted to identify the species that adhere ftrst into stainless steel surfaces at 96'C from turbulent
flows of whey (inlet fluid temperatures of 68 and 73'C). The heating contact times were varied between
4 and 210 s. At 4 s, XPS analysis showed the first layer being made of proteinaceous material. For a
fluid inlet temperature of 73'C, a lag phase of 150 s occurred before deposit aggregates were observed
under SEM. These aggregates were identifted with X ray elemental mapping as being made of protein
and calcium. No phosphorus was found in any experiments. At 60 minutes contact time, both calcium
and phosphorus were found at the interface deposit-stainless steel.

INTRODUCTION
Fouling lowers the efficiency of a food process and can endanger product sterility of the
product. The problems of fouling are discussed in more detail elsewhere [1, 2]. The literature
so far is not conclusive as to which species (proteins or minerals) adsorbs to the surface first
[3,4]. If it were possible to extend the induction period prior to fouling then plant operation
might be improved and cleaning costs reduced. Here, experiments to study the initial stages of
fouling are described.

MA TERIALS AND METIIODS

Whey protein concentrates (WPC) (35 and 75% protein) were dissolved to 1% protein
concentration in distilled water. The test fluid was preheated before passing through the heat
exchanger. For contact times below 3 minutes, a rig was constructed to allow turbulent flow by
gravity at a Reynolds number ca. 15000. For longer contact times a second rig was built
operating at a Reynolds number for the fouling fluid of 6250. The inlet fluid temperature was
varied from 68'C to 73'C, and the wall temperature kept constant at 96'C. Fouled tubes were
removed, dried and cut into small sections for Scanning Electron Microscopy (SEM), X ray
microanalysis and XPS. Further details can be found in [3, 5].

RESULTS
Up to 120 s (73'C inlet temperature) the surface of the material became smoother.
Through naked eye the surface initially looked yellow, then. became green and blue as a result
of an interference film. At 150 s the surface looked whitish to the naked eye and under the
SEM, aggregates appeared on the surface in a random way (size range 0.2-0.7 !lm). From this
 809
point the surface became completely covered. For a fluid inlet temperature of 68°C, however
no aggregates were found on the surface even after 210s. XPS studies showed that, as the
contact time increased (up to 210 s for 68°C, and 150 s for 73°C) the peaks corresponding to
the metallic elements decreased and the carbon and nitrogen peaks increased. The presence of
protein was followed through the carbon and nitrogen peaks. Figure 1 shows the atomic
percentages for carbon, nitrogen and oxygen obtained for WPC75 adsorbed at 68°C and 73°C
for different contact times. At both 68°C and 73°C the surface was already covered after 4 s.
As the contact time increased the percentages of the three elements approached the theoretical
values for 6-lg: carbon 65%, oxygen 19% and nitrogen 16%, becoming almost identical at 150
s, 73°C. Sulphur and calcium were detected in trace amounts (0.8 and 0.6% respectively) at
150 s, 73"C. No phosphorus was found.
Figures 2a and 2b show elemental maps obtained for calcium and phosphorus for
WPC35 after 60 minutes contact time, 73"C. Both elements are concentrated at the interface
deposit-stainless steel surface. Figure 3 shows an elemental map for iron obtained for 150s
73"C. The detection of iron is a good indicator of how the surface had become covered with
deposit: the higher concentration of iron, the thinner the adsorbed foulant layer. Elemental
maps for calcium and sulphur coincided with XPS results for the same specimen. Phosphorus
was not found at 150s.
DISCUSSION
SEM analysis showed that for an inlet temperature of 73°C, there was a lag phase up to
150 s, in which no aggregates appear on the surface. This lag phase probably corresponds to
the induction period. During this lag phase the surface became covered with a proteinaceous
film, which has been seen under SEM [4]. The presence of whey proteins on the surface was
successfully identified with XPS through the organic carbon and organic nitrogen. No sulphur,
calcium or phosphorus were detected during the lag phase observed at 73°C or up to 210 s at
68°C. X ray elemental mapping prove to be a very useful technique in showing the
composition and distribution of elements on deposits formed at the end of the induction period.
The surface analysis results indicate that proteins are most likely to be the first species
to adhere to the surface. The heterogeneous distribution of minerals along the deposit, in which
minerals eventually become concentrated at the interface deposit-surface is still a unanswered
question. Our work showed conclusively that if calcium or phosphorus formed part of the first
layers, their amount was insignificant in comparison to the amount of protein. Thus, protein
adsorption controls the first stages of fouling. Any surface modification to the heat exchanger
most affect protein binding sites.
ACKNOWLEDGEMENTS
MTBB wishes to acknowledge financial support from AFRC.
REFERENCES
1. Fryer, Pl., Schreier, P.J.R., Belmar-Beiny, M.T. (1993). This conference
2. Schreier, P1.R., Toyoda, I., Belmar-Beiny, M.T. and Fryer, PJ. (1993). This conference.
3. Belmar-Beiny, M. T., & Fryer, P. J. (1992a). Bulk surface effects on the initial stages of
whey fouling. Tmnsactions of the Institute of Chemical En&ineers, 70(part C),
193-199.
4. Belmar-Beiny, M. T., & Fryer, P. 1. (1992b). A study of the initial stages of deposition from
whey protein solutions. &tropie, 28(169), 51-58.
5. Belmar-Beiny, M., Gotham, S., Paterson, W., Fryer, P., & Pritchard, A. (1993). The effect
of reynolds number and fluid temperature in whey protein fouling. Journal of
FoogEngineering, 19, 119-139.
810
711
,h -- ....
hO t?,;r'-t-, -0 ~J 68' C
t!.
:'\(1 --G--- 068'(;
- ~--C 68'C
.J(I 73'C
--E)-N

~
~J --0 - - 0 73'C
.\0
-6- C 73'C
0--
::: ~________ :~_~:~~~:~''''''''B
II +.-.-.-,..,..,-.-r,..,..,,.,..,..,..,-.-.-,..,...,-.-.-.,......,~.,-,-.-.
II 211 .J(I 1111 110 ItKI 120 I.J() 1611
Timt' (s)

Fij!.ure L XI'S alomic percenlaj!.es for carllon. oxy~en and nilro~en on rouled
slainless sleel surface (AlS1321) with WI'(: 75 al differenl contact limes

'0
and inl..1 tt'mperalures; walliemperalure 9t\T. Plmtoelt'ctron anJlle or
emis.'iicm or 30' ..-ilh I't'SPKt the ... rflQ.

(A) (8)

fiJlure 2. X ray elemental maps ror (a) calcium and (h) phosphorus on fouled
stainless steel surface (AISI 321) with WPC35 for 60 minutes contact time.
(Reynolds 6250, temperatures: inlet 73'C and oullet 83'C, wall
temperature 97'C). Dwell time 200 ms. m~nirlcatilm 1000 X. The Jlrey
ICaIr is liven in the r",ure, iI reads rrom left to ri&ht.

t'ij!.ure 3. X ray elemental map ror iron on fouled stainless stt't'l surface (AISI 321 I
with WI,(' 75 for 150 s ('(Intact time. (Temperatures: inlet 73(' and o"tI,'1
75 (', wall temperalure 9t\(,). ""ell lime 21KI ms. maj!.nilkalion IIHIII X.
(~n'~ st.:alt· as in fi~urt' 2.
 ADSORPTION OF PROTEIN ONTO STAINLESS STEEL PARTICLE
SURFACE AND ITS DESORPTION BEHAVIOR

HIROSHI ITOH, TADASHI NAGAI*, TAKES HI SAEKI*,

T AKAHARU SAKIYAMA, and KAZUHIRO NAKANISHI
Department of Biotechnology, Faculty of Engineering, Okayama University,
3-1-1, Tsushima-naka, Okayama 700, Japan, *Institute for Fundamental Research,
Suntory Co., 1023-1, Yamazaki, Shimamoto-honmachi, Mishima-gun, Osaka 618, Japan

ABSTRACT

The adsorption and desorption of ~lactoglobulin were studied using stainless steel particles.
The amount of fl-Iactoglobulin adsorbed was influenced by various factors such as
temperature, pH, incubation time, and the protein concentration. The desorption behavior was
analyzed by feeding NaOH or proteolytic enzyme solution into a glass column packed with the
stainless steel particles on which the protein had been adsorbed and subsequently measuring
its elution profile at the outlet. The amount of the protein remaiming on the stainless particles
decreased accoring to a first-order kinetics at the initial stage of desorption, and then gradually
reached a constant value. The desorption rate constant and the remaining amount were studied
under various conditions.

INTRODUCTION

Cleaning is quite an important operation in food processing particularly from a viewpoint of

sanitation of manufacturing processes. To optimize cleaning conditions, the mechanism for
adsorption of soils onto the surface of the equipment and its desorption behavior should be
clarified. We studied here various factors affecting the adsorption of ~lactoglobulin and its
desorption behavior using NaOH or proteolytic enzyme solution was analyzed under various
conditions. ~Lactoglobulin is one of the main soil components when milk products are heat-
treated.

MATERIALS AND METHODS

In most cases, ~lactoglobulin (3x crystallized, Sigma Chemical Co.) dissolved in 2 ml of 35

mM phosphate buffer, pH 6.85, containing 100 mM NaCI was incubated with 2 g of stainless
steel particles (100 meshes in mean size) with vigorous shaking under various conditions.
The amount of protein adsorbed was measured from the difference in concentration of the
solution before and after incubation. The stainless steel particles have a large surface area
(0.17 m2/g) to measure the amount of protein adsorbed with accuracy.
812
About 1.5 g-portion of the stainless steel particles on which fl-Iactoglobulin had been
adsorbed was packed into a glass column (1.5 cm cp x 5 cm). The phosphate buffer was
allowed to flow for 2 h, and then NaOH or proteolytic enzyme (Protin AC, Daiwa Kasei Co.
Ltd., Osaka) solution was fed usually at 50°C. The protein concentration in the eluent was
measured by the Lowry-Folin method. After 2-h feeding, the stainless steel particles were
withdrawn from the column and the remaining amount of protein was directly measured. The
amount of protein remaining on the surface at any time during the desorption process was
calculated by cumulative summation of the protein eluted.

RESULTS AND DISCUSSION

Adsorption of f\-Iactoglobulin
The adsorption equilibrium of ~-lactoglobulin at 25°C was similar to the Langmuir type
adsorption isotherm, and the maximum amount of protein was adsorbed in the thickness of
monolayer. When the temperature exceeded 65°C, the amount of protein adsorbed increased
by several times. At 70°C the amount of protein adsorbed was 1.3 mg/g (7.6 mg/m2) at the
protein concentration of 10 mg/ml. The increase in the adsorbed amount with temperature is
closely related to the denaturation of ~-lactoglobulin [1]. When the pH of the solution was
lower than pH 6, the amount of protein adsorbed considerably increased, indicating the effect
of an electrostatic interaction between the protein and the stainless steel surface. The~­
lactoglobulin which had been denatured at 70°C showed an adsorption behavior similar to that
for obtained without pre-incubation, which might imply that the adsorption takes place at an
instance when the protein suffers from denaturation.

Desorption behavior of f\-lactoglobulin

The elution behavior of ~-lactoglobulin with NaOH or enzyme solution was measured after
feeding the phosphate buffer for 2 h. The amount of protein remaining on the stainless steel
particle decreased accoring to a first-order kinetics at the initial stage of desorption, and then
gradually reached a constant value as shown in Fig. 1.

-'as
c
=0.62 cm 3/min
-..
15... flow rate
c. 0.0 50°C Concentration of
0 NaOH
c
:::J 0 O.01N
0
E -1.0
G
C)
•I:l.
O.02N
O.OSN
c
'c ...
'i O.1N
E
...
CI)

CI)
-2.0 0 O.SN

>
1a

-
"i
a:
.E -3.0
0 10 20 30
Time (min)
Figure 1. Relative amount of protein remaining on the stainless steel surface during
desorption with NaOH.
 813
With NaOH solution, the first-order rate constant for desorption, kd, was dependent on
the NaOH concentration, temperature and flow rate. From the temperature dependence of k<f
the activation energy was estimated to be 210 kJ/mol. The kd increased with 0.5th power of
the linear flow rate, indicating the effect of mass transfer. The desorption behavior with the
enzyme solution was different in some points from that using NaOH solution. The activation
energy was 30 kJ/mol, which was much lower than that for the desorption with NaOH
solution. The remaining amount of the protein after 2-h treatment was much less in the case of
desorption with enzyme solution than in the case with NaOH solution at the same pH as

-
shown in Fig 2.

~
en 220
200 50°C
.=, 180
C
160
S
ec. 140
--
o
C
120
100
:::J
o 80
E 60
to
en 40
C
,-
c 20
'n; 0
E 6 8 10 12 14
CD
a: pH
Figure 2. Amount of protein remaining after 2-h treatment with NaOH (0) or proteolytic
enzyme solution (e).

CONCLUSIONS

The adsorption and desorption behaviors of f3-lactoglobulin were studied using stainless steel
particles of 100 meshes in mean size. In particular, the rate of desorption was studied by
analyzing the elution profile of protein desorbed from the column packed with the stainless
steel particles. Furthermore, the different behaviors in desorption with NaOH and proteolytic
enzyme solution were shown.

REFERENCES

1. Arnebrant, T., Barton, K. and Nylander, T., Adsorption of a-Iactoalbumin and ~­

lactoglobulin on metal surfaces versus temperature. J. Colloid Interface Sci., 1987, 119,
pp. 383-390.
 SUPERCRITICAL FLUID EXTRUSION - A NEW PROCESS

S.S.H. RIZVI and S.J. MULVANEY

Institute of Food Science
Cornell University, Ithaca, NY 14853

ABSTRACT
A new hybrid process which combines the attributes of supercritical fluids with extrusion processes
is described. Based on the commonality of high pressure operation for both processes,
supercritical fluid extrusion (SCFX) offers numerous added advantages over the conventional
process. Major features and unique system configurations for the process along with
microstructures of resulting products are presented.

INTRODUCTION
Food extrusion is a well established unit operation. It is used extensively in the manufacture of a
large number of products worldwide. The process provides a unique means of utilizing mechanical
and/or thermal energy to obtain uniformly processed products of defined characteristics.
Developed primarily for the purpose of making puffed snacks from relatively dry «20% water)
granular cereal ingredients, the use of this technology now covers a wide spectrum of consumer
foods and related products (Harper 1979). Examples of such products include ready-to-eat
breakfast cereal, soup bases, pregelatinized starch, textured vegetable proteins, pasta and pet
foods, among many others.

In general, the manufacture of extruded direct expanded products involves conversion of

mechanical energy into heat by the turning screws. The temperature of the extrudate rises over
150°C under high shear, causing plasticization of the feed ingredient. The molten and plasticized
mass is then forced through a die and the extrudate is puffed upon exit because of the phase
change of liquid water into steam. Product attributes are controlled by manipulation of the specific
mechanical or thermal energy inputs and residence time, as adjusted b~ such variables as moisture
content, barrel temperature, screw speed, feet rate, degree of barrel fill, etc.

In view of the above, typical cooking extrusion processes for direct expanded products,
operating under rather harsh environments of temperature and pressure, Impose the following
limitations. 1. Heat sensitive ingredients (proteins, flavorings, etc.) are excluded from the
formulations. 2. Water's role is undesirably coupled. It is both the plasticizer for the dough and
blowing agent for expansion. 3. Low in-barrel moisture causes extensive barrel and screw wear and
produces undesirable dextrins as a result of high energy input. 4. Expansion is controlled by phase
change of water, causing uncontrolled expansion with open cell structure. 5. The driving force for
expansion Is limited and cannot be independently varied.

The commonality of high pressure between supercritical fluids and extrusion processing and
their own specific attributes provide a means to overcome some of the basic limitations of the
conventional processes indicated above. Termed supercritical fluid extrusion (SCFX), the process
involves introduction of supercritical carbon dioxide (SC-C0 2) carrying micronutrients, flavorants and
colorants soluble in the fluid phase into a cooked and COOled dough contained within an extruder
of modified configurations. The details of the process have been reported elsewhere (3,4) and a
system schematic is shown in Figure 1.

While various other supercritical fluids can be advantageously exploited as well in this
process, SC-CO is generally the fluid of choice because of its greater solvency, moderate critical
temperature a;;[pressure, nontoxicity and availability (2). A few typical pressure profiles obtained
with a TX 52 corotating twin-screw extruder consisting of 14 heads are shown in Figure 2. The low
 815
shear (pressure) configuration permits the process to occur at subcritical conditions for CC?2 whUe
the medium shear configuration is designed for above critical conditions for CO 2 to prevah within
the extruder barrel.

Figure 1. Schematic diagram of the SCFX process showing a supercritical fluid system
--
Schematic: of Superc:ritic:al Fluid Extrusion System

coupled to a modified extrusion system.

14
- - - Low shcar/3.94mll1 die
12 ---+-- Medium shear/3.94mm die
"2 10 ----- !\'lcdiulll shear/) hule rcstridur
Q.
:5
OJ
8

S 6
'"'"~ 4
Q.

oL-,-,-~~ __ ~~~~~~~~~~~

o 1 2 3 4 5 6 7 8 9 10 11 12 13 14

<:fulD~~ ~

Head Number

Figure 2. Typical pressure profiles and the corresponding zones for the SCFX process.

A scannin!;! electron micrograph of both the crosssection and surface of SCFX produced
corn curl is shown In Figure 3. The most noticeable features are the smooth surface and the closed
thick-walled cells. These unique features can be effectively utilized to create a new generation of
surface coated, coextruded or honey-comb structured products of superior quality.

1. Supercritical fluids can be used to simultaneously add solutes, lower the in-barrel dough viscosity
and puff the extrudate. 2. Heat-sensitive proteins and other ingredients can be successfully utilized
to make expanded products. 3. High moisture Qow viscosity) extrudates reduce barrel and screw
wear. 4. The role of water as a plastiCizer and as a blowing agent is decoupled. Expansion is
controlled independently of moisture content. 5. Control of the expansion process resides in two-
phase flow of supercrltical fluid-dough system. 6. Die shaping characteristics become simHar to
injection molding process because of variable forces for product expansion. 7. Formation of
carbonic acid when SC-C02 dissolves in the aqueous phase of dough can be exploited to change
rheological properties and taste of the extrudate. 8. The extruder can be operated as a controlled
bioreactor to Induce desirable reactions such as hydrolysis of starch without any residual acidity in
the final product. 9. The use of supercritical fluids also offers the possibility of in-barrel extraction
of undesirable components prior to puffing.
816
CONCLUSIONS

The use of a supercritical fluid in conjunction with conventional extrusion processes is likely to
further enhance the extensive capability and range of application of extruders. Very special and
Innovative extruded products can be obtained using SCFX which can lead to new trends in the
manufacture of snack foods, breakfast cereals, pasta and baked goods. Further research is needed
both in engineering solutions to the many challenging problems this new technology poses and in
pursuing the science Involved.

SFX PREPARED EXPA D D CORNMEAL (35% MCWS)

(a)
Cross-section

xc . 0.37 9/ cm J

(b)
Surface

Figure 3. Scanning electron micrograph of the SCFX produced com extrudates.

The SCFX process is envisioned to offer the following specific advantages over the
conventional extrusion operations.

REFERENCES

1. Harper, J.M. 1979. Food Extrusion. Critical Reviews in Food Science and Nutrition 11 :155.
2. Rizvi, S.S.H., Benado, A.L., ZoIlweg, J.A. and Daniels, J.A. 1986. Supercritical fluid
extraction: Fundamental principles and modeling methods. Food Technology 40:55-64.
3. Rizvi, S.S.H. and Mulvaney, S.J. 1992. Extrusion processing with supercritical fluids. U.S.
Patent 5, 120, 559.
4. Mulvaney, S.J. and Rizvl, S.S.H. 1993. Extrusion processing with supercritical fluids. EQQQ
Technology (in press).
 RATE PROCESSES IN SUPERCRITICAL FLUID EXTRACTION

BEN J. MCCOY
Department of Chemical Engineering,
University of California, Davis, CA 95616

ABSTRACT

Rates of supercritical fluid (SCF) extraction with C02 depend

on (a) partitioning of the extract between the natural matrix
and the fluid, (b) rate of mass transfer of extract from the
interior of the matrix to the fluid, and possibly (c)
solubility of the extract in the supercritical fluid. These
issues are discussed and illustrated by calculations.
Investigations of the decaffeination of coffee beans
demonstrate the concepts. Decaffeination was measured as a
function of C02 flow rate, temperature and pressure. Soaking
the raw beans in water prior to decaffeination enhanced the
rate of extraction. The rate of decaffeination increased with
both pressure and temperature. The mathematical model
describes the external and intraparticle diffusion resistances
and the distribution of caffein between water and
supercritical C02. The partition coefficient for caffeine
distributed between water and supercritical C02 depends on
temperature and pressure.

INTRODUCTION

supercritical fluid extraction, especially with C02, is the

process of choice in many extractions of natural products,
e.g., flavor from hops, essential oils from spices, and
caffeine from coffee beans or tea leaves. Peker et al. (1992)
presented a theoretical model with the capability to describe
quantitatively a range of such physical extraction processes
(notation is shown in Table 1). The general case is based on
porous spherical particles whose pores are filled with a
liquid. For a thin, gradientless bed of particles (or well-
stirred vessel), the dimensionless differential equations with
their initial conditions are
dx/d9 + x/a = -~(x - my) (1 - a)/Q, x(9=O) = 0
818
dy/d9 = ~(x - my) (P + (1 - P)K), y(9=O) = 1/[P+(I-P)K

dF/d9 = X/(1 - a), F(9=O) = 0

since the concentrations were quite low for the the flow
system used by Peker et al. (1992), the solubility of the
solute in the SCF C02 was not a limiting factor. For an
enclosed batch system, the second term in the first equation,
which represents the flow of extract out of the vessel, is set
to zero, Le., x/a = o. Peker et al. (1992) gave analytical
solutions for x, y, and F, but the system of differential
equations is easily solved numerically, e.g., by a Runge-Kutta
method. By means of the coefficient m, the model accounts for
the partitioning of the extracted solute between the matrix
medium and the SCF C02. The equilibrium coefficient K allows
for adsorption-desorption effects. The mass transfer
coefficient ~ accounts for intraparticle diffusion and
external mass transfer resistances.
Coffee beans are always saturated with water during SCF
decaffeination. This allows the caffeine crystals to dissolve
completely in the water (K « I), and transfer more easily to
the C02. For dry coffee beans the extraction process is
extremely slow, since the solid caffeine must desorb from the
plant tissue (Peker et al., 1992). For caffeine distributed
between C02 and water the partition coefficient m increases
wi th both temperature and pressure, thus increasing the rate
of decaffeination. A correlation that describes the data
presented by Peker et ale (1992) is

In principle, a coefficient m can also represent the

partitioning of a solute between the SCF C02 and a nonpolar
liquid.

T = 313, 323, 337, 353 K P = 10.3, 13.8, 16.5, 19.3 MPa

LI. .1 .1

.01 .01

M .001 .001
313 K 10.3

.0001 .0001
0 20 40 60 80 100 0 20 40 60 80 100
9 9

Figure 1. Effluent concentration, x, and cumulative fraction,

F, for SCF C02 decaffeination; effect of changing T and P when
a=O.371 and P=O.515.
 819
Figure 1, based on parameter values given in Table 2,
shows the effects of increasing temperature and pressure,
respectively. The mass transfer coefficient ~ decreases with
P and T. The two sets of curves in Figure 1 are similar since
the values of ~ and m are of similar magnitude for the two
sets. The effluent concentration x of caffeine immediately
jumps to a value that slowly declines with time as the
caffeine is depleted. Coffee-bean decaffeination usually
requires about ten hours. The cumulative fraction approaches
unity as caffeine is exhausted.

TABLE 1. Notation
Bi = Biot number (kfR/De)
c = extractable solute concentration in the C02, g/ml
ci = solute concentration in the pore volume of the matrix,
g/ml
Co = initial total solute concentration, g/ml
De = effective intrap~rticle diffusion coefficient for solute
in the matrix, cm /s
F = cumulative fraction of solute extracted
kf external mass-transfer coefficient, cm/s
kp = combined mass-transfer coefficient (15kf/R)/(5+mBi), cm/s
K equilibrium adsorption coefficient
m = equilibrium partition coefficient of solute distributed
between C02 and the matrix: (conc of solute in C02)/(conc
of solute in matrix)
R = radius of spherical particles, cm
t = time
x = dimensionless solute concentration in C02 (c/co)
y = dimensionless solute concentration in matrix (ci/co)
a void fraction in bed
~ = porosity of the particles
~ dimensionless mass-transfer coefficient (kpr)
e = dimensionless time (t/r)
r total bed volume/volumetric flow rate, s

TABLE 2. Values of parameters for Figures 1 and 2.

T, K P, MPa ~ m
313 13 .8 1.07 0.00089
323 13 .8 0.88 0.0036
337 13.8 0.68 0.0073
353 13.8 0.36 0.021
323 10.3 1.08 0.0013
323 16.5 0.79 0.0054
323 19.3 0.35 0.0215

REFERENCES
1. Peker, H., Srinivasan, M.P., Smith, J .M., McCoy, B.J.,
Caffeine extraction rates from coffee beans with supercritical
carbon dioxide. AIChE J., 1992, 38, 761-770.
 MEMBRANE SEPARATION OF SUPERCRITICAL FLUID MIXTURE

KOZO NAKAMURA, TERUHIKO HOSHINO, AKITOMO MORITA

MASANORI HArrORl AND RYUJI OKAMOTO
Department of Biological and Chemical Engineering, Gunma Univ.
Kiryu, Gunma 376, Japan

ABSTRACT

the ceramic membrane as well as the polymer membrane were used to examine the possibility
of mem brane separation of the supercritical n uid mixture. The permeation rate of carbon
dioxide exhibited a maximulll peak ncar the critical pressure when it was measured at a constant
temperature with change of pressure. Polyethylene glycol 400(PEG 4(0), PEG 600 and triolein
were used as the model solutes wilh the maximum concentration of 1%. The rejection was
ncgath·c in the composite polymer J11cmbrane(NTOS-2100) and approached zero with increase
in the permeation nux. The rejection of PEG 400 was from 0.5 to 0.8 at 40'C and :20 MPa in
the thin porous silica membrane with the avemge pore size less than 0.6 nm.

INTRODUCTION

The supercritical carbon dioxide(SeCOz) is a safe solvent with several excellent properties such
as low viscosity and high diffusivity. The solubility of solutes in seeo:! is, however, much
lower than that in appropriately chosen organic solvents, and the solvent ratio usually becomes
large in See02-extraction. It means that the energy-saving recovery of see02 is necessary in
a largc scalc extraction. The membrane could be useful for this IHlfjx)se, allll it will be also
applied to the reach.)r of enzymatic reaction in sce02. The research works on the membrane
proccssing of supercritical nuid(Sef-) were scarcely rqxlI"ted[l,2J. and it is the jlllrposc uf this
study to pursue the possibility of the membrane separation of SeF mixture using the spedally
prepared ceramic membmne as well as the commercial polymer membmne.

MATERIALS AND METHODS

(l)Membranes and modules: The polymer membrane was the composite membnllle of the
 821
polyimide ultrafiltration membrane with the skin layer of cross-linked silicone. This
membrane(NTGS-2100) was developed by Nitto Denko Corp. for separation of organic gases.
The circular flat membrane with diameter of 72 mm was used in the experiments. The thin
porous silica membranes were prepared by the sol-gel technique using a cylindycal porous, a-
alumina substrate(average pore diameter: 1 f1. m, outside diameter: 1 cm, porosity:about 50%)[3].
Three membrane with the average pore size of 1.5 nm(A), 0.6 nm(B) and less than 0.6 nm(C)
were used in the experiments, and the length was 51 mm. The modules were designed by
ourselves and prepared by the manufacturing company of high pressure vessels.
(2)Materials: The carbon dioxide in a high pressure cylinder was of food grade and purchased
from the local supplier. Polyethylene glycol 400 and 600 were the reagents of WAKO PURE
CHEMICAL INDUSTRIES, Ltd., and triolein was the product of SIGMA with approximate
purity of 99%.
(3)Methods: The experimental apparatus and the method were almost as same as those
explained in the reference[2].

RESULTS AND DISCUSSION

(l)permeation rate: Fig.l(a) is an example of the permeation rate when the membrane is
ceramic. The permeation rate takes the maximum value near the critical pressure and then
dropped to the almost constant value. This change is remarkable in the membrane with the
larger pore size. The permeation rate is proportional to the reciprocal of kinematic viscosity as
shown in Fig.l(b), which is expectative for the permeation of fluid through the porous
membrane. Its dependence on the kinematic viscosity, however, becomes weaker beyond the
critical pressure. The appearance of the peak followed by the sharp fall was also observed in the

,....., s 5

:1;1 ·
C1I Key Momlnne Key Membrane TIt) i
0..
.., 0
"
(J A

...
E
:::::
4 I).

0
B .....
Po.
~
4 •
().
B
0 50
<J
:
0 <I) 0 c 40 1<» ()...~
E J N
<» 1~'~
~ ~ 3
IU o () /1 l:.l:.~_"-
...
t;
2
8
..; 2 .;;. /A6~-'-
c /0 I
.!: ~
Gi
GI
E 1
,
: Pc = 7.38 MPa
--
....,
()
()
•
Y ~
0 ........ 0
~h-o-g-o%-
!
'"'
IU
0.. ':./ T = 40°C
~o""""""'" I
0 o
0 10 20 30 0.0 0.2 0.4 0.6 0.8 1.0 1.4
Average pnssure [Mra]
lIv X 10. 7 [51m 2]

(a) Effect of Pressure (b) Effect of kinematic viscosity

Fig.! Effect of pressure and kinematic viscosity on permeation rate of CO2 in ceramic
membranes.
822
0.6 r---~---.---~---,

0.4

.....,
7
~
0.2 ,...... O.S

c
.2 .~ 0.0 .............................•7" .•......................
ti ..0.2
...,.cu
cu 1i e_ ~
Clt: -0.4 ..... e r--.----.-----,
~ Kc)·

•...
-0.5 ~Icmbnnc
SaI ••e : PEa 400
A q ..... , : 0.1-1
.0.6
C pPoIP., : 14.1
-0.8 L-_~~_--' _ _----,-_ _...J
·1.0
0.0 O.S 1.0 0.0 0.2 0.4 0.6 0.8
Permeation flux (mol/m 2 s1 Transmembrane pressure [MPa]
(a) Polymer membrane (b) Ceramic membrane

Fig.2 Rejection of polyethylene glycol in Polymer membrane(a) and ceramic membrane(b).

polymer membrane and could be related to the condensation of SCC02 to narrow the pores in
the membrane.
Fig.2(a) and (b) show the rejection of PEG 400 and 600 measured using the polymer
membrane and the ceramic membrane respectively. The rejection was negative in the polymer
membrane probably due to the solubility of PEG in the polymer membrane being better than
that of SCC02. The sieving effect of the membrane became remarkable to retain the solute well
as shown in Fig.2(b) when one lot of the ceramic membrane with the average pore size
probably less than 0.6 nm (not determined) was used. The effect of temperature, pressure,
trans-membrane pressure and solute concentration was also measured, but the explanation of
the results have to be omitted as space is limited.

CONCLUSIONS

1. The permeation rate of SCC02 became maximum near the critical pressure and fell sharply to
the almost constant level in the polymer membrane and in the ceramic membrane with large
pore size.
2. The rejection of PEG 400 and 600 was negative in the polymer membrane, while PEG 400
was well retained by the ceramic membrane the average pore size of which was supposed to
be less than 0.6 nm.

REFERENCES

1. 1.M.Martinet et.al.: Proc. Int. Symp. on SCFs, Tome 1, 1988, p.343-347

2. K.Nakamura: Fouling and Cleaning in Food Processing(ed. by H.G. Kessler and D.B.
Lund), 1989, p.258-267
3. M.Asaeda et.al.: Proc.IFMEST92 on Porous Materials (to be published in 1993)
 ADSORPTION ISOTHERMS FOR SUPERCRITICAL CARBON
DIOXIDE ON PROTEINS AND POLYSACCHARIDES

*
KOZO NAKAMURA, TERUHIKO HOSHI NO, YOSHITSUGU SUZUKI
AND NORIHIKO YOSIZAWA
Department of Biological and Chemical Engineering,
Gunma University, Kiryu, Gunma 376, Japan
* Department of Agricultural Chemistry, Faculty of Agriculture.
University of Tokyo, Yayoi. Bunkyo-ku, Tokyo 113. Japan

ABSTRACT

The adsorption isotherms for C02 on several proteins and polysaccharides were measured at
pressures of 0-29.4 MPa by two different microbalances, gravimetric and piezoelectric. The
adsorption isotherms measured gravimetrically all exhibited the maximum peak near the critical
pressure Pc and remained constant at the higher pressure, the level differing each others. The
piezoelectric microbalance using the quartz crystal (9 MHz, At-cut) could measure the
"absolute" amount of fluid sorbed in the adsorbent which was larger especially at the high
pressure than the "surface excess" adsorption measured gravimetrically. The density of
adsorbed layer was calculated using the data of the surface excess and the absolute adsorption
isotherms.

INTRODUCTION

The adsorption phenomenon of high pressure gas or supercritical fluid is fundamental to the
application of expansion of biomaterials or sterilization of the microbial cells, and it is also
related to enzymatic reactions and chromatographic or membrane separation. But the adsorption
of supercritical fluid on the biological materials has not been well studied. In this study, the
adsorption isotherms for carbon dioxide on several proteins and polysaccharides were
measured at the pressures of 0-29.4 MPa by the different microbalances, gravimetric and
piezoelectric.

MATERIALS AND METHODS

All materials used in this study were dry proteins and polysaccharides. The apparatus used for
the adsorption experiments by a piezoelectric quartz crystal is shown in Fig. 1. The apparatus
824

Figure 1. Experimental apparatus for piezoelectric microbalance.

by the quartz spring has been described in the literatures [1, 2]. The sample was coated
on the surface of a 9 MHz At-cut quartz crystal and the frequency change was measured by a
frequency counter to follow the time course of the weight change caused by sorption. The
principle of piezoelectric quartz crystal microbalance is referred to elsewhere [3].

RESULTS AND DISCUSSION

Figure 2 shows the adsorption isotherms for C02 on polysaccharides (corn, potato, dextrin and
cellulose) measured by the quartz spring at 313 K. In Fig. 2, the adsorption became maximum
just above Pc and constant at the higher pressure. This profiles are similar to those of proteins
reported previously [1, 2]. The high level adsorption of cellulose can be explained by its
micro-structure.
Figure 3 shows the adsorption isotherms for C02 on proteins measured by the piezo-
electric microbalance and the quartz spring at 313 K. The piezoelectric microbalance could

~ 6.0
~
P=7.38 o Corn A Celluulose
MPa o Potato 6 Dextrin
~
~ 4.0
A-
0 A-
U
0 0
"C
u 2.0
...
.D
0
'"
"C 313 K
< 0.0
0 10 20 30
P [MPa]

Figure 2. Adsorption isotherms for CO2 on polysaccharides.

 825
40

tIi!
30
~
M
0
u 20
"'0
.8...
0 10
'"
«
"'0

0
0 10 20 30
P [MPa]

Figure 3. Adsorption isotherms for C02 on proteins at 313 K.

xcf 0.5
"'0
!a Casein 313 K
c: o. 10 20 30
P [MFa]

Figure 4. Density of adsorbed layer for casein at 313 K.

measure the "absolute" amount of fluid in the adsorbent which was larger especially at the high
pressure than the "excess" adsorption measured by the quartz spring. The relation between the
absolute adsorption 11 W and the excess adsorption 11 W' is expressed as 11W = 11 W' + Pb
(I1WI Pa ). The density of the adsorbed layer Pa can be calculated from this equation, that is,
Pa = Pb 11WI ( 11W - 11W') where Pb is the bulk density of C02. The density of the adsorbed
layer for casein was calculated at 313 K using the data of Fig. 3, and it is shown together with
the bulk density in Fig. 4. There still appeared the maximum peak in the density of the adsorbed
layer at the pressure where the adsorption became maximum. The peak density of l.03xl03
kg/m 3 was approximately equal to the hypothetical density estimated from the Dubinin-
Nikolaev's equation which was recommended for estimation of the adsorbed layer above the
critical temperature [4]. At the near-critical point, there will be occurring the clusters of C02
molecules which may be more condensable than the molecules. Therefore, the formation of the
adsorption maximum at the near-critical point may possibly correspond to the formation and the
disappepearance of cluster.

REFERENCES
1. Nakamura, K., Hoshino, T. and Suzuki, Y., Biosci. Biotech. Biochem., (submitted).
2 Nakamura, K., Hoshino, T. and Ariyama, H., Agric. Bioi. Chem., 1991,55,2341-2347.
3. Stockbridge, C.D., In Vacuum Microbalance Techniques, ed. H.K.Behendt, Plenum Press,
NY, 1966, vol.5, pp. 147-190.
4. Suzuki, M., Adsorption Engineering, Elsvier, Oxford-NY-Tokyo, 1990, p. 44-46.
 APPLICATION OF SUPERCRITICAL C02 FOR FOOD PROCESSING

z.KNEZ, M. SKERGET
University of Maribor, Faculty of Technical Sciences,
Dept. of Chemical Engineering, SLO-62000 Maribor,
Smetanova 17, Slovenia.

ABSTRACT

The influence of the process parameters to the composition of the extracts and mass
transfer coefficients for extraction of paprika are presented.
The equilibrium solubilities of capsaicin (pungent component of paprika) in
liquid and supercritical C02 were measured by the static analytic method. The data
were correlated by use of the thermodynamic model relating the solubility to the solvent
density and a model for the dependence of the enhancement factor on the density of the
solvent. The data were also correlated with the Peng-Robinson equation of state.

INTRODUCTION

High pressure technology offers the industry an enormous opportllnity to develop novel
products of high value. High isostatic pressure is able to inactivate microorganisms and
enzymes and thus can be applied for preservation of some foodstuffs which is usually
performed by high temperature treatment (1).
Recently supercritical fluids have been applied as a solvent for non extractive
applications in high pressure micronisation, in chromatography and as a chemical and
biochemical reactions media (2). On the other hand, supercritical fluids have been used
as a solvent for a wide variety of extractive applications.
For the design of a sub- or supercritical fluid extraction system, data are required
on the pressure and temperature required for extraction and separation, the type and
quantity of the solvent, the recirculation rate, and energy consumption. This information
can be obtained from phase equilibria, mass transfer measurements, and from the
thermodynamic analysis of the conditions in the extraction and separation step (3).
In the literature (4) the process for production of paprika oleoresin with
supercritical C02 are relatively well described, however, there are practically no data on
the solubility of capsaicin and no mass transfer data for extraction of paprika.
 827
EXPERIMENTAL

Apparatus and experimental procedure for determination of mass transfer coefficients

and of solubility of capsaicin C02 was already described (5,6).

RESULTS AND DISCUSSION

Mass transfer coefficients

Paprika was extracted in the temperature range 20-60oC and in the pressure range 80 to
450 bar and the kinetics of extraction were followed.
Mass transfer coefficients were determined for the constant rate periods where
steady state mass transfer prevails(3), and are presented in Table 1.

TABLE 1
Parameters for the extraction of paprika and mass transfer coefficients.

Diameter of equivalent sphere d = 0.15xlO-3m

Specific area As = 27200 m2/m3
Solvent C02
Temperature 42°C

Pressure (bar) 90 90 150 300 400

Solvent flow rate (k~) 23.5 32.8 25.78 30.45 32.7
Mass transfer coeff. (10 m/s) 0.53 1.47 0.56 0.37 0.30

Equilibrium data
The experimental equilibrium solubility data for capsaicin in carbon dioxide were
determined at the temperatures of 25, 40 and 60°C and in the pressure range from 70 to
400 bar. The equilibrium mole fraction (yi) of the solute in the supercritical C02 as a
function of system pressure is presented in Figure 1. The data were correlated with
thermodynamic correlations such as In y vs. In pr and log E vs. pr (5), and with the Peng-
Robinson equation of state. The physical data for capsaicin were determined by the
Lydersen group contribution method.
The constants Qa,Qb and acentric factor in the Peng-Robinson equation of state
were for capsaicin not evaluated from critical data. They were obtained by regressing
the solubility data measured at 25 and 40°C. On this way the PREOS was modified. The
omega-parameters were »set free«, so they have different values from those of PRo The
results are summarised in Table 2. It can be seen that the temperature does not much
influence the coefficient kij, which could be the consequence of the way of calculation.
Fig. 1 shows that the approximation of the experimental data is good for 25°C,
where the average absolute relative deviation is 4.63% but it is worse for 40°C, where
AARD is 30.07%.
828
TABLE 3
Properties of Capsaicin estimated by regressing the experimental data
with the modified PREaS

Property Value of property

Qa 0.13797
Qb 0.15968
5.89741
~(m~mol) 0.89142
kij(298K) -0.27900
kij(313K) -0.27700

•
35
V)
• 25C 40C t:. 60C - PREOS t:.
Ul 30 t:.
0
- 25 t:. t:.
:x: t:.
~ 20
0
..g 15
~

-:E
t.l:l
<1)
0
10
5
0
50 100 150 200 250 300 350 400
Pressure (bar)
Fig. 1. Solubility isotherms of Capsaicin as a function of pressure and PREaS corr.

CONCLUSION

The measured mass transfer coefficients for extraction of paprika and equilibrium
solubilities of capsaicin in C02 can be useful in design of paprika extraction plant.

REFERENCES

1. Hayashi, R, Utilization of pressure in addition to temperature in food science and

technology, In High pressure and Biotechnology, Eds. C. Balny, R Hayashi, K.
Hermans, P. Masson, John Libertey, Grand Mote, 1993, pp 185-92.
2. Nakamura, K., Hoshino, T., Novel Utilization of Supercritical Carbon Dioxide for
Enzymatic Reactions and other processings. Int. Workshop, Jakarta, Sept. 1991
3. Brunner, G., Mass separation with supercrit. gases, Int. Chern. Eng" 1990, 30,
191-205.
4. Coenen, H., Hagen, R, Knuth, M., German Pat. DE 3114593 C1, 1982.
5. Knez, Z., Steiner, R, Solubility of Capsaicin in dense C02, The Journal of
Supercritical Fluids, 1992,5,251-5.
6. Knez, Z.,Posel, F., Hunek, J., Golob, J., Extraction of plant materials with supercriti-
cal C02. In Proc. II. Int. Conf. on Sup. Fluids,ed.M.Mc Hugh, Boston, 1990, pp.
101-4.
 SUPERCRITICAL FLUID EXTRACTION OF AROMA COMPOUNDS FROM
AROMATIC HERBS (Thymus 7J'gis and Coriantbwm sativum).

A. Lopes Cardoso, M. Moldao-Martins, G. Bernardo-Gil*, M.L. Beirao da Costa

Dept. of Food Sci. & Tech. Inst. Sup. Agronomia, Tapada da Ajuda, 1399 Lisboa Codex.
*Dept. ofChem., Inst. Sup. Tecnico, Av. Rovisco Pais, 1096 Lisboa Codex, PORTUGAL

ABSTRACT

Supercritical fluid (SCF) extraction was compared to steam and Clevenger distillation, in terms
of yield and composition of extracts, when applied to thyme and coriander leaves. In thyme
extracts, higher production yields were always obtained by the Clevenger extraction method.
Using SCF extraction high yields were also obtained but the extracts included other kinds of
compounds. In respect of chemical composition of steam and Clevenger distillation extracts
show similar profiles. Supercritical extracts present, besides the same components as that of
steam distillation, non aromatic compounds. All the extraction methods tested produced very
low yields of extracts when applied to coriander leaves, nevertheless the patterns of the
chromatograms are quite different among those methods.

INTRODUCTION

Aromatic herbs are quite abundant as flavoring ingredients in the productions of food. Many of
these plants only produce large quantities of flavor compounds in a very short period of the
vegetative cycle. So, an important task is to find the most suitable extraction procedure and to
produce an extract of interesting composition, with good yield.
Thyme and coriander leaves are the herbs commonly produced in Portugal. In a
previous work [1] the evolution of thyme aromatic compounds throughout the vegetative cycle
were studied and it was found that the richer extracts were produced at the flowering and post-
flowering stages.
C~ supercritical fluid extraction is an interesting process to be applied in the food
industry as it produces solvent-free extracts and is non-toxic [2]. On the other hand it should
be applied only when high added values products are expected.
The aim of present work is to study whether the SCF extraction used under several
experimental conditions can produce aroma from coriander and thyme. The yields and the
composition of the extracts were compared to those obtained by steam and Clevenger
distillation.
830
MATERIAL AND MEmODS

Raw materials
Thymus zygis spp silvestris was coUected in the north of Portugal during the flowering period,
and was stored in the dark at the room temperature for a week.
Coriandrum sativum leaves were purchased from commercial sources.

Extraction methods
SCF extraction was conducted under the following experimental conditions:
Thyme: T=313 Kat 10, 15, and 20 MPa.
Coriander leaves: T=308 K at 9.5, 12.5 and 20 MPa.
Steam and Clevenger distillation were carried out at the atmospheric pressure for 30 min.

Methods of analysis
GC analysis was performed with lIP 5890 instrument equipped with a lIP-5 (5% diphenil, 95%
dimethylpolysyloxane (50 m x 0.32 mm; coating thickness 0.17 ,.un) and a FID.
Analytical conditions: injector temperature 473 K; detector temperature 523 K; oven
temperature 333 KIlO min; oven temperature programmed 337 K- 453 Kat 2 Klmin, then
453 K - 473 Kat 10 Klmin, then isothermal at 473 K for 50 min; flow rate of carrier gas 2
mlImin.
Relative concentration were evaluated using the peak areas, without correction for the
response factor.
GC-MS analysis were performed with lIP-589O instrument equipped with a WAX
capillary column (30 m x 0.32 mm; coating thickness 0.1 ,.un) and 7.00 eMvolts mass
selective detector.
Analytical conditions: injector temperature 503 K; oven temperature 333 Kl15 min,
oven temperature programmed 333 K - 493 Kat 2 Klmin, then isothermal for IS min; flow
rate of carrier gas 2 ml He/min; source 70 eVolts.
Identification of compounds: standard of most of components were used. When the
standards are not available mass spectrum of unknown compound were compared to the
library.

RESULTS AND DISCUSSION

SCF extraction of thyme showed that the maximum yield was obtained for 313 K at 15 MPa,
althougth the other two yields were quite similar. This yield (6.36%) is considerably higher
than the one obtained by steam and Clevenger distillation (0.6 % and 0.7 %, respectively). This
difference is explained by a high extraction of other than aroma compounds like waxes,
carotenoids and chiorophyls.
The difference in composition of the extracts obtained by the different methods is
shown in Figure 1. It can be seen that geranyol and geranyl acetate are almost exclusively
extracted in the distillation procedures.
 831
35
30

25 • liMP.
.15MP.
2.
•-1!. 15
• liMP •

10 • s. OtItlL
• Cln. Ot..
5
0
a-
'1.... cy. . . .

Figure 1. Comparative extract composition of thyme produced by cfifIereot methods

a-Pinene is present to a similar extent in the SCF extracts obtained at 313 K at 10 MPa
and in the product of Clevenger distillation. p-Cymene is almost absent in the SCF extract
obtained at 10 MPa (0.07010) and is very low in the product obtained at 15 MPa (1.5%)
although it is present in a high content on the other extraction procedures (11 % for Clevenger,
5.6% for steam distillation and 7.2% for 20 MPa). Thymol is extracted at similar levels by all
the tested methods.
Based on the resuhs obtained in the extraction of coriander leaves, it must be stated
that the yields were too low for all the methodologies to be quantified. SCF extraction showed
that the maximum yield was obtained for 308 K at 12.5 MPa, althoagh the other two yields
were quite similar. The difference in composition of the extracts obtained by the different
methods is shown in Figure 2. Steam distillation produced a richer extract, mainly composed
by monoterpenes, aldehydes and alcohols. Monoterpenes appear at a high composition in the
supercritical fluid extract obtained for 308 K at 9.5 MPa. The SCF extracts are rich for higher
molecular weight compounds, which have not yet been identified.
UI
16
14 89.5MPa
12
.1l..5MPa
10
•
~
S .1OMPa
6
4 • S&. Dbt.
2
0 <lev. DBt.

!• II 1
I

Figure 2. Comparative extract composition of coriander produced by difterent methods

REFERENCES

1. Moidio-Martins, M .M., Beirao-da-Costa, M.L.and Bernardo-Gil, M.G.,Comparative

extraction procedures of volatile compounds from Thymus Sigis. 8th world Congo of Food
Sci. and Techn., Toronto, September 1991.
2. Caragay, A.B.and Little, A.D., Supercritical fluids for extraction of flavours and fragrances
from natural products. Perfumer & Flavourist, 1981,6,43-55.
 Separation of aroma components from soy sauce
by continuous supercritical CO 2 extraction

Yoshihisa KITAKURA*, Hitoshi IMAMURA * * , Saburo HAYAKAWA**

Mitsutoshi HAMAN 0 * , Hikotaka HASHIMOTO*

* Kikkoman Corporation, 339 Noda, Noda-shi, Chiba 278, JAPAN

** Nippon Sanso Corporation, 4-320 Tsukagoshi, Saiwai-ku,
Kanagawa 210, JAPAN

ABSTRACT
Tradi tional soy sauce making process have not been able to
separate aroma compounds from body component, which are re-
garded as useful food resources.
In this investigation, soy sauce aroma components were
successfully separated from soy sauce by using a continuous
supercri tical CO 2 extraction system and the concentrate of
soy sauce aroma was extracted at the same time. These new high
quality food resources have not been available using tradi-
tional processes.

INTRODUCTION

Soy sauce is one of the Japanese traditional seasonings and is

indispensable for Japanese-style cooking. Improvement of the
separating technologies enables us to make new types of soy
sauce. For example, electri~al dialysis makes it possible to
provide lower-salt soy sauce, which is suitable for preventing
excess sodium intake; spray-drying makes it possible to pro-
vide powdered soy sauce, which is used for instant noodle
soup, and other dried foods [1]; membrane makes it possible to
provide non-pasteurized soy sauce, which has neither cooking
flavor nor color increase [2].
This investigation is a part of an effort to develop a
new type of soy sauce. A continuous supercritical CO 2 extrac-
tion system was expected to provide new soy sauce which has
less aroma but has delicious taste and soy sauce aroma con-
centrate at the same time. Those new types of soy sauce are
regarded as useful food resources.
 833
MATERIALS AND METHODS

Materials
In this investigation, a soy sauce model system simulating
real soy sauce was applied. It is comprised of a solution of
salts, acids, amino acids, and several soy sauce aroma compo-
nents in water. Each aroma component is added at the concen-
tration of 100 ppm. This soy sauce model system is easy to
analyze.
Extracting conditions and aroma concentrating conditions
were investigated using this soy sauce model system and the
examination using real soy sauce followed.

Extraction
A continuous supercri tical CO 2 extracting pilot plant with a
counter flow extracting tower was used. To judge the extract-
ing conditions, the elimination of isoamylalcohol, furfurylal-
cohol, methionol, 2-acetylpyrrole was caliculated;

Final aroma concentration of solution

Elimination 1-
Initial aroma concentration of solution
Quantitative analyses were carried out by gas chromatography.

Concentration
Since aroma components are very thin, it was decided to con-
centrate soy sauce aroma components into solvent. An absorbing
column was attached to the continuous supercri tical CO 2 ex-
tracting pilot plant.
We investigated the recovery of aroma components from
aromatic CO 2 into absorbing solvent. First, absorbing pres-
sure and temperature were examined. Next,· various solvents
were examined. The concentrating conditions were judged by
aroma concentration in absorbing solvent after experiments
with the same absorbing period.

RESULTS AND DISCUSSION

Extraction conditions
A Raschig ring was decided to be used because of its high
elimination data. The longer extracting tower achieved higher
elimination. As for extracting temperature, 40' C and 60' C gave
higher elimination than 20·C. Extracting pressure was respon-
sible for the elimination of furfurylalcohol and methionol.
These components are more easily taken away under higher pres-
sure.
Lower CO 2 flow rate and lower food material flow rate
show higher elimination and furfurylalcohol and methlonol are
mostly eliminated under these conditions.

Concentration conditions
Atmospheric pressure, and 50, 80, and 200 kgf/cm 2 were exam-
ined to trap aroma concentrate. Among those pressures, 50
834
kgf /cm 2 was most sui table for our purpose. As for tempera-
ture, 40· C is more effective for aroma recovery than 30· C.
Water, 50% (w/w) , and 99.4% (w/w) ethanol solutions were
examined as solvents. Water is not suitable for solvent, be-
cause it does not absorb aroma components efficiently. On the
other hand, the volume of 99.4% ethanol reduced during con-
centration. It appeared that CO 2 gas blows out ethanol, so
99.4% ethanol is not suitable for our purpose. It was decided
to use 50% ethanol solution as the solvent.

Production
An extracting tower 1500 mm long with a Raschig ring in it was
prepared and real soy sauce aroma was extracted under the
condition of 200 kgf/cm 2 , 40·C, and CO 2 flow rate of 12NI/min.
Soy sauce was fed under flow rates of 15 ml/min and 5 ml/min.
Aroma concentration of soy sauce becomes less than 1 ppm.
Very weak soy sauce aroma was smelled when soy sauce was fed
at 15 ml/min but less aroma at 5 ml/min.
Aroma components were absorbed into 100 ml of 50% ethanol
solution under the condition of 50 kgf/cm 2 , 30· C. Almost all
aroma components in solvent are concentrated up to 10 or 30
times as dense as those in soy sauce. Loss of solvent was only
2 or 7 ml, so we expected this condition may be practicable.

CONCLUSIONS
From the above experiments, it is concluded that less aroma
soy sauce and soy sauce aroma concentrate can be obtained
using this pilot plant under the conditions suggested by the
present investigation using a soy sauce model system.

ACKNOWLEDGMENT

This investigation was carried out under the governmental

project of "The Japanese Research and Development Association
for High Separation System in Food Industry" and was men-
tioned in its report published in book form [3].

REFERENCES
1. Hamano,K. ,Behaviors of Water and Quality on the Dehydrated
foods. Nippon Shokuhin Kogyo Gakkaishi, 1982, 30, 125-132.

2. Hamano,K. and Suzuki,K., Nama-shoyu no Tokusei to Sono Sei-

zo-gijutu. Japan Food Science, 1991, 30, 53-57.

3. Renzoku-chorinkai CO 2 Chusyutu-ho ni yoru Shoyu no Ko-fuka-

kachi-ka Gijutu no Kaihatsu. In Kinousei Shokuhin Sozai no
Koudo Bunri Seisei to Kaihatsu, Syokuhin Sangyo High
Separation System Gijutsu Kenkyu Kumiai, Tokyo, 1992,
pp.453-478.
 SUPERCRITICAL CARBON DIOXIDE EXTRACTION OF CAROTENOIDS FROM
CARROTS

MOTONOBU GOTO, MASAKI SATO AND TSUTOMU HIROSE

Department of Applied Chemistry, Faculty of Engineering, Kumamoto University, 2-39-1
Kurokami Kumamoto 860, JAPAN

ABSTRACT

Carotenoids were extracted from freeze-dried and raw carrots with supercritical carbon dioxide.
For the freeze-dried carrots carotenoid existing only about lO,...m in the outer shell of each
carrot particle was rapidly extracted and the rest was extracted very slowly. However for the
raw carrots carotenoid was completely extracted and the extraction rate increased with pressure
and ethanol as an entrainer. HPLC analysis of the extracts indicated that extracted carotenoids
are mainly a- and l3-carotene.

INTRODUCTION

Carotenoids are one of the major natural pigments which are widely distributed in nature.
Carotenoids, in particular l3-carotene, are important for food and pharmaceutical industries,
since they are a precursor to vitamin A and a coloring material. The extraction of carotenoids
from natural materials by supercritical carbon dioxide extraction process has been studied [1-
3]. The objectives of this paper are to evaluate the effect of water present in the carrot cells on
the extraction process and to study the extraction kinetics of carotenoids from carrots with SC-
C02.

EXPERIMENT

Carrots were diced into 5mm cubes and freeze-dried. Freeze-dried carrots were ground and
sieved into average size of diameter of 0.26mm, 0.47mm and 1.12mm. Fresh ground carrots
were used as raw carrot sample. The carotenoid content in the carrots was determined by
absorbance measurement at 448nm of ethanol extracts with cell disruption.
Freeze-dried or raw carrots were placed in an extractor (1Ocp x13mm). Pure C02 or C02
with 1, 3 and 5wt% ethanol was used as solvents. Extraction was carried out at 313, 323 and
333K and at pressures of 7.8 - 29.4MPa. The effluent concentration of carotenoids in SC-
C02 was monitored continuously with a high-pressure UV detector at 436nm for dried carrots /
pure C02 system. During the extraction run, the extracts were dissolved in ethanol
periodically and the amounts of carotenoids were measured by the absorbance. The
constituents of the extracts were determined by HPLC with YMC C-18 column with
ethanol/acetonitril (40/6OvIv).

RESULTS AND DISCUSSION

836
Initial carotenoids contents were about 7.OxlO-3 and 2.OxlO-2 kglkg-dry for freeze-dried and
raw carrots, respectively. Carotenoids must be partially denatured during the freeze-drying
process. J}-carotene is unstable, e.g., cis isomerization accelerated by heat and light [3]. The
absorbance may decrease with the denaturation to reduce carotenoids contents.

0.012

0.010 raw carrots

• 313K-14.7MPa
~ 0.008 • 313K - 29.4MPa
"0
I
0> • 313K-14.7MPa,
c:1!
0> 0.006 3wt%EtOH
~
~ 0.004 freeze-dried carrots
o 313K-14.7MPa
0.002 JJ.. 313K - 29.4MPa
D 313K-14.7MPa,
3wt%EtOH
0.1 0.2 0.3 0.4 0.5
C02 [Nm1
Figure 1. Effect of pressure and entrainer on the cumulative extraction histories from raw and
freeze-dried carrots (Dp < O.lmm, flow rate = 1.1xlO-7 m 3/s at pump).

~ - carotene

(J. - carotene
Xanthophylls
i 313K and 14.7MPa
i 448nm

I
-"1
Q 10 20 30 40

Retention time [min]

Figure 2. HPLC chromatogram of the extracts from raw carrot with SC-C02.

The influences of pressure and entrainer are shown in Figure 1 for both freeze-dried and
raw carrots. The extracted residue was colorless for raw carrots, while it was still colored for
freeze-dried carrots. For extraction from raw carrots, carotenoids were extracted completely
and the increase in pressure and the addition of entrainer enhanced the extraction rate, because
of increased solubility of carotenoids [3]. On the other hand, for extraction from freeze-dried
carrots, about 70% of carotenoids were extracted and neither pressure nor entrainer affected the
extraction histories. The discrepancy between the amounts extracted after the complete
extraction and the original contents for raw carrots may be owing to denaturation of carotenoids
during the extraction or the measuring process. Although the initial extraction rate for freeze-
 837
dried carrots was faster than that for raw carrots, the extraction rate for freeze-dried carrots
decreased soon.
HPLC chromatograms showed that pigments extracted were mainly a- and f}-carotene
(Figure 2), and the relative amount of f}-carotene was decreased by the freeze-drying process.

0.0010

0.0008
~ Dp
"0
I
-/l.- -,../% -
C)
0.0006 _",lJ.--'- 0-'- -0- 0.26mm
rn
~
".,.I::r-- . .c.~ ....o··-··· -1:.- 0.47mm
........
~
_- 0 - - -- 0- 1.12mm
3: 0.0004 - -<>-- S.OOmm

0.0002
A.. __ " ' - <>- --.-.~-.---~.-.-­
,. .... - Q-- ....... -yc

0.01 0.02 0.03 0.04 0.05

C02 [Nm3 ]

Figure 3. Effect of particle size on the cumulative extraction histories for freeze-dried carrots
( 313K, 14.7MPa, flow rate = 1.1xlO-7 m3/s at pump).

For the extraction from freeze-dried carrots, the influence of particle size on the extraction
histories is shown in Figure 3. In each run, only the carotenoids existing in about 1OlA-m of the
outer shell of each particle, which corresponds approximately to the size of a broken carrot cell,
could be extracted rapidly. The rest part was extracted very slowly. To extract carotenoids
completely the carrot cells should be broken for freeze-dried carrots.

CONCLUSION

The cell membranes of freeze-dried carrots prevented the extraction of carotenoids with SC-
C02. For complete extraction of carotenoid from freeze-dried carrots it is required disruption
of the carrot cells. For raw carrots cell disruption was not essential.

ACKNOWLEDGMENT

This work was supported by a Grant-in-Aid for Scientific Research (No. 04238106) from the
Ministry of Education, Science and Culture, Japan.

REFERENCES

1. Lorenzo, T.V., Schwartz, S.J. and Kilpartric, P.K., Supercritical fluid extraction of
carotenoids from Dunaliella algae, Proc. of 2nd Int. Symp. on Supercritical Fluids,
1991, 297-298.

2. Favati, F., King, J.W., Friedrich, J.P. and Eskins, K., Supercritical CO 2 extraction of
carotene and lutein from leaf protein concentrates., J.Food Sci., 1988, 53, 1532-1536.

3. Cygnarowicz, M.L., Maxwell, R.J. and Seider, W.O., Equilibrium solubilities of b-

carotene in supercritical carbon dioxide, Fluid Phase Equilibria, 1990, 59, 57-71.
 THE PERFORMANCE OF PREPARATIVE SUPERCRITICAL-FLUID
CHROMATOGRAPHY OF LIPIDS AND RELATED MATERIALS

SHUleHl YAMAMOTO AND YUJl SANO

Department of Chemical Engineering,
Yamaguchi University, Tokiwadai, Ube 755, JAPAN

ABSTRACT
The plate height(HETP) and the mobile phase velocity(u) relationships on ODS-silica
gel columns of various particle diameters (5 rv 30l1m) were measured. Both the liquid
(methanol-water) and the supercritical carbon dioxide were employed as the mobile phase.
Naphtalene and its derivatives as well as unsaturated fatty acids were used as test sam-
ples(tracers). The stationary phase diffusion coefficient, the axial dispersion coefficient
and the distribution coefficient were determined from the H ET P - u data for various
experimental conditions and discussed.

INTRODUCTION
Supercritical-fluid chromatography(SFC) has several advantages against conventionalliq-
uid chromatography(LC) for the preparative and process-scale separation of lipids and
related materials[l]. However, it is not easy to optimize the chromatographic separation
process as a large number of operating and column variables must be tuned to obtain the
maximum productivity[2]. In addition, SFC has two more important operating variables,
temperature T and pressure p, which are usually not important in LC. It is therefore
needed to investigate the fundamental separation mechanism in SFC over a wide range
of experimental conditions in detail.
In this paper, the plate height(HETP) and the mobile phase velocity( u) relationships
on ODS-silica gel columns of various particle diameters (dp = 5 rv 30l1m) were measured.
Both the liquid (methanol-water) and the supercritical carbon dioxide (SC-C0 2) were
employed as the mobile phase. Naphtalene and its derivatives as well as unsaturated
fatty acids were used as test samples(tracers). The stationary phase diffusion coefficient,
the axial dispersion coefficient and the distribution coefficient were determined for various
experimental conditions and discussed.

RESULTS AND DISCUSSION

H ET P - u curves obtained for SFC and LC are shown in Figs.1 and 2. These data were
analyzed on the basis of the following equation[3,4].
 839
HETP = 2(DL /u) + d~HKu/[30Ds(1 + HK)2] (1)
where DL=axial dispersion coefficient, Ds=stationary phase diffusion coefficient, K =
distribution coefficient and H = volumetric phase ratio. The results are shown in Ta-
ble 1. It is seen that the K Ds/ Dm values and the Pe( = udp/ Dd values for SFC are
comparable to those for LC. When the H ET P - u values of SFC and LC are re-plotted
as the reduced HETP, h(= HETP/dp) - the reduced velocity v=(udp/Dm), they were
expressed by a single curve(data not shown).
From these results it may be concluded that high separation efficiency at high flow
velocities can be expected for SFC because the molecular diffusion coefficient is high and
the viscosity is low(pressure drop is low).
These data were obtained at low sample loadings where the effect of sample volume
or concentration to the peak width is negligible. Further investigation is needed on the
elution behavior at high (or overloaded) sample loadings.
HETP . HETP -;--c::--;--=-;---;---;-:--:-::=-;------,
mobil. ph ... = 85% methanol + 15% water
( pm ( mob". pha .. : sc-co, 10MPa. 40 " C (I'm]
temperature = 30 0 C
column: ODS-BOT s
100 0 = 0.5 mg/mL naphtal.n.
column: ODS-BOT.
V = 0.5 mg/mL 2,6dimethylnaphtal.ne
100 o = 0.5 mg/mL naphtal.n. V
0
[:, = 0.2 mg/mL 2,7dim.thylnaphtalene V = 0.5 mg/mL 2,6dimethylnaphtalene
[:, = 0.2 mg/mL 2,7dimethylnaphtalene
° v6

dp = 2Ol'm
°
.°
50 50
v
»
tI ...8

°IN °V II v
o v g 0
0 0 dp = 5/lm
6
d p = 5pm ~ I!t 40- X

0 0
50 100 10 20
linear mobile phase velocity u [em/min] linear mobile phase velocity u [em/min]

Fig.1 H ET P vs. mobile phase velocity u(SFC). Fig.2 H ET P vs. mobile phase velocity u(LC).

Table 1 Values of the parameters obtained from the H ET P - u data

K Ds X 10 6 Dm X 106 KDs/Dm
solute Pe mobile
[-J [cm 2 /sJ [cm 2 /sJ [ - J [ - J phase
naphthalene 1.6 2.0 12 0.28 1.6 85%methyl-
2,6 dimethyl napht.halene 2.8 1.4 10 0.41 1.8 alcohol
2,7 dimethyl napht.halene 2.8 1.5 10 0.42 1.5 30°C
napht.halene 3.9 6.6 160 0.16 2.1 SC-C02
2,6 dimethyl naphthalene 5.3 5.5 137 0.21 2.2 10MPa
2,7 dimethyl napht.halene 5.3 6.3 137 0.24 1.8 40°C
Molecular diffUSIOn coeffiClent(D m ) values were calculated by the Wilke-Chang EquatIOn.
Column:ODS-80Ts (d p = 20p.m, 4.6mm diameter and 15 em length, Tosoh, JAPAN)

REFERENCES
1. Engelhardt,R., Gross, A., Mertens, R. and Petersen,R.,J.Chromatogr., 1989,477,169.
2. Yamamoto,S., Nomura"M. and Sano,Y.,J.Chromatogr., 1990, 512, 77.
3. Yamamoto, S., Nakanishi, K. and Matsuno,R., Ion Exchange Chromatography of Proteins,
Marcel Dekker, New York, 1988.
4. Yamamoto,S., Nomura,M. and Sano,Y.,J.Chromatogr., 1987,394,363.
 CONTINUOUS PROCESSING OF MILK FAT WITH SUPERCRITICAL CARBON DIOXIDE

S.S.H. RIZVI and A.A. BHASKAR

Institute of Food Science
Cornell University, Ithaca, NY 14853

INTRODUCTION

Milk fat fractionation is an area of current research which holds potential for development of novel
ingredients with varied functional properties based on their unique physical and chemical
characteristics. As a first approximation, the ability to manipulate solvating power of supercritical fluids
by varying density, offers a feasible approach to customized fractionation of milk fat by the
phenomena of selective distillation and extraction. Research on milk fat fractionation with supercritical
carbon dioxide (SC-C02 ), as with most other commercial uses of this technology, has thus far
concentrated mostly on batch systems. With fluids, such as milk fat, which can be continuously
pumped at high pressures, it is reasonable to anticipate that the processing time can be minimized
and the economics made more favorable (1).

PHASE EQUILIBRIA

Studies on SC-C02 phase equilibria are needed to provide a deeper understanding of the proceSSing
conditions so that separation can be influenced on a more selective basis. This information can then
be used to obtain new products (fractions) in a most cost-effective manner.

An extensive study of fluid-liquid equilibria for milk fat using a static method has been done
by Yu et al. (2). They correlated experimental fluid liquid equilibria data of milk fat with SC-C0 2 using
the Peng-Robinson (PR) equation of state (EOS) with Panagiotopoulos and Reid (PR) mixing rule.

In order to understand the separation of cholesterol from triglycerides, Yu et al. (2) used
selectivity as an index. They found that the highest selectivity of cholesterol occurred at pressures
between 8-12 MPa at 313.15 and 333.15 K. Yet, the highest selectivity value obtained is not sufficient
enough for clean removal of cholesterol. Use of in-line adsorbents has been shown to remove over
80% cholesterol (3).

FRACTIONATION

Using a continuous countercurrent SC-C02 system (Figure 1), we have fractionated milk fat into six
fractions, Sl-S5 and a flavor fraction (00) (Table 1). The short-chain (C4 - CS) and medium-chain
(C10 - C12) fatty acids increased from Sl - $5. The long-chain (C14 - C18) and the unsaturated fatty
acids decreased gradually from Sl - S5. The unsaturated to saturated fatty acid ratio decreased from
Sl - S5, with a range of 0.75 - 0.47 as compared to 0.57 for milk fat.

The triglycerides followed the same trend as the fatty acids. The low-melting (LMT) and
medium-meitinQ (MMT) triglyceride concentration increased from Sl - S5, while the high-melting
(HMT) triglyceride decreased from Sl - S5. The last fraction (S6) is a flavor fraction enriched with
lactones which are four times the concentration in milk fat. The cholesterol content decreased in the
raffinate (Sl) by 51% but increased in the fractions S3,S4,S5.

The solid fat content is an indication of hardness for fats and oils. The percentage of solid and
liquid fat at different temperatures calculated from thermograms of milk fat and two of its fractions are
shown in Figure 2. The solid fat content increased in the order of S5-S4-S3-milk fat-S2-S1 fractions.

ECONOMICS

Singh and Rizvi (4) conducted a detailed economic analysis for separation of milk fat into four
fractions. The calculated conversion cost for such a process was only 14¢ jkg. This cost is
considerably less when compared to other commercial uses of SC-C02 (Table 2).
 841
AMF

14

Figure 1. Schematic diagram of pilot-scale continuous SC-C02 processing system.

TABLE 1
Otemical composition of milk fat and its fractions with SC-C02•

Feed Raffmatel Extract Flavor

Milk fat SI S2 S3 S4 S5 S6
Temp(OC) 40 40 60 75 60 60 4
Press(psi) 3500 3500 3500 2500 1000 500 7
Yield(wt%) 100 21.0 15.0 48.0 4.0 11.0 1.0
FA(s)l:
C4-C8 8.55 1.22 5.97 10.14 10.67 12.42 N/A
CIO-C12 4.60 1.95 4.17 5.34 5.91 5.88 N/A
CI4-CI8 86.85 96.83 89.86 84.52 83.42 81.69 N/A
Unsat 31.28 41.57 34.37 28.19 27.23 26.32
Sat 55.oI 55.26 55.49 56.33 56.19 55.59
Unsat/Sat 0.57 0.75 : 0.62 0.50 0.48 0.47
TRG(S)2:
C24-C34 16.72 Traces i 10.18 18.82 24.30 26.39 N/A
C36-C40 50.85 17.07 i 49.94 56.19 53.62 54.22 N/A
.£~?':9.L ......... ~~:?~... .... §.f:2.Ll...?~·.~~....... ~:??...... ~~:Q§..... J2.·}2.. .... li!.A ....
ChoP 240.6 117.6 i 234.6 251.8 363.6 353.7 N/A
(mgf 100,1;) i
ChoP change %) - -51.1 -2.5 +4.7 +50.7 +47.0
Carotenoids4 314 768 N/A N/A N/A N/A
(IU/lOO 2)
Lactones5 61.74 11.90 N/A N/A N/A N/A 217.7
fug/g)

IFatty acids; 2TrigIycerides; 3Otolesterol compared to original milk fat; "Carotenoids

determined as JI..carotene; 5Lactone determined as summation of d-I 0, d-12, d-14 &
d-16Iactones; 6Not analyzed
842
TABLE 2
Comparitive economics ofvarious processes utilizing SC-C02

Reference System studied Capacity Convers. Capital Major costs

(tons/yr) cost cost item
~¢/kl!l ~M$l
Passey Batch system, 1000 70 6.1 Labor &
(1991) defatting of peanuts electricity
(oil as byproduct)

Leyers Semi-continuous, 11,000 41 22.0 Utilities

(1991) coffee decaffeination (steam)

Singh and Rizvi Continuous, 10,000 14 6.2 Utilities

(1993) fractionation of milk
fat into 4 streams

100
~ 80
..s
cf!
....
"CI
60
40
..... - SlMilk
---.-- fat
'0
\J'.) --0-- S5
20
0
220 240 260 280 300 320
Temperature,K

Figure 2. Variation in solid fat content with temperature of milk fat and its fractions.

