Recent Advances On Membrance Processes For The Contration of Fruit Juices A Review

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 22

Journal of Food Engineering 63 (2004) 303–324

www.elsevier.com/locate/jfoodeng

Recent advances on membrane processes for the concentration


of fruit juices: a review
B. Jiao a, A. Cassano b,*
, E. Drioli b

a
Citrus Research Institute, Chinese Academy of Agricultural Sciences, CRI-CAAS c/o Southwest Agricultural University,
Beibei, Chongqing 400712, China
b
Institute on Membrane Technology, ITM-CNR, c/o University of Calabria, via P. Bucci, cubo 17/C, Rende (CS) I-87030, Italy
Received 27 January 2003; accepted 8 August 2003

Abstract
Fruit juices have been traditionally concentrated by multi-stage vacuum evaporation, resulting in a loss of fresh juice flavors,
color degradation and a ‘‘cooked’’ taste due to the thermal effects. The promising alternative is reverse osmosis concentration.
However, it cannot reach concentrations larger than 25–30Brix with a single-stage RO system due to high osmotic pressure limi-
tation, which is quite below the value of 45–65Brix for standard products obtained by evaporation. Technological advances related
to the development of new membranes and improvements in process engineering have been proved to overcome this limitation. New
membrane processes, including membrane distillation and osmotic distillation, and integrated membrane processes are still being
identified and developed in concentrated fruit juice processing to improve product quality and reduce energy consumption. Recent
advances and developments of the use of membrane processes for concentrating fruit juice are reviewed and discussed in this paper.
Major attentions are focused on the application of new membrane processes and integrated membrane systems.
 2003 Elsevier Ltd. All rights reserved.

Keywords: Fruit juices; Concentration; Membranes; Reverse osmosis; Direct osmosis; Membrane distillation; Osmotic distillation; Integrated
membrane processes

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304

2. Reverse osmosis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304


2.1. Performance of RO membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
2.2. Effect of operating parameters on permeate flux and flavor rejection . . . . . . . . . . . . . . 305

3. Direct osmosis concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307


3.1. Process fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
3.2. Performance of the DOC process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307

4. Membrane distillation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309


4.1. Process fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
4.2. Effect of membrane types on the permeate flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
4.3. Effect of feed juice concentration on the permeate flux. . . . . . . . . . . . . . . . . . . . . . . . 310
4.4. Effect of operating temperature on the permeate flux . . . . . . . . . . . . . . . . . . . . . . . . . 310
4.5. Effect of flow rate on the permeate flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
4.6. Effect of UF pretreatment on the permeate flux. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
4.7. Effect of MD on the retention of fruit juice components . . . . . . . . . . . . . . . . . . . . . . 310

*
Corresponding author. Tel.: +39-0984-492011; fax: +39-0984-402103.
E-mail address: a.cassano@itm.cnr.it (A. Cassano).

0260-8774/$ - see front matter  2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfoodeng.2003.08.003
304 B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324

5. Osmotic distillation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310


5.1. Process fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
5.2. Effect of operating conditions on the OD flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
5.3. Effect of OD on the retention of fruit juice components . . . . . . . . . . . . . . . . . . . . . . . 312
5.4. OD membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
5.5. Effect of UF pretreatment on the evaporation flux. . . . . . . . . . . . . . . . . . . . . . . . . . . 314
5.6. Process design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315

6. Integrated membrane processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316

7. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322

1. Introduction covery (Gadea, 1987; Medina & Garcia, 1988; Paulson,


Wilson, & Spatz, 1985; Pepper, 1990).
To reduce the storage and shipping costs, and to Recently, technological advances related to the de-
achieve longer storage, fruit juices are usually concen- velopment of new membranes and improvements in
trated by multi-stage vacuum evaporation. This process process engineering have been proved to overcome this
results in a loss of fresh juice flavors, color degradation limitation. New membrane processes including mem-
and a ‘‘cooked’’ taste due to the thermal effects. Since brane and osmotic distillation and integrated membrane
consumers generally prefer the flavor, aroma, appear- processes might contribute to the improvement of con-
ance and mouth feel of freshly squeezed juices, scientists 
centrated fruit juice processing (Alvarez et al., 2000;
and processors have tried to develop new techniques for Calabro, Jiao, & Drioli, 1994; Cassano et al., 2003;
retaining such characteristics of freshly squeezed juice in Cuperus, 1998; Drioli, Calabr o, & Wu, 1987; Drioli,
the concentrate and ultimately in the reconstituted juice. Jiao, & Calabr o, 1992; Girard & Fukumoto, 2000;
The juice industry has developed complex essence re- Hogan, Canning, Peterson, Johnson, & Michaels, 1998;
covery, careful process control and blending techniques Jiao, Molinari, Calabr o, & Drioli, 1992; Petrotos &
to produce a good quality concentrate that is acceptable Lazarides, 2001; Vaillant et al., 2001). This paper will
to consumers, but still easily distinguishable from fresh provide a review of the recent significant progress on
juice. Many efforts have been devoted to develop im- membrane processes for concentrating fruit juices, in-
proved methods such as freeze concentration, sublima- cluding the use of reverse osmosis, direct osmosis con-
tion concentration and membranes (ultrafiltration and centration, membrane and osmotic distillation, and
reverse osmosis) for concentrated juice processing integrated membrane processes.
(Chen, Shaw, & Parish, 1993; K€ oseoglu, Lawhon, &
Lusas, 1990). The most promising alternative is mem-
brane concentration. 2. Reverse osmosis
Membrane processes as microfiltration (MF), ultra-
filtration (UF) and reverse osmosis (RO) have been 2.1. Performance of RO membranes
widely applied to the dairy, food and beverage industry
after the discovery of asymmetric membranes by Loeb Fruit juice concentration by RO has been of interest
and Souriragin in the early 1960’s. The potential of de- to the fruit processing industry for about 30 years. The
veloping RO process as a concentration technique to advantages of RO over traditional evaporation are in
remove water from fruit juices has been of interest to the low thermal damage to product, reduction in energy
fruit juice industry for about 30 years. The advantages consumption and lower capital investments (Merson,
of RO process over traditional evaporation are in lower Paredes, & Hosaka, 1980), as the process is carried out
thermal damage to product, increase in aroma retention, at low temperatures and it does not involve phase
less energy consumption and lower equipment costs. change for water removal. The retention of juice con-
One main disadvantage of RO has its inability to reach stituents, especially flavors, and the permeate flux, re-
the concentration of standard products produced by garding RO performance, are two major factors, which
evaporation because of high osmotic pressure limitation. are related to the type of membranes and the operating
Studies have shown that the final concentration of fruit conditions used during the process.
juices is limited by membranes and equipment to about Early studies by Merson and Morgan (1968) found
25–30Brix with the most efficient flux and solute re- that oil-soluble aromas of orange juice were easily re-
B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324 305

tained using cellulose acetate membrane, while there centration (Chou et al., 1991). The plate and frame
were losses of water-soluble aromas. Membrane reten- configuration had 11.5% permeation of n-hexanal
tion of some other juice constituents (sugars, acids, compared with 0.5% for the spiral wound. This was
potassium, calcium, magnesium, and phosphorus) was probably due to greater membrane packing density and
shown to be excellent (Peri, Battisti, & Setti, 1973). membrane area of the spiral wound (3.90 m2 of total
After that, a considerable research has been carried membrane surface area) when compared with the plate
out for the concentration of a variety of fruit juices, and frame (0.36 m2 of total membrane surface area).
including apple, pears, grapefruit, kiwi, pineapple, pas- The spiral wound system gave lower concentrations of
sion fruit and tomato juice (Bowden & Isaacs, 1989; n-hexanal in the permeate because of increased rejection
Braddock, Nikdel, & Nagy, 1988; Medina & Garcia, and/or of increased flavor compound absorption due to
1988; Palmieri, Dalla Rosa, Dall’Aglio, & Carpi, 1990; greater membrane surface. Braddock et al. (1988) re-
Paulson et al., 1985; Sheu & Wiley, 1983). These works ported that volatile compounds except for methanol,
have mainly focused on the test of different types of ethanol and traces of limonene, were not detected in
membranes and the effects of operating conditions on measurable quantities in the permeate, during the con-
the permeate flux and retention of juice compounds. centration of orange, grapefruit and lemon juices in the
It has been found that polyamide has greater retention range of 22–25Brix at pulp contents of 7–10%, using a
of flavors and other constituents as well higher fluxes composite tubular RO membrane (ZF99, Patterson
than cellulose acetate membranes (Chua, Rao, Acree, & Candy International Ltd., England, 99% NaCl rejection,
Cunningham, 1987; Fukutani & Ogawa, 1983a, 1983b; surface area 0.9 m2 ). However, a 17–30% loss of volatile
Sheu & Wiley, 1983). Recovery of apple juice volatiles peel oil (measured as limonene) was found if the mem-
during RO concentration to 20Brix was found to be brane system is not totally closed during recirculation of
about 80% when using a high resistance (HR) membrane the process stream to effect concentration.
(Sheu & Wiley, 1983). Chua et al. (1987) compared a
commercial spiral wound cellulose acetate membrane 2.2. Effect of operating parameters on permeate flux and
(CA-99) and an experimental polyamide membrane (PA- flavor rejection
99, 99% salt rejection) for concentrating apple juice using
a pilot-scale RO unit. The PA-99 membrane exhibited 
Alvarez, 
Riera, Alvarez, and Coca (1998) studied the
higher flux and concentration rates than the CA-99 permeation of a model apple juice aroma solution
membrane. In addition, the PA-99 membrane retained through a RO spiral wound aromatic polyamide mem-
45% of the total odor-active volatiles in the feed, while brane (MSCB 2521 R99, Separem, Biella, Italy, surface
the CA-99 membrane retained 23%. Medina and Garcia area 1.03 m2 , NaCl rejection 99.2%). Rejection was
(1988) reported that the overall recovery of sugars, acids observed to increase with transmembrane pressure
and flavor-volatile components in orange juice was ap- (15–35 bar) and feed flows (200–600 l/h). The major
proximately 93% with a spiral wound 192 MSO2-P compounds responsible for apple flavor such as ethyl-2-
polyamide membrane (Osmonics, Minnetonka, MN, methyl butanoate, ethyl butanoate and hexanol were
effective surface area 1.28 m2 ). Sugar recovery (>98%) rejected to a degree of more than 90% under the best
was higher than acid recovery (85%). Due to the unequal operating condition (35 bar and 600 l/h). The retention
membrane rejection of sugars and organic acids, the of ethyl acetate and isopentanol having a negative effect
Brix/acid ratio increased, making sweetness more per- on apple flavor was 80.2% and 98.4%.
ceptible, thus improving one of the quality indicators. It was observed that an increase in processing tem-
Chou, Wiley, and Schlimme (1991) compared the per- perature tends to increase the permeation rate of vola-
meation of apple juice volatiles by polyether-urea (TFC tiles, but this effect seems to depend on the type of
SS, Fluid Systems, Allied-Signal, Inc., San Diego, CA, volatiles to some extent. Recoveries decreased from
surface area 1.95 m2 ) and polyamide (TFCL HP, Fluid 41.7% to 29.2% for hexanal and from 56.6% to 50.8%
Systems, Allied-Signal, Inc., San Diego, CA, surface area for ethyl-2-methylbutyrate during RO concentration of
1.95 m2 ) thin-film composite membranes during RO. apple juice using a spiral wound polyamide membrane
There was two to three times less permeation of hexanal at 20 and 40 C, respectively (Chou et al., 1991). How-
and ethyl-2-methylbutyrate into the permeate of TFCL ever, the recovery of these two flavor compounds in the
HP when compared with TFC SS. Thus, apple juice retentate increased by approximately 20% with the in-
concentrated using the polyamide membrane would crease of operating pressure from 35 to 50 bar. Increased
demonstrate a more intense flavor than juice concen- pressure can result in higher flux, reduced process time
trated using the polyether-urea membrane. Polyamide and consequently smaller flavor compound losses at-
membranes are also known to have a longer useful life tributed to volatilization and adhesion to the membrane.
than cellulose acetate. The permeate flux has been affected mainly by
Different membrane configuration may also affect the transmembrane pressure (TMP), followed by tem-
retention of flavor compounds during apple juice con- 
perature and flow rate (Alvarez, 
Alvarez, Riera, &
306 B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324


Alvarez, 
1997; Alvarez, Riera, Alvarez, & Coca, 2001; viscosity and density of the solution on the membrane
Fukutani & Ogawa, 1983a; Sheu & Wiley, 1983; Sheu & surface. Thus, the mass transfer coefficients of the