CONCLUSIONS

The quality and not necessarily the quantity of fat is today's health issue. The very nature of milk fat,
Its unique triglyceride profile and flavor constituents coupled with SFE, can insure milk fat a long and
economically viable future as a raw material to the wand market. Additionally, SFE of milk fat can
supply the dairy industry Itself with modified fats needed to offer alternative product lines formulated
in response to the demands of the scrutinizing consumer. Further, it can be seen that the generally
held view of SFE being expensive is valid for batch/semi-continuous operations and not for
continuous processing of biomaterials.

REFERENCES

1. Rizvi, S.S.H. Supercritical fluid processing of milk fat. Newsletter #6. Northeast Dairy Foods
Research Center. Cornell University, Ithaca, NY. 1991, 11, 1-4.
2. Yu, Z.R., Rizvi, S.S.H., and ZoIlweg, J.A. Auid-liquid equilibria of anhydrous milk fat in
supercritlcal carbon dioxide. J. Supercritical Auids. 1992b, 123-129.
3. Rizvi, S. S. H., Um, S., Nikoopour, H., Singh, M., and Yu, Z., Supercritical fluid processing of
milk fat. In Engineering and Food, Vol 3, eds. W.E.L Spiess and H. Schubert, Elsevier Applied
Science Publishers, New York. 1989. p.145-160.
4. Singh, B. and Rizvi, S. S. H. Design and economic analysis for continuous countercurrent
processing of milk fat with SC-C02. J. Dairy Sci. (In review). 1993.00-00.
 SEPARATION OF ANTIOXIDATIVE FERRUGINOL FROM JAPANESE CEDAR
BARK BY SUPERCRITICAL CARBON DIOXIDE

HAJIME OHINATA, NAOKO INOUE, YOSHIO YONEI

Tobacco Science Research Laboratory
Japan Tobacco Inc.
6-2, Umegaoka, Midori-ku, Kanagawa 227, Japan

ABSTRACT

Extraction and separation of ferruginol as an antioxidant from

Japanese cedar bark using supercritical carbon dioxide (SCC02)
was studied. After being ground, the cedar bark was
extracted with SCC02 and the ferruginol was then fractionated
from the extracts using preparative-scale-supercritical fluid
chromatography (P-SFC) with a 50mm i.d. column (TMS,15~m).
Ferruginol, 0.4% in content in the cedar bark, was
concentrated to 16% by SCC02 extraction and to 66% by P-SFC.
50-70% of ferruginol in the cedar bark was recovered as
ferruginol fractions by a composite procedure of SCC02
extraction and P-SFC.

INTRODUCTION

Lipid peroxidation is known as one of the major factors

causing deterioration of food during storage and processing.
To inhibit lipid autoxidation of foods, natural antioxidants
are being used commercially, but there remain some problems in
their use as food additives. They are usually extracted with
organic solvents such as ethanol, acetone and hexane, but
these organic solvents are inflammable and present problems of
physiological safety.
SCC02 as a solvent is very suitable for use in the food
industries because C02 is an inert, odorless and tasteless gas
with low toxicity, and it can be removed completely from
extracts at room temperature.
Ferruginol(8,11,13-abietatriene-12-ol), a natural antioxidant,
was isolated and identified from Japanese cedar bark. Its
antioxidative activity was similar to that of a-tocopherol.
In order to develop an extraction and separation technique for
ferruginol from cedar bark, a composite procedure of SCC02
extraction and SFC was studied.
844
MATERIALS AND METHODS
Sample preparation
The cedar bark which had been stripped from a tree was air-
dried at room temperature and was then ground to particles of
0.4mm mean diameter by a single-disk refiner. The content of
ferruginol in it was 0.37% and its moisture content was 10%.

Apparatus and procedure

SCC02 extraction experiments of the ground cedar bark were
carried out under various pressures and temperatures in the
ranges of 9-30MPa and 15-60·C in the type of semi-batch
apparatus which is commonly used. Figure 1 shows a schematic
diagram of the P-SFC used in this study. It was designed and
assembled by the author's group and was equipped with a 1000mm
x 50mm i.d. stainless steel column packed with a tri-methyl-
silane modified silica(TMS) and heated electrically. The
pore size of the packing materials was 120A and the particle
size was 15~. The TMS was selected from among HPLC's based
on the preliminary experiments.

18

1.C02 cylinder, 2.C02 reservior, 3,5,19.Heat exchanger, 4.C02 pump, 6.Feed pump,
7. Feed , 8.Air-actuated 3-way ball valve, 9.2-way ball valve, 10.Column, 11.UV detector,
12.Pressure reducing regulator, 13.Separator, 14.Flow controlled valve, 15.Air-actuated 2-
way ball valve, 16.Pressure controlled valve, 17.Fraction collector, 18.Gas cleaner
Figure 1. A schematic diagram of preparative-scale
supercritical fluid chromatograph.

The sample was injected directly as an ethanol solution into

the C02 stream by a piston pump or a sample injector
consisting of stainless tubes and valves. At the column
outlet line, a small fraction of the C02 (1f/min) was split,
one portion of which went to a UV detector for a chromatogram.
Based on this chromatogram, four fractions were collected by
the operation of 3 pairs of air-actuated 2-way ball valves.

RESULTS AND DISCUSSION

The extract containing 16-19% of ferruginol from the cedar
bark was obtained by SCC02 extraction. The recovery of
ferruginol showed maxima with increasing pressure at each of
 845
the temperatures tested and these values were about 80% at any
temperature tested. In consideration of the SFC conditions
described below, the SCC02 extraction conditions were
determined at 15MPa and 40·C. Under these conditions, the
extract yield was 1 .8g/100g-material and the content of
ferruginol in the extract was 16%. The recovery of
ferruginol from the cedar bark was about 80%.

2.56.---..,...,.---------'30 a;'
~ ~
c M~
~ 10 ~
E W
< Obd~~iL~~__~~~~~ ~
o 30 60 90 120 ~
Retention time [min]

Figure 2. A typical chromatogram and cut point arrangements.

(injection 3.4g, UV at 280nm, linear velocity 0.3cm/s)

Figure 2 shows a typical chromatogram obtained from the

extract mentioned above and cut point arrangements. The
optimal SFC conditions, a pressure at the middle of the column
of 15MPa and the temperature of 60·C, had been previously
determined by an analytical-scale SFC on a 250mm x 4.6mm i.d.
column (TMS, 5~). However, because injected sample could not
be completely eluted at 15MPa, after the ferruginol fractions
had been collected, the pressure was increased to 27MPa for 10
min and held there for 2 hours as the injected sample could be
completely eluted at this pressure. Table 1 shows the
analytical results of the four fractions shown in figure 2
with a capillary gas chromatography.

Table 1
The analytical results of the fractions (injection 3.4g)
Fraction Recovery of Content of Recovery of
No. solute[%] ferruginol[%] ferruginol[%]
Fr . 1 4.8 11 . 2 3.0
Fr.2 20.0 65.6 71.6
Fr.3 5.1 48.9 13.7
Fr.4 70.1 3.1 11.7

In Fr.2 shown in Table 1, a fraction containing 66% of

ferruginol could be fractionated from the extract and the
recovery of ferruginol was then 72%. Its recovery from the
cedar bark was finally 57%. By using a longer column or
smaller packing materials, it is thought that it will be
possible to further concentrate ferruginol in the fractions.

ACKNOWLEDGMENT
We would like to thank the Japanese Research and Development
Association for High Separation System in Food Industry for
its support and approval of this presentation.
 MOISTURE ADSORPTION ISOTHERMS AND REACTION
CHARACTERISTICS OF IMMOBILIZED LIPASE IN
SUPERCRITICAL CARBON DIOXIDE

KOZO NAKAMURA~ TOSHIO SUGIYAMA AND HIDEKI SUDO

Department of Biological and Chemical Engineering,
Gunma University
Kiryu, Gunma 376, Japan
-I\- Department of Agricultural Chemistry. Faculty of Agriculture.
University of Tokyo. Yayoi. Bunkyo-ku. Tokyo 113. Japan

ABSTRACT

The water adsorption characteristics of several carrier particles as well as the

immobilized lipase were measured in supercritical carbon dioxide (SCC02 ) by
the flow method. The maximum sharp peak appeared near the critical pressure
when the adsorption of water was measured at sot at various pressures. The
data of the adsorption experiments could be correlated with a single curve for
each material when the concentration of water in seeo 2 was normalized by the
solubility. The lipase of Rhizopus delemar adsorbed less water than ePG 75
and Duolite A-568 with a specific surface area as large as 150 to 200m2 /g.

INTRODUCTION

The acidolysis reaction catalyzed by lipase is crucially influenced by the water

concentration in SeC02 ,an excellent reaction medium and it was reported that
the water content in the immobilized lipase could be directly related to the
activity of enzymatic ester synthesis in a limited amount of water. In this
study, the water adsorption characteristics of several carrier particles, as well
as the immobilized lipase, were measured to elucidate the effect of water on the
reactions of esterification and also to identity good carrier particles for the
preparation of immobilized lipase with high activity.

MATERIALS AND METHODS

(1) Materials: The adsorbents used were anionic exchange resin (Duolite A-
568), porous glasses (CPG-75, -1000, -3000), ceramics (SM-10, NGK
Insulators Ltd), lipase of Rhizopus delemar (Seikagaku Kogyo, 600 U/mg) and
immobilized lipases (Lipozyme IM20, IM60, 1M, Novo Industry A/S). The
physical properties of those adsorbents, except the enzyme, are shown in Table
1.
 847
(2) Measurement of water adsorption: The flow method was used to measure
the amount of water adsorbed in the materials. Moistened carbon dioxide was
fed into the column packed with an adsorbent, and the water concentration was
continuously measured at the exit of the pressure regulator by the moisture
meter (AQUAMATIC+, MEECO Inc.). The blank test was done with the empty
column to measure the amount of water adsorbed on the walls of the pipes and
the column, and this amount was substracted from the total amount adsorbed.
The dimensions of the column were 4.6 mm in inner diameter and 50 mm in
length.
(3) Immobilization of lipase: The lipase was immobilized on CPG by the
silane coupling method and on SM-I0 by the adsorption method with
glutaraldehyde cross-linking. The activity was assayed by the acidolysis of
triolein (5.6 mM) with stearic acid (17.6 mM) under the condition: sot, 29.4
MPa and total water concentration 28 mM.

RESULTS AND DISCUSSION

Water Adsorption
When the amount of water adsorbed on Lipozyme IM20 was measured in the
SCC02 at almost constant water concentration (0.027 to 0.036 mM), the
maximum adsorption peak (0.06 wt%) appeared near the critical pressure and
was followed by a sharp decline to zero adsorption. The water adsorption soon
returned to a level of 0.02 wt% and then gradually decreased with increase in
pressure. This gradual decrease is supposed to occur by the solubility of water
in SCC0 2 increasing with pressure. The water adsorption isotherms measured
at various temperatures and pressures were well correlated using the
concentration normalized by the solubility, as in Fig. 1, although the data
taken at the near-critical pressure deviated from those taken at higher
pressures.

Table 1
Physical properties of immobilizing materials

Mean pore
Pore volume Specific surface Mean diameter
Materials diameter
[cm3/g] area [m2/g] [pm]
[A]
Duolite A-568 180 0.80-0.85 200 460
CPO 75 74 0.47 152.7 146
CPO 1000 1010 0.79 21.8 122
CPO 3000 2869 1.06 8.9 120
SM-1O 365 0.56-0.80 77.7 180
848
8
3.0
...at
~ IM20,IM60

••
19.6MPa.SO"C
1M, A-568
2.5 Key Absorbents
6
•..

/0:
... Lipase

-
'"0 ~
SM-1O
~ 2.0

-
X 0 CPG75

61.7 37.2 64.4 '"0 II CPG1000

~ 4 14.7 86.7 96.1
X
1.5
0 CPG3000
93.9
19.6 96.7 114 142 ~
0 0
::c:'" Adsorbents: IM20, A-568 0
-0
~ P t ::t:'"
.J:J
.... Key [MPa] -0
["e) ~

•
0
'"
-0 9.8 40,50,70 ~
« til

«
0 .J:J
14.7,19.6 40,50,70
0
0 2 4 6 8 0.5 1.0 1.5 2.0
Nonnalized water concentration in Nonnalized water concentration in
SCC02 ClCo X 1()3 [-] SCC02 C/Co X 1()3 [-]
Fig. 1 Water adsorption on several carrier particles and immobilized lipase

Table 2
Specific activity of immobilized enzymes
Specific activity
Amount of
immobilized protein Protein base IMEbase
Immobilization [U/mg-protein] [U/g-IME]
Carriers
efficiency [-]
[mg/g- [mg/m2_
Hydrolysis Esterification Hydrolysis Esterification
support] support]
CPG-75 0.20 21.9 0.14 1.47 0.14 32.2 3.14
CPO-lOOO 0.48 52.8 2.42 0.54 0.28 28.3 14.8
CPO-3000 0.29 30.7 3.45 1.49 0.50 45.7 15.3
Duolite A-568 0.20 22.9 0.30 1.03 0.60 23.7 13.8
1M - - - - - 43.4 60.2

Activity of immobilized enzyme (Table 2)

The high esterification activity for 1M could mainly be ascribed to the lipase of
Mucor miehei, and the water adsorption per unit surface area might be also an
important factor for specific activity, although the water adsorption isotherms
were measured at too low concentrations of water to estimate the amount of
water adsorbed on each immobilized enzyme under the reaction condition.

CONCLUSIONS

(1) A sharp change of the isothermal water adsorption was observed near the
critical pressure.
(2) The water adsorption isotherms in SCC02 were correlated with a single
curve for each adsorbent using the normalized water concentration.
 ENZYME REACTION IN SUPERCRITICAL FLUID

YASUSHI ENDO, KENSHIRO FUJIMOTO

Faculty of Agriculture, Tohoku University, Sendai 981, JAPAN
KUNIO ARAI
Faculty of Engineering, Tohoku University, Sendai 980, JAPAN

ABSTRACT

Enzymatic interesterification between tricaprylin and methyl

oleate was carried out in supercritical carbon dioxide
(SC-C0 2 ) using an immobilized lipase. A temperature-
controlled reflux column was attached to the SC-C0 2 reaction
system to preferentially extract methyl caprylate, a
byproduct formed during enzymatic interesterification. The
selective removal of methyl caprylate by SC-C0 2 extraction
from the reaction mixture promoted the incorporation of oleic
acid to triglycerides. The enzymatic interesterification in
SC-C0 2 was affected by moisture and pH. The supplement of
water or phosphate buffer in proper quantity activated the
enzymatic interesterification. Chiral esters from secondary
alcohols and short-chain fatty acids could be also produced
by the SC-C0 2 system using a lipase.

INTRODUCTION

Recently the supercri tical carbon dioxide (SC-C0 2 ) has been

used for the extraction of oils and flavor compounds in
biological materials, and the medium of enzymatic reactions
instead of water or organic solvents. In fact, our group
applied SC-C0 2 system to the removal of cholesterol from
butter oil [1], the extraction of oil with low acid value
from rice bran [2], and the defatting of fish meats [3]. On
the other hand, the utilization of microbial lipases has been
attempted as catalysts for producing useful triglycerides and
esters. The purpose of this research was to produce the
useful triglycerides and various esters efficiently by
applying the SC-C0 2 system using immobilized lipases.
850
Enzymatic interesterification between tricaprylin and methyl
oleate
A schematic diagram of the SC-C0 2 bioreactor system using an
immobilized lipase is shown in Figure 1. A temperature-
controlled reflux column was set in the SC-C0 2 reaction
system to preferentially extract methyl caprylate produced
during the enzymatic interesterification. A mixture of
tricaprylin and methyl oleate (1:3, 30g) together with
Lipozyme IM20 (Mucor miehei, Novo Ind. Japan) as an
immobilized lipase at the concentration of 5 or 10%, were put
in a reactor, and then incubated at 40<>C with agitation in
SC-C0 2 under the pressure of 10MPa for 12hr. The incubation
was also attempted with CO 2 extraction at 5L/min.
Oleic acid levels in triglyceride after enzymatic inter-
esterif ication by a closed or a flowing SC-C0 2 system are
shown in Table 1. The level of oleic acid incorporated in
triglycerides with flowing SC-C0 2 was lower than that with a
batch type after interesterification although a SC-C0 2
extracted about 70% of methyl caprylate produced during the
enzymatic reaction.
The water was supplied to a reactor in the range of
0-1.2mL/hr to confirm whether the water activity in an
immobilized lipase was reduced by SC-C0 2 extraction. As shown
in Table 1, the supplement of water (O.6mL/hr) increased the
oleic acid level in triglycerides during lipase-catalyzed
interesterification to the same level as that obtained by a
batch type. This observation demonstrated that the removal of
moisture in an immobilized lipase by SC-C0 2 impaired the
enzyme activity. Phosphate buffer at pH6.98 and pH8.04 were
supplemented to a reactor. Enzymatic interesterif ication was
notably enhanced by the supplement of phosphate buffer
(pH6.98) compared with water (Table 1). These results show
that adjustment of pH besides supplement of water is
necessary for enzymatic interesterification of oils when
using a flowing SC-C0 2 system.

Enzymatical production of chiral esters from short-chain

fatty acids and secondary alcohols in supercritical carbon
dioxide
Short-chain fatty esters provide the specific flavors which
are common in tropical fruits. Their flavor characteristics
depend on the chiral structure. Our group attempted the
enzymatical production of flavoring chiral esters from
short-chain fatty acids (C6-10) and secondary alcohols (C6-8)
in SC-C0 2 using immobilized lipases. Lipozyme (Mucor miehei,
Novo Ind. Japan) and Lipase OF (Candida cylindracea, Meito
Sangyo) catalyzed the production of esters from short-chain
fatty acids and secondary alcohols in SC-C0 2 as well as in
n-hexane. Lipozyme showed higher activity for octanoic and
decanoic acid than hexanoic acid. Moreover, catalytic
activity of the lipase depended on the chain length of
secondary alcohols (2-octanol>2-heptanol>2-hexanol). Lipase
 851
OF catalyzed to produce esters from both 2(R)- and 2(S)-
hexanol, while Lipozyme showed activity only for 2(R)-hexanol
and catalyzed to produce 2(R)-hexyl hexanoate.

B:Buffer(Water) C:Compressor
CY:C02 cylinder FM:Flow meter
GC:Gas cooler H:Heater
P:Pump R:Reactor
RC 5 TABLE 1
RC:Reflux column S:Separator Interesterification in SC-CD2
using an immobilized lipase

Incorporated oleic acid

in triglycerides (%)

Batch 28
CY Flowing 17
Water
O.36.L/hr 24
O.60mL/hr 30
1.20mL/hr 18
Buffer
pH6.98 39
pH8.04 34
Figure 1. Supercritical carbon dioxide
(SC-C02) bioreactor system.

REFERENCES

1. Shishikura, A., Fujimoto, K., Kaneda, T., Arai, K. and

Saito, S., Agric. BioI. Chern., 1986, 50, 1209-1215.
2. Zho, W., Shishikura, A., Fujimoto, K., Arai, K. and
Saito, S., Agric. BioI. Chern., 1987, 51, 1773-1777.
3. Fujimoto, K., Endo, Y., Cho, S.-Y., Watabe, R., Suzuki, Y.,
Konno, M., Shoji, K., Arai, K. and Saito, S., ~ EQQd~,
1989, 54, 265-268.
 SUPERCRITICAL CARBON DIOXIDE AS PROCESSING MEDIUM FOR
ENZYMATIC INTERESTERIFICATION

B.MOSHAMMER, R.MARR, A.BILADT, F.FROSCHL, W.PREITSCHOPF

Institute of Thermal Process and Environmental Engineering,
Technical University Graz, Inffeldgasse 25, 8010 Graz, Austria

ABSTRACT

Supercritical carbon dioxide (SC-C02) is used as processing medium for the enzymatic
interesterification of D,L-Menthol with triacetin catalysed by the lipase from Candida cylindracea.
Studies on the reaction and solubility-measurements were performed in a batch-mode reactor. The
activity of the lipase was maintained to a high percentage after the exposure to SC-C02. The kinetic
parameters and the conversion-rate could be highly influenced by changing the pressure and the
water-content of the system. A new continous process based on enzymatic reactions in SC-C02 was
developed and constructed for further studies.

INTRODUCTION

The industrial application of biocatalysis has attracted attention with the knowledge that enzymatic
catalysis can take place in nearly anhydrous media (1,2). By using supercritical fluids, especially
supercritical carbon dioxide (SC-C02), as solvents for enzymatic reactions a new wave of research in
the field of biocatalysis in unconventional-media has emerged (3,4) resulting in new aspects for the
industrial biocatalysis as the demand for solvent-free as well as enantiomeric-pure products gets more
and more distinct. The special properties of the enzymes as biocatalysts and the well-known
advantages of SC-C02 as solvent combined in one process lead to a continous production/product-
recovery-process with a promising large application field in the food- and pharma-industry.
In this work SC-C02 and the lipase of Candida cylindracea are used for the study of a new
reaction-system in SC-C02. The model-system is the transesterification and racemic resolution of D,L-
Menthol with triacetin to L-Menthyl-acetate, Diacetin and D-Menthol.
The main objectives of the research-programme on enzymatic catalysis in SC-C02 at TVT/UT-TU
Graz consist of a study about the effects on the enzymatic reaction in SC-C02 in a batch-reactor-
system, the collecting of solubility-ctata and the comparison to organic solvents. The aim is the
processing in a continous-rnode with an integrated product-separation in a closed circuit based on the
enzyme-catalysis in SC-C02.

EXPERIMENTAL

The basic studies on the potential effects on the reaction and the SOlubility-measurements are realized
in a Batch-reactor-system, which consists of a high-pressure-pump, an enzyme-reactor or an
 853
equilibrium-cell with a sapphire-window and a sampling-unit.
The Batch-system is a part of our High-Pressure-Enzyme-Reactor-System (HP-ERS) having been
developed and constructed by our own (see figure 1). The special feature of the HP-ERS-plant, seen
in figure 1, is on the one hand the online-SFC-unit, as an appropiate online-technique is favourable in
the case of monitoring the reaction and the process. On the other hand a product/substrate-
fractionation in the down-stream-phase, based on the different solubilities of the substances at different
temperature/pressure-conditions, is provided in two steps.
The batch-reactions and the sOluibility-measurements are carried out at constant pressure and
temperature by magnetic-stirring. The production of L-Menthyl-acetate is followed by manual
withdrawals of aliquots from the reaction-system using a HPLC-valve with a sampling loop of 100 IJ.I
instead of the online-SFC-unit. The samples are then analyzed offline gaschromatograhically. The
online-SFC-technique is used and optimized by determining the solubility data of the substrates and
products in C02.

-------------------------------,
:
I
~-

MK: Mixing Chamber

ER: Enzyme-Reactor
SFC: Supercritical-F1uid-
Chromatograph
AI: Separator 1
A2: Separator 2 Carbondioxide
Substrate / Product

Figure 1. High-Pressure-Enzyme-Reactor-System at TVT/UT - TU Graz.

RESULTS AND DISCUSSION

The typical run of the studied racemic resolution of D,L-Menthol in SC-C02 shows in the beginning of
the reaction an increase of the converSion-rate, which reaches after a certain time a steady-state. The
reaction-rate, the time to the equilibrium and the achievable conversion were found to be dependent
of several external effects. The best studied parameters till now are the water content of the reaction-
medium and the pressure-effect.
It is a fact that water for enzymatic activity in SC-C02 is needed, but the amount of water differs
from system to system and by using organic solvents (5,6). The water-content of the studied reaction
influences distinctively the conversion- and reaction rate. The optimum value of water in the fluid is
854
situated between 0.06 % and 0.1 % (w/w) water in the fluid. Increasing the pressure at constant
temperature leads to a drastic reduction of the conversion rate whereas the reaction rate at different
pressure-steps is comparable. In contrast to high pressures up to 200 bar, small pressure changes
near the critical region of the C02 lead to significant changes in the conversion rate corresponding to
the properties of fluids in the critical region, where small changes in temperature and pressure result
in great changes of density and diffusion directly effecting the mass transfer and the solubility
parameters. As the study on the reaction demonstrates a substrate-inhibition of D,L-Menthol, it can be
assumed additionally that by increasing pressure an increasing substrate-solubility leads to a less
conversion-rate compared to reactions in the critical region.
The lipase of Candida cylindracea has been used for the first time in SC-C02 for enzymatic
conversions. The experiments show a good stability and activity of this lipase. After the exposure to
SC-C02 a deactivation, ranged about 20 %, is observed. A clear tendency can be deduced; the higher
the pressure and the time of exposure to SC-C02, the higher the deactivation of the lipase is found.
Additionally the deactivation is dependent of the number of ventilation-steps and the kind of ventilation.
The described reduced activity of the lipase after the C02-exposure is confirmed by fluorescence-
emmission-spectra of exposed enzymes. There is a shift in the height of the maximum of the
tryptophane-fluorescence which is due to a partial unfolding of the proteins, according to (7),
depending on the incubation time and used pressure.
The performed studies on the reaction result in a good and promising base for the continous
processing in the new developed High-pressure-enzyme-reactor-plant.

ACKNOWLEDGEMENTS

The authors wish to express their cordial thanks to Amano Enzyme Europe Limited (Milton Keynes,
U.K.) for their generous donation of lipase-samples. We thank JWolfgang carrying out the batch-
experiments and Univ.Doz.Dr.A.HermeUerfTU-Graz for the experimental work on the fluorescence-
emission-spectra. This work was supported by FWF (Vienna, Austria).

REFERENCES
1. Randolph, TW., Blanch, H.w., Clark, D.S., Biocatalysis in Supercritical Fluids. In Biocatalysis
for Industry, ed. J.S.Dordick, Plenum, London, 1991, pp 219.

2. Zaks, A. and Klibanov, A.M., Enzyme-catalyzed processes in organic solvents,

Proc.NatI.Acad.ScLUSA, 1985,82,3192-3196.

3. Aaltonen, 0., Rantakyla, M., Biocatalysis in supercritical C02, Chemtech, 1991,240-48.

4. Nakamura, K., Biochemical reactions in supercritical fluids, TIBTECH, 1990,8,288-92.

5. Dumont, T., Barth, D., Perrut, M., Enzymatic reaction kinetic comparison in organic solvent
and in supercritical C02, 2nd International Symposium on High Pressure Chemical
Engineering, 1990, Erlangen.

6. Marty, A., Chulalaksananukul, W., Condo ret, J.S. Wiliemot, A.M., Durand, G., Comparison
of Iipase-catallysed esterification in supercritical carbon dioxide and in n-Hexane,
Biotechnol.Letters, 1990, 12, 11-16.

7. Kasche, V., Schlothauer, R., Brunner, G., Enzyme denaturation in supercritical C02:
Stabilizing effect of S-S bonds during the depressurization step, Biotechnol.Letters, 1988,
18/8, 569-74.
 ENZYME INACTIVATION BY PRESSURIZED CARBON DIOXIDE

MURAT O. BALABAN, S. PEKYARDIMCI, C. S. CHEN, A. ARREOLA, M. R. MARSHALL

Food Science and Human Nutrition Dept. University of Florida
Gainesville, FL. 32611, USA.

ABSTRACT
Pectinesterase in orange juice, polyphenoloxidase in apple juice, and various
polyphenoloxidases in model systems were used in the determination of the
kinetics of inactivation of the enzymes by carbon dioxide treatment at
different pressures, temperatures, pH values, and times. Controls were used
for pressure, temperature and pH. Kinetics parameters are presented, effects
of the treatment on the enzyme molecules examined, and implications to juice
processing are discussed.

INTRODUCTI ON
Supercritical carbon dioxide (SC CO ) is used in food processing in extrac-
tion, separation / fractionation of lipids, essential oils, and pigments [1].
Another application of pressurized CO 2 is the inactivation of enzymes in foods
and beverages [2,3]. Polyphenoloxidase (PPO) causes browning in fruits,
vegetables and juices. Different sources of PPO such as mushroom, potato and
shrimp have different activities and different thermal resistances.
Pectinesterase (PE) causes cloud loss in orange and pineapple juices. These
enzymes can be thermally inactivated. However, this causes undesirable changes
in thei r fl avor and aroma. Therefore, non-thermal i nact i vat i on of these
enzymes is of great interest. Our purpose was to determi ne the effect of
pressurized CO 2 treatment conditions such as time, temperature, pH and
pressure on the inactivation of PE in Valencia orange juice, PPO in apple
juice, and PPO from various plant and animal sources in enzyme solutions in
pure water. Inactivation kinetics of these two enzymes were also determined.
Preliminary investigations of CO 2 treatment on secondary structure of PPO is
presented.

MATERIALS AND METHODS

Mushroom tyrosinase was purchased from Sigma (St. Louis, MO). Fresh Florida
spiny lobster (Panulirus argus) PPO was purified in our lab [4]. PPO from non-
sulfited brown shrimp (Panaeus aztecus), and from Russet potato tubers were
similarly prepared. Florida apples were crushed and pressed ~t 400 kPa at room
856
temperature, and SC CO 2 treatments were performed immediately after juice
extraction. Fresh squeezed Valencia orange juice (OJ) was frozen in cans by
the FMC Corp. (Lakeland, FL). Before treatments, OJ was thawed with tap water
while in the unopened can. PPO activity was determined by the pyrocatechin
assay [5], and that of PE by the titration method [6].
Mushroom PPO in pure water was treated at a) atmospheric pressure by
bubbling CO 2 through the solution at 33, 43 and 50°C, with T and pH controls;
b) 2 MPa ana 5.5 MPa by placing the solution in a pressure vessel immersed in
a constant temperature (T) bath (33, 43 and 50°C) with pH controls; c) 33.7
MPa at 35, 45 and 55° with T controls. Periodic samples were taken and PPO
activity determined. Spiny lobster PPO was treated at atmospheric pressure at
33°C, 38°C, or 43°C. T and pH controls were also run. Lobster, brown shrimp
and potato PPO solutions were treated at 5.8 MPa and 43°C. T controls were
run. Apple juice was treated at 33.7 MPa, 4SoC for 4 hrs, with T controls.
Orange juice was treated at ~ressures between 6.9 MPa and 34.4 MPa,
temperatures between 35°C and 60 C, for times from 15 to 180 min.

RESULTS AND DISCUSSION

The 1st order inactivation rate constants (k) for mushroom PPO treated with
CO 2 were calculated, and the energy of activation computed (Table 1).
Table 1.
Energy of activation for mushroom PPO inactivation at different CO 2 pressures.
E (KJ/mol)
Pressure T control pH control CO 2 treatment
Atm. pressure 93.9 194.7 50.0
117.6 46.8
2 MPa 26.1 44.8
5.5 MPa 19.1 37.4
33.7 MPa 110.3 43.1

Results show a drast i c decrease in the E values of CO treated sampl es

compared to T controls. There is some vari abil ity in the results since
different batches of the same commercial enzyme was used. There is also a
general trend of decreasing E values of the treated PPO as pressure increases.
Spiny lobster PPO treated at atm. CO 2 gave E=42.9 KJ/mol, while T
control gave E=26.2 KJ/mol. A single protein band at the same position was
observed (255 kD) for the control and CO 2-treated PPO on the nondenatured PAGE
gel, meaning that there was no alteration of the protein with CO2 , Secondary
structure changes may be responsible for the loss of activity. Ireatment at
5.81 MPa resulted in a very rapid inactivation of spiny lobster PPO : after
1 min there was 2% of original activity left. Spiny lobster PPO was more
sensitive to CO 2 than all other enzymes used in this study.
Brown shrimp PPO treated at 5.8 MPa and 43°C yielded k=0.79 min-l, and
T control gave k=0.0082 min-l. Potato PPO was the most resistant to CO2
treatment at 5.8 MPa and 43°C (k=0.061 min-l). T control gave k=0.0023 min-l.
Far UV circular dichroic spectra of shrimp, potato and lobster PPO showed a
change between untreated, and CO 2 treated samples (Table 2). Lobster and brown
 857
shrimp PPO showed the most noticeable changes in composition of Q-helix and
random coil. Only minor changes in secondary structure occurred in potato PPO,
which may explain its higher resistance to CO 2 treatment.
Table 2.
%Secondary Structure Estimates of Spiny Lobster, Brown Shrimp, and Potato PPO
from Far UV Circular Dichroic Spectra.
PPO Q-helix j3-sheet j3-turn randan coi 1
Lobster non-treated 24.4 26.2 21.4 29.9
CO 2-treated 19.7 25.9 15.2 39.3
Brown shrimp non-treated 20.1 22.3 15.2 42.4
CO 2-treated 29.6 18.9 18.2 33.3
Potato non-treated 14.8 34.6 28.4 22.2
CO 2-treated 17 .8 35.9 25.9 20.4

Treatment of apple juice at 33.7 MPa resulted in an increase in PPO

activity after an hour, followed by a drastic decrease by 4 hrs (8 % of
original activity). k was calculated as 0.0154 min-l. T control resulted in
k = 0.0007 min- l
Analysis of PE inactivation kinetics in orange juice for T control
samples resulted in E = 166.6 KJ/mol. Orange juice has 3 isoforms of PE. Z
values (increase of temperature necessary to decrease the decimal reduction
value D by 90%) for these were 6.5°C for PE-I and PE-III, and 11°C for PE-II
(7), or 6.5°C and 10.5°C (8). Our z value was 8.8°C, which is between reported
values of heat stable and heat labile forms of PE, and represents a "lump-sum"
value of z. Treatment at 31 MPa resulted in E = 97.4 KJ/mol.

REFERENCES
1. Rizvi, S. S. H.; Benado, A. L.; Zollweg, J. A.; Daniels, J. A.,
Supercritical fluid extraction: Fundamental principles and modeling
methods. Food Technol. 1986, 40, 55-65.
2. Taniguchi, M.; Kamihira, M.; Kobayashi, T., Effect of treatment with SC CO2
on enzymatic activity. Agric. Biol. Chern. 1987, 2, 593-594.
3. Haas, G. J.; Prescott, H. E.; Dudley, E.; Dik, R.; Hintl ian, C.; Keane,
L., Inactivation of microorganisms by CO 2 under pressure. J. Food
Safety. 1989, 9, 253-265.
4. Chen, J. S., Balaban, M. 0., Wei, C. I., Gleeson, R. A., and Marshall, M.
R., Effect of CO 2 on the inactivation of Florida Spiny Lobster
polyphenol oxidase. J. Sci. Food Agric. 1993, 61, 253-259.
5. Traverso-Rueda, S.; Singleton, V.L., Catecholase activity in grape juice
and its implications in winemaking. Am. J. Enol. Viticult. 1973, 24,
103-106.
6. Rouse, A. H.; Atkins, C. D., Pectinesterase and pectin in commercial
citrus juices as determined by methods used at the Citrus Experiment
Station. University of Fla. Agric. Exp. Stn. Bulletin. 1955, 570.
7. Versteeg, C,; Rombouts, F.M.; Pilnik., Purification and some
characteristics of two pectin esterase isoenzymes from orange. Lebensn.
Wiss. U. Tech. 1978, 11, 267-274.
8. Wicker, L.; Temelli, F., Heat inactivation of pectinesterase in orange
juice pulp. J. Food Sci. 1988, 53, 162-164.
 PURIFICATION OF ORGANIC ACIDS BY GAS ANTI-SOLVENT
CRYSTALLIZATION

AKIHIRO SHISHIKURA
Process Development and Engineering Center, Idemitsu Petrochemical Co., Ltd.,
1-1 Anesaki-kaigan, Ichihara, Chiba 299-01, Japan.

ABSTRACT

Citric acid (CA) has successfully been separated from fermentation broth by a novel and unique purification
process, which is characterized by organic solvent extraction and precipitation using compressed carbon dioxide
(C0 2) as an anti-solvent. Also, CA could be separated from other organic acids which are the by-products of CA
fermentation.

INTRODUCTION

Citric acid (CA) is an important compound used as an acidulant in food and beverages. CA
is generally produced by the fungal fermentation with mainly Aspergillus nigar, and is purified
by a firmly established process known as the method of calcium salt precipitation. However,
this process includes several batch treatments which require large amounts of chemical reagents.
More calcium sulfate is formed as an industrial waste than the weight of CA. Though these
negative factors have a significant influence on production costs, improvements in this process
have not yet been practically accomplished.
We previously proposed a novel and unique CA purification process which was
characterized by organic solvent extraction and precipitation using compressed carbon dioxide
(C0 2 ) as an anti-solvent (1). This paper addresses the outline of the new CA purification
process and fractionation of the organic acid mixture by gas anti-solvent crystallization.

MATERIALS AND METHODS

The broth of CA and the acetone solution of crude CA were prepared by the same
previously reported methods (1).
The experimental apparatus for gas anti-solvent crystallization was equipped with three
volumes (80, 250 and 2000 ml) of transparent glass-type level gauges (LG). The precipitation
conditions of impurities and organic acids were measured by batch treatments. CO2 was
dissolved in acetone solutions from the bottom of the LG. In the case of continuous treatment,
the acetone solution and CO2 were previously mixed in the mixing line, and the LG was used as
a settler.
 859
Analysis for CA and other organic acids was carried out by mean of HPLC. Sugar,
minerals and water contents were determined by the standard methods of the Japan Food
Additives Association(2).

RESULTS AND DISCUSSION

Precipitation of Impurities from Acetone Solution of Crude CA.

The CA concentrations in the solute extracted with acetone from the condensed fermentation
broth (44.9wt% CA, 20.1 wt% Water) reached 91 wt% by removing the majority of impurities
(mainly sugar). However, this value was not sufficient for the food additive grade of CA.
Therefore, in order to further remove the residual impurities, we applied gas anti-solvent
crystallization as the second step in our purification process.
Figure 1 shows the pressure vs. concentrations of CA (A) and impurities (B) in the
supernatant at 30 <C. CA started to deposit at 28 kg/cm 2 • On the other hand, impurities started
to deposit at 9 kg/cm 2 G. Most of the impurities were separated as precipitates from the solution
in the vicinity of 15 kg/cm 2 without any deposition of CA. These results suggest that the
effective separation of impurities from CA solution is possible at a pressure below 28 kg/cm 2 ,
because of the difference in the pressure needed to start the deposition of the CA and the
impurities. The purity of CA dissolved in the supernatants was increased to 99.6 wt%. The
results of continuous crystallization were similar to those of the batch treatment shown in Fig.
1.
25

Solute Cone. = 24.5 wt% (A) 4

~3
~20 !
! 19.8 wt%
0.8
1:!
5
~ '"E
gI). h
.,." 14.1 wt%
0.6
<I)

g :g.5"
::I 0;
<I)

.5 ~

«10 10.6 wt%

...u
0
0.4

'""
..§I
U '-
U"
0 5 0.2
0
U
u"
0

0 0 ...........
L.....----'-~--'---~L.....____'_~--'--- O
0 10 20 30 40 50 0 10 20 30 40 50 60
Pressure (kg/em 2G) Pressure (kg/em 2G)

Figure 1. Effect of the pressure on the concentrations of CA (A) and impurities (B)
in the supernatant at 30 <C.

Selective Crystallization of Organic Acids by Gas Anti-Solvent Crystallization.

To separate and crystallize CA from by-products [oxalic (OA) and malic (MA) acids], we
attempted gas anti-solvent crystallization.
Figure 2 shows the effect of the relative concentration on the pressure for the start of
deposition for several organic acids at 30<C. The starting pressure for deposition decreased
with increasing relative concentration in all acids. In the case of the crystallization of CA from
the model acetone solution [CA 18.7wt% (C/Csat=0.89), OA 1.0 (0.03) and MA 0.4 (0.02»).
the crystal grains ofCA could be obtained with 99.9% purity and 99.6% recovery at 54 kglcm 2
and 30<C. The particle size and the size distribution of the precipitated CA could be controlled
by the rate of pressure rise (the introdllcing rate of CO 2 ) and the standing time. The
860
crystallization of CA was finished within 1-3 min.

Novel CA Purification Process.

Figure 3 shows the flow diagram of a novel CA purification process. First, the fermentation
broth of CA is filtered to remove microorganisms (F) and is dried to adjust its water content to
about 10-20 wt% by multiple effect evaporation (C-l). CA is subsequently extracted with
acetone from the condensed broth (SE). The residual impurities are then removed as
precipitates from the acetone solution of CA using the anti-solvent effect of compressed CO2 at
25 kglcm 2 and 30~ (PC-I). The deposited impurities are readily separated from the
supernatant using a settler (SET-1). CA is crystallized from the acetone solution by anti-solvent
crystallization with CO 2 at 50 kg/cm 2 (PC-2). Crystal grains of pure CA can be rapidly
obtained by simple pressure regulation.

0.8

~O.6
S:2
uO.4

0.2

10 20 30 40 50 60
Pressure for Deposition Start (kg/em2 G)

Figure 2. Effect of the relative concentration (the saturation ratio) on the pressure
for the start of deposition for several organic acids at 30~.

SF

3
Figure 3. The flow diagram of novel CA purification process.
(SF) submerged fennentation tank, (F) filter, (C-l) multiple effect evaporator, (SE) solvent extractor, (PC-l,2)
pre-crystallizer, (SET-I ,2) settler, (DR) dryer, (I) water, (2) impurities, (3) residual impurities, (4) oxalic acid,
(5) acetone and (P) product.

REFERENCES

(1) A. Shishikura, H. Takahashi, S. Hirohama and K. Arai,J. Supercri. Fluids, 1992,5,

303-312.
(2) "Food Additives Official Hand Book", p4-62, Japan Food Additives Association, 1986.
 EFFECT OF HIGH PRESSURE ON ACTIVITY OF SOME OXIDIZING ENZYMES

YOSHIO AOYAMA, MASASHI ASAKA, and RITSUKO NAKANISHI

Biological Chemistry Division
Toyo Institute of Food Technology
23-2-4, Minami-hanayashiki, Kawanishi, Hyogo 666, JAPAN

ABSTRACT

The inactivation effects of high pressure treatment on some

oxidizing enzymes were investigated, compared with thermal
inactivation. Glucose oxidase and ascorbate oxidase were
inactivated irreversiblly above 300 MPa and the inactivation
was followed by first-order reaction. The activation volume of
inacti va tion of these enzymes was determined from the
pressure dependence of the rate constants. Tyrosinase and
super oxide dismutase were stable against high pressure, but
the former was thermolabile and the latter thermostable. Thus,
these enzymes are divided into three types: thermostable and
pressure-stable enzyme (superoxide dismutase), thermolabile
and pressure-stable enzyme (tyrosinase), thermolabile and
pressure-labile enzymes (glucose oxidase and ascorbate
oxidase). Thermal treatment is more effective than high
pressure treatment for irreversible enzyme inactivation.

INTRODUCTION

The residual enzyme activity in food processing seems to be

one of the major problems. The effects of high pressure treat-
ment on enzyme activity were different among enzymes 0)-(3).
Although many stUdies have been performed about hydrolizing
enzymes, there are only few studies about oxidizing enzymes
related to food quality. So we studied effects of high
pressure treatment on the activity of some oxidizing enzymes.
The inactivation of the enzymes was investigated kinetically
as compared with thermal inactivation.
862
MATERIALS AND METHODS

The enzymes were purchased from Wako Pure Chemical Industries,

Ltd. and used without purification: glucose oxidase from
Asperg:illus n:igar, ascorbate oxidase from cucumber, tyrosinase
from mushroom, superoxide dismutase from bovine erythrocyte.
High pressure treatments were performed with high pressure
test machine MFP-7000 (Mitsubishi Heavy Ifldusties, Ltd).
Enzyme activity was determined as follows (4}-(6}:glucose
oxidase, colorimetry with glucose and o-J.lan.ls.lJ.ln and peroxi-
dase; ascorbate oxidase, oxygen uptake by oxygen electrode;
tyrosinase, colorimetry with tyrosine; superoxide dismutase,
colorimetry with xanthine-xanthine oxidase/cytochrome c.

RESULTS AND DISCUSSION

The inactivation curves of glucose oxidase and ascorbate

oxidase at different pressures are shown in Fig.I. The plot
indicates that the inactivaton is first-order reaction.
As pressure increases, the inactivation velocity of these
enzymes increases. However, the extrapolation to 0 min does
not show 100lJll of enzyme activity, showing very short time
treatment of pressurization and depressurization could cause
inactivation of the enzyme. The activation volumes for
inactivation of these enzymes were determined from the
pressure dependence of the rate constants (Table 1).
Tyrosinase was stable against high pressure treatment(Fig. 2},
but unstable against heat treatment (data not shown).
Superoxide dismutase was stable against both pressure and heat
treatment. These studies indicate that enzymes are divided
into three types: 1} thermostable and pressure-stable,
2) thermolabile and pressure-stable, 3) thermolabile and
pressure-labile.

TABLE 1

Effects of pressure and temperature on the inactivation of

the oxidizing enzymes and activation volume (AV#) and
activation energy (Ea)

T,o/lo.a AV# Ea P T10 b

("C ) (ml /mol) (kcal /mol) (MPa)

Glucose oxidase 440 56 -50 56 80

Ascorbate oxidase 410 56 -38 46 145

• The pressure or temperature which caused loss of half

of the activity in 10 min.
b Pressure increase for the inactivation equivalent to
temperature increase of 10"C.
 863
100 100 (b)

g g

1O~~:::::
...">>. ...">
>.

;:
u t
" "
"..
..,:::J
in
-0- 300MPo
....- 350MPo
"..,
:::J
in
II
a: -.0.- 400MPo a: -.0.- 400MPo
-J.- 450MPo -A- 450MPo
5
-0- 500MPo - 0 - 500MPo

-.- 550MPo -.- 550MPo

o 5 10 15 20 25 30 o 5 10 15 20 25 30
Time (min) Time (min)

Figure 1. Inactivation of glucose oxidase(a) and ascorbate

oxidase(b) by high pressure treatment at 20 ~ .

150r-------------------------~

0 - 0 300 MPo (20 "C)

. - . 400 MPa ( , )
/;'--/;. 500 MPa .( • )
A--'" 600 MPa ( • )
O--a 700 MPa ( • )
...---... 600 MPa (35 "C)
0---0 700 MPa ( • )
.............. 600 MPa (40 "C)
0 ......·0 700 MPa ( '" )

Time (min)

Figure 2. Effects of high pressure treatment on tyrosinase.

REFERENCES

1. Asaka, M. and Hayashi, R., Activation of polyphenoloxi-

dase in pear fruits by high pressure treatment. Agric.
BioI. Chern. « 1991, 55, 2439-2440.
2. Hara, A., Nagahama, G., Ohbayashi A. and Hayashi, R.,
Effects of hjgh pressure on inactivation of enzymes and
microorganisms in non-pasteurised rice wine (namazake).
Nippon Nogeikagaku Kaishi, 1990, 64, 1025-1030.
3. Ko, W. C., Tanaka, M., Nagashima, Y., Taguchi, T. and
Amano, K., Effect of pressure treatment on actomyosin
ATPase from flying fish and sardine muscle. ~ Food Sci.
1991, 56, 338-340.
4. Nakamura, T., Purification and properties of ascorbate
oxidase from cucumber. ~ Biochem., 1968, 64, 189-196.
5. McCord, J. M. and Fridovich I, Superoxide dismutase ;
an enzymatic function for erythrocuprein (hemocuprein).
~ Diol. Chern., 1969, 244, 6049-6055.
6. Swoboda, B. E. and Massey, V., Purification and proper-
ties of the glucose oxidase from Aspergillus niger.
~ BioI. Chern. « 1965, 240, 2209-2215.
 APPLICATION OF STERILIZATION TECHNIQUE BY HYDROSTATIC HIGH
PRESSURE FOR GREEN TEA DRINK

TADAKAZU TAKEO, HITOSHI KINUGASA, *MASAMI ISHIHARA,

*KENJI FUKUMOTO & *TETSU SHINNO

ITO-EN Co. Ltd., Central Research Institute

*Nippon Steel Co., R & A Laboratory I

ABSTRACT

Green tea extracted with water contained a few viable micro-

organisms, which were steri I izad by pressurization at 400 MPa
amd room temperature for 30 min.
Thermoduric spores of Baci I Ius species added to green tea
infusion were not steri I ized at 400 MPa and room temperature,
but steril ized at 700 MPa and 80 C. degree. However, thermo-
duric spores survived at the pressure of 200-300 MPa were
effectivly inactivated during storing tea drink at room
temperature. It was revealed that thermoduric spores injured
by hydrostatic pressure were inactivated by tea tannin
(catechinJin tea infusion.
By hydrostatic pressure treatment at 700 MPa and 80 C. degree
for 30 min, the induction of off-flavor and the changes of
chemical compounds im green tea drink were effectivly de-
pressed. These results showed that the pressurization tech-
nique at 300 MPa and below 100 C. degree might be possible
to use as a new pasturization technique on making green tea
drink.
 865
INTRODUCTION
On the production o~ green tea drink, one o~ important
problems is to make tea drink having~resh ~Iavor. Usually,
tea drink are steri I ized by retort pasturization. By this
treatment, tea ~Iavor is deteriorated and strong retort smel I
is produced. Many trials were done to improve tea drink
~Iavor, but su~~icient results were not gotten yet.
Recently, it has been reported that microbes in ~ood and
beverage were inactivated by a high pressure treatrnewnt. On
th i s time, low mo Iecu I ar compounds were not e~~ected.

There~ore, it was considered that the high pressure treatment

might be a use~ul steri I ization technique to make ~Iavorous

~ood or beverage.

EXPER I MENTALS
Material: Green tea in~usion prepared with hot water was used.
Instrument: Per~ormance o~ hydrostatic pressure equipment
made by Nippon Stee I. Capac i ty o~ pressur i ng chamber; 1 000 mi.
Ma~. pressure; 700 MPa. Ma~. temp.; 80 C. degree. Pressur-
ing: Compressor type.
Microbial assay: Thermoduric spores belonging to Bci I Ius sp.
were separated ~rom vegetative cells by heating culture
medium and inoculated in tea in~usion. Microbes in tea
in~usion were assayed by counting the numbers o~ colony
growing on agar plate a~ter incuation.
Chemical analysis: Volatile compounds in tea drink were ana-
lyzed by GC and catechins and VC were determined by HPLC.

RESULTS
1) Microbes in tea in~usion made with water were steril ized
by the pressure treatment at 300 ~a under roon temp ..
2) Thermoduric spores belonging to Bcillus sp. showed high
resistibil ity ~or hydrostatic pressure. However, al I
these spores were i nact i vated per~ect I y under the
pressure o~ 700 MPa at 80 C. degree.
866
3) In tea in~usion, thermoduric spores were becoming sensi-
ble a~ter the pressure treatment. As shown in ~ig. 1,
spores survived a~ter the pressure treatment were gradual-
ly inactivated and sterilized a~ter several days in tea
in~usion. This e~~ect was raised stronger by the combi-
nation o~ pressure and heating as shown in ~ig. 2.
4) It was revealed that the inactivation e~~ect ~or spores
~ound in tea in~usion was due to the steri I ization e~~ecct

~or microbes o~ tea catechin dissolved in tea in~usion

(~ig.3) .

5) Aroma c~nds and chemicals in tea drink did not su~~er

any changes by the pressure treatment and green tea drink
treated by the high pressure kept ~resh ~Iavore.
S. subt i Ius s. cereus 10 6 200 MPa 300 MPa
"- 10 6
t: fiOC
OJ
..C>

'"=
II!!

10 3 . . 20min
.-..C>'" 10 3
0

.-..,
.....

II!! 10 0
SOl 2 3 weeksS 0 1 2 3
(Stored period after treatment) Fig 2 Survival numbers of
t-t(}-()-o S.cereus in tea infusion
S. coagu lance after pressure treatment

~~hOsPbate
S.licheniformis \ buffer
10 O:EGCG
o \ tea
SOl 234 SOl 2 3 4 5 weeks ~catecbin
Fig 1. Survival numbers of S. sp. spores 0
in tea infusion after pressure treatment s o 1 2 3 weeks
(300 MPa, 30 C degree. 20 min) Fig 3 The antimicrobial
S:original infusion effect of tea catecbin
O:pressure treated .infusion
t:tea infusion.O:tea infusion witbout catecbins
 STUDY OF IDGH PRESSURE EFFECf ON INACTIVATION OF
Bacillus stearothermophilus SPORES

I. HAYAKAWA, T. KANNO AND Y. FUnO

Department of Food Science and Technology, Faculty of Agriculture, Kyushu University
6-10-1 Higashi-ku, Fukuoka-shi 812, Japan

ABSTRACf
To establish a new sterilization method with minimal heating, the effect of elevated
pressure on bacteriostasis was studied using a heat-resistant spore of Bacillus
stearothermophilus. After exposure to 800 MPa for 60 min at 60 °e, the active spore count
decreased from 106 to 102 /m/. However, exposure to the same pressure at room
temperature did not cause any significant change. The synergistic effect of high
pressurization on the bacteriostatic action of sucrose fatty acid ester at low concentration
(<10 ppm) was pronounced with sucrose palmitic acid ester but not with sucrose stearic
acid ester. Repetitive pressurization was a more effective method for spore inactivation. Six
times of 5-min pressurization with 400 MPa at 70 °e decreased the spore count from 106 to
l02/ml, and with 600 MPa, complete inactivation was achieved.

INTRODUCTION
Spores are more heat resistant than living cells. D-values of B. stearothermophilus spores
range from 0.1 to 14 min at 121°e. At present, sucrose fatty acid esters (SEs) are being
used as additive in canned liquid coffee, because they inhibit the germination of bacterial
spore contaminant, which would otherwise germinate at the lesser heat processing
temperature employed in canning of liquid coffee and other foods. Some studies have
already been done on the effects of SEs on bacteriostasis and at high concentration (100-
1000 ppm) and on the germination of B. cereus, B. licheniformis, and B. circulars (Taki et
al., 1990). To assess the practical applications of high pressure technology to food
processing and to clarify some critical conditions on high pressure sterilization, pressure
resistant characteristic properties of the spores were studied using a strain of B.
stearothermophilus. We also deemed it necessary to further clarify the integrated effects of
pressurization and a method like repetitive pressurization without SEs, temperature, and
low level addition of SEs (<10 ppm) on the sterilization of B. stearothermophilus spores
under high pressure.

MATERIALS AND METHODS

Microorganism and culture method. Spores of B. stearothermophilus IFO 12550 were
used. The spores were collected by centrifugation after 5 days' cultivation at 55°e in a soil
868
medium, consisting of (g): bacto peptone (5), meat extract (3), bacto yeast extract (1),
MnS04 (0.1), bacto agar (25), soil extract (250), H20 (750), and pH adjusted to 7.2 by
NaOH.
Sucrose palmitic acid ester supplements. Sucrose fatty acid esters(P-1695 and S-1695
(Mitsubishi Kasei Food Co., Ltd., Tokyo) were used as bacteriostatic agents. The spore
suspension (1 ml) was treated with 0.1-10 ppm of SE, which was the concentration range that
could not affect spore germination under normal conditions.
Pressure application. The spores were placed in a germ-free tube before pressurization. A
pressure apparatus (P max :1000 MPa, Yamamoto Suiatu Kogyosho Co., Ltd., Osaka) was
used. Continuous and oscillatory pressurization were compared in this study. Survivors after
pressurization were counted after 7 days incubation at 55 ·C by plate culture in nutrient agar
media (Biken Chemical Co., Ltd., Tokyo). Heat treatment was used in tests of spore germina-
tion.

RESULTS
Temperature Dependence. Pressurization under 400-1000MPa for 20 and 60 min at 20·C
gave no particular effect on spore germination. On the other hand, upon the pressurization
under 800 MPa at 60 ·C, the spore survivors slightly decreased from 106 to 104 -5/ml
after 20 min and about 102/ml after 60 min (Fig. 1).
6
~10A~~--~~----~--~~~

§ 20 min & 60 min at 20·C

8 4
'flO
.....~ At 60 ·C, for 20 min

~
00 At 60 ·C, for 60 min . /
102~__~~__~~__~~__~~__~
o 200 400 600 1000
Pressure (MPa)
Fig. 1. Effect or pressure, temperature and treatment time on survivors of
B. stearothermophilus spores.

20 min, 600 MPa, at 60 ·C

• P-1695
,. ......... , S-1695

10.1 10° 10 1 10 2
Concentration (ppm)
Fig. 2. Effect of concentrations of SEs on survivors of B. stearothermophilus
spores at 600 MPa pressurization.
 869
Effect of Sucrose Fatty Acid Ester. Pressurization for 60 min with increasing concentration
of S-1695 enhanced survivability, but at such longer exposure P-1695 seemed unable to cause
protective action against pressure sterilization(Fig. 2).
Repetitive Pressurization. Repetitive short pressurization(5 min internal 70°C) was
found very effective to inactivate the spore. The effect of continuous pressurization for 60
min under 600 MPa at 60°C was similar to that of six times of oscillation under 400 MPa
at 70°C. Six times of the repetitive pressurization under 600 MPa at 70°C resulted in
complete inactivation (Fig. 3).

1rr ~----~----------------~
C'
e
';:t
106

0
= lOS
= 10 4
~
~
10 3
....
0
~

t 10 2
O:400MPa
=
"-l
.:600MPa
10 1
at 70°C
5minxD
0
0 2 4 6 8
Numbers of pressurization (D)
Fig. 3. Effect of pressurization times on survivors of B. stearothermophilus spores.

DISCUSSION

This research was undertaken in order to establish a new inactivation method for the spore by
high pressure (400-1000 MPa). The B. stearothermophilus spore is one of the strongest end-
spore of aerobic bacteria. If this spore can be sterilized by pressurization, food deterioration
by microorganisms can be eliminated. Strong heat-resisting spores also showed strong
pressure durability, as the spores were never affected by 1000 MPa for 60 min under room
temperature. Even if, upon exposure for 60 min under 60°C, some survivors (102/ml) were
still found. Repetitive pressurization was more effective as the complete inactivation was
achieved by 6 times of 5-min repetitive pressurization with 600 MPa at 70°C. In this case,
every spore was completely destroyed. This phenomenon may be due to the following two
reasons: (1) the adiabatic explosion velocity of spore cell wall and that of high pressure water
upon the release of highly pressurized, and (2) water permeability into the spore cell wall
were promoted by the rise in temperature (70°C). Of course, some physical changes could be
~iven onto the spore cell wall by this significant temperature difference (from 20°C to 70
C).

ACKNOWLEDGMENT

The authors acknowledge Mr. K. Yoshiyama for his guidance on the spore observation by a
scanning electron microscope. This work was supported by research grants from Mitsubishi-
Kasei Foods Corporation, and Nestle Science Promotion Committee is acknowledged with
pleasure.

REFERENCES

Taki, Y., Awao. T., Asada, K. aDd Tsujimoto., S., Sterilization of Bacillus sp. Spores by
Hydrostatic Pressure. In "pressure-Processed Food Research and Development". (Ed.), R.
Hayashi. Sanei Publishing. Co. Ltd., Kyoto, 1990, pp. 143-155.
 DEVELOPMENT OF FIBER-CONTAINING NON-EXPANDED PRODUCTS
BY EXTRUSION COOKING PROCESS

WENCHANG CHIANG MING-JYH SHIH ANN-RONG YIAO JYH-HUAR WENG

Graduate Institute of Food Science and Technology
National Taiwan University, Taipei, Taiwan, R.O.C.

ABSTRACT
Wet soybean residue was directly used as a raw material to mix
with starchy and proteinaceous substances to make cereal flakes
and wet-type texturized products with a twin-screw extruder. The
cereal flakes could be made from the pellets which were baked in a
band infrared oven. While using a two-stage cooling system linked
up with the extruder, a good fibrous texturized product containing
dietary fiber could be manufactured.

INTRODUCTION
Wet soybean residue, a by-product of soymilk and tofu processing
plant, contains high dietary fiber content (1). It is a good
raw material for producing low calorie, high dietary fiber and
cholesterol-free products. Our laboratory has used the mixture of
rice flour and dried soybean residue to make directly expanded
extrudates (2). Wet soybean residue contains about 80% moisture
and is very hard to dry. Although if it is mixed with rice flour
to reduce initial moisture content, the dehydration of the mixture
becomes easier (3), we tried to use wet soybean residue directly
to manufacture non-expanded extrudates in this study.

MATERIALS AND METHODS

Soybean residue was washed, soaked, ground and then filtrated to
get wet soybean residue. Corn flour, rice flour, high gluten
wheat flour, soy protein isolate (SPI), sucrose and sodium chloride
were also used in this experiment.
A co-rotating and intermeshing twin-screw extruder ( Clextral
BC-45 , France) was used with a barrel length of 100 em. The screw
profile was adopted with two reverse screws positioned in the
middle part. The barrel was divided into four sections. The
first barrel was cooled using tap water, and the barrel
 871
temperatures of other sections were controlled using an induction
heater or tap water. There were two holes with a diameter of 0.4
em on the die plate for cereal flakes processing. A special
cooling system including two cooling stages and one central
cooling pipe was linked up with the extruder.
Cereal flakes was milled to pass through a 60 mesh screen,
and used to determine water solubility index ( WSI ) by the method
of Anderson et al. (4). The supernatant, water-soluble fraction,
was also used to analyze water soluble carbohydrate ( WSC ) by the
phenol-sulfuric acid method (5), using the standard curve of
glucose to convert the data. Hedonic scale was adopted to evaluate
the acceptability of cereal flakes from 20 panels.
The hardness and chewiness of wet-type texturized products
were calculated by the texture profile analysis using a rheometer
( Fudoh NRM-2020J, Japan). A piano wire ( adaptor No.30 ) of the
rheometer was used to cut the sample from the parallel and vertical
directions to measure tear strength separately. Dietary fiber
content of the product was analyzed by the enzymatic method (6).

RESULTS AND DISCUSSION

using the obtained extrusion conditions in the previous paper (7),
the mixture of the basic recipe ( corn flour : wet soybean residue
: rice flour = 81.8 : 10 : 8.2 , on the total weight basis) and
the additive of 5% SPI, 5% sucrose or 5% sodium chloride was
extruded under feed moisture 33.5%, feed rate 26.4 kg/h, screw
speed 150 rpm, and the 2nd, 3rd and 4th barrel temperatures being
90, 130 and 40·C respectively to prepare non-expanded extrudates
( pellets). After cooling, the pellet was flaked to the thickness
of about 0.15 em and pre-dried at 40·C to reduce the moisture
content around 25%. The pre-dried flake was then roasted at 140·C
for about 5 minutes in the infrared oven. Both WSI and WSC
values were lower than 20%, and all the cereal flakes had
acceptable appearance and mouth-feel quality according to the
sensory evaluation.
The raw material ( wet soybean residue : SPI : high gluten
wheat flour = 60 : 30 : 10 , on the total weight basis ) was
extruded under feed moisture 60%, feed rate 25 kg/h, screw speed
160 rpm, the barrel temperatures in the 2nd, 3rd and 4th sections
being 145, 185 and 165·C respectively. As shown in Table 1,
cooling water rate affected the rheological property and tear
strength of the texturized product a lot. If controlling the 1st
stage, 2nd stage and central pipe of cooling water rate at 1, 2
and 1 L/min respectively, a good fibrous texturized product could
be obtained. Because the higher the Fv/Fp value is, the better
the fibrous formability is. The soluble and insoluble dietary
fiber content in the raw material was 3.6% and 12.3% respectively,
but it was 8.5% and 13.8% in the texturized product. The increase
of dietary fiber might be resulted from the polymer degradation and
starch-protein interaction during extrusion.

ACKNOWLEDGMENT
Grateful acknowledgment is made for the financial support from the
National Science Council of the Republic of China.
872
TABLE 1
Effect of cooling water rate on the rheological property
and tear strength of the wet-type texturized product.
cooling water rate Rheological property Tear strength
at each part, L/min
Hardness Chewiness Fv Fv/Fp
1st 2nd Central (kg) (kg) (kg/cnt)
0 2 0 0.97 0.75 1.08 1.10
1 1.10 0.93 1.45 0.90
2 1.01 0.89 1.40 0.95
1 1 0 0.82 0.53 1.20 0.77
1 0.92 0.73 1.35 0.70
2 0.97 0.79 1.46 1.10
1 2 0 0.75 0.60 1.40 0.70
1 1.22 1.04 1.21 0.43
2 0.99 0.80 1.16 1.10
2 2 0 0.95 0.53 1.14 0.74
1 0.89 0.66 1.07 1.02
2 0.98 0.78 1.12 1.23

REFERENCES
1. Chiang, W., Shih, M. J. and Chen, S., Effect of soaking and
grinding conditions on quality of soymilk and physical
properties of bean residue. ~ Chinese Agric. Chern. Soc.,
1988, 26, 165-72.
2. Wu, T. P. and chiang, W., Optimal extrusion conditions for
manuacturing rice expanded products containing dietary fiber.
~ Chinese Agric. Chern. Soc., 1991, 29, 17-25.
3. Shih, M, J., Perng, M. Y. and Chiang, W., Drying of the
mixture of soybean residue and rice flour. Food Sci., 1991,
18, 286-94.
4. Anderson, R. A., Conway, H. F., Pfeifer, U. F. and Griffin, E.
L., Gelatinization of corn grits by roll- and extrusion-cooking.
Cereal Sci. Today, 1969, 14, 4-7, 11-12. .
5. Dubois, M., Gilles, K. A., Hamilton, J. K., Rebers, P. A. and
Smith, F., Colorimeteric method for determination of sugar and
related substances. Anal. Chern., 1956, 28, 350-6.
6. Prosky, L., Asp, N., Schweizer, T. F., Devries, J. W. and Furda,
I., Determination of insoluble, soluble and total dietary fiber
in foods and food products: Interlaboratory study. J. Assoc.
Off. Anal. Chern., 1988, 71,1017-23.
7. Shih, M. J. and Chiang, W., Physicochemical changes of corn-
based semi-product during extrusion processing. J. Chinese
Agric. Chern. Soc., 1992, 30, 454-61.
 THE HEAT DENATURATION PROCESS OF SOYBEAN PROTEIN
BY TWIN SCREW EXTRUDER

YOSHINOBU AKIYAMA
Research Laboratories,
Kyodo Milk Industry Co.,Ltd.
180 Hirai,Hinode-cho,Nishitama-gun,Tokyo 190-01 JAPAN

ABSTRACT
Finely-fibered texturized tofu were prepared from tofu by use of twin screw
extruder. The relations between the extrusion process parameters and the
intensity of protein heat denaturations are mentioned. The screw
revolution and material feed rate were controlled as variable process
parameters in order to make extrudates have the different residence time.
SDS-PAGE was applied to the protein denaturation analysis. Four major
bands 7Sa,7S,B,11S acidic and 11S basic were detected. They were decreased
in proportion to residence time in the extruder. Cooking value(Cv) was
defined as the index of denaturations, which indicates the integral of net
heated time and temperature. The correlation coefficients between 7S a,
7S~,11S acidic, lIS basic and corresponding Cv were -0.98,-0.84,-0.82 and
-0.81 respectively. It was suggested that Cv indicate the the degree of
heat denaturation of proteins so accurate that it could be the rational
process parameter in manufacturing of protein texturization by extruders.

INTRODUCTION
Textured vegetable products have been manufactured as meat substitutes,
and it is expected that their consumption will increase. However,
it seems that the qualities of textured vegetable products such as taste,
flabour and texture are not enough to satisfy the consumers.
For successful texturization of vegetable proteins, the control of
heat denaturaions in the extruder barrel and of flowing conditions of
extrudate in the die section must be very important. The relations
between residence time and protein denaturations were measured.
The numerical model derived from the theory of heat sterilization was
applied to estimate the amount of heat denaturation.
874
MATERIALS AND METHODS

Tofu were dehydrated partially and extruded by the twin screw extruder.
Screw revolutions and material feed rates were controlled in order to make
extrudates have the different residence time. To evaluate the heat
denaturations. SDS-PAGE was applied to texturized tofu. Fine structures
were examined by SEM.

RESULTS AND DISCUSSION

Table 1 shows the relations between the residence time and changes in
four major subunits of soyprotein. All subunits decreased in proportion
to residence time and 7Sa was most sensitive to heat.

TABLE 1
Residence time and changes in subunits
SPL No. Residence 7S(X' 7S{1 11 S 11 S
time sec acidic bas i c

1 o 19. 2 1O. 5 11. 8 15. 7

2 145 13. 6 8. 0 8. 2 11. 2
3 180 13. 5 9. 3 9. 6 11. 3
4 240 1O. 5 7. 9 7. 6 11. 3
5 280 1O. 3 7. 9 7. 2 9. 7
6 300 10. 9 9. 0 8. 5 1O. 2
7 375 6. 8 4. 9 7. 7 1O. 5
8 400 6. 0 3. 2 5. 0 11. 4
9 450 5. 3 5. 3 6. 8 8. 0

If the protein heat denaturation mechanism is regarded same as those

of heat sterilization. it must be of the first order in reaction.
In Eq-l. () 2 is maximum temp. of material. () 1 is the lower
critical temp. of denaturation and Z is constant. Amount of denaturation is
given by S. Final amount of denaturation during extrusion cooking will be
given by Eq-2. where tl is the beginning time of denaturation. and t2 is
terminating time. Cooking value Cv stands for. therefore. the degree of
denaturation during extrusion.
Fig.l shows the relations between Cv and subunit 7S a.
The correlation coefficient was -0.98. It is supposed that about 15 %
of nat i ve 7S a wou I d be degraded or assoc i ated. Changes in denaturated
subunit are in inverse proportion to Cv lineally.
 875
82- 81
S = 10 Z ( 1)

82- 81
v= Jt 2 10
Z dt (2)

t1

20 y = 18.1-0.10Cv
r = -0.98
~

:-
E-

•
.....:J
0
a:l
;::::l
.....:J
10
0
IZl
Z
~
E-
o
c=
0-

0 50 100 150 Cv
Figure 1. The relations between Cv and 7Sa.

CONCLUSIONS

It is concluded that subunit solubilities are closely connected with

residence time and changes in amount of denaturated protein correlate
with Cv. Cv stands for the degree of protein denaturation so accurate
that it could be the rational index as extrusion parameter.
 THE DEVELOPMENT AND CONTROL OF COLOUR IN EXTRUSION COOKED
FOODS SIMULATED USING A MODEL REACTION CELL

LISA BATES, JENNIFER M. AMES AND DOUGLAS B. MACDOUGALL

Department of Food Science and Technology,
University of Reading, Whiteknights, P.O. Box 226, Reading RG6 2AP, UK

ABSTRACT

A reaction cell was designed to allow the rapid, small scale examination of colour development
which occurred in an extrusion cooker. It simulated the temperature, time, rapid heating and
pressure experienced during extrusion cooking, but not the mechanical shear. A starch-
glucose-lysine mixture was extrusion cooked and heated in the reaction cell, and the colour of
the samples was analysed. pH had the most significant effect on colour development;
temperature, moisture and residence time were also significant. Reaction cell results were used
to accurately predict L * and a* values of samples extruded at a predetermined pH.

INTRODUCTION

Extrusion cooking of expanded snacks is an intermediate moisture, high temperature and high
mechanical shear operation. These conditions favour the Maillard reaction which reduces lysine
availability and leads to colour development. Many previous studies have shown the effect on
lysine availability, while only a few have examined the effect on colour [1]. This study uses a
reaction cell to model colour development in an extruder over a range of conditions.

MATERIALS AND METHODS

A mixture of wheat starch type A, D(+)- glucose and L-Iysine monohydrochloride (96:3:1,
m:m:m) was cooked in either a twin screw corotating extruder or in the reaction cell.
Tristimulus CIE L *a*b* values of ground and sieved samples were obtained and hue angle
 877
values, e, tan- 1(b*/a*) were calculated. Statistical evaluation of the results was done by
analysis of variance and regression analysis using SAS software. Further experimental details
are available on request from the authors.

RESULTS AND DISCUSSION

Preliminary experiments were performed at 3 levels of pH, 3 levels of moisture, 3 temperatures

and 2 target mean residence times. The most significant parameter affecting colour was pH,
followed by temperature, pH I temperature interaction and moisture. Residence time had the
least significant effect. Statistical analysis ofthe reaction cell data is summarised in Table 1.