Wiley, 1984). Alvarez et al. (1997) reported that the aroma compounds are lower, and their concentrations
permeate flux, through a tubular polyamide RO mem- on the membrane surface as well as their permeation
brane (AFC 99, Paterson Candy International Ltd, increase. Good agreement between experimental and
England, 99% NaCl rejection, total effective surface area estimated data was observed for most aroma com-
0.9 m2 ) increased by increasing TMP (at a constant pounds and operating conditions used.
concentration). The TMP also affected the maximum In spite of the high selectivity and solute retention
concentration that can be obtained with the limit being capacity of RO membranes, this process has a signifi-
19, 23, and 27.5Brix at 35, 45 and 55 bar, respectively. cant drawback. The high osmotic pressure of fruit juice
The high permeate flux was obtained at high feed ve- precludes its concentration to the required level of total
locities, but it considerably diminished for values higher soluble solids. Studies (Medina & Garcia, 1988; Pepper,
than 1.5–2 m/s and did not affect the maximum con- Orahard, & Merry, 1985; Sheu & Wiley, 1983) have
centration that can be obtained. It was also showed that shown that for cellulosic and non-cellulosic membranes
the permeate flux decreased as concentration increased the most efficient flux and solute recovery are at a con-
due to the increase of juice osmotic pressure and its centration lower than 30Brix. This suggests that RO
viscosity. For apple juice concentration, the average system should be viewed as preconcentration step with
processing capacity increased about 3–4% for every 1 C other technologies like freeze concentration or evapo-
increase in operating temperature between 20 and 60 C ration combined process; it has the advantage of re-
(Sheu & Wiley, 1983). At higher temperature, the ducing energy consumption and increasing production
membrane permeability coefficient is higher, the diffu- capacity (Gadea, 1987; Sheu & Wiley, 1983).
sivity coefficient in the solution increases and the vis- Gadea (1987) described the first commercial RO plant

cosity coefficient decreases (Alvarez et al., 1998). for concentrating orange juice using AFC99 membrane
In addition, enzymatic pretreatment of orange juice as first step, before final freeze concentration. The feed
(with pectinase) increased permeate flux and did not flow rate could be varied between 4200 and 9700 l/h of
affect solute recoveries (Medina & Garcia, 1988). De- natural orange juice with a water removal rate of 2000 l/
pectinization of orange juice increased RO flux initially h. The membranes showed an excellent retention of juice
up to 15% hydraulic recovery, but flux decreased beyond constituents as illustrated in Table 1. The organoleptic
this point. Fukutani and Ogawa (1983a) found clarified characteristics, examined by specialists in taste tests
Satsuma mandarin juice had better flux than cloudy
juice (3% v/v). On the contrary, both clarified and
cloudy pineapple juice were found to have similar flux Table 1
Retention of orange juice constituents in RO concentration (Gadea,
during RO (Bowden & Isaacs, 1989). Permeate flux de- 1987)
creased as feed flow velocity decreased below 2 m/s, but
Juice strength Permeate
there was little difference between 2 and 3 m/s velocities.
Dissolved solids (Brix) 11.2 0.0
The discrepancy may be due to different pulp charac-
Glucose (g/l) 26.6 0.1359
teristics of the two juices. Fructose (g/l) 28.4 0.1552

Recently, Alvarez et al. (2001) proposed the prefer- Saccharose (g/l) 35.1 0.0815
ential sorption-capillary flow model for the prediction of Citric acid anhydrous (%, w/w) 0.85 0.01
the rejection of aroma compounds and permeate flux Vitamin C (ppm) 480 6.0
Ash (%) 4.1 0.11
during the RO concentration of model solutions simu-
Sodium (ppm) 59 2
lating apple juice through the MSCB 2521 R99 spiral- Potassium (ppm) 1427 47
wound membrane (Separem, Biella, Italy, surface area Phosphorous (ppm) 200 2
1.03 m2 , NaCl rejection 99.2%). Water and aqueous Aspartic acid (mg/100 ml) 28 0.35
solutions of apple juice aroma were used to determine Asparagine (mg/100 ml) 45 1.0
Proline (mg/100 ml) 80 2.4
the parameters included in the model. Then, the model
Glycine (mg/100 ml) 1.5 Traces
was tested with aqueous solutions of glucose and aroma Alanine (mg/100 ml) 9.9 1.5
compounds. Experiments were performed at different Valine (mg/100 ml) 1.8 0.034
TMP (15–35 bar), feed flow rates (200–600 l/h), Rey- Isoleucine (mg/100 ml) 0.0 Traces
nolds numbers (51–190), temperatures (15–30 C) and Leucine (mg/100 ml) 0.5 Traces
Phenylalanine (mg/100 ml) 3 0.04
glucose concentrations (9.1–17.0Brix). It was observed
c-Aminobutyric (mg/ml) 24 0.46
that the permeate flux and aroma rejection increased Histidine (mg/ml) 1.1 0.062
with feed flow rate and decreased with the glucose Ornitine (mg/ml) 0.8 0.32
concentration. This is due to the reduction of concen- Lysine (mg/ml) 2.9 0.14
tration polarization by increasing feed flow rate. A rise Arginine (mg/ml) 70 0.42
Total nitrogen (g/l) 1363 11
in the feed concentration involves an increase in the
B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324 307

Table 2
Advantages and disadvantages of RO applied to the concentration of fruit juices
Advantages Disadvantages
Technical aspects
• Broad industrial scale application • Fouling phenomena
• Low temperatures • High pressures
• Combination with vacuum evaporation and with systems • Necessity of an inactivation enzyme pre-treatment
of vapour recompression, already commercially available • Juice concentration limited at 22–23Brix
• Loss of aroma compounds during the process
• Difficulty to concentrate solutions with high level of suspended
solids
Economical aspects
• Energetically and economically convenient in comparison • Membrane replacement at high cost
with the thermal evaporation • High cost of working process

(Leathered Food RA), as well as the physical/chemical Stirring improved the osmotic flux by 67%. However,
specification determined in the laboratory, showed no the diffusion of some salt through the membranes was
difference between natural and the reconstituted reverse observed in the grape concentrate, when using a satu-
osmosis concentrate. rated brine solution as the OA solution.
Advantages and disadvantages of the RO process for OSMOTEK, Inc. of Corvallis developed the direct
the production of concentrated juices are summarized in osmosis technique as an efficient alternative to thermal
Table 2. evaporation for concentrating food products. Beaudry
and Lampi (1990a, 1990b) reported the performance of
the Osmotek direct osmosis process for concentrating
3. Direct osmosis concentration fruit juices. They utilized new generation thin film
composite (TFC) RO membranes that were properly
3.1. Process fundamentals modified. The overall membrane thickness was 25–85
lm, while the membrane’s top selective layer remained
Direct osmosis concentration (DOC) is another similar to that of tight RO membranes (MWCO less
membrane process capable of concentrating fruit juice at than 100 Da), thus impeding passage of other food in-
low temperatures and low pressures, thereby maintain- gredients besides water. More than 99.9% of the total
ing original flavor and color characteristics of the fruit. acid in orange juice and the color of red raspberry juice
The principle uses an osmotic agent (OA) solution to were rejected by the membrane with a nominal MWCO
establish an osmotic pressure gradient across a semi- of 100 Da. Furthermore, a tubular membrane module
permeable membrane and thus remove water from a with specific design was constructed to provide housing
single strength fruit juice. An osmotic agent is generally for the sensitive ultrathin direct osmosis membranes. It
a solid highly soluble in water, hygroscopic, non-toxic, improved turbulent flow and minimized flux resistances
inert toward the flavor, odor and color of the foodstuff, rising from concentration polarization and membrane
and which does not pass through the membrane. Gen- fouling, allowing an increase in the osmotic flux. This
erally, the higher the concentration of the dissolved module configuration formulated a flow path (inside the
solids and the lower the molecular weight of the dis- membrane) with a continuous change in flow direction,
solved solids, the higher the osmotic pressure. The most providing at the same time effective support to the
frequently employed constituents are: sodium chloride, membranes. A similar flat module was also developed.
sucrose or glycerol, cane molasses or corn syrup. The These new developments were outlined in a US patent to
OA solutions must have an osmotic pressure greater produce fruit juices (Herron, Beaudry, Jochums, &
than that of the concentrated fruit juice. For example, Medina, 1994). Some theoretical aspects of this method
the osmotic pressure of a 74Brix high fructose corn as well as its practical application concerning membrane
syrup is about 270 bar that is greater than 90 bar for type, membrane configuration and suitable methods to
42Brix pulpy orange juice. support the thin membrane, were discussed in the patent
document. The maximum osmotic flux of 5–6 l/m2 h was
3.2. Performance of the DOC process obtained through new 100 Da membranes for several
kinds of liquid food samples including orange and
Popper, Camirand, Nury, and Stanley (1966) first raspberry juice. In addition, no transfer of osmotic
used DOC to concentrate grape juice from 16 to agent through the membrane was detected. A fructose/
60Brix. An average osmotic flux of about 2.5 l/m2 h was glucose solution (approximately 74Brix) such as high
obtained with a flat sheet cellulose acetate membrane. fructose corn syrup served as the osmotic agent that
308 B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324

flows countercurrent to the fruit juice. The membrane to 0.3 l/m2 h, while the OA solution decreased from 69 to
assembly of this invention allows for rapid turbulent 60Brix (Beaudry & Lampi, 1990b). The flux to make
flow, without excessive pressure drop, and a relatively 42Brix pulpy orange juice was similar. The time re-
long region of contact with the membrane without quired to concentrate raspberry juice to 45Brix ap-
fouling or concentration polarization even with fluids proximately doubled (5 h vs. 10 h) when the processing
containing 75% or more of dissolved solutes and 35% temperature decreased from 26 to 8 C (Wrolstad et al.,
suspended solids. In Fig. 1 the simplified DOC flowsheet 1993).
is reported. Raspberry concentrates (45Brix) prepared by Os-
Unlike RO process, pressure differences across the motek’s DOC system were compared with conventional
membrane in DOC are negligible and flux depends on evaporative technology (i.e., centrifugal film evaporator)
the difference in osmotic potential. The only hydraulic (Wrolstad et al., 1993). Concentration by either pro-
pressures required are those (approximately 2 bar) nec- cesses resulted in small anthocyanin pigment losses,
essary to pump the juice and the OA solution over the small increases of polymerized pigment and no signifi-
membrane surfaces (Milleville, 1990). Juices containing cant flavor differences as compared to the single-strength
large amounts of both dissolved and suspended solids juice. Concentration of red raspberry juice using either
can be concentrated with minimum fouling because the osmotic or evaporative processes gave juice concentrates
solids are not forced against the membrane. The fruit of good quality comparable to commercial samples.
juices can be fed to the module in a continuous, multi- Herron et al. (1994) also provided evidence that the
stage process. At each stage the fruit juice is recycled DOC can produce juice concentrate with superior
through a heat exchanger (if refrigeration is desired) and quality in comparison to that yielded by a conventional
then through the DOC modules. The juice is kept under vacuum evaporator. These studies gave an answer to
a nitrogen or argon blanket to assist in flavor retention. earlier claims for a negative effect of the DOC process
The OA solution is fed either once countercurrently on the quality of the juice concentrate (Bolin & Salunke,
through the entire process or is recycled continuously 1971). In those days, the prolonged processing times due
through an evaporation/heat recovery system. If there is to the poor process performance was probably the rea-
a large quantity of very concentrated solution that is son for the observed quality deterioration.
diluted at the plant, and if this solution could be used as Recently Petrotos, Quantick, and Petropakis (1998,
the OA solution, the dilution can be performed within 1999) studied the effect of several membranes and pro-
the juice concentration modules. This would result in cess parameters on the performance of DOC during the
substantial cost savings for the juice concentration concentration of tomato juice. These studies suggest that
process (Girard & Fukumoto, 2000). Typically, the OA the membrane thickness and the viscosity of the osmotic
solution will be diluted and reconcentrated using evap- agents are of greater importance. A thinner membrane
oration. Therefore, the OA solution rather than the juice and a low-viscosity osmosis medium (like NaCl brines)
is exposed to the high temperature, allowing the im- are expected to yield a better performance. The findings
provement of concentrate quality. indicated that the use of viscous carbohydrate syrups as
Under recirculation conditions, the flux can be osmotic media has to be abandoned in favour of highly
maintained at a steady rate indicating minimum fouling. osmotic and low-viscosity salt brines. The pretreatment
But as the juice becomes more concentrated, its osmotic of the juice by plain filtration, microfiltration or ultra-
pressure increases; the flux then decreases due to the filtration was also found to have a significant effect on
decrease of osmotic pressure differences between the OA the value of the observed osmotic fluxes. The ultrafil-
solution and the juice. During a run, the flux of a de- tered juice gave the best performance than that of all
pectinized raspberry juice at 30 C decreased from 1.37 other filtrates with a 135% increase of the direct osmosis
flux.
Petrotos and Lazarides (2001) developed a new os-
motic apparatus with a flat configuration (Fig. 2). The
membrane sugar (OA)
Product
flow key of the system is an osmotic cell of special configu-
concentrate
water ration to promote turbulence. Using the apparatus
equipped with a commercially RO membrane, with an
heat
exchanger overall thickness of 260 lm, a tomato juice was con-
concentration
cells evaporator centrated from 5 to 16Brix at low pressure and room
temperature. The obtained average flux of 4.5 kg/m2 h is
Recirculation water
product loop
close to the values reported by Herron et al. (1994), al-
inlet though the membrane used in this study was much
thicker than the membranes used by Herron.
Advantages and disadvantages of the DOC process
Fig. 1. Simplified DOC flowsheet (Herron et al., 1994). are summarized in Table 3.
B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324 309

Stirrer

Brine
stock tank

Membrane
Part A
Part B
Tomato juice
stock tank

Pressure Pressure
Pressure Pressure Control valve
Control valve
meter meter

Pressure Pressure
Pressure Pressure
Control valve Osmotic Control valve
meter meter
cell
Membrane Membrane
pump pump

Fig. 2. The DOC schema for the concentration of fruit juice through flat sheet membrane module (Petrotos & Lazarides, 2001).