TABLE 1
Preliminary Reaction Cell Study

VARIABLES & L* a* b* e
DEGREES OF R-Square: 0.93 R-Square: 0.95 R-Square: 0.98 R-Square: 0.96
FREEDOM F Value pa F Value pa F Value pa F Value pa
pH 2 107.49 *** 191.71 *** 419.93 *** 241.68 ***
Moisture 2 10.27 *** 0.13 NS 10.00 *** 7.41 **
Temperature 2 39.37 *** 37.13 *** 48.29 *** 31.40 ***
Time 1 3.60 NS 5.24 * 9.69 ** 14.38 ***
pH*Moisture 4 0.38 NS 1.62 NS 5.13 ** l.10 NS
pH*Temp 4 4.39 ** 12.67 *** 5.44 ** 11.52 ***
pH*Time 2 1.85 NS 1.31 NS 2.73 NS 2.97 NS
Moist*Temp 4 1.35 NS 1.19 NS 2.95 * 1.99 NS
Moist*Time 2 1.22 NS 1.80 NS 0.67 NS 0.80 NS
Temp*Time 2 0.05 NS 1.03 NS 0.07 NS 1.96 NS

pa (significance level) = *** at p<0.001, ** at p<0.01, * at p<0.05, NS not significant.

A detailed study of the most significant factor, pH, was then carried out, holding the other
factors constant at 15% moisture, 140°C and 32s. Overall, more colour developed in the
extruded samples than in the reaction cell samples. This may be due to differences in the mode
of heat transfer, or in material fluidity, between the 2 systems. However, the increase in colour
with increasing pH was very similar. Plots ofL *, a* (see Figure 1) and hue angle against pH
gave very similar gradients for extrusion and reaction cell samples. In contrast, the gradients
were not similar for the b* plots, possibly due to the different stages of colour development in
the 2 systems. Using linear regression of the L * and a * responses from the reaction cell, L * and
a* were predicted for the mixture extrusion cooked at pH 6.3. The predicted values were 63.9
878
and 5.6 respectively. The ranges of extrusion experimental data were 60.5 -70.2 (L*) and 5.2-
7.3 (a*). Thus, the predicted responses at pH 6.3 were within the experimental ranges.

90

80 I

o
CD
J I
: 70
>

60
.. Reaction Cell Samples
• Extrusion Samples

50
3 4 5 6 7
pH

• Reaction Cell Samples

• Extrusion Samples

~ 5
: 4
> I
«
as
3
2
1
I
o ,.
-1
3 4 5 6 7
pH

Figure 1. Changes in (a) L * and (b) a * values with increasing pH.

REFERENCE

1. Berset, C., Colour. In Extrusion Cooking, eds. C. Mercier, P. Linko and J.M. Harper, Am.
Assoc. Cereal Chern., St Paul, MN, USA, 1989, pp. 371-85.
APPLICATION OF EXTRUSION-COOKING FOR FEED PREMIXES STABILIZATION

LESZEK MO~CICKI & STANISLAW MATYKA

Department of Food Process Engineering. Agricultural University
Doswiadczalna 44. 20-236 Lublin. Poland

ABSTRACT
The premixes (microelements. vitamins and chemotherapeu-
tics) are the most expensive components and play important role
in feed production. Usually added in small amounts.thay can be
incorrectly mixed with other components because of its bulk
density and suppleness to segregation. Moreover. the water ab-
sorbtion and chemical activation taking place during the stora-
ge of microelements can reduce the quality of the feed.
In the paper. the results of premixes stabilization are
reported using the extrusion-cooking technique.
The extrusion-cooking process conditions and its influence
on physical and chemical products properties are discussed to
show how the stabilization of feed premixes can be achived with
minimum energy costs and maximum quality effects.

MATERIALS AND METHODS

The typical mineral premix. applied usually to the produc-

tion of fodder mixture for farm animals, was used in this rese-
arch. It includes microelements as: Fe, Cu. Zn and Mn. 50 kg
samples of mineral premixmixed with a vegetable carrier [corn,I
variant and the horse bean (vicia faba),II variant] in the dif-
ferent proportion~ of the mineral to the carrier (30/70; 35/65;
40/60; 45/55; 50/50 and 55/45) were prepared for the experi-
ments. Each sample of the mixture was exposed to extrusion-
cooking to gain homogenic extrudates. The extrusion-cooking was
conducted on the single screw extrusion-cooker of TS-45 polish
producSion. with the application of thermal treatment from 140
to 170 C. the 0 6 mm die, the screw of the 1:3 c.r. and with a
screw speed of 1.66 * l/s.
The prepared mixtures were extruded without the addition
of water. The obtained product was blended to the size of the
bruised grain. the size of the most commonfraction of the typi-
cal components included in typical feed mixture.
Homogeneity of extrudates was estimated on the basis of
the analysis of the Zn. Cu. and Co content in the probe taken
in the initial, middle, and final phases of extruding of the 50
880
the analysis of the Zn. Cu. and Co content in the probe taken
in the initial. middle. and final phases of extruding of the 50
kg propor tion of the previously mentioned mixtures. The ele-
ments mentioned above were present in the mixture in widely
varying proportions. What is more. in order to estimate the ho-
mogeneity and resistance to selfsegregation. the corn-
extrudates was crumbled then fractio nized by means of sieves
into three fractions of the 1 mm; 0.5 mm and 0.1 mm grain.
The Zn. Cu and Co content was analyzed in each fraction
by the technique of ASA Atomic Absorbtion. The obtained numeri--
cal data were described statistically. counting average values.
standard deviations and coefficients of variability. Student's
Test was applied to compare average values.

RESULTS
Table 1 presents the results of the Zn. Cu and Co designa-
tion.both in corn and horse bean extrudates.The data in the
table indicate that the average content of the elements in the
probes. taken in the initial. middle. and final phases of ex-
trusion-cooking. did not differ significantly from the prescri-
bed content (p>0.05) in the entire range of the examined rela-
tions of the mineral premixed with the vegetable carrier.
It needs to be emphasized that the variability of the con-
tent of the elements appeared to be extremely small and. in the
case of Zn.it did not exceed 2%. The variability of the concen-
tration of Cu did not exceed 7% in any case. and Co did not
exceed 8%.
Bearing in mind that these results were also influenced by
the degree of blending before the extrusion-cooking. the obtai-
ned coefficients of variability V testifies to the high repeat-
ability of the composition of the product. This was also con-
firmed by the stability of the composition of the corn extruda-
te after blending and fractionating with the sieves.
In the 50/50; 45/55 and 40/60 mixtures. the relationship
of the mineral fraction to the organic one. the coefficients of
variability V did not exceed 2% for Zn. 5% for Cu. and 5.5% for
Co.
According to the above discussion we may conclude that
during extrusion-cooking. the mineral fraction dilutes regular-
ly in the whole mass of the organic carrier. stabilizing the
product in this way. The blended extrudates. which. even after
the self-segregation of the grain during the transport. pre-
vents its primary mineral content.
The examination of the physical features of the stabilized
feed premixes showed that their bulk density approximated the
value of the typical cereal feed components. Moreover. the ca-
pability of absorbtion of water or the tractability of dissolu-
tion decreased in relation to the qualities of non-extruded
materials. Accordingly. the process of extrusion-cooking. even
on that scale. appeared to be promising.
 881
TABLE 1
2n, Cu, Co concetration in corn and horse bean extrudates [g/kg]

Mineral/ Components Extrudates Composi tion

Carrier Corn Horse bean
M SO V% M SO V%

2n 39. 100 0.245 0.62 38.310 0.802 2.09 39.420

30170 Cu 3.950 0.090 2.27 3.950 0.082 2.04 4.050
Co 0.153 0.008 4.98 O. 155 0.004 2.58 0.160

2n 45.360 0.681 1. 50 45.270 0.328 0.72 45.990

35/65 Cu 4.670 0.999 2.14 4.630 0.311 6.71 4. 720
Co 0.177 0.011 6.29 0.184 0.006 3.48 O. 187

2n 51. 630 1. 118 2.16 52.410 0.336 0.64 52.560

40/60 Cu 5.320 0.100 1.88 5.200 0.229 4.40 5.400
Co 0.204 0.014 6.99 0.205 0.005 2.44 0.214

2n 59.030 0;980 1. 66 58.430 0.608 1. 04 59. 130

45/55 Cu 6.030 6.l00 1. 65 5.920 0.142 2.39 6.070
Co 0.232 0.005 2.21 0.229 0.018 7.96 0.240

2n 65.490 0.499 0.76 65. 180 0.724 1.11 65.700

50/55 Cu 6.650 O. 136 2.04 6.710 0.121 1. 79 6.750
Co 0.249 0.015 6.18 0.261 0.003 1. 23 0.267

2n 71. 170 0.879 1. 22 72.210 0.626 0.87 72.270

55/45 Cu 7.220 0.204 2.83 7.310 0.373 5.10 7.420
Co 0.280 0.013 4.72 0.293 0.006 2.07 0.294

TABLE 2
2n, Cu, Co concentration in blended corn extrudates [g/kg]

Mineral/ Components Fraction M SO V Composition

Carrier A B C (%)

2n 64.80 63.30 62.30 63.47 1.2297 1. 94 65.700

50/50 Cu 6.43 6.65 6.48 6.52 0.1153 1.77 6.750
Co 0.23 0.25 0.23 0.24 0.0115 4.88 0.227

2n 57.60 58.30 58.20 58.04 0.3763 0.65 59.130

45/55 Cu 5.98 6.10 6.00 6.03 0.0643 1. 07 6.070
Co 0.22 0.23 0.23 0.22 0.0061 2.69 0.240

2n 49.50 50.50 51. 20 50.43 0.8934 1. 77 52.560

40/60 Cu 5.59 5.32 5.07 5.32 0.2600 4.88 5.400
Co 0.18 0.20 0.19 0.19 0.0104 5.43 0.214
 EXTRUSION OF Castanea sativa

F. SILVA, A. CHOUPINA, 1. M.N. SOUSA and M. L. BEIRXO da COSTA

*S.A.C.T.A.I Instituto Superior de Agronomia, Technical University of Lisbon
Tapada da Ajuda, 1399 Lisboa Codex. Portugal

ABSTRACf

In Portugal, chestnut (Castanea sativa) is mainly conswned unprocessed. Only a small amount
of the exceeding production is used in the candy industry. The objective of this study was to
find an alternative industrial utilisation for chestnuts, by extrusion-cooking.
The properties of chestnut starch were studied by the Brabender amylograph. The
amylose/amylopectin ratio and the starch granules dimensions were obtained by light
microscopy. The performance of chestnut flour in single screw extrusion cooking was studied
setting a response surface method (R.S.M.). All the responses linearly decreased with initial
moisture content and die temperature being independent of the screw speed.. The best
conditions were low initial moisture and low die temperature for crispness, but the expansion
ratio at this conditions was very poor. To overcome this, a new R.S.M. was set considering
the incorporation of com grits to enhance expansion.
The extruded starches showed an interesting preservation of the integrity of the granules
with a huge swollen effect in the case of chestnut and some disrupted granules mainly among
the com starch.

INTRODUCflON

Chestnut is used in sweets and candy specialities. In Portugal its industrial use is still quite
insufficient. Extrusion-cooking is known to be a process of high versatility for starchy
products. Chestnut presents a high percentage of starch, consequently in this paper it was tried
to assess the possibility of extruding chestnut flour. There are very few references on chestnut
starch and one of our concerns was its characterisation.
Extrusion conditions for chestnut flour and this mixed with com grits at different ratios
were optimised using objective textural evaluation as dependent variables.

MATERIALS AND METHODS

Dry chestnuts and com grits are commercially available. Both were milled and passed by a
0.71 mm sieve.
 883
Extru.date Production and Evaluation
A single screw laboratorial 20 DN Brabender extruder was used, operating at 3: 1 compression
ratio, using a die of 3 mm diameter. Feed hopper speed was 40 r.p.m. and temperatures in the
barrel respectively T 1=T2= 423 K.
A response surface methodology (RS.M.) with a central composite rotatable
experimental design was used to optimise extrusion of chestnut and corn grits mixtures using
as independent variables moisture (Xl), ranging from 12 to 20%, die temperature (X2), from
423 to 473 K, and percentage of chestnut flour in the mixture of chestnut plus corn grits (X3),
from 20 to 100%. The extrusion conditions of chestnut flour was optimised using the same
methodology. The independent variables were moisture (Xl), ranging from 10 to 18%, die
temperature, T3 (X2), from 423 to 473 K, and screw speed (X3) from 120 to 220 r.p.m.. The
dependent variables used in both cases were expansion ratio measured with a micrometer
gauge, rupture strength evaluated in a TAX-T2 Texturometer (Stable Microsystems) with a
Wamer-Bratzler cell, and consistency of the slurries of extrudates flour determined in a
Brabender Amylograph.

Starch Characterisation
Starch in chestnut flour was quantified by the Lintner polarimetric method (AOAC,
1990). Starch size and shape granules were studied by light microscopy. Gelatinization
properties were studied by Amylograph method using convenient dilution (8.5% starch in the
suspension). Amylose/amylopectin ratio was determined by G.P.C. using Sepharose CL-2B
eluted with NaN3 (0.02% v/v) at a flow rate ofO.5ml.min-l.

RESULTS AND DISCUSSION

Chestnut starch granules when studied by light microscopy showed an average size
similar to corn starch. Nevertheless, after extrusion chestnut starch show a much higher
swelling degree keeping simultaneously its integrity. It can also be seen that inside the
granule, components are melted. Corn starch granules are much more susceptible to rupture.
When subjected to amylograph evaluation, chestnut starch began gelatinization earlier
than corn starch (532 K and 343 K, respectively) showing higher values for maximum
consistency (545 U.B.) than corn starch (230 U.B.). After extrusion, chestnut flour showed
that some extruded starch had not yet been gelatinised what was evidenced by a new increase
in amylograph consistency.
The results from the first set of experiments, with the chestnut/corn grits extrudates
were fitted to a second order polynomial equation by a multiple regression analysis. When the
response is expansion ratio (E.R), the respective equation is:

Y = 1.683 - 0.1698 Xl - 0.293 X2 - 0.301 X3 + 0.188 Xl X3 (1)

with a R2 = 0.89. The equation for the rupture strength (RS.) as the dependent variable is:

Y = 4.274 - 2.471 Xl - 1.227 X3 2 (2)

with R2= 0.97. From the above equations one can see that both E.R. and R.S. of the
extrudates decreased with moisture (Xl), die temperature (X2) and chestnut flour
incorporation (X3) or moisture and chestnut incorporation, respectively. Nevertheless, the
experiments with 100% of chestnut gave extrudates with some expansion (1.67) and good
organoleptic characteristics. Our next step was the optimisation of the extrusion of the
chestnut flour de per si using this time as a third variable the screw speed. The respective
equation for E.R. was:
884
Y = 1.683 - 0.173 Xl - 0.166 X2 (3)

with R2= 0.97 and for R.S. :

Y = 2.166 - 0.849 Xl - 0.891 X2 (4)

with R2= 0.79. From these equations it can be seen that the screw speed did not affect
the E.R. or R.S. and the die temperature had a negative influence on both characteristics.
These results are in agreement with previously reported work (1,2) on corn grits extrusion.
The higher E.R. value (1,9) was obtained for the extrudate produced with 10% moisture
and 175° C die temperature, this extrudate also showed the higher rupture strength (5.18 N)
that seems to be related with good crispness.
The consistency of chestnut / corn grits extrudates flour at 20°C, used as another
response for the RSM, is an index of the degree of gelatinization undergone during extrusion
and is expressed by equation:

Y = 292.94 + 102.71 Xl - 191.80 X3 + 58.91 X1 2 + 68.63 X3 2 (5)

with an R2 = 0.90
When extruded alone, the starch of the chestnut flour shows a consistency at 20°C as
expressed by:

Y = 559.48 + 74.64X1 + 127.79 X2

with an R2 = 0.79. From the above equations one can see that the consistency of the
extrudates slurries is linearly dependent on moisture in both cases. For the mixtures it also
depends on chestnut· flour incorporation but on the chestnut extrudates consistency is
dependent on die temperature.
From GPC results the presence of amylose and amylopectin was determined to be 43
and 57% respectively in chestnut starch. It was also possible to assess that the molecular mass
of the chestnut amylopectin is higher than the one of the corn starch (7413 kD and 3890 kD).
The difference in the amylose molecular mass is not so big being higher for corn starch (346
kD,490kD).

CONCLUSIONS

It is possible to obtain expanded extrudates from chestnut flour or chestnut flour / corn grits
mixtures with acceptable textural properties specially at low moisture and low die
temperatures. Chestnut starch has a higher amylose / amylopectin ratio than corn starch,
having simultaneously an interesting sweet taste and preservation of the integrity of the starch
granules in the extruded products.

REFERENCES

1. Launay, B. and Lisch, 1M., Twin screw extrusion cocking of starches. J.. Food Eng., 1983
2: 259.

2. Chinnaswamy, Y. and Hanna M.A.,. Optimum extrusion cooking conditions for maximum
expansion of corn starch. J.. Food Sci., 1988,53: 834.
 EFFECTS OF PREPARATION PROCEDURES AND PACKAGING MATERIALS ON
QUALITY RETENTION OF CUT CHINESE CABBAGE

EERO HURME, RAIJA AHVENAINEN, EllA SKYTI'A, MARGARETA HAGG",

MIRJAMI MATTILAt, RAIlA-LIISA HEINIO AND ANNA-MAllA SJOBERG
Technical Research Centre of Finland (VTI), Food Research Laboratory, P.O.Box 203,
FIN-02151 Espoo, Finland, "Agricultural Research Centre, FIN-31600 Jokioinen, Finland,
tDepartment of Food Technology, P.O.Box 27, FIN-00014 Helsinki University, Finland

ABSTRACT

The effects of different washing procedures, packaging materials and long-term storage before
processing on the quality retention of packed cut chinese cabbage were studied. A shelf life of at least
7 days at 5 °C could be achieved by two separate washings, the first of which could contain chlorine,
and by highly permeable packaging films (02 perm. 5200 - 5800 cm3/m 2 24 h 101.3 kPa, 23°C, RH
o %). Long-term storage before processing was detrimental to the quality of the packed samples.

INTRODUCTION

The shelf life of prepared cabbages can possibly be prolonged by using highly permeable films for
preventing anaerobic respiration and by using chemical treatments for controlling microbial
contamination. Several senescence reactions occur in cabbages after harvesting and therefore long-term
storage before processing can affect the quality maintenance of prepared cabbages. The aim of this
study was to determine the effects of different processing procedures on the shelf-life of packed cut
chinese cabbage. In addition, the effects of long-term storage before processing were studied.

MATERIALS AND METHODS

All external parts of the cabbages (Kingdom) were first removed. The cabbages were then cut into 5
mm thick strips, washed (1 min/wash) in 5 °C water and/or in 0.01 % chlorine solution (as sodium
hypochlorite), spin dried and packed. Five packaging types and two packaging sizes were used (1.0 kg
cabbage/package, vol. 3000-3500 cm3, film thickness 40 ,urn; PE-film impregnated with ceramic
powder, PE-EVA-film and OPP-film, 1.5 kg cabbage/package, vol. 5000 cm 3; PE-paper-lid + PP-tray
and PET-carton package). The O2 permeabilities of the films were 5800, 5200, 1200 and 1500 cm3/m2
24 h 101.3 kPa, 23°C, RH 0 %, respectively. The packages were stored at 5 °C for up to 7 days.
886
The percentages of CO 2 and O2 in the package headspace were determined with Servomex PA 404
and Servomex 507A analysers, respectively. All analyses were carried out from four replicate packages.
Microbiological determinations using pour plate techniques were carried out on duplicate sample
packs. Aerobic plate counts (APC) were obtained after incubation at 30°C for 3 days (Plate Count
Agar, Difco), the number of lactic acid bacteria (LAB) after incubation at 30°C in 5 % CO2 for 3 days
(MRS agar, Oxoid), and coliforms after 24 h at 37°C (Violet Red Bile Agar, Difco).
Sensory evaluations of the samples were performed by trained panel members. First, the odour and
appearance of the samples were evaluated by 5 panelists, followed by flavour evaluations by 10
panelists (10-20 minutes after the first evaluation). All analyses were carried out on duplicate samples.
Samples were scored on a scale of 0 to 5 (5 = excellent, 2 = unacceptable for retailing and 0 = unfit
for consumption). The minimum level of acceptability for human consumption was 1.5.
~-carotene and ascorbic acid concentrations were analysed using HPLC (1,2). The residual free
chlorine concentrations (samples CIW in Table 1) were analysed spectrofotometrically 16 hours after
packaging (3). The detection level for free chlorine was 0.1 mg/kg.
The respiratory activity of the cut cabbages was measured by the flow-through system. Air was fed
into an airtight chamber via one port and allowed to exit from another port. The 02 and CO 2
concentrations of the outcoming gas were determined. Measurements were terminated when both the
O2 and the CO 2 concentrations had stabilized (four replicates).

RESULTS AND DISCUSSION

Washing in two separate washing waters improved the maintenance of sensory quality of the cut
chinese cabbage (Table 1). The quality was maintained better when chlorine was used in the first
washing water. The washing methods did not, however, significantly affect the microbial load, the CO 2
production and 02 consumption, the ascorbic acid and ~-carotene concentrations (12 mg/lOO g fresh
weight and 29,ug/l00 g fresh weight, respectively, after 1 days of storage and 10 mg/lOO g fresh
weight and 17,ug/lOO g fresh weight, respectively, after 7 days of storage) or the respiration rates (13
ml CO2/kg h) of the samples. No residual free chlorine was found in chlorine-treated samples.
Packaging in highly permeable films (PE and PE-EVA) improved the quality retention (Table 2). The
rapid quality loss of the cut cabbage packed in OPP-film and PE-paper-lid + PP-tray was obviously due
to the high CO 2 concentration in the packages. Drying was the main factor reducing shelf life of the
cabbages packed in PET-carton packages.
Long-term storage before processing was detrimental to the quality keeping of the cut cabbage
(washed once in water, packed in OPP-film). For example, the odours of the samples made from
cabbages stored for 3 months were scored clearly better than those of samples made from cabbages
stored for 5 months: after 7 days of storage the scores were 2.0 and 1.5 (season 1991) and 1.1 and 0.7
(season 1992), respectively. Obviously, this was due to respiration, enzymatic and microbiological
activity of the cabbages. The ~-carotene content of the uncut fresh cabbages decreased between 3 and
5 months of storage from 19 to 12 (season 1991) and from 12 to 9 (season 1992) ,ug/100 g fresh
weight.

REFERENCES

1. Ollilainen, V.M., Heinonen, M., Linkola, E., Varo, P. and Koivistoinen, P., Carotenoids and retinoids in
Finnish foods:meat and meat products. !:. Food Compo Anal., 1988, 1, 178-88.
2. Speek, A, Schrijver, J. and Schreus, W.H.P., Fluorometric determination of total vitamin C and total
isovitarnin C in foodstuffs and beverages by high-performance liquid chromatography with precolumn
derivatization. !:. &!!£:. Food Chern., 1984, 2, 352-5.
3. Test Method VTT-4426-91, VTT Food Research Laboratory, Espoo, Finland, 1991.
 887
TABLE 1
The effects of washing method and storage time on the quality retention of cut chinese cabbage
packed in OPP-film. W = washed in water, WW = washed twice in water, CIW = washed first in
chlorinated water (0.01 % free chlorine) and then in pure water. The amount of washing water was
3 l/kg product. The cabbages were stored for 3 months before processing.

Quality Storage timeLwashing method

attributes 1 day 3 days 7 days
W WW CIW W WW CIW W WW CIW

CO 2 (%) 4 ±l 4 ±l 4 ±o 10 ±o 9 ±O 9 ±l 15 ±l 15 ±l 15 ±l
O 2 (%) 14 ±l 15 ±O 15 ±O 4 ±O 7 ±l 6 ±l 1 ±O o ±O o ±O

APC 5.7 nd nd nd nd nd 6.8 6.8 6.8

lAB 3.8 nd nd nd nd nd 5.9 5.9 5.7
Coliforms 1.5 nd nd nd nd nd 4.4 4.7 4.7

Odour 4.2 ±O.2 4.1 ±O.l 4.1 ±O.l 3.1 ±O.l 3.4 ±O.l 3.3 ±O.2 1.1 ±O.l 1.9 ±O.4 2.3 ±O.4
Appear. 4.4 ±O.5 4.5 ±O.4 4.4 ±O.4 4.0 ±O.4 4.0 ±O.4 4.0 ±O.4 3.4 ±O.9 3.4 ±O.9 3.4 ±O.9
Flavour 4.0 ±O.5 3.9 ±O.5 4.1 ±O.3 3.3 ±O.5 3.4 ±O.7 3.2 ±O.4 na na 2.8 ±O.7

TABLE 2
The effects of packaging material and storage time on the quality retention of cut chinese cabbage.
Cut cabbages were washed in water before packaging (W in Table 1). OPP = OPP-film, EVA =
PE-EVA-film, PE = PE-film impregnated with ceramic powder, Pap = PE-paper-lid + PP-tray, Car
= PET-carton package (not sealed). The cabbages were stored for 3 months before processing.
Quality Storage timeLQackaging material
attributes 1 day 3 days 7 days
OPP EVA PE Pap Car OPP EVA PE Pap Car OPP EVA PE Pap Car

CO 2 (%) 7 6 5 5 2 13 9 8 11 1 20 10 9 20 1
±1 ±l ±2 ±l ±l ±l ±l ±1 ±2 ±O ±O ±2 ±1 ±l ±O

O2 (%) 9 10 12 12 20 1 1 1 2 20 0 0 0 0 20
±l ±2 ±3 ±1 ±O ±l ±O ±O ±l ±O ±O ±O ±o ±o ±o

Odour 4.0 4.1 4.1 4.2 3.9 2.8 3.1 3.3 3.6 3.4 1.8 2.6 2.8 1.7 2.4
±O.6 ±O.2 ±O.4 ±O.4 ±O.4 ±O.2 ±O.4 ±O.3 ±O.6 ±O.3 ±O.5 ±O.3 ±O.2 ±O.8 ±O.3

Appear. 4.3 4.2 4.2 4.2 4.3 3.7 3.6 3.6 3.7 3.8 3.0 3.3 3.3 3.3 1.1
±O.3 ±O.3 ±O.3 ±O.2 ±O.3 ±O.5 ±O.5 ±O.4 ±O.5 ±O.3 ±O.4 :to.3 ±O.3 :to.3 ±O.3

Flavour 3.8 4.0 4.0 3.8 3.9 3.4 3.4 3.8 3.6 3.6 na 3.3 2.9 na na
±O.4 :to.4 :to.3 :to.4 :to.3 ±O.7 ±O.7 ±O.3 ±O.4 ±O.5 ±O.8 ±1.0

nd = not determined, na = quality not acceptable for human consumption.

The O 2 and CO2 concentrations (%) and the results of the sensory evaluations are presented as
mean values ± standard deviations. The microbial counts are presented as mean values (log cfu/g).
 EFFECTS OF PREPARATION PROCEDURES AND PACKAGING MATERIALS ON
QUALITY RETENTION OF GRATED CARROTS

EERO HURME, RAIJA AHVENAINEN, EIJA SKYTIA, MIRJAMI MAlTILA·,

MARGARETA H.AGGt, RAIJA-UISA HEINIO AND ANNA-MAI1A SJOBERG
Technical Research Centre of Finland (VTf), Food Research Laboratory, P.O.Box 203,
FIN-02151 Espoo, Finland, ·Department of Food Technology, P.O.Box 27, FIN-00014 Helsinki
University, Finland, tAgricultural Research Centre, FIN-31600 Jokioinen, Finland

ABSTRACT

The effects of different washing procedures, packaging materials and long-term storage before
processing on the quality retention of packed grated carrots were studied. A shelf life of at least 8 days
at 5 °c could be achieved by washing the peeled carrots in 0.01 % chlorine solution and in 0.5 %
citric acid solution. The packaging materials tested did not affect the shelf life of grated carrots. Long-
term storage before processing was detrimental to the quality of the packed samples.

INTRODUCTION

The shelf life of packed grated carrots can possibly be prolonged by using chlorine wash for reducing
microbial contamination, by using citric acid wash for reducing enzyme activity and by using highly
permeable films for preventing anaerobic respiration. Furthermore, the effectiviness of chlorine wash
can be improved by high washing temperatures. The aim of this study was to determine the effects of
different processing procedures on the shelf-life of packed grated carrots. In addition, the effects of
long-term storage before processing were studied.

MATERIALS AND METHODS

The carrots (Navarre) were washed and all heavily contaminated parts were removed. The carrots were
then peeled with carborundum, washed (Table 1), grated into 3 mm thick strips, gently sprayed with
water, spin dried and packed. Five packaging types and two packaging sizes were used (1.0 kg
carrot/package, vol. 3000 cm3 , film thickness 40.um; PE-film impregnated with ceramic powder, PE-
EVA-film and OPP-film, 1.6 kg carrot/package, vol. 5000 cm3; PE-paper-lid + PP-tray and PET-carton
package). The O 2 permeabilities of the films were 5800, 5200, 1200 and 1500 cm 3jm 2 24 h 101.3 kPa,
23°C, RH 0 %, respectively. The packages were stored at 5 °C for up to 8 days.
The percentages of CO 2 and O2 in the package headspace were determined with Servomex PA 404
and 507A analysers, respectively. All analyses were carried out from four replicate packages.
 889
Microbiological determinations (pour plate techniques, duplicate sample packs) carried out were:
1) aerobic plate counts (APC) after incubation at 30°C for 3 days (Plate Count Agar, Difco), 2) lactic
acid bacteria (LAB) after incubation at 30°C in 5 % CO 2 for 3 days (MRS agar, Oxoid), 3) yeasts and
molds (Y & M) after 5 days at 25°C (YGC Agar, Difco).
Sensory evaluations of the samples were performed by trained panel members. First, the odour and
appearance of the samples were evaluated by 5 panelists, followed by flavour evaluations by 10
panelists (10-20 minutes after the first evaluation). All analyses were carried out on duplicate samples.
Samples were scored on a scale of 0 to 5 (5 = excellent, 2 = unacceptable for retailing and 0 = unfit
for consumption). The minimum level of acceptability for human consumption was 1.5.
a- and f3-carotene acid concentrations were analysed using HPLC (1). The residual free chlorine
concentrations (samples ClW in Table 1) were analysed spectrofotometrically 16 hours after packaging
(2). The detection level for free chlorine was 0.1 mg!kg.
The respiratory activity of the grated carrots was measured by the flow-through system. Air was
fed into an airtight chamber via one port and allowed to exit from another port. The O2 and CO2
concentrations of the outcoming gas were determined. Measurements were terminated when both the
O2 and the CO 2 concentrations had stabilized (four replicates).
The O 2 and CO 2 percentages and the results of the sensory evaluations are presented as mean values
± standard deviations and the microbial counts as mean values (log cfu/g) (Tables 1 and 2).

RESULTS AND DISCUSSION

Washing the peeled carrots in water containing 0.01 % chlorine or 0.5 % citric acid (5°C) improved
the maintenance of sensory quality of the packed grated carrots (Table 1). The same effect was
obtained by washing the peeled carrots first in chlorinated water (30°C) and then in pure water (5°C).
The washing methods did not, however, significantly affect the microbial load, the CO 2 production and
O2 consumption, the a- or f3-carotene concentrations (2 mg/loo g fresh weight and 8 mg/l00 g fresh
weight, respectively, just after grating and 2 mg/lOO g fresh weight and 6 mg/lOO g fresh weight,
respectively, after 8 days of storage) or the respiration rates (10 ml CO 2/kg h, measured from samples
W, CI and CA) of the samples. No residual free chlorine was found in chlorine-treated samples.
Obviously, the chemical treatments lowered the enzymatic activity of the grated carrots. The packaging
materials tested did not significantly affect the shelf life of grated carrots (Table 2). Drying slightly
reduced the shelf life of the grated carrots packed in PET-carton packages. The shelf life retention of
the samples OPP (Table 2) was clearly better than that of the similarly processed samples W (Table 1).
This was possibly because of better manufacturing practices (e.g. hygiene) used in the processing of
the samples OPP (the samples OPP were prepared in the laboratory and the samples W in an industrial
processing line).
Long-term storage before processing was detrimental to the quality keeping of the grated carrot
(not washed, packed in OPP-film). For example, the odours of the samples made from carrots stored
for 2 months were scored better than those of samples made from carrots stored for 7 months: after 8
days of storage the scores were 1.4 and 0.7 (season 1991) and 1.4 and 1.2 (season 1992), respectively.
Obviously, this was due to enzymatic activity of the carrots, because the long-term storage did not
affect the a- or f3-carotene contents (2-3 mg/lOO g fresh weight and 7-8 mg/100 g fresh weight,
respectively), the microbial load or the respiration rate (4 ml CO 2/kg h) of the peeled whole carrots.

REFERENCES

1. Ollilainen, V.M., Heinonen, M., Lifikola, E., Varo, P. and Koivistoinen, P., Carotenoids and retinoids in
Finnish foods:meat and meat products. I:. Food Compo Anal., 1988,1, 178-88.
2. Test method VTI-4426-91, VTI Food Research Laboratory, Espoo, Finland, 1991.
890
TABLE 1
The effects of washing method (3 l/kg carrot, 1 min/wash) of peeled carrots and of storage time on
the quality retention of grated carrot packed in OPP-film. NW = not washed, W =washed in 5 °C
water, CI = washed in 5 °C 0.01 % free chlorine solution (as sodium hypochlorite), CA = washed
in 5 °C 0.5 % citric acid solution, ClW = washed first in 30°C chlorinated water (0.01 % free
chlorine) and then in 5 °C pure water. The carrots were stored for 2 months before processing.

Quality Storage timeLwashing method

attributes 2 days 4 days 8 days
NW W CI CA CIW NW W CI CA CIW NW W CI CA CIW

CO2 (%) 8 9 9 8 11 15 17 16 14 14 31 34 29 33 29
:to :tl :tl :tl :tl :t2 :tl :t2 :tl :tl :t2 :t3 :t2 :t5 :t3
O2 (%) 7 5 5 7 3 0 0 0 0 0 0 0 0 0 0
:to :t2 :t2 :tl :t2 :to :to :to :to :to :to :to :to :to :to

APC 5.1 5.0 4.5 6.1 4.9 nd nd nd nd nd 6.7 6.6 6.4 5.9 6.3
LAB 4.4 5.2 4.3 4.4 4.0 nd nd nd nd nd 7.7 7.8 7.6 7.7 7.7
Y&M 2.9 3.0 3.0 2.7 3.0 nd nd nd nd nd 3.8 5.0 3.7 3.5 3.5

Odour 4.0 4.1 4.0 4.2

4.1 2.8 3.3 3.5 3.5 3.8 1.4 1.3 2.9 2.9 2.6
:to.2 :to.4 :to.2 :to.1
:to.2 :to.1 :to.5 :to.5 :to.5 :tOA :to.8 :to.2 :to.5 :to.3 :to.6
Appear. 4.2 4.2 4.2 4.2
4.3 3.9 3.9 3.9 3.9 4.1 3.6 3.6 3.6 3.6 3.7
:to.2 :to.2 :to.2 :to.2
:to.2 :tOA :tOA :tOA :tOA :to. I :to.3 :to.3 :to.3 :to.3 :to.2
Flavour 3.5 3.5 3.6 3.8
3.7 3.2 3.3 3.5 3.3 3.5 na na 2.9 3.0 3.1
:to.5 :to.3 :tOA :tOA :to.3 :to.5 :to.5 :tOA :to.5 :tOA :to.5 :to.8 :to.6

TABLE 2
The effects of packaging material and storage time on the quality retention of grated carrots. Peeled
carrots were washed in 5 °C water before grating and packaging. The carrots were stored for 2
months before processing. OPP = OPP-film, PE = PE-film impregnated with ceramic powder, EVA
= PE-EVA-film, Pap = PE-paper-film + PP-tray, Car =PET-carton package (not sealed).
Quality Storage time/packaging material
attributes 2 days 5 days 8 days
OPP EVA PE Pap Car OPP EVA PE Pap Car OPP EVA PE Pap Car

CO2 (%) 10 6 7 7 1 19 10 13 17 2 25 15 15 27 2
:t2 :tl :tl :tl :to :tl :to :t2 :t2 :tl :t3 :tl :t4 :tl :tl
O2 (%) 2 4 1 9 20 0 0 1 2 20 0 0 1 0 20
:t3 :t2 :to :t3 :to :to :to :t2 :t3 :to :to :to :t2 :to :to

Odour 3.9 3.7 3.0 4.1 3.7 2.9 3.0 3.1 3.5 2.9 3.0 2.8 2.8 2.9 3.1
:to.3 :to.3 :to.3 :tOA :tOA :tOA :tOA :to.3 :to.S :to.5 :tOA :tOA :to.6 :tOA :to.8
Appear. 4.3 4.3 4.3 4.3 4.3 4.3 4.3 4.4 4.4 4.2 4.4 4.4 4.4 4.4 3.7
:to.3 :to.3 :to.3 :to.3 :to.3 :to.7 :to.7 :to.6 :to.6 :to.5 :to.6 :to.6 :to.6 :to.6 :to.6
Flavour 3.6 3.5 3.7 3.9 3.8 3.3 3.2 3.3 3.0 3.4 3.1 3.4 3.1 3.0 3.1
:to.6 :to.5 :to.S :to.6 :to.6 :to.5 :tOA :tOA :to.6 :to.7 :to.7 :to.6 :to.9 :to.6 :to.6

nd = not determined, na = quality not acceptable for human consumption.

 STORAGB TBCIINOLOOY ON DBllYDBA.TED 8WOBD BRAN

ZHANG MIN KONG BAOHUA WANG CBBNZHI YANG YONGU

(Nortbeut AgrIcu1turaI Collep, HarbID 160080,P.B. ChIna)

AB8'l'1U.CT

In 0liIJ paper, a&onge teBtI on dedydnted IWOrd been were COIldueted, and
the belt a&onge ieeImoIogy WI8 deiermiDed. Then, the 8ierlIIzation mechanlmt of
the methoda before Itonge and the efteefIJ of temperature, moisture, and puking
on a&onge are diIewIJed.

INTRODUCTION

Every y-.r, a great quantity of dellydrated vegetable8 alllorb moilhue beQue

of improper Itonge wide, ean lead to the _ of nutrition and pigment, even to
the growtb. of mold8 and tile pl'elJeDt'le iD8eetI wide, may Que Ole deeay (C.Z.
Wang,1990). 80 it ill very imporiment to study Ole a&onge of deIlydrated veget-
ablel for eIpOrt.
The dehydrated IWOrd beaD, wide, is an important delydrated produd, wu
selected in 0liIJ study to obtain the btwt stonge teeJmology.

1.lfateriall
Dellydrated IWOrd been with 8% moiliure CODtent WI8 uaed u tile 8Ulple for
the a&onge teItI. The IlaDdHDg methodl before a&onge are maiDIy tieoDdatlon,
&teriIizatlon, and radiation •
2.Methodl
.... of Ole teItI: I're8Il IWOrd bean wu pleled from tile vegetable prden
at our eollege, then pretreated (lOme of the 8Ulple WI8 added wlOi Ole baete-
rieide) and deIlydrated.The 109 8Ulple WI8 put into Ole boUle wIOl Ole eo~
poDding .turated IOlution lor the a&onge 8&. Meanwlile,Ole deoDdiHr wu
put into lOme of Ole boWel, and Ole mouOl of Ole boWe wu aeaIed wIOl vue-
892
line. Then lOme of Ole 8IIIIlples were irradJated. The stonge te&t.a wen earried
out at a temperature of 9O-3O"C. The eontentll of ehlorop.yD and vitamin C(Ve)
WeI'e WI8d 18 Ole test eriteria, and Ole number of molds and baeteria were alIIO
eoD8idered.

RBJULTS .AND ANALY8IS

The primary test Dve sIlown tOt Ole main faeton wide. ailed Ole above
four eriteria are Ole meOiod of pretreatment before &tonge, activity of water,
and time of stonge. The tMee levels of every faetA)r were IU'I'IUIg8d in Table 1.

Table 1 The faeton and levels of Ole orthogonal

storing test for de.ydrated nOM bean

faeion A B C
pretreatment water time of stonge
before storage activity (montlul)
levels

1 deoxygenation 0.86 0

S sterilization 0.76 8

8 radiation 0.68 6

The re&ult. and aoalyail of Ole orthogonal test WeI'e given in Table S(omitted).
For eomparison, normal teelmology te&t (lCJ#) WII arranged.The re&ult. preIIeDted
in Table 2 sIlow Olat Ole I'tlIUltll of te&t 1CJ# are poor in Ole four eriteria,1IO Ole
improvement of Ole teelmology ill needed.
The result. in Table S sIlow tOt Ole order of importanee of Ole faeton and
their belt levels are A1 , ~, 11. for Ole eontentll of ehlorop.yD and V.. (Xy,Kvv),
and Ba, Aa(C.) for Ole number of molds and bacteria (X.,Kx ).
To judge Ole eriteria synOietleaDy,Ole proper powen for every eriterion Dve
been determined (y .E.He,l986).The powen of Ole eontenfa of ehlorop.yll and V..,
an4 Ole number of molds and baeteria are 0.8,0.2,and 0.26 respeetively.The values
witIl "." in Table S are Ole eompre.enslve :re8IlIta. The order of importanee of Ole
faeton and Ole belt levels are Be, ~, A1 after synOietie evaluation, tot is,Ole
greatest effect on Ole eriteria i8 water activity, Ole aeeond one i8 stonge time
and Ole leut one iIJ method of pretreatment before storage. The . . water aetivity
and &torage time are, Ole betier Ole eompre.enslve erIteria. Water activities Dve
IitUe effect between 0.68-0.76, whDe stonge times bve Httle effect between o-a
montlul. Deoxygenation before norage ... Ole belt effect on Ole eriteria, whDe
steriIlzation and radiation bve Httle effect on Ole resuIta.
AD tIling8 eonsidered, Ole held teelmology of &torage is 18 follows: (»ntrolling
 893
the moisture in dehydrated sword bean before stonge,keepiDg the relative humidity
)ftIB than 76%, deoIidJzIng the paekage, and keeping the stonge time to )ftIB than
3 months.

DISCUSSION

The eomparative analysis of the sterilization meehanism under three methodB of

pretreatment before stonge.
It WII obaerved that the main eaWle of the deeay 01 dehydrated nord bean wu
that the number 01 molda and buteria were beyond the National standard. It hu
been reported (N.W.Dtwroier,I989) that bIeterielde,deoxygenation,and radiation ean
efIeetively inhibit the growth 01 mohta and bae&eria, but the ftII1lIt8 wiD vary with
different material, baeterieide, deoxydizer, and irradiating dGIeI. Deoxygenation
hu the mOlt aipilieant efIeet under the test eonditioDll. 8terilization and radi-
ation have _ effeet, which may be beeaWle the eoneenntioDll 01 bacterielde and
dole of irradiation were too low. However,the primary teBt8 bave shown that ehIo-
ropllyn and V., wiD be destroyed heavily if the dole of irradiation II up to 8
kOy,and the recommended eoneenntlon of dehydrated lJOdJum aeetate (baeterieide)
In vegetable proeeadng II 0.1%. Therefore, deoxygenation II the _est and mOlt
elleetive method of pretreatment.

CONCLUSION

1.The belt teehnology 01 storing dehydrated sword bean II to eontrol moilture

belore storage, keeping the relative humidity to _ than 76"., deoDHzing the
inside of the paekage, and keeping the storage to IftIB than 3 months.
2.The sterilization meehanism 01 dJIIerent pretreatmenil before stonge hu been
analyzed eomparatively.Deoxygenation II eoD8ide:red II the .rest and the most eII-
eetive method.

I.C.Z. Wang, The developing protJpeet and eountermeuu:re of lood irradiation

in China, Isotope (ChineBe), 1990(2).
2.L.W.Zhang,The study on IreIJh . ' 8 stonge, the muter paper 01 Northeast
AgrleuItunI eonege, 1988.
3. Y.E.He et ai, The testing deaignJng 01 the agrIeuIturaI maehJDery,
the Mwinery IndUltry PrtlII,I986.
 EFFECTS OF SEVERAL TREATMENTS ON THE QUALITY OF COLD STORED
'CLEMENTINE MANDARINS'

F. PAZIR AND M. AZAK

Food Engineering Department of Engineering Faculty,
Ege University,
35101 Bornova-Izmir, Turkey

ABSTRACT

The effects of harvesting type (careful and rough harvesting), dipping into fungicide solution (TBZ +
imazalil) at different temperatures (20, 40, 50, 60°C for 2 min.) and wax application on the quality of
cold stored 'clementine' mandarins were investigated. Mandarins were stored at 6°C and 80-85% RH
for 3 months. Juice content, titratable acidity, and ascorbic acid content decreased throughout the
storage period. The inverted sugar content increased during the storage period although the changes of
total sugar and soluble solids of mandarins were not significant. Surface deformation and off-flavour
were detected in the ones which were dipped into fungicide solution at 60°C. Rough handling increased
the amount of decayed fruits.

INTRODUCTION

One of the most important factors which lengthens the storing life of the cold stored citrus fruits is the
way of harvesting. Most damages, during or after harvest are mainly due to some strikes, in other
words composed by impact forces, which are produced by dropping from a height. 1
Citrus fruits should be picked by trimming them off the branches, instead of plucking, since the risk
of decay increases two times following plucking and carriage. 2
Bruises due to the rough harvesting increase the rate ofrespiration of the fruit, with a great decay.4
Washing prior to storing and application of fungicide and waxing are effective practices in the
prevention of the fruits decaying. Chilling injury is one of the main problems and such damage can be
reduced considerably by hot-dip procedures, and the usage of TBZ increases the efficency of the
procedure. 6

MATERIALS AND METHODS

Clementine mandarins were harvested in December, 1991 and 1992. Fruits were then randomised into
12 treatment units of approximately 200 fruits each.
The mandarins were separated into two lots; careful and rough handling. In careful harvesting, the
fruits were picked with clippers and placed in the picking box slowly. In rough harvesting, fruits were
also picked by clippers but without considering their stem lengths and were allowed to drop from up to
50 cm heights. The treatment units are shown in Table 1.
Mandarins were dipped into fungicide solution for 2 min. All of the fungicide solutions and wax
contained 2000 ppm TBZ-750 ppm imazalil.
 895
Table 1 Treatment units which were applied in
the research
A. Careful harvesting B. Rough harvesting
a Control a Control
b 20°C b 20°C
c 40°C c 40°C
d 50°C d 50°C
e 60°C e 60°C
f Wax. f Wax.

Fruits were then held in wooden boxes and stacked in the cold room in the two replicate groups. All
groups were stored for 12 weeks in a cold room with fan forced air circulation which operates at
6 ± 1°C. For all the groups, weight loss, juice content, titratable acidity, pH, ascorbic acid, total
inverted sugar, color values (L,a,b) and sensory scores were measured every three weeks. Sensory
analysis was made according to the diagram of Karlsruhe, but the scores have been changed to 1-9.
The 60°C treatment units were not included in the research of 1992.

RESULTS AND DISCUSSION

At the end of three months storage in both years, the juice content (8.2-14.4%), titratable acidity
(60-63%) and ascorbic acid (29.7-34.2%) had decreased. These decreases were highly significant
(p < 0.05). The reduction in the titratable acidity tends to occur as a result of the use of citric
acid in respiration and in the synthesis of amino acids. 3
Hunterlab colour values of the peel were measured. L (lightness) value and alb (orange colour)
value decreased 3.3-4.5% and 6.8-7.5% respectively. L values of the 600C processed samples were
rapidly decreased during the storage. The reduction of L and alb values were 20% and 33%
respectively. This can be accounted for by browning as a result of casting surface oil off the peel and
therefore oxidation. 5 In the 50°C processed samples orange color changed into yellow and alb value
decreased by 20%. The changes were found insignificant in all the samples except in those two
samples. The sensory analyses of the treatment units were by appearance, color and taste. When
examining the appearances of the mandarins of 1991, it was observed that during the process of storage
the peel was removed from fruit segments, with sponginess occuring. In the samples of 1992, there was
no considerable sponginess and other characteristics resembled those of the results in 1991. In the
samples which were dipped at 60°C, some browning was detected due to surface deformation. After
studying the samples in view of color it was observed that waxed samples preserved their color
followed by the control, 20°C, and 40°C sample. It was found that 50°C treated orange color of the
samples turned to yellow and 60°C processed ones were browned. In all of the samples loss of taste was
observed because the acid-sugar balance was changed due to decrease in acidity. Off-flavour,

Table 2 The sum of the decay and weight loss at different temperatures and waxing according to
rough and careful handling at the end of storage for two years. (%)
1991 1992
Treatment Careful harvest Rough harvest Careful harvest Rough harvest
units I.M 2.M 3.M I.M 2.M 3.M I.M 2.M 3.M I.M 2.M 3.M
Cont. 1.8 9.5 53.2 12.1 28.0 64.6 2.6 4.0 17.0 3.6 6.9 21.6
20°C 1.7 7.2 14.5 6.7 13.8 29.9 2.8 4.3 8.2 2.4 3.8 8.5
40°C 1.4 4.9 36.1 1.3 3.4 32.9 3.3 4.8 8.9 3.9 5.1 9.5
50°C 2.9 7.4 51.4 2.7 5.5 39.7 3.0 4.1 8.0 2.8 4.1 9.9
60°C 2.6 11.3 69.3 2.6 10.5 74.5
Wax. 0.2 13.9 48.4 3.1 11.1 56.1 2.5 4.9 11.2 1.9 3.1 17.1
M: months.
896
especially was observed in 60°C treated samples. Weight loss and decaying were observed in all
samples during the storage. The mandarins which were processed at 20°C showed the least weight loss
and decay for each storage year. It was observed that waxing and fungicide application protected the
fruit quality well until the late second month. Decaying suddenly increased in the fruits following this
period when the effect of wax processing decreased. The rates of decaying were relatively less in all the
treatment units in 1992, because of the falling rain and slight frost before the harvest of 1991. The
decaying and the weight loss results are given in Table 2.

REFERENCES

1. Fluck, R. C. and Ahmed, E. M., 1985. Measurement by compression test of impact damage to
citrus fruits. 1. Texture Studies 4, 4-500.
2. Grierson, W. and Ben-Yehoshua, S., 1987. Storage of citrus fruits. Chapter 20, in Fresh Citrus
Fruits p. 484. Avi Publishing Company, Connecticut.
3. Kara<;ali, I., 1990. Protection and marketing of orchard products. Ege University Publishing, Izmir.
4. Munoz-Delgado, J. A., 1987. Problems in cold storage of citrus fruit. Int. 1. Refrigeration, 10,229-
233.
5. Usai, M., Arras, G., Fronteddu, F., 1992. Effect of cold storage on essential oils of peel of
thompson navel oranges. 1. Agric. Food Chern. 40, 271-275.
6. Wild, B. L., 1990. Hot dip treatments reduce chilling injury during storage of citrus fruit at 1°C.
Food Research Quarterly, 50, 36-40.
 POSTHARVEST TREATMENTS ON QUALITY OF MANGO FRUITS
(Mangifera indica (L) var. ZILL) STORED AT LOW TEMPERATURE

MARIA AMALIA BRUNINI KANESIRO, JOSE FERNANDO DURIGAN,

RAUL ROBERTO DE SOUZA FALEIROS, DENISE MARIA ANDRIOLLI
Department of Technology,
College of Agricultural Science and Veterinary "Campus" Jaboticabal
San Paulo State University (UNESP)
(14870) Jaboticabal, SP, Brazil

ABSTRACT
The effect of postharvest treatments by using hot-water and benomyl
solution and combination of both on the quality of mango fruits var. "Zill"
during storage at 12 0 C, 95% RH and at normal atmosphere (24-27 0 C, 65-85%
RH) were investigated. The parameters analyzed were weight, peel color,
total soluble solid, starch content, titrable acidity, soluble carbohydrate
and general appearance (presence of rottenness in the peel). Results
allowed to conclude that the treatments did not interfere with the fruits'
ripening process, which was showed by the increase in total soluble solid
and decrease of titrable acidity and starch content; that neither heat or
fungicide treatments were able to control the rottening grocess on fruits
stored under low temperature; and that storage at 12 C was the most
suitable to increase the shelf life these fruits.

INTRODUCTION
Mango fruits are becoming one of the most important commodity of Brazil,
which recently has become the sixth producer-country of mango fruits[l].
With the increase of exportation and fresh consumption of mango fruits,
many techniques of postharvest treatments and storage condition have been
studied in order to extend the shelf life maintaining the quality and
reducing the incidence of storage diseases. Among these techniques, hot
water treatment and low temperature storage have been reported to be
effective against latent infection of Collestotrichum gloeosporioides Penz,
which is causative of anthracnose, and that hot water treatment is more
effective when a fungicide is incorpored[2,3,4,5,]. Then, the objective of
this study was to investigate the effect of postharvest treatment on the
ripening quality and anthracnose development in mango fruits cv. Zill
during storage.
898
MATERIAL AND METHODS
Mango fruits cv. "Zill" were harvested at unripe stage from the
Agricultural Experimental Station of the College of Agricultural Science
and Veterinary "Campus" of Jaboticabal/UNESP, Jaboticabal City, San Paulo
State, Brazil and transported to the Laboratory of Technology, where they
were randomly divided into four experimental groups of 80 fruits. After,
experimental groups of 80 fruits were im~ersed in one of the following
postharvest treatments: hot-water at 55 C for 10 minutes (HWT); 0.2%
benomyl solution for 5 minutes (FT); and hot-water at 55 0 C for 10 minutes
and subsequent 0.2% benomyl solution for 5 minutes (HWFT). One experimental
group of 80 fruits was untreated and was used as control (Control). After,
each experimental groups were separated in two lots of 40 fruits. One lot
of 40 fruits of each experimental group was stored in refrigerated-room at
12 0 C, 95% RH and the second lot of 40 fruits of each experimental group was
stored in room at normal atmosphere (24-27 0 C, 65-85% RH). For each stored
lot of mango fruits, 20 fruits were separated and used to estimate the
weight loss, peel color and presence of rotten in the peel.
The ripening qualities of the fruits were measured at harvesting time and
during storage. Weight fruit, weight fruit pulp, peel color, general
appearance, titrable acidity (TA), total soluble solid (TSS), soluble
carbohydrate and starch content were measured as the component of ripening
quality. Weight loss was measured by weighing of each fruit and expressed
in %. The pulp weight was calculated from the difference between fruit
weight and the weight of peel and seed, and expressed in %. The surface
peel color was examined visually and assessed on a five index: 1= green
color; 2= more green color than yellow color; 3= equal amounts of green
color and yellow color; 4= more yellow color; and 5= yellow color with red
color areas. The disease was also examined visually and assessed on a five
index: 1= healthy ; 2= more healthy areas than rotten areas; 3= equal
amount of healthy and rotten areas; 4= 2/3 of the surface peel with rotten;
and 5= full rotten or more 2/3 of the surface peel with rotten. Titrable
acidity was measured by titration method using O.lN NaOH and the results
was expressed in g of citric acid/100g of pulp. Total soluble solid was
measured directly in the expressed jHice by using a digital refractometer,
and the results was expressed in Brix. The soluble carbohydrate was
determined by using the phenol-sulfuric method[6] and the values were
expressed in g of glucose/ 100g of pulp. Starch content was hydrolyzed into
reducing sugar, which was analyzed by using phenol-sulfuric method[6], and
the starch content was calculated by multiplying 0.9 of reducing sugar
content.

RESULTS
The results of this study gave evidence that storage temperature had
effect on the ri peni ng and storage li fe of treated and untreated "li 11"
mango fruits. The storage at 12 0 C delayed the ripening and extended the
storage life of the fruits until 26 days, while storage at normal
atmosphere reduced the storage life of the fruits, accelerated the ripening
and the senescence of the fruits (Table 1). The visual examination of
treated and untreated- stored fruits showed that the fruits had greater
peel color changes and disease development, and that the effect of HWFT on
chlorophyll change was less detrimental than others treatments in fruits
stored at 12 0 C (Table 1). It was observed that neither heat or fungicide
 899
treatments used in this experiment were able to control the incidence of
anthracnose development, but these treatments associed with low temperature
storage might retard the incidence of this disease.
Based on the pattern of changes that occurred in the chemical
properties of stored fruits (Tablel), it could be deduced that the
postharvest treatments did not affect these properties. Also, it could be
observed that there was no difference in the degree of ripening between
treated and untreated fruits in both storage condition.

TABLE 1
Changes in "Zi 11" mango fruits quality at the end of storage

Quality Initial Cont*ol maggo HWT manso FT Manso HWFT manso

index NA 12 C NA 12 C NA 12 C NA 12 C
--------------------------------------------------------------------------
Storage life(days) 11 26 11 26 11 26 11 26
Weight loss(%) 100.0 86.5 91.2 86.7 89.9 87.7 92.2 84.9 90. 1
Pulp weight(%) 67.1 75. 1 81.5 77.8 75.8 79.8 82.6 78. 1 80. 7
Peel color index 1.6 5.0 4.8 5.0 4.0 5.0 4.0 5.0 3.8
Disease index 1. 1 4.0 3.8 3.5 4.0 4.0 3.5 4.3 3.8
TA(g61OOg ) 1.3 0.2 0.6 0.2 0.8 0.2 0.9 0.2 0.8
TSS( Brix) 7.8 15.5 16.0 15.0 14.0 15.0 16.5 18.5 15.5
Glucose(g/100g) 2.6 9.6 9.8 9.4 9.8 8. 1 7.7 8.5 7.0
Starch(g/100g) 13.5 5.0 5.0 5.4 5.0 6.8 6.5 5.7 7.7
* NA= normal atmosphere storage

REFERENCES
1. FAO. Production yearbook. Roma, 1988, 42, pp. 218-219
2. Quimio, T.H., Mango anthracnose and low temperature storage, Philippine
Agriculturist, 1974, 58, 192-199.
3. Dodd, J. C., Jeffries, P. and Jeger, M.J., Management strategies to
control latent infection in tropical fruit, Aspect of Applied
Biology, 1989, 20, 49-56. In Dodd, J.C., Bugante, R., Koomen, 1.,
Jeffries, P. and Jeger, M.J., Pre-and post-harvest control of mango
anthracnose in the Philippines. Plant Pathology, 1991, 40, 576-583.
4. Thompson, A.K., The development and adaptation of methods for control of
anthracnose. I n Dodd, J. C., Bugante, R., Koomen, 1., Jeffri es, P. and
Jeger, M.J., Pre- and post-harvest control of mango anthracnose in the
Philippines. Plant Pathology, 1991, 40, 576-583.
5. Dodd, J.C., Bugante, R., Koomen, I., Jeffries, P. and Jeger, M.J., Pre-
and post-harvest control of mango anthracnose in the Philippines, Plant
Patho logy, 1991, 40, 576-583.
6. Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A. and Smith, J.K.,
Colorimetric for determination of sugar and related substances.
Analytical Chemistry, 1956, 28, 350.
 FLAVOR CHANGB OF GRAPB JUICB DURING PROCBSSING

HIDEAKI OHTA, YOICHI NOGATA and KOH-ICHI YOZA

Chugoku National Agricultural Research Institute
Ministry of Agricu1ture,Forestry and Fisheries
Fukuyama, Hiroshima 721, Japan

ABSTRACT

The flavor change of Concord (Vitis labrusca L.) grape juice

during processing was examined. During manufacture of
concentrated grape juice, 99.3% of the total peak area of
headspace volatiles detected in fresh grapes had disappeared.
The loss was greatest during the heating process (53%), followed
by concentrating process by evaporator (27%) and clarifying
process with enzymes (18%). On the other hand, both the heating
process to extract color and clarifying process provided a
larger amount of flavor extracted with sol vent. Methyl
anthranilate was abundant during the heating process to extract
color, but it was reduced to half after the concentrating
process.

INTRODUCTION

Little information concerning the flavor change of grape juice

during processing has been reported, although flavor loss -and
change of fruit juices have been considered as one of the most
important factors affecting the quality of fruit Ju~ce.
Manufacture of concentrated grape juice involves two unique
processes such as heat treatment for extracting color and
filtration for removing argo1. In this study, we clarified the
flavor change of Concord grape juice during processing.

MATBRIALS AND MBTHODS

The juices investigated were sampled during the commercial scale

processing, including fresh grapes, heating to extract color
(fruit blancher, 70°C, 15 min), extracting, pasteurizing,
clarifying, concentrating and the final product. Analysis of
 901
headspace volatile components was performed by gas
chromatography (GC) according to the Tenax trapping technique
[1]. The flavor compounds from the sample juice recovered by
continuous liquid extraction using with n-pentane and methylene
chloride (2: 1) [2]. The extract was dehydrated with sodium
sulfate, then concentrated in vacuo (2S0C, 10 mmHg) for GC
analysis. 10 % benzyl acetate was used as internal standard.
Flavor compounds were analyzed quantitatively by GC using a
Hitach 163 with a flame ionization detector. A WCOT glass
capillary column (70 m x 0.50 rom I.D.) coated with PEG 20M was
used. The column oven temperature was programmed from 60 to
190°C at 3°C/min. Nitrogen was used as the carrier gas at a flow
rate of 2 ml/min with a splitting ratio of 10:1 for headspace
analysis and 20:1 for solvent extract analysis, respectively.
GC-mass spectrometry was performed as described previously [2}.
RBSULTS AND DISCUSSION

The presence of 27 peaks was demonstrated in the headspace of

fresh grapes. Nineteen of these peaks were identified by
comparing and matching the mass spectra and GC retention times.
Figure 1 shows the relative changes in total peak area of
headspace volatiles and peak area of ethyl crotonate during the
processing of grape juice. During manufacture of concentrated
grape juice, 99.3% of the total peak area of headspace volatiles
detected in fresh grapes had disappeared. The loss was greatest
during the heating process (53 %), followed by the concentrating
process by evaporator (27%) and clarifying process with enzymes
(18%). The quantity of ethyl crotonate, a key aroma component
[3], decreased in a similar manner.
100% 100%

Total peak area Ethyl crotonate

46.7% 47.4%

28.6% 24.2%

1.8% 1.2%

ABC DE AB CD E
Figure 1. Retention of total peak area of headspace volatiles
and ethyl crotonate from Concord grape juice during
processing.
A: Fresh, B: Heating to extract color, C: Clarifying with enzymes,
D: Concentrating, E: Final product.

The yield of flavor extract concentrate from the juice

during the heating process to extract color was 3.22 mg kg-I,
followed by fresh grapes (2.56 mg kg-I), clarifying with enzymes
(2.37 mg kg-I), concentrating process (0.52 mg kg-I) and final
product after filtration (0.50 mg kg-I) .
902
We detected 111 peaks in the solvent extract and identified
49 compounds. Figure 2 shows the change of four compounds, which
contributed to the Concord grape flavor [3], during processing.
As expected, 2-methyl-3-buten-2-ol and ethyl crotonate, the low-
boiling compounds, decreased markedly during the heating,
clarifying and concentrating processes, showing the similar
behavior in the headspace analysis. On the other hand, methyl
anthranilate, which is well-known as a foxy flavor [3] and high-
boiling compound, was abundant during both the heating process
to extract color and clarifying process, but it was half after
the concentrating process.

Methyl anthranilate

Ethyl Crotonate

Ethyl anthranilate

Figure 2. Retention of some flavor compounds from Concord grape

juice during processing.
See Figure 1 for A, B, C, D and E.

These findings suggested that efforts should be made to

recover the flavor from juices during both the heating process
to extract color and concentrating process.
R.BPBR.BNCBS

l. Ohta, H., Tonohara, K., watanabe, A., Iino, K. and Kimura,

S., Flavor specificities of satsuma mandarin juice extracted
by a new-type screw press extraction system. Agric. Biol.
Chern .. 1982, 46, 1385-1386.
2. Ohta, H., Nogata, Y., Yoza, K. and Maeda, M., Glass capillary
gas chromatographic identification of volatile components
recovered from orange essence by continuous liquid
extraction. ~ ChramatQsr., 1992, 607, 154-158.
3. Stern, D.J., Lee, A. McFadden, W.H. and Stevens, K.L.,
Volatile from grapes: Identification of volatiles from
Concord essence. ~ Agric. Food Chern., 1967, 15, 1100-1103.
 PRODUCTION OF POWDERY PRODUCTS OUT OF CONCENTRATED FRUIT-
JUICES AND OTHER FOOD AND BIO-PRODUCTS

T.Gamse, R. Marr, F.Froschl, E.Moschitz

Insitute of Thermal Process and Environmental Engineering, TU Graz

Head of the institute: o.Univ.Prof.Dipl.-lng.Dr.techn.R.MARR
Inffeldgasse 25, 8010 Graz, AUSTRIA

1.) Basic project idea:

At the Institute of Thermal Process and Environmental Engineering, TU Graz,

experiments for the development of a new process were carried out. Because of the
high percentage of water included in fruits there is the negativ effect on the storage
and the water has to be transported. If it is possible to remove this water, there will
be less transportation weight and the possibility for a longer dated storage. The new
idea of this process is to freeze the fruit juice in fine drops and to remove the water
in a subsequent vacuum dryer by sublimating it.

2.) State of the project:

For the realization of the basic idea a simple discontinous plant was built at the
institute and in a feasibility study experiments with different raw materials were carried
out.
The fruit-juice it is compressed in a storage tank (see figure 1). Leaving this
vessel the fruit-juice is dispersed in fine drops into the crystallize with the help of a
nozzle. The necessary performance for freezing will be reached by dropping the CO2-
pressure from 60 bar to atmospharic conditions. By this flash evaporation the CO2
904

j' ,,[5. lff

withdraws the crystallizer and the fruit juice the 5 6 CD.

equal amount of heat and the fruit juice drops free-

ze immediately. It is necessary to coordinate the
flow rate of the CO2 and of the fruit juice. The
1... subcooler
intermediate product, which is collected at the filter 2 ... crystallizer
3 .. .filter cloth
cloth at the bottom of the crystallizer, is very fine 4... receiver for fruit pulp
5 ... ring conduit with 4 boreholes
crystalline and is put into the vacuum dryer. 6... dispersing nozzle for fruit pulp
Figure 1: flow sheet of the dispersion

Figure 2 shows the vacuum dryer with the vacuum pump and the condenser.
The pressure of the dryer has to be lower than the tripple point of water (5 mbar)
because otherwise the water cannot sublimate and the intermediate product would
melt. If that happens no powder can be produced. A very important factor in this part
is the efficiency of the vacuum pump.
The condenser has to make sure that
5
no water reaches the vacuum pump.
If the seperation is insufficiently water
1... vacuum dryer
reaches the oil of the pump, which 2 ... stop valve
3 ... cooling trap
has the effect of a decreasing pump- 4,6 ... air release valve
5 ... vacuum pump
efficiency and an increasing pressure Figure 2: flow sheet of the vacuum-dryer

in the vacuum-dryer.

Another important point is the synchronization of the heating in the vacuum-

dryer and the efficiency of the vacuum-pump and the condenser. If there is too much
heating power the condenser and the vacuum pump cannot remove all the water and
therefore the pressure will increase. On the other hand the time for drying decreases
with increasing heating because more water sublimates from the frozen product.

Limiting factors for this process are the percentage of solid matter and of sugar
in the raw material. With low percentage of solid matter a higher amount of water has
to be removed in the vacuum-dryer and the drying time increases. But the dispersion
of the fruit juice can only be achieved if there is a certain percentage of solid matter
in the raw material. The other limiting factor is the percentage of sugar in the raw
material. With increasing amount of sugar the problems with drying this product
 905
increase because of the bonding of water with sugar and it is difficult to remove the
whole water.

Until now powdery products were produced out of sieved tomatoes, apple
purre, orange pulp and liver. Commercial products (apple, orange and tomatoes) were
used as raw material, which have the disadvantage of a high percentage of sugar and
also preservatives are added. The produced powder was free-flowing and there were
no problems to redissolve the powder in water. Further the resolved powder shows no
loss of flavor in comparison to the raw material.

3.) Aim of the project

The next step of the project has to be the installation of a continous plant.
Figure 3 shows a flow sheet of such a plant. The fruit pulp is pumped with a viscous
material pump through a nozzle into the crystallizer. The necessary CO2 for refrigera-
ting capacity is compressed after the expansion with the help of a compressor to 60
bar. After passing the condenser it is liquid again and gets into the condensate tank.
e the cycle for the CO2 is closed.
P1 ... pump for fruit pulp
Further appliances for the inlet P2 ... C02 compressor
P3 ...compressor for refrigerant
P4 ... vacuum pump
and the outlet to the vacuum 1... crystallizer
2 ...vacuum dryer
dryer are necessary. The use of P1 3 ... C02 receiver
K01..C02 condenser
a heat pump is optimal, because K02 .. cooling trap
DR1 .. nozzle for refrigerant
the amount of heat for the dryer
for sublimating the water is equal
to the amount of coldness, which POWDER

is needed in the condenser. Figure 3: flow sheet for the continuous plant

The data of the discontinuously experiments cannot be transmitted directly for

a continuous plant unit. With this plant it should be possible to vary different important
parameters in dependence on the raw material and these data can then be used for
a scale up.
 PHYSIOLOGICAL ROLES OF MEMBRANE ALTERATION IN GAMMA
IRRADIATED POTATO TUBER

S.TODORIKI and T.HAYASHI

National Food Research Institute, Ministry of Agriculture, Forestry & Fisheries,
Tsukuba, Ibaraki, 305 Japan

ABSTRACT

Gamma-irradiation of potato tuber brought about the increase of electrolyte leakage

from the disks . This increase of membrane permeability began just after irradiation
and continued for a few days after irradiation, and then decreased to the original
levels. Lipids in irradiated potatoes were not significantly decreased by irradiation.
However, for a few weeks storage after irradiation, phospholipid and linolenic acid
content in galactolipid increased.

INTRODUCTION

Gamma-irradiation of potato tuber has been _ commercially utilized for the purpose
of sprout inhibition. Not only sprouting inhibition, irradiation has been reported to
bring about many other physiological changes related to membrane changes such
as increase in respiration, accumulation of sugars etc [1]. However, very little was
known about the effect of gamma -irradiation on the membrane lipid and property
of potato tuber. This study was undertaken to clarify the effect of
gamma-irradiation on lipid composition and membrane property of potato tuber in
vivo.

MATERIALS AND METHODS

Irradiation of potato tuber

Potato tuber (solanum tuberosum cv. dejima ) were obtained from a local market
in Tsukuba and irradiated with a Gamma -cell 220 (2.1Tbq of Co-60, 4.6x103Gy/h,
AECL) at a dose of 0.5kGy, unless otherwise stated, and stored at 5·C in a dark
room.

Extraction and fractionation of potato lipid

Lipid extraction was done according to the method of Bligh-Dyer after the boilingof
tissues in 2-propanol for 5 minutes. Total lipid extract were fractionated into
 907
neutral, glyco-and phospholipid fractions on a silicic acid column and lipid component
in glyco and phospholipid were separated by thin layer chromatography.

Lipid analysis
Fatty· acid methyl esters was prepared from each lipid fraction in HCI-methanol
and analyzed on a gas- chromatograph (GC 14A) equipped with capillary column
(DB 225 J & W Inc Ltd 0.25mmx30m) and an FlO. Pentadecanoic acid was used as
an internal standard.
Quantifion of free fatty acid were carried out by the fluorescence measurement
of ADAM derivative of free fatty acid after the separation by HPLC. The oxidative
lipid degradation was evaluated by measuring TBA value of the extracted lipid.

Measurement of electrolyte leakage from potato disk

Ten disks(25mm diameter , 5mm thickness) were taken from one tuber, and rinsed
with distilled water. The disks thus prepared were incubated at 27·C in 100ml of
distilled water with gentle shaking, and the electrical conductivity of the water
was measured with a Conductivity Meter ES12 (Horiba Ltd) at 15(C15) and 75 min
(C75)' The conductivity of the water incubated with frozen-thawed disks were
measured to determine total electrolyte(C t ). The rate of electrolyte leakage was
expressed by; 100x (C75 -C15 )/C t .

RESULTS AND DISCUSSION

Membrane lipid composition of Irradiated potato tuber

Total fatty acid content of potato tuber irradiated at 0.5kGy were Slightly
decreased 1 day after irradiation, but it recovered within a few days after
irradiation (table 1). There was neither significant loss of poly unsaturated fatty acid
nor accumulation of free fatty acid and lipid- hydroperoxide up to the dose of 2 kGy.
Long term storage after irradiation brought about the increase of phospholipid
content rather than the reduction. Fatty acid composition of glycolipid , especially
Monogalactocyl diglyceride (MGDG) and digalactosyl diglyceride(OGOG), were
significantly changed by irradiation. The percentage of linoleic acid decreased
accompanied by the increase of linolenic acid with increasing dose[2]

TABLE 1
Fatty acid content of ittadiated potato tuber

treatment Fatty acid (11 mo1/100g. f w)

unirradiated (Oh) 138.02
unirradiated(10d) 139.96
0.5kGy(Oh) 138.84
0.5kGy(6h) 138.94
0.5kGy(12h) 130.10
0.5kGy(24h) 114.16
0.5kGy(4Bh) 122.06
0.5kGy(72h) 136.24
0.5kGy(10d) 140.98
908
Effect of gamma -Irradiation on membrane permeability of potato tuber
The increase in membrane permeability of irradiated potato tuber measured by
increased rate of electrolyte leakage from potato disks(fig.1). Membrane permeability
increased until 2 days after irradiation with increased radiation dose then decreased
to the initial level.