Table 3
Advantages and disadvantages of DOC applied to the concentration of fruit juices
Advantages Disadvantages
Technical aspects
• Low temperatures, low pressures • New technology that requires an evaluation at industrial level,
• No fouling problems, constant permeate flux in time flexibility to be evaluated
• Possibility to obtain high TSS concentration in the final • The long life of the membrane to be evaluated
product • Relatively low permeation (1.8–2.5 l/m2 h)
• Modularity, easy scale-up
• Possibility to treat solutions with high level of suspended solids
• Possibility of using modules in series, the same unit can
concentrate several different products
Economical aspects
• Low cost of membrane replacement • High investment costs
• High energy consumption

4. Membrane distillation The theoretical aspects and potential applications of


the process have been extensively studied (Bandini,
4.1. Process fundamentals Gostoli, & Sarti, 1991; Calabr o et al., 1994; Kimura,
Nakao, & Shimatani, 1987; Schofield, Fane, & Fell,
Membrane distillation is a relatively new membrane 1987, 1990). The most suitable materials for MD
process in which two aqueous solutions, at different membranes include polyvinyldifluoride (PVDF), poly-
temperatures, are separated by a microporous hydro- tetrafluoethylene (PTFE) and polypropylene (PP). The
phobic membrane. In these conditions a net pure water size of micropores can range between 0.2 and 1.0 lm.
flux from the warm side to the cold side occurs. The The porosity of the membrane will range from 60% to
process takes place at atmospheric pressure and at 80% of the volume and the overall thickness from 80–
temperature that may be much lower than the boiling 250 lm, depending on the absence or presence of sup-
point of the solutions. The driving force is the vapour port. In general, the thinner the membrane and the
pressure difference between the two solution–membrane greater the porosity of the membrane, the greater the
interfaces due to the existing temperature gradient. The flux rate. The membrane configurations used include flat
phenomenon can be described as a three phase sequence: sheet, spiral wound and hollow fiber; the latter has re-
(1) formation of a vapour gap at the warm solution– ceived the greatest attention.
membrane interface; (2) transport of the vapour phase Because MD can be carried out at the atmospheric
through the microporous system; (3) its condensation at pressure and at a temperature which can be much lower
the cold side membrane–solution interface (Calabr o than the boiling point of the solution, it can be used to
et al., 1994; Curcio, Barbieri, & Drioli, 2000; Drioli & concentrate solutes sensitive to high temperature (e.g.
Wu, 1985; Jariel, Reynes, Courel, Durand, & Dornier, fruit juices), also at high osmotic pressure. Therefore
1996; Jiao, Molinari et al., 1992; Jiao, Calabr
o, & Drioli, MD has received a great attention as techniques for
1992). concentrating fruit juices.
310 B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324

4.2. Effect of membrane types on the permeate flux vere than that of high crossflow velocity (Franken,
Mulder, & Smolders, 1990).
Drioli et al. (1992) reported the concentration of or-
ange juice by MD, using a commercial plate PVDF 4.6. Effect of UF pretreatment on the permeate flux
membrane (Millipore Corp.) with a nominal pore size of
0.22 lm and a laminated hydrophobic microporous It was found that ultrafiltration (UF) of the single
membrane (G0712) with a pore size of 0.2 lm manu- strength juice resulted in an increase in MD flux and
factured by Gelman Science Tech. Ltd. The flux of that the flux remained essentially constant during an
PVDF membrane was remarkably higher than of G0712 approximately two-fold concentration (Drioli et al.,
membrane. Curcio et al. (2000) reported that the single 1992). The MD flux of the unfiltered juice, by contrast,
strength apple juice could be concentrated to 64Brix by decreased steadily over the same concentration range.
MD using a PP hollow fiber membrane module (ENKA The improvement in MD flux after UF was attributed to
Microdyn MD-020-2N-CP, membrane area 0.1 m2 , a reduction in juice viscosity as a result of pulp and
nominal pore diameter 0.45 lm). An overall flux of 1– pectin removal.
1.5 kg/m2 h was obtained in this study.
4.7. Effect of MD on the retention of fruit juice
4.3. Effect of feed juice concentration on the permeate flux components

Permeate flux decreases with an increase of feed juice The preliminary studied on the concentration of or-
concentration. This phenomenon can be attributed to ange juice by MD (Drioli et al., 1992) showed that the
the reduction of the driving force due to the decrease of PVDF membrane has very good retention of orange
the vapour pressure of the feed solution and to the ex- juice compounds such as total soluble solids, sugars and
ponential increase of the viscosity of the juice solution. organic acids; the rejection of sugars and organic acids
At high concentration ratios, fluxes are higher in the were 100%. Vitamin C decreased by 42.1%. This de-
MD process than in the RO (Drioli et al., 1992). gradation has mainly associated with high temperature
and oxidation. As a result, the operating temperature in
4.4. Effect of operating temperature on the permeate flux MD process must be maintained as lower as possible.
Investigations have been carried out into the rela-
The MD flux gradually increases with an increase of tionship between the structure of hydrophobic mem-
the temperature difference between feed juice and the branes and the loss of volatile organic flavour/fragrance
cooling water. At lower temperature differences a de- components from liquid foods in MD process. The re-
creasing of the permeate flux is observed due to the sults have suggested that the best volatiles retention for
lower vapour pressure difference between the two side of a given amount of water removal in membrane distil-
the membrane (Drioli et al., 1992). lation is provided by membranes having an open fibrous
Lagan a, Barbieri, and Drioli (2000) reported that structure rather than discrete pores.
MD was able to concentrate apple juice up to 64Brix
with fluxes of 1 kg/m2 h. The viscosity of the juice at high
concentration induces severe polarization phenomena. 5. Osmotic distillation
However, temperature polarization is more important
than concentration polarization and it is located mainly 5.1. Process fundamentals
on the feed side. The presence of the osmotic effects on
juice stream generates flux inversion. A small DT of 4 C Osmotic distillation is a recent membrane process
was sufficient to compensate this osmotic effect. (Lefebvre, 1988), also known as osmotic evaporation
(Deblay, 1995), membrane evaporation, isothermal
4.5. Effect of flow rate on the permeate flux membrane distillation or gas membrane extraction
which has been successfully applied to the concentration
Studies demonstrated that the increase of flow rate of liquid foods such as milk, fruit and vegetable juice,
determines an increase of the permeate flux in MD instant coffee and tea and various non-food aqueous
process (Drioli et al., 1992). The shearing forces gener- solutions.
ated at high flow rate cause a lower accumulation of This technique can be used to extract selectively the
particulates such as pectin and cellulose on the active water from aqueous solutions under atmospheric pres-
membrane surface and thus reduce membrane fouling. sure and at room temperature, thus avoiding thermal
Besides, a lower crossflow velocity causes a lower Rey- degradation of the solutions (Kunz, Benhabiles, & Ben-
nolds number, thus hindering the heat transfer from the Aim, 1996). It is therefore particularly adapted to the
bulk of the solution to and from the membrane surface, concentration of heat-sensitive products like fruit juices
and an effect of temperature polarization are more se- (Deblay, 1995; Durham & Nguyen, 1994; Gostoli, 1998;
B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324 311

Johnson, Valks, & Lefebvre, 1989). The process involves maintain the water vapour pressure gradient. OD, on
the use of a microporous hydrophobic membrane to the other hand, does not suffer from any of the problems
separate two circulating aqueous solutions at different mentioned above when operated at room temperature.
solute concentrations: a dilute solution and an hyper-
tonic salt solution. If the operating pressure is kept 5.2. Effect of operating conditions on the OD flux
below the capillary penetration pressure of liquid into
the pores, the membrane cannot be wetted by the solu- Sheng et al. (1991) studied the effect of operating
tions. The difference in solute concentrations, and con- conditions (juice flow rate, juice concentration and
sequently in water activity of both solutions, generates, temperature) on the OD flux during the concentration of
at the vapour–liquid interface, a vapour pressure dif- apple, orange and grape juice through a PTFE mem-
ference causing a vapour transfer from the dilute solu- brane with pore size of 0.2 lm and an overall thickness
tion towards the stripping solution. The water transport of 100 lm. The results showed that the OD flux de-
through the membrane can be summarized in three creased with the increase of juice concentration and it
steps: (1) evaporation of water at the dilute vapour– depends strongly on the osmotic pressure difference Dp
liquid interface; (2) diffusional or convective vapour between the aqueous streams. When Dp decreased of
transport through the membrane pore; (3) condensation 33%, from 416 atm (low juice concentration) to 280 atm
of water vapor at the membrane/brine interface (Courel, (high juice concentration), a five-fold decline in OD flux
Dornier, Herry, Rios, & Reynes, 2000; Gostoli, 1999; was observed.
Hogan et al., 1998; Lefebvre, 1992; Mengual, Zarate, Vaillant et al. (2001) evaluated the potential of OD
Pena, & Velazquez, 1993; Sheng, Johnson, & Lefebvre, for concentrating clarified passion fruit juice on an in-
1991). dustrial scale, taking into account the relevant impact
The typical OD process involves the use of a con- on the overall product quality. A pilot plant containing
centrated brine at the downstream side of the membrane a 10.2 m2 of PP hollow fibres module was used to con-
as the stripping solution. A number of salts such as centrate passion fruit juice up to a total soluble solids
MgSO4 , CaCl2 , K2 HPO4 are suitable. As shown in Fig. (TSS) content higher than 60Brix at 30 C. Tangential
3, potassium salts of ortho- and pyrophosphoric acid velocity, temperature and concentration of solutions
offer several advantages, including low equivalent significantly affected OD flux. An average evaporation
weight, high water solubility, steep positive temperature flux of almost 0.75 kg/m2 h was obtained with water,
coefficients of solubility and safe use in foods and 0.65 kg/m2 h when juice was concentrated to 40Brix and
pharmaceuticals (Deblay, 1995; Hogan et al., 1998; Le- 0.50 kg/m2 h when it reached 60Brix. A long-term trial,
febvre, 1992; Michaels, 1998). lasting 28 h, was successfully carried out without
As compared with RO and MD process, the OD membrane fouling. Osmotic evaporation can also be
process has the potential advantage which might over- conducted as a multistage procedure, giving a constant
come the drawbacks of RO and MD for concentrating evaporation flux of around 0.62 kg/m2 h when juice was
fruit juice, because RO suffers from high osmotic pres- concentrated from 14 to 60Brix. Sensory quality and
sure limitation, while in MD some loss of volatile vitamin C content were well preserved in the concen-
components and heat degradation may still occur due to trated juice. It was observed that the OD flux decay was
the heat requirement for the feed stream in order to more attributable to the reduction of the stripper con-
centration at a low TSS of the feed juice, while it mainly
depends on viscous polarization (juice viscosity) when
35 juice concentration reached a value higher than 40Brix.
Cassano, Jiao, and Drioli (2002) and Cassano et al.
30 (2003) confirmed these observation during the concen-
Vapor pressure of water, mbar

25
tration of clarified kiwifruit juice and orange juice by
osmotic distillation. Operating factors that are expected
20 to enhance the performance of the OD process are the
use of higher temperature than ambient (to increase
15
the driving force, decrease the viscosity and to en-
Sucrose
10 K2HPO4
hance the solute diffusivity) and improvement of the cell
NaCl2 hydrodynamic conditions in the OD module (Courel
5 CaCl2 et al., 2000; Hogan et al., 1998). Examples of this include
0 the use of a novel helically wound hollow fibre module
0 10 20 30 40 50 60 70 80 and a transverse flow hollow fibre module (Hogan et al.,
Wt. % Solute
1998). Courel et al. (2000) summarized in detail the ef-
Fig. 3. The vapour pressure/concentration relationships for sugar and fect of the operating condition on the OD flux during
salt solution at 25 C. the concentration of sucrose solution (Table 4). The
312 B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324