5
0 0 kGy
--
:;:.g
0
4
• 0.15 kGy
0
A 1 kGy
0
T"" 3
X

--
+-'
()
l() 2
T""
()
I
l()

-
I"-
() 1

0
12345 7 21 42

Storage period (day)

Figure 1. Electrolyte leakage of potato disks.

Conclusions

Gamma irradiation brought about the increase in membrane permeability of potato

tuber immediately after irradiation. During the storage period after irradiation
lipid composition of tuber altered, and the membrane permeability which once
damaged was recovered.

References

1. Thomas, P. CRC Critical Reviews in Food Science and Nutrition,19,1984,

pp. 343-349.

2. Hayashi, T. Todoriki, S. and Nagao, A. Effect of gamma-irradiation on membrane

permeability and lipid composition of potato tubers. Env. Exp. Bot. , 1992, 32(3)
pp. 265-271.
 DEVELOPMENT OF A TIME-TEMPERATURE INDICATING DEVICE
USING PHOSPHOLIPASE

S.H. YOON, C.H. LEE, D.Y. KIM, J.W. KIM, and KH. PARK
Department of Food Science & Technology and Research Center for
New Bio-Materials in Agriculture, Seoul National University, Suwon, 441-744, Korea

ABSTRACT

As an attempt to create a time-temperature indicator (TTI) which performs

reproducible reaction with confidence in monitoring storage life of frozen foods, a
TTl using phospholipase was developed and applicability of the TTI to frozen pork
was investigated. Smaller standard deviations of the reaction rate constants for
phospholipase system over the temperature range between 30°C and - urc than those
for lipase system indicated that the former is more reliable than the latter. Good
correlation between the color change of the TTl and the TBA value of frozen pork
at fluctuating storage temperature suggested the potentiality of the TTl for
predicting the storage life of frozen foods.

INTRODUCTION

Various lipid hydrolyzing enzymes such as lipase could be used for a TTl, but
stable substrate emulsion is highly required to obtain a reproducible reaction rate of
the enzyme[1]. The emulsion of triglyceride, a substrate for lipase, has been known
to be unstable. It is practically impossible to obtain a uniform droplet size in the
emulsion and the emulsion cannot be stored over very long periods, especially at low
temperatures. Therefore, the device using unstable emulsion system is not reliable
for predicting enzymatic reaction to an acceptable degree of accuracy and
reproducibility. On the other hand, phospholipid retains both hydrophilic and
lipophilic portions in the molecule and is capable of maintaining a stable emulsion
system. Commercial preparations of phospholipids derived from soy bean oil are
used extensively in manufactured foods.
In present study, we report development of a new type of TTl using
phospholipid-phospholipase system. The device was evaluated for the reproducibility
of the enzyme reaction and the application of the system for predicting frozen pork
quality change along the storage time and temperature fluctuation is also discussed.
910
MATERIALS AND METHODS

Preparation of'ITIs.
Two types of TTl were prepared; one based on lipase-triglyceride system and the
other phospholipase-phospholipid system. The TTIs were composed of an enzyme,
substrate emulsion. salts, antifreezing agents, pH indicators, and emulsion stabilizer.
The substrate emulsion for lipase was prepared by sonicating the mixture of 1.25 m1
olive oil and 23.75 m1 of 10% (w/v) gum arabic solution for 1.5 minutes (Fisher,
Desmembrator Model 300). The substrate emulsion for phospholipase was prepared
by sonicating 25 m1 of 3% (w/v) soya lecithin for 3 minutes. In order to make the
most appropriate pH indicator mixture for the TTI, bromothymol blue, neutral red,
and methyl red were mixed at the ratio of 8 : 2 : 0.1 (v/v).
Reaction rate of lipid hydrolysis.
The hydrolysis reaction was initiated by adcling a predetermined amount of each
enzyme into the reaction solution of pH 8.0. The amount of free fatty acid liberated
by enzymatic hydrolysis was determined by pH-stat method. One unit of enzyme
activity was defined as the amount of enzyme which produces 1 f1. M of free fatty
acid per minute. Production of free fatty acid by the enzymes followed the pattern
of first-order reaction and the reaction could be expressed as the following equation:

where [A] is the concentration of substrate, Ira is the reaction rate constant, and t is
the reaction time.

RESULTS AND DISCUSSION

The reaction rate constant and reproducibility of phospholipase reaction

system.
The pH change in each TTI was monitored over the temperature between 30°C and
-lSOC. The pH of the reaction at temperatures between O°C and the above dropped
sharply during the first hour of the reaction and then slowed down significantly
(data not shown). It could be due to the limiting amount of substrate in the
reaction. The reaction rate constant, Ira, was calculated for each reaction
temperature by assuming that it is a first order reaction. The reaction rate constant
was largely dependent on the reaction temperature (Table 1) and were used to
predict the time for color change of the TTl Standard deviations of the initial
reaction rate constant for lipase and phospholipase with 95% confidence were
determined from the experiments that were repeated 6 times at each temperature by
pH-stat test (Table 1).
From the result, we concluded that reproducibility of the reaction system of
phospholipase was better than that of lipase, and it was more agreeable to employ
phospholipase syste than to employ lipase system for monitoring food quality change
during storage at very low temperature.
 911
Table 1.
The reaction constants of lipid hydrolysis reactions by two lipases
at various temperatures.

Phospholipase Lipase
Temp(OC) k O/min) SD k (l/min) SD
30 0.0012919 0.0001459 0.00195287 0.000228
20 0.0006749 7. 942E-05 0.00134049 0.000282
10 0.0002048 3. 875E-05 0.00079098 0.000210
0 3.706E-05 8. 561E-06 0.00031010 0.000116
-5 1. 498E-05 4.081E-06 0.00024044 5. 79E-05
-10 2. 416E-06 4. 528E-07 2. 3691E-05 6. 58E-06
-15 8. 772E-07 1. 769E-07 7. 6187E-07 1. 56E-07
-18 5. 524E-08 1. 779E-08 9. 5927E-08 1. 2E-08

* Standard Deviation
Correlation between the TTl and quality change of frozen pork.
The degree of color change of the TTl was divided into three ranges; 1) between
pH 8.0 and 7.5 usable, ii) between pH 7.5 and 7.0 uncertain, iii) between pH 7.0 and
6.5 unusable ranges. The TBA value of frozen pork during storage was also
divided into three ranges; i) 0.0 and 0.5 usable, ii) 0.5 to 1.0 uncertain, and iii) 1.0 to
1.5 unusable ranges[2]. The amount of the enzyme in the TTl was adjusted so that
the pH of the reaction mixture reaches 7.5, 7.0, and 6.5 when the TBA value of
frozen pork became 0.5, 1.0, and 1.5, respectively, depending on the storage time and
temperature. Thereby, the storage life of the frozen pork could be estimated from
the color of the TTl. Both the changes of the TTl color and the TBA values of
frozen pork were in good correlation with the predicted data. (data not shown).
Therefore, we concluded that the TTl using phospholipid-phospholipase system is
reliable for monitoring the quality deterioration of frozen pork.

CONCLUSIONS

A new type of TTl using phospholipid-phospholipase system was developed to

monitor quality loss of frozen foods during storage. It was determined that the
phospholipase system carried out more reliable reaction than the lipase system at all
the temperatures tested. The pH of the TTl and the TBA value of frozen pork
were diveded into three ranges to indicate usability of frozen pork. The color change
of the TTI along the pH change correlated well with the change in TBA value of
frozen pork. Therefore, it was concluded that the TTl developed in this study
could be applied to other frozen foods.

REFERENCES

1. Taoukis, P. S., Fu, B., and Labuza T. P., Time-temperature indicators. Food
Tech., 1991, 45, 70-82.
2. Truner, E.W.,Paynter,W.D.,Montie, E.J.,Bessert, J.M., Struck, G.M., and Olson, F.C.,
Use of the 2-thiobarbituric acid reagent to measure rancidity in frozen pork. Food
Tech., 1954, 8, 326-330.
 HEAT TREATMENT EXPERIMENTS ON HAY TO ELIMINATE
POSSmLE CONTAMINATION WITH HESSIAN FLY

SHAHAB SOKHANSANJ AND H.C. WOOD

College of Engineering, University of Saskatcnewan
Saskatoon, Saskatchewan, S7N OWOCanada

ABSTRACT
Alfalfa chops were exposed to hot air in a rotary machine to demonstrate that the heat
treatment kills the insect Hessian fly [Mayetiola destructor (Say)]. The laboratory results
showed that an exposure to 60 C for three minutes results in the total destruction of the
pupae stage of the insect. Tests on a full scale rotary dryer showed that the chopped hay
being dried in these machines reaches to the critical temperature of 60 C and remains at
these temperatures for at least three minutes, and these temperature and time meet the
kill conditions.

INTRODUCTION

Japan has permitted imports of heat treated chopped hay from Canada on the basis that
the product temperature reaches 90 C for at least three minutes during artificial drying
(Sokhansanj et al, 1990, Sokbansanj and Wood 1991). This high temperature was
specified because experimental thermal kill data for Hessian fly [Mayetiola destructor
(Say)] was not available at the time.

Experiments were conducted on a full number of insects but in a smaller rotary drum
(Sokhansanj et al., 1993). About 48000 insects were heat treated in the drum and about
16000 insects were used as control. The results showed that a temperature of 58 C is
critical for the survival of the insects. A theoretical analysis showed that the stem
temperature of the infested plants during dehydration in a production dryer substantially
surpasses the critical temperature of 58 C.

A series of tests were conducted on a production rotary drum to show that

temperatures inside the drum can be measured experimentally; and the internal
temperatures can be related to the exhaust temperature. This paper describes the dryer,
method of testing and test results.
 913
DESCRIPTION OF THE DRYER AND TEST PROCEDURES
The hay chop dryer was a rotary drum as shown schematically in Figure 1. The drum is
6.22 m long and ·1.8 m in diameter and the body of the dryer is insulated. The dryer
rotates at about 7 r/min. The source of heat is direct fired natural gas. Air flow through
the dryer is about 500 m 3/min. The throughput of the dryer varies from 3 tonne per hour
to 6 tonne per hour.

Chops out
Chops
8 0
0
7 -.- AIr
0
6 4 Chops
0 0 0 o
1 2 ~ 4 5

Figure 1. Schematic of the dryer showing the locations of the thermocouples

Instrumentation
Five thermocouples were inserted into the drum such that the tip of each thermocouple
was inside the dryer, about 10 cm away from the wall. Hard Teflon tubing was used to
hold each thermocouple in an upright position. Figure 1 shows the location of these
thermocouples along the dryer. Two thermocouples were inserted into the inlet of the
exhaust duct; one measured dry bulb temperature (thermocouple # 6) and the second one
measured the wet bulb temperature (thermocouple #7). For wet bulb temperature, the
junction of the thermocouple was wrapped in a wetted felt cloth. One thermocouple
measured the outside ambient temperature close to the drum wall. Thermocouple wires
were connected to a Campbell Scientific data logger which had been installed and secured
on the body of the dryer drum. The data logger was pre-programmed to scan each channel
every 10 seconds and store the acquired data on a data storage chip.

Procedure
Three tests were conducted on September 17, 1992. Prior to the tests, the dryer had been
running normally. For preparation for the measurements, the dryer was stopped,
thermocouples were installed, the data logger was initiated and the dryer was re-started to
operate normally. The exhaust temperature was set to a desired value and this temperature
was maintained automatically by gas flow rate to the burners. The exhaust temperature
was set at 71 C for about 25 minutes (test A), followed by dropping the exhaust
temperature to 61 C for 20 minutes (Test B), and finally increasing the temperature to 66
C for about 15 minutes (Test C).
During the tests, samples of hay chops from the inlet and from the end of the cooler
drum were collected.
914
RESULTS

Temperatures - Figure 2 shows the plot of six dry bulb temperatures (thermocouples 1-6
shown on Figure 1). The three test durations are designated test A (exhaust temperature at
71 C), test B(exhaust temperature at 61 C), and test C (exhaust temperature at 66 C).
Thermocouple 6, marked by heavy line, shows the dry bulb temperature of the air at the
exhaust.

OOr-------------------------------~

~+-~Hr----------r_------_r----~~

OM'r-~~----------+--------r----~~

1~r-~~~------+-------+-----4-~
ti
~ 70
~~+--4~~~~~~~~~~~--~~r-4

~ OO+-~Ir-~-----+~~~~--~~~~~

50~~~~~~~~~~~~~~~~~
o 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 M
DURATION OF TEST, minutes

Figure 2. Temperatures in the drum during drying of chopped hay at three settings of the
exhaust air temperature: duration A at 71 C, duration B at 61 C, and duration C at 66 C

The variation of the exhaust temperature in each setting, is about ±2 C.

Thermocouples 4 and 5 registered low temperatures as these thermocouples sensed the
temperature of the fresh chops entering the dryer. The largest temperature sensed was that
by thermocouple 3 that was located at about the middle of the drum length. The average
moisture content of the fresh chops during the test was 14.8% and the average dry chop
moisture content was about 8.6%. The production throughput was about 5 tonnes per hour
during the experiment. The relative humidity of air measured at the outlet of the dryer (6
and 7 on Figure 1) was 15% during test A (71 C), 45% during test B (61 C) and 25%
during test C (66 C).

REFERENCES
Sokhansanj, S., H.C. Wood, J.W. Whistlecraft, and G.A. Koivisto. 1993. Thermal
disinfestation of hay to eliminate possible contamination with Hessian fly. Postharvest
Biol0I:Y and Technol0I:Y (Feb 10 1993).
Sokhansanj, S., V.S. Venkatesan, H.C. Wood, IF. Doane, and D.T. Spurr. 1992. Thermal
kill of wheat midge and Hessian fly (Diptera: Cecidomyiidae). Postharvest Biolof:Y and
Technol0I:Y 2(1992);65-71
Sokhansanj, S., and H.C. Wood. 1991. Simulation of thermal and disinfestation
characteristics of a bale dryer. Dryinl: TechnolQI:Y - an International Journal 9(3):643-
656
 BACTERIOCIN PRODUCING LACTIC ACID BACTERIA
USED FOR BIOPRESERVATION OF FOOD

Birthe Jelle & Nicolai Peitersen*

Chr. Hansen's Laboratorium Danmark AIS, 10-12 Bege AIle, P.O.Box 407, DK-2970
Hersholm, Denmark
* Chr. Hansen's Laboratorium Danmark AIS, Siber Hegner Service Building Isogo,
Nihon Siber Hegner K.K., IMD. 6-1 Shin Isogo-Cho, Isogo-Ku, Yokohama 235, Japan.

INTRODUCTION

Lactic acid bacteria are able to inhibit other microorganisms by producing organic acids
hydrogen peroxide and sometimes other more specifically inhibitory substances like
bacteriocins. Bacteriocins are proteins or protein complexes with bactericidal activity
against organisms usually closely related to the producer organism.
Nisin, produced by Lactococcus lactis, is a polycyclic peptide, a so-called lantibiotic (Fig.
1). It is the best known and most extensively studied antimicrobial which tends to have a
wider spectrum of activity than the majority of other bacteriocins produced by lactic acid
bacteria. It inhibits the growth of several gram positive bacteria including the outgrowth
of Bacillus and Clostridium spores.

ABA =Amino Butyric Acid

DHA =Dehydroaianine DHB =Dehydrobutyrine 1.8 -Methyldehydroaianinel
ALA- S- ALA = Lanthionine ABA- S- ALA = /3 -Methyllanthionine COOH

Figure 1. Structure of Nisin (from Gross and Morell, 1971 (1»

916
Numerous studies have been made on the mode of action of both susceptible vegetative
cells and spores (2-5). Nisin is believed to affect the germination stage by inhibiting the
swelling process (2,6). The effect against the spore formers is very important, especially
for the dairy industry, where it has proved to be a most effective preservative in processed
cheese spreads in preventing spoilage caused by Clostridium tyrobutyricum.

The use of nisin to prevent spore formers during production of hard and semi-hard cheese
has been reviewed by Hurst (7). Experiments with Gouda cheese have indicated that
approximately 40IU/g (lppm) of nisin is sufficient to prevent the butyric acid fermentation,
even at a relatively high number of clostridial spores (6/ml) in the cheese milk (8).

APPLICATION OF NISIN PRODUCING LACTOCOCCI

A number of nisin producing strains of Lactococcus lactis subsp. lactis has been screened
at our laboratory for acidification and nisin production in reconstituted skim-milk. In
general, they acidify very slowly due to limited proteolytic activity, which also limits the
production of nisin. Addition of casein peptone (0.5 %) to the milk results in a considerable
increase in both the acidification rate and the nisin production (from 100 IU/ml to 500
IU/ml) during incubation for 18 hours at 30°C. These results are particularly of interest
if the nisin producer can be applied as a bulk starter in the cheese manufacturing process.

Laboratory experiments combining a nisin producer and a thermophilic starter normally

used in cheese production have been carried out in skim-milk using the procedure
(time/temperature profile) for hard cheese processing, but without addition of rennet.
Results from these experiments indicate that the nisin producer shall be applied as a bulk
starter. When the nisin producer is added directly into the cheese milk (inoculum 107
CFU/ml) in the preripening step the growth is limited and nisin is not produced in
detectable quantities during the process. The thermophilic starter added at 30°C (inoculum
approx. 107 CFU/ml) causes a so fast acidification that the nisin producer is inhibited.

Application of the nisin producer as a bulk starter together with the thermophilic starter
after the preripening step, gives an initial nisin concentration in milk of approx. 16IU/ml.
In this experiment the thermophilic starter is inhibited by the nisin and nearly no
acidification occurs during the process. It was therefore necessary to adapt the
thermophilic starter to higher concentrations of nisin. This was done by repeated transfer
into milk containing increasing amounts of nisin. Eventually, the adapted strain was able
to grow in and acidify milk even in the presence of 150 IU/ml nisin. The experiment was
repeated using the nisin producer together with the nisin resistent strain. Then the
acidification rate increased and the end pH in the milk (after approx. 30 hrs) is nearly the
same as for thermophilic starter applied alone, but the lag phase is extended by approx.
7 hrs. The growth of both organisms in the milk increased during the cooling process and
further, nisin is produced which result in a final nisin concentration in the milk of 24
IU/ml.

A nisin concentration of 24 IU/ml is not sufficient to inhibit the germination of clostridial

spores. However, all our experiments have been carried out in milk without the renneting
process in which a 10 times concentration of the initial nisin content (16 IU/ml) in the milk
 917
can be expected. It is thus concluded that the strains developed may be applicable for
making hard cheese.

ACKNOWLEDGEMENTS

The authors wish to thank Dorte N0rgaard and Karina Jensen, two dairy scientists, who
made most of the practical work as part of their Dairy Engineering thesis.

REFERENCES

1. Gross, E. and Morell, I.L., The structure of nisin. J. Am. Chern. Soc., 1971,93,
4634-5.

2. Campbell, L.L. and Sniff, E.E., The effect of subtilin and nisin on the spores of
Bacillus coagulans. J. Bacteriol., 1959, 77, 766-70.

3. Ramseier, H.R., Die Wirkung von Nisin auf Clostridium butyricum prazm. Archiv
fUr Mikrobiologie, 1960, 37, 57-94.

4. Morris, S.L., Walsh, R.C. and Hansen, J.D., Identification and characterization of
some bacterial membrane sulphydryl groups which are targets of bacteriostatic and
antibiotic action. J. BioI. Chern., 1984,259, 13590-4.

5. Henning, S., Metz, R. and Hammes, W.P., Studies on mode of action of nisin. Int.
I. Food Microbiol., 1986,3, 121-34.

6. Lipinska, E., Nisin and its applications. In: Antibiotics and antibiosis in agriculture.
ed. M. Woodbrine. Butterworths, London. 1977, pp. 103-30.

7. Hurst, A., Nisin. Adv. Annl. Microbiol., 1981,27, 85-123.

8. Hugenholtz, J. and deVeer, G.J.C.M., Application of Nisin A and Nisin Z in Dairy

Technology. In: Nisin and Novel antibiotics. Ed. G. Jung and H.-G. Sahl. ESCOM
Science Publishers, The Netherlands, 1991, pp. 440-56.
 STUDIES ON THE CONTROL OF THE GROWTH OF
SACCHAROMYCES CEREVISIAE BY USING RESPONSE SURFACE METHODOLOGY
TO ACHIEVE EFFECTIVE PRESERVATION AT HIGH WATER ACTIVITIES

AN-ERL KING
Dept. of Food Science, National Chung-Hsing Univ.
Taichung, Taiwan 40227, ROC

ABSTRACT

water activity (0.93-0.97), pH (4-6), and sorbic acid

concentration (0-50 ppm) were combined to study their effects
on the growth of Saccharomyces cerevisiae. A three-variable
and three-level design method, analyzed by response surface
methodology (RSM), was used. The growth could be inhibited at
water activity 0.94 and pH 4 with a sorbic acid concentration
of 25 ppm. Predicted conditions of inhibition and
experimental values were found to be consistent.

INTRODUCTION

Saccharomyces yeasts ferment sugars and often cause the

spoilage of fruits and fruit products [1]. The water activity
of fruit products might be lowered to avoid their spoilage
[2]. However, flavors and characteristics are always affected
appreciably by the reduction of water activity [3].
Preservation under higher water activities is therefore to be
desired. Microbial inhibition might be effectively achieved
at higher water activities using the "hurdle effect", in which
several food preservation processes are applied simultaneously
[4]. In this study, preservation techniques of water activity
control and pH adjustment were combined with the addition of
sorbic acid [5] to effectively inhibit the growth of
Saccharomyces cerevisiae under higher water activities.
Response surface methodology was used to study the effects of
these three factors on the responses of the growth parameters,
i.e. lag time, growth rate, and maximum population, and a
three-variable and three-level design method was adopted.
 919
MATERIALS AND METHODS

The culture strain obtained from the culture collection was

maintained on Sabouraud agar slants as stock culture.
Inoculated media were incubated in a rotary shaker at 27±1°C
for the experiment. In the main experiment, media consisting
of 1% peptone solution were used. water activity, pH, and
sorbic acid concentration were adjusted to the desired values
according to the RSM experimental design. A three variables-
three levels design was used [6]. Table 1 lists the real
experimental values of various factors (Xi) adopted at
different levels.

TABLE 1
Process variables and their levels in the three variables-
three levels response surface design

Independent Symbol Levels

variables

Coded* Uncoded Coded* Uncoded

water activity X1 aw 1 0.97

0 0.95
-1 0.93
pH value X2 pH 1 6
0 5
-1 4
Sorbic acid (ppm) X3 HS 1 50
0 25
-1 0

uncoded value - 0.5x(high value+low value)

*coded variable =
0.5x(high value-low value)

Lag time, growth rate in log phase, and maximum

population in stationary phase of S. cerevisiae grown in
various media were selected as responses. RSREG procedure in
SAS software was used to conduct the RSM analysis. A second-
order polynomial expression of three variables was fitted.
Optimization was performed graphically.

RESULTS AND DISCUSSION

All the model regressions were found to be significant and

lack-of-fits were not. This indicated that these fitted
models could represent responses appropriately. Regression
coefficients obtained from SAS analysis in the second-order
920
polynomial were not all significant at a 5% level. Non-
significant coefficients were deleted when graphing was
conducted. On the other hand, since pH was less significant
on the growth rate and maximum population compared to the
other two factors, water activity and sorbic acid
concentration were selected for plotting and pH was set to be
constant at 4, 5, and 6, respectively. In the contour plot
with pH=4, a region where the reciprocal of lag time fell
below zero could be located. Meanwhile, the yeast could not
grow at pH 4 with a water activity of 0.93 and a sorbic acid
concentration of 25 ppm, and the growth rate was zero. When
superimposing the contour plots of pH 4, the regions predicted
for yeast inhibition were found not to overlap completely.
This might be due to the errors caused in the experiments.
For the reduction of errors, the overlapped region was adopted
as the region for growth inhibition. Verification was
conducted using a point in the region. Point of X1 = -0.8,
X2 = -1, and X3 = 0 was used for the test. The experimental
result and the estimated value coincided and the model was
proved to be adequate.

REFERENCES

1. Potter, N.N., Food Science, AVI Publishing Co., Westport,

1986, pp. 140-~

2. Levi, A., Gagel, S. and Juven, B.J., Intermediate-moisture

tropical fruit products for developing countries. II.
Quality characteristics of papaya. ~ Food Technol., 1985,
20, 163-75.

3. Chirife, J., Ferro Fontan, C. and Benmergui, E.A., The

prediction of water activity in aqueous solutions in
connection with intermediate moisture foods. IV. A
prediction in non-electrolyte solutions. J. Food Technol.,
1980, 15, 59-70. -- ----

4. Leistner, L., Rodel, W. and Krispeien, K., Microbiology of

meat and meat product in high and intermediate moisture
range. In Water Activity: Influence on Food Quality, ed.
L.B. Rockland and G.F. Stewart, Academic Press, New York,
1981, pp. 855-916.

5. Cerrutti, P., Alzamora, S.M. and Chirife, J., A multi-

parameter approach to control the growth of Saccharomyces
cerevisiae in laboratory media. ~ Food Sci., 1990, 55,
873-40.

6. Box, G.E.P. and Behnken, D.W., Some new three level designs
for the study of quantitative variables. Technometrics,
1960, 2, 455-475.
 THEORY AND APPLICATIONS OF A NEW VISCOMETER
BASED ON ANNULUS LIQUID FLOW

KANICHI SUZUKI
Department of Food Science,
Faculty of Applied Biological Science, Hiroshima University,
1-4-4 Kagamiyama, Higashi-Hiroshima, 724 Japan

ABSTRACT

A new viscometer was developed by applying the flow theory of fluid in an annular
channel. Pressure drop and shear stress were expressed as a function of flow rate,
viscosity, and the dimensions of the annular channel. Both of direct measurement of wall
shear force and pressure drop measurement were acceptable for evaluating viscosity with
known flow rate. The viscometer measured wide range of viscosity with high accuracy.
Non-Newtonian flow properties were also measurable. It was available for an on-line
viscometer for liquid food processes.

INTRODUCfION

Liquid food processings need an on-line viscometer that measures wide range of flow
properties accurately for quality and process controls. The viscometer also requires as
simple structure as possible for sanitation. However, most conventional on-line
viscometers have complicated structure, and require a bypass from the pipelines to
introduce liquid foods for viscosity measurement. Thus, I have developed a new
viscometer by applying the flow theory of fluid in an annular channel. This study
presented the theory, structure and measured results of the new viscometer.

THEORY AND STRUCTURES

When a fluid flows through an annular channel between two coaxial circular cylinders
of radii Rand IC R (0 < IC < 1 XFig.l, (a», the relationship between pressure depression
.6.P for length L and flow rate Q in laminar flow region is expressed as follows!),
AP =8 !.I. LQ I:rtR4[(1 - K4) - {(1 - K2)2/1n(lIK)}] (1)
922
where !! is viscosity of the fluid. The shear stress acting on the surface of inner cylinder
'ti is also a function of .6.P as,
'tj =AP R [K - {(l - K2) / (2K In(lIK)}] / (2L) (2)
The shear force exerted by the fluid on the surface of inner cylinder F
F =- 2 :n; KR L 'tj (3)
By combining equations (1), (2), and (3), F is expressed as a function of Q, JA., and
dimensions of the annular channel. These relations suggest that two types of viscometer
are possible if one measures f:1 P or F for length L, and correlated with Q. Schematic
illustrations of the two viscometers constructed in this study are shown in Fig.l (b), (c).
I.Horizontal annular channel viscometer; a liquid flows through a horizontal
cylindrical annulus. A differential pressure manometer measures pressure depression for a
fixed cylinder length. Viscosity of the liquid can be evaluated from Eq.(I). This
viscometer can be used as a perpendicular viscometer, too.
2.Perpendicular annular channel viscometer; a load sensor measures the shear force F
on the surface of perpendicular inner cylinder of length L. The load sensor measures also
line pressure of fluid and the difference of forces acting on the upper and lower surface
of the inner cylinder. The former is measured by another load sensor, and subtracted. The
latter value is theoretically correlated with F as follows,
:n; (KR)2 AP = -F / [ 1 - {(l - K2) / (2K2 In(lIK)}] (4)
The true value of F can be evaluated by correcting the measured force value using
Eq.(4). The values of F, Q and the dimensions of the annular channel give the viscosity.
For non-Newtonian fluids, the relationship between the shear stress and apparent shear
rate calculated from the measured apparent viscosity and the shear stress gives the flow
behavior.

(a) (b) (c)

load sensor

differential pressure

Fig. 1 Schematic illustration of fluid flow in an annular channel (a), and two types of
viscometers examined( (b), (c».

RESULTS AND DISCUSSION

Both viscometers measured viscosities of low viscous liquids (e.g., water, sugar solutions)
 923
accurately, as well as of high viscous liquid foods (e.g., tomato puree, refinery molasses)
by adjusting the radius of inner cylinder and/or sensitivity of load sensor or manometer
(Fig.2). For non-Newtonian liquid foods, both viscometers measured apparent viscosities
at different flow rates or shear rates. Simple structure and short length of these
viscometers ensure easy installation on any process pipelines without large pressure
drops, and real-time measurement of viscosity is possible. Computer monitoring of
change in viscosity is also possible. The viscometers are suitable for laboratory use, too.

..... x 10 3
0.6 &! 10
b. 60% Sucrose soln. • 59% Sucrose soln.
A 52% Sucrose 8
Z 0.5 0 57% glycerine
~ • water
<I 6
~
0.4
•water (a)
rlIil
U
4 (b)
I

Z 2

e
~
t,) 0
0.3 10 20 -6
~ x10
~

~ 0.2
8 108
rlIil
~ c:.: 6 (c)
;J
~
== 0.1 4
•A
~ ~
Sauce A
~ 2
0
Sauce B
2 3 4 -5 1 2 3 -6
x10 x10
VOLUMETRIC FLOW RATE Q (m 3 Is)

Fig. 2 Comparison of experimental results of flow rate Q vs. shear force F for the
perpendicular viscometer (a), and Q vs. pressure difference L:::..P for the horizontal one
(b) with theoretical values (solid lines), and examples of non-Newtonian liquid foods (c).

CONCLUSIONS

1\vo new viscometers were developed by applying the fluid flow theory in an annular
channel. The viscometers with simple structure measured wide range of viscosity with
high accuracy.

ACKNOWLEDGMENT

The author wish to express his sincere thanks to Mr. S. Nakai, Technical Official,
Hiroshima University, who contributed to make the experimental apparatus.

REFERENCE

1. R.B.Bird, W.E.Stewart, & E.N.Lightfoot : Transport Phenomena, pp.51-54, (John

Wiley & Sons, Inc., New York, 1960).
 AN EXTENDED HOT-WIRE METHOD FOR MONITORING FLUID VISCOSITY
AND THE ONSET OF GEL FORMATION

TOMOSHIGE HORI, YASUHIKO SHIINOKI and KENSUKE ITOH

Technical Research Institute,
Snow Brand Milk Products Co. Ltd.
1-1-2 Minamidai, Kawagoe, Saitama 350-11, Japan

ABSTRACT
The practical feasibility of a hot-wire method for monitoring the change
in viscosity and gel-setting point of fluids was critically examined on
the basis of the theoretical relationship for free-convection heat
transfer around a circular cylinder. By comparing the temperature of a
hot-wire probe placed in a gelatin solution and in whole blood, this
method is proved to be extremely effective for assessing coagulation.

INTRODUCTION
A new hot-wire technique that will detect the onset of gel formation
accurately, in line and in real time as the change in fluid viscosity
has been established since 1985 for true automation in the cheese
industry. [1] The method uses the phenomenon of free-convection heat
transfer to the surrounding fluid from a linear heat source. The rate
of convective heat transfer from a solid to a fluid is inversely
related to its viscosity, so that in the steady state, the heat source
temperature will be lowest with fluids of the least viscosity. This is
because, at equilibrium, a lower viscosity fluid possesses a thinner
boundary layer at the interface with the solid heat source. During gel
formation, the viscosity of a fluid changes, often very rapidly.
By using a hot-wire probe, we have been able to detect milk
clotting during cheese production without disturbing the coagulum. [2]
The probe placed in rennet-treated milk detected the clotting time,
which is defined as the time at the inflection point of the time versus
hot-wire temperature curve.
The objective of this study is to establish an extended hot-wire
technique with strong potential for practical application in determining
the viscosity change of a coagulating process fluid in both industry
and clinical medicine, utilizing the practical characteristics of free
convection heat-transfer around a circular cylinder placed in fluids
with a wide range of both viscosity and thermal conductivity.
 925
THEORETICAL BASIS
The following functional relationship in the equilibrium state of
laminar free-convection heat transfer provides the basis for practical
applications of the hot-wire method:
f(Nu, Gr, Pr) = 0 (1)
where Nu, Gr and Pr are the Nusselt, Grashof and Prandtl numbers,
respectively. If a fluid such as a gelatin solution consists
principally of water, and if the solid content remains constant during
the gelling process, the heat-transfer relationship can be obtained
from Equation 1 as follows:
J.I = f(tJ. es ) (2)
where J.I is the kinematic viscosity, and tJ. e s is the temperature
difference between the surface of a linear heat source and the
surrounding fluid.
The equation for conductive heat transfer in a cylindrical hot-wire
probe with a built-in linear heat source results in a practical
relationship between the surface temperature of the probe and the
built-in hot-wire temperature. [3]
(3 )

where tJ. e w is the hot-wire temperature referred to the fluid

temperature, Q is the heat generated in a unit length of the heat
source, and C is a numerical constant for the hot-wire probe. This C
value is inversely proportional to the probe's thermal conductivity,
which is practically constant with temperature.
Equations 2 and 3 assure the interchangeability of hot-wire probe
and indicate conclusively that the temperature of the built-in hot wire
can detect the onset of gel setting directly by the change in fluid
viscosity.
In addition to Equation I, the following relationships for a
vertical linear heat source are of great use for extending the
applications of the hot-wire method:[2]
Nu = (2£ /d) / {In(1 + 20 /d)} (4 )

o = (d/2) {exp (2 A / ad) - I} (5 )

where 0 is the calculated thickness of an imaginary layer Of stagnant

fluid related to the temperature profile in the boundary layer, £ and
d are respectively the length and diameter of the heat source, A is
the thermal conductivity of the fluids, and a is the heat transfer
coefficient at the surface of the heat source.
Both A and a in Equation 5 are measurable by the present hot-
wire method, whi Ie the £ /0 value calculated from the experimental
relationship for the Nusselt number of a vertical plate, Nu p , such as
Nu p = 0.638Grl/4Prl/2 (0.861 + Pr)-1/4 (6)
[4] is useful for estimating the tJ. ew behavior in a potential process
fluid.
926
MATERIALS AND METHODS
A hot-wire probe (£ =5cm, d=2mm, Q=5W/m) was placed in a 3% gelatin
solution (G-2656 from Sigma Co., volume=200cm 3 ) to obtain the fluid
temperature versus 11 () w curve, the solution being cooled at 15°C/hr.
The activated partial prothrombin time of human whole blood diluted
10 times was determined at 37~, monitoring the built-in hot-wire
temperature of a probe designed for clinical use (£ =4mm, d=0.6mm,
Q=3.8W/m) against time.
RESULTS AND DISCUSSION
Both the gel-setting temperature of the gelatin solution and the
clotting time of whole blood were determined with inherent simplicity
(Figure 1).

(a) (b)

. 30

,
~ 11 u
L.L.i L.L.i
u u
z z~
~
a::: a:::
~
L1..
L1..
10 COAGULATION
~
It 26
Ci
./
Ci
~
~
a::: a:::

~
::J
~ 9
a:::
~
0-
~
0- 22
:::a: :::a:
I=! I=!
8
21 22 23 24 25 26 a 50 150 250
FLUID TEMPERATURE, ·C TIME, SEC

Figure 1. Measurement of the gel-setting point of (a) a gelatin solution

and (b) human whole blood by the hot-wire technique.
CONCLUSIONS
The hot-wire method offers strong potential for practical application in
determining gel-setting point by monitoring the change in fluid
viscosity in both the food industry and clinical medicine.
REFERENCES
1. Hori, T., Objective measurement of the process of curd formation
during rennet treatment of milks by the hot-wire method. J. Food
Sci., 1985, 50(4), 911-917.
2. lfcflr[, T., 1]RIances in Food Engineering, ed. by Singh, P. and
Wirakartakusmah, A., CRC Press, FlorIda, 1992, pp. 475-90.
3. Hori, T. and Itoh, K., Food Hydrocolloids, ed. by Nishinari, K.,
Plenum Pub. Co., New York, 1993, In prInt.
4. Ostrach, S., An analysis of laminar free-convection flow and heat
transfer about a flat plate parallel to the direction of the
generating body force. NACA Report. 1953, 1111. 1-17.
 USE OF AN IN-LINE VISCOMETER IN THE MANUFACTURE OF SKIM MILK
POWDER

COLM O'DONNELL *, NIALL HERLIHY**, BRIAN MCKENNA ***

* Dept. of Agricultural and Food Engineering, Univ. College Dublin, Dublin 2, Ireland.
** Waterford Foods pIc, Dungarvan, Co. Waterford, Ireland.
*** Dept of Food Science, Univ. College Dublin, Dublin 4, Ireland.

ABSTRACT

Factors affecting skim milk concentrate viscosity were examined in an industrial milk powder
plant using an in-line viscometer. It was concluded that using an in-line viscometer instead of
a density meter to control the degree of skim milk concentration in an evaporator affords an
opportunity to significantly reduce steam consumption and minimise evaporator fouling by
ensuring skim milk concentrate of 100 cPs at 100/s viscosity is constantly produced.

INTRODUCTION

In the manufacture of skim milk powder (smp), falling film evaporators are used to
concentrate skim milk prior to spray drying. The specific steam consumption of a modern
multiple effect evaporator with thermal vapour recompression is 0.1 kg steam per kg water
removed, while the specific steam consumption of a two stage spray drier is between 2.0 and
2.2 kg steam per kg water removed[1]. Thus considerable reductions in steam consumption
can be achieved during smp production by maximising water removal in an evaporator prior
to spray drying. However the degree of concentration in an evaporator is limited since the
maximum recommended viscosity of skim milk concentrate (smc) in falling film evaporators
to avoid pipeline blockages and excessive fouling levels is 100 cPs at 100/s shear rate[2].
In the dairy industry, the level of skim milk concentration in evaporators is generally
determined with reference to smc density and hence a fixed total solids percentage (%TS).
However problems may arise with pumping, storage or atomisation of smc due to smc with
too high a viscosity being produced, particularly when the protein lactose ratio (PLR) is high
or smp with a high heat specification is being produced.
The experimental work reported in this paper has been undertaken to investigate the
main parameters influencing smc viscosity and to examine the feasibility of using an in-line
viscometer to control the level of skim milk concentration in an evaporator.
928
MATEWALS AND METHODS
Research trials were carried out over a two year period in an industrial smp plant which
consisted of a 6 effect falling film Laguilbarre evaporator and a Niro two stage spray drier. A
Brookfield ITIOO in-line viscometer was installed after the evaporator to measure viscosity.

RESULTS

LOW HEAT MEDIUM HEAT HIGH HEAT

Figure 1. Effect of skim milk PLR and skim milk: preheat treatment level on the viscosity
(cPs) at 100/s shear rate of smc at 49% total solids +/-0.25%

35 40 45 so 55 60 65 70
TEMPERATURE DEGREES CELCIUS

1- SO.3 %TS - 49.6 %TS - - - 47.2 %TS

Figure 2. The effect of temperature and smc %TS on the viscosity of smc with a high heat
specification and a PLR of 0.75
 929
Table 1. The effect of temperature and agitation on age-thickening of skim milk concentrate.

Smc viscosity cPs at 100ls shear rate

Time (minutes) 60°C (agitated)
o 64 63 62.5 62.5
5 66.5 66 65.1 63
15 72.5 80.2 97.2 81
30 134 153 189 148

DISCUSSION

Figure 1 illustrates that if smc %TS and PLR are kept constant, smc viscosity may more than
double with variation in preheat treatment. This rise in viscosity is caused by increased
voluminosity of whey proteins due to heat denaturation. A similarly large increase in smc
viscosity is shown to occur in figure 1 when variations in the PLR occur. This is due to the
greater voluminosity of milk proteins over an equal mass of lactose particles. Large
variations in the PLR from 0.62 to 0.91 are a characteristic of Irish milk production. Figure
1 clearly illustrates the difficulty in selecting a suitable smc density value to control the level
of skim milk concentration in an evaporator. Variations in the skim milk preheat treatment or
PLR are likely to cause the production of smc with a viscosity in excess of the maximum
recommended value, or alternatively smc is produced which is below 100 cPs at 100ls shear
rate and hence the maximum amount of water removal is not taking place, which
significantly increases the steam consumption per tonne of smp produced.
Figure 2 shows that smc viscosity reduces with increasing temperature up to 65°C.
This reduction is due to the decrease in the viscosity of the serum. An increase in the smc
concentration level is also shown to sharply increase smc viscosity.
Table 1 shows that the age thickening rate of smc increases with temperature. It is
therefore important to minimise the residence time for smc between the smc heat exchanger
and the spray drier atomiser to prevent pipeline blockages occurring.
Frequently during the two year test period the in-line viscometer was checked for
calibration against a laboratory viscometer especially near the end of long production runs.
On no occasion was evidence of internal fouling or loss of accuracy detected.

CONCLUSIONS

It can be concluded that the installation of an in-line viscometer rather than a density meter
in an evaporator to control the level of skim milk concentration affords an opportunity to
significantly reduce steam consumption and prevent excessive evaporator fouling especially
in situations where large variations in skim milk PLR or preheat treatments occur by
ensuring smc of 100 cPs at 100ls shear rate is constantly produced.

REFERENCES

1. Fergusson P.H., Developments in the evaporation and drying of dairy products. J. of the
Society of Dairy Technology, 1989, 42, pp. 94-101.

2. O'Donnell c.P., Ph.D. thesis, Univ. College Dublin, 1992.

 DEVELOPMENT OF SHEAR STRESS BASED SENSOR TO MEASURE DRYING
RATE AND ITS APPLICATION TO SNACK DRYING AUTOMATION

s.c. SHIN, I.]. YOO and J.K. CHUN

Department of Food Science and Technology, College of Agriculture and Life Sciences,
Seoul National University, Suwon, Korea

ABSTRACT

A sensor based on shear strain was developed to measure the moisture content and
volumetric changes of food chips in rotating drum dryer. The sensor was installed in pilot
scale perforated drum dryer and studied for its characteristics in snack chips dehydration.
The shear strains were correlated with the moisture content of chip and bulk volume of
food. For the application of the sensor to the automation of drying process, transducing
device and process controller were designed and fabricated.

INTRODUCTION

In snack industry rotary drier is widely used because of its advantage of the rapid and
evenly drying capability. The dryer serves to adjust the final moisture content of the snack
chip, which was produced in the first drying stage to moisture content around 20 % and
tempered to equalize the moisture concentration and textual properties throughout chip.[1]
Humidity sensor is widely used to estimate moisture content of chip and in some food
industry samples are withdrawn intermittently to analyze the moisture of food. In snack
drying environment, crust and dust produced form a film layer on the surface of sensor and
interfere the air circulation through the sensing element, and cause the inaccuracy and delay
of the response of the measurement Another difficulty faced is to transmit the data from
the sensor installed inside of the rotating drum to outside. This study was investigated to
measure the shear strain of the chip layer and correlate shear strain with the moisture
content and Volumetric changes of the material under drying operation, and aimed to apply
the sensor technology to the automation of the snack drying process.

MATERIALS AND METHODS

Snack chip: Rectangular snack chip (3 x 3 x 30 mm) was prepared from flour sheet
and then the wet chip of 40% moisture content was dried in first dryer to 19 - 20 % in a
continuous belt drier. Mter being tempered in storage room, the dry product was used as
the drying raw material in drum dryer.
 931
Drwn dryer: A pilot drum dryer of 1/ 90 scale downed in capacity is consisted of a
cylindrical drum (70 cm diameter, 50 em length). The perforated drum shell has 3 mm holes
to pass air stream. Hot air of 75 C was blown into the bottom of the rotating drum(4 rpm)
in cross-flow[l, 2].
Shear strain sensor: A dual beam spring element type strain sensor was fabricated
with steel alloy sheet which was used in lumber saw. Fig. 1 shows the strain transducer
mounted with strain gages(KFC-5-CI-II) to form full bridged circuit. The output signal
was processed to meet 0 - 5 volt DC while load was applied in pilot dryer.

aluminium bar

strain gauge

drum shell

Figure 1. Schematic diagram of drum of snack dryer and strain sensor

Data processing and controller: Strain developed from moving bed of chip upon
rotation of drum was measured with the strain sensor installed at inner surface of the drum
shell. Data were sent to the process controller through a data transmitting device
implemented at the end of the drum shaft. The device had 4 stationary carbon brush units
to collect signals from the rotating copper rings which were soldered to lead-cable of the
sensor. The information about the process variables were sent to the dryer controller, which
was specially designed with Intel MCS8052AH as CPU to perform the necessary
mathematical analysis and controI[3].

RESULTS

The radial velocity profile of chips in opposite direction to the revolving shell of drum can
produce shear stress gradient between the circular layers of chip.
For a given material recipe of
snack pallet (flour, potato
starch etc ), strain may be
described as;
1.2

t
't=q'(MNS)
j
where M is moisture content
o. of material, N is rpm of drum
and S is surface
characteristics of chip.
~ 0.6

o 40 ,., 120 160 200

lDcat i on of sensor (degree)

Figure 2. Strain curve obtained with snack chip loaded in drum.

932
The strain values obtained by the sensor were useful data to estimate the moisture content
of drying material since surface properties closely relating to shear stress was affected by
the moisture content A sinusoidal strain curve as shown in Fig. 2 illustrates that the
height and width of curve were influenced by the weight and volume of the material in the
rotating drum.
From the measured strain of the chip having different moisture contents at a given working
volumes, the relationship between moisture content (M %) and strain (-em" shear stress at
peak of curve, kg) was described as the following linear straight line equation:

M= -42.2 + 0.79 'tm (2)

And working volume (m3 ) was also represented with the strain by the following second
degree power equation(3):

v = 19.524 + 26hm + fr. m2 (3)

or the volume also was estimated by arc length which was measured by the sensor. The
coefficient of the above equations were experimentally detennined for the given operation
conditions.[41
Drying curve: For the application of automation of snack drying, operation program.
was prepared to perform the measurement of the strain and convert into moisture content,
and was resided in ROM of the controller. The process data such as operation time,
temperature, moisture content and
volume of material were processed
in the controller for the local
control of dryer or sent
to the microcomputer.
1.6
Fig. 3 shows the snack drying
curve constructed by the sensor
j 1.5 and clearly illustrates the
feasibility of the practical
i
.£
application in the drying
1.4 automation system.

~ 1.3

1.2

o z 6 8 10
Tiae(hr)

Figure 3. Snack drying curve obtained with the strain sensor and the drying controller

REFERENCES

1. Shin, S.C., Development of shear stress sensor for the measurement of food drying rate
and its application to snack drying process under the microcontroller based automation
system, MS thesis, Seoul National Univ., Korea(l989)
2. Yoo, LJ, Automation element of snock drum drying process, MS thesis, Seoul National
Univ., Korea(l99l)
3. Intel Co, BASIC-52 User's Manual(l983)
4. Shin, S.C., Yoo I.J. and Chun, J.K, Development of shear stress sensor for the
measurement of food. drying rate, submitted to Korean ]. Food Sci.Tec1uwL(l993)
 MEASUREMENT OF FLUID THERMAL CONDUCTIVITY
WITH A STEADY STATE HOT WIRE METHOD

YASUHIKO SHIINOKI, TOMOSHIGE HORI and KENSUKE ITOH

Technical Research Institute of Snow Brand Milk Products Co., Ltd.,
1-1-2 Minamidai, Kawagoe, Saitama 350, Japan

ABSTRACT

Thermal conductivity measurement apparatus was developed based on the steady state
hot wire method. The measurement cell was made up of coaxial stainless cylinders and a
stainless sheathed hot probe. The results of numerical analysis suggested that the effect
of convection heat transfer was negligible small when the clearance between the outer
cylinder and the hot probe was less than 0.75mm. The average deviation of the mea-
sured thermal conductivity from known values was about 2 %. Thermal conductivity of
skim milk, W /0 and O/W emulsions were measured with the apparatus. The results
indicated that this method is applicable to the measurement of each total solid content
of skim milk, water or oil content of emulsions.

INTRODUCTION
Thermal conductivity of foods depends on their structure and chemical composition.
Many measurement techniques have been discussed by Mohsenin [1], and others. The
transient hot wire method has been recommended for most food applications because of
its simplicity [2]. However, this method has several drawbacks in applying to a practical
process. We have developed the thermal conductivity measurement apparatus, based on
steady state hot wire method, for the practical process control. The principle of this
method has already been established and the measurement apparatus has been devel-
oped [3], but that apparatus was not applicable to on-line measurement. The purpose of
this research is to clarify the heat transfer in the measurement cell and to show that the
apparatus presented in this paper can be used as a process sensor in food manufacture
processes.

MATERIALS AND METHODS

Numerical analysis
The temperature and velocity profile in the measurement cell were calculated with
934
STREAM ver. 2.6 (Software Cradle Co., Ltd.) numerical analysis program which is
based on finite volume method.

Thermal conductivity measurement

Figure 1 shows diagram of the measurement cell. Temperature of the heating resistor
was directly calculated from the electrical resistance of itself. The length of the heating
resistor was 30mm and the hot probe diameter was 2.5mm. Temperature difference
between the heating resistor and the thermostated fluid (.6.0w) was measured using 16
liquid samples whose thermal conductivity is known. Heat amount of generated from
hot probe was kept constant and the jacket was maintained at 25°C through the mea-
surement.
Applications in food production process
Total solid content of skim milk was measured with the apparatus. Skim milk were pre-
pared by dissolving four commercial skim milk powders. Water or oil content of emulsion
was measured.

RESULTS AND DISCUSSION

Thermal conductivity measurement of the standard sample
As a result of numerical analysis, the calculated isotherms were parallel to the hot probe,
and the velocity by natural convection was negligible small when the clearance between
the outer cylinder and the hot probe was below O.75mm. The clearance between outer
and inner cylinder was set to this value . .6.0w was measured with the apparatus using
liquid samples. The relationship between .6.0w and the reciprocal thermal conductivity
was confirmed to be linear as shown in Figure 2. The average deviation of estimated
thermal conductivity was within 2 % from known values. The computed temperature
difference .6.Bw was in good agreement with the experimental results.

Water
~,Out
~

Water
In Heating
Resistor

Sample

.hgure :L: TypIcal relatIOnshIp between

Figure 1: Diagram of measurement cell
.6.Bwand t
 935
Applications to food production process
Total solid content of skim milk: Figure 3 shows the relationship between the total
solid content of skim milk and the calculated value from 6.(}w. The correlation coefficient
was 0.998.
Water or oil content of emulsion: 6.(}w values on W /0 and O/W emulsions were mea-
sured and the thermal conductivity was calculated. Figure 4 shows the dependence on
<Pd (volume fraction of dispersed phase) of Ad/Ac (thermal conductivity ratio of dispersed
phase and continuous phase). The thermal conductivity of emulsions was in good agree-
ment with Maxwell Eucken's equation [4] (solid lines in figure 4).

a
N

u a
....;
--<
......
-0
--<
'"d

a
d
o 5 10 15 20 25 30
0.0 0.1 0.2 0.3 0.4 0.5
Total solid content [%] pd[ 1 I

Figure 3: Relationship between total

Figure 4: Dependence on <Pd of Ad/ Ac
solid content of skim milk and calcu-
(0 :W /0, 0 :O/W)
lated values

CONCLUSION
Thermal conductivity measurement apparatus based on the steady state hot wire method

° °
was developed. The results of the measurement of total solid content of skim milk, water
or oil content of W / and /W emulsions indicated that the measurement method can
provide high potential for practical applications.

REFERENCES
l. Mohsenin, N. N., Thermal Properties of Foods and Agricultural Materials, Gordon
and Breach Science Publishers, Inc., New York, 1980.
2. Sweet, V. E. Experimental values of thermal conductivity of selected fruits and
vegetables. J. Food Sci., 1974, 39, 1080-1083.
3. Arakawa, K., Tanaka, Y., Kubota, H. and Makita, T., Thermal conductivity of
freon mixtures. Proc. 5th Japan Symposium on Thermophysical Properties, 1984,
125-128.
4. Eucken, A., Allgemeine GesetzmaBigkeiten fur das Warmeleitvermogen ver-
schiedener Stoffanrten und Aggregatzustande. Forsch. Gebiete. Ingenieur., 1940,
11, 6-20.
 OPTICAL SENSORS (UV, VIS' NIR) FOR THE DETERMINATION OF
CONNECTIVE TISSUE, LIPID AND PROTEIN FUNCTIONALITY IN MEAT

H.J. SWATLAND
Department of Food Science,
University of Guelph, Guelph, ontario N1G 2Wl, Canada

ABSTRACT

New developments in using fibre-optics (FO) to measure

commercially important properties of meat are described. UV
fluorescence is used to detect collagen and elastin in meat,
which are important sources of toughness. Spectrophotometry has
been enhanced by using multichannel FO to measure at different
angles and positions relative to the point of illumination in
the meat.

INTRODUCTION

Prediction of the composition of meat by rapid, non-destructive,

optical methods is useful for the grading and sorting of fresh
meat, and for controlling the functional properties of meat
slurry in further processing using feed-back and feed-forward
control. FO were used first to predict the water holding
capacity of pork [1], but many other applications have now been
developed inVOlving ultraviolet (UV), visible (VIS) and near-
infrared (NIR) light. Colorimetry and VIS and NIR
spectrophotometry of food systems through FO are now well
established and routine. The objective here is to high-light
new developments of this technology.

UV FLUORESCENCE

Biochemical Types I and III collagen have different fluorescence

emission spectra and optimum separation is obtained with
excitation near 370 nm. Type I fibres emit a pre-quenching
spectrum for longer than Type III fibres because they are larger
in diameter and have a core that takes longer to quench. Thus,
the emission spectrum contains information on collagen fibre
diameters when duration and intensity of excitation are
controlled. The principle has been used to develop a hand-held
probe for detecting meat toughness, as shown in Figure 1.
 937

II
U
C

..
II
u .5
II
L
a
::l
I.&.

0
0 50 100

Distance mm
Figure 1. Signal from a beef carcass in a meat cooler using a
hand-held FO probe to detect tough meat.

POLARIZED LIGHT

No applications have been attempted yet using polarized light,

but basic research has shown that pH-related light scattering in
meat, which has been known for years but never explained,
originates from the scattering of light by the myofilament
lattice of myofibrils.

200
e
C
• •••
II
u 190 •
C
II
L •
II
'to
'to 180
c •
"r•
••
.r. ••
....
II
Q. 170
4.5 5.5 6.5
pH

Figure 2. Effect of pH on meat birefringence.

 938
In myofibrils, light splits into two components that travel
at different velocities, the ordinary ray (0) and the
extraordinary ray (E), with 0 ~ E. Birefringence, which may be-
or +, is given by n E - no. Retardation, the decrease in
velocity of light caused by interaction with.the medium, may be
detected as phase retardation, interference caused by path
difference E ~ O. The path difference of a depth of muscle (rm)
was measured by ellipsometry with a de Senarmont compensator.
rm changed with pH (Figure 2), and thus is a major factor in pH-
related light scattering in meat.

SPATIAL MEASUREMENTS AND GONIOSPECTROPHOTOMETRY

Light scattering in meat affects the pattern in which different

wavelengths penetrate the meat, so that pH-dependent properties
such as water-holding capacity may be predicted by both
spectrophotometry and goniophotometry. When both are used
together, the accuracy of prediction is improved [2). Figure 3
shows the FO window pattern of a multi-channel probe used for
the combined measurement of subcutaneous fat depth and the
marbling, colour, water-holding capacity and connective tissue
content of the meat.

< PROBE DIRECTION·

Light Guide
Group Large UV Fibre

Cc> (>
(>

~ Side Arm Fibres

(>
(>

Figure 3. FO window pattern of multi-channel probe.

REFERENCES

1. Swatland, H.J., Fiber optic spectrophotometry and the wetness

of meat. ~ Food Sci., 1982, 47, 1940-3.

2. Swatland, H.J. and Irie, M., Wavelength-position matrices in

the fiber-optic analysis of meat. ~ Comput. Assist.
Microsc., 1992, 3, 149-55.
 QUICK DETERMINATION OF FAT CONTENT IN BEEF LONGISSIMUS BY
NEAR-INFRARED SPECTROSCOPY WITH A FIBER OPTIC PROBE

MITSURU MITSUMOTO, SHINOBU OZAWA, TADAYOSHI MITSUHASHI,

KIMIYUKI SHINOHARA*, KENICHI TATSUBAYASHI*
Chugoku National Agricultural Experiment Station, 60 Yoshinaga, Kawai-cho,
Oda-shi, Shimane-ken 694, Japan
* Nireco Corporation, 2951-4 Ishikawa-cho, Hachioji-shi, Tokyo 192, Japan

ABSTRACT

Fat content in beef longissimus thoracis was compared with near-infrared (NIR)
spectroscopy readings using a fiber optic probe. A partial least square regression
was used to find the equation which would best fit the data. The correlation
coefficient between optical densities and fat content was 0.975 (SE=2.1 %). A NIR
spectroscope fitted with a fiber optic probe should enable quick determination of fat
content in the longissimus thoracis at the quartering section during beef carcass
grading.

INTRODUCTION

Marbling score of the longissimus thoracis at the site of quartering is the most
important characteristic for carcass grading in Japan. Fat content in the muscle
closely relates (r=0.93) with marbling score. Since conventional methods for
determining fat content are time consuming, a more rapid and accurate technical tool
is desired. Recently, near-infrared (NIR) spectroscopy has been developed for
analyzing chemical compositions of foods. The objective of this work was to evaluate
NIR spectroscopy using a fiber optic probe as a means of determining fat content of
beef.

MATERIALS AND METHODS

940
Longissimus thoracis from forty-five Japanese Black steers were used. A 5.5 cm
diameter x 6 cm deep sample was cut from the muscle using a template cutter and
placed in a specially designed sample cup to prevent interference from outside light.
Fiber optic spectra measurements (680 - 1235 nm) were performed by a Neotec
Model 6250 Spectrophotometer. Scannings were performed on both sides of each
sample to obtain the average value of individual beef cuts. A white ceramic disk was
used as reference. Scanning reference was performed with a specially designed
reference cup which had 2 mm of air distance between the fiber optic probe and the
ceramic disk. Data were recorded at 2 nm intervals and 50 scans / 25 sec were
averaged for every sample. Data obtained were saved as log 1/R, where R is the
reflectance energy, and then mathematically transformed to second derivatives to
reduce effects of differences in particle size and sample composition.
Fat content was determined by ether extraction.
A partial least square regression was used to find the equation which would
best fit the data.

RESULTS AND DISCUSSION

The correlation coefficient between optical densities and fat content was 0.975
(SE=2.1 %). Ben-Gera and Norris (1) reported a correlation coefficient of 0.974
between fat and AO.D. (1725 nm - 1650 nm) in 2 mm-thick emulsions of meat
products using NIR absorbance. Iwamoto et al. (2) reported a multiple correlation
coefficient of 0.996 for fat in ground pork using NIR reflectance. Kruggel et al. (3)
reported that multiple correlation coefficients for fat were 0.92 in emulsified beef and
0.81 in ground lamb using NIR reflectance. Lanza (4) reported that multiple
correlation coefficients were 0.998 for fat in emulsified pork and beef by NIR
reflectance. Although our samples were not emulsified or ground, a high correlation
coefficient for fat was obtained.
The conventional determination of fat content takes at least 3 days including
time for grinding, freezing, freeze-drying, ether extraction and weighing. On the other
hand, NIR determination of fat content can be done within ten seconds. A NIR
spectroscope fitted with a fiber optic probe should enable quick determination of fat
content in the longissimus thoracis at the quartering section during beef carcass
grading.

CONCLUSIONS

Fat content of beef longissimus thoracis at the quartering section could be quickly
determined by the near-infrared spectroscopy with a fiber optic probe. This would
lead objective grading of beef carcasses.
 941
39

R=O.975

--
(ft
CI)
30
SE=2.1 %

-
:J
~ 21

...ca
"tJ
CI)

-:J
-ca
(,)
12
0

3
3 12 21 30 39

Fat content (%)

Figure 1. Relationship between fat content determined by ether extraction and
values calculated by NIR spectroscopy (n=45).
R: correlation coefficient, SE: standard error

REFERENCES
1. Ben-Gera, I. and Norris, KH., Direct spectrophotometric determination of fat and
moisture in meat products. J. Food Sci., 1968, 33, 64-67.

2. Iwamoto, M., Norris, KH. and Kimura,S., Rapid prediction of chemical

compositions for wheat, soybean, pork and fresh potatoes by near infrared
spectrophotometric analysis. Nippon Shokuh;n Kogyo Gakkaishi., 1981, 28, 85-
90.

3. Kruggel, W.G., Field, A.A., Riley, M.L., Radloff, H.D. and Horton, KM.,
Near-infrared reflectance determination of fat, protein, and moisture in fresh meat.
J. Assoc. Off. Anal. Chem., 1981, 64, 692-696.

4. Lanza, E., Determination of moisture, protein, fat, and calories in raw pork and
beef by near infrared spectroscopy. J. Food Sci., 1983,48,471-474.
 INDIVIDUAL SUGAR CONTENT CONTROL BY THE USE OF
F.T.-I.R. SPECTROSCOPY COUPLED WITH AN A.T.R. ACCESSORY.

Veronique BELLON*, Celine VALLAT*, Olivier PAUWELS **

* CEMAGREF, BP 5095, 34033 MONTPELLIER Cedex 1, France.
** ARD, 27 rue Chateaubriand, 75 008 PARIS, France.

ABSTRACT

This paper describes the use of F.T.-I.R. spectroscopy coupled with an AT.R. accessory to
quantify the individual sugars (glucose, maltose, matodextrines ... ) in a mixture extracted from
a process of starch hydrolysis. In a first step, binary and ternary model mixtures are prepared.
Multivariate mathematical processing are applied to these spectra recorded from 1300 to 850
cm-I. The results show a very good discrimination of the different sugars: in the binary
mixture, the standard error of prediction for glucose (ranking from 100 to 200 glkg) and
maltose (ranking from 250 to 290 g/kg) are respectively equal to 1.. 66 g/kg and 1.07 glkg. In a
second step, the experiment is successfully run on real samples.

INTRODUCTION.

Starch enzymatic hydrolysis can be a means to produce small polymeres or monomers of

glucose: maltose (DP2), maltotriose (DP3), maltodextrine (DPn). The precision demanded by
sugar industry customers for respectively glucose, maltose, maltotriose and DPn are typically
8g/kg, 109/kg, 5g/kg, 5g/kg on samples containing over 800g/kg dry matter (in the so-called
60DE mixture). The concentrations are currently determined by H.P.L.C. which is a tedious
and time consuming method. Our purpose is to develop an original method able to measure the
concentrations of the different sugars fast enough and within the precision requirements cited
before. Mid infrared spectroscopy (M.I.R.) could be specific enough for our application.
Attenuated Total Reflection (ATR) [1] can be used to analyze visquous samples [2].

MATERIAL AND METHODS

Material
The F.T.-I.R. spectrometer is a BRUKER I.F.S. 25, equipped with a ZnSe Attenuated Total
Reflectance accessory. The spectra are recorded from 1350 to 800 cm-l with a 4cm-l
resolution and averaged on 100 scans.
 943
Three experiments have been done with model and real solutions (60 Dextrose
Equivalent or 60 DE) transformed by adding different quantities of sugar. The characteristics
ofthe samples in the different experiments are given in Table 1.

TABLE 1
Characteristics of the samples used in the three experiments.

Exp Binary Ternary Real Samples

Indiv. Sugars Binary Model Ternary Model True samples
Glucose (glKg) 100-200 90-160 196-216
Maltose (glKg) 250-290 80-120 300-330
Maltotriose (glKg) 80-120 120-138
DPn (glKg) 177-197
Nr of samples 55 55 54

Mathematical processing

The spectra are related to the concentrations of individual sugars thanks to 3 kinds of
mathematical processings: Mulilinear Stepwise Regression (M.L.R.), Principal Component
Regression (P.C.R.) and Partial Least Squares (P.L.S.).

The initial set is always divided into 2 sub-sets: the calibration and the validation sets.
The performances of the model are respectively given by the Standard Error of Calibration
(SEC) and the Standard Error of Prediction (SEP).

RESULTS

As the individual sugars of interest are gluocse polymers (glucose, maltose and maltotriose and
DPn i.e. sugars with a glucose polymerisation degree over 3), their spectra are very similar (Fig
1), what explains that the spectrum analysis will need powerful processing such as multivariate
calibrations.

0,9,---------------------------~,~~------------,

0,8
".--
ci 0,7
Q 0,6
Ul
t':l 0,5
Z
~ 0,4
\
cr
g 0,3
~ 0,2 •- - --... ...

0, ~ C==§~~~~__+------_+-------__:=-::=-:==::::J
1294 1214 1134 1054 974 894
WAVE NUMBERS (em-1)

Figure 1. Infrared spectra of glucose and its polymers and of real sample: - - glucose, ...
maltose, __ (thick) maltotriose, __ (thin) DPn, _. _ Real sample.

As shown in Table 2, in binary models, best results are got with P.L.S. mathematical
processing in which 3 axis have been computed. In the case of multilinear stepwise regression,
the wavelengths have to be selected according totheir correlation with sugar content. But as
shown in Table 2, stepwise regression is not satisfying and will be then abandonned.
944
TABLE 2
Standard Error of Prediction (SEP) for glucose and maltose determination in binary solution.

S.E.P. PLS PCR Stepwise Regression

(g/Kg)
Glucose 1.66 2.07 9.09
Maltose 1.07 1.43 5.06

Table 3 shows the results of multivariate analysis on ternary models and on real
samples.

TABLE 3
Performances of PLS2 and PCR for the quantification of individual sugars within the ternary
model mixture and real samples.

Experiments Ternary Real samples

Sugars PLS PCR PLS PCR
SEC SEP SEC SEP SEP SEP
Glucose 2.38 2.26 2.37 2.26 2.17 1.7
Maltose 4.12 5.18 4.89 5.27 4.71 5.04
Maltotriose 2.41 4.60 4.14 4.19 n.d. n.d.
DPn 2.91 3.11
Total Sugars 3.16 2.30 3.25 2.14

These performances, although not so good as the ones of binary models, meet the
requirements of the sugar industry in the ternary models. However, in the real samples,
maltotriose cannot be detected within the limits of the industry ..

CONCLUSION

Fourier transform A.T.R. spectroscopy processed by muktivariate methods shows very good
results for the quantification of individual sugars in aqueous mixtures either in model or in real
mixtures. Further experiments must be run in order to confirm the potential of this method for
sugar analysis I.e. to test the influence of contaminants (proteins and salts), temperature ...

BIBLIOGRAPHY

1 Belton, P.S., Saffa, A.M., Wilson, R.H., Use of Fourier transform infrared spectroscopy for
quantitative analysis: a comparative study of different detection methods. Analyst, 1987, 112,
1117-1121.

2 Cadet, F., Bertrand, D., Robert, P., Maillot, J, Dieudonne, J., Rouch, C., Quantitative
determination of sugar cane sucrose by multidimensionnal statistical analysis of their Mid-
Infrared attenuated total reflectance spectra. Applied spectro., 1991,45 (2) 166-172.
 ELECTRICAL CONDUCTIVITY IN AVOCADO AS MATURITY INDEX.

Mar Montoya, V. L6pez-Rodriguez, J.L. De La Plaza*.

Dept. Physics of Materials, Facultad de Ciencias, UNED, Madrid-SPAIN
*Dept. Fruit and Vegetables, Instituto del Frio, CSIC, Madrid-SPAIN.

ABSTRACT

Electrical conductivity, firmness, skin and pulp colour measurements were carried out in
avocado (persea americana. Mill) cv. 'Hass' during fruit ripening at +20°C. Conductivity
measurements were taken by an impedance bridge and two needle electrodes inserted into the
whole fruit. Respiration rate and ethylene production of the fruits were also determined.

Electrical conductivity increased slightly during the first five days. Afterwards, it grew rapidly
and reached a maximum on the seventh day. A drop in electrical conductivity was observed
when the fruits became overripe by about day 10. The onset of a steep rise in conductivity was
followed by softening and colour change in the fruit and it reflected the beginning of the
ripening process. On the other hand, the electrical conductivity peak arose at the same time as
the ethylene peak. These results have allowed us to set the electrical conductivity as a maturity
index simple and rapid to measure.

INTRODUCTION

The avocado fruit has a high rate of post-harvest respiration and limited shelf life [1]. It is
important to be able to recognize the maturity stage of fruits. In order to find a physical test,
simple and rapid to evaluate, some electrical properties in the low frequency range (100 Hz -
100 kHz) have been tested without success up to now because of the instability of registered
data [2] and because the results taken at the lower frequency tested were not supported by those
taken at the higher frequency [3].

Preliminary work [4] showed an enhanced technique for measuring electrical conductivity in the
whole fruit which approaches the above-mentioned problems. Here, we measured this electrical
parameter during avocado ripening as well as others physical properties. The time course of
ripening was followed by the respiration rate and the ethylene production.

MATERIAL AND METHODS

Firm and unripe avocado fruits (persea americana. Mill. cv. Hass) of uniform size were
harvested and then kept at 20°C and 80% relative humidity for up 11 days.
946
Respiratory intensity and ethylene production were calculated from the flowing system derived
from the Pratt and Mendoza method [5]. CO2 and ethylene analysis were determined by gas
chromatography of atmosphere samples of 1m! as described by Merodio and De la Plaza [6].

Electrical conductivity was determined by the method of M. Montoya et al [4]. This parameter
was measured at 20 kHz over samples of five fruits. Four measurements were taken at the
equatorial diameter of each fruit.

Skin and pulp colour were measured with a Hunter Lab Colorimetric, over an illuminated
surface area of 13 mm. The expressions lOab/L and (L*b)/lOO were used as skin and pulp
colour index respectively.

Firmness measurements were achieved with an "Instron" 1140 Universal Testing Meter. The
penetrometer used was a double metallic plate designed by De La Plaza [7]. Five fruit were
sampled and three measurements were taken at the equatorial diameter of fruit.

RESULTS AND DISCUSSION

Ethylene production by mature fruit remained below 0.5 ILl/kg-h for 4 days after harvest.
Ethylene production began to increase between days 4 and 5 and continued until the climacteric
(Figure 1)
350,-----,----------------------,0.4
_ _ C02
175 ~ Ethylene
300
............. Conductivity
s
"---
rn
250 -140
? .c
I
0.3 »
I
Ql) Ql) ~
.!<:
~200 "--- ....,
o~
Ql)
"2. 105 o
S ;:J
-0
>=i
N 150 ~
o
0 ::r::
N
o
u U 70 0.2 cr;
o
100 os::
....,
oQ)
35
50 E£l

OL---~0~~~~~~3~~4~~5~~-L*-~8~~-L~~

Days at 20 C
Fig.l Respiration rate, ethylene production and electrical conductivity during avocado
fruit storage at 20°C.

Accordingly, Figure 2 shows how physical properties tested, such as firmness, skin and pulp
colour, changed suddenly starting from the fifth day. These parameters developed, either
increasing (skin colour) or decreasing (pulp colour and firmness), until the fruit overshot the
climacteric. Electrical conductivity, however, followed a pattern similar to the ethylene
production one (Figure 1), that is, increased slightly during the first five days. Afterwards,
conductivity grew rapidly and reached a maximum on the seventh day. A drop on electrical
conductivity was observed when the fruit became overripe by about day 10.
 947
100

.......
14.8
8.0
*-)j{- *-*"* Firmness

----
Skin c.
0
....:l 0 Pulp C. 80
~
~
.-<
* ---
~
ro :0 13.6 - -,... - -- - - if<,
0 0.0

d* Z
><
Q)
50 ~
X
"0
.s
Q) UJ
"C! UJ

.S
Q)

-8.0 12.4 ;:::

;...
;::J r... 8;...
0 ;::J 40
0() 0 ~
0()
;:::
s.::
[fJ. -16.0 £<
;::J 112
.
P-. 20

- 2 4.0 -"--~--H:~0i--'-+-'---+--"--+3-'---:4i--'-+5--'---!5i--'--!'7--'--""'8!;---J'---f
9'--J..""'1"'0--'---;'1 P
Days at 20 C
Flg.2 Evolution of some physical properties of 'Hass' avocado during storage at 20°C.

CONCLUSIONS

Avocado fruit ripening is characterized by a sharp increase in ethylene production which reaches
a maximum at the same time as the climacteric peak in respiration. In practice, therefore,
changes in ethylene production can be used to estimate the time course of ripening in avocado
fruit. In the same way, our data show that electrical conductivity during ripening can be used
successfully as a physical maturity index, the determination of which is simple and rapid and
does not require expensive complicated equipment.

BIBLIOGRAPHY

1. Biale, J.B. and Young, R.E., The Avocado Pear. In The Biochemistry of Fruits and their Products,
Vol II, Academic Press, London, (1971)

2. Lozano, Y.K., Rotovohery, J.V. and Gaydon E.M., EtUde de caracteristiques pomologiques et
physico-chimiques de divers cultivars d' avocats produits en Corse, Fruits, 42 (5), 305-315, (1987).

3. Weaver, E.M. and Jackson, H.O., Electrical impedance, an objective index of Maturity in peach,
Canad. J. Plant. Sci., 46, 323-326, (1966).

4. Montoya, M., L6pez-Rodriguez, V. and De La Plaza, J.L., An enhanced Technique for Measuring
the Electrical Conductivity of Intact Fruits, Food Science and Technology, in press.

5. Pratt, H.K. and Mendoza, D.B., Colorimetric determination of carbon dioxide for respiration studies,
HortScience, 14 (2), 175-176, (1979).

6. Merodio, C. and De La Plaza, J.L., Interaction between ethylene and carbon dioxide on controlled
atmosphere storage of Blanca de Aranjuez pears, Acta Horticulturae, 258, 81-88, (1989).

7. De la Plaza, J.L., Calvo, M.L., Iglesias, M.C. et Luechinger, R., La Lyophilisation des avocats en
tranches, Proc. XIVth Int. Congo of Refrigeration, Moscu, 3, 679-689, (1975).
NEW ELECTRICAL METHOD FOR DENSITY SORTING OF SPHERICAL FRUITS

KORO KATO
Department of Agricultural Engineering, Faculty of Agriculture,
Kyoto University, Sakyo-ku, Kyoto 606-01,Japan

ABSTRACT
A new electrical dry method for density sorting of spherical fruits, which measures
the volume by electrical capacitance and mass by electronic balance, was proposed.
A exact relation between the capacitance of concentric double spheres and the
volume of the inner sphere was used for measuring the volume of spherical fruit.
The accuracy of measured volume was very high. The new automatic and
continuous density sorting system for sorting hollow heart and over-mature
watermelons based on this study was developed and tested. A new packing house
for sorting hollow watermelons using this dry method was constructed and operated
successfully.

INTRODUCTION
Density is one of the properties which can be used for non-destructive quality
evaluation of fruits, vegetables and nuts. Density sorting has been employed by
farms for some agricultural products since ancient times. However the
conventional method is an inefficient wet method using brine solution which
encounters complicated problems such as cleaning and drying the products and
it is not practically used in automatic sorting system in the packing house. There
is a change in density with maturity, cavity, soluble solid contents, damage, peel
thickness, etc. If the dry method can be developed, density sorting will have a great
potential for automatic sorting. Thus the dry method for density sorting of spherical
fruits, which measures the volume by electrical capacitance and mass by electronic
balance, was proposed.

Fig.1 Relation between density and cavity section of watermelons.

 949
MATERIALS AND METHOD
Relationships between density and the internal quality of citrus fruits, Japanese
pears and watermelons were investigated. It was apparent that the density of the
watermelon was related to the cavity section (cavity volume) as shown in Fig.l. The
relationships between density, cavity and ripeness were arranged and illustrated in
Fig.2. The watermelon with a density of 0.94 to 0.96 was fully ripe, high in sugar
content and had the best taste.

CAVITY NON-CAVITY

RIPENESS

1.00 0.95 0.90 0.85 0.80

DENSITY OF WATERMELON p (g/mI)
(Watermelons harvested after (0 to 50 days from flowering)
Fig.2 Relation between density, cavity and ripeness of watermelon.

It is very important yet c s= 4 7t e 0 err I r 2/ ( r-2- r I)

difficult to measure the fruit e r : Re I ati ve pe rm it ti vity 0 f air gap
volume precisely on the SO
c;:: r2=150mm
density sorting line. Q,
r2: Radius of outer spher
There is an exact ., .0 r2=300mm

C!2
U
relationship between the
electrical capacitance(Cs) of ""
u
:z: 30
concentric double spheres -<
Eo-
and the radius (or volume) u
-< 20
of the inner sphere as 0..
-<
u
indicated in Fig 3. 10
This principle can be
applied for measuring the
volume of spherical and oval O~0=-~~~5~O~~~~I~O~O~~~·~15~O~~
fruits having the ratios(k) of RAD IUS OF [NNER SPHERE r I (mm)
short to long axis in Fig.3 Capacitance of concentric double s~here.
the range of 0.86-1.16,
with small error within
2/1000 as shown in u"' Prolate Ratio of sh9rt to long axis k=r/c<1
'-
Fig.4, if the outer C E= S'n- € 0 € r C e / I n I ( 1 + e) / ( 1 - e) I. e =/W
u"' ). 005
sphere diameter is Oblate k>1
much greater than that '"'- C E=4n-€o€,re/sin-J. e=~
o~ J. 004
of the inner sphere. Ow Ct:Capacitance of spheroid
The new automatic e::~ Cs:Capacitance of spheres
and continuous density ;;2"'\. 003

$
r : Radius in the direction of X,Y
sorting system for ~~
sorting hollow heart ~~)'002 '0.c :Radius in the direction of

u~
and over-mature <w (
I,
watermelons based on u"" ~g:J.OOI J-.
this principle was c::-
"'-,
developed as shown in J. 000
O. 8
Fig.5. O. 9 \. 0 1.1 I. 2
RATIO OF SHORT TO LONG AXIS OF SPHEROID k=r/c
Fig.4 Capacitance ratio of spheroid to sphere.
950
Watermelon was put on a cup carrier with the fruit stalk on the horizontal plane
pointing in forward direction(Fig.6). The volume of watermelon is measured while
passing through the polygon external electrode tunnel, with conveying speed of
0.22m/s. The conductive rubber sucker electrode(Fig.6 & Fig.7), which is in contact
with the fruit surface by suction, and the precision capacitance meter(Fig.7) were
developed. The capacitance between the fruit and the external electrode was
measured with a high frequency of over IMHz to make good conduction through
the fruit skin. The relation between the volume and the output of precision
capacitance meter is shown In Flg.6. The density is calculated by computer from
volume data and mass data which is measured by electronic balance before entering
the tunnel while moving.
>
~ml
..,
~
External
f- electrode
~1500
I>iI
0..,
<u
!:;~ HIll
~t:
f-~ Conductive I
~~ 500 rubbe r sucke r I
f-U (Internal
5....
o 0 L-:-_--:-:~el~e~c~tr~o~d~e)~~~~~~L
o fro) lam l!ml
VOLUME OF WATERMELON (m I)
Flg.S Experimental system for electrical Fig.6 Relation between watermelon
density sorting of watermelon. volume and capacitance meter output.

RESULTS AND DISCUSSION

The test results of the new automatic and continuous density sorting system for
sorting hollow heart and over-mature watermelons was very satisfactory. The
accuracy of measured volume was very high, approximately 0.4%, when the
orientation of watermelon was maintained. The threshold value of density of
watermelons can be set in multi steps, according to the cavity volume. It became
clear that watermelon can be classified nondestructively according to the cavity
volume ratio by this electrical dry method of density sorting.
A new packing house for sorting hollow heart watermelon using this dry method
was constructed and operated. The system (Fig.8), which was controlled by a
computer, consisted of two sorting lines that successfully performed at a capacity
of 2800 watermelons /hr.

Fig.7 Conductive rubber sucker electrode Flg.8 Watermelon volume measurement

and precision capacitance meter. apparatus in the packing house.
 HIGHLY SENSITIVE AND RAPID DETERMINATION OF DIACETYL IN
LIQUID FOODS BY ELECTROCHEMICAL METHOD

N.Horikawa, K.Hayakawa, Y.Yamada, O.Miyawaki*

Research Institute, Kagome Co.,Ltd., 17, Nishitomiyama, Nishinasuno-cho, Nasu-gun,
Tochigi 329-27, Japan, *Department of Agricultural Chemistry, Faculty of Agriculture,
University of Tokyo, 1-1-1, Yayoi, Buokyo-ku, Tokyo 113, Japan

ABSTRACT

Measurement of diacetyl(DA) in liquid foods is very important for its role as a flavor
component with a low threshold. A sensitive electrochemical method for the determination
of DA was developed. After sample was subjected to a pretreatment process such as
filtering, centrifugal separation or distillation if necessary, DA was transformed to 2,3-
dimethylquinoxaline (DMQ) by the condensation reaction with o-phenylenediamine(OPD)
and electrochemical method was applied to determine DMQ with glassy carbon electrode as
a working electrode. By use of differential pulse voltammetry(DPV), a detection limit of
DA was 0.6~M(0.05ppm) with assay time of 15minutes. Optimum conditions for the assay
were 4.5-5.0 for pH, 40-50°C for the temperature in the electrochemical measurement,
and 8-10 minutes for the condensation time. DA in drinks could be determined by this
method and the reliability was compared with the colorimetric method.

INTRODUCTION
DA seriously affects drinks and foods quality. Although DA is a flavor-providing
component in fermented dairy products such as yogurt, it adversely affects the quality of
processed drinks such as tomato juice and some fruit juices. In either situation it is
important to measure the content of DA. Chromatography methods of measurement of DA
have the advantage of providing high accuracy, but they require troublesome operations and
take a long time for measurement. It is therefore the objective of our investigation to
develop an accurate and rapid method of measuring DA in drinks and foods. This paper
describes a DPV method for determination of DA. DPV method is compared with the
colorimetric method based on the Voges-Prokauer reaction.

METHODS

Procedure: 5mL of sample solution was combined with 5mL of 1mM OPD solution in
100mM acetate buffer(pH5.0). The mixture was stirred for 1Ominutes, then DA was
transformed to DMQ by the condensation reaction. The mixture was purged with nitrogen
(50mL/min.) to remove dissolved oxygen for lOminutes and DPV measurement to
determine DMQ by the electrochemical reaction was carried out. Yanaco polarographic
952
analyzer P-l100 was used for DPV measurement with glassy carbon electrode as a working
electrode(i.d. 5mm), scanning between -0.2V and -0.9V vs Ag/AgCI.
Calculation method of peak current; In the calculation of peak current, normal peak current
method and partial peak current method were used. In the normal peak current method, the
start point and the end point of the reduction peak of DMQ were selected manually. In the
partial peak current method, the data points of ±3OmV of the peak top were fixed as the both
ends of current peak. The reduction peak potential for DMQ was about -0.62V at pH5.0.
RESULTS AND DISCUSSION

Effect of condensation time; A standard solution containing IIJlM DA mixed with ImM
OPD solution was subjected to DPV measurement with condensation time varied from 0 to
20minutes. In the presence of an excess of OPD the condensation reaction was rapid and
finished in 6minutes at 30°C.
Effect of OPD concentration; A standard solution containing IIJlM DA was mixed with
OPD solution with concentration varied from O.OlmM to 5mM. After stirring for
lOminutes, the reaction mixture was subjected to DPV measurement at 40°C. The
condensation reaction with the OPD concentration higher than 0.5mM reached the
equilibrium.
Effect of pH ; The pH value of sample solutions seriously affected the electrochemical
reaction rate. Dependence of the normal peak current of DA derivatives on pH was
investigated. Britton-Robinson buffer were used for the adjustment of pH in the range 2-8.
The highest peak current was observed at pH5.0. It was effective for the highly sensitive
determination of DA to adjust the pH in the range 4.5-5.0.

Effect of temperature at DPV measurement: The peak current of DA derivatives increased

with increase in temperature at DPY measurement with lOJlM DA as a standard. Although
increase in temperature was effective for high sensitivity, the experimental error might be
increased at a high temperature like 70-80°C because boiling point of DA is 88°C. Thus the
optimum temperature for the assay was estimated to be at 4D-50°C.

Stability of peak current : A freshly polished electrode gives the highest sensitivity in
electrochemical measurement, since the sensitivity decreases gradually by the pollution at
the electrode surface. In the present case, however, the working electrode was stabilized by
the immersion in a sample solution for lOminutes. The repeatability of the peak current
both in the normal peak current method and the partial peak current method, was
determined for tomato juice and for tomato juice filtrate with DA added at 5.5JlM. Relative
standard deviation for the normal peak current method was 2.9%-7.7% and that for the
partial peak current method was 4.6%-13.0%.
Recovery of DA added in tomato juice: DA was added to distilled water and tomato juice
and analyzed by DPV method. All the calibration curves were linear to DA concentration
(Fig.I). DA recoveries in tomato juice were not much different from those in distilled water.

Comparison of DPV method with colorimetric method: DA concentration in off-flavored

pineapple juices were determined by DPV method with and without distillation as a
pretreatment. The normal peak current method and the· partial peak current method were
agreed well with the results by the colorimetric method. The colorimetric method was more
reliable than HPLC and GC method. Correlation coefficients between DPV method with
distillation and the colorimetric method were 0.984 and 0.996, for the normal peak current
method and for the partial peak current method, respectively (Fig2). These correlation
coefficients decreased to 0.948 and 0.945, respectively, when the distillation pretreatment
 953
was abbreviated. Thus, the distillation pretreatment was proved to be effective to increase in
the accuracy in the measurement.

CONCLUSION
DA in liquid foods could be detected sensitively and rapidly by DPV method with glassy
carbon as a working electrode. DA was transformed into DMQ by the condensation reaction
with OPD before the measurement. Optimum conditions for the assay were determined to
be 4.5-5.0 for pH, 40-50°C for the temperature in the electrochemical measurement, and
8-lOminutes for the condensation time. DA in off-flavored pineapple juice could be
determined by this method, the reliability of which was demonstrated by comparison with
the colorimetric method.

2.0 r-----;-----..,...------, r-------------__,.

f----e- Normal peak current: Distilled water
- - Y = -0.00897 + 0.0760x R= 0.9973
~ 1.5 .. . e,.- .. Normal peak current: Tomato juice
1: ....... Y = 0.240 + 0.0821 x R= 0.9952
~
:; 1.0 ··············---r·······················_-_··
o -+- Partial peak current: Distilled water
.l:J
.If::
111 - Y = 0.00883 + 0.0235x R= 0.9980
rf. 0.5 ..•... Partial peak current: Tomato juice
....... Y = 0.0509 + 0.0280x R= 0.9935

30

Fig.1 Relationship between concentration of DA added and peak current

Condensation reaction time: 10min.
Measurement temperature: 40°C

10
-e- Normal peak current method
>
0- - - Y = 0.795 + 1.13x R= 0.9844
Cl 8
>. - -~ - Partial peak current method
.J:J
"0 - - - - Y = 0.0388 + 1.19x R= 0.9957
Q)
c:
6
:1
'E~ /: , ,
~::1.
$~
4 I................ ;.............. ;..............
Q)
"0
t.i
c:
2 ---"""'-i-
8
«
Cl
O .........-'-'-..J....I.~~'-'--'-'--"'-'-.........-'-'-.........

o 2 4 6 8 10
DA conc. determined by colorimetry
(11M)
Fig.2 Relationship between diacetyl concentration determined by DPV
method and that determined by colorimetric method in pineapple
juice with distillation pretreatment
 NONDESTRUCfIVE QUALITY EVALUATION OF MELONS BY
ACOUSTIC TRANSMISSION CHARACTERISTICS

S.HAYASHI, J.SUGIYAMA*, K.OTOBE*.

Department of Agricultural Technology, College of Technology, Toyama
Prefectural University, Kosugi-machi, Toyama 939-03, Japan, * National
Food Research Institute, Ministry of Agricultural, Forestry and Fisheries,
2-1-2 Kannondai, Tukuba, Ibaraki 305, Japan

ABSTRACf

Acoustic impulse responses of muskmelons and watermelons were studied to

evaluate their quality. We found that the impulse waveform of acoustic
signals induced by impact was transmi tted along the surface with uniform
velocity in both fruits. The velocity varied drastically with the ripeness
of muskmelons. The relationship between the velocity and hardness of flesh
of muskmelons was examined. Two kinds of waves, however, were observed in
watermelons: one was the wave transmitted along the surface with constant
velocity, and the other was the wave going through the inside. The latter,
however, was not found to occur in internally cracked watermelons.
INTRODUCfION

We obtained acoustic signals at points which divide the equator of the

sample into 24 equal parts(see Figure 1). It became evident that the
impulse waveform of acoustic signals induced by impact was transmitted
along the surface of the equator with uniform velocity [1].
a lOms

1 t Impact
3 23

5/: ~

~
8
Flesh

11 13
Cross section of the equator

Figure 1. Transmission of wave forms along the equator of a muskmelon.

 955
MATERIALS AND METHODS
MATERIALS
Thirty six muskmelons with various degrees of ripeness were used.
Normal and internally cracked watermelons were also examined.
METHODS
We developed a system for determining the transmission velocity of
acoustic signals. The system consists of 2 microphones, an AID converter
(100kHz sampling frequency) with a simultaneous sample-and-hold circuit and
a computer(see Figure 2). The lag time, which is the difference between
two microphone signals, was detected using cross-correlation techniques.
Transmission velocity was determined by dividing the distance between the
microphones by the lag time(see Figure 3) .

Photoelectric
switch

Pendulum

Mpx.
Simultaneous Personal
sampling hold circuit computer
Figure 2. Measuring system of acoustic impulse response.
(a) Data acquisition o~ two acoustic signals

-1~------~-- __-L~______~______~
o Time (ms) 20
-
(b) ~xtraction of 256 data points 1 (c) Calculation of cross
~nfront of and behind the peak correlation and lag time
2
Lag time 0.13 ms
Transmission velocity
o r---,-----',;----- 38 . 5 ml s
(l ! - - - H - - \ J r - - -

-1~~--~~4-~
-1 '--'-'-_-'---'---...::::..J
o 1.4 2.8 0.5 o1.0
Time (ms) Lag Time(ms)
Figure 3. An example of the signal processing.

RESULTS

Experimental results in muskmelons were as follows : (1) The microphones

should be located on the equater 45° - goa from the impact point 0°. (2)
Optimum microphone-to-sample distance was found to be 1 - 4 mID. (3) Five
hundreds and twelve data points with 100 kHz sampling frequency were
needed to calculate the accurate lag time by cross-correlation techniques.
956
(4) The velocity varied drastically with ripeness of fruits. The
correlation coefficient between transmission velocities and fruit hardness
was 0.832. The transmission velocity of edible muskmelons ranged from 37
to 50 m/s(see Figure 4).
1000
........
'H
~800
en
gJ 600 (J)
correlation
.§ ~ coefficient 0.832
H +'
..@ 400 'rl
+'
::l
en
Calculations based on 512 data points.
e 200
'rl

~
r-~--r-~------~

45 65
Transmission velocity (m/s)
Figure 4. Relationship between transmission velocities and fruit hardness.
Two kinds of waves on the signal of a watermelon were observed: one
was the wave transmitted along the surface with constant velocity, and the
other was the wave going through the inside. The latter, however, was not
found to occur in internally cracked watermelons [2J.
Shock sound
lOms

II Impact
23 Peel
,j

/ / ,I
/11'
/ //'
./ 1/:
,/ I / I
19 I---~...JIV
7 /
I / I
I I
/ I

10
11
13 12
Cross section of the equator f-I-n-t-e-rn-a-l. . II,
13 wave ,,' ~e wave

Figure 5. Transmission of wave forms along

the equator of a normal watermelon.

REFERENCES

1. Sugiyama, J. and Usui, S., Nondestructive quality evaluation of

muskmelons by acoustic impulse response. Trans.Soc.Instrument and
Control Eng., 1990, 26, 367-374.
2. Hayashi, S., Sugiyama, J., Otobe, K., Nondestructive internal crack
detection of watermelons by acoustic signals. Bulletin of Toyama
Prefectural University, 1992, 2, 115-119.
 NONDESTRUCTIVE INTERNAL DEFECT DETECTION OF AGRICULTURAL PRODUCTS USING
SECONDARY ULTRASONIC WAVE

S.TANAKA, K.MORITA, S.TAHARAZAKO,

Chair of Agricultural Systems Engineering, Faculty of Agriculture,
Kagoshima University, Korimoto, 1 chome, Kagoshima 890, Japan

ABSTRACT

The purpose of the present work is to develop a method for nondestructive

internal defect detection of agricultural products using secondary ultra-
sonic wave. The experimental apparatus consists of a function generator,
a driver and detector, and a FFT Analyzer. The internal defects of
several products and the ripeness of kiwi fruit were measured. Resul ts
obtained were as follows: (l)The amplitudes of waveform with respect to
time for defective products were smaller than that for normal products.
(2)There was a correlation between the amplitude of waveform and firmness
of kiwi fruit.

INTRODUCTION

The requirement for nondestructive inspection of agricultural products is

strict. Because it is necessary to detect internal defects of many
agricul tural products rapidly, simply, accurately, safely and with low
cost. The purpose of the present work is to establish the fundamentals of
a method for nondestructive detection of internal defect of various
agricultural products using secondary ultrasopic wave instead of primary
wave.

MATERIALS AND METHODS

The internal defects of watermelon, Japanese radish (longish common type

and cv. Sakurajima), potato and apple, and the ripeness of kiwi fruit were
measured. The experimental apparatus consists of a function generator, a
driver and detector, sample holder and a FFT analyzer.
The products except for potato were held between the driver and the
detector by a sample holder. Potato was held by hand. A pulse of
sinusoidal secondary ultrasonic wave was input one side of a sample and
the transmitted waveform was received on the other side. The waveform was
analyzed with a FFT analyzer.
958
RESULTS AND DISCUSSION

1. On the detection of defective watermelon

Twenty normal fruits and forty-three defective fruits were inspected using
the ultrasonic apparatus. After each fruit was divided into two parts,
and the internal defect was studied.
Figure 1 shows waveforms with respect to time for watermelons, the
left is for a normal fruit and the right is a defective one. There are
obvious differences of amplitude and transmission time between normal and
defective fruit. The amplitude of a defective fruit is smaller than that
of a normal fruit, and therefore this technique should be able to detect a
defective fruit.

NORMAL DEFECTIVE
!>

-32+--r~r-.--.--r-.--.--.-.--+
o 2 4 o 2 4
TIME, ms TIME, ms

Figure 1. Waveforms with respect to time for watermelons

2. On the detection of bruised apple

Apples with artificial bruises greater than 1 inch diameter were in-
spected. Figure 2 shows the powerspectra of apples, the left is for a

NORMAL BRUISED

-80+-.--r-r~-.~~~~r-4
o 50 100 0 5 100
FREQUENCY, kHz FREQUENCY, kHz

Figure 2. Power spectra of apples

 959
normal apple and the right is for a bruised fruit. There are differences
in the output at the frequency band above 40--50 kHz, and therefore this
technique was able to detect a bruised apple. However, in this case, the
resul ts were obtained by holding the driver and detector to the bruised
part by hand. This is a problem to be solved in the future.

3. On the measurement of the ripeness of kiwi fruit

It can be considered that there is some correlation between the ripeness
of a kiwi fruit with thin peel and its firmness. The authors compared the
firmness of kiwi fruit measured with a pressure tester with a transmitted
waveform of secondary ultrasonic wave. Forty kiwi fruits with almost the
same firmness were left in a room at normal temperature, and five samples
were tested per day for eight days.
Figure 3 shows waveforms with respect to time for kiwi fruits, the
left is a sample of 570g for firmness and the right is a sample of 240g
for firmness. From the fact that the amplitude for a sample which is less
firm is small, it is assumed that a sample with small amplitude is riper.

1.6

FIRMNESS 570g FIRMNESS 240g

:>
~

....:l
Pt:l
~ I~
:>
Pt:l
....:l

-1.6 0
2 4 o 2 4
TIME, ms TIME, ms

Figure 3. Waveforms with respect to time for kiwi fruits

CONCLUSIONS

The authors conclude from the experimental results described above that
the method of using secondary ultrasonic wave is able to detect the
internal defects of selected agricultural products and the ripeness of
kiwi fruit.
The information provided by this method is less than that of X-ray
CT, but for practical usage, this method offers some possibility.
WIRELESS IMAGE TRANSMIT UNIT AND HIGH SPEED INSPECTION SYSTEM
FOR INNER SIDE OF CONTAINER

MASARU HOSHINO
Packaging Research Institute,
DAI NIPPON PRINTING Co.,Ltd.
591-10 Kamihirose,Sayama-City,Saitama,350-13 Japan

ABSTRACT
The Wireless Image Transmit(WIT) unit is -the device to transmit
plural video image data through closed pairs of special loop
antennas shielded against electromagnetic wave.
It is designed for the rotary-carrier inspection system, and
gives higher performance in speed and precision than
conventional line-type or index-type inspection system.
Our inspection system with WIT unit for laminated tube enables
inspection of inner side of shoulder part and nozzle part, also
accomplishes high speed operation at 190 tubes per minute.

INTRODUCTION
As the demand for the product quality is getting higher and
severer, manufacturers are aware of necessity to inspect all
products. Automatic inspection system replaces traditional
visual inspection in recent years for its convenience.
WIT unit improves inspection accuracy and processing capacity
of image inspection device combined with rotary-carrier system.
An inspection system for inner surface of shoulder part and
nozzle part of laminated tube has been developped and the
followings are the fundamental outline of the system.

HARDWARE COMPOSITION
Followings are the composition of 12 heads rotary-carrier tube
inspection system.

Inspection head
Inspection head consists of coaxial optical fiber light and 250
mm long bore-scope directly connected to CCD camera.
The bore-scope axis is shifted from the tube axis and the bore-
scope is designed to insert into the tube during inspection.
 961

•
Ja

t'". ~•. I
.'
., II
II

l'f
~~

.
(

Figure 1. Figure 2. Figure 3. Antenna

Inspection system Inspection head

:;(_ ... _... __ .--_.. _.. -_ ... _...ROTARY

_. --------.----.__ .. --_._ . ----.
UNIT
- - .. -_ ... _*; -.... ---_.. -----------_.
FIXED UNIT
- -_. ---. ----_ .. -_. -------).;
*IH:INSPECTION HEAD *II:IMAGE INSPECTION UNIT
*VC:VIDEO CAMERA *IR:IMAGE SIGNAL RECEIVER
*IT:IMAGE SIGNAL TRANSMITTER *ST:SYNCHRONOUS SIGNAL
*SR:SYNCHRONOUS SIGNAL RECEIVER TRANSMITTER
* IA:IMAGE SIGNAL ANTENNA
* SA:SYNCHRONOUS SIGNAL ANTENNA
*IHPS:INSPECTION HEAD POSITION SENSOR

Figure 4. Block diagram of WIT unit.

WIT unit
Figure 4 shows block diagram of WIT unit.
WIT unit is generally divided into two systems, image data
transmitter system and positioning data system.
Image data system consists of image signal system and
synchronous signal system. Each system is composed of
transmitter unit, receiver unit and antennas.
Transmitter unit consists of image-selector, modulator and
mixer. It modulates four channels video signals into one RF
signal.
Antennas are inpedance matched and shielded against
electromagnetic noises from outside. Its loop shape enables
constant transmission of signal during rotation. To perform
mutual communication of image signal and synchronous signal,
962
two cocentric loop antennas separated electromagnetically are
used.
Receiver unit consists of distributor and demodulator. It
recovers RF siganl to video signals.

Carrying and handling unit

Tube packages with caps are carried by plastic tube holders in
which the tube stands with the cap bottom.
During inspection, ring guide supports tube opening to fix the
rotating axis.
The tubes revolve together with the inspection heads around the
machine axis and they also repeat the motion of rotate
themselves 90 0 and rest under the head.
Tube loading is performed with the screw-star wheel
synchronized with the tube holder. At unloading unit,
defectives are rejected by air suction and the rest return to
line.

RESULTS AND DISCUSSION

Image transmission
Stable clear image is obtained without any noise interference
or electric wave leak. As plural image signals are modulated
differently, they are mixable and can be demodulated and
separate without any interference.

Inspection accuracy
This system can inspect 0.2 diameter of black spots when a
precise bore-scopes are used. As the bore-scope axis is shifted
from tube axis, inner side of nozzle part can be inspected
easily and efficiently.

Processing capability
The combination of WIT unit and rotary-carrier system performs
high speed operation. This system is capable of inspection at
the speed of up to 300 tubes per minute when appropriate
carrier unit is combined.

CONCLUSION
It is confirmed that WIT unit performs positive improvement in
tube inspection system at following points.
l.As the camera and the tube move together on inspection stage,
their relative position can be set up freely. This enables
effective inspection of complex shape -product.
2.Less restriction to video camera type as input source expands
the inspection field and enables more accurate inspection.
3.Commercial inspection device shows its potential efficiency
by combining with WIT unit and it can achieve high speed
inspection.
APPLICATION OF A BIOGAS BUBBLE COUNTER TO FERMENTA TION PROCESSES

YJ. LEE·, N.E. CHOI, D.H. WOO, K.M. KIM and J.K. CHUN
Department of Food Science and Technology, College of Agriculture and Life Sciences,
Seoul National University, *Nong-Shim Co. Ltd, Korea

ABSTRACT

A biogas bubble counter was designed to count the number of bubble produced from a
fermentation with photo-interrupter and IC chips. Gas production rate curve obtained with
the sensor represented the growth associated microbial activities without disturbing the
process environment. This system was applied successfully to monitor the Kimchi, alcohol
and methane fermentation and found to be very practical to trace the progress of
fermentation. To estimate gas production quantitatively, a measuring device of a individual
bubble size was developed. A fermentation controller was built with one chip
microcontroller to conduct the bubble counting and control the fermentation process.

INTRODUCTION

In most natural fermentations, the microbial flora and their activities are unpredictably
varied depending on the raw materials and process variables. Once substrate is fed into
the fermentation vessel and it is cured in a tightly packed state until the end of process.
Therefore it is hard to estimate the state of the fermentation and the quality of the product.
Conventionally broth is taken for the analysis of the process but it results in the
disturbance of the fermentaion environment, which is undesirable and rather harmful.
Chun et al.[l] developed a biogas detector which could measure the number of gas bubbles
produced during the curing process and they fabricated the biogas detectable sensor
consisting of photo-interrupter and IC counter chip, and illustrated the possibility of its
application in Kimchi, Tackjoo(alcohol brewing) and methane fermentation studies. [1, 2, 3,
4, 5,6]. This study is aimed to replace the conventional quality control method by a new
monitoring and control system in the fermentation industry.

METHODS AND MATERIALS

Biogas bubble counter: Chun's bubble counter was used to count bubbles having mean
volume of 0.02 cm3. Pulse signal obtained from the sensor was processed to convert to
964
Number of bubbles.[2, 4]
Fermentation materials: Kimchi, Tackjoo(alcohol) and methane fermentations were
tested with the new sensor. (a)Kimchi: Chinese Cabbages of good quality and medium size
were prepared by conventional method[2,3]. (b}Tackjoo: Korean rice-brewing was prepared
with Koji and cooked rice in Korean traditional method[5]. (c}Methane: Rice straws-cut
digested with various alkaline were used as substrate with CIN ratio ac!justed.[6]
Fermentation controller: We used Motorola 68705 chip (EPROM) for the CPU in our
controller and built the measurement module of bubble and temperature on the main PCB.
The controller has necessary actuator control module for heater and agitator of fermentor.
and it has a serial communication device for the monitoring system.

RESULTS
The gas production rate was analysed on the basis of the measured gas bubbles during the
Kimchi fermentation process. Number of bubbles acquired with the monitoring system was
plotted against the fermentation time and analysed their fermentation patterns of Kimchies.
The curve patterns well reflected the influences of the fermentation variables and
parameters such as material type, temperature and salt concentration.

--- radish
-e- cabbage

o 20 30
Fennentation period(hours)

Figure 1. Kimchi fermentation curves constructed by the bubble sensor

When the sensor was applied to Tackjoo brewing in a open vessel fermentor, a special
bubble collecting device was attached to the sensor. Ethanol content was in good
agreement with the gas production measured with the sensor as shown in Fig.2.

'C'
$
~
'1
"*~
7 c:
& 0
·til
~c: E
0
''8 4
8c:
::I 0
"0 0
e
Co
"0
.s=
0
0
2
<:

Fennentation period(day)

Figure 2. Gas and alcohol production curves in Tackjoo brewing

 965
In methane fermentation the monitoring of gas production rate is more important than the
analysis of gas composition, and the fermentation process is so sensitive to environmental
changes that minor disturbances are undesirable. Therefore we tried to use the sensor in
methane fermentor in an on-line manner to avoid the disturbance. Fig 3 shows the

10 100

>.
- Gas rate -I- Cumulative volume
«I .....
"t:I 8 80 cGI

-...
......
C
E
GI
g
::I
«I
c
6 60
«I
0 «I
:;:; 01
0
::I
"t:I 4 40 ~
E ~
0- :;
«I
«I
E
::I
(!) 2 20 0

0 0
0 5 10 15 20 25 30
Fermentation period(day)

Figure 3. Methane gas production rate and cumulative curves

cumulative and gas production rate during the fermentation period under the enviromental
control with the one chip controller. We also applied this sensor successfully to the
multi-fermentor system, where the various alkaline treatment effects were investigated for
several months.

REFERENCES

1. Lee, Y.]. and Chun, J.K. Development of gas production measurement system by bubble
counting during fermentation, Korean J. Food Sci. Techno!. 25(3)(in print,1993)
2. Lee, Y.J. and Chun, J.K., Development of pressure monitoring system and pressure
changes during Kimchi fermentation, Korean J. Food Sci. Techno!. 1990,22(6),686.
3. Choi, N, E., Application of Microcontroller to Plotting of Kimchi Fermentation Curve,
MS thesis, Seoul National Univ., Korea, 1989.
4. Woo,D.H., Development of the Bubble Size Measurement Sensor and its Application to
the
Monitoring of Kimchi Fermentation Process, MS thesis, Seoul National Univ., Korea, 1989.
5. Kim, K.M., Development of Single Chip Microcomputer-based Auto-Measurement and
Automatic Control Methods for Tacldoo Fermentation, MS thesis, Seoul National Univ.,
Korea, 1989.
6. Hwang, C.H., The On-Line Monitoring System for Methane Fermentation Process, MS
thesis, Seoul National Univ., Korea, 1989.
7. Choi, N, E., Woo, D.H. and Chun, J.K., Development of the bubble size measurement
sensor and its application to the monitoring of Kimchi fermentation, Korean J. Food
 COMPUTER AIDS IN FLEXIBLE FOOD MANUFACTURING

CHRISTINA SKJOLDEBRAND
SIK, Box 5401, S--40229 Goteborg, Sweden

ABSTRACT

The future of food production plant is being intensely discussed in Sweden at present. The industry is
facing tough international competition with the removal of trade barriers and deregulations. At the
same time customers are demanding highest quality and greater varieties of food.
Tomorrow's plant will have to be more flexible, capable of producing products that may have only
limited duration on the market. Many different products must be produced in the same plant - often
with many shifts a day.
The consequences of these trends will be better production planning, production uniformity targeted
towards predetermined quality short start-up and stoppage times, and minimum bottle necks.
Integrating computers will be one solution to achieve the needed flexibility. It must be possible for
the food industry to automate further by enhancements without whole plants or production lines
becoming obsolete.
This paper will present results from projects that have been carried out in Sweden at the moment.
The results are - among other things - computer programmes that have been designed for production
planning in a bakery, a meat factory, and a fine chemicals factory.

BACKGROUND

Quality assurance will be more and more important in the future food production plant. Together with
the demand on increased productivity and better use of raw material the structure of production plant
layout and planning will change.
Plant will have to be more flexible, capable of producing products that may have only limited
duration on the market. Many different products must be produced in the same plant often with many
shifts a day. The plant must have the ability to shift to another product, or accomplish a variation on
the original recipe. The results will be reduced made-to-order inventory, lower costs to the
manufacturer and better prices and quality for the consumer. Global marketing will be based on rapid
response (Springer 1990, Le Maire 1990).
The consequence of these trends will be more effective production planning, production uniformity
targeted towards predetermined quality, short start-up and stoppage times and minimum bottlenecks.
Integrating computers and automation will be one solution to achieve the needed flexibility. It must
be possible for the food industry to automate further by making enhancements without whole plants or
production lines becoming obsolete.
Integrating computers into the food manufacturing process is complex and requires more than just
overbuying hardware and software as in the past. The operator in the plant has to be involved in the
development.
 967
THE DUP PROGRAMME

Since 1987 the Swedish Government through the Swedish National Board for Industrial and Technical
Development (NUTEK) has funded a research and development action area programme where
applications of new information technology are tested in the process industry. The programme is called
Development of User Friendly Process Operation Systems (DUP).
The goal of the DUP R&D programme is to develop process control systems that cultivate and
utilise the skill of the operator and facilitate a broadening of his or her role and at the same time
contribute to a better and more even product quality by improved use of raw materials and energy and
increased productivity.
Three different industrial areas are emphasised. These are the paper and pulp industry, the chemical
industry and the food industry. The DUP programme is divided into two sub-programmes: basic
studies and case studies. The basic studies are coordinated with the case studies and are intended to
give support to their projects.
This programme is a six-year programme which means that it will end this summer (1993). The
Government has, however, decided to extend the programme for 4 years more in order to give
information about the results to representatives in the whole process industry. These information
activities will be carried out in many ways.
The most important project within the DUP food industry programme concerns support and control
systems to food industries with flexible production systems. A tool box will be developed. This box will
contain different software for computer aids in a food production plant with flexible production.

FLEXIBLE MANUFACTURING SYSTEMS

Flexible manufacturing systems (FMS) as a concept have been around for many years. They are
directly related to the computer era and the technologies that have evolved. FMS is a computer
controlled array of semi-independent work stations and an integrated material handling system
designed to produce a family of related products with medium variety and medium production volumes
of each (Clayton 1987). The concept is often mentioned in connection with the engineering industry
where it signifies a production plant system that has the ability to change production conditions. In the
food industry, however, the development of FMS has been slow because managers are cost oriented
rather than asset oriented and reluctance has left them with cheaper but unreliable equipment. The
FMS will be very interesting in the food industry as, for example, batch processing will be the main
type of production unit in this industry.
The traditional batch production process machine actual use has been observed to be only 6-8% of
the available machine time. In contrast 45-55% machine utilisation is achieved in dedicated highly
specialised transfer lines in which movement is in lockstep along a production line. For a typical
machine product the cost using the most efficient mass production methods is less than, often much less
than, 10% of the cost of batch production.
Why is there renewed interest in batch production? The increased demand on product quality and
productivity of the production can be fulfilled by using, among other things, different computer
systems. Flexibility is another keyword; the ability to use a manufacturing facility in such a way that
one more piece of equipment with related auxiliary support devices for materials handling is
integrated. The production organisation and working organisation are as important.

INCREASING FLEXIBILITY

As said before, tomorrow's plant will have to be more flexible, capable of producing products that may
have only limited duration on the market. Many different products must be produced in the same
plant. The customers are demanding higher quality and greater varieties of food.
This will demand a higher productivity and efficiency from the production plant. The most important
968
issue for the food industry in order to increase flexibility is to develop equipment and control systems
that optimize and control production lines and quality. These have to be user friendly. The operators
have to be educated both in order to increase their knowledge and to take more responsibilities. It is
important that they understand their role, their responsibilities and that they can influence their
working conditions.
 AUTOMATION OF CHEESE AND YOGURT MANUFACTURING PROCESSES

G. S. Miltal
School of Engineering, University of Guelph;
Guelph, Ontario, Canada NIG2Wl

ABSTRACT

Milk coagulation and curd firmness in commerc!al operations is usually determined subjectivel},
either manually or visually. However, many factors that affect curd fmnness do not remain
constant. Curd fIrmness is influenced by acidity, heat treatment or cold storage of milk prior to
cheese making, variation of calcium or inorganic salts in milk. Cheese is more uniform when the
curd is cut at a constant fIrmness determined instrumentally. Line heat source probe, generally
used to determine thermal conductivity, was used to determine coagulation time of curd and
yogurt. This instrument can be used to automate cheese and yogurt manufacturing. The
automation will maximize cheese and yogurt yield and provide optimum control of cheese
moisture.

INTRODUCTION

Curd fmnness in commercial operations is usually determined subjectively, either manually or

visually. Some vats are cut automatically after a specifIed time. More objective instrumental
determination of coagulation characteristics, curd ftrmness, and cutting time should refIne cheese
making, maximize cheese yield, and provide optimum control of cheese moisture. The overall
objective was to develop techniques for on-line monitoring and control of the cheese and yogurt
making processes.

METHODS AND MATERIAL

A line heat source probe was used to measure coagulation of milk for curd and yogurt production.
The probe (Fig. 1) was inserted into the milk sample containing enzyme (rennet) for curd making
or bacterial culture for yogurt manufacturing. A constant current of 200 rnA was applied to the
heater wire. The probe temperature was recorded at 2 s interval on a data logger. The thermal
conductivity of the milk remains constant during curd/yogurt formation (1).
970

1....- - - S.G OZDOlID-----1.~1

0.88 mm. O.D.

TBERKOCOUPLlC
JUNCTION

Figure 1. A line heat source probe

RESULTS AND DISCUSSION

Cheese/Curd
The milk is coagulated either by addition of rennet or acid or both. The coagulated milk (curd)
is cut into cubes and excess water is expelled by continued action of the rennet, by acid
development, by manual/mechanical stirring, and by heating at a desired temperature. The
coagulation time (CT) is generally used to characterize changes in the physical state of
coagulating milk. The CT of milk has been measured by using various instruments. Only a few
equipments are appropriate for in-line use in a dairy plant. The syneresis in renneted milk is
initiated by cutting the curd. Cutting curd at optimum curd fIrmness is important to minimise loss
of milk solids and for proper drainage of whey (2).

The CT of milk was measured at 32 0 C and adding single strength rennet (Dairyland Food
Lab., Waukesha, WI) at a rate of 0.2 mUL of milk (3). Fig. 2 shows a typical plot between probe
temperature minus initial milk temperature (dT) and time. A sharp increase in this temperature
difference was noted at the time of milk coagulation. The tangent to the inflection point on the
17

o
u I
15
D
)"~
/

13 i
i
111
/
lD C
oagu IaUon
' .
Ume
r-·-·--·-~
9 ! . . .
o 5 10 15 20 25 30
Time, min
Figure 2. Measuring the coagulation time of the renneted milk
 971
.1T vs. time provided CT. Thus, the hot wire temperature could detect the onset of milk clotting
in line and in real time without disturbing the milk coagulum as also observed (4).

The curd cutting time has not been measured by using this probe. With further work it
may be feasible to determine the proper curd cutting time for various situations.

Yogurt
There are two predominant types of yogurt produced--set yogurt and stirred yogurt. After
pasteurizing, the sterile milk is homogenized, and then cooled to about 41 0 C to 43 0 C before
fermentation. It is inoculated with equal numbers of LactobaciJIu..'. bulgaricus and Streptococcus
thermophilus at about 2% level. The automation of stirred yogurt ig feasible by the sensor tested
in this work. Fig. 3 shows a typical plot obtained for the probe temperature vs. time during yOg~lrt
manufacturing. The time at which probe temperature jumped to a high value provided coagulation
time for the yogurt. This should also be evaluated to continuous proces!; yogurt compared to batch

I
operation.
43 '~----------- __________

o
u

r
r-----~--j
0.15
___ ____ ____
~5~----~---~--~~--~--~
1.85 2.7 3.55 4.4
~
5.25
~

6.1
4

Time, h

Figure 3. Measuring the coagulation time during yogurt manufacturing

REFERENCES

1. Hori, T., Effect of rennet treatment and water content on thermal conductivity of skim
milk. J. Food Sci., 1983,48, 1492-6.

2. Bynum, J.F. and Olson, N.F., Influence of curd flrmness at cutting on cheddar cheese
yield and recovery of milk constituents. 1. Dairy Sci., 1982, 65, 2281-90.

3. Sharma, S.K. 1992. Kinetics of Enzymatic Coagulation and Aggregation of Ultrafiltered

Milk. Ph.D. thesis, Univ. of Guelph, Canada.

4. Hori, T., Miyawaki, O. and Toshimasa, Y., In-line measurement of milk clotting by a hot
wire method. In Engineering and Food, Vol. 1, ed. W.E.L. Spiess and H. Schubert,
Elsevier Applied Sci. Publ., London, 1990, pp 743-51.
 AUTOMATION OF SHRIMP QUALITY EVALUATION

MURAT BALABAN, SENCER YERALAN, YMIR BERGMANN, W. STEVEN OTWELL

Food Science and Human Nutrition Dept. University of Florida
Gainesville, FL. 32611, USA.

ABSTRACT
Current quality evaluation of shrimp involves subjective determination of
melanosis, and manual measurements of count per unit weight, ratio of largest
to smallest shrimp, and presence or absence of broken parts, foreign
materials, etc. With the globalization of shrimp trade, objectivity,
repeatability and standardization of quality evaluation is necessary.
Computerized image analysis was used to evaluate count, uniformity ratio, and
weight estimation. Weight of individual shrimp were predicted from surface
area. Different correlations were tried to relate weight to surface area.
Analysis of errors from automated evaluations is presented.

INTRODUCTI ON
Shrimp fisheries, imports and processing are important to Florida (92% of U.S.
total value, 1990). About 70% of the processed shrimp is imported. Therefore,
objective standards in trade will help in maintaining a high quality and safe
shrimp supply. Access to rapid, objective and repeatable quality evaluations
will help regulatory agencies in assuring high quality and safety that the
consumer expects from a high-priced food. Current shrimp evaluation uses semi-
quantitative sensory methods (visual, smell, texture). Development of an
automated and objective qual ity evaluation device using computer vision, rapid
ammonia determination, and texture measurement will improve standardization
of shrimp quality, decision making in purchases from remote locations, and
help in setting standards in marketing. With the accumulation of objective
and repeatable quality data, prediction of quality from different locations,
different species, different seasons etc. will also be possible. Machine
vision has been tried in grading and orienting oysters [1, 2], in recognizing
fish species [3, 4], and in shrimp processing [5-7]. Automation of count,
uniformity ratio and weight estimation of shrimp is presented in this paper.

MATERIALS AND METHODS

A VCR camera was coupled to a ComputerEyes/RT color frame grabber (Digital
Vision, Dedham, MA) installed in an IBM PC. White shrimp was purchased intact,
frozen. A wide size range was selected. Headless tiger shrimp was purchased
 973
as a 2 kg frozen block. The package displayed a count of 21-25 per 453 g. A
5 mm x 5 mm black metallic piece served as an area reference square. Shrimp
were placed one by one on a vision platform with the reference square. A
computer program determined the surface area of shrimp relative to that of the
reference square. Shrimp weight was determined by a balance. Sixty-three white
shrimp were processed this way. Then, a} heads, b} shell except the tail and
last segment, and c} all shell and tail were removed. Between each operation,
shrimp were processed as before, resulting in four groups of view area-weight
data for each shrimp. Tiger shrimp was separated into two batches of 50
shrimp each. These were processed as above, resulting in two independent sets
of view area-weight data for headless, peeled, and tail off forms. The data
were fitted the following equations: Linear, Y=A+BX, Power, Y=A exp(BX},
Forced power (F. Power): Y=A XI . S, where X=view area (mm2), Y=experimental
weight (g). The coefficients A, B, and R2 were determined. The estimated
weight of each shrimp, as well as the total weight estimation were calculated
for each equation fitted. The count and uniformity ratio of shrimp were also
determined for the actual data, and for the predictions of various equation
fits. For the tiger shrimp, the A and B values for each equation fit for one
set were used to estimate the weight of the other set, and vice versa.

RESULTS AND DISCUSSION

The A, B, and R2 values are not given due to space restrictions. Experimental
and estimated total weights, counts, and uniformity ratios based on various
fits for white shrimp are shown in Table 1. For intact shrimp, the power and
forced power equations are very close. For the linear fit there is a trend of
decreasing intercepts and similar slopes as shrimp is deheaded, peeled and
tail taken off. For the power fit, A's increase, while B's are around 1.5.
For the forced power fit, exponent of the area is 1.5, and it increases with
decreasing shrimp size (intact, peeled, tail off). Table 2 shows experimen-
tally determined, and estimated total weight, count and uniformity ratios for
batch 1 tiger shrimp. Batch 2 data is similar (not shown). In this table the
vi ew area of the batch 1 was used with the parameters of batch 2 (" By batch
2 parameters"). Since batches 1 and 2 came from the same box, their parameters
should be interchangeable, and the parameters of one should accurately be used
to predict the properties of the other. The only significant difference is in
the peeled tail off form, where estimated weight of one batch by the
parameters of the other is different by about 2%.
The weight, count, and uniformity ratio of white and tiger shrimp can
be accurately evaluated by computer vision. We need to evaluate other product
forms (such as butterflied), and other species to determine the maximum error
associated with the "vision weighing" system.

REFERENCES
1. Diehl, K.C., Awa, T.W., Byler, R.K., van Gelder, M.F., Koslav, M. and
Hackney, C.R. 1990. Geometric and physical properties of raw oyster
meat as related to grading. Transactions of the ASAE. 33: 1270-1274.
2. Tojeiro, P. and Wheaton, F. 1991. Oyster orientation using computer
vision. Transactions of the ASAE. 34:689-693.
3. Wagner, H., U. Schmidt, and J. H. Rudek. 1987. Distinction between species
of sea fish. Lebensmittelindustrie 34: 20-23.
4. Pau, L.F. and Olafsson, R. 1991. Fish quality control by computer vision.
Marcel Dekker, Inc. New York.
974
5. Marel, 1991. Model L-I0 Vision Weigher for shrimp processing. Reykjavik,
Iceland.
6. Ling, P.P., Searcy, S.W. and Grogan, J. 1988. Adaptive thresholding
techniques for shrimp images. Presented during the 1988 international
summer meeting of the ASAE. June 26-29, 1988. Rapid City, SO.
7. Ling, P.P., and Searcy, S.W. 1989. Feature extraction for a vision based
shrimp deheader. Presented during the 1989 international winter meeting
of the ASAE. December 12-15, 1989. New Orleans, LA.

TABLE 1
Experimentally determined, and estimated total weight, count and uniformity
ratio values for different forms of white shrimp.
Form Experi menta 1 Linear Power F.Power
Intact Total weight (g) 1762.6 1762.3 1756.9 1768.3
Count (shrimpj453 g) 16.0 16.0 16.0 15.9
Uniformity ratio 2.45 2.52 2.36 2.36
Headless Total weight (g) 1217.7 1218.6 1217.5 1225.9
Count 23.5 23.5 23.5 23.4
Un iformi ty rat i 0 2.34 2.45 2.38 2.45
Peeled Total weight (g) 1083.8 1084.3 1082.6 1091.8
Count 26.4 26.4 26.5 26.2
Uniformity ratio 2.36 2.42 2.36 2.41
Tail off Total weight (g) 1008.2 1007.9 1005.2 1011.7
Count 28.4 28.4 28.5 28.3
Uniformity ratio 2.41 2.48 2.41 2.55

TABLE 2
Experimentally determined, and estimated total weight, count and uniformity
ratio values for different forms of tiger shrimp, batch 1.
Form Experimental Linear Power F.Power
Headless Total weight (g) 803.16 803.16 802.37 801. 40
By batch 2 parameters 807.91 807.08 806.90
Count (shrimpj453 g) 28.3 28.3 28.3 28.4
By batch 2 parameters 28.1 28.2 28.2
Uniformity ratio 1.34 1. 28 1.28 1.39
By batch 2 parameters 1.25 1.25 1.39
Peeled Total weight (g) 713.71 713.71 713.05 712.33
By batch 2 parameters 714.53 713.94 712.93
Count 31.8 31.8 31.9 31.9
By batch 2 parameters 31.8 31.8 31.9
Un iformi ty rat i 0 1.34 1.30 1.29 1.39
By batch 2 parameters 1. 29 1.29 1.39
Tail off Total weight (g) 655.67 655.67 654.95 653.86
By batch 2 parameters 666.96 666.53 669.56
Count 34.7 34.7 34.7 34.8
By batch 2 parameters 34.1 34.1 33.9
Uniformity ratio 1.35 1.31 1.30 1.43
By batch 2 parameters 1.32 1.32 1.43
 AUTOMATIC CONTROL OF THE BISCUIT BAKING OVEN PROCESS

G. Trystram (1), M. Allache (2), F. Courtois (1)

(1) ENSIA, Food Process Engineering Department, France
(2) CTUC, Massy, France

ABSTRACT
The automatic control of a biscuit baking oven is studied from an experimental point of view.
Sensors are adapted and identification is performed. A pole placement strategy is developped
to control the colour of biscuits.

INTRODUCTION
The control of the baking oven process is important to improve process efficiency and the
quality of baked cereal products. Few papers are dedicated to this problem. To minimize the
energy losses and to maximize bread quality an approach with manual control is tested (2). A
strategy for colour control based on air velocity control is discussed in (5). Some applications
with feedforward and coupled control of moisture and colour are presented (4). The project
presents a good opportunity to study sensors and associated control law of baking oven. First
results are presented.

MATERIALS AND METHODS

The Baking oven process

A pilot plant is used to study the dynamic of the oven. It is a 15 meter long oven indirectly
fired with multiple burners, heated with natural gas. 4 combustion chamber are equipped, in
the roof and base, with 34 independently controlled burners. The band is 0.65 meter width.
Six zones are available, and five exhausts permit the control of air velocity and of atmosphere
composition inside the baking chamber. A Programmable Logic Controller interfaced with a
computer is used for oven control and data handling. Classical control functions are
implemented on the PLC (1, 6). This oven is used, in this research, for the baking of biscuits.

Sensors

In order to study baking oven control, sensors are adapted or developed specially for
measurement quality of biscuits. Due to the drying phase occurring during baking, the final
moisture content of the biscuit is an important parameter. A near infra-red analyser is used on
line (Infrared Engineering, MM55G). This sensor is located at the exit (figure 1) of the oven.
976
Studies were performed in order to establish the best location and the accuracy of the
measurement: 19b.
Moisture is not the only quality parameter for the biscuit. From a consumer point of
view, other characteristics must be taken into account. The colour is one of these
characteristics. Two measurement systems are studied. Firstly, image processing is compared
with laboratory measurement (using Minolta CR200 colour measurement system). A good
correlation is established between grey level analysis through image processing and
luminosity (L): 0.987 is the correlation coefficient. Measurement is performed in a few
seconds and is available for 5 biscuits at a time. Secondly, a real time measurement system is
used. It is the Colourex (Infrared Engineering) which is able to measure the colour of the
product in the CIE Lab coordinates. This sensor is located at the exit of the oven, near the
moisture sensor. Distance between product and sensor is 0.03 m. Comparisons are made
between laboratory measurement (Minolta and Gardner spectrometer) and Colorex. A very
good correlation is obtained: 0.998.

Due to the cyclic derive of the band, strong and artificial disturbances are added to the
real time measurements. As an example, Figure 1 presents a recording of colour with and
without filtering.

Method
No publications are available for the description of the dynamics of moisture or colour in
oven. The first part of this work is the systematic analysis of this dynamic behaviour. To do it,
a point is chosen which is characterised by: baking time 8 min, air temperature (setpoint of
the controllers): 200°C in each zone for roof and base, air exhaust profile: 20,40, 80,0, (in
9b of control input). On the basis of this process tuning which is used to establish
°
reproducibility of trials, step variations are performed for each control variable: 6
temperatures of air, 5 air exhausts and velocity of the band. Different step amplitudes are
used. After filtering, the data are studied in order to establish the transfer functions between
each control variable and moisture content and colour respectively. The controllers are
designed on the basis of these transfer functions. Firstly the right control variable is chosen, in
order to control moisture first and colour secondly. The control law is then established and
parameters are calculated. Simulations of the performances are then performed.
RESUL TS AND DISCUSSIONS

Figure 1 presents the comparison between measured colour variations versus time, due to air
temperature step modification in zone 5, and predicted colour using dynamic model as
identified from experimental measurements. The two curves are in good agreement. This
numerical transfer function is a very simple model which is easy to determine from specific
experimental results.

This method has been applied for all the measurable variables (water content, colour).
All oven transfer functions are known. Air exhausts appear to be bad control variables. This is
probably due to a bad design of the fan exhausts. The most significant variables are the air
temperatures. Control variables are chosen from these results. For example for colour control,
air temperature in zone 5 appears to be a good choice. The transfer function is used to
determine the controller. A pole placement strategy is prefered and parameteres are calculated
using PIM software (3). The oven responds quickly to this setpoint change. The effects of
disturbances are well reduced. This illustrates the efficiency of the pole placement strategy.
These results should be validated directly on the oven itself, but this study is an illustration of
the capability of control science to provide progress in biscuit baking oven process.
 977
100 Luminosity

80

60

40
Model
20+---~---'----~---r--~~--r---~--~--~
o 20 40 60 80
Sampling time

Figure 1: Recording of colour of biscuit at the end of the oven. Comparison between
measured and modelled colour.

CONCLUSION

Even if a baking oven is a complicated process (distributed parameters system) it is

established that progress with sensors is possible specially with on line moisture and colour
sensors. Work is in progress for on line thickness measurements. On the basis of these
measurements, the dynamics of moisture and biscuit colour is established. Pole placement
controllers are tested and simulations are performed. In the near future pilot validation will be
performed. This establishes that automatic control of biscuit qualities is possible.

REFERENCES

1. Emprun, C., Brunet, P. and Trystram, G., La cuisson des produits cerealiers dans un four
it regulation automatique, Congres innovation energetiques et industries alimentaires,
AFME,1990
2. German, H., Meuser, F., Z.F.L., 1987,38,25-30.
3. Landau 1., Identification et commande, Hermes, paris, 1989.
4. MacFarlane, 1. Automatic control of food manufacturing systems, 1988, Elsevier applied
sciences, London.
5. Sato A., Sato, Y., Yamanoi, K., Method of controlling the baking of foods, United states
patent, 1990,4.963.375.
6. Brunet, P., Savoye, 1., Trystram, G., Rapeau, F., Flexibilite d'un four de biscuiterie par
regulation automatique, Congres innovation energetiques et industries alimentaires,
AFME,1990,
 STUDY OF HANDLING TECHNIQUES FOR THE SOFT AND
PLASTIC SUBSTANCE SUCH AS FOOD

SHINZO MAMMOTO
Mayekawa MFG.CO.,LTD.
Food process sect.
13-1 Botan 2, Koto-ku, Tokyo 135, Japan

ABSTRACT

For handling irregularly shaped, soft-textured materials such

as food products, it is necessary to provide robots which
have functions similar to human eyes and hands. The aim of
this study is to construct a control system for grasping
movement of mechanical gripper using visual and force
sensors. Applicabilities of this system to typical Japanese
food products such as tofu and kamaboko are confirmed.

INTRODUCTION

Sophisticated functions such as those provided by the human

hands and eyes are required in order to achieve a clear
grasp of irregularly shaped, soft-textured materails such as
food products. For this reason, most of the dishing up
processes, are performed manually.
In order to contribute the automation of food processing
plants of the future, this study is aimed at construction of
a system capable of grasping irregularly shaped,
soft-textured materails such as food products.
The proposals contained herein make full use of modern
computer and electronic technology and studies of
visualization systems and grasping movement control systems.

PROCEDURE AND RESULTS

The construction of an automatic handling system to irregula

rlyshaped, soft-textured materials such as food products, is
obtained by means of the integration of the following basic
technologies.
 979
(1) Technique for measurement system of mechanical
characteristics of the objective materials and
development of the data base system.
(2)Technique for pattern recognition of irregularly shaped
materials such as food products by visual sensor.

(3)Technique for reduction of the image processing time by

means of hard ware and soft ware of computer.

(4)Technique for a real time control system for grasping

movement of mechanical gripper with force sensor.

Table 1 shows the mechanical charactristics of typical food

products in Japan. Food products have the peculiar
mechanical characteristics because of viscoelasticity.

Figure 1 shows the composition of functional devices for

visualization processing. The processing time for pattern
recognition and visual information data was less than a few
seconds.

TABLE 1
Grasping characteristic of typical food products
Narre of food products
Item
Kanaboko Processed Cheese Tofu
Shape Rectangular parallel-piped
Shape of gripper Flat plane
Setting displacement 3 (mn0
Rate of compression 20 <% )
Grasping 1l1JVement speed Maximum grasping pressure (g)
0.7~s 966 774 258
1.1 ~s 1092 804 Z76
3.3~s 1054 876 234
Grasping IOOvement speed Stress relaxation rate eX)
0.7~s 15.5 32 26
1.1 ~s 18.7 40 30
3.3~s 22.0 51 38
 980
Basic pattern
Internal MennO'
device

Video carrera Picture data ~PI/O

c<IIlPile pattern recognition -+ Systan control port
processing device device
calculate ~RS-23X
device port
,
'. __ ... _---------------------.----------------.------------.---------------
,

Lighting
Figure I. Composition of functional devices for
visualization processing.

CONCLUSION

We have studied the necessary technologies for FMS (Flexible

manufacturing system) of food processing. We then
manufactured a test machine capable of grasping, transfer
and positioning, and confirmed the usefulness of this system
by typical food products such as tofu and kamaboko.
These technologies can be applied to most of the handling
work at food production facilities which is performed
manually.

ACKNOWLEDGMENTS

This study was done at research and development of

Intellectualized Food Processing Technology Co.,LTD. which
was invested by The Japan Key Technology Center.
We are deeply grateful to the members concerned.

REFERENCES

I. K.Tanie, T.Fukuda and N.Kitamura, Flexible handling by

gripper with consideration of charactristics of objects.
PROCEEDINGS OF THE IEEE INTERNATIONAL CONFERENCE ON
ROBOTICS AND AUTOMATION, San Francisco, California, April
7-10, 1986

2. K.Hirota, Y.Arai and Y.Hachisu, Real time pattern

recognition and its application to robot control, Hosei
Univ. Techno. Report NO.23,P 117
 DYNAMIC MODELLING AND SIMULATION OF FOOD PROCESSES

J.J. BIMBENET, G. TRYSTRAM, A. DUQUENOY, F. COURTOIS

A. LEBERT *, M.L. LAMELOISE, F. GIROUX, M. DECLOUX
ENSIA, Food Process Engineering, Massy, France
* INRA, CR V, Theix, France

ABSTRACT

Based on the experience of the Food Process Engineering department at the ENSIA, several
ways for building a dynamic model of food processes are presented and discussed. Examples of
each approach are proposed.

INTRODUCTION

Food process engineering has three inherent objectives: to understand the phenomena which
exist during processing, to design unit operations and to control them. Modelling offers a good
opportunity to study and analyse the processes. Most of the models are steady state models, in
which the time is not considered. This is not sufficient when qualities of food and heat and mass
transfers are studied together. A dynamic model is often necessary because most of the
phenomena are quickly performed and kinetics and transformation are carried out coupled with
transport phenomena. The effects of time become important, and it is necessary to describe the
phenomena versus time and operating conditions. For process control studies, dynamic
modelling is also important. Chemical engineering takes dynamics into account and simulation
packages were developed for dynamic simulation (Speed-up, Process, Batches, ... ). In these
libraries, some food processes are included, but the lack of models for physical parameters is a
problem. Even if some parameters are available, variations with respect to water content,
temperature and food components are not included. It is therefore necessary to develop original
approaches. Different ways are available to build use dynamic models of a process. At the Food
Process Engineering department, at the ENSIA, we have developed some studies in order to use
modelling and simulation of food processes, specially for automatic control purposes. Several
approaches were developed and this paper is a summary of these approaches illustrated with
examples.

Three approaches are considered. First, the descending approach consists of the consideration
of fundamentals laws. From assumptions about phenomena, the model structure is developed
and validated with experiments. Different techniques are encountered. Often fundamental laws
are not easy to model and then hybrid or analogue models are developed. They are simpler and
easier to use with a computer. Secondly, knowledge is not always available. Then the ascending
approach is possible. The model is built from data and techniques like Residence Time
Distribution and identification of transfer function. Neural networks are also available for
982
dynamic modelling. Thirdly, for process control purposes, an approach based upon the events
that occur on the process is possible. Specific graphical methods like Petri nets or Sequential
Functions Charts are available.

EXAMPLES
Knowledge based model; Dynamic adjustment of retort time
To establish the time-temperature requirements to achieve the sterilization of a canned product,
one must run several pilot trials with variable retort times, using a trail-and-error procedure,
until the desired sterilization value is obtained. The heating time to be applied at the industrial
scale is deduced from the last trial. A computer programme as been developed by ENSIA and
CTCPA (Technical Center for Caning of Agriculture Products) to follow the evolution of the
temperature at the centre of cans placed inside a pilot retort. The interpretation of the evolution
allows a real time evaluation of the heat transfer characteristics j and fh, and also heat transfer
coefficients and apparent diffusivities of each can. Considering the can which appears to be the
more difficult to heat up, the programme predicts its thermal behaviour during the forthcoming
treatment: end of heating and also cooling. Based on a numerical simulation of the process, the
prediction forecasts to know what could be the best time to stop the heating phase. Thus it is
possible to proceed to trials that are very close to the optimal solution, saving time and giving a
better knowledge of the variability of the sterilization value around the objective value. The
programme is now being fitted into the hardware system for the temperature measurement and
sterilization value evaluation of the French firm COMEUREG which will commercialize it under
the name OPTIBAR.

Dynamic modelling and simulation of continuous chromatographic separations

Glucose and fructose separation from invert sugar may be achieved by adsorption
chromatography on a strong cationic resin under Ca++ form. Since UOP established the
principle of continuous chromatography in the 60s, glucose and fructose separation has been
one of the most studied applications. However, optimisation of design and operating
conditions, given quality and productivity requirements, as well as monitoring strategy are
important questions which remain be answered. We have studied each of the phenomena
responsible for this separation so as to build a predictive model of fixed-bed chomatography.
Adsorption equilibrium isotherms determined by frontal analysis over a large range of
concentration (0-400 gil) were found linear and independent. Mass transfer resistance was
shown to be essentially internal and to follow an homogeneous diffusion model, equivalent in
the case of a linear system to a first order law with time constant td. Flow pattern in 50 p. 100
cm long columns, was well represented by the axial dispersed plug-flow model, with high
peclet number. The chromatographic column transfer function could then be derived from the
MCE model (Mixing Cell in series with mass Exchanges). Numerical resolution was achieved
by the Fast Fourier Transform algorithm. Simulated curves were compared to experimental
ones, in the case of pulse response experiments, and of elution curves; they showed satisfactory
agreement. This linear fixed-bed chromatographic model has been used to simulate the
performances and the dynamics of a continuous separator, the Simulated Moving bed adsorber.
Both steady state and transient behaviour have been successfully investigated, for design and
monitoring purposes.

Analog model; Production of yeast from whey by fermentation

Production of yeast from whey is performed on an large scale air lift fermenter. Physiological
modelling of the culture is coupled with mass balance to simulate the dynamic behaviour of the
fermenter. Three species are combined in the fermenter. The description of the physiological
states of these species is much too difficult. Another approach is to consider an equivalent
culture, which is a theoretical one, with an analogous physiological behaviour. This equivalent
culture does not exist, the physiological states are functional states, and conditions of evolution
from one state to another are described for the limiting factors of the culture; substrate, biomass
and oxygen. From this description, a classical mass balance is developed to predict versus time
the dynamic evolution of each variable of the culture. The model is based upon knowledge
laws, but from a microbiological point of view, it is not a real model, because we create a
 983
species who is not a real one. The model is a description of functional states and not of real
states, but this approach permits the use of fundamental transport laws for mass balances.
Identification of unknown parameters is performed from literature and using optimization
procedure from experimental results (error: 7%).

Compartmental model: corn dryer modelling

Corn is, in France, the second agricultural product after wheat. The objective of this work is to
build up a simulation tool helpful for the design of dryers and control algorithms optimized with
regards to energy, grain flow and qUality. The influence of a thermal shock on wet-milling
quality is modelled. The quality equation thus defined is used in a dynamic model of com
drying based on a compartmental method (two compartments) and developed from the thin layer
to the industrial dryer. It permits the derivation of the characteristic equations for the dynamic
behaviour of the grain under the influence of air temperature and moisture. The model, adjusted
on drying kinetics under constant conditions, is used to predict the steady state of any dryer. It
allows the modelling of any transient phenomena happening in industrial dryers : jumps of air
temperature, drying breakdowns, cooling, condensation, air recycling. This model is also used
to predict the dynamic behaviour of dryers when a disturbance is applied and thus to test the
applicability of control algorithms. From the thin layer to the industrial dryer, simulations are
compared to experimental results. A 5% mean error on the predicted moisture content of dried
corn is obtained. The error on the wet-milling quality prediction is of the same order as the error
due to the experimental procedure followed.

Identification: case of drum dryer

Automatic control science proposes different ways for dynamic modelling. A very simple
structure is chosen and parameters are calculated from experiments. These experimental
approaches are very useful, and many software packages exist which have identification
algorithms. Such an approach is used for dynamic modelling of drum drying. Drum drying is a
complicated process in which moisture content of dried product depends on steam pressure
inside the drum, and drum velocity. Because the measurement of moisture content is established
(using a near infrared on-line analyser) it becomes possible to perform identification. The model
chosen is a recurrent one, moisture is modelled as a function of moisture at previous sampling
times and drum velocity at present and previous time. It is the equivalent of a second order
transfer function. This dynamic model is used to determine the design of a moisture controller.
This technique is a quite simple one, the model is linear and is applicable in the experimental
range used.

Neural networks applications

From 1985, artificial intelligence proposes new approaches for modelling. Use of neural
networks is one such technique. Because of its structure, a neural network is able to take into
account non linearities which are one of the characteristics of food processes. Several
applications are performed in our research department for product formulation, drying or
microfiltration. For example, for drying of solids with hot air a comparison between classical
approaches and neural networks is made. Characteristic drying curves are obtained on an
experimental basis, moisture evolution through time is observed depending on operating
conditions like air velocity, moisture and temperature. For corn drying, previous work proposes
different approaches. A neural network is built. Inputs are the operating variables and time.
Output is only the moisture of dried corn. Two hidden layers are used with respectively 3 and 4
cells. Four experimental curves are used for learning in the neural network, and five curves for
validation of the model. The network gives the same results as previous classical methods.
Error is small; 5% maximum. Neural networks are a good opportunity when no previous
knowledge about the structure of the model is available.

CONCLUSION
Differents ways are studied for dynamic modelling of food processes. Many different methods
exist. The design of a dynamic model for food processes becomes, specially when real time
measurement are available. Thus it is possible to study coupled phenomena in process control
applications such as, for example, transport and reaction phenomena.
 Neural Network control for extrusion cooking

K. Uemura, S. Isobe, A. Noguchi.

National Food Research Institute,
Tsukuba, Japan

ABSTRACT

Neural Network (N.N. for short) controller was designed and examined for extrusion
cooking. The feed-forward and feed-back loops of N.N. were found to simulate the
extrusion cooking. The control was realized by two kinds of N.N. that minimized the
output deviation from the target value. The dynamic controlling can be achieved in the line
of quick learning, feed-back-calculation of N.N. and renewed teaching data. N.N. will
open the way for the AI control of extrusion cooking.

INTRODUCTION

The Extrusion cooking is one of the most versatile food process owing to its capability to
convey, mix, homogenize, cook, gelatinize and denature of the food material, and to
perform many other conventional food manufacturing operations and even biochemical
and chemical processes. Nevertheless, the automation and control of extrusion cooker
faces several difficulties because of the complex and unclear interaction between the
operational parameters. Neural Networks(N.N. for short) are one of the Artificial
Intelligences. The new control theory is applied all the controller s of home electrical
equipments, because of the novel, the cost, the power, and the obscure. N.N. has strong
points that are automatic designing and tuning. Then the switch over from the Fuzzy to
N.N. is carried out. The practical aim of the present work was to apply N.N. to control of
an extrusion cooking, and thus reduce the problems associated with the construction of
fuzzy control systems.

The design of the Controller

Fig.1 represents an articulated extrusion cooking as a black-box which has three
input(control) ports and four output(sensing) ports. The Figure illustrates the same
 985
input/output set as N.N. with 10 neurons in the hidden layer. The example product
investigated was flat bread produced with a Clextral BC 21 twin-screw extruder. The
chosen input variables were mass moisture content, mass feed rate, and screw speed, and
the output variables product expansion ratio, main motor current, pressure at the die outlet,
and product bulk density. The neurons in the input layer transfer the input values within a
range of [0, 1] to the appropriate neurons in the hidden layer. Teaching data for the N.N.
consisted of 15 experimentally obtained data sets. Both the input and the output data were
normalized between the range of values [0, 1].