Table 4 graphic analysis of 20–35 volatile components showed a


Summary of the relative effect of the various operating parameters loss of about 32% of volatile components in orange and
studied on the OD flux (Courel et al., 2000)
about 39% in passionfruit juices (Tables 5 and 6). As
Operating parameters Effect on vapour flux seen from Table 5, during the OD processing of orange
Solute content )10% Csalt )10% to )40% juice, the sample A had two components, hexanol and
+10% Csugar )8% to )95%
nonanal increased and the sample B had six components
Fluid velocity +10% Qsalt +2% to +6% (mostly alcohols) which increased. For both orange and
+10% Qsugar Not quantified passionfruit, sensory evaluations using triangle differ-
Mean temperature +10% T +6% to +12% ence tests showed a significant difference between the
initial juice and the reconstituted concentrate. In orange
Presence of DT DT ¼ 10 C Pure water: +75%
55–65 w/w% sucrose: juice, a trained taste panel was used to evaluate six
6 10% specific flavour characteristics in the two juices. The
Csalt ¼ concentration of brine solution, Csugar ¼ concentration of sugar characteristics evaluated were green, fresh orange, fru-
solution, Qsalt ¼ velocity of brine solution, Qsugar ¼ velocity of sugar ity/floral, peel oil, cooked/processed and terpene. There
solution, T ¼ temperature. were no significant differences found by the panel be-
tween the juices in any characteristic, but the average
flavour score for each characteristic was slightly lower in
solute content, both salt and sugar, is the most influ- juice from concentrate than in the initial juice.
encing variable on mass transfer. Transfer of three juice aroma compounds (ethyl bu-
tyrate, hexyl acetate and hexanol) in very diluted aque-
5.3. Effect of OD on the retention of fruit juice compo- ous solution was recently evaluated during the
nents concentration by OD with PTFE plate membrane under
various operating conditions (Ali, Dornier, Duquenoy,
Barbe, Bartley, Jacobs, and Johnson (1998) tested & Reynes, 2002). The results showed that the transfer of
nine types of hydrophobic microporous membranes for volatiles depends on the nature of compound but also
their influence on the retention of a range of volatile on the operating conditions. By decreasing the circula-
organic species when model aroma aqueous solutions tion velocity or the temperature, it is possible to slow
were subjected to osmotic distillation. Similar studies down the transfers significantly. Experimental kinetics
were also carried out on Gordo grape juice and Valencia can be characterized by a global coefficient of transfer
orange juice. GC-MS head-space analyses of the feed which connects flux to the bulk concentrations. There-
materials, coupled with SEM and image analyses of the fore, a more detailed approach of the phenomena which
used membranes, indicated that lower organic volatiles govern the mass transfer of juice volatiles must be un-
flux to water flux ratios occurred when pore sizes at the dertaken to better optimize the process.
membrane surface were relatively large. Membranes
having relatively large pore sizes at the surface have 5.4. OD membranes
been shown to be associated with higher organic vola-
tiles retention per unit water removal than those with The most well-known module designed for OD is
smaller surface openings. A simple model based on the Hoechst-Celanese Liqui-Cel membrane contactor
00
differences in their resistances to organic volatiles (Liqui-Cel Extra-Flow 2.5 · 8 , effective surface area
transport has been proposed. Accordingly, pores with 1.4 m , effective area/volume 29.3 cm2 /cm3 , fibre potting
2

larger diameters at the membrane surface allow greater material polyethylene, maximum transmembrane dif-
intrusion of the liquid feed and brine streams with a ferential pressure 4.08 bar, temperature operating range
resulting increase in the thickness and hence resistance 1–40 C) containing microporous polypropylene follow-
of the boundary layer at the pore entrance. On the basis fibres of Celgard membrane. These fibres are approxi-
of this study it can be concluded that the utilization of mately 0.3 mm in external diameter with a wall thickness
membranes with large surface pore diameters may be of about 0.03 mm; they have a mean pore diameter of
beneficial in laboratory and industrial applications of about 30 nm and a porosity of about 40%.
osmotic distillation in which the retention of volatile Durham and Nguyen (1994) studied the concentra-
flavour/fragrance components is desirable for product tion of tomato juice by OD, using the Gore-tex PTFE
quality. membrane and the Gelman 11104/2TPR (a crosslinked
Shaw et al. (2001) evaluated concentrated orange and acrylic fluorourethane copolymer). A fouling effect, due
passionfruit juices prepared by OD distillation in terms to some lipid constituents of the tomato juice, was re-
of retention of flavors. Both juices were concentrated ported. Furthermore, several membrane cleaning
three-fold to 33.5 and 43.5Brix, respectively, in a pilot methods were evaluated. The Gelman 11104/2TPR was
scale osmotic evaporator containing 10.3 m2 of PP found to be suitable for the concentration of tomato
hollow fibers. Quantitative headspace gas chromato- juice by OD. It would be cleaned with P3 Ultrasil 56
B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324 313

Table 5
Quantities (in mg/l) of 35 volatile compounds in orange juice before and after osmotic evaporationa (Shaw et al., 2001)
Compounds Sample A Sample B
b
Initial Conc. Initial Conc.b
c
Acetaldehyde 3.3 1.9 4.0 3.7
Methanolc 24 12 24 9.5
Ethanolc 429 292 535 419
Ethyl acetatec 0.20 0.12 0.25 1.05
2-Methypropanol 0.01 0.11 0.04 0.18
Butanol 0.06 0.03 0.25 0.16
Penten-3-ol 0.06 0.03 0.02 0.01
Penten-3-one 0.04 0.02 0.07 0.03
Penten-2-ol trd trd tr tr
Ethyl propinate 0.01 0.01 0.01 0.02
Methyl butanoate 0.03 0.01 0.04 0.01
3-Methylbutanol 0.06 0.05 0.29 1.04
2-Methybutanol 0.01 0.01 0.04 0.01
Pentanol tr tr tr tr
Hexanal 0.63 0.01 0.64 0.12
(E)-2-hexenal 0.62 0.62 0.62 0.61
(Z)-2-hexenale 0.26 0.15 0.18 0.12
Hexanol 0.07 0.20 0.04 0.4
Heptanal 0.05 0.01 0.0 tr
a-Pinene 1.53 1.35 0.99 1.14
Sabinene 0.18 0.16 0.09 0.09
Myrcene 8.3 7.4 6.3 7.23
Ethyl hexanoate 0.05 0.03 0.05 0.03
Octanal 0.81 0.4 0.48 0.22
a-Phellandrene 0.28 0.25 1.05 1.07
Limonene 276 247 204 240
Linalool 1.42 1.10 1.55 0.93
Nonanal 0.11 0.18 NDf NDf
Ethyl octanoate 0.36 0.14 0.25 0.10
4-Terpineol 0.21 0.17 0.33 0.24
Decanal 0.87 0.76 0.42 0.31
Carvone 0.23 0.10 0.10 0.05
Perillaldehyde 0.08 0.05 0.06 0.03
Caryophyllene 0.37 0.36 0.40 0.42
Valencene 5.6 3.4 16.7 8.7
Total oilg 420 300 250 280
a
Values determined by dynamic headspace GC unless otherwise noted.
b
Juice reconstituted to the Brix value of the initial juice prior to analysis.
c
Values, determined by static headspace GC with a polar column.
d
tr ¼ trace, too small to quantify.
e
A minor amount of (E)-2-hexenol eluted with this component.
f
ND ¼ not detected.
g
Determined by the Scott oil method.