Output

Flower Feed Expansion

b·
Current
Water Feed

Presser b
Screw
Densi ty

a· a Input

Fig.l Block diagram of N.N. for Extruder Fig.2 Principle of the reverse calculation

Theoretical analysis of N.N. controller

we can calculate with the same routine. Fig.2 shows the principle of the reverse. If the state
of present stay at the point(a), then the output is (b). Then if you want to increase the
output (b) to (b l ), following the response curve, you can get the renewal input data(a l ) . The
result become as the renewal control factors.

Result
After 1000 times learning by B.P, the mean square difference decrease less than 1.0. The
results of learning were illustrated by 3-dimensional contour graphs as shown in Fig. 5. In
each graph the X-, Y- and Z-axes represented mass feed rate, feed moisture, and
expansion ratio, respectively, with one of the input variables, in this case screw speed, kept
constant at various levels. Whenever you set control condition to input ports of the N.N.,
after feed forward calculation, the result of the Extrusion cooking is appeared on output
ports. In other words, the calculation realizes the simulation of the Extrusion cooking
between the input and the output. The simulation has following tree merits. The first one is
a clearing the operation. The second one is working well in the training for beginner
operators. The last one is good for improvement of the controller. When the controller
running and learning in line at the same time, we must decide to add ,change or ignore the
new condition to the teachers sets and gradually improve the controller as follows.
986
z z

x x
(a) Hidden layer is consist of (b)Hidden layer is consist of
5 neurons. 20 neurons.
x-axis is Water, y-axis is Flower and z-axis is Expansion.
Fig. 3 3D-Response surface of N.N.
Multi Reverse calculation
If each input values are set on the middle point of a preset range, the expansion ratio for
example will be about 4.65 times. If a more expanded product would be desired, then the
reverse calculation, the controller will fined out the revised input set. In this case, we can
get the set of input parameters that are related to mark 4.99 times expansion. But, the
expansion ratio was never global maximum point but local maximum point. We used
complexed N.N. for avoidance the local maximum. Fig.3(a) and (b) shows the response
surface that are build up by few-hidden
layers node and large nodes

D
Data base eng i ne N. N. engine Fuzzy engine

D D
respectively. The more the number
increase, the more unevenness the
(P-IB EtherNet EtherNet
Eth .. Not to IS response surface change. When the
c=~ reverse calculation is executed with
few nodes N.N. in the above example,
the result is improved until the
expansion become 5.52times. After that,
next reverse calculation that uses large
number nodes N.N. is started from the
renewed point. Mter all, we get a set of
input parameters that is related to
5.65times expansion.

Fig. 4 Dispersed N.N. control system.

Conclusion
A neural network was constructed which could efficiently learn from prior extrusion
cooking data sets. The complexed reverse calculation of the N.N. makes it possible to
simulate and control the extruder simultaneously. We are currently working on the build
up the dispersed AI controller system (show in FigA) that are consists of Database, more
than one N.N. engine and Fuzzy engine. While they work independently, they work in
cooperation with each other.

REFERENCES
1. Linko, P., Uemura, K., Y.-H. Zhu and Eerikainen, T., Application of network models
in fuzzy extrusion control. Trans IChemE, vo170, parte, Sept. 1992, pp. 131-137.
 FUZZY TECHNIQUES FOR PROCESS STATE ESTIMATION

VALERIE 1. DAVIDSON, RALPH B. BROWN and GORDON L. HAYWARD

School of Engineering, University of Guelph
Guelph, Ontario NIG 2Wl Canada

INTRODUCTION

For most food and biological processes, control information is available in three forms:
continuous numerical data from sensors (e.g., temperatures, flowrates), discretely sampled data
from the quality control laboratory or at-line analysis, and the operators' intermittent
observations of process conditions and product characteristics. Operators usually express their
observations in linguistic terms, frequently using words that are unique to the process, and
which conventional automatic control systems cannot use as inputs. However, fuzzy set theory
allows quantification of linguistic descriptors and makes it possible to integrate the operators'
observations into the overall control strategy. In fact, a fuzzy logic/expert system controller
allows all types of inputs to be processed. In this sense, working in the fuzzy domain is
somewhat analogous to using the Laplace transform in conventional control. In the fuzzy
domain all arithmetic and logical operations can be performed on mixed input types, and
control rules that would make no sense in the traditional crisp numeric domain can be applied.
For example, inputs like "temperature is high" and "exposure time is long" may be combined
to yield the consequent "product is overheated". The resulting control decision, in this case
"increase the product flowrate" must then be defuzzified (i.e., inverse transformation) to
produce a numerical value for the setpoint change in flowrate.

Our objective is to develop computer-assisted control systems for processes that

integrate human judgement values with electronic sensor inputs. Operators are an important
link in the control strategy because their observations and interpretations are critical, and
frequently cannot be replaced with on-line instrumentation. Computer-assisted control
enhances the operator's decision-making process. In particular, the benefits are: recorded
process history, improved forecasting, more consistent control actions among operators, and
more flexibility for accommodating different products with the same process equipment. Two
examples are used to demonstrate these concepts.

MOISTURE CONTROL SYSTEM FOR A CONTINUOUS DRYER

This system was developed to improve the current statistical process control (SPC) for a dryer
on an industrial bakery line and to integrate the monitoring aspects of SPC with process
knowledge that could assist the supervisor in making control decisions. The current quality
control practice on this bakery line is to sample the product at the point of packaging, and
988
to analyse the moisture content at-line. Moisture values are plotted on a control chart which
shows the target moisture content (5.5 %) as well as the acceptable limits (4 to 7 %). Simple
rules are stated on the control chart to guide the operator in making decisions about the
process state, however some rules are vague and the operators are frequently so busy that
they only notice results that are clearly unacceptable. The SPC protocol is simply a
monitoring function. It is a tool to alert the operator to the need for control action but can not
provide any recommendations as to what change is appropriate. Not surprisingly, control
actions are not consistent among operators.

To be consistent with the SPC model, the numeric moisture values were transformed
into three variables: an error value (difference between target and actual moisture), a variance
estimate (the squared error for the last four observations) and an estimate of the derivative
error (a linear regression through the last four observations). The control logic was expressed
in the form of production rules. Two rule bases were compared. The fIrst set of rules were
based on proportional control action. The second rule set incorporated a derivative action.
The manipulated variable was the setpoint for dryer air temperature and in the initial
prototype equal changes are made across all three drying zones.

In the fIrst rule base, the control action was proportional but it was moderated by the
variance estimate. The measured variable (moisture content) was noisy. In our fuzzy
controller, proportional control action was recommended only if the moving variance estimate
became "medium" or "large". As the variance estimate increased, the gain of the controller
also increased. This control strategy has been tested during actual production on the bakery
line. Figure 1 shows the recommendations of the fuzzy production rules as well as the
operators' actions during a ten hour operation. Early in the test, the operator made a setpoint
change that was incorrect. The fuzzy rules catch this problem at the next observation and
continue to recommend lower dryer temperature until the operator makes setpoint changes
about half-way through the test. After these changes, the average moisture content moves
much closer to the target value for the remaining 5 hours of the test.

A discrete proportional control algorithm was also defIned based on the numeric value
of error. The controller gain (Kc = 3.0) was the average of the fuzzy gains for
medium variance (2.0) and large variance (4.0). A deadband of +/- 0.3 % error was also
included to be consistent with the fuzzy value of zero error. As shown in Figure 1, this
controller recommends frequent control actions, even during the last half of the test.

CONTROL SYSTEM FOR A BATCH-COOKING PROCESS

The control strategy for a batch process is quite different from that for continuous processes
because the batch process is never at steady state. Key process variables are monitored to
determine if the process trajectory is normal or abnormal. If it is abnormal, it is desirable to
make compensating changes as early as possible so that the desired product characteristics are
achieved at the end of the batch. If abnormal process dynamics can not be detected and
corrected, there will be undesirable batch-to-batch variation and possibly batches of product
that are completely unacceptable.

A control strategy was developed for a batch-cooking process in a pilot-scale

smokehouse. The process was typical in that heat-transfer conditions were not uniform over
the oven volume. This resulted in variable product heating rates with a range of normal
 989
8 r---------------------------------------------~
actual moisture target moisture operator temperature
-e--
iemperat~ ·rec·~m.nend: "(fuzzy) te"mpemi"ure" recommend," (P) ................. ................. . 150

6 ......................................................................................................................... .

--- - - 145
.......... !. ./\/. ........................ : .. ·····ft····················· ....................................... .
1 ~ ft."': .1:" : i!
~ I
il ".:"
•• ,
: ...··'·.i \. "i \ t·
:\.
:i
: - \ I.
.,,\ ".:j \!\!\~ \il
! \/\
'i \
i
i
\! \ i \ l H
Y V \: i\!
2 ........~.......
' ../'
+ ................................\~ ..-.·H~,....
i:.: .:::
.....-..,---...,.....,
A",. I
i: :I : '\
. ___...----\r-?,
i \ ; ,\ ,I 140
\::: ir·'/ 1Vi i! 'i\" \ .' :
i; \,' v
ii' \ r-"
S

~i
V :. \I
V V
o ~----~-------~--------~--------~------~
o 2 4 6 8 10
Time (hours)

Figure 1: Continuous dryer production test

process trajectories instead of a single heating curve. The product heating rate (at the centre)
was described by first-order transfer functions. Two time constants were defined at each fan
speed: 'tmin (fastest heating rate) and 'tmax (slowest heating rate). Time constants were
detennined empirically over a range of typical operating conditions. During cooking, internal
temperatures were monitored at several locations. The transfer function was used to estimate
nonnal temperature limits (based on 'tmin and 'tmax ) for the current process time. If the
measured temperatures were within these limits, the fuzzy estimator predicted the remaining
process time to ensure a minimum thennal exposure. A set of rules predicted the final
product quality (moisture loss and texture) based on the estimated range of temperature-time
trajectories. If the quality estimates were unacceptable, the control system suggested remedial
actions (changes in air temperature and/or fan speed). The control system also recommended
changes if the measured temperature(s) are outside the normal limits. The operator could
accept or ignore the recommendations of the control system.

This control system integrates conventional control techniques with fuzzy mathematics.
The process model is in a transfer-function fonn which explicitly dermes the effects of air
temperature and fan speed on heating rates, as well as the product temperature-time histories.
This model is fuzzy because, at each operating condition, a range of time constants defines
the possible internal temperatures. The second fuzzy component of the control system is the
prediction of product quality. Fuzzy variables are convenient for describing food quality
attributes in linguistic terms. Fuzzy rules are used to make inferences about the impact of
process changes on product quality. In these inferences, a rule-based model is acceptable and
more tractable than equation-based models.
 JIT AND CIM CONCEPTS, APPLIED TO A LARGE SCALE CATERING
PRODUCTION PLANT

TOON MARTENS
ALMA University Restaurants, Catholic University ofLeuven
E. Van Evenstraat 2 C, B-3000 Leuven, BELGIUM

ABSTRACT
Japanese assembly industry has shown that quality, cost and market flexibility can be
improved by applying the TIT-philosophy (material flow) and integration of computers, to
speed up information processing from machine control to strategic planning. The Catholic
University ofLeuven has build a new production plant for "sous-vide" meal components, that
tries to integrate the newest ideas in production management and food technology.
Although the general concepts of TIT and CIM were usefull, the implementation was very
difficult. This was due to the biological variation of ingredients, the wide variety of products,
the impossibility to automate certain tasks like tasting and lack of standardisation of hardware
and software.
This experience was the basis to setup new research programs on computer aided design,
computer aided quality control and intelligent equipment.

INTRODUCTION

Because of the high investment and operational cost of a new central kitchen that had to re-
place the classical warm kitchen, a study was done to identify the consumer needs and the
technologies that can meet this requirements. We found that the product design criteria that
were identified by different authors [1 ][2] were also true for the students and the personnel of
our university. The major concern was the variety. More than 70% of the students want each
day more than 5 menu items, about 20% would like to have more than 10 menus. The second
concern was the quality. Sensory quality and freshness were the most important criteria for
buying a product, especially the quality of vegetables was very important. Last but not least,
the price should be as low as possible. We conclude that the new production should produce
a wide variety of high quality products, free of additives, fresh and with a low cost. For
quality reasons we became interested in the "sons-vide" technique, that has been applied by
many French and Belgian chefs to improve quality in high class retaurants. We made some
small scale experiments and start thinking how we could apply this technique on an industrial
scale. We started discussions with researchers of the department of production management
about layout, automation and logistics.
 991
JUST IN TIME

Just In Time is not a well defined method or technique but a way of thinking. It is a set of
ideas that are usually described with a number of zero's and one 1 : "zero defects, zero
setup-time, zero breakdown, zero inventory, zero lead time and a lot size of 1".
In a HT-approach all kinds of waste are eliminated and special attention is given to factors
that influence market flexibility and/or productivity. Market flexibility is extreme in the
catering business : diversity of products and fluctuations in numbers are very high in com-
paraison with other sectors. Another characteristic is that the catering business is very labor
intensive and that productivity is underdevelopped due to the lack of planning and
automation. For this reason HI is very well suited for the catering industry. We found that
the following points were very important in the implementation of HT :

a) order and cleanliness

There is a direct relationship between the quality of the environment and the quality of the
products. For this reason we decided to build a new plant instead of renovation. No
intermediate stocks were allowed at different workplaces.
b) zero defects
In a HT quality program the main attention is not focused on the output of the production but
on product and process design. Quality problems should be avoided, this is certainly true for
microbial problems. We stopped control of end products and started with HACCP-analysis
and computer aided process design. Quality is in hands of the operator, who is also responsi-
ble for preventive maintenance and cleanliness.
c) uniform production load
HT tries to synchronize the rhythms of production with the rhythm of consumers needs. The
"sous-vide" system allows to level the peaks. HT environment tries to make the most out of
the available people and not of the machines. For this reason the personnel has to be multi-
skilled and automation will help them : "computeraided cooking".
d) layout
The layout of the building has been studied with different computer layout-programs
(CRAFT) to obtain a fast product flow, minimal space and minimal movements of materials.
Material handling of the finished goods has been automated to guarantuee FIFO and minimize
the work in cold storage.
e) reduction of setup times
For an operator it is impossible to remember all the settings of the machines for more than
1000 recipes. Also for safety reasons time-temperature of cooking is controlled by a process
computer.
f) production control
To deliver just-in-time a good planning system is needed that is also very flexible because
quantities are changing until the last minute. To optimize the usage of people and machines a
scheduling software has been written to plan the batches in the right sequence.
g) network of suppliers
Suppliers have to be selected that can deliver just-in-time but also can guarantee the microbial
and quality norms.

COMPUTER INTEGRATED MANUFACTURING

ClM integrates the information flow between the different departments of a factory : engi-
neering and product design, marketing and sales, purchasing, production planning and
control, maintenance [3].

a) computer aided process design

ALMA develops an expert system software for computer aided process design of minimally
processed foods as a part of an EC-research program (FLAIR AGRF 0047) The effects of the
992
changes in product formulation, process parameters (time, temperature, ... ) the microbial
safety and texture can be easily evaluated.
b) production planning and control
A Materials Requirement Planning software has been selected and adapted to our needs. The
main problems were the variability in yield of the raw materials, frequent changes in the
planning, substitution of ingredients depending on price.
A scheduling software has been written that tries to produce as much orders as possible
within the shelf-life limits and capacity limits. The sequence of the batches is optimised in or-
der to optimise the usage of the packaging machine, the bottleneck of the production.
c) computer aided manufacturing
To automate recipe handling a batch control software has been selected and adapted
(FERRANTI PMS). The machine steering was implemented on a PLC (SIEMENS). Since no
equipment was available that could be integrated in the batch control system, we had to
program the PLC ourselves. The pyramid of information processing has 3 layers : PLC, real-
time process computer and a UNIX-system at the highest level.
d) computer aided inventory control
Every batch has to be split in a different number of packs for each restaurant. Because of the
high diversity, collection of all products for a restaurant is very time consuming. Also for
safety reasons it is important that residence time in central storage can be controlled and
FIFO is guaranteed.

CONCLUSIONS

An attempt was made to implement the concepts of JIT and CIM as far as possible. The in-
stallation is now running in full production for one academic year and we could achieve a
substantial reduction of labour cost (ca. 50%). The changes in the recipes for the "sous-vide"
system were very consuming. Also the lack of standardisation in the way cooks prepare a
recipe caused many discussions. Software and hardware for recipe handling are not flexible
and standardised. No equipment was available that could be interfaced very easily. We found
that there is a lot of pep talk on JIT and CIM and that a lot of development is needed to make
this JIT and CIM-concepts working in the food industry.

This study is in part supported by the European Commission, Food Linked Agro-Industrial
Research Programme.

REFERENCES

[1] Fox, R., Plastic Packaging. The consumer preference of tomorrow. Food Technology
43 (12),84, 1989

[2] Keuning, R., Food ingredients for the 90's. Chapter 8 in Food for the 90's edited by
G.G. Birch, G. Campbell-Platt and M.G. Lindley, Elsevier Applied Sciences, London
1990.

[3] Scheer, A., CIM Computer Steered Industry, Springer Verlag, 1988
DEVELOPMENT OF NOVEL AGITATING DEVICE FOR BIOREACTORS

K. Maruyama, M. Ohmi, M. Imai, S. Urushiyama Department of Chemical Engneering,

Division of Chemical & Biological Science, The Tokyo University of Agriculture
and Technology. 2-24-16,Nagamachi, Koganei-C, Tokyo 184 ,Japan.

ABSTRACT

A novel agitating device called the "Cross Flow Agitator" was developed for
bioreactors. The change of flow direction with flap angles were observed.By change
of flap angles could successfully govern the flow direction even if the revolution
direction of the cylinder is the same. Such flow direction inversion character is
desirable to induce good mixing and circulation of multiphase fluid in bioreactors.
The one dimensional energy spectra in the vessel were measured. The energy
dissipation in the vessel was not localized compared with the case of turbine
impeller.

INTRODUCTION

Development of a novel agitating device for bioreactors is important for the

advancement of biochemical process systems. Agitation is effective for oxygen
solubilization 1) and homogenization of a culture medium, and it has been adopted
in various bioreactors for culture of animal and plant cells and for enzyme
reactions. Agitation, however, involves some practical issues such as damage of
cultured cells due to fluid turbulence 2 ) and formation of stagnant space in reactors.
In this paper, we propose a novel agitating device "cross flow agitator" for
bioreactors. Simple manner to change of circulation flow nearby the cross flow
agitator and energy dissipation in the agitating vessel were investigated
experimentally.

EXPERIMENTAL

Flow direction in the vessel

The schema of the experimental apparatus is shown in Figure 1. The agitating
device was composed of the cylinder of cross flow agitator and two flaps. The cross
flow agitator was composed of fifteen small wings. The cylinder of cross flow
agitator was installed horizontally in the vessel. The diameter and axial length of the
cross flow agitator were 100 mm and 296 mm respectively. The dimension of the
agitating vessel was 224 mm(W), 230 mm(H) and 347 mm(L). The change of flow
994
direction was seen using tracer particles which were photographed with a camera.

Measurement of one-dimensional energy spectra

A schematic diagram of the experimental apparatus is shown in Figure 2. The
diameter and axial length of the cross flow agitator were 100 mm and 130 mm
respectively in the vessel. When energy spectra was measured, the cross flow
agitator was set up vertically, since accurate power consumption in the vessel could
be measured simultaneously. The flow rate was detected by a probe electrode under
the diffusion-controlling conditions. An autocorrelation function was calculated
based on change of flow rate. The energy spectra were calculated with a Fourier
transform of autocorrelation function. For comparison the case of a six-blade
turbine impeller, the power consumption of the cross flow agitator was adjusted to
be equal to that of the turbine impeller. The energy spectra were measured both in
the axial and radial directions.

l :cross flow :tgitator

2:probe
3:Pt electrode
4:motor
S:bridge circuit
6:squarer circuit
l:cross flow agitator 7::tmplifier
2: flaps r---'----, 8:cllmputcr
3:agitating vessel
4:motor

Fig.l Experimental apparatus Fig.2 Experimental apparatus

The flow direction was observed
in the direction of the arrow

RESULTS AND DISCUSSIONS

Flow direction in the vessel

Flow direction in the vess el us ing tracer
particle are shown in Figure 3. The
key (0 0 , 40 0 ) express the flap angle of
cross flow agitator as shown in this
Figure 3. the tracer particles were introd-
uced at the particular flank of the cylind-
er and exhausted out from the cylinder.
According the loci of particles, the fluid
was introduced into the cylinder along
the small wings of cylinder. From this
motion of particles, shear stresses dur-
ing agitation will be reduced. The flow
direction nearby the cross flow agitator
was changed drastically with flap angles
even if the direction of cylinder is the Fig.3 Loci of tracer particles in the vessel.
same. This change in flow direction Flap angles (0',40') , (40',0')
 995
is desirable to suppress the formation of stagnant space in the vessel.

Energy spectrum
The spectral distribution in the case of impeller used, presented in Figure 4,
apparently changed in the axial direction. On the other hand, the distribution of the
cross flow agitator slightly changed in the axial direction presented in Figure 5. In
accordance with these results, the large energy dissipation in the local region did
not occur in the agitating vessel, especially near the cross flow agitator.

, ,
10 I 10

, II
10 t--- 0", _ - Measuring point r- ,
-
10 Measuring point -

ch~
1----1- " r-

\
-

"
-
10°
~~~
1--/'""\- .~ -

P'\\
-
I , .
~ ~
~ "-
Hj' o~ ~
'"°0,
'?~ ~

~
\
°~
Hi' Hj' ~
~
~
\

10'
1\
Hj' 10' 10'
10° 10' 10' 10'
k[cm]
k [cm]
Fig.3 Energy spectra in turbine impeller
FigA Energy spectra in cross flow agitator

CONCLUSIONS

A novel agitating device "cross flow agitator" was demonstrated. The shift of flow
direction and the one-dimensional energy spectra in the agitating vessel were
presented. The direction of flow near the cross flow agitator was changed by the
flap angles, and this appears to suppress the formation of stagnant space. The
energy dissipation in the vessel was not localized as with the turbine impeller.
These characteristics are expected to reduce the biological damage due to shear
stresses during agitation. The cross flow agitator shows the desirable characteristics
for a practical bioprocess agitating device.

REFERENCES

1 Asai, T. and T. Kono , Estimation of oxygen absorption coefficient and power

consumption in a stirred tank fermenter. J.Ferment.Technol.,1982,60,265-268
2 Kawase, Y., Design for bioreactors. Chem.Eng.(in Japanese), 1987,7,646-650
TOTAL ENERGY MANAGEMENT SYSTEM (TEMS) AT FOODSTAFF FACTORY
MODERATED ENVIRONMENTAL IMPACT PRODUCTION SYSTEM

SHIGERU SAKASHITA
Deputy ManagerlEnergy Management Block
Mayekawa Mfg.Co.,Ltd.
13-1,Botan 2, Koto-ku, Tokyo, Japan

ABSTRACT

In modern society, energy saving is an important theme in the operation of any production
facility, as is the importance of limiting environmental pollution caused by release of C02.
NOx and SOx into the atmosphere. Maximizing utilization of primary energy is a key
solution to the above problems. Effective utilization of primary energy at a factory begins
with the introduction energy-saving production engineering, for example,
• Saving input energy by recovering waste heat,
• Improving energy utilization efficiency by introducing co-generation
• Storing energy by some means such as thermal storage
All the above require instrumentation systems that assure that the hard ware used functions
effectively. The above engineering is consistent not only with the energy saving goal but also
the economic objective of a factory, that is ,to minimize the cost of energy used in production.
The development of TEMS ( Total Energy Management System)has been based on full
consideration of the above and fully operational systems are actually in use at a beer brewery
, a sake brewer and a ham production plant. The following can be mentioned as common
characteristics of the above foodstuffs plant:
• Imbalance in daytime and nighttime energy load,
• Use of various energy sources such as electric power, steam, warm water and cold
water(cold heat),
• High sensitivity to quality control aspects such as taste and sanitation.
The case of factories characterized by the above, the management of energy flow and storage
using combinations of energy saving devices closely connected to the production process and
the use of thermal storage units and computerized instrumentation systems greatly enhance
efficiency.

What is TEMS

TEMS is the abbreviation for "Total Energy Management System. "As the name implies, it is
a system aimed at achieving comprehensive andeffective utilization of energy.
We call the system "comprehensive" because it covers the entire factory and also the
management of various utilities such as electricity, steam, warm water and cold heat. As
well, TEMS also involves the total flow of production, transport, storage, consumption and
recovery of energy. TEMS save energy, recover waste heat and stock and or discharge
utilities based on estimated energy consumption levels. Such management is achieved by of
AI(artificial intelligence)and a heat accumulator.
 997
Basic Policy of TEMS

Basically, TEMS consists of the following three factors:

1) Bringing a factory close to an a closed energy system by recovering and reusing waste
heat.
2) Expending effective utilization of primary energy through co-generation.
3) Shifting and effectively utilizing energy and equipment through use of a heat
accumulator and AI.

Closed system orthodox system

Vapor re-compression

Heat pump Boiler System

Figure 1. Closed energy System

Figure 2. TEMS for Brewery

 998
TEMS Instrumentation and Control System

Nonnally,1EMS uses the control system shown in Figure 3.

The Process control computer and the equipment sequence controller are independently
mounted. Here, process control based on production management as well as sequence control
and loop control based on feedback are carried out.
The system is basically a local control structure.
TEMS is controlled by a work station level computer. Optimum operation is determined upon
receipt of production planning data and process information input into the process computer
through signal communications. The value settings are then dispatched to the sequence
process controller. The set values are renewed every minute to one hour.
The TEMS control system has the following four functions:
1) Utility load forecasting(energy demand forecast) ,
2) Optimum operation control,
3) Abnonnality diagnosis,
4) Statistical analysis.

[Q]
1 -I
. - - - - - -Work station- - - - - - ,

AI
demand estimation
I Air temperature I -~

--
r - - - - - - - , , - - - - - - - - - - , ,,/.,(
~
steam decision , Sequence controller
,-------,' ice energy
I Production schedule I,~ on
warm water --to optimum - ,~: ~p:r~ti~n _m~d~ ,, '----_co_n_tro_ll_e_r
,selection of and 1 loop
power _ _---'
condition
1other data 1- ~ others
, ~

,,
,

Statical Analysis of Plant data

and Abnormality diagnosis

Figure 3. TEMS Control System

Conclusion

The total energy management system introduced here with first developed in 1987. Its high
energy efficiency is unparalleled. The TEMS controlling computer system serves an
important role in shifting energy. Problems do remain which must be surmounted, the most
important being man/machine type problems.
Since the ability of the computer to communicate details of optimization and abnormality
diagnosis to human beings is insufficient at present, the result is those high performance
computer functions as a sort of 'black box.' The ironic result of this is that it may fall to
attract the interest of human beings.
At present, the main development theme is improvement of the man/machine system to a
capability level that takes advantage of the expressive power of the computer.
In closing may say that if this paper provides the reader with any useful advice, I would be
most pleased.
A BLACKBOARD AND OBJECT ORIENTED FERMENTATION PLANT MODEL

VIRGINIE NIVIERE
Imeca, BP 94, 34800, Clermont-L'Herault, France
PIERRE GRENIER, JEAN-MICHEL ROGER, FRANCIS SEVILA
Cemagref, BP 5095, 34033, Montpellier cedex 1, France
MORAD OUSSALAH
EerielLeri, parc scientifique Georges Besse, 30000, Ni'mes, France

ABSTRACT

A model of simulation of wineries has been proposed. Each elementary set of knowledge used
in the model has been defined as an entity gathering constant data, variable parameters, and
expert calculation functions. The model has been structured hierarchically, and the interest of
blackboards has been discussed. It has been implemented in object oriented programming and
the language C++ has been chosen. This expert simulation software produces dynamic balances
of resources.

RATIONALE

During the period of grapepicking, Enologists have to manage resources as varied as humam
power, grapes, process equipments, and refrigeration power. Refrigerating grapes is a
necessity in elaboration of most quality wines. A simple idea for optimising the use of these
resources is the dynamic simulation of a winery operation. We have developed a knowledge
based system with a hierarchical model of objects for simulation of wineries operation [1, 2].

RESULTS

Design of the model

Modeling: Wine-making follows one or several process lines, each one being a
succession of operations such as, for instance, grapes reception, pressing, thermal transfer,
settling, centrifugation, maceration, fermentation, or storage. Each operation gathers a set of
units making the same kind of transformation on the materials. We have imagined a hierarchy
between units, operations, and process lines, with respectively three levels of abstraction. At
1000
the first abstraction level, a set of knowledge has been defined with six objects, respectively
Process-Unit, Input-Flux, Output-Flux, Material (processed by the unit), Phenomenon (taking
place in the unit), and Equipment (container of the unit: a tank, an exchanger, ... ). At the
second level of abstraction, we have defined the objects Operation, Operation-Input-Flux and
Operation-Output-Flux. At the third level of abstraction, we have the Process-line, containing a
list of operations.

Dynamic simulation: The knowledge described at the first level of the hierarchical
model is a discret event system. A continuous system may have the activities "On" and "Off".
A discontinuous system may have the activities "Standby", "Refilling", "Storage", "Draining".
The logics of state variables evolution of the objects of a system rely on its activity regulated
by a set of functions.

Blackboard: for further extension of our knowledge based system, we have designed a
blackboard architecture. One knowledge source, which contains the hierarchical model already
implemented writes lists of tasks to achieve on an object called the Blackboard , and another
knowledge source at present time under development, will work on these lists.

Validation
At Saint-Genies-Ies-Mourgues (France), we have studied a process line for red winemaking
where the alcoholic fermentation is split in two operations, maceration without temperature
control, and fermentation in liquid phase with temperature control.

-t ...., ........
Variable values at
:
TIME
~ 23tb
a
/III
VAT NUMBER
~ 26

a
RE~GERAnONNEEDS
44842 kcal/h
~
/III
~

~
=
~
~
IIC
/III

~ Curser
:
to< Uc:del
nME (h)
1 360 720

Figure 1. Example of computer output. Simulation of vats and refrigeration needs at the
Cooperative winery of Saint-Genies-Ies-Mourgues (France)
during the 1991 harvest.
 1001
We have estimated the fluxes of grapes and musts through the various operations and
units, and we have compared the computer simulation to these estimations. We could observe
that the material fluxes were well simulated: the number of vats simulated was equal to the
observed number by one unit, and this difference was due to the assumption that maceration
was three days long when actually it was fluctuating. An example of computer output is
presented on Figure 1.

PERSPECTIVES

We want to simulate more precisely the planification of the tasks. We can make the knowledge
based simulation work by cycles of 24 hours. While the simulation is run a first time for a given
day, the list of tasks to be achieved on this day is written and executed on a provisional basis.
Then this list of tasks can be ordered, the simulation run for the second time, and the modified
list of tasks executed. Tasks planification is a problem of Flexible Constraints Solving, and
adequate theories are applied [3].

CONCLUSIONS

Dynamic simulation of equipments and refrigeration power requirements in wineries makes an

excellent optimisation tool in the winery. Flexible Constraints Solving associated with fuzzy
logics will permit in a near future to improve the simulation.

ACKNOWLEDGEMENTS

We thank Mr Orange, Director of the Winery of Saint-Genies-Ies-Mourgues (France), for

welcoming us during the 1991 grapepicking, and Dr Lopez, Professor at the University of
Lerida (Spain), for his help in the model validation.

REFERENCES

1. Grenier, P., Feuilloley, P., Sablayrolles, J.-M., Development of an expert system for the
optimization of the wine alcoholic fermentation. International Conference on Agricultural
Engineering, March 2-5, 1988, Paris, European Society of Agricultural Enginneers, paper
n088.399.

2. Niviere, V., Grenier, P., Roger, J.-M., Sevila, F., Oussalah, M., Intelligent simulation of
plant operation in the wine industry. Journal of Food Control. In revision, 1993.

3. Dubois, D. and Prade, H., La theorie des possibilites. Edition Masson, 1988.
 TREATMENT OF SWEET-POTATO PROCESSING WASTEWATER USING UASB
REACTORS. I - START-UP OF THE PROCESS

DE MARIA, M.O.; DEL BIANCHI, V.L. and MORAES, 1.0.

Department of Food Engineering and Technology - UNESP
P.O. BOX 136, 15054-00, Sao Jose do Rio Preto, SP-Brazil

ABSTRACT

Three different DASB reactor designs were evaluated at ambient temperature, to determine the efficiency of
COD removal, volatile acidity, total alkalinity and pH, with distinct initial sludge concentration and
diverse pH and nutrient correction solutions. Each reactor was operated with anaerobic sludge originating
from a VASB plant of an orange juice factory. At the end of each operation phase of the reactors, an
average efficiency of COD removal greater than 80% was obtained.

INTRODUCTION

The DASB concept was developed by Lettinga and coworkers (1) in 1971, and showed the potential of
anaerobic digestion systems using DASB reactors, leading to satisfactory results with industrial
wastewaters from different sources. The fundamental property of a DASB reactor is the wastewater passage
through a high concentration region of anaerobic microorganisms, called a sludge blanket, in upflow.
This way, the wastewater suffers degradation, with the biodegradable components converted into biogas.
This biogas is separated from the treated effluent for later elimination or utilization as energy source.
In this work, the start-up of three distinct reactors, with some design differences, was studied. Each
reactor was used for the treatment of wastewater from the sweet-potato starch extraction laboratory
process.

MATERIALS AND METHODS

The reactors used in this work were called DASB I, DASB II and DASB III. VASB I reactor was built
of plastic material, with a useful volume of 2.5 liters, in cylindrical shape. DASB II and DASB III
reactors were made from glass, with square section and volumes of 3.4 and 4.7 liters, respectively. They
have some differences in the design of their phase-separator systems. To control the treatment process and
to verify its efficiency, some periodic analyses were done, including COD, pH, total alkalinity and
volatile acidity. These were done in the affluent and in the effluent of the reactors, according to the
methods described in the literature (3).

RESULTS

Based on the medium characteristics of the brute wastewater (COD = 5880.0 mg/l; pH = 5.40; total
nitrogen = 157.4 mg/l and total phosphorus = 153.2 mg/l) and using the optimum relation of 100/5/1
(COD/N/P) observed by Patza et al. (2), it was verified that correction of the nitrogen ratio and pH of the
system affluent was needed. At first, this was done with NH4HC03 and Ca(OH)2 solutions, respectively.
1.2 liters of sludge were inoculated into the DASB I reactor (48% of its volume) and the operation
began with a hydraulic retention time (HRT) of 24 hours, until the 16th day of operation, when it was
reduced to 18 hours.
 1003
On the other hand, one liter of the same sludge was inoculated into the UASB II reactor (27% of its
volume) and the operation began with an HRT of 24 hours, until 38 days, when the HRT was reduced
again, now to 12 hours. In this phase, the affluent nitrogen ratio was corrected with Nl4HC03, and the
pH was Ca(OH)2/NaOH.
One liter of anaerobic sludge was inoculated into the UASB III reactor (21 % of its volume). The
nitrogen ratio and the pH were corrected at the start of the process, with NH4HC03 and Ca(OH)2. The
operation began with an HRT of 24 hours until 18 days, when it was reduced to 12 hours, until 40 days.
At the end of the 55th day of operation, it was observed that the total alkalinity of the system was falling,
a fact which reflected the global efficiency of the system. Considering this, it was decided to correct the
pH and nitrogen with a solution containing NH4HC03/Na2C03/NaHC03. This change caused a rise in
the global efficiency of the system, in terms of COD removal, from 68.3% to 86.7%, three days after
changing this parameter. The average results obtained for all reactors are presented in Table 1.
The initial sludge concentration in the reactor must also be considered. A high concentration results
in a higher system efficiency, because of the presence of a higher number of active cells in the reactor.
Table 2 shows these data.

TABLE 1.
Average COD affluent, COD effluent and COD removal data for each operation phase ofUASB 1, UASB
II and UASB III reactors.

Average value UASBI UASBII UASBIII HRT

COD affluent (mg/!) 1248.9 1843.5 3754.8
COD effluent (mg/!) 117.0 256.9 589.5 24h
COD removal(%) 90.3 85.4 82.1
COD affluent (mg/l) 1277.5 1267.3
COD effluent (mg/l) 237.5 241.3 18h
COD removal (%) 80.7 79.8
COD affluent (mg/!) 1841.8 1616.0
COD effluent (mg/!) 334.7 323.1 12h
COD removal (%) 82.0 79.9

TABLE 2
Influence of initial sludge concentration on the average COD removal, for the three reactors, with an HRT
of 24 hours.

Reactor Initial sludge concentration Average CD removal

UASBI 48% 90.3%
UASB II 27% 85.4%
UASB III 21% 82.1%

CONCLUSIONS

For all reactor start-ups, a fast adaptation was observed of the sludge bacteria to the new substrate used,
and a rapid changing of process parameters, such as organic loading ratio, HRT and correction solutions,
to pH and nitrogen ratios. This fact shows the versatility of the system, considering its adaptation
capacity to new conditions and its application to different kinds of wastewaters.

REFERENCES

1. LETTINGA, G.; KLAPWUK, A. and HOBMA, S.w. - Use of upflow sludge blanket (USB) reactor
concept for biological wastewater treatment, especially anaerobic treatment. Biotechnology and
Bioengineering, 22. pp 699-734,1980.
1004
2. PATZA, M.G.B.; PAWLOSKY, U. and GABARDO, M.T. - Tratamento anaer6bio de vinhoto de
mandioca pelo reator de leito de lodo granulado. In: 12th Brazilian Congress of Environmental and
Sanitary Engineering, 1983. 32p.

3. Standard Methods for the Examination for Water and Wastewater. American Public Health
Association, 1965. 769p.
 TWO STAGE CONTROLLED ANAEROBIC BIO-REACTOR FOR DIGESTING MIXED
DAIRY WASTE

C.L. Hansen*, S.H. Hwang

Department of Nutrition and Food Sciences and Biological and Irrigation Engineering, Utah State
University, Logan, Utah, 84322-8700, USA

ABSTRACT

As part of an effort to optimize each phase of a multi-phase anaerobic fermentation, a cheese

processing waste mixture (chemical oxygen demand (COD) ;;;: 25,000 mg/L) was digested in
batch fermentations to optimize acid formation. Three different influent concentrations (5800,
10,000 and 17,000 mg soluble COD (SCOD)/L); n= 10 for each concentration), were used in the
acid forming fermentations. A mixture of ethanol (0.1 M) and acid solution including: acetate
(0.7 M), propionate (0.1 M) and butyrate (0.2 M), (SCOD = 13,300 mg/L) was digested in the
batch methane fermentation experiment (n=8). The acid forming fermenters produced the
maximum amount of volatile organic acids (VOA) in about 30 h. The production of acids was
directly dependent upon influent concentration. The maximum production of each acid in mg/L
was: acetate, 3000, propionate, 2500 and butyrate, 3200. The methane fermenter removed about
70% of the influent SCOD in 20 d.

INTRODUCTION

Fermentation involves a series of catabolisms to produce methane which can provide fuel for
many unit operations. First, one group of anaerobes, the acidigens produce volatile organic
acids (VOA) from complex organic material. The VOA is utilized by a second group of
bacteria, the methanogens to produce methane. Some research has been done on biological
production of VOA from biomass (l, 2) and the results are encouraging, but this data is not
sufficient to design an optimal fermenter. The purpose of this research was to learn more about
production of VOA in the acid forming stage of fermentation of cheese processing waste.

MATERIALS AND METHODS

Acidogens: Cheese processing waste, with composition as given previously (3) was
obtained from Gossner Foods, Logan, Utah in one batch during November, 1992 after which it
was divided into smaller portions and frozen at ~-25°C for use later on. This was thawed just
prior to use and diluted with distilled water to three different concentrations (5,800, 10,000 and
1006
11,000 mg Soluble CODIL) used in the non-mixed batch reactors. Thirty fermentations (125 mL
serum bottles; T= 35°C)were run for the acid forming stage; ten each for each influent
concentration. Two N NaOH was added when necessary to adjust the pH to 6.0 in the
beginning of the experiment.
Methanogens: An acid solution and ethanol solution ( 0.1 mole acetate, 0.1 mole
propionate, and 0.2 mole n-butyrate, 0.1 mole ethanol; SCOD = 13,300 mglL) similar in
composition to that used in separating the bacterial strain was tested for the methanogenic phase.
Eight batch fermentations (125 mL serum bottles; T=35OC, pH adjusted to 1.0) were run. A
concentrated nutrient and trace mineral solution was added to both phases to give a final
COD:N:P ratio of 500:5: 1 (4, 5). The mixture of substrate and microbial culture was put in the
125 mL serum bottles.
Sample Analysis: Soluble COD was measured by the ampule method (6). Solids
concentration was determined according to Standard Methods (1).
A Varian gas chromatograph (series 2100) with Carbopac~ column (Supelco Inc., 1-1825)
and thermal conductivity detector was used to determine acetate, propionate and n-butyrate
concentration.

RESULTS AND DISCUSSION

Figures 1-3 show the results of the acidogen experiments. Maximum concentration ofVOA was
achieved in about 30 h. After this same length of time, pH had stabilized to 4.8-5.0, volatile
suspended solids production and soluble COD were also constant (data not shown). Maximum
organic acid production for all acids was dependent on influent COD concentration. The
experiment was designed to have the three COD concentrations set up in a ratio of 1:2:3: final
COD values deviated slightly from these ratios due to difficultly of making exact COD
concentrations with actual industrial waste. The amounts of acetate, propionate and butyrate
produced followed about the same ratios in proportion to the influent SCOD. Experiments are
continuing to determine the kinetics of the acid forming stage.
Soluble COD destruction and biomass production for the methanogenic reactor is shown in
Figure 4. About 10% of the influent VOA and ethanol was removed in this less than optimum
fermentation. Work is also continuing in production of methane from VOA in our laboratory;
we have found that in a continuous fermentation of VOA to methane in a complete mix reactor,
~95% of VOA can be removed (data not shown).

3000

~
co
§,2000

j
< 1000
A 5.8 gCODIl A 5.8gCODIl
10.0 g COD It • 10.0 g CODII
17.0 g CODII --0- 17.0gCODIl
O~~-.~-,--~r-~~~~
30 60 90 120 150 o 30 60 90 120 150
Period (hrs) Period (hrs)

Figure 1. Acetate production versus time. Figure 2. Propionate production versus time.
 1007
14.0 650

3000
12.0
600

~10.0 2
'-'
C\ SCOD Ef
550 ........
0 VSS
C) 8.0 fIl
fIl
en
>
500
6 5.8 g COOII 6.0
• 10.0 g COOII
--0- 17.0gCOOII
4.0 450
30 60 90 120 150 0 10 20 30
Period (hrs) Period (day)

Figure 3. n-Butyrate prodution versus Figure 4. Soluble Chemical Oxygen Demand

time. reduction and biomass production with time
in the methane forming batch reactor

REFERENCES

1. Chiruvolu, C. and Engler, C.R. 1992. Biomass Conversion to Organic Acids by Rumen
Microorganisms. American Society of Agricultural Engineers paper No. 92-6566, American
Society of Agricultural Engineers, St. Joseph, MI.

2. Kisaaiita, W.S., Pinder, K.L. and Lo, K.V. 1987. Acidogenic Fermentation of Lactose.
Biotechnology and Bioengineering, 30: 88-95.

3. Hansen, C.L, S.H. Hwang. 1992. Two Stage Anaerobic Digestion of Cheese Processing
Waste. Am. Soc. of Ag. Engr. Paper # 92-6605, presented at the ASAE Winter Meeting,
December 15-18, Nashville, TN.

4. Zeikus, J. G. 1977. The biology of methanogenic bacteria. Bacteriological Reviews.

41(2):514-541.

5. Stronach, S. M., Rudd, T., and Lester, J. N. 1986. Anaerobic digestion processes in
industrial wastewater treatment. Springer-Verlag press. Berlin: New York.

6. Adams, V. D., Cowan, P. A., Pitts, M. E., Porcella, D. B., and Seierstad, A. J. 1981.
Analytical procedures for selected water quality parameters. Utah Water Research
Laboratory, Utah State University, Logan, Utah.

7. APHA-AWWA-WPCF. 1989. Standard methods for the examination of water and

wastewater. 17th Edition. American Public Health Association., Washington, D. C.
 ACTIVATED SLUDGE TREATMENT OF MEAT PROCESSING WASTEWATER

A.P. ANNACHHATRE AND S.M.R. BHAMIDIMARRl

Department of Process and Environmental Technology
Massey University, Palmerston North, New Zealand

ABSTRACT

Feasibility of aerobic activated sludge process for treatment of meat industry

wastewater has been demonstrated through the operation of a model activated sludge unit.
High degree of nitrification and COD removal of more than 85% for loads upto 3.2 kg
COD/(m3.day) were achieved for reactor operated at SRT values of 3-13 days. Reactor
operation at SRT of 6 days or more was highly stable and yielded near complete nitrification.
Under these conditions aerobic treatment is likely to be more attractive than the conventional
anaerobic lagoon treatment.

INTRODUCTION

New Zealand meat industry processes up to 40 million animals each year. A typical meat
processing unit in New Zealand can produce wastewater upto 10,000 m3/day with a pollution
load equivalent to a city of around 200,000 inhabitants(1). A common practice in New
Zealand for meat waste disposal is primary treatment followed by effluent irrigation (2). Many
researchers have studied COD removal from meat waste through aerobic treatment (1,2) but
no conclusive data on concomitant nitrification is available. Accordingly, this research
demonstrates feasibility of aerobic activated sludge treatment for meat processing wastewater
through the operation of a model activated sludge unit for COD removal and nitrification.

MATERIAlS AND METHODS

Wastewater and sludge source:

Primary treated meat wastewater (Weddel Feilding, Feilding, New Zealand) was settled
overnight to remove suspended solids and the supernatant was then used for further
experimentation. Mixed liquor from an activated sludge treatment plant treating fellmongery
effluent was acclimated to the slaughterhouse wastewater in a continuous reactor. Sludge from
this reactor served as seed for all chemostat and activated sludge experiments.
 1009
Aerobic activated sludge process:
Experimental setup essentially consisted of fresh feed storage tank, 15 litres Brunswick
aemtion tank, sludge settling tank and pumps for fresh feed, sludge recirculation and sludge
wastage. 1:1 diluted fresh feed with COD 600-750 mg/l was fed continuously to the aeration
tank while the effluent from the reactor was withdrawn continuously and fed to the settling
tank. The settled biomass was recirculated back into the reactor, while overflow from the
settling tank was discarded. Sludge recirculation ratio of 1:1 was maintained throughout the
investigation. Sludge was withdrawn continuously from the reactor and wasted in order to
control Solids Residence Time(SRT) in the reactor. Investigations were performed with SRT
of 13, 9, 6 and 3 hours at fixed HRT of 10 hours in the aeration tank. Finally, reactor
performance was assessed at 5 hours HRT and 6 days SRT. Adequate air supply was
maintained to the aeration tank so as to maintain DO concentration of about 6.5 mg/l. Fresh
feed and reactor contents were analyzed daily for COD, NH4-N, N0 2 -N and N0 3-N. Besides
TKN and fat content of the reactor feed and effluent were also analysed occasionsly.

RESULTS AND DISCUSSION

COD removal:
Kinetic parameters K., !In., Y and kd were estimated as 176 mgCOD/l, 1.14 day·l, 0.4 mg/mg
and 0.043 dati respectively from continuous chemostat experiments (2). Figure 1 shows
influent COD ranging from 650-750 mgll while that of effluent to be always less than 100
mg/I, yielding COD destruction always over 85%. Figure 1 also shows COD removal rates
achieved. Highest COD destruction rates in the range of 3-3.2 kg COD/(m 3.day) are recorded
for HRT and SRT of 5 hours and 6 days respectively.

r-----------------------------~~

40

Figure 1 : COD and NH4-N concentration

Nitrification:
Figure 1 shows the influent NH4-N concentration ranges from 18-35 mg/l. The effluent NH4-N
concentration is remarkably steady at less than 6 mg/l. Likewise, N02-N buildups upto 10
mg/I and high effluent N03-N concentration in the range 15-35 mg/l (Figure 2) also confirm
concomitant nitrification activity during activated sludge treatment. However, Figure 1 and
Figure 2 also bring out that reactor performance with 3 days SRT is rather vulnerable to
1010
40,-------------_-.4o

8
:;:;20
o 20
'-
.oJ

~ 10 10

Effluent N02-

~~--~2O~--4~0~--OO~---8~0~-~I~
Days of Operation

Figure 2 : Effluent N02-N and N0 3-N concentrations

drastic variations in influent NH4-N concentration, is more prone to higher NOz-N buildups
(upto 5-7 mgll) and requires more time to stabilize to lower N0 2-N levels thus confirming
tendency for incomplete nitrification. In contrast reactor performance with 6, 9 and 13 days
SRT is much more stable with low N02-N buildups (always less than 5 mgll) and consistently
higher N0 3-N concentration (22-35 mg/l) in comparison with SRT of 3 days (15-25 mgll),
thus signifying near complete nitrification.
These data suggest that although the reactor operation yielded satisfactory COD
removal at 3 days SRT, its nitrification performance was vulnerable to variations in influent
feed. In contrast, reactor operation at SRT of 6 days or more yielded high COD removal with
steady and near complete nitrification. Figure 1 and 2 also show reactor performance at HRT
of 5 hours and SRT of 6 days. For these operating parameters, both COD removal and
nitrification are satisfactory. COD removal above 85% with loads upto 3.2 kg COD/(m3.day)
are achieved with low NH4-N and N02-N concentrations (always well below 5 mgll) and high
N0 3-N concentration (upto 33 mgll). TKN analysis of feed and effluent also confirmed more
than 90% removal.

CONCLUSIONS

This study establishes that aerobic activated sludge treatment of meat industry wastewater is
an effective way of COD removal and nitrification. Reactor operation at 6, 9 and 13 days
SRT is much more stable and yields near complete nitrification as compared to reactor
operated at 3 days SRT. Under theses conditions, aerobic treatment is likely to be attractive
than conventional anaerobic lagoon treatment.

REFERENCES

1. Bhamidimani, S.M.R., Appropriate industrial water management technologies : The

New Zealand Meat Industry. Wat.Sci.Tech.• 1991, 24(1), 89-95.

2. Annachhatre, A.P. and Bhamidimarri, S.M.R., Aerobic Treatment of Meat Industry

Wastewater. In: Chemeca 93, 1993 (In press).
 ANAEROBIC TREATMENT OF HIGH STRENGTH DAIRY PROCESSING WASTE
IN AN UPFLOW ANAEROBIC SLUDGE BLANKET REACTOR

B GRAUER, S M R BHAMIDIMARRI and R L EARLE

Department of Process and Environmental Technology
Massey University
Palmerston North, New Zealand

ABSTRACT

An upflow anaerobic sludge blanket (UASB) reactor was used to study the anaerobic digestion of
whey permeate. The 25L reactor was capable of reducing a soluble organic load of 15 kg COD/m3.d
by 95% while an increased load of 20 kg COD/m3.d resulted in 90% COD reduction. The methane
production rate increased linearly with the loading rate up to a loading rate of 15 kg COD/m3.d beyond
which there was no increase in methane yield. An analysis of the sludge granules revealed that the
accumulation of inorganics in the granules was significant and the specific activity of the granules
could be adversely affected by the concentration of calcium and magnesium salts, in particular the
phosphate and carbonates. The concentration of methane in the biogas was approximately 52% for
organic loads up to 15 kg COD/m3.d but beyond this value the methane concentration dropped to a
low 41%. The yield of methane was 4.1 m3/m3.d for organic loading rates of up to 15 kg COD/m3.d.

INTRODUCTION

Large quantities of whey are generated as byproducts of the New Zealand cheese and casein industries.
With the exception of a small quantity of sweet whey, it is considered a waste product, which is in
general disposed of by spray irrigation. However, this strategy is no longer acceptable as its
environmental impact is better understood. The anaerobic treatment of whey permeate from the
ultrafiltration of lactic acid precipitated casein was successfully demonstrated with loads of 7.7 kg
COD/m3.d in a fluidized bed digester (1) and 4.3 kg COD/m3.d in a sludge blanket digester (2).
Although the performance of the fluidized bed digester was shown to be superior, operation of this
system is complicated by the development and retention of the biofilm on the support surface in
anaerobic reactors (3).

The sludge blanket reactor, on the other hand is a high rate digester which is simple to operate and
more robust against fluctuations in operating conditions (4). Granulated sludge in UASB reactors has
been shown to perform well even after extended periods of storage (4) and therefore the system is
better suited to treat the waste streams from seasonal operations such as dairy processing.

The digestion performance of a UASB reactor treating lactic whey permeate is presented in this paper.
The quality of biogas generated and the composition of the sludge granules are also discussed.
1012
MATERIALS AND METHODS

Digester Operation
A 25L digester was constructed from QVF glass column sections and included a gas collection
mechanism which minimised the sludge carry-over. The digester was charged with approximately 10
kg of granulated sludge which had been previously acclimated to lactic whey permeate. The average
granule size was 2.2 mm in diameter. The whey permeate feed, with a COD of 48000 mgIL was
supplemented with 1.0 gIL of agricultural grade urea and 50 f.Al./L of 10% (v/v) defoaming agent. The
flow rate through the column was 1 Umin with recycle ratio ranging from 500-700.

Analytical Methods
Chemical Oxygen Demand (COD) was determined by digestion of samples with 0.035 M ~Cr207 in
concentrated sulphuric acid for two hours using a Hach COD apparatus (Hach, Ames, USA).

Biogas was measured using a Brand wet-test gas meter. Methane concentration in the biogas was
determined using a Varian Aerograph 900 gas chromatograph with stainless steel column packed with
80/1 00 mesh Porapak S.

Volatile Solids (VS) were measured according to Standard Methods (5).

The elemental composition of granules was determined by Inductively Coupled Plasma Emission
Spectroscopy whilst CO-3- was determined by the loss of mass on combustion at 1000°C.

RESULTS AND DISCUSSION

COD Removal
The digester was operated over a range of COD loading rates varying from 2.5 to 22 kg COD/m3.d.
The COD removal across the granular sludge bed was measured in terms of both the soluble and
insoluble components. As shown in Figure 1 the performance of the UASB is significantly higher
than that reported in the literature (2) with the soluble COD reduction greater than 95% up to a
loading rate of 15 kg COD/m3.d. Even at 20 kg COD/m3.d loading rate, a 90% soluble COD removal
was consistently achieved.

The total COD reduction was 90% up to 12 kg COD/m3.d decreasing to 74% at a loading rate of 22
kg COD/m3.d.

A COD mass balance over the entire period of digester operation of 237 days not accounting for COD
incorporated in biomass and CO2 revealed an 88.5% COD recovery through the process.

Methane Generation
The methane production rate as a function of organic loading rate is shown in Figure 1. The methane
production rate of STP increased rapidly as the COD loading rate increased to a maximum of 4.1
m3/m3.d at a loading rate of 15 kg COD/m3.d, but the methane yield decreased as the organic loading
rate increased further. This reflects the incomplete digestion resulting in the accumulation of volatile
fatty acids. Indeed, the concentrations of propionate and acetate were found to increase rapidly as the
loading rate increased beyond 15 kg COD/m3.d. The overall methane yield was found to be 0.3 m3/kg
COD removed.

Granule Composition
The granule elemental analysis revealed that there was an accumulation of inorganic matter is the
granules. The major components were found to be Ca (0.36 kg/kg), P (0.17 kg/kg) and CO-3- (0.07
kg/kg) indicating calcium phosphate deposition. Interestingly, there was no significant potassium
 1013
accumulation (0.001 kglkg) although the potassium concentration in the penneate was high (1.37 gIL).
These results imply that the specific activity of the granules could decline in UASB reactors operating
over prolonged periods of time.

100 5

'"'
4 -c
~
,

c..
!-
CI)
3
c=
£
c=
~
2 c::
.9
U~

]
1 c..
-:t
:=
U
70+-------~--------~------~r_------_+------__4
o 5 10 15 20 25
Loading Rate (gCOD.1- 1.d-])
Figurel. Performance of UASB digester.

CONCLUSIONS

The high rate digestion of strong dairy processing waste is shown to be feasible. Organic loading rates
of up to 15 kg CODm3.d can be treated to achieve 95% soluble COD and 85% total COD removal.
High methane yields of up to 4.1 m3/m3d are possible under optimal operating conditions. The results
presented in this study confinn that a UASB reactor is simple to operate and is superior in
perfonnance to the biofilm systems. However, the specific activity of the granules is likely to decrease
with time if the waste stream contains significant concentrations of inorganic salts.

REFERENCES

1. Boening, P.H. and Larsen, V.F. Anaerobic fluidized J:>ed whey treatment. Biotechnol. Bioeg.
1982,24,2539-2556.
2. Archer, R.H., Larsen, V.F. and McFarlane, P.N. Anaerobic digestion of whey penneate: A
comparison of three reactor designs. Poster paper presented at the 3rd International
Symposium on Anaerobic Digestion, Boston, August 14-19, 1983.
3. Henze, M. and Harremoes, P. Anaerobic treatment of wastewater in fixed film reactors - a
literature review. Wat. Sci. Tech., 1983, 15, 1-101.
4. Lettinga, G., van Velsen, A.F.M., Hobma, S.W., de Zeeuw, W. and Klapwijk, A. Use of
upflow sludge blanket reactor concept for biological wastewater treatment, especially for
anaerobic treatment. BiotechnoL Bioeng., 1980, 22, 699-734.
5. American Public Health Association. Standards methods for the examination of water and
wastewater (14th edition), Washington, D.C., (1975).
 ANAEROBIC TREATMENT OF DISTILLED WASTE WATER OF WHEAT 'SHOCHU'
USING TWO-PHASE FERMENTER

K. Morita. S. Taharazako. Y. Nishi

Chair of Agricultural Systems Engineering. Faculty of Agriculture.
Kagoshima Universi ty. Korimoto. 1 chome. Kagoshima 890. Japan

ABSTRACT

Distilled waste water of wheat ·shochu" which contained high concentra-

tion of organic pollutants was anaerobically fermented with practical
scale two-phase fermenter.The purification of digestive waste water. the
productivity and utility of biogas were also investigated. The per-
formance of the fermenter was evaluated using COD removal rate. removal
efficiency of COD and BOD. The generated biogas rate. the produced
biogas ratio and the concentration of methane gas were measured. and the
efficiency of biogas generation was evaluated.

INTRODUCTION

Waste from food processing operations has traditionally yielded severe

pollution problems. Especially. it is difficult to purify the waste
water containing high concentration of organic pollutants. In the field
of food processing waste management. it is important to develop an
improved and practical system to remove major pollutants such as COD.
BOD. and uti 1 ize the produced biogas. This present study was undertaken
to investigate the feasibility of anaerobic treatment using a Tvo-phase
methane fermenter for distilled waste water of "shochu".

MATERIALS AND METHODS

Figure 1 shows a schematic diagram of the methane fermentation system.

The two-phase fermenter(MBB) continuously loaded reactor consists of two
cylindrical tanks (total volume 6.6m 3 • phasel:phase2=1:9) equipped with
separate biogas collector. The reactor was kept at about 38°C and 7.4pH
in this study. This experiment was conducted over a 3 month period.
Hydraulic retention time(HRT) was about ten days and influent s~bstrate
 1015
concentration of COD from 10.000 to 14.000ppm. BOD from 20.000 to 37.000
ppm. SS fromm 6.000 to 8.000ppm were tested. The reactor performance was
evaluated based on COD removal and biogas production. COD removal rate
and biogas production rate were calculated with the following equations:
-dS/dt = K.(S-Sr)
where. S : COD after anaerobic treatment(ppm)
Sf: undecomposable COD (ppm)
KI: COD removal rate (1/d)
t retention time(d)
dG/dt = K2 (Gt-G)
where. G volume of actual generated biogas(m 3 )
Gt: volume of theoretical generated biogas(m 3 )
K2 : biogas production rate(l/d)

CD Inlet CD Phasel Gas holder

® Phase2 @ Outlet
® Biogas @ Biogas
(phasel) (phase2)
~
~or~ ~sl;:ine
c:::JI!:.t::::J L~ Gas burner
~ O.451113 /h

TW(rphase
feI'llenter
Waste water tank 6.6m 3
Aeration tank
20111 3

Fig.l Block diagrll.l of fermentation system and schematic diagram of tW(rphase fenaenter
RESULTS AND DISCUSSION

Table 1 shows the change of COD concentration for anaerobic fermenta-

tion The process was started using 4.0m 3 of seed anaerobic sludge from a
municipal waste treatment plant. The initial load consisted of sweet
potato shochu waste water with low concentration of COD. Steady-state
fermentation was obtained by a gradual increase of COD loading during
five months. In these periods(A. B. C). COD removal efficiency were 66.3-
79.0% at ten days of retention time and COD removal rate.K •. varied from
0.112 to 0.166.
Table 2 shows the change of BOD concentration and effect of deterrent on
anaerobic fermentation. The efficiency of BOD removal was low for high
concentration of BOD. Dramatic changes were not observed in NH.+. VFA
and T-N. SS removal efficiency greatly changed from 44.8 to 84.3%.
Table 3 shows the change of biogas production and methane concentration.
In these periods. volume of biogas was more than 10m 3 /d and volume ratio
(phase2/total) was approximately 87%. Methane concentration maintained
1016
Table 1 Change of COD concentration for anaerobic fermentation

period COD (ppm) removal volume loading HRT K1

(d) charge discharge COD(%) COD(kg/m 3 ) (d)

A(16d) 16,590 3, 470 1, 65 79. 0 10. 0 o. 161

B(35 d) 13,740 2, 820 1.44 77. 6 9. 7 o. 166
C(41d) 11. 900 3, 970 1, 20 66. 3 10. 0 o. 112
A: sweet potato and wheat shochu waste water
B: wheat shochu waste water
C: N by low pressure distillation

Table 2 Change of BOD concentration and effect of deterrent on anaerobic fermentation

period Temp pH BOD (ppm) removal NH.+ VPA T-N SS(ppm) SS(%)
(d) (Oe) charge BOD(%) (ppm) (ppm) (ppm) charge removal

A(16d) 38.5 7.35 25, 600 46.1 1. 450 2,000 18,200 84.3
B(35d) 38.2 7.41 25,500 57.9 1.990 1. 900 8,500 68.1
C(41d) 38.6 7.49 35, 200 51. 6 1. 970 5, 325 3,000 6,100 44.8

Table 3 Change of biogas production and methane concentration

period biogas(m"/d) ratio(%) CH. (%) K. production ratio

(d) phase1 phase2 phase2/total phase1 phase2 a (m' /kgCOD) /3 (m" /m"waste)

A(16d) 2.63 11. 53 81. 4 35.0 67.5 o. 153 1. 27 21. 0

B(35d) 1. 78 11. 93 87.0 40. 2 67.3 0.152 1. 28 17.5
C(41d) 1. 38 10.03 87.9 50.7 64.4 0.052 1. 38 16.4

67% during phase 2. Biogas production rate, K., greatly decreased during
period C, but biogas productivities a & /3 did not show any remarkable
change.

CONCLUSIONS

This present study indicated that an anaerobic treatment system using a

two-phase fermenter was practical to use for the disposal of high
concentrated organic waste, and the produced biogas containing a high
percentage of methane gas could be used as an alternate energy source.
 PLEUROTUS PRODUCTION IN COFFEE BEAN WASH WATER

Jose E. Sanchez V.
Centro de Investigaciones Ecol6gicas del Sureste
Apdo. Postal 36, Tapachula, Chiapas 30700 Mexico

Antonio M. Martin
Department of Biochemistry, Memorial University of Newfoundland
St. John's, Newfoundland, Canada AlB 3X9

ABSTRACT

In this work the growth of the fungus Pleurotus ostreatus in the form of pellets on coffee
wash water is studied. By supplementing the medium with non-proteinaceous nitrogen, a
biomass production of 3.05 gIL is obtained. The fungal growth causes a 25 % tannin content
reduction and a 43 % BOD reduction.

INTRODUCTION

The run-off water from the processing of coffee beans is an environmental pollutant.
Generally its COD value ranges between 5000 and 35,000 mgIL. It could be of interest to
use this waste water as a culture medium for growing P leurotus in the production of fungal
pellets. Pleurotus, among other fungi, has been grown in submerged culture using various
culture media [1,2,3]. In this work we have studied the technical viability of cultivating
Pleurotus in the water used to depulp coffee.

METHODS

Half a kilogram of fresh, ripe coffee berries were washed and de-pulped manually in 3.5 L
of water. The water thus obtained was separated from the pulp and the beans, centrifuged,
filtered, and finally sterilized by filtration (0.22 pm Millipore filters). The INIREB-8 strain
of Pleurotus ostreatus was inoculated into the sterile coffee wash water (CWW) and
1018
incubated for 10 days at room temperature, with constant stirring (50 rpm). The soluble
sugar, total nitrogen (Kjeldahl) and tannins contents, as well as the pH, COD and BOD of
the wash water were determined during fungal growth. At the end of fermentation, the pellet
diameters and the dry material obtained were measured.

RESULTS AND DISCUSSION

When coffee wash water sterilized in an autoclave was used as medium, there was no
growth, most likely due to degradation of carbohydrates and nitrogen, and to sulphur loss
from the culture medium. Over 8 days of fermentation, the pH of the medium underwent an
increase from a value of 5.5 to 7.9 due to the fungal metabolism, while the concentration of
soluble sugars decreased from 1.20 to 0.45 %. Nevertheless, at the end of 8 days of
fermentation, an important carbohydrate remnant remains, indicating incomplete fermentation.
The nitrogen concentration in the medium at that time was 0.058 %.
At the end of 8 days of growth on CWW, the concentration of biomass produced, in
the form of small pellets measuring 1.5 to 2 cm in diameter, reached 2.1 gIL. In terms of
yield and productivity, this amount is smail, since Martin and Bailey [3], working with
Agaricus campestris, obtained between 3 and 4 gIL of biomass in 8 days of fermentation on
hydrolysed peat. The lower result obtained here could be explained as a function of the
nitrogen content of the waste water, since Sugihara and Humfeld [1] reported nitrogen values
of around 1 gIL as optimal for the growth of Agaricus in submerged culture. In the present
study, the initial nitrogen content was 0.089 %.
As seen in Table 1, when the medium was supplemented with 3 different sources of
nitrogen (yeast extract, nitrogen without amino acids and nitrogen without sulphur-containing
amino acids), an improvement in biomass production was observed in most cases at the end
of 10 days of fermentation. In this time period, the carbon source was not entirely consumed
(there was a 0.16 % remnant, on average). This could be due to a limiting factor in CWW,
probably related to its low phosphorus and sulphur contents. Pleurotus utilizes tannins as a
carbon source, and so the concentration of these compounds is reduced from 0.16 to 0.12
mgIL during fermentation (an average reduction of 25 %), implying that this fungus has
anti-polluting activity.
In Table 2, the reductions of the BOD and COD in the CWW throughout the process
of medium preparation and microbial growth are observed. At the start, due to the solid
wastes that the water contains, the BOD and COD values reach levels on the order of 8,000
mgIL and 21,000 mgIL, respectively, giving a BOD/COD ratio of 0.38. This indicates a high
content of non-biodegradable pollutants in the CWW. Centrifugation eliminated most of these
waste products, and sterilizing filtration produced further elimination, leaving a BODS value
of 1,350 mgIL. Stirring and aerating, as well as the growth of P. ostreatus in the medium,
brought about a reduction of this value on the order of 43 %, leaving the run-off water with
a final BOD value of 770 mgIL, representing a more acceptable level in terms of pollution.
The reduction of BOD and COD via fungal metabolism has previously been reported [4].
Coffee bean wash water contains high levels of pollutant materials, which for the
most part are non-biodegradable (BOD/COD ratio 0.38). The treatment used leaves relatively
low values of BODS and COD in the run-off. To assure good growth it is necessary to
supplement the medium with nitrogen, and probably phosphorus and sulphur, after
centrifugation and sterilization.
 1019
TABLE 1
Pleurotus ostreatus growth on coffee bean wash water supplemented by nitrogen sources
(10 days of fermentation).

Culture Medium Final pH Biomass Produced Soluble Sugars N2 (%)

Supplementation (gIL of medium) Residue (%)
Yeast Extract 8.27 1.99 0.17 0.070
N2 without AA 6.09 3.05 0.17 0.055
N2 without S-AA 5.60 2.87 0.13 0.075
No Additions 7.25 2.21 0.15 0.074

TABLE 2
BODS and COD variation in coffee bean wash water during the different steps of the process.

Step BODS (mgIL) COD (mg/L)

Initial sample 8,076 21,000
After centrifugation 2,285 6,000
After sterilization 1,350 3,000
After growth 770 800

ACKNOWLEDGEMENTS

The Rosario Izapa Experimental Station (INIFAP-SARH) is acknowledged for the

cooperation provided to this study. The authors wish to express their thanks to M. del R.
Hernandez Ibarra, L.A. Calvo Bado and P. Bemister for their assistance.

REFERENCES

1. Sugihara, T.P. and Humfeld, H., Submerged culture of the mycelium of various
species of mushroom. Appl. Microbiol., 1954,2(3), 170-172.

2. Martin, A.M., A review of fundamental process aspects for the production of

mushroom mycelium. J. Food Proc. Eng., 1986,8, 81-96.

3. Martin, A.M. and Bailey, V.I., Growth of Agaricus campestris NRRL 2334 in the
form of pellets. Appl. Environ. Microbiol., 1985, 49(6), 1502-1506.

4. Church, B.D., Erickson, E.E. and Widmer, C.M., Fungal digestion of food processing
wastes. Food Technol., 1976, 27, 36.
 BY PRODUCTS FROM FOOD INDUSTRIES: UTILIZATION FOR
BIOINSECTICIDE PRODUCTION.

IRACEMA DE O. MORAES
Universidade Estadual Paulista -UNESP/IBILCE-DETA
CP 136 CEP 15054 000 S.Jose do Rio Preto SP Brazil

DEISE M. F. CAPALBO
Centro Nacional de Pesquisa de Defesa da Agricultura
CP 69 CEP 13820 000 Jaguariuna SP Brazil

REGINA DE O. MORAES
Universidade Estadual de Campinas-UNICAMP/FEAGRI
CP 6011 CEP 13081 970 Campinas SP Brazil

ABSTRACT

Agricultural residues and wastewaters from food, beverages and paper

industries, are feasible substrates to produce microbial insecticides through
Bacillus thuringiensis fermentations. These substrates are sources of carbon and
nitrogen that are essential components to culture media composition. Since 1970
our research group are studying both the submerged (1) and semisolid
fermentations (2) and two patents (3,4) of the processes were developed. Corn
steep liquor and sugar cane molasses were determined as the principal components
of the culture media. This paper deals with the process, advantages and problems
and the use of low cost raw material availlable in Brazil. The group intends to
develop local technology to produce Bacillus thuringiensis, just to use against
agricultural pests that are responsible to 40% losses in field, harvested and stored
products. These losses cause price increments in the final product and contribute
to the maximize hungry and poverty problems, in Brazil.

INTRODUCTION

The practical use of entomopathogenic microorganisms to crop protection is

possible when an industrial scale production of the organism is developed. That
is the case of Bacillus thuringiensis, one of the most studied and commercially
important entomopathogenic bacterium. Brazil is using imported products and our
group is trying to develop local technology to massal production (5, 6), based in
the use of low cost raw material, by products, residues or wastewaters from food,
beverage or agroindustries to compose the culture media.
 1021
MATERIAL AND METHODS

Bacillus thuringiensis Berliner was maintened in nutrient agar slants to be used

as inoculum, for the different media composition. Standard medium composition has
sugar cane molasses and corn steep liquor, as stated in BR Patent (3).
Culture media were composed as follows:
MSG - wwater from monossodic glutamate industry + tap water;
SCG - monossodic glutamate industry ww + sugar cane molasse;
CWG - coconut water + ww from monossodic glutamate industry;
YBI - liquid residue from beer industry + tap water;
These media were adjusted to 8.3 g total carbohydrates.
MCF - cassava flour industry ww (75%) + tap water( 25%).
pH was adjusted to 7.3, before sterilization, 15 min. 121°C.
The fermentation process were runned in shaker ( flasks of 250 ml, with 50 ml
culture medium sterilized ), fermentors of 5 I and 10 I working volume.
Temperature was controlled at 30 0 C, agitation at 150 and 300 rpm, respectivelly
and 1 vvm aeration, for fermentors.
Optical density (600 nm), pH, viable spore count/colony forming units-CFU (7),
were acomplished. Bioassay was performed against Anticarsia gemmatallis, to know
about toxicity of the bioinsecticide obtained.

RESULTS

The fermentation processes, present the following results expressed on Table

1.

TABLE 1: Comparison of the fermentation processes development.

MEDIUM Harvest time pH Absorb. ores

(h) (aD) (CSiU/mll
MSG 0 6,5 4,8 < 103
4 6,7 5,5 <103
7 6,8 6,4 <103
17 7,7 7,5 1,8x106
19 8,1 9,6 5,5 x 109
24 8,2 11 ,1 > lOll
SCG 0 6,6 5,3 <103
5 5,7 6,6 <103
29 6,1 17,3 1,8 x 106
45 6,8 17,1 7,7 X 106
53 7,3 15,1 2,8 x 107
CWG 0 6,1 5,3 <103
2 6,1 5,7 <103
21 5,5 16,0 2,4 x 104
45 6,8 21,9 1,3 x 107
66 7,3 19,3 3,9 x 108
1022

MEDIUM Harvest time pH Absorb. S~ores

(h) (aD) (C Ulml)
YBI 0 7,0 3,7 <10 3
4 6,6 4,2 <103
22 7,8 7,1 1,2 x lOll
25 8,0 7,7 > lOll
28 8,0 10,0 > lOll
41 8,4 8,7 > lOll
MCF 0 6,5 10,6 <103
6 5,9 15,4 <103
46 6,1 21,8 6,0 x 105
70 6,9 19,5 4,4 x 106
76 7,3 20,5 2,0 x 109

AKNOWLEDGEMENTS
The authors express their gratitude to FUNDUNESP, CNPq, OEA and FAPESP

REFERENCES
1. Moraes, 1.0., Capalbo, D.M.F., Del Bianchi, V.L., Sifuentes, L.E.,
Technical aspects of the use of wastewater to get useful products through
fermentative processes. IV Latinamerican Congress of heat and mass
transfer, Chile, 1991, 1 , 93-96.
2. Capalbo, D. M. F . and Moraes, I. o. , Production of proteic protoxin by Bacillus
thuringiensis in semisolid process. 12th Simposio Anual da Academia de
CMncias do Estado de S. Paulo. Campinas SP Brazil, 1988, 2 , 46-55.
3. Moraes, J.O., BR Patent 7608688, Brazil, 1976.
4. Moraes, J.O., BR Patent 8500663, Brazil, 1985.
5. Moraes, I. o. and Capalbo, D. M. F., Study of continuous fermentation for
obtaining bacterial insecticide to control agricultural pests. IV Japan-Brazil
Symposium on Science and Technology, Brazil, 1984, II, 248-255.
6. Moraes, I. 0., Production and utilization of Bacillus thuringiensis for crop
protection in Brazil. International workshop on Bt and its applications in
developing countries. Egypt, 1991. -
7. Thompson, P.J. and Stevenson,K.E. Mesophilic aerobic spore formers.In
Compendium of methods for the microbiological examination of foods.
Speck, M.L. 2. ed., Washington, Am. Public Health Assoc., 1984, 19,
211-220.
 NOx REMOVAL IN WATER BY BIOCHEMICAL REACTION

TADATAKE OKU, MITSUTAKA KONDO, HITOSHI SATO, TOSHIYUKI NISHIO,

TEIICHIRO ITO, MIKIO HOSHINO* and HIROSHI SEKI*
Laboratory of Bio-organic Chemistry, College of Agriculture and Veterinary Medicine,
Nihon University, 34-1, Shimouma 3-chome, Setagaya-ku, Tokyo 154, Japan,
*The Institute of Physical and Chemical Research, 2-1, Hirosawa,
Wako, Saitama 351-01, Japan

ABSTRACT

NOx removal in water by the reduction of N02- and NO to NH4+ was examined by using
denatured cytochrome c (cyt c) by heating at 100°C for 30 min or by Co-60 gamma-ray
irradiation with a dose of 10 kGy. Sodium dithionite and methyl viologen were,
respectively, used as a reducing agent and an electron carrier in the reduction system. The
conversion yield of N02- to NH4+ were ca. 100% and ca. 50% in N2 atmosphere and air,
respectively. The ESR and optical studies have shown that nitrosyl complex of cyt cis
produced as an intermediate for the reduction of N02-. The conversion of NOr to NH4+ is
a net 6-electron and 8-proton reduction process in the same manner as nitrite reductases.
Denatured cyt c is suggested to be used as a model of nitrite reductase for catalytic removal
of NOx in water.

INTRODUCTION

Recently, removal of NOr in water has been one of the important research area in
environmental science because of the necessity of reducing water pollution. In nature,
ferredoxin-nitrite reductase (EC 1.7.7.1) which reduces N02- to NH4+ (N02- + 8H+ + 6e-
~ NH4+ + 2H20) has been found in higher plants l ) and green algae2). Purification method
and some properties of nitric oxide oxidoreductase (BC 1.7.99.3) and nitric oxide reductase
(BC 1.7.99.2) have been studied with denitrifying bacterium Alcaligenes S-6 3 ) and
Alcaligenes /aecalis 4), respectively. Although the reduction of NOr to NH4+ by
photoinduced enzyme (EC 1.9.6.1)5) or water soluble iron porphyrin 6) as well as the
reduction of NOr to NO by NADH analog7 ) has been investigated, no nonenzymatic
reduction of N02- via NO to ~+ by denatured protein has so far been reported. The
present work shows that N02- is reduced to NH4+ by a chemical reducing agent in the
presence of denatured cyt c and methyl viologen.
1024
MATERIALS AND METHODS

Cytochrome and its denaturation. Horse cardiac cyt c was obtained from Sigma
Chemical Company and purified by gel filtration with Bio-gel P-lO. Denaturation of cyt c
in an aqueous solution was made by Co 60 gamma-ray irradiation with a dose of 10 kGy or
by heating at 100°C for 30 min.

Reduction of nitrite by denatured cyt c. The reaction mixture which was composed of
1.5 ml ofO.2M buffer (0.75mmole, pH3-9), 2 ml ofO.OlM sodium nitrite (20J.1mole), 2.5 ml
of 0.003M methyl viologen (7 .5J.1ffi0le), 2.5 ml of 50J.1M cyt c, and 1.5 ml of 0.96M sodium
dithionite (0. 36J.1mole) dissolved in an aqueous solution of 0.29M sodium
hydrogencarbonate, was incubated at 30°C. One ml of the reaction mixture was shaken
vigorously by Vortex until the remaining methyl viologen was completely decolored.

Assay of nitrite and ammonium ions. N02- and NH4+ concentrations in the reaction
mixture were determined by using diazo-reaction method and HPLC, respectively.

Spectrometric measurement of iron-nitrosyl complex. Optical and ESR spectra of iron-

nitrosyl complex were measured by a JOEL FE 3AX X-band spectrometer and a MILTON
ROY 3000 spectrophotometer, respectively.

RESULTS AND DISCUSSION

Reduction of nitrite to ammonia in nitrogen atmosphere and air. In nitrogen

atmosphere, a decrease of N02- in the reaction mixture was observed with time; after 45
min the remaining of NOr was 3% and all NOr have disappeared after 60 min. The
formation of NH4+ increased as the time proceeds; the yields were 92% after 45 min, 98%
after 60 min and 100% after 75min.

~ 100 N2, NH4

~
!;i 80
:::E

..
II:
0

.
~

x
60
z
AIR,NH4·
CJ 40
~
z
:cc
:::E
w 20
II:
0
Z

30 TIME (min)SO 90

Figure 1. N02- remaining and NH4+ formation when N02- was reacted with denatured cyt
c in the presence of sodium dithionite and methyl viologen
in N2 atmosphere and air at pH 7.0.
 1025
In the presence of air, the rate ofN02- reduction was slightly slow, compared to that in N2
atmosphere. The formation of NH4+ was only 50-55% after 75 min. From these results,
in the presence of air, the reaction other than the reduction of N02- to NH4+ was expected
as in the case of nitric oxide reductase4 >.

pH Dependence of nitrite reduction. In the range of pH3-9, the rate of reduction of

N02- to NH4+ at lower pH was much faster than that at higher pH in both N2 atmosphere
and air.

Detection of iron-nitrosyl complex. A characteristic ESR of FeIII.N=O complex of

denatured cyt c was observed near 3,300 G. Optical absorption due to the formation of
Feill-N=O was observed around 570 nm.

Reaction mechanism of nitrite reduction. From above results, the reaction mechanism
of N02- reduction by denatured cyt c in the presence of a reducing reagent and an electron
carrier was proposed as follows.

This reaction is a net 6-electron and 8-proton reduction process as in the case of feredoxin
nitrite reductase 1•2>. These results suggest that the denatured cyt c can be used as a model
of nitrite reductase for catalytic removal of NOx in water.

This work was partly supported under Nihon University Research Grant for 1992.

REFERENCES

1. Cardenas, B. J., Rivas, J. and Moreno, C. G., Purification and properties of nitrite
reductase from spinach leaves. FEBS Letters, 1972,23, 131-132.

2. Zumft, W. G., Ferredoxin: Nitrite oxidoreductase from Chlorella purification and

properties. Biochim. Biophys. Acta, 1972,276,363-375.

3. Kakutani, T., Watanabe, H., Arima, K. and Beppu, T., Purification and properties of a
copper-containing nitrite reductase from a denitrifying Bacterium, Alcaligenes /aecalis
strain S-6. J. Biochem., 1981,89,435-461,463-472.

4. Miyata, M., Studies on denitrification XIV. The electron donating system in the
reduction of nitric oxide and nitrate, ibid., 1971,70,205-213.

5. Willner, I., Lapidot, N. and Liklin, A., Photoinduced enzyme-catalyzed reduction of

nitrate (N03-) and nitrite (N02-) to ammonia (NH3). J. Am. Chem. Soc., 1989, 111,
1883-1884.

6. Barley, M. H., Takeuchi, K. 1. and Meyer, T., Electrocatalytic reduction of nitrite to

ammonia based on a water soluble iron porphyrin. ibid., 1986,108,5876-5885.

7. Fukusumi, S. and Yorisue,T., Acid-catalyzed one electron reduction of nitrite to nitric

oxide by an NADH analog and 1,1'-dimethylferrocene in the absence and presence of
dioxygen. Chem. Letters, 1990,871-874.
 RIPFADI: IBERO AMERICAN NETWORK ON PHYSICAL
PROPERTIES OF FOODS

JOSE MIGUEL AGUILERA

RIPF ADI Coordinator
Department of Chemical Engineering
Universidad Cat6lica de Chile
P.O. Box 306, Santiago, Chile

ABSTRACT
This paper discusses the objectives and modus operandi of RIPFADI, an
Ibero American Network of Physical Properties of Foods. Launched in
1992, RIPFADI already enrolls 88 groups from 11 countries and 263
researchers. Its objective is to promote scientific interactions between
researchers of the region working on thermal, rheological, electrical,
diffusion, thermodynamic and mechanical properties of foods and other
important physical parameters.
INTRODUCTION
CYTED is an international cooperation program in Ibero America (Latin
America plus Spain and Portugal) among research groups of universities,
R&D centers and innovative companies. Its aim is to achieve scientific
and technical results transferable to the production system and social
development plans of the region to assist in improving the quality of life
of people, technological modernization and ultimately, the economic'
development of Ibero American countries. Currently, twenty on'e
countries participate in several of CYTED's sixteen subprograms, one of
which is Food Technology and Preservation. A first successful project of
this subprogram was finished in 1991 and dealt with intermediate
mois ture foods and combined methods technology [1].
Within CYTED a network is an association of research units of
private or public organizations sharing scientific interests and activities in
a common subject. Its objectives are to foster a continuous and stable
scientific interaction, exchange technical and scientific information, assist
in the formation of human resources and in the transfer of technology and
helping in strengthening of research groups. Moreover, networks should
be a nuclei for multinational projects and a means of coordinating
research at the international level.
 1027
JUSTIFICATION AND OBJECTIVES OF RIPFADI
The agriculture and food sector plays a major role in the economy of
countries of the region. The formidable double task of food technology is
providing local consumers with wholesome and nutritious foods at a
reasonable price while simultaneously exporting high-quality, value-
added products that are in demand in industrialized countries. Technology
and scientific knowledge in the latter case has to be state-of-the-art in
order to compete in equal terms with foreign suppliers.
There are several reasons jus tifying the exis tence of a network like
RIPFADI:
• Data on physical properties of foods are essential to design new
processes, optimize unit operations and to develop local technologies.
• Needed information on local agricultural products and foods may not be
available because they are not studied in industrialized countries.
• Local food companies are usually small and cannot afford research
facilities to generate their own data.
• Research facilities in some countries may not be equipped to perform
determinations on certain physical properties.
• Many products are of similar nature (e. g., tropical fruits) and
duplication of research efforts should be avoided.
• Researchers in similar areas need to work in groups beyond critical
mass.
• Synergistic effects are expected when research is coordinated and
cooperation schemes are implemented.
• The almos t common language of the network (all countries except Brazil
and Portugal speak Spanish) facilitates communication, dissemination of
res ults and fulfillment of exchange programs.
RIPF ADI is expected to become an international forum for
reporting research results and convey them to the productive system. It
should also facilitate contacts between research groups and eventually
provide access to an international data base on physical properties of
foods.

A DIAGNOSIS OF RESEARCH IN PHYSICAL PROPERTIES

OF FOODS
As a first step in the formation of RIPFADI a survey was conducted to
assess the local capabilities of research groups willing to participate and
their main areas of interest. In Table I a matrix of physical properties and
products researched by groups is presented.
The fruits/vegetables group is the most cited, probably due to the
indigenous nature of the products and their high perishability. Thermal,
rheological and thermodynamic (water activity) are the most frequently
mentioned properties under research. This initial survey also
demonstrated the existence of some well staffed groups and the
availability of modern equipment, which predicts good possibilities for
high quality research and available training centers in the future.
1028
TABLE 1
Matrix of physical property/product in research of institutions associated
to RIPFADI (figures represent the number of groups working on a topic)

Meat/Fish Dairy Fruits/Veg Cereal Other

Prod.
Thermal 25 10 31 16 9
Rheological 14 22 24 18 16
Electrical 2 3 2
Diffusion 4 22
8 10 6
Micros tructure 17 10
6 5 6
Thermodynamic 17 24
13 25 16
Mechanical 19 13 18 15 10

OPERATION AND FUTURE OF RIPFADI

As of today there are 88 research groups from 11 countries, representing
263 researchers, associated to RIPFADI. The CYTED program finances
most of the coordination efforts, including publication of a newsletter,
scientific exchange activities, technical courses and workshops, technical
publications and the annual meeting of national coordinators.
Participating countries through their national research councils, continue
funding and organizing independently the research work based on
scientific merit and provide financing for local coordination.
Administration procedures for RIPFADI are minimum and quite simple,
which is a plus for participants in the network. The system largely relies
on the enthusiasm of those who collaborate that are also the direct
beneficiaries.
Plans for 1993 include consolidation of the network, publication of
an inventory of participating institutions and groups and printing of a
book listing almost 300 abstracts of papers in international journals
authored by local researchers. In order to become a truly collaborative
experience some groups will be assisted in bringing up the level of their
scientific background via courses and technical exchange visits. In turn,
it is expected that beneficiaries of these actions will trickle down the
experiences in their countries.
As RIPF ADI promotes collaboration and creates contacts in the
region it will become a nucleus for developing food engineering research,
particularly in Latin America.

REFERENCES
1. Aguilera, J.M. and Parada, E., CYTED-D AHI: An Ibero American
project on intermediate mois ture foods and combined methods
technology. EQQdR.e.s....ln1l.., 1992,25,159-165.
 RELATIONSHIPS BETWEEN PHYSICAL PROPERTIES AND
CHEMICAL COMPONENTS OF OKINAWAN BROWN SUGAR

TAKAYOSHI AKINAGA, YOSHIHIRO KOHDA

Associate professor/Professor/Department of Bioproduction,
College of Agriculture, University of the Ryukyus,
1 Senbaru, Nishihara, Okinawa 903-01, Japan

ABSTRACT
Some physical properties and chemical components of brown sugar produced in
Okinawa were investigated to improve the quality standard and manufacturing
methods. Hardness, viscosity, color, moisture content, water activity, and
density of brown sugar were measured as the components of physical
properties. Content of K, P, Ca, Mg, Na, N, and Cl in the brown sugar were
measured. As the result of investigation, it became clear that the physical
properties of brown sugars varied with the island where the sugar was
produced; and P, N, and Ca significantly affected the quality of sugar.

INTRODUCTION
Brown sugar is one of the uncentrifuged sugars and a special product of
Okinawa prefecture, and it have been manufactured only on small islands.
Brown sugar manufacturing is the main industry that have been supporting
the economy of those islands. However, it was reported that the quality of
Okinawan brown sugars varied with the districts where it was produced and
among layers in one package(1,2). Final products were inspected by the
inspector under his experience and the traditional quality standard at each
manufactory. A knowledge of quality standard is important and essential
engineering data for quality control of brown sugar. Little literature on
the quality of brown sugar was found because brown sugar is a local
product. It is hoped to produce brown sugar which has constant quality.
The subject of this study was to contribute to the improvement of the
quality standard and manufacturing method of brown sugar.

MATERIALS AND METHODS

Brown sugars were collected at all manufactories in Okinawa prefecture
and used for investigation of their qualities. TABLE 1 shows
manufactories, islands, grades, and manufactured dates.
1030
Physical properties
Hardness was defined as the maximum penetrating resistance and measured by
a universal testing machine using 8 mm in diameter rigid plunger at 10
mm/min in penetrating velocity. Viscosity was measured by a Shimadzu model
CFT-500 flowtester under preheated temperature of 75°C, preheat time of 5
seconds, and pressure of 5884 KPa using a cylindrical specimen which
recomposed from powdered brown sugar. Viscosity was calculated by Hagen-
Poiseuille's equation. Surface color of brown sugar was measured by a
Minolta model CR-200b chroma meter. Moisture content was measured by the
vacuum oven method. Water activity was measured by a Rotronic model DT2-1
hy~roscope of 25°C. Density was measured by a Beckman model 930 air
comparison pycnometer and a electronic balance.

TABLE 1
Manufactured data of brown sugars
Manufactory Island Grade Date
Kohama Sugar Ltd Kohama Excellent Jan. 28, 1988
Mi yako Sugar Ltd Tarama Excellent Feb. 12, 1989
Aguni Agri. Coop Aguni First Jan. 17, 1989
Hateruma Sugar Ltd Hateruma First Dec. 16, 1988
Iheya Agri. Coop Iheya First Jan. 13, 1989
Iriomote Sugar Ltd Iriomote First Jan. 14, 1989
Miyako Sugar Ltd Tarama First Nov. 28, 1988
Yonaguni Agri. Coop Yonaguni First Jan. 19. 1989

Chemical components
Brown sugars produced in Okinawa were composed of 10 to 11 layers. Samples
were prepared from all layers and mixed after broken into pieces with a
knife. K, P, Ca, Mg and Na were extracted by the wet-decompose method. K
was measured by the flame photometry. P was measured by the Bernard-
molybdic acid method and spectrophotometry; Ca and Mg were measured by the
nuclear absorption photometry. N was measured by the semi-micro Kjeldahl
method. Cl was measured by the silver nitrate titration method after ashed.

RESULTS AND DISCUSSION

TABLE 2 shows the average values of physical properties. and TABLE 3
shows average values of chemical components. TABLE 4 shows results of
multiple regression analysis between physical properties and chemical
components. From the multiple regression analysis, it became clear that P
content affected the pH, viscosity and density, K content affected
hardness, Nand Na content affected the color, Ca content affected water
activity, and Mg content affected the pH of brown sugar.

CONCLUSION
Content of inorganic composition such as P, K, N, Ca, Mg, and Na in
brown sugar might be introduced in the quality standards of brown sugars.
 1031
TABLE 2
Average values of physical properties
Island Grade. M.C. Brix pH Hard. Visco. Hue W.A. Density
(iaWB) (0 ) (-) (MPa) (KPaS) a(-) (-) (g/cm 3 )
Kohama E 5.33 97.8 5.51 2.39 8.50 5.99 0.686 1.606
Tarama E 4.28 98.8 5. 10 2.22 53.04 6.65 O. 702
Aguni F 4.58 97.1 6.45 1.85 89.62 4.57 0.684 1.572
Hateruma F 4.35 98.8 5.97 1.16 12.22 5. 78
Iheya F 5.56 98.4 6.22 4.48 5.58 6. 13 0.669 1.628
Iriomote F 4.88 96. 7 5.89 0.99 25.03 5.36 0.672 1.625
Tarama F 4.59 98.6 5.88 5.41 15.59 6.52
Yonaguni F 4.67 96.6 5.39 1.6118.73 6.84 1.605

TABLE 3
Average values of chemical composition
Island Grade P205 K20 N Ca Mg Na Cl
(ia) (ia) (ia) (mg/g) (mg/g) (mg/g) (mg/g)
Kohama E 0.080 0.740 O. 145 1.106 7.039 0.253 0.649
Tarama E 0.053 1.499 0.213 1.519 7.226 0.534 0.668
Aguni F 0.153 1.309 0.076 1.370 5.654 0.158 0.698
Hateruma F 0.061 1.362 0.217 0.827 4.926 0.229 O. 733
Iheya F 0.056 0.826 O. 148 0.687 4.908 O. 157 0.800
Iriomote F 0.046 1.558 0.172 0.991 4.902 O. 147 0.676
Yonaguni F 0.066 1.479 0.317 0.862 7. 150 0.282 0.880

TABLE 4
Multiple correlation coefficients
pH Hard. Visco. W.Act. Hue(a) Chroma Density
P 0.573 0.592 -0.929
K -0.716 0.339 -0.506
N 0.739 0.982 -0.337
Ca 0.440 0.630 -0.260
Mg -0. 793
Cl -0.231
Na 0.436 0.535
M.C. O. 187 0.319
Brix - 0.403 0.364

REFERENCES
1. Okadome, H., Akinaga, T., Kohda, Y., Izumi, H. and Ueno, M., Fundamental
studies on physical properties of brown sugars, Proccedings of 49th
annual meetings of JSAM, 1990, 323-324.(in Japanese)
2. Akinaga, T. and Kohda, Y., Basic studies on the storage properties of
Okinawan brown sugar -hardness, moisture content and pH-, Journal of
Kyushu blanch of the JSAM, 1989, 38, 58-61. (in Japanese)
 THE EFFECT OF PROCESSING VARIABLES ON THE PRODUcr QUALITY
OF SOYA PLANTAIN BABY FOOD

P.O. OGAZIt, H.O. OGUNDIPE2, F.A. OYEWUSe, S.A.O. ADEYEMI4

1. RMRDC, P.M.B. 12873, Lagos, Nigeria

2. IITA, P.M.B. 5320, Ibadan, Nigeria
3. FIIRO, P.M.B. 21023, Ikeja, Lagos, Nigeria
4. NIHORT, P.M.B. 5432, Ibadan, Nigeria

ABSTRACT

Appropriate technologies were used to produce soya-plantain baby food enriched with minerals
and vitamins. Plantain was peeled, and dried using cabinet dryer. Soybean seeds were dried,
dehulled, milled and extruded. The extruded soybean and plantain flour were blended with
sugar, multi-vitamin and calcium carbonate and subjected to proximate analysis and cooking
tests. It was found that a mixture of 60% plantain flour, 32% soybean grit and 8% sugar
produced a blend whose proximate analysis showed protein 15.8%, fat 8.0%, carbohydrate
70.8% and energy 457.4 kca1!100g. There was no significant difference between the soya-
plantain foods produced using extruded and non-extruded soybean grits.

INTRODUCTION

Solid foods are introduced to children as from six months upwards. Most of these solid foods
are made from our local crops like maize, rice, sorghum, millet, yam, cassava and plantain.
But the problem of calorie-protein-mal-nutrition has always existed in rural and urban children
between the age of four and eighteen months because the solid foods are not well balanced in
the amount of protein, fat and carbohydrate content. In response of this obvious challenges of
infact nutrition in Nigeria, a result-oriented research and development effort was initiated to
produce a weaning formula that would meet local taste and siut Nigerian custom using plantain
and soybean as the base raw materials.

MATERIALS AND METHODS

The main raw materials used for the experiment were green plantain, soybean, multi-vitamin
and calcium carbonate. The stages involved in the development of soymusa are the production
of plantain flour and soybean grit (extruded and non-extruded); formulation, laboratory
production and preliminary sensory evaluation.
 1033
Plantain Flour Production: Green mature plantain fruits were hand peeled and sliced using
knife or automatic dicing machine to an appropriate thickness of about 15mm. The slices were
dried in a cabinet dryer. The drying temperature was 120°C for the first 2 hours and 80°C for
the remaining 3 - 4 hours of drying. The drying was completed when the moisture content of
the dried plantain was below 10%. The dried plantain was milled into flour using hammer
mill.
Production of Non-Extruded Soybean Grit: 25 kg of raw soybean was weighed out and heated
in a cabinet dryer at 110°C for one hour and cracked in a disc attrition mill for 40 minutes.
The cracked beans were subjected to air aspiration for 15 minutes using wooden aspirator. The
dehulled beans were washed and cooked for 4 hours. The cooked beans were filter pressed and
dried in tunnel dryer at 150°C for 1~ hours.
Production of Extruded Soybean Grit: Dried soybean seeds were dehulled, winnowed, milled
into fine powder, and extruded using the insta pro 600 Jr. model extruder. The feed intake was
set to ensure a discharge of between 300 - 350 kg/hr at a temperature range of 140 - 150°C.
The approximate residence time is about 30 seconds. All extruded soybean were dried into
grit, hammer-milled, sieved, packaged.
Extruded and non-extruded soybean were blended with plantain flour and sugar respectively
as follows: ABC D E
Plantain Flour 1500g 1458g 1393g 1313g 1222g
Soybean Grit 400g SOOg 600g 700g 800g
Sugar l00g 12Sg 150g 175g 200g
The blended mixtures were milled in a roller mill to produce a homogenous mixture of uniform
particle size. The milled samples were sieved using vibro sieving machine of aperture 315
microns. Multi-vitamin and calcium carbonate were added to the sieved blend and blended
further for ten minutes using tumbling blender. The composition of the final blend is plantain
flour 60%, soybean flour 32%, sugar 8%, multi-vitamin 0.15%, calcium carbonate 0.84%.
Three desert spoon fulls of each blend were reconstituted with small quantity of cold water to
obtain a homogenous liquid mixture. Boiling water was poured into the reconstituted mixture
and cooked for about 3 minutes with continuous stirring to avoid any lumping until well
cooked. The texture of the gel was consistent and smooth.
The moisture content of the blend was determined by drying in forced air oven at 130°C for
one hour. Protein content, crude fat, crude fibre and ash contents were determined according
to the AOAC Methods (1). Potassium, sodium, calcium and phosphorus were determined
according to the method described by Pearson 1976 (2). Carbohydrate was determined by
difference method.
The two samples produced using extruded and non-extruded soybean were submitted for
sensory evaluation. The samples were compared for attributes of colour, flavour, consistency,
mouthfeel and taste using 9 points hedonic type questionnaire (William 1982 (3». The data
collected were analysed by the student t-test to determine if the samples differed significantly.

RESULTS

The results of the proximate analysis are shown in Table 1. From the results, it was decided
that plantain flour (60%), soybean flour (32%) and sugar (8%) will be maintained as the blend
provides the required protein and energy level for the healthy growth of babies. The results
1034
from the sensory evaluation of the two samples showed that the extruded and non-extruded
products did not differ significantly from any of the attributes for which they were evaluated.

TABLE 1
Proximate Analysis of Plantain/Soybean/Sugar Blends

A B C D E
Moisture % 5.5 6.0 5.3 6.2 6.3
Fat % 3.6 4.5 5.3 8.0 8.6
Protein % 8.7 10.5 12.0 15.8 16.5
Carbohydrate % 82.5 76.8 73.51 70.8 70.4
Energy/Kcal/l OOg 417.7 420.7 419.74 457.4 458.7
Fibre % 0.83 1.57 1.78 1.0 1.17
Ash % 2.50 2.77 2.73 2.85 3.04
Potassium % 0.1 0.2 0.1 0.3 0.2
Sodium % 0.07 0.07 0.07 0.06 0.06
Calcium % 0.30 0.40 0.35 0.35 0.35
Phosphorus % 0.13 0.28 0.15 0.15 0.16

DISCUSSION

Plantain and soybean can be produced easily in the country. Plantain is a good source of
carbohydrate, while soybean is superior to all other plant foods as source of protein and have
good balance of the essential amino acids. The proximate analysis of plantain and soybean
have shown that both contain all the necessary ingredients required for the formulation of a
weaning food. The other required ingredients vitamins and minerals have been added as
supplementary to the ones contained in the crops. The preliminary sensory evaluation has
shown that the extruded and non-extruded products did not differ significantly from any of the
attributes for which they were evaluated.

CONCLUSION

This study has shown that acceptable weaning food can be produced from plantain and soybean
using appropriate technology. For a small-scale production, the non-extrusion method which
is labour intensive with long process time can be adopted. For medium-large scale production
the use of extrusion method which requires more capital for the equipment but minimum
operational cost is recommended.

REFERENCES

1. AO.AC. Methods of analysis of the Association of Officials Analytical Chemists,

Washington (1975).
2. Pearson, D., The Chemical Analysis of Foods. Published by Churchill Livingstone,
Edinburgh, 7th ED. 1976, pp. 19-24.
3. Williams, AA Scoring methods used in the sensory analysis of foods and beverages at
Long Ashton Research Station, I. Food Tech., 17, 1982.
 PREDICTING TOFU PRODUCTIVITIES OF SOYBEAN VARIETIES
BY TOFU GEL CENTRIFUGATION

S.-J. TSAI, T.L. HONG and s.c.s. TSOU*

Department of Food Science, National Chung Hsing University
250 Kuo Kuang Road, Taichung, Taiwan, R. o. C.
*Analytical Laboratory, Asian Vegetable Research and Development Center
(AVRDC) P. o. Box 42, Shanhua, Tainan, Taiwan, R. o. C.

ABSTRACT

Ten soybean varieties were selected to study their processing properties for
tofu. Tofu yields varied from 70.83g to 98.70g per 30g of soybean. The
yield of tofu was loosely correlated with the solid content of tofu (r<0.5).
A tofu gel centrifugation method was developed as a substitution method for
soybean quality evaluation in laboratory. A high regression coefficient was
found on~ield of tofu between the centrifugation method and conventional
method (R =0.84). Texture of tofu gels prepared by the centrifugation method
was also correlated with tofu made from varieties by the conventional method.
Regression coefficients obtained for softness and jelly strength were 0.64
and 0.66, respectively. Experimental results suggested that the gel
centrifugation method could be used to predict tofu productivity with small
size of soybean in laboratory.

INTRODUCTION

The quality of tofu is characterized by the yield, color and texture. It is

also known that the raw material of soybean can affect these characteristics.
Several investigations have been conducted in identifying the soybean
qualities that are related to its processing properties (1)(2)(3).
For laboratory operation, a centrifugation method as an evaluation
technique for determining tofu making properties of soybean ,,,ith small sample
size was developed.

MATERIALS AND METHODS

Soybean samples
Ten soybean varieties grown in the experimental field of AVRDC were used.

Preparation of Conventional Tofu and Small-scaled Tofu (S-tofu)

250 g of soybean were washed and soaked. The soaked beans were removed from
the water and ground for 5 min. by a Waring blender at the rate of 9:1 (water
to bean). The filtrate was boiled for 5 min. and cooled down to 85°C before
coagulated with calcium sulfate. After 20 min., it was poured into a wooden
mold (15x15x6 cm) and covered with a piece of cheese cloth. A pressure of
62.2 g/cm2 was put on the bean curd for 30 min. 30 g of soybean were used to
make S-tofu. The procedures were the same as in the preparation of
conventional tofu. Only the size of wooden mold was reduced to 7x7x4 cm.
1036
Centrifugation Method
30 ml of soymilk was poured into a centrifuge tube and heated in boiling
water for 5 min., and then poured into another centrifuge tube which
contained 2 ml of water and calcium sulfate and incubated at 80°C for 20 min.
The bean curd formed in the tube was cent.rifuged at various relative
centrifugal forces (ReF) for 10 min. The supernatant was removed and the gel
was weighed as the gel yield.

Evaluation of Tofu Texture

Texture of the tofu was evaluated by breaking test with a rheometer (Fudo
NRM-2010J-CW) according to Tsai et al (4).

TI+----< (om)-----I

Softness: b/a (mm/g)

a (g) Jelly strength: a(g)

statistical Analysis
-
Figure 1. Rheogram of tofu texture

All the experiments were performed in duplicate tests. By the SAS

statistical package, simple regressions were used to obtain a regression
model.

RESULTS AND DISCUSSION

Yields of Conventional Tofu and Small-scaled Tofu (S-tofu)

The results showed that there were a lot of differences in tofu yields
(563.30 - 956.34g /250g soybean) were shown among the varieties (Table 1).
The S-tofu yields of the ten samples ranged from 70.83g to 98.70g per 30g of
soybean; whereas the highest and the lowest tofu yields were KS#9 and AGS302,
respectively. A regression analysis was also made between yields of
conventional tofu and S-tofu, and the regression coefficient (R2) obtained was
0.86. There was a low correlation (R2<0.5) between the yield and solid of
conventional tofu and S-tofu. However, S-tofu making, as well as traditional
tofu, was too time-consuming to be suitable for mass screening.

TABLE 1
comparison of yields and solid contents among conventional tofu,
S-tofu and gel.

Yield ( g) Solid (t)

Variety
Tofu* S-tofu Gel** Tofu* S-tofu Gel**

AGS66 859.74 92.32 25.21 15.15 14.36 11.87

AGS129 756.04 89.99 22.14 14.29 11.01 11. 67
FWCPT 637.51 83.01 17.72 13.88 11.35 ·10.58
PMT 644.46 82.84 17.70 12.12 11.98 9.55
CS#1 613.58 74.20 16.37 19.61 19.65 12.52
Clary 647:38 75.92 14.37 20.65 21. 24 15.16
KS#5 735.94 87.10 17.80 15.42 16.50 12.28
AGS302 563.30 70.83 14.12 18.66 19.05 13.72
HKT 620 •. 79 81.60 16.07 17.01 17.50 12.31
KS#9 956.34 98.70 26.70 17.19 17.84 12.45
* Conventional tofu. ** Gel was obtained by the centrifugation method.
 1037
Establishment of the Centrifugation Method
Irregular shaped gels were formed under the RCFs from 55 xg to 223 xg, and
great variation of gel yields occurred at the range of 1397 - 2738 xg. The
RCF of 894 xg resulted in the lowest C.V. \ (1.40) and was selected to make
gels for further experiments (Table 2).

TABLE 2
Gel yield and rotor speed by using the centrifugation method.
RCF* (xg) Rotor speed (rpm) Gel yield** C.V. ***
%

55 1000 2s.14±0.97 3.86

223 2000 17.17±0.43 2.54
503 3000 13.81±0.23 1. 70
894 4000 11. 61±0 .16 1.40
1397 5000 10.02±1.09 10.81
2012 6000 8. 24±1. 03 12.50
2738 7000 8.67±1.32 15.22
* relative centrifugal force. ** mean±S.D.(standard deviation)
*** coefficient of variation.

Yields Comparison of S-tofu and the Gels Obtained by the Centrifugation

Method
Varieties KS#9 and AGS302 respectively demonstrated the highest and the
lowest yields of s-tofu as well as gel (Table 1). The regression coefficient
(R2) of yields between S-tofu and gel was 0.84, no matter how a low
relationship in solid content they might be. Texture of tofu gels were also
evaluated by rheological measurement as it was considered to be an another
important quality indicator of tofu. The regression coefficient of jelly
strength and softness between gel and S-tofu were 0.64 and 0.66,
respectively. This result indicated that the centrifugation method could be
a suitable way of evaluating the gel qualities.

CONCLUSIONS

The best way of evaluating soybean quality for tofu-making is to process tofu
directly with the conventional method. Although small-scaled tofu method
could be applied in soybean variety screening, it is time-consuming in the
sample preparation. A simpler method, gel centrifugation, was therefore
developed with a high regression coefficient in tofu yields among soybean
varieties.

REFERENCES

1. Wang, H.L., Swain, E.W. and Kwolek, W.F. 1983. Ef~ect of soybean
varieties on the yield and quality of tofu. Cereal Chem. 60(3): 245-
248.
2. Saio, K., Kamiya, M., and Watanabe, T. 1969. Food processing
characteristics of soybean 11S and 7S proteins. Part I. Effect of
difference of protein components among soybean varieties on formation
of tofu-gel. Agr. Biol. Chem. 33(9): 1301-1308.
3. Skurray, G., Gunich, J., and Carter, O. 1980. The effect of different
varieties of soybean and calcium ion concentration on the quality of
tofu. Food Chern. 6:89.
4. Tsai, s.-J., Lan, C.Y., Kao, C.S., and Chen, S.C. 1981. Studies on the
yield and quality characteristics of tofu. J. Food Sci. 46:1734.
USE OF THE MODERN TECHNOLOGY IN IMPROVEMENT OF CHINESE TRADITIONAL FOOD
TOU-NAO PROCESSING

SHEN GOU-OI, CHENG CHUANG-Jl, LI SHAO-Jl*

Shanxi Agricultural University, Shanxi Taigu, 030801, P.R. China
*Taiyuan Industrial University, Shanxi Taiyuan, 030002, P.R. China

ABSTRACT

'Tou-nao', the Chinese traditional food was invented by the famous artist, scholar and Chinese doctor,
Mr. Fu Shan, who lived during the Ming dynasty about 350 years ago, for his old mother as a morning
drink in winter. It has continued through many families and restaurants. The ingredients of Tou-nao,
which include wheat flour, essential mutton, Chinese yam and some other auxiliary materials, are
satisfactory for old people's health from the modern nutritional point. Commercial formulae and
procedures for Tou-nao are initially documented and the product is described from a traditional point
of view. Then experiments to improve the pretreatment of the materials and the procedure in the
laboratory are introduced.

MATERIALS AND METHODS

Tou-nao is a nutritious porridge. It is manually made and sold in restaurants for winter mornings. The
formula and procedure used have been handed down from ancient times. It has been the habit for
many old people to take a bowl of Tou-nao every morning in winter to keep good health. The process
for making Tou-nao has received little systematic study despite its long history. The vast historical
record and oral instruction about the magical effect of Tou-nao indicate that it is especially fit for weak
old people and patients marked by deficiency of vital energy and lowering of body resistance. There
are many records about the Tou-nao curing diabetes and maintaining a good status. The ingredients
of the Tou-nao include wheat flour, essential mutton, Chinese yam and some other auxiliary materials.
In the process for making Tou-nao, steam cooked wheat flour, pretreated Chinese yam chips and
cooked shredded mutton are boiled together in a soup where some clean sheep bones have previously
been boiled. Then some other auxiliary condiments are poured into the product to fit a personal taste.
From a traditional viewpoint, an ideal Tou-nao has the following characteristics: brainlike shape and
white colour cream without any muttony flavour. It has been verified by practice that the magical
effect of the Tou-nao is mainly due to the correct prescription and adding of Chinese yam. In different
areas of China, the Chinese yam is used with rice or seeds of Job's tears to make other nutritious
porridges such as: Shen-xian (celestial) Zhou (porridge) and Zhu-yu (pearl and jade) Zhou. These
porridges have significant effects on some patients suffering from dyspepsia, night sweats, listlessness,
impotence and premature ejaCUlation. As a food material the Chinese yam is rich in protein (9.4%,
dry) which is in a balanced amino acid pattern with a lysine content of 320 mg/lOO g. Its effect is
related to the chemical and amino acid compositions of its protein. From Table 1 it can be found there
are balanced components and many kinds of trace elements, zinc, selenium and others.
Many studies have indicated that lysine is a limiting component of cereals resulting in a low
biological value. The biological deficiencies of the protein quality do not really cause problems and
 1039
Table 1 The chemical and amino acid composItIOns of the Chinese yam
(mg/IOO g fresh)
Moisture 87.4% Tryptophan 57 Lecine 120 Mg 20
Protein 2 Histidine 28 Cystine 25 Fe 0.3