(Henkel, UK). The OD flux was maintained and there polymers such as PTFE or PVDF or to make laminate
was no salt leakage during repeated fouling/cleaning membranes that prevent liquid intrusion without im-
trials. peding vapour transport (Hogan et al., 1998; Mansouri
Some fruit juices, especially citrus juices, contain peel & Fane, 1998; Michaels, 1999). It was found that the
oils and other highly lipophilic flavor components that hydrophilic polymer film of adequately high intrinsic
reduced their surface tension and promote wetting of water permeability and virtual impermeability to mac-
hydrophobic surfaces. Components of solutions used to rosolutes and colloids, which can be fabricated in suffi-
clean membrane devices often contain surface active cient thickness, is suitable for this laminate (Michaels,
agents and/or residuals from these solutions, that might 1999). Such hydrophilic polymers include those com-
also promote membrane wetting. Such liquids, in most posed of esters and ethers of cellulose; crosslinked gel-
cases, will promote wet-out and liquid penetration into atin; gelatinized starch; chitin; agar; alginic acid;
PP membranes (Hogan et al., 1998). Development work crosslinked polyacrylamide; poly (hydroxyethyl) meth-
currently is underway to attempt to produce hollow acrylate; crosslinked polyvinyl alcohol. Michaels (1999)
fiber microporous membranes from more hydrophobic reported that commercially available cellophane
314 B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324

Table 6 namely perfluro-2,2-dimethyl-1,3-dioxole (PDD), can be


Quantities (in mg/l) of 22 volatile compounds in passionfruit juice formed into non-porous gas membranes, which provide
before and after osmotic evaporationa (Shaw et al., 2001)
acceptable transmission rates for OD. Osmotic distilla-
Compounds Initial Conc.b tion utilizing such high free volume, non-porous gas
c
Acetaldehyde 2.5 2.1 membranes, as a thin layer on the feed side of com-
Methanolc 16 12
mercial microporous membranes, can concentrate pulpy
Ethanolc 307 57
Ethyl acetatec 0.31 0.11 fruit juices and especially, limonene-containing juices,
2-Methypropanol 0.06 0.03 such as orange juice. The process can be operated at
Butanol 0.06 0.05 high flux for long durations between membrane clea-
2-Methybutanol 0.38 0.22 nings. Additionally, less contamination of the strip so-
Methyl butanoate 0.007 0.003
lution into the feed juice occurs to provide a high quality
Hexanald 0.06 0.02
Hexanol 0.14 0.09 concentrate (Bowser, 2001).
a-Pinene 0.27 0.23 Surface morphology changes of hydrophobic Celgard
Benzaldehyde 0.004 0.002 2500 and Accurel 1E-PP membranes during initial
Sabinenee 2.23 1.54 contact with water were examined using scanning elec-
Myrcene 0.69 0.58
tron microscopy (SEM) and image analysis (Barbe et al.,
Ethyl hexanoate 0.13 0.08
Octanal 0.032 0.008 1998). The surfaces of both membranes showed
Hexyl acetate 0.017 0.010 increases in several of the following morphology pa-
Limonene 25 21 rameters after 72 h of contact: porosity, pore area, pore
c-Terpinene 1.37 1.14 length, pore breadth, pore equivalent diameter and pore
(E)-linalool oxide 2.2 1.2
spread factor. This was attributed to non-wetting in-
Linalool 0.29 0.19
4-Terpineol 0.04 0.03 trusion of the water meniscus into some pores with a
Total oilf 30 20 resulting enlargement of pore entrances. The results also
a
Values determined by dynamic headspace GC unless otherwise suggested that the larger pores became larger and the
noted. smaller pores became smaller. A model based on more
b
Juice reconstituted to the Brix value of the initial juice prior to effective intrusion of liquid water into the larger pores
analysis. and a resulting lateral displacement of the fibrils that
c
Values determined by static headspace GC with a polar column.
d formed the boundaries with neighbouring, smaller pores
Ethyl butanoate also present.
e
b-Pinene also present. was invoked to explain the latter observation. Celgard
f
Determined by the Scott oil method (Scott & Veldhuis, 1966). 2500 membrane was also subjected to contact with 30%
w/w CaCl2 solution but no significant morphology
change was observed in this medium after 72 h. The
membranes, as thin as 12.5 lm, would probably be absence of a morphology change was attributed to a
suitable for this purpose. The hydrogel-film-side of the lower degree of intrusion of the CaCl2 solution relative
laminate should be in contact with the feed liquid or to that of water due to the higher surface tension of the
solution to be concentrated. former.
Mansouri and Fane (1999) developed modified hy-
drophobic membranes that are tolerant to oil feeds (e.g. 5.5. Effect of UF pretreatment on the evaporation flux
limonene solution) by coating the feed side of the
commercial membranes (including the Celgard 2500, UF of single strength Gordo grape juice using
Millipore GVSP and the UPVP) with a thin layer of membranes with nominal pore diameters of 0.1 lm or
polyvinylalcohol (PVA). The tolerance of laminates was less (Osmonics Sepra CF flatsheet unit, membrane area
evaluated against oil containing feeds in short and long 0.0155 m2 ) result in appreciable osmotic distillation flux
term experiments. For concentrations up to 1 wt.% increases over that observed for juice not subjected to
limonene solution laminate membranes remained un- the UF (Bailey, Barbe, Hogan, Johnson, & Sheng,
changed when the experiment continued for up to 24 h. 2000). This has been attributed to a reduction in the
In contrast, membranes without the hydrogel coating viscosity of the concentrated juice-membrane boundary
were wetted out very rapidly by the oily feed. Due to layer where the solute concentration is highest and the
their high water content and the small thickness of the effect of protein removal is expected to be most pro-
hydrogel, the water permeabilities for laminate mem- nounced. In addition to the advantage of a higher OD
branes in most cases were not significantly lower than flux in permeate concentration, the fermentable sugars
the ones for uncoated substrate. This means that lami- content of standard 68Brix concentrate can be achieved
nates with comparable permeability and superior resis- at a significantly lower Brix value, thereby providing a
tance to organic solutions should be fabricated. possible means of reducing the handling of highly vis-
Recently