Lysine 64 Arginine 178 Serine 121 Mn 0.12
Methionine 23 Tyrosine 47 K 213 Zu 0.27
Glutamic acid 307 Alanine 87 Na 18.6 Cu 0.24
Isoleucine 78 Aspartic 152 Ca 16
Threonine 57 Valine 67 P 34
Phenylalanine 57 Glycine 55 Cw 0.55

they may be improved by: (1) supplementing with synthetic limiting amino acid(s) to upgrade the
nutritive value, and (2) complementing one food material with another to increase protein quality as
well as quantity. The Chinese yam is an ideal natural additive for this purpose. From the modern
nutritious viewpoint there are gluelike substances, choline, amylase and polysaccharides in the Chinese
yam. The Chinese yam can be used as a staple food material for the everyday diet of many people. In
the last decade the output of Chinese yam has greatly increased in China (about 7500-10 000 kg/ha).
The Chinese yam is a stem-tuber crop. There are several species of it. The moisture of the fresh stem
tuber is about 85-87%. The cuticle is very thin (less than 0.1 mm) in the fresh state and there are many
fibrous roots scattered on the surface of the cuticle (about 4--6 sticks/cm 2 with 1-1.2 mm diameter).
The fibrous roots extend into the pulp of the stem tuber about 1.2-1.5 mm deep. Under the nearest
cuticle there is the precious gluelike substances which are very important for its nutrition and special
effect in adjusting the physiology of the human body.
The traditional dehydration procedure for the Chinese yam is as follows: washing-peeling-shaping-
sun drying, or washing-steam heating-peeling-shaping-sun drying. Another special pretreatment
procedure is as follows: washing- peeling-sulphur smoking-air drying-kneading alternately (pass
through much times). The product has a column-like shape and bright white colour with fine and close
texture. It has a good mastication property after rehydration and is a traditional export to Southeast
Asian countries.
Several problems exist in the pretreatment of the Chinese yam: (1) the loss of the precious
substances is very high (about 17-20%), and (2) the efficiency of the traditional pretreatment is very
low. Every year the loss of the Chinese yam is very high due to rough treatment.
According to the construction property and considering the difference between the cuticle and the
fibrous roots we have treated it in a laboratory and induced a new procedure as follows: washing-hot
air blowing-cutting the fibrous roots helped by vacuum apparatus-classing-laser melting the vestiges
of the fibrous roots-flash heating with ultrasonic generator in a vessel filled hot water (about 100°C)-
alkali treatment-scraping and cool water blowing-shaping (to cut into chips with 3-3.5 mm thickness)-
drying in microwave oven-packaging. The cuticle and the vestige of the fibrous root have significantly
different colours. The vestige of the fibrous root is deeper brown. The laser apparatus can easily hit the
vestiges under the guidance system of optics according to the colour property and melt them out in very
short time. Comparing this procedure with the traditional procedure, the loss of the precious
substances diminishes from 17% to 5%. The mastication after rehydration is better than that of the
procedure treated by much kneading and without sulphur vestige. The loss of other nutritious
substances is also lower, especially the loss of the carotene and vitamin C.

CONCLUSION

(1) As the average age of the world population becomes greater and greater, it is important to study
special food to fit old people and to find more green resources for the food material. (2) The Chinese
food culture follows the principle: the food and the drug are from the same resources. Is it important to
develop the world food culture to dig up these precious heritages? (3) There are many natural
resources which have an abundance of nutritious substances. Combining these materials with
1040
conventional food materials not only increases the food quality but also improves the food quality.
(4) To extend the food resources and to diminish the loss of precious food materials, modern
technology will play more important part in the treatment of the materials.

REFERENCE

1. Wang Guang-ya, The Food Composition Table, 1991, Renmin Sanitary Publishing Co.
 FROZEN SfABIUZEDMINCE AS A SOURCE OF PACIFIC WHI11NG SURIMI

Simpson, RI., Kolbe, E2., MacDonald, G.A3 ., Lanier, T.e., and Morrissey, M.T I.
1 Dept. of Food Science, Oregon State University, Corvallis, OR, 97331. 2 Dept. of
Bioresource Engineering, Oregon State University, Corvallis, OR, 97331. 3 CRI Crop
and Food Research, PO Box 5114, Nelson, New Zealand. 4 Dept. of Food Science,
North Carolina State University, Raleigh, NC, 27695.

ABSfRACT

Pacific whiting is available off the U.S. West Coast for about six months each year.
A plan to extend the period of shore-based surimi production from whiting was
investigated. Headed/gutted fish, and mince stabilized with cryoprotectants, were
frozen, stored for up to six months, then processed into surimi. Measurements of
texture and color were compared with those for a surimi control sample. Frozen
mince stabilized with 6% sucrose and stored at -20 C resulted in production of
0

good quality surimi. Pilot scale yield and freezing rate studies indicated the
potential feasibility of commercial-scale production.

INfRODUCIlON

Pacific whiting CMerluccius productus) are in abundance off the U.S. West Coast for
about six months per year; annual landings approach 200,000 tons. This fish is
characterized by a soft texture and presence of a myxosporidian parasite linked to
rapid and severe enzymatic softening when cooked.
Surimi is thus an attractive process because of the effectiveness of well-mixed
enzyme inhibitors. In any case, acceptable quality of frozen fillets and surimi
requires that the fish be rapidly chilled on board and processed within a day or two
of capture.
Previous work has demonstrated the feasibility of making suitable surimi from
a frozen mince of New Zealand hake stabilized with sucrose [1]. Such a process
could lengthen the season for onshore surimi production on the U.S. West Coast.
The objectives of this study were to evaluate the potential of producting good
quality surimi from frozen Pacific whiting, and to identify important process
variables that might influence commercial feasibility.
1042
MATERIALS AND ME1HODS

All samples were made from rapidly chilled whiting processed within 24 hours of
harvest. Samples included headed and gutted (H&G), unstabilized mince (UM),
mince stabilized with sucrose (6-12%) and 0.2% polyphosphates (SM), and surimi.
Preparation of surimi with 8% cryoprotectants followed a standard process [2].
Samples were vacuu~n packed, frozen in a blast freezer, then stored at -20· C, -34· C,
and -50· C until testing. At intervals from 0 - 6 months, samples were made into
surimi and compared in quality with surimi stored at -34· C as a control.
Dimethylamine (DMA) content was determined by a modified copper
dimethyldithiocarbamate colorimetric procedure. Color of surimi gels was measured
using a Minolta Chroma Meter CR-300 which reports L*, a*, b* color coordinates.
Samples were compared for whiteness calculated as 100-((100-L*)2+a*2+b*2)1!2.
To measure gel forming ability, frozen surimi was tempered and chopped
with salt in a vacuum cutter, adjusting moisture content to 78%, salt content to 2%.
The resulting past~ was extruded into metals tubes, cooked at 90·C for 15 min then
cut into samples and tested to failure in the torsion mode, as described by Simpson
et al. [2].
Yield and freezing rate effects were tested over a 3 month period under
conditions described by Simpson et al. [3].

RESULTS

Good quality surimi was made from mince stabilized with 6-12% sucrose (SM) and
stored at -20 • C for up to six months. Failure strain remained above 2.2 (Fig. 1).
Values of 1.9 and above indicate surimi that will pass the double fold test.
Surimi made from headed and gutted fish (H&G) and from unstabilzed mince
(UM) was of poor quality (Fig.l). DMA formation in UM was also excessive (Fig.2).
Whiteness of surimi made from mince was less than that of control (77 vs.
81.3) but still considered commercially acceptable.
Surimi yield made from pilot-scale lots of 15 kg of mince was comparable to
that of the control. Minimal differences were noted in surimi gels made from SM
when frozen over a range of freezing times, to 2 days.

PROCESS IMPUCATIONS

Initial results suggest the potential to: land whiting quickly at nearby coastal
processing plants where it can be minced, mixed, and frozen; partially process and
freeze mince on mid-size factory trawlers which have no surimi processing ability;
regulate and lengthen shore-based surimi process schedules; utilize continuous drum
freezers or batch vertical plate freezers; manufacture new products from frozen
mince, using reduced, or alternative stabilizers.
Future efforts will determine suitable stabilizer alternatives and levels, frozen
storage stability of different product forms, and economics of production.
 1043
3 Failure Strain 300 DMA Content (ug/g)
+UMG-:ZOOC
9O"CCook I '* UM C!I-6O"C
*H&GG-:ZOOC
250 +8% Sucrose G-:ZOOC
.6% Sucrose O-:W"C

200

150

100
+12% Suerose G-:ZOOC
oe-UMG-:ZOOC
0.5 -H&G G-:ZOOC 50
'*8% Sucrose G-:ZOOC
+SUrimIO-:W"C

50 100 150 200 50 100 150 200

Storage Time (days) Storage Time (days)

Figure 1. Failure Strain of Figure 2. DMA levels during

surimi vs. storage time frozen storage

REFERENCES

1. MacDonald, G.A., Wilson, N.D., and Lanier, T.C. 1990. Stabilized mince: an
alternative to the traditional surimi process. IN Proc. of the IIR Conf. on Chilling
and Freezing of New Fish Products. Sept 18-20. Torry Research Station,
Aberdeen, Scotland.

2. Simpson, R., Kolbe, E., MacDonald, G.A., Lanier, T.C., Morrissey, M.T. 1993a.
The feasibility of producing surimi from partially processed and frozen Pacific
whiting (Merluccius productus). In Review, J. of Food Sci.

3. Simpson, R., Kolbe, E., MacDonald, G.A., Lanier, T.C., Morrissey, M.T. 1993b.
Process variables affecting surimi made from frozen stabilized mince of Pacific
whiting (Merluccius productus). In Review, J. of Aquatic Food Prod. Tech.

ACKNOWLEDNEMENTS

This publication is the result in part of research sponsored by Oregon Sea Grant
with funds from the National Oceanic and Atmospheric Administration, Office of Sea
Grant, Department of Commerce, under grant no. NA89M-D-SG108 (project no.
R/PD-57) and from appropriations made by the Oregon State Legislature. The U.S.
Government is authorized to produce and distribute reprints for governmental
purposes, notwithstanding any copyright notation that may appear hereon.
 PRODUCTION OF BRASILIAN IMITATION CHEESE WITH A
PARTIALLY HIDROGENATED VEGETABLE FAT

MIRNA L. GIGANTE*; SALVADOR M. ROIG**

*DETA - UNESP - CX.P. 136, 15054-000 Sao Jose do Rio Preto - SP - BRASIL
**FEA - UNICAMP - CX.P. 6121, 13081-970 Campinas, SP - BRASIL

ABSTRACT
Brasilian fresh cheese (minas type) were prepared from skimmed milk
to which partially hydrogenated vegetable fat or milk fat was added.
The cheeses were compared as to the transfer level of milk
componentes yield and sensory evaluation.
Results showed that there was. a significant difference (P(0,05)
between the fat and total solids transfer level, but there was not a
significant difference between the transfer level of proteins and ashes.
Cheese with vegetable fat showed 2,86% increase in yield as compared
to those made with milk fat. In the sensory evaluation a difference was
detected between the cheeses, but there was no preference for neither one.

INTRODUCTION
Imitation cheese is a product in which milk fat is substituted by
vegetable fat, on a protein system from milk origin or being a vegetable
fat used in a product produced with caseinates, vegetable proteins,
stabilisers, emulsifiers, flavour and colour (5).
Presently different imitation cheeses are being produced as: mussarela,
cheddar, swiss, colby, gouda, etc., and are commercialized in different
countries as England, United States, Sweden and Australia.
"Minas Frescal" cheese is a widely produced cheese in Brazil, and
for this reason was choosen for this feasability study.

MATERIAL AND METHODS

Four set of cheeses were prepared (4) with skim milk and milk fat
added (3,5% - control cheese - CC) or partially hidrogenated vegetable fat
added (3,5% - imitation cheese - IC). The transfer level values were
calculated with reference to milk and cheese whey weight and composition
and yield was expressed as cheese kg per 100 milk kg employed for fixed
 1045
cheese moisture. It was used "t" Student's test to verify the existence
as significative difference between transfer level values and yield.
Milk, cheeses and cheese whey (CC and IC) were evaluated with respect to
total solids (TS), total proteins (TP) ash (A) (1) fat (F) (2) content
,md InC'tos€' (L) by difference. A 20 persons panel was used and run a
Triangular alld Paread - comparisons - preference tests, (3) for difference
rmd pI'r> ferencf' r'e:,;p(~" Live 1 y •

RESULTS AND DISCUSSION

The milk employed had normal composition with 11,94 TS, 3,55% f, 2S5%
TF, 0,7% f\ and 4,71% L.
TiJp tTclll~;f",r level value for fat was in average 89% for control and
94% for imitation cheese. There was a significative difference at 5%
level,l':1I1cl1 I'esulted in d transfer level value for TS also statiscally
di fel'prJ t. <I t" ')~~ 1 evel felT' the illli tation cheese, as can be seen on figure 1.

60-----'
[l] CC
§IC
-- -
-:--- ,-
t--
r-- - t--

~
~ - f=
!: 50
~ ,- I-
--
r- 1=
I--
(/)
I- 1== ~ 1=
...J
I-
1= ~
t--
1=
I- ~
f--
c-
c-
I-- ~
I--
l- t-- '--
1-
c-
I-- l- I-- '--
I-- ~

I- I-- 1--- r-
1--' I-- I--
40 I--

2 3 4

PROCESSING

Figure 1. Transfer level of total solids (TLTS-%) for control (CC) and
imitation (IC) cheese

There was no significative difference on the transfer level value for

TP and A either for control cheese and imitation cheese.
Wi ttl control and il1li taU on cheese were obtained yields respectively
of 14,69 and 15,11 kg cheese/lOO kg milk signifying an yield increase of
2,86%.
1046
On the sensorial evaluation, 11 from 20 panelist indicated the
different sample, 8 and 3 respectively prefered control and imitation
cheese. As result it was verifyed to exist a significative difference
between treatments, but do not exist statistical significative preference
at 5% level among them.

CONCLUSIONS

1. It was shown the feasabili ty of "Minas Frescal" imitation cheese

production.

2. Cheese whey of imitation cheese presented lower fat content as compared

to control cheese whey.

3. Imitation cheese presented higher transfer level value for fat and total
solids.

4. There was a significative difference (P(0,05) for transfer level value

total solids for imitation and control cheeses.

5. There was not a significative difference for transfer level value for
imitation and control cheeses.

6. Imitation cheeses presented yield increase of 2,86 with respect to

control cheeses.

7. Sensorial analysis indicated to exist significative difference (P(O,05)

for imitation and control cheese, however was not detected preference
for neither one.

REFERENCES

1. A.O.A.C. Official methods of analysis of the Association of Official

Analytical Chemists, 14 ed., Arlington, 1984, p. 1141.

2. Atherton, H.V. and Newlander, J.A., Chemistry and testing of dairy

products. 4 ed., Westport, Avi, 1981, p. 386.

3. Moraes, M.A.C., Metodos para avalia~ao sensorial dos alimentos. 5 ed.,

Campinas Ed., UNICAMP, 1985, p. 85.

4. Oliveira, J.S., Queijos: fundamentos tecnologicos. Sao Paulo, Ed. ,

UNICAMP, 1986, p. 146.

5. Walder, D., Imitation dairy produts - the identity problem. British

Food Journal 90 (3), 1988, pp. 117-119.

Authors thanks to Dr~. Maria Helena Damasio and to Refinadora de Oleos

Brasil sl A.
IMPROVEMENT OF BREAD FLAVOR BY A DAIRY SUBSTRATE
TREATED WITH ENZYMES AND LACTIC ACID BACTERIA

MASASHI YAMAMOTO, REIKO WATANABE,

TSUTOMU KANEKO AND HIDEKI SUZUKI
R&D Department, Central Research Institute, Meiji Milk Products
Co., Ltd.
1-21-3 Sakae-cho, Higash imurayama-sh i, Tokyo 189 Japan.

ABSTRACT

Iso-amy1a1cohol (iso-AmOH), iso-butyl alcohol (iso-BuOH) and ethyl caprylate

(EtOcapry) were the main aroma components in bread, and the content
of these components with a dairy substrate treated with several
enzymes and fermented with lactic acid bacteria (FCT) increased
from 1.2 to 1.6 times those of control bread. The increase in iso-
AmOH and iso-BuOH content in bread with FCT appeared due to increased
leucine and valine content in dough by adding FCT. The increase
in EtOcapry content may have been caused by the caprylic acid found
in FCT. The addition of the hydrolysate of milk protein and milk
fat to dough is thus shown to greatly improve bread flavor.

INTRODUCTION

In Japan, bread is widely produced by the sponge dough method, or

short-time fermentation method, of straight dough. Generally, aroma
components such as iso-AmOH and iso-BuOH in bread are present in
small amounts. Thus, a method was sought for improving bread flavor.
This was found possible by a dairy substrate treated with protease
and lipase and fermented with two strains of Lactococcus lactis
(FCT). The mechanism for improvement was studied.
MATERIALS AND METHODS

Preparation of bread
FeT (2.8% milk protein and 38% milk fat) and skim milk powder (Meiji
Milk Products Co., Ltd. Tokyo Japan) were used. FCT was prepared
by treating with commercial protease from Aspergillus, lipase from
1048
kid and lamb and strains two strains of 1. lactis subsp. lactis.
Agitation was carried out at 100 rpm during incubation at 35°C
for 48 hr. FCT was then heated at 90°C. The flour, yeast, yeast food,
sal t, sugar, and shortening were used to prepare the bread by the
sponge dough method.
Determination of flavor components in bread crumbs
This parameter was determined by head space gas chromatography (HS-
GC) (1) wi th modification. Extracts of bread flavor were prepared
by distillation and extraction methods and analyzed by gas chromatography-
mass spectrometry (GC-MS). n-Butylalcohol was used as the internal
standard. Individual flavor components were identified by relative
retention times with standards and GC-MS analysis.
RESULTS

Analysis of bread flavor

Acetaldehyde, ethylalcohol, n-propylalcohol, iso-BuOH, and iso-AmOH
were the main components. Fusel oils such as iso-BuOH and iso-AmOH
were the most important of these components.
Improvement of bread flavor by FCT
Table 1 shows the effects of FCr on aroma component production.
Iso-AmOH and iso-BuOH were the main aroma components in bread. The
addition of FCT raised their relative amounts by 1.6 and 1.2 times
that of control bread. EtOcapry and ethyl caprate (EtOcapra) content
in bread crumb with FCT was 1.6 and 1.3 times that of control bread.

TABLE 1
Effects of FCT on the production of aroma components in
bread crumb
Component None 5% FCr
iso-AmOH 21. 7 J1 g/g 34.4 J1 g/g
iso-BuOH 18. 0 "g/150g 22.3 "
EtOcapry 45. 6 J1 59. 6 J1 g/150g
EtOcapra 27.5 " 33.9 "

Changes in iso-AmOH and leucine content in dough

during fermentation
Baker's yeasts produce alcohols such as iso-AmOH and iso-BuOH from
amino acids through the Ehrlich mechanism. Changes in iso-AmOH and
leucine concentrations in dough sponge during fermentation
(Fig. 1) were investigated. Leucine in dough reacted wi th baker's
yeast during sponge dough fermentation. However, its content in
mixing dough with FCr exceeded that without FCr. At the final proofing
step, this reaction occurred rapidly with the production of iso-
AmOH.
 1049
-: 60 Sponge Mixing Final Baking 60
~ dough dough proofing ~
., .. .1
c::....
ot.)
~ ~ ~

11
z: r:...

40

.......
,.....
...... 20
::l.

=
•
Q

41
I

...
0
III

Time (min)

Figure 1. Changes in iso-AmOH and leucine concentrations in dough

sponge during fermentation

DISCUSSION

Iso-AmOH, iso-BuOH, EtOcapry, and EtOcapra were the main aroma components
in bread. Their amounts increased from 1.2 to 1.6 times those of
control bread. Amino acids such as leucine and val ine, which are
present in FCT, are precursors of iso-AmOH and iso-BuOH, and they
reacted with baker's yeast during dough fermentation to increase
iso-AmOH and iso-BuOH content in bread. Baker's yeast contains
an enzyme which converts caproic acid to EtOcapra. Thus possibly,
increase in EtOcapry contents may have resulted from the capric
acid found in FCT. Lactic acid bacteria also contribute to the hydrolysis
of the substrate and produce lactate. It is evident from the above
that the addition of the hydrolysate of milk protein and milk fat
to dough greatly improves bread flavor.
REFERENCES

1. Kaneko, T., Suzuki, H. and Takahashi, T., Diacetyl Formation

and Degradation by Streptococcus lactis subsp. diacetylactis
3022. !&r. BioI. Chern., 1986, 50(10), 2639-2641.
 INTERACTION BETWEEN INHIBITORS OF
CALCIUM PHOSPHATE FORMATION AND CALCIUM ION

KAZUTAKA YAMAMOTO, HITOSHI KUMAGAI, TAKAHARU

SAKIYAMA, HIROYUKI OGAWA, AND TOSHIMASA Y ANO

Department of Agricultural Chemistry, The University of

Tokyo, Bunkyo-ku, Tokyo 113, Japan

ABSTRACT

The interaction between the inhibitors of calcium phosphate formation and calcium ion was
investigated. First, the induction time in base titration experiments was measured to evaluate
the inhibitory activity against calcium phosphate formation. Inhibitors such as citrate, alginate,
and poly-L-glutarnate delayed the induction time. It was suggested that not only the monomer
concentration but also the molecular weight of the inhibitor was the important factor for the
inhibitory activity. Second, calcium ion activity was measured with a calcium ion selective
electrode. The calcium ion activity gradually decreased in the same concentration range as
tested in the base titration experiments.

INTRODUCTION

Calcium phosphate precipitation leads to clogging of the membrane pores, resulting in a lower
flux during ultrafiltration of milk and whey. Moreover, the formation of calcium phosphate is
also important from a physiological point of view, e.g. biological calcification of bones and
teeth. Some inhibitors of calcium phosphate formation affect the biological calcification.
However, it is not clearly understood how the inhibitor behaves in the inhibition of calcium
phosphate formation. In this study, the inhibitory activity against calcium phosphate formation
was evaluated by the induction time in the base titration experiments, and the calcium ion
activity was measured to evaluate the interaction between the inhibitors and calcium ion. In
addition, the effect of molecular weight of the inhibitor on the inhibitory activity was
investigated using oligo- and poly-L-glutamate.
 1051
MATERIALS AND METHODS

Materials
Poly-L-glutamate (PLG) (Mw = lx104 and 6xl04; Sigma Chemical Co.), alginate (GI-L type
; Kimitsu Chern. Ind., Tokyo), and citrate (Kanto Chemicals Co. Ltd., Tokyo) were used.
Oligo-L-glutamate (n = 10, purity 48%) was synthesized by the solid phase method using
Fmoc protecting group. 1

Inhibitory activity against calcium phosphate formation

The induction time, at which the transition of amorphous calcium phosphate to crystalline is
known to occur,2 was measured in base titration experiments as an index of the inhibitory
activity3. Details of experimental procedure are described in our previous paper.4

Calcium ion activity

Calcium ion activity was measured by the ion meter connected to calcium ion selective
electrode and to reference electrode (double junction type).

RESULTS AND DISCUSSION

Figure 1 shows the effect of concentration on the induction time. As shown in Fig.l (a), the
plots for polyelectrolytes were sigmoidal, while the plot being concave upwards in the case of
citrate. These results suggest that the inhibitory mechanism of polyelectrolyte inhibitor is
different from that of a low molecular weight inhibitor. In Fig. 1 (b), the inhibitory activity of
PLG was highest, and that of oligo-L-glutamate was lower than that of PLG. However, L-
glutamate, the monomer of PLG, had almost no inhibitory activity. This result indicates that
not only the monomer concentration but also the molecular weight was the important factor for
the inhibitory activity of oligo- and poly-L-glutamate.

0
0
Citrate
Alginate (M/G=2.4)
t--- ,-
•
t';.
Glu

•
PLG (MW~1 x1 04 )
Alginate (M/G=0.S3)
t';. PLG(MW~1 x1 04 )
0 PLG (MW~6x1 04 )
0
105 lOS
~
()
Q) U • (GIU)10
Q)
.!!!... .!!!...
Q) t';.~ 'DL'>&
E DO Q)
:;::; E
:;::;
c:
•
0 10 4 0 c: 104
:;::; 0
:;::;
()
lJ. a
•
....
::l ()
-0
.E 0
•
::l
-0
.E • •
• A
• ••
0 t';.

••
0 0 0
10 3 l! l! l! r;; iii lcf ~ ~ t';. trl.!~

10 1 l<f 103 10 4 101 102 10 3 10 4

Concentration IIlM]
Concentration IIlM]
(a) (b)
Figure 1 Effect of concentration on the induction time. Polyelectrolyte concentrations were
expressed by the monomer concentrations.
1052
Figure 2 shows the effect of concentrations on the calcium ion activity. The calcium ion
activity gradually decreased in the presence of inhibitors, indicating that the inhibitors interacted
with calcium ion.

~
.§.
5

4
o
•
~
.
o D
~
o D

~
.S;
3
t5co o Citrate
c
.Q 2 D Alginate (M/G=2.4)
E • Alginate (MIG =0.53)
:l
·0 b. PLG (MW~1x1 0 4 )
(ij
()

0
10° 10 1 102 103 104
Concentration [IlM]

Figure 2 Effect of concentration on the calcium ion activity. Total calcium concentration was
5mM in O.2M Tris-Hel (pH 7.4)

As shown in Figs. 1 and 2, the calcium ion activity decreased in the same concentration
range tested in the evaluation of the inhibitory activity. This result indicates that the interaction
between the inhibitors and calcium ion plays an important role on the inhibition of calcium
phosphate formation.

REFERENCES

1. G. B. Fields and R. L. Noble, Solid phase synthesis utilizing 9-fluorenylmethoxycarbonyl

amono acids, Int. J. Peptide Protein Res., 35, 161-214 (1990).
2 A. L. Boskey and A. S. Posner, Conversion of amorphous calcium phosphate to
microcrystalline hydroxyapatite. A pH-dependent, solution mediated, solid-solid conversion, L
Phys. Chern. 77: 2313 (1973).
3. J. L. Meyer and E. D. Eanes, A thermodynamic analysis of the amorphous to crystalline
calcium phosphate transformation, Calcif. Tissue Res. 25: 59 (1978).
4. K.Yamamoto, H. Kumagai, T. Sakiyama, C.-M. Song, and T. Yano, Inhibitory activity of
alginates against the formation of calcium phosphate,.BioscL Biotech.. Biochem., 56, 90-93
(1992).
 STUDIES ON THE OPTIMUM CONDITIONS TO UTILIZE BIOLOGICALLY
ACTNE PEPTIDES DERIVED FROM FOOD PROTEINS

MASAAKI YOSHIKAWA and HIROYUKI FUJITA*

Department of Food Science and Technology, Kyoto University, Kyoto 606, and
*The Nippon Synthetic Chemical Industry Co. Ltd., Ibaraki-shi, Osaka 567, Japan

ABSTRACT

As a model study to effectively utilize biologically active peptides derived from

food proteins, the optimum conditions for the release of inhibitory peptides for
the angiotensin-converting enzyme (ACE) were investigated. Peptides which
showed apparent inhibition of the enzyme were classified into three groups: true
inhibitors, substrates and pro-drug type peptides which are converted from the
substrates to the true inhibitors in vivo. Mter intravenous administration, all of
these peptides suppressed the blood pressure elevation caused by the angiotensin
I injected. Mter oral administration in spontaneously hypertensive rats (SHR),
however, only the true inhibitors and pro-drug type peptides lowered blood
pressure. The thermolysin digest of "Katuo-bushi", dried bonito which contained
many kinds of true inhibitors and pro-drug type peptides showed long-lasting
anti-hypertensive effects. On the other hand, those obtained by proteases from
digestive tracts were ineffective in lowering the blood pressure of SHR after oral
administration. Thus, because of its good taste and anti-hypertensive effect, the
thermolysin digest of dried bonito is an ideal food stuff for a physiologically
functional food.

INTRODUCTION

Many kinds of biologically active peptides are released from food proteins during
their enzymatic digestion [1,2]. The angiotensin-converting enzyme (ACE) is a
dipeptidyl carboxypeptidase which catalyses the conversion of angiotensin I to
angiotensin II, a strong pressor. Therefore, inhibitors for ACE show anti-
hypertensive effect [3]. Many inhibitory peptides for ACE have been isolated
from the enzymatic digests of food proteins [4-7]. However, we found that some of
them were ineffective in lowering blood pressure of spontaneously hypertensive
rats (SHR) after their oral administration. In order to obtain orally effective
peptides, factors affecting the anti-hypertensive activity were studied and
conditions for the enzymatic hydrolysis of proteins were optimized.
MATERIALS AND METHODS

The inhibitory activity for ACE was measured by the method of Cushman and
Cheung and expressed by ICSO [8]. The inhibitory activity was also measured after
1054
preincubation of the peptides with ACE for 3 hrs. The anti-hypertensive
activities of peptides against injected angiotensin I (100 ng/Kg) were measured
with a pressure transducer which was cannulated into the carotid artery of
normotensive rat. The anti-hypertensive effects of peptides were also measured
by the tail cuff method using a VR-SOOO (Veda Seisakusho) after oral
administration of the peptides. Stability of the inhibitory peptides for
preincubation with ACE was tested by HPLC equipped with an ODS column.
RESULTS AND DISCUSSION

Peptides showing inhibitory activities for ACE were obtained from the
thermolysin digests of dried bonito and chicken muscle and the peptic digest of
ovalbumin (Table 1). These peptides are classified into three groups; 1) The true
inhibitors of which 1CSO values are not altered by preincubation with ACE, 2)
Substrates for ACE, of which ICSO values are increased by the preincubation with
ACE, and 3) Pro-drug type peptides which are converted from substrates to true
inhibitors for ACE by digestion with ACE itself or other proteases in vivo. After
intravenous administration, all of the peptides suppressed the hypertensive
effect of angiotensin I. On the other hand, only the true inhibitors and the pro-
drug type peptides lowered the blood pressure of SHR after oral administration.
The pro-drug type peptides required a longer time to achieve their maximal
effect than their activated forms. This 2-hr delay may be explained by the time
required for the absorption of slightly larger peptides and that required for the
enzymatic conversion into true substrates in vivo. The thermolysin digest of
dried bonito, which contained many kinds of true inhibitors and pro-drug type
peptides, showed long-lasting anti-hypertensive effects after oral
administration. On the other hand, the digests obtained by gastrointestinal

Table 1
Relationships between IC 50 for ACE and Anti-hypertensive Effects

ICsO (flM) Stability Anti- Blood Pres-

Peptides origin for ACE by An-I by sure by p.o.
-Preinc. +preinc. HPLC (% ) i.v. (max • .1mmHg)

Inhibitor type
IY Bonito 2.1 1.9 100 + -19 2hrs
LW OVA 6.8 6.6 100 + -22 2hrs
IKW Chicken 0.21 0.18 >95 + -17 4hrs
IKY Bonito 2.2 2.4 >95 + -17 4hrs
IKP Bonito 6.9 6.7 >95 + -19 4hrs
LKP Bonito 0.32 0.3 >95 + -18 4hrs
IWH Bonito 3.5 3.5 100 + -30 4hrs
IVGRPR Bonito >300 >300 100 + -17 6hrs
Pro-drug type
IWHHT* Bonito 5.8 3.5 0 + -26 6hrs
LKPNM* Bonito 2.4 0.76 80 + -23 6hrs
IVGRPRHQG** Bonito 2.4 23 0 + -14 8hrs
Substrate Type
FFGRCVSP OVA 0.4 4.6 0 + 0
FKGRYYP Chicken 5.8 34 0 + 0
.
*: Activated by ACE , **:Activated by Trypsin •
Underlined peptides are activated forms of pro-drug type peptides.
 1055
proteases are ineffective in lowering the blood pressure of SHR after oral
administration (data not shown). This means that we cannot expect such an anti-
hypertensive effect by merely eating the fish. In this context, digestion by
microbial proteases during the food proceSSing step has an important meaning
for the expression of potential anti-hypertensive activity.
By controlling the hydrolysis with thermolysin at 80°C, we can decrease
the amount of enzyme and prevent the growth of bacteria. Furthermore,
thermolysin digest has no bitter taste while other digests did taste bitter. Fishy
flavor has been another problem in utilizing fish protein hydrolysates as a food
constituent. This problem could be overcome by using Katuo-bushi, a Japanese
traditional seaoning material made of dried bonito as a starting material. By
using residues of dried bonito, which are obtained in large quantity after the
industrial flavor extraction, we can obtain a hydrolysate with the anti-
hypertensive effect and no specific taste.
Thermolysin was also effective in releasing inhibitory peptides for ACE
from zein [6] and opioid peptides from gluten [9]. However, thermolysin digest of
ovalbumin showed very weak inhibitory activity for ACE (data not shown). This
means that the most suitable protease to be used is variable depending on
substrate proteins and biological activities to be aimed.
Another factor to affect oral avairability of peptides is a physical state of
the pep tides. Suetuna and Osajirna observed improvement in the oral availability
of ACE inhibitors by emulsification in egg yolk [5]. However, we could not
observe the same effect in our inhibitors. This may be probably caused because
our peptides are small enough to be absorbed without emulsification. In fact, we
observed an improvement in the oral availability by emulsification of a vaso-
relaxing octapeptide, ovokinin, which we isolated from the peptic digest of
ovalbumin (unpublished data).
REFERENCES

1. Brand, V., Teschemacher, H., Henschen, A. and Lottspeich, F., Novel opioid
peptides derived from casein (~-Casomorphins) I. Isolation from bovine casein
peptone. Hoppe-Seyler's Z. Physioi. Chern, 1979,360, 1211-6
2. Yoshikawa, M. and Chiba, H., Biologically active peptides derived from food and
blood proteins. In Frotiers and New Horizons in Amino Acid Research. ed.
K. Takai, Elsevier Science Publishers, Amsterdam, 1992, pp. 403-9
3. Cushman, D.W. and Ondetti, M.A., Inhibitor of angiotensin-converting enzyme.
In Progress in Medicinal ChemiStry, ed. G. P. Ellis and G. B. West., Elsevier /
North-Holland Biochemical Press, Amsterdam, 1980, 17, pp. 42-104
4. Maruyama, S. and Suzuki, H., A peptide inhibitor for angiotensin I converting
enzyme in the tryptic hydrolysate of casein. Agric. BioI. Chern., 1982,46, 1393-4
S. Suetuna, K. and Osajima, K., Blood pressure reduction and vasodilatory effect in
vivo of peptides originating from sardine muscle. Nippon Eiyo Shokwyo
Gakkaishi. 1989,52,47-54
6. Miyoshi, S., Ishikawa, H., Kaneko, T., Fukui, F., Tanaka, H. and Maruyama, S.,
Structure and activity of angiotensin-converting enzyme inhibitors in an
a-zein hydrolysate. Agric. BioI. Chern., 1991, 55, 1313-8
7. Yokoyama, K., Chiba, H. and M. Yoshikawa, Peptide inhibitors for angiotensin I
converting enzyme from thermolysin digest of dried bonito. Biosci. Biotec.
Biochern., 1992,56, 1541-5
8. Cushman, D.W. and Cheung, H.S., Spectrophotometric assay and properties of
angiotensin-converting enzyme of rabbit lung. Biochem. Pharmacol., 1971,20,
1637-48
9. Fukudome, S. and Yoshikawa, M., Opioid pep tides derived from gluten: their
isolation and characterization. FEBS Lett., 1992,296, 107-11
 ANTI-PLATELET PRINCIPLE FOUND IN THE ESSENTIAL OIL OF GARLIC
(A11iu. sativua L.) AND ITS INHIBITION MECHANISM

TOYOHIKO ARIGA, TAIICHIRO SEKI, KIYOSI ANDO, SACHIYUKI TERAMOTO AND

HIROYUKI NISHIMURA*
Department of Nutrition and Physiology, Nihon University
School of Agriculture and Veterinary Medicine,
3-34-1 Shimouma, Setagaya, Tokyo 154, Japan
*Department of Bioscience and Technology, Hokkaido Tokai University,
Sapporo 005, Japan

ABSTRACT

Garlic oil and its component methyl allyl trisulfide (MATS) inhibited
platelet aggregation induced by arachidonic acid (AA); however, the
aggregation induced by AA metabolites was not inhibited by these Allium
components. The Allium components inhibited production of thromboxane Ba
(TXBa), 12-hydroxyheptadecatrienoic acid (HHT), and prostaglandin Ea
(PGEa), as well as 12-hydroxyeicosatetraenoic acid (12-HETE), although
the AA release reaction was accelerated to some extent, indicating that
the inhibition would be caused by a metabolic impairment from AA to its
metabolites. The direct interaction between MATS and cyclooxygenase was
not revealed. MATS was absorbed by platelets, and some of it reached to
the platelet organella.

INTRODUCTION
The plants that belong to Allium family have been widely used as
nutlt'itious and physiologically functional vegetables. Among the
functions, antithrombotic activity afforded by essential oil of garlic is
the most reputed [1]. The authors initially found MATS as an active
principle for platelet aggregation inhibition in garlic oil, and
suggested that the AA cascade in platelet is one of the possible sites of
being blocked by MATS [2, 3]. In this paper, the interaction of MATS
with human and animal platelets is extensively studied.

MATERIALS AND METHODS

The essential oil of garlic was prepared by using a Linkens-Nickerson's
apparatus, and its component MATS was isolated by a preparative gas
chromatography. Platelet aggregation was measured by an aggregation meter
(DP-247E, Sienco Inc., U.S.A.). For the assaying of antiaggregatory
 1057
activity, the essential oil or MATS was added to the platelet-rich plasma
(PRP) from human or rabbits prior to the addition of an inducer. In
radiolabeling, rabbit platelets were labeled with 14C-AA (NEN Research
Products) according to Bills et al. [4]. 14C-AA-derived radioactive
metabolites were separated by an HPLC system (Hitachi 655A, Tokyo)
equipped with a column of Inertsil ODS (GL Science Co., Tokyo). For
identification of the radioactive eluates, authentic materials obtained
from Dupont/NEN Research Products were analyzed by the same system. 35S-
MATS was synthesized chemically by using sodium sulfide, 35S (Amersham,
England), allyl bromide and methyl bromide.

RESULTS AND DISCUSSION

Effects of Garlic oil and MATS on the AA-Releasing Mecbanisa.
14C-AA-Iabeled rabbit platelets were stimulated with thrombin, and the
amounts of the label released from various phospholipid moieties of the
platelets were measured. In the presence of garlic oil, platelet aggrega-
tion was suppressed, whereas the total amount of the label released,
which would be free AA and/or its metabolites, was markedly increased.
The most prominent phospholipid liberating the label was phosphatidylcho-
line (PC). MATS also enhanced AA liberation from PC.

Inhibitory Effect of Garlic Oil and MATS on the AA Metabolisa in

Platelets.
We performed quantitative analysis of the 14C-AA metabolites generated in
washed human platelets treated or not with garlic oil or MATS. The plate-
lets prepared could be aggregated fully in response to 50 #M AA contain-
ing 14C-AA, and could be inhibited by either garlic oil (2 #g/ml) or
MATS (2.6x10- 8 M) by 50~ of the maximum aggregation of the control plate-
lets. MATS or garlic oil seemed to inhibit both cyclooxygenase and lipox-
ygenase, since they suppressed production of the metabolites TXB2 and
PGE2, as well as 12-HETE. Cyclooxygenase appeared to be more susceptible
to MATS than lipoxygenase, since a MATS concentration sub-inhibitory for
12-HETE production (10- 8 M) caused appreciable inhibition of TXB2 and PGE2
formations.

Garlic Oil and MATS Do Not Inhibit T~-Induced Aggregation.

Platelet aggregation induced by either TXA2 or its analogues was not
inhibited by garlic oil or MATS. LASS, a labile aggregation-stimulating
substance prepared from human platelets [4], U-46619 (a PGH2 mimic), and
STA2 (a TXA2 analogue) were able to induce platelet aggregation even in
the presence of garlic oil or MATS. These results strongly suggest that
the Allium components specifically inhibit the metabolic pathways from AA
to its aggregatory metabolites, PGH2 or TXA2.

Effect of MATS on the enzyaatic reaction of cyclooxygenase.

In ex vivo experiments, garlic oil or MATS was found to interfere the
pathway from AA to TXA2. However, there is no evidence that such foreign
chemicals interact with enzyme molecules, and cause inactivation. Then,
we studied to know how MATS affects cyclooxygenase activity in in vitro
reaction system. As shown in Figure 1, the cyclooxygenase was inhibited
very weakly by MATS. In this experiment, the substrate 14C-AA remained
after reaction was measured (n=3; ±SD). The inhibition would not be
specific for the enzyme, since the complete inhibition was not obtained
even at the concentration higher than 10- 5M MATS, by which platelet
aggregation was fully suppressed [2]. Demonstration of real target(s) of
MATS within platelets should await further study.
1058
100~ ____________________--.

50
~
0
0
.....
• .-4
• .-4
,.0 0

~
-50

-100+---.--.---.---,--,..---.--.---'
-6 -4 -2
\0 \0 \0

MATS concentration (M)

Figure 1. Inhibitory effect of MATS on cyclooxygenase activity.

MATS-Uptake by Platelets.
Ingestion of garlic or its essential oil or even an active principle,
MATS, gives rise to decreased platelet aggregability. However, no evi-
dence has been obtained to indicate that platelets absorb such garlic
components. We studied the in vitro uptake of MATS using synthesized 35S-
MATS and washed human platelets. The majority of 35S-MATS was absorbed by
platelets. Distribution to the membrane reached to 92% of the total label
absorbed, and only 8% was shared by mitochondria, microsomes and cytosol.
This suggests that the antiaggregatory principle in garlic may positively
be absorbed by the platelets in vivo.

ACKNOWLEDGEMENT

This work was supported by grant from the Ministry of Education, Science
and Culture (Grant-in-Aid for Scientific Research (C), 03806015), Japan.

REFERENCES

1. Fenwick, G.R. and Hanley, A.B., The genus Allium. In CRC Critical
Reviews in Food Science and Nutrition, CRC Press, Inc., London, Vol.
23, Issue 1, pp. 1-73.

2. Ariga, T., Oshiba, S. and Tamada, T., Platelet aggregation inhibitor

in garlic. Lancet, 1981, i, 150-1.

3. Bills, T.K., Smith, J.B. and Silver, M.J., Metabolism of [14C]arachi-

donic acid by human platelets. Biochim. Biophys. Acta, 1976, 424, 303-
14.

4. Willis, A.L., Vane, F.M., Kuhn, D.C., Scott, C.G. and Petrin, M., An
endoperoxide aggregator (LASS), formed in platelets in response to
thrombotic stimuli. Prostaglandins, 1974, 8, 453-507.
 LIPID PEROXIDE DECREASING ACTIVITY OF MICROBIAL CELLS

MASANORI ITO and KAZUOKI ISHIHARA

Institute for Intestinal and Environmental Microbiology
Advance Co.Ltd.,1-35-2 Shimoishihara,
Chofu-shi,Tokyo 182

ABSTRACT

Microbial cells decreased linoleic acid hydroperoxide (LPO)

in phosphate buffer, by binding LPO and reducing hydroperoxide
group of LPO. The LPO-decreasing activity of microbial cells
was not so affected by digestive enzyme or bile.

INTRODUCTION

Peroxidatants in foods can cause damage to artery and other

tissues because they are potential source of free radicals.
It is important to decrease the level of peroxidants in foods
and intestinal contents.
In this paper, we describe decreasing activity of
microbial cells to linoleic acid hydroperoxide (LPO) in vitro.

MATERIALS AND METHODS

Preparation and determination of LPO

LPO was prepared as described by Terao et a1 (1). Concentration
of LPO was determined by TBA method (2) and absorbance at 233nm
due to conjugated diene of LPO (3).

LPO decreasing activity of various microbes

Living cells or heat-treated (115"C,10min.) dry cells of
microbes were added to 2ml of 40mM phosphate buffer (pH6.8)
containing LPO (250-500 ~g/ml), and stood for 1 hour at room
temperature. After centrifugation, concentration of LPO in the
supernatant was measured.

Action of the cells of S. serev.is.iae on LPO

The heat-treated cells of S.cerev.is.iaewas washed with water 3
times and freeze-dried. The dried materials(500mg) was added to
10ml of LPO solution (500 ~g/ml). The mixture was centrifuged
and amounts of LPO in the supernatant and acetone extract from
1060
the precipitate were measured. The acetone extract, LPO and
reduced substances of LPO with NaBH4 or Na2S203 were developed
on silica gel TLC plates with the developing sol. ethyl ether/
acetic acid, 100:1 (vol/vol) and spots were detected by conc.
H2S04 with heating.
Proteolytic digestion of heat-treated cells of S.cerev1s18e
The heat-treated cells of S.cerev1s1ae were treated with
pepsin, trypsin and chymotrypsin at 37"C for 2.5h, respectively,
and the LPO decreasing activity of indigestible parts of cells
were measured.
RESULTS AND DISCUSSION

LPO decreasing activity of various microbes and Scatchard

analysis
The LPO decreasing activity was observed among all strains of
microbes tested (TABLE 1). There is little difference in LPO
decreasing activity between living cells and heat-treated cells.
It can be considered that the LPO decreasing activity of
microbial cells is physico-chemical action. According to
Scatchard plots it was indicated that the adsorption of LPO to
microbial cells was nonspecific binding.

Action of the cells of S. serev1s18e on LPO

Percentage of LPO remaining in the supernatant calculated by
using Abs. at 233nm was about the same as that calculated by
using TBA method (TABLE 2). TBA value and Abs. at 233nm of the
acetone extract from the precipitate were 16% and 82% of
initial value, respectively. These data suggest that the heat-
treated cells of S. cerev1s1ae have both hydrophobic binding
ability and chemical-reducing ability which can reduce the
hydroperoxide group but can not affect on conjugated diene
structures of LPO. Reduction of LPO by heat-treated cells of
S.cerev1s1ae was confirmed by TLC analysis (TABLE 3).

TABLE 1
LPO Decdreasing Activity of Various Microbes

Microbes LPO Qecreasing Activity Scatchard

( Jlg/mg-dry cells) analysis

E. faeca.l1s AD1001 28
E. faes1um AD1060 30 nonspecific
L. ac1dopb1.lus AD0006 36
L. case1 AD0013 32
L. reuter1 AD0002 27
L. p.lantarum AD0010 23
B. b1f1dum AD0060 29
B. breve AD0058 31
B. frag1.l1s ATCC2528 19
S. cerev1s1ae 167-12 43 nonspecific

*Not done
 1061
Influence of proteolytic digestion and presence of bile on LPO
decreasing activity
The LPO decreasing activity of the S.cerev1s1ae cells was not
so affected by treatment of enzymatic digestion or by the
presence of 0.1% bile.
These findings suggest that microbial cells are able to
lower the level of LPO in foods and intestinal systems.

TABLE 2
Action of the Cells of S. cerev1s1ae on LPO

% of LPO
Method
Extracts from cells Supernatant Reduced

Abs. at 233nm 82 18 o
TBA Method 16 17 67

TABLE 3
TLC Analysis

Rf Value of Main spot

LPO 0.82
Acetone extracts* 0.78
Reduced of LPO with Na2S203 0.78
Reduced of LPO with NaBH4 0.78

* Reduced substances of LPO with heat-treated cells of

S. cerev1s1ae .

REFFERENCES

1.Terao,J.and Matsushita,S.,The isomeric compositions of

hydroperoxides produced by oxidation of arachidonic acid with
singlet oxigen. Agric.Biol.Chem.,1981,45(3),587-93.

2.0hkawa,H.,Ohishi,N.and Yagi K.,Reaction of linoleic acid

hydroperoxide with thiobarbituric acid. J.Lipid Res.,1978,19,
1053-7.

3.Terao,J.and Matsushita,S.,Products formed by Photosensitized

oxidation of unsaturated fatty acid esters. J.Am.Oil Chem.Soc.
1977,54.234-9.
EFFECf OF WATER AND ALCOHOL ON TIlE FORMATION OF INCLUSION COMPLEX BETWEEN
CYCLODEXTRINS AND d-LIMONENE BY TWIN SCREW KNEADER

HIDEFUMI YOSHII', TAKESHI FURUTA2, TAKASHI KOBAYASHI 3, TOSHIMI NISHITARUMIZU 3,

HIROSHI HIRAN04 AND AKIRA YASUNISHI 2
'Department of Biochemical Engineering, Toyama National College of Technology, Toyama 939, Japan.
2Department of Biotechnology, Tottori University, Tottori 680, Japan.
3Kurimoto Co. Ltd., Osaka 559, Japan
4Food Engineering Laboratory, Toyama Food Research Institute, Toyama 939, Japan

ABSTRACf

To enhance the storage stability of essential oils such as d-limonene, the molecular inclusion
complex powder of d-limonene in a.- or ~-cyclodextrin was made by using a twin screw kneader. The
influence of water and alcohol content on the formation of the inclusion complex was studied in comparison
with the inclusion complex by the micro-aqueous method. There were many differences in the quantity of
the inclusion complex between the two methods, particularly at a low water content. For a.-cyclodextrin, the
addition of alcohol inhibited the formation of the inclusion complexes.

INTRODUCTION

Powdery encapsulation of a hydrophobic liquid flavor is quite important to improve its storage stability, as well
as to expand its application field. A popular method for the encapsulation is preparation in solution of
cyclodextrin. In this method, the specimen is dissolved into an aqueous solution of cyclodextrin to form the
inclusion complex in crystalline form [I, 2]. The crystalline complex is separated and dried by an adequate
method. However, with so called kneading method, the specimen is mixed directly in a mixer to be
transformed into the inclusion complex during kneading. This method is superior to the former, because no
separation process is required and less energy will be needed in the dehydration process. In this study, the
molecular inclusion of d-limonene (as a model of orange oil) in ~-cyclodextrin was accomplished by using a
twin screw kneader at a low water content. We have reported that the hydrophilic linear alcohol had increased
the formation of the inclusion complex of d-limonene with ~-cyclodextrin [3]. We examined the effect of water
and alcohol content on the formation of inclusion complex between d-limonene and cyclodextrins.

MATERIALS AND METIIODS

 1063
Materials
a- and f3-cyclodextrin (a-CD and f3-CD) was from Ensuiko Sugar Chemical Co. Chloroform stabilized with

amylene was from Kanto Chemical Co., Inc. Ethanol of special grade was from Nakalai Tesque.

Prepanttion of Inclusion Complex Powder by '!Win Screw Kneader

Twenty grams of the mixed powder of f3-CD was moistened by distilled water, followed by the addition of
d-limonene. After being mixed gently in a glass beaker, it was supplied into a twin screw kneader (KRC-S 1,
Krimoto Ltd.). The kneaded wet samples were taken at 30 and 60 min after kneading, followed by being dried
in vacuo at 70 0 C for 15 hours. The content of d-limonene in the powder was analyzed by a gas
chromatograph, using a solvent extraction procedure reported previously [3]. The inclusion fraction of d-
limonene was defined as a molar ratio of d-limonene in the complex powder to CD.

RESULTS AND DISCUSSION

/l-CDld-Limonene System
Figure 1 shows the molar ratio of the inclusion complex between f3-CD and d-limonene (inclusion fraction)
formed by kneading for 60 min under various initial moisture contents. The solid line represents the
experimental results of the inclusion fraction by the so-called micro-aqueous method reported previously.
There are marked differences in the inclusion fraction between them, particularly below the water content of 10.

'0' 1.0
.l!
:p
000
II 0.8
'0
E 0.6
......
Q
(.)
0.4 - Micro-Aqueous
~c
CD
c
0.2 o Kneading FIGURE 1 Formation of
0
00 Inclusion Complex of f3-CD by
E 5 10 15 20 25 Kneading (Comparison with
::::i
Waterfl3-CD (molar ratio) Micro-Aqueous Method)

Figure 2 shows a typical example of the influence of addition of linear alcohol in the case of methanol
and pentanol. The amount of each alcohol added is I or 5 molar times as much as that of f3-CD. For the
addition of methanol, the inclusion fraction does not depend on the moisture content, and shows a much higher
value than that by the micro-aqueous method shown by the solid line. However, at a higher methanol content,
the difference between the kneading and the micro-aqueous method becomes less and tends to coincide with
each other. For pentanol, on the other hand, the inclusion fraction depends markedly on the amount of pentanol
added at a water content of less than 10, though the inclusion fraction shows no dependencies on the quantity of
pentanol added in the case of the micro-aqueous method. Moreover, in the case of pentanol/f3-CD = 5, the
inclusion fraction by the two methods are nearly equal. This means that with the increase in the alkyl chain
1064
'0 1.0 r---------,,.--------..------..., .------...,
~ •o • • • MeOH(1)

•
.
0.8 o
---.
~
o
o
"
MeOH(S)
"0
E 0.6
...... • PntOH(1)
c .-~
o
t 0.4
-
/' Micro-
MeOHIJ3-
ueous [J PntOH(S)

ic 0.2 , ~HlJ3-CD=S
[J
FIGURE 2 Effect of Alcohol
o Content on the Fonnation of
E
:J Inclusion Complex by
10 15 Kneading
Waterlf3-CD (molar ratio)

length of alcohol added the inclusion becomes less formed particularly at lower water content. These results
suggest that the shear stress created by kneading at these conditions may promote the formation of the inclusion
complex. Figure 3 shows the relationship between the kneading torque (shear stress) and the inclusion fraction
of d-limonene. The inclusion fraction increases sharply with the increase in the kneading torque.

'0 1.0 r----..-----,r-----,..------,..-----,;-----,

.~
tii 0.8
"0 o
FIGURE 3 Correlation of .§. 0.6
Inclusion Fraction with c
Kneading Torque ~0.4
lJ
ic 0.2
o
E
:J 1 234 5 6
Torque (Kg m)
a-CD/d-Limonene System
The effect of ethanol on the formation of the inclusion complex was examined at a constant water content of 15.
When the ethanol content was larger than 2 times the moles of a.-CD, the inclusion was inhibited by ethanol
and no inclusion complex could be formed. This was similar to the result observed in the micro-aqueous
method.

Acknowledgement
This work was supported by the IIjima Memorial Foundation for the Production of Food Science and
Technology, and a Grant-in Aid for Scientific Research from the Ministry of Education, Science and Culture.
We also appreciate Ensuiko Sugar Refining Co. for the gift of cyclodextrins.

REFERENCES

1. Szejtli, J. Szente, L. and Banky-Elod, E., Molecular Encapsulation of Volatile, Easy Oxidizable Labile
Flavour Substances by Cyclodextrin, A eta Chim iea Scientiarurn H ungarieae, 1979, 101, pp.27 -46
2. Reineccius, G.A. and Rish, S.J., Encapsulation of Artificial Flavors by J3-Cyclodextrin, Perfumer and
Flavorist, 1986, 11, pp.3-6
3. Furuta, T., Yoshii, H., Miyamoto, A. Yasunishi, A. and Hirano, H., Effect of water and alcohol on the
formation of inclusion complexes of d-limonene and cyclodextrin, Suprarnoleeular Chern istry, 1993, 1,
pp.32l-325
 ENTRAPMENT OF LIQUID LIPIDS INTO POWDERY MATRIXES
OF SACCHARIDES AND PROTEINS

RYUICHI MATSUNO, JUN IMAGI, and SHUn ADACHI

Department of Food Science and Technology, Faculty of Agriculture,

Kyoto University, Sakyo-ku, Kyoto 606-01, Japan

SUMMARY

The emulsifying activity (1), the high stabilizing activity of emulsion (2) and the formation
of a fine dense skin layer during drying (3) were the properties of agents that effectively
entrap liquid lipids. Gum arabic and gelatin having the three properties were effective.
Addition of an agent having a property to a base agent lacking the property improved the
entrapment.
Oxidation of entrapped liquid lipid was retarded. However, extent of retardation
depended on kinds of lipids and of entrapping agents. Oxidation processes of some
combinations of lipids and entrapping agents were expressed by a kinetic model including
oxygen diffusion through dehydrated entrapping agents, but most were not. Ethyl
eicosapentaenoate was also stabilized by the entrapment.

INTRODUCTION

The most popular method for encapsulation is emulsification of lipids in a solution of

entrapping agents followed by spray drying. By this encapsulation, the lipids might gain
various advantages, that is, the autoxidation of lipids is retarded, the enhanced stability
allows full utilization of the lipids and so on.!) To determine the best conditions under
which to encapsulate a lipid for a particular purpose, a large number of entrapping agents and
combinations needs to be tested under an enormous range of conditions. A convenient
method for assessing encapSUlation has been introduced, in which a single lipid emulsion of
about 5 ,..,.e was dried and a resultant dried powder was subjected to the analyses for the
amount of lipids emerged at the surface and the oxidation process of lipids.

METHODS

Drying method for a single droplet of lipid emulsion

A 5 ,..,.e droplet of emulsion containing 15% lipid and 15% entrapping agent(s) was
suspended on a glass globule 1 mm in diameter fixed on a chromel-alumel thermocouple
1066
junction (100 ~lm diameter) or on a glass filament, and dehydrated by a single droplet drying
apparatus described elsewhere.2)

Measurement of lipid appeared at the surface of dehydrated droplet

A droplet was usually dehydrated for 15 minutes. Encapsulation efficiency was then tested
by soaking in 0.5 mR n-hexane for 3 s, and the amount of extracted lipid was determined
by gas chromatography. It was assumed that the lipid extracted during the first 3 s was lipid
not covered by the entrapping agent.

Method for following the oxidation process of lipids encapsulated

Methyl linoleate (ML), linoleic acid (LA) and ethyl eicosapentaenoate (BE) were used as
model lipids susceptible to oxidation. Methyl oleate (MO) was selected as the internal
standard. Equal amounts of the model lipids and MO were emulsified in a solution of
entrapping material(s), then a 5 !-lR droplet of each emulsion was suspended at the lower end
of a glass filament and dehydrated. The dehydrated droplets on the glass filaments were
placed upright in a desiccator with controlled humidity and temperature. Periodically a dried
droplet was removed and analysed to determine the ratio of the model lipid to MO.

RESULTS AND DISCUSSION

Screening of entrapping agents

Encapsulating agents suitable for use should have the following properties: 1) high
emulsifying activity, 2) high stabilizing activity of emulsion, and 3) a tendency to form a fine
dense skin layer during drying. The third property might be assessed by the rate of
isothermal drying. Maltodextrin, puUulan, gum arabic and gelatin have the third property.
Various entrapping agents have been used to encapsulate ML, and then the encapsulation was
evaluated. Table 1 summarizes the measurement of the exposed lipid. We define 'optimal'
encapsulation as corresponding to a surface lipid level of less than 1 %. As shown in Table
1, excellent encapsulation was achieved using gum arabic or gelatin alone, both of which
were excellent emulsifiers and had the third property. Maltodextrin and pullulan, which
might have the third property, but had no emulsifying activity, were not suitable when used
alone. Egg albumin and sodium caseinate, which had high emulsifying activity but not the
third property, were also not effective. Glucose, maltose and mannitol lacked all three
properties and were not useful at all.
To improve emulsifying activity, a surfactant of low or high molecular weight was
added. In general, the addition of a high molecular weight surfactant had positive effects;
sodium caseinate gave the best results. The amount of surface lipid with maltodextrin and
lecithin was 25.9%. However, when xanthan gum, which is highly viscous in solution, was
added to this system, the amount of surface lipid was nearly zero. Thus, this combination
(15% ML, 14% maltodextrin, 0.5% lecithin and 0.5% xanthan gum) was the optimal
combination for encapsulation.

Retardation of lipid oxidation by encapsulation

ML encapsulated with a-cyclodextrin (a-CD), maltodextrin and pullulan was extremely
resistant to autoxidation. ML encapsulated with sodium caseinate was rapidly oxidized after
a week. Although gum arabic and gelatin were good encapsulating agents, their abilities to
retard oxidation were not good. Albumin was not a good entrapping agent with respect to
the retardation of ML oxidation.
 1067
a-CD was a good material for retarding lA oxidation. lA encapsulated in maltodextrin
and pullulan matrixes was rapidly oxidized, unlike ML. The reason for this remains unclear.
The oxidation of lA encapsulated with albumin was relatively rapid, as for ML, whereas lA
encapsulated with sodium caseinate oxidized more slowly than ML. Gum arabic and gelatin
exhibited a similar retardation effect on the oxidation of both ML and lAo
EE is very susceptible to oxidation due to its high degree of unsaturation. When EE was
encapsulated in a.-CD, which greatly retarded the oxidation of both ML and lA, 65% was
lost by oxidation within two weeks. However, the remaining 35% was unoxidized thereafter.
Pullulan significantly retarded the oxidation of EE. Maltodextrin and gum arabic did not
prevent EE oxidation.
To elucidate theoretically the reason why the oxidation is retarded, the oxgen diffusion
equation and lipid oxidation equation were solved simultaneously, and calculated and
experimental average concentrations of unoxidized lipid were compared. Some experimental
results correlated well with the calculations, but most of the experimental results indicated
that lipids were more stable than predicted by the simulation. Therefore, other mechanisms
must be investigated.

REFERENCES
1) Matsuno, R. and Adachi, S., Lipid encapsulation technology - techniques and
application to food. Trends in Food Sci. & Technol., 1993, 4(8), 256-261.
2) J. Imagi, Kako, N., Nakanishi, K., and Matsuno, R., Entrapment of liquid lipids in
matrixes of saccharides. J. Food Eng., 1990, 12, 207-222.

Table 1 Amount of Methyl Linoleate Exposed at the Surface of Dehydrated Samples of the
Emulsion Solution of Entrapping Agent with 0.5% Surfactant, 0.5% Stabilizer, or Neither.

Lipid exposed as percentage of total lipid

Surfactant or stabilizer (B)

Entrapping None LE Sugar Tween CMC Gum arabic Casein XG a-CD

agents (A) (+XG*) ester 80 0.5% 5.0%

Glucose 22.9 3.3 0.3 39.1 5.1 3.1 0.5 0.7 19.0 47.3
Mannitol 34.7 6.9 0.4 38.2 24.4 26.1 1.9 0.5 28.6 40.9
Maltose 32.2 3.9 1.0 59.0 4.9 5.2 0.8 2.2 21.4 36.4
Maltodextrin 1.8 25.9 8.0 30.5 1.2 2.0 0.7 1.1 4.3 16.2
('" 0)
a-CD 15.9 13.0 13.5 12.0 7.4 16.3 0.2 13.6 11.1
Pullulan 2.5 0.8 0.4 48.9 0.2 1.1 0.9 0.4 16.7 28.5
Gum arabic 0.3 0.9 42.0 52.3 0.4 1.0 0.6 1.1
Egg albumin 5.6 8.7 9.4 53.3 3.9 3.0 3.5 3.5 6.7
Casein 3.7 4.8 7.3 6.2 0.8 2.1 1.5 1.5
Gelatin 0.3 2.9 22.6 35.1 0.5 0.9 0.2 0.6 0.8

LE, lecithin; CMC, carboxymethylcellulose; casein, sodium caseinate; XG, xanthan gum; a-CD, 0.-
cyclodextrin; non, neither surfactant or stabilizer. The concentration of methyl linoleate in the
original emulsion solution was 15%. Total concentration of entrapping agent was 15%. The
concentrations of A and B were 14.5% and 0.5%, respectively, unless otherwise noted.
* The concentration of XG was 0.5%.
 AUTHOR INDEX

Adachi, S. 480, 582, Arai, S. 48, 176 Belton, P.S. 15

1065 Araki, H. 507 Bergmann, Y. 972
Adeyemi, S.AO. 1032 Ariga, T. 1056 Berk, Z. 433
Adler-Nissen, J. 295, Ariuchi, N. 203 Bernardo-Gil, G. 212,
555 Arreola, A 855 829
Agemura, c.K. 218 Asaka, M. 861 Bhamidimarri, S.M.R
Aguilera, H.G. 710 Ateba, P. 301 618, 1008, 1011
Aguilera, J.M. 1026 Autio, K. 72, 161 Bhandari, B. 433
Ahvenainen, R 775, Axelson-Larsson, L. Bhaskar, AR 840
885, 888 775 Bifani, V. 152
Akashi, T. 689 Azak, M. 894 Biladt, A 852
Akinaga, T. 1029 Azuma, N. 594 Bimbenet, J.1. 334, 981
Akiyama, Y. 873 Biswal, RN. 394
Allache, M. 975 Borcoski, D. 152
Alonso, AA 721, 724 Bae, S.K. 373, 382 Bourlier, C. 439
Altarriba, A 415 Balaban, M.D. 307, 855, Bressa, F. 474
Altomare, RE. 298 972 Bringiotti, R 134
Amachi, T. 766 Ban, N. 265 Britt, 1.1. 274, 716
Amarowicz, RA 621, Banga, J.R. 721, 724, Brosio, E. 227
627 730 Brown, RB. 987
Ames, J.M. 876 Barana, AC. 549 Buhri, AB. 283
Anantheswaran, RC. Barbanti, D. 474 Biiltermann, R 698
707, 781 Barros, G.V. 549
Ando, K. 1056 Basheer, S. 588
Andrieu, J. 233 Bates, L. 876 Cakaloz, T. 403
Andriolli, D.M. 897 Bauer, W. 448 Cal-Vidal, J. 194
Anese, M. 286 Beelman, RB. 781 Cano, M.P. 501
Anifantakis, E. 358 Behringer, R. 695 Capalbo, D.M.F. 606,
Anjos, C.AR 796 Beido da Costa, M.L. 1020
Annachhatre, AP. 1008 206, 212, 829, 882 Cardelli, C.F. 710
Aoki, T. 609 Bell, L.N. 489 Cardoso, AL. 829
Aono, K. 719 Bellon, V. 942 Cavenagui, M.E. 549
Aoyama, Y. 861 Belmar-Beiny, M.T. 24, Chang, Y.-H. 459
Arai, K. 849 805, 808 Chao, R.R 456
 1069
Chen, c.-c. 772 1002 Galan-Wong, J.L. 543,
Chen, C. -CO 772 Diosady, L.L. 671 546
Chen, C.S. 855 Doi, T. 93 Gallardo, J.M. 721
Chen, S.-Y. 772 Dumoulin, E. 433, 439 Gamse, T. 903
Chen, W.J. 665 Duplan, J.e. 233 Giarola, T. 194
Cheng, C.-J. 1038 Duquenoy, A 981 Gigante, M.L. 1044
Cheryan, M. 677 Durigan, J.F. 897 Giroux, F. 421, 981
Chiang, W. 870 Gladden, L.F. 471, 498
Chinachoti, P. 129 Gohtani, S. 203
Cho, H.-Y. 453 Earle, R.L. 1011 Gontard, N. 790
Choi, N.B. 963 Eerikiiinen, T. 36 Goto, M. 835
Chou, C.-c. 540 Endo, M. 271 Gould, lV. 280
Choupina, A 882 Endo, Y. 849 Grauer, B. 1011
Chun, J.K. 930, 963 Enomoto, A 733 Grenier, P. 999
Chuy, L.E. 191 Guidolin, F.R. 549
Clark, J.P. 262 Guilbert, S. 397, 421,
Clement, L. 439 Faleiros, R.R.S. 897 790
Collignan, A 400 Faria, J.AF. 796 Gumbe, L.O. 123
Cooley, H.J. 111 Farkas, B.E. 63 Guzman, M. 576
Comillon, P. 233 Flesland, O. 385
Corradini, C. 170 Franz, P.R. 280
Correa, c.P. 546 Fregert, J. 748 Haas, J. 698
Correia, L.R. 513 Friis, A 295 Hiigg, M. 885, 888
Corrieu, G. 558, 564 Froschl, F. 852, 903 Hagura, Y. 87, 167, 250,
Coumans, W.J. 430 Fryer, P.J. 24, 471, 498, 253, 507, 510
Courtois, F. 334, 975, 754, 760, 805, 808 Hakoda, M. 477, 689,
981 Fujii, M. 531 733
Cuq, J.-L. 790 Fujii, T. 105, 114, 117 Hamada, T. 615
Fujimoto, K. 849 Hamano, M. 179, 832
Fujio, Y. 57, 522, 867 Hanafusa, N. 224
D'Ubaldo, A 227 Fujita, H. 1053 Hansen, c.L. 1005
Dalla Rosa, M. 474 Fujita, S. 173 Hanzawa, T. 313, 713
Datta, AK. 325 Fujiwara, K. 588 Hara, K. 483
Davidson, v.J. 987 Fujiwara, T. 751 Harada, F. 624
De Baerdemaeker, J. Fukumoto, K. 864 Harada, T. 600
200, 462, 465, 701, Fukuoka,~. 236, 239 Harding, S.E. 206
727 Fukushima, H. 733 Harkonen, H. 72
De Cordt, S. 692 Fukushima, Y. 525, 615 Harris, J.L. 668
De La Plaza, J.L. 945 Fukuta, Y. 388 Hasegawa, M. 624
De Maria, M.O. 1002 Furukawa, T. 656 Hashimoto, A 316, 361,
de Massaguer, P. 704 Furusho, S. 636 504, 742
De Melo, S. 194 Furuta, M. 739 Hashimoto, H. 641, 832
De Pontes, AE.R. 549 Furuta, T. 418, 436, Hassager, O. 295
Debenedetti, P.G. 30 1062 Hata, K. 736
Dec1oux, M. 981 Fuse, O. 525 Hatanaka, K. 799
Dejmek, P. 90, 644 Fushijima,~. 662 Hattori, M. 820
Del Bianchi, V.L. 549, Hayakawa, I. 57, 522.
1070
867 Imagi, J. 480, 1065 Kato, Kaoru 656
Hayakawa, Kan-ichi Imai, M. 492, 633, 993 Kato, Koro 948
418, 704 Imamura, H. 832 Kato, S. 630
Hayakawa, Kiro 650, Imou, K 78, 140 Kauten, RJ. 218
951 Imura, N. 367 Kawagoe, Y. 328, 424
Hayakawa, S. 832 Inoue, M. 492 Kawai, G. 525
Hayashi, H. 75 Inoue, N. 843 Kawai, H. 173
Hayashi, N. 57 Inoue, T. 477 Kawase, Y. 612
Hayashi, S. 954 Inukai, T. 105 Kawasome, S. 203
Hayashi, T. 739, 906 Ishida, N. 241 Kaymak, F. 403
Hayes, RJ. 280 Ishiguro, T. 480 Kelley, R 262
Hayward, G.L. 987 Ishiguro, Y. 650 Kerkhof, P.J.AM. 430
Heinio, R-L. 885, 888 Ishihara, K 69, 209, Kessler, H.G. 695
Heldman, D.R 456 1059 Kessler, U. 448
Henderson, T. 307 Ishihara, M. 864 Ketelaars, AA.J. 430
Hendrickx, M. 692 Ishii, K 176 Kikuchi, M. 271
Herlihy, N. 927 Ishikawa, K 388 Kikuchi, S. 630
Hertlein, J. 787 Ishikawa, S. 388 Kikuchi, T. 636
Hikosaka, T. 247 Ishizaki, A 552, 561 Kim, D.U. 310
Hill, S.E. 206 Ishizuka, H. 674 Kim, D.Y. 909
Hirano, H. 1062 Isobe, S. 984 Kim, J.W. 909
Hirose, T. 835 Isono, Y. 686 Kim, KM. 963
Hisanobu, Y. 483 Itani, Y. 60 Kimura, K. 736
Ho, e.T. 516 Ito, M. 1059 Kimura, Y. 582
Hollewand, M.P. 471 Ito, T. 1023 King, A-E. 918
Honda, T. 504 Itoh, H. 811 Kinugasa, H. 864
Hong, T.L. 1035 Itoh, K 924, 933 Kiranoudis, C.T. 340
Hori, T. 924, 933 Iwaniuk, B. 409 Kitagawa, K. 624
Horibe, K. 352 Kitakura, Y. 832
Horikawa, N. 951 Kitatsuji, E. 597
Hoshino, M. 960, 1023 Jadan, A 364 Kluza, F. 376
Hoshino, T. 343, 820, Jelle, B. 915 Knez, Z. 826
823 Jiang, J. 90 Ko, W.e. 665
Hosokawa, T. 259 Jouppila, K 188 Kobamoto, N. 519
Hu, W. 325 Kobayashi, H. 624
Huang, V.T. 155 Kobayashi, T. 1062
Huang, x.J. 713 Kobori, S. 591
Hung, Y. 126 Kameoka, T. 352 Kohda, Y. 1029
Hurme, E. 775, 885, 888 Kaneko, T. 1047 Kohno, S. 579
Hwang, S.H. 1005 Kanekuni, N. 659 Kohyama, K 108, 120
Hyman, S.M. 262 Kanesiro, M.AB. 897 Koike, S. 674, 680
Kanno, T. 867 Kolbe, E. 1041
Kano, H. 241 Komatsu, Y. 483
Kano, K 579 Kometani, T. 597
Ichiba, J. 424 Kantelinen, A 72 Kondo, M. 1023
Igami, H. 650 Karathanos, V.T. 197 Kong, B. 891
Igarashi, H. 742 Karbstein, H. 9 Kono, H.O. 60, 247
Ikeda, K. 736 Kasuga, Y. 451 Koseoglu, S.S. 683
 1071
Kostaropoulos, AE. 197 Lizarraga, V. 501 McCarthy, M.J. 215
Kosugi, Y. 594 L6pez Leiva, M.H. 576 McCoy, B.J. 817
Kramer, K 427 L6pez-Rodriguez, V. McKenna, B. 927
Kubota, K 87, 250, 253, 945 Medrano-Roldan, H.
507, 510 Lotz, M. 221 543, 546
Kumagai, Hitomi 176 Lu, L.T. 507 Meerdink, G. 445
Kumagai, Hitoshi 105, Merson, R.L. 292, 757
114, 117, 176, 1050 Mihori, T. 132, 239
Kumakura, K 656 Ma, X. 412 Miki, H. 96
Kunisaki, S. 766 MacDonald, G.A. 1041 Miller, KS. 63
Kurasawa, H. 66 Macdougall, D.B. 876 Min, S.G. 370
Kurihara, I. 636 Maesmans, G. 692 Min, S.H. 603
Kurokawa, M. 507 Magalhaes, M.A 495 Mineshita, T. 54
Kuwata, T. 636 Mallikarjunan, P. 331 Mitchell, J.R. 206
Kyereme, M. 707 Mammoto, S. 978 Mitsuhashi, T. 939
Mannapperuma, J.D. Mitsumoto, M. 939
784 Mittal, G.S. 301, 331,
Labuza, T.P. 191, 489 Marchesseau, S. 790 969
Lagarde, S. 439 Marinos-Kouris, D. 340 Miura, S. 117
Lai, T.H. 665 Maroulis, Z.B. 340 Miyamoto, T. 271
Lameloise, M.L. 981 Marr, R. 852, 903 Miyata, T. 361
Lang, W. 406 Marsaioli, A Jr. 796 Miyawaki, O. 149, 373,
Lanier, T.G. 1041 Marshall, M.R. 855 382, 600, 951
Lasseran, J.C. 334 Martens, T. 990 Moe, T. 555
Latrille, E. 558 Martin, A.M. 537, 1017 Mogi, K 588
Laureano, O. 102 Maruyama, K 993 Moldao-Martins, M.
Laurent, M. 233 Masduki, A. 567 212, 829
Le Maguer, M. 349, 394 Masi, P. 134, 304 Mollah, M.R. 280
Lebert, A 334, 981 Massaguer, P.R. 495, Montoya, M. 945
Lee, C.H. 909 710 Moraes, 1.0. 534, 549,
Lee, C.M. 99 Masuda, Y. 733 606, 1002, 1020
Lee, S.-L. 540 Masumoto, Y. 87, 510 Moraes, R.O. 606, 1020
Lee, S.W. 468 Matagi, K. 343 Mori, A 609
Lee, Y.J. 963 Matias, E.C. 102 Mori, M. 751
Lenart, A 409 Matsuda, O. 367 Morishima, H. 78, 140,
Lerici, C.R. 170 Matsuda, R. 483 328
Lewicki, P.P. 137, 185 Matsumoto, S. 271 Morishita, T. 486
Li, C. 412 Matsumoto, Sachio 164 Morita, A. 820
Li, S.-J. 1038 Matsumoto, W. 591 Morita, K 126, 957,
Li, T.-Q. 215 Matsuno, R. 391, 480, 1014
Liebenspacher, F. 230 582, 597, 1065 Morita, N. 268
Lin, S.-S. 772 Mattila, M. 885, 888 Morrissey, M.T. 1041
Linders, L.J .M. 445 Mattila-Sandholm, T. Moschitz, E. 903
Linko, P. 36, 573 775 Moscicki, L. 879
Linko, S. 36 Matyka, S. 879 Moscoso, M. 364
Linko, Y.-Y. 573 McCarthy, KL. 215, Moshammer, B. 852
Liu, S. 754 218 Motai, H. 615
1072
Motegi, T. 525 Noritomi, H. 630 Paulson, AT. 716
Moyano, P. 152 Notsu, T. 582 Pauwels, O. 942
Mukae, T. 733 Nt1nez-Lemos, A 415 Pazir, F. 894
Mulvaney, SJ. 814 Peitersen, N. 915
MylHirinen, P. 161 Pekyardimci, S. 855
Myllymaki, O. 161 O'Donnell, C. 927 Perez-Galindo, A. 543,
Odake, S. 355 546
Odawara, T. 343 Perez-Martin, RI. 721,
Nabetani, H. 650, 680, Ogawa, Hidejiro 241 724, 730
686 Ogawa, Hiroyuki 1050 Perret, B. 564
Naczk, M. 627 Ogawa, Y. 689 Persson, KM. 647
Nagahama, K 630 Ogazi, P.O. 1032 Pessoa, AL. 549
Nagai, K 477, 733 Ogiyama, T. 736 Pfeifer, J. 695
Nagai, T. 811 Ogundipe, H.O. 1032 Piazza, L. 134, 304
Nakai, T. 504 Ohinata, H. 843 Picque, D. 558
Nakajima, M. 588, 650, Ohlsson, T. 18, 319, 322 Pigache, S. 570
680, 686 Ohmi, M. 993 Pinnavaia, G.G. 289
Nakamura, K 343, 477, Ohsaki, K 525, 656 Pittia, P. 170, 286, 289
689, 733, 820, 823, Ohta, H. 900 Pittroff, M. 256
846 Ohtake, H. 567 Pizzi rani, S. 474
Nakamura, T. 54 Ohtomo, H. 636 PomaraIiska-Lazuka, W.
Nakanishi, K 391, 653, Okado, T. 361 137, 185
811 Okamoto, K 253 Pongsawatmanit, R. 149
Nakanishi, R 861 Okamoto, R 820 Portier, K 200
Nakano, K 630 Okamoto, Y. 579 Poutanen, K 72, 161
Nakanuma, H. 93 Okazaki, T. 507 Powell, RL. 215
Naoe, K 633 Okemoto, H. 641 Preitschopf, W. 852
Narabe, H. 624 Oku, T. 1023 Prosetya, H.M. 325
Nezu, T. 591 Omura, H. 259 Prussia, S. 126
Ngadi, M.O. 513 Ooshima, K 531 Pyun, y'-R 453
Nicolai, B.M. 701, 727 Orlando, C. 286
Nicoli, M.C. 289 Osajima, Y. 793
Niranjan, K 337 Oshita, S. 316, 361 Qu, D. 516
Nishi, Y. 1014 Osorio, F. 152
Nishimoto, J. 96 Otobe, K 954
Nishimura, H. 1056 Otwell, W.S. 307 Rajagopalan, N. 677
Nishinari, K 108, 120 Oussalah, M. 999 Raman, L.P. 677
Nishio, K 486 Oyewusi, F.A 1032 Rao, M.A 111
Nishio, T. 1023 Ozawa, S. 939 Raoult-Wack, AL. 397,
Nishitarumi, T. 1062 400
Nishizawa, Y. 567 Razavi, S.K 668
Niviere, V. 999 Padly, F.B. 588 Rhee, C. 468
Noda, K 766 Pain, J.-P. 754 Ribeiro, C.AA 549
Nogaki, H. 659 Park, KH. 757, 909 Rios, G.M. 397
Nogata, Y. 900 Park, SJ. 310 Risman, P.O. 319
Noguchi, A 310, 984 Parkkonen, T. 72 Riva, M. 304
Nomura, K 624 Patil, RT. 84 Rizvi, S.S.H. 814, 840
 1073
Robles, e.M. 543, 546 Seki, H. 1023 Sokhansanj, S. 84, 406,
Rodriguez-Padilla, C. Seki, T. 1056 912
543, 546 Senoussi, A 433 Song, X. 385, 427
Roger, J.-M. 999 Sensidoni, A 286 Sonomoto, K. 579
Roig, S.M. 1044 Seo, Y. 78, 140, 328, Sousa, I.M.N. 102, 206,
Ronnegard, E. 90 424 882
Roos, Y. 188 Sereno, AM. 158, 182 SpieS, W.E.L. 370, 376,
Rosen, C. 644 Seto, H. 176 603
Rossi, M. 573 Severini, C. 289 Stapley, AG.F. 471
Roy, S. 781 Sevila, F. 999 Stark, R 716
Rubiolo de Reinick, A Seymour, J.D. 215 Steven Otwell, W. 972
379 Shahidi, F. 621, 627 Stoforos, N.G. 745
Shen, G.-Q. 1038 Str,mmen, I. 385, 427
Sherkat, F. 668 Sudo, H. 846
Sa, M.M. 158, 182 Shi, Z. 561 Suehisa, T. 639
Saeki, T. 766, 811 Shibata, M. 388 Suematsu, S. 483
Sagara, Y. 78, 140, 328, Shibauchi, Y. 799 Sugiyama, J. 954
424 Shih, M.-J. 870 Sugiyama, T. 846
Sahashi, Y. 674 Shiinoki, Y. 924, 933 Suortti, T. 72, 161
Saif, S.M. 81 Shilton, N.C. 337 Suprinyanto, 522
Saigo, H. 483 Shima, M. 582 Suter, D.A 81
Sakai, N. 313, 418, 713 Shimada, S. 236 Suzuki, A 173
Sakanishi, K. 54 Shimiya, Y. 277 Suzuki, H. 1047
Sakashita, S. 996 Shimizu, K. 561 Suzuki, Kanichi 167,
Sakiyama, T. 146, 811, Shimizu, M. 316, 492, 250, 253, 921
1050, 504, 633, 742 Suzuki, Kazuaki 674
Sakurai, H. 176 Shimoda, M. 793 Suzuki, Tatsuru 630
Salva, TJ.G. 534 Shimoyamada, M. 388 Suzuki, Tetsuya 739
Sanchez V., J.E. 1017 Shin, S.e. 930 Suzuki, Y. 823
Sano, Y. 451, 639, 838 Shindo, J. 96 Swatland, H.J. 936
Saravacos, G.D. 197, Shinno, A 864 Symns, R 262
340 Shinohara, K. 939 Synowiecki, J. 627
Sasaki, K. 277 Shirai, Y. 391
Sastry, S.K. 769 Shishikura, A 858
Sato, H. 1023 Shoeman, D. 489 Tabata, H. 659
Sato, Masaki 835 Shoji, I. 66 Taharazako, S. 957,
Sato, Masayuki 736 Shuto, I. 167 1014
Saurel, R 397 Silva, F. 882 Takagi, M. 268
Sawada, H. 292 Silva, M. 364 Takahashi, K. 763
Sawai, J. 504, 742 Simpson, R. 1041 Takahashi, M. 69
Schreier, PJ.R 24, 805, Singh, RP. 63, 283, Takama, K. 739
808 730, 784 Takaya, T. 108
Schrevens, E. 200 Sjoberg, A-M. 885, 888 Takayama, R 69
Schroen, e.G.P.H. 585 Skerget, M. 826 Takeo, T. 864
Schubert, H. 9, 244, SkjOldebrand, C. 346, Takeuchi, T. 259
256,698 966 Takizawa, H. 739
Schuchmann, H. 244 Skyttii, E. 885, 888 Tambunan, AH. 328
1074
Tamez-Guerra, R 543, Vaara, T. 573 Yamada, S. 525
546 Vall at, C. 942 Yamada, Y. 650, 951
Tanaka, K. 552 Van Buren, J.P. 111 Yamamoto, A 54
Tanaka, S. 957 Van Impe, J.F. 701 Yamamoto, K. 1050
Tanaka, T. 799 van der Linden, J. 42 Yamamoto, M. 1047
Tanaka, Takaaki 653 Van der Padt, A 585 Yamamoto, S. 451, 639,
Tanaka, Y. 630 Vandewalle, X.K. 465 838
Tang, C.Q. 132 van't Riet, K. 445, 585 Yamamoto, T. 209
Tang, J. 274 Varsanyi, 1. 778 Yamano, Y. 203
Taniguchi, M. 531 Vavra, C. 683 Yamaoka, A 588
Tatsubayashi, K. 939 Velazque, C.G. 543, 546 Yan, Y. 891
Taylor, S.M. 498 Verlinden, B.E. 462, 465 Yanagihara, N. 271
Teixeira, A 307 Verreydt, J. 200 Yanniotis, S. 358
Teramoto, S. 1056 Vervaeke, F. 200 Yano, Takuo 567
Thai, C. 126 Verzegnassi, B. 227 Yano, Toshimasa 1, 105,
Thorvaldsson, K. 346 Vidal, D. 415 114, 117, 146, 1050
Tobback, P. 692 Viikari, L. 72 Yao, Z. 349
Todoriki, S. 739, 906 Yasuda, M. 519
Tollner, W. 126 Yasunishi, A 436, 1062
Tomizuka, N. 594 Wanasundara, P.KJ.P.D. Yeralan, S. 972
Tonakawa, M. 561 621 Yiao, A-R. 870
Torii, T. 355 Wang, C. 412, 891 Yokoo, M. 680
Torikata, Y. 265 Wang, H.-H. 528 Y onei, Y. 843
Tosello, RM. 495 Wang, 1. 704 Y 00, I.J. 930
Toyoda, I. 805 Wang, S.S. 516 Yoon, S.H. 909
Toyoda, K. 143 Wang, Y. 78, 140 Yoshii, H. 436, 597,
Tozuka, H. 751 Watanabe, A 659 1062
Tdigardh, C. 748 Watanabe, H. 132, 236, Yoshikawa, M. 1053
Tragardh, Ch. 644 239 Yosizawa, N. 823
Tragardh, G. 647 Watanabe, K. 388 Yoza, K. 900
Tripetchkul, S. 561 Watanabe, R 93, 271, Yuo, S.-H. 772
Trystram, G. 421, 570, 1047
975, 981 Watase, M. 108
Tsai, S.-J. 1035 Weisser, H. 221, 230, Zarmpoutis, J. 358
Tsao, J.-C. 528 787, 802 Zhang, G. 510
Tsou, S.C.S. 1035 Wen, W. 239 Zhang, L. 268, 760
Tsoubeli, M. 489 Weng, J.-H. 870 Zhang, M. 412, 442, 891
Tsujimura, M. 612 Wirtanen, G. 775 Zheng, X. 516
Tsukada, T. 418 Witrowa, D. 137 Zhu, Y.-H. 36
Tung, M.A 274, 716 Woo, D.H. 963 Zuber, M. 802
Turunen, M. 573 Wood, H.C. 912

Ueda, T. 552 Xu, N. 442

Uemura, K. 310, 984
Urushiyama, S. 993
 SUBJECT INDEX

acid forming fermenter 1005 arachidonic acid 1056

acoustic impulse 954 aroma 522
activated sludge 1008 components in bread 1047
activation aromatic
energy 772 herb 212
volume, inactivation 861 herbs 829
adsorbed layer 823 L-ascorbic acid 268
adsorption 823 aseptic
protein 811 fi1ling, beverages 802
agarose 108 processing, particulate foods 745
agglomeration, jet 244 aspartame 489
aggregative organism 567 precursor 686
agricultural pests 1020 ATR 942
AI control 984 automate 966
air lift fermentor 570 automatic
alcohol 564 control 975, 981
alfalfa 84 handling system 978
L-amino acid production 600 automation 969
a-amylase 534 autoxidation, kinetics 480
anaerobic fermentation 1005, 1014 avidin, egg white 639
angiotensin-converting enzyme, inhibitor
1053
annulus liquid flow 921 baby food, enriched 1032
ANOYA procedure 415 Bacillus
antiallergic function, rice-based foods subtilis 534
48 thuringiensis 543, 546
antimicrobial activity 766 bacterial
antioxidation 477 insecticide 606
antiseptics 763 spores 695, 698
antiviral function, rice-based foods 48 bacteriocin 915
apparent water diffusivity 403 baked food 277
apple 126, 152, 200 bakery
juice, aseptically packed 772 oven 298
shrinkage 137 product 197, 203
1076
baking 268, 975 carotenoid 537
banana 501 carotenoids 835
batch carrageenan 108
cooking process 987 carrot
crystallizer 391 cubes 754
fermentation 558 juice 603
retort 724 carrots 888
beer, pulse-treated 736 casein 185
beverage 662 castor oil hydrolysate 540
beverages 489 catalyst removal 683
aseptic filling 802 catechins 483
sterilization 736 cell membrane disruption 445
biogas 1014 cellobiose 573
bioinsecticide 543, 546 ceramic
biological reduction 603 carriers 609
biologically active peptides 579 filters 662
bioprocess engineering 36 membrane 659
bioreactor 993 ceramics powder slurry 742
biscuit, failure behavior 134 cereal flakes 870
bitterness 486 change, flavor 900
bread 140 cheese
dough 265 processing waste 1005
viscoelasticity 78 vegetable fat 1044
brittle fracture properties 253 chemical oxygen demand 1002
broccoli 784 chestnut 882
brown sugar chicken muscle 513
chemical components 1029 chromatography, supercritical fluid 838
physical property 1029 citric acid 858
bubble 963 citrus fruits 486
butter, viscoelasticity 75 classifier, turbo-type 247
cleaning 24
co-rotating disc scraped surface heat
cabbage 885 exchanger 295
caking 191 coagulation 924, 969
calcium codfish 239
ion 1050 coenzyme recycling 600
phosphate formation, inhibitors 1050 coffee
calorimeter 763 solution 424
canned wash water 1017
food 713 cohesiveness, cooked rice 66
tuna 721 cold storage 701
canola 671 collapse temperature 155
cap-stemming 280 color
capillary viscometer 54, 57, 63 classification 200
carbon dioxide gases 733 development 876
 1077
colour change 286 flavor 793
complex viscosity 111 dewatering 397
computer aided DHA, docosahexanoic acid 630
design 990 diacetyl 951
simulation 721 differential scanning calorimetry 158,
concentrated fruit juice 650 283
conductance method 742 diffusion 227
conductivity, fluid thermal 933 moisture 236, 239
cook-chill processing 465 diffusivity 197
cooking aroma 430
potato 462 moisture 430
rate 510 water 409
ratio 510 digestibility, protein 286
correlation length 114 digital signal processing 265
cream 271 dispersed system 146
creep test 78 disruption
cross flow stabilisation, droplet 9
agitator 993 microbial cells 733
filtration 653, 659 dissolved oxygen 546
membrane filtration 656 DNA 117
crumbliness, bread 129 Doehlert network 400
cryo-shattering 250 drum dryer 930
cryoprotectant 224, 1041 dryer, vacuum 903
crystallization 188, 194 drying
curd 969 automation 930
cycling phenomenon 271 characteristics, wheat 406
cyclodextrin 436, 1062 enzymes 451
polymer 641 L. diacetylactis 448
cyclomaltodextrin glucanotransferase micro-organisms 445
573 polyethylene terephthalate 796
cytochrome c, denatured 1023 dynamic
model 981
modelling 570
dairy processes 42 dewatering 421
DDC control 561
decaffeination 817
y-decalactone 540 edible oil refining 683
degradation effective moisture diffusivity 415
vegetable, cooking 462 egg yolk phosphatidylcholine 624
oil 63 electrical conductivity 769, 945
dehydrated sword bean 891 electrochemical method 951
denaturation 224 electro-ultrafiltration 689
density sorting 948 electron beams 739
desorption, protein 811 electronic balance 948
deterioration 778 electrophoresis, gel 117
1078
emulsification 9 process control 36
emulsion 164, 167, 170, 173 rate 564
enzyme reactor 585 fermented
end-over-end rotation 716 bevarage 555
energy tofu (tofuyo) 519
saving 996 film
spectra 993 bags 525
entrapment, liquid lipids 1065 edible wheat gluten 790
enzymatic laminated 793
reaction 689 finite element
synthesis 686 analysis, 2D 471
enzyme method 146, 727
membrane reactor 686 firmness 140
production 531, 534 bread 129
proteolytic 811 fish meat 259
xylanolytic 72 flavor
EPA, eicosapentaenoic acid 630 change 900
equivalent time 453 deterioration 793
esterification 492 retention 433, 436
ethanol flaxseed 621
fermentation 561 flexibility 966
oxidation 597 flow
production 609 laminar 748, 760
ethyl eicosapentaenoate 1065 power-law 760
expert simulation 999 property 54
extension, biaxial 274 rate, particles 751, 754
extraction 212 solid-liquid mixtures 754
concentration 627 turbulent, whey 808
selective 633 visualization technique 707
supercritical fluid 817, 826, 829, fluid
832, 835, 840, 843 motion 215
extruder, single screw 218 viscosity 924
extrusion 516, 814, 870, 873, 876, 879, fluidized bed 427
882 drying 448
dynamic control 36 fermentor 531
cooking 984 reactor 618
food
engineering 1
far-infrared drying 442 processing 295
fat quality 453
content 939 specified health use 48
transfer 301 storage 456
fatty acid 594 foods, frozen 909
fatty acids, polyunsaturated 674 forced convection 716
fermentation 963 forming capacity 176
 1079
fouling 24, 647 green tea 864
kinetic model 805 growth-inhibitory effects 742
surface 808
Fourier number 456
fractionation, ~-lactoglobulin 636 handling, automatic system 978
fracture stress 132 hay 912
freeze 903 hazel-nut 289
concentration 391 heat
drying 134 capacity 233
freezing 194 denaturation 873
fresh cheese, Brasilian 1044 processing 483
friability 134 pump dryer 427
fried food 283 transfer 298, 745
frozen coefficient 713
food 149, 221, 233 conductive 727
foodstuff 253 convective 757
fruits 945 treated dry cell 1059
spherical 948 treatment 912
frying 513 heating 277
deep-fat 301 microwave 707
FT-IR 942 Herschel-Bulkley model 57
fungus 1017 high
fuzzy control 987 pressure 861, 864, 867
homogenization 256
temperature-short time process
GAB equation 182 704
gamma rays 739 water activities 918
gas holding tube 751
anti-solvent crystallization 858 hot-wire
cell 203 anemometry 748
permeation 787 method 924, 933
pressure release 733 HPLC 624
gel humidity 170
food polymer 274 hybrid process 814
layer 477 hydration 164, 224
strength 206 hydrogel, microstructure 117
gelatination reaction 459 hydrogen peroxide, gasified 799
gelatinized wheat flour 468 hygrophysical change 418
gelation 498 hygroscopicity 179
glass transition 129, 155, 158, 188, 191 hygrostrain-stress 418
glutaminase, immobilized 615
glutathione 209
gluten 790 ice
grape 900 crystals agglomeration 391
grasp 978 fraction 149
1080
Image 241 oxidation 474
analysis 203, 972 thermal conductivity 513
data 960 kinetics 304
immobilized aspartame degradation 489
cell 618 autoxidation 480
cells 615 decomposition 209
impedance spectroscopy 143 esterification 492
impregnation 397 fouling 24
soaking process 421 heat gelation 498
in line removal 585 isomerization reaction 483
in-container thermal processing 18 retrogradation 468
in-flow thermal processing 18 thermal death 698
inactivate rate equation 451
inactivation
enzyme 855, 861 lactic acid 558
kinetics 448, 695, 855 bacteria 915
L. plantarum 445 fermentation 552, 555
indirect heating 698 Lactobacillus 701
individual sugar content 942 Lactococcus lactis 552
infant formula 191 lactose 188
insect 912 laminated tube 960
inspection system 960 large scale 570
interesterification 588, 591, 849, 852 catering 990
intergrated bioreactor system 579 preparation 624
internal defect, nondestructive detection lees 656
957 length scales 15
interparticle force 247 Lentinus edodes 525
ion exchange lethality 710, 757
chromatography 639 thermal process 704
column, multi-stage 636 thermal processing 727
irradiation 906 levan, hydrolized 69
irreversible thermodynamics 394 d-limonene 1062
isomerization, reaction kinetics 483 linoleic acid hydroperoxide 1059
lipase 492, 852
immobilized 591, 594, 846, 849
Jam 182 modified 588
hypocaloric 102 lipid 480
jelly strength 96 oxidation 474
jet agglomeration 244 lipids oxidation 591
juice, grape 102 liquid
foods 951
liquid extraction 674
kamaboko 96 particle system 292
kinetic lisophospholipid 173
model, fouling 805 low temperature 897
 1081
Luikov's model 418 curd 93
L-lysine 582 fat 840
lysozyme, recovery 639 globule 54
hydrolysate 1047
processing 24, 805
magnetic resonance imaging 215, 218 protein, hydrolysate 1047
method 471 whey 576
Maillard reaction 876 milling, wet 256
maize 123 mixer, intelligent 265
maltose production 504 mixing system 262
mandarins 894 mochi 87
mango 897 modelling 42, 543
marmalade 486 moisture
mass transfer 644 adsorption 846
maturation 143 control, dryer 987
maturity index 945 distribution 197
Maxwell model 81, 84 permeability 790
measure drying rate 930 sorption characteristics 406
measurement, cell density 567 molecular encapsulation 436
meat 936 monitor 963
low fat 250 monitoring 558
processing wastewater 1008 mungbean starch 665
mechanical emulsification 9 mushrooms 781
melons 954
melting 516
membrane 906 NAD(P)H regeneration 597
concentration 576 natural flavor 540
emulsification 167 near-infrared
enzyme reactor 686 analyzer 975
separation 585, 674, 680, 820 spectroscopy 939
technology 677, 683 network, physical property 1026
transport equation 650 neural
metabolic heat 763 network 984
methane production 1011 networks 36
methyl allyl trisulfide 1056 nisin 915
micro cut blender 259 nitrate 603
microbial reductase 1023
cells, disruption 733 NMR 15, 227, 230, 233, 241
insecticide 1020 pulsed-field-gradient 236, 239
microencapsulation 433 non-destructive 936
micro filtration 662 internal defect detection 957
microorganisms 256 quality evaluation 954
microwave 501, 504, 507, 1038 non-enzymatic browning 474, 501
heating 707, 796 non-invasive measurement 15
milk non-Newtonian flow 612, 921
1082
NOx removal 1023
ODS-silica gel 838
ohmic
heating 760, 769
thawing 307
oil
frying 63
hydrolysates, sunflower 680
oligopeptides 582
oligosaccharide 573
oligo saccharides 576
omega-3 fatty acids 627
optic probe 939
optical
fiber sensor 93
sensor 936
optimization
air dehydration 730
fermentation parameters 555
quality parameters 42
stochastic algorithm 730
thermal processing 730
orange juice factory 1002
osmo-convective drying 409
osmotic
concentration 394
dehydration 400, 403
oxidation
kinetic 474
lipid 1065
oxygen diffusion 480
Pacific whiting 1041
packaging 230, 778
materials 787
glass bottles 802
perforated polymeric 784
modified atmosphere 781, 784
papaya pulp 495
particle
diameter distribution 167
formation 30
motion 292
particles, flow rate 751, 754
particulate foods
.. aseptic processing 745
sterilization 769
pasta cooking 304
pasteurization 18
pectin, methoxyl 102, 111
pectinesterase, 495
Peng-Robinson equation, state 826
peptides, biologically active 1053
percolation theory 105, 114
permeability
moisture 790
oxygen 793
permeation
gas 787
measurement 787
peroxidants, foods 1059
pH-stat 561
Phaffia rhodozyma 537
phospholipase 909
photometric assay 567
physical property, food 1, 1026
physiological functions, foods 48
pigment production 537
plant
food 262
production 966
plant tissue 143
plastics 766
platelet aggregation 1056
poly-L-glutamate 1050
polyethylene terephthalate 796
polysaccharide 612
polyunsaturated fatty acid 477
population balance theory 9
pore size, ultrafiltration 647
porous structure 277
potato 507
irradiated tuber 906
shrinkage 137
potatoes, mashed 775
powder classification 247
practice, food engineering 1
 1083
pressure, controlling 724 cake 87
pressurized carbon dioxide 855 glutinous 66
processing 900 parboiled 81
product quality 403 ripening 519
protein RO membrane reactor 600
fish myofibrillar 99 roasting process 289
hydrolysate 176 rye flour 72
isolate 671
lupin 206
powders 30 Saccharmyces cerevisiae 918
recovery 668 sawdust media 525
proteins 633 scale-up 612
pulsed scaling law 105, 114
nuclear magnetic resonance 221 scraped surface heat exchanger 295
ultrasonic velocimetry 748 seal
puncture test 126 blubber 627
purification 641, 858 integrity 775
secondary ultra wave 957
semi-solid fermentation 606
quality sensorial analysis 212
change, kinetics 456 sensory evaluation 1044
evaluation 972 separation 843
parameters, product 42 cryo-mechanical 132
retention 885, 888 low fat meat 250
mambrane 820
shelf life 200, 772, 778, 781
rapeseed 671 shochu, waste water 1014
raspberry 152 shrimp 972
refining 677 block 307
rehydration, vegetable 412 shrinkage 394, 406
relaxation time 81 apple 137
rennet 93 silica retention 647
residence time distribution 745 SIMAN 262
response surface methodology 918 simulation 301
retort 716 winery 999
packages 775 skim milk 90
pouches 710 powder 927
retrogradation rate 468 soaking 400
reverse process 397
micelles 633, 689 soft-textured material 978
osmosis 650, 680 softness 155
Reynold's stress 644 solid state fermentation 528, 531
rheological parameters 510 solvation regions 30
Rhizopus 528 solvent extraction 621
nce sorghum 528
1084
sorption isotherm 182, 185 sublimation dehydration 424
soy substrate
protein 120 fermentation 549
isolate 57 removal 618
sauce 179, 615, 656, 659, 832 sugar solution 194
soya-plantain baby food 1032 sultana 280
soybean 259, 1035 supercritical
curd 120, 132 carbon dioxide 823, 826, 832, 835,
feed 236 840, 843, 846, 849, 852
protein 579, 873 extraction 630
residue 870 fluid 30, 814, 820
tempeh 522 chromatography 838
soybeans 668 extraction 817
specific resistance 653 surimi 96, 99, 1041
spontaneously hypertensive rat 1053 suspension 215
spores, Bacillus stearothermophilus 867 sweet potato 504, 549
spray drying 433, 439 sword bean, dehydrated 891
stability
emulsion 170, 173
glutathione 209 temperature 292
stabilization 879 low 897
starch 161, 185, 516 program 692
gelatination 459 sensitive products 427
gelatinization, wheat grain 471 tenderness value 412
statistical treatment 522 tensile strength 60
steam ternary diffusion process 430
fusion 719 texture 87
saturated 802 change 462
sterilization 18, 721, 724, 864, 867 development, vegetable 465
beverages 736 eating 126
packaging containers 799 theory, food engineering 1
particulate foods 769 thermal
pulsed discharge 736 conductivity 146, 149
radiation 739 conversion 161
sterilizer 719 death, kinetics 698
sterilizing process 713 diffusivity 152
stickiness 66 inactivation 495, 701
stiffness 84 process 286, 692
storage 140, 289, 891, 894, 897 lethality 704
life 909 processing 710
stability, oils 1062 lethality 727
structural parameters 424 optimization 730
structure property 283
dry powder 60 stability, enzymes 451
spray 439 thin-layer drying 442
 1085
time scales 15 viscoelastic properties 274
time-temperature viscoelasticity
integrator 692 bread 78
profile 707 butter 75
tofu 120 viscometer
gel centrifugation 1035 in-line 927
productivity 1035 new 921
tomato 241 viscosity 161, 206
total concentrate 927
energy management system 996 intrinsic 69
solid content 933
Tou-nao 1038
transition point 108, 111 waste water 549, 665
sol-gel 105 food industry 1020
triacylglycerol synthesis 594 water 227, 230
triglycerides 588 activity 176, 179, 221, 695
trypsin-catalyzed synthesis 582 transfer 421
tuna flesh 253 unfreezable 99
turbulent diffusion 644 uptake, rice 459
two-fluid atomizer 439 wheat flour 268
noodle 507
whey
ultra high temperature 719 permeate 1011
ultra-violet rays 799 protein concentrate 498
ultrafiltration 665, 668 turbulent flow 808
pore size 647 Whipping, continuous 271
ultrasonic 1038 wine-making 564
wave 957
unfrozen water 158
unsteady state diffusion 415 xylan 72
upflow anaerobic sludge blanket xylose 552
reactor 1002, 1011

yeast 597
vacuum, dryer 903 cell suspension 653
variable temperature history 453 yoghurt, drinking 90
vegetable
cheese 519
drying 442 zeolite 766
oil 677 zeta potential 164
velocity profile 218 Zymomonas mobilis 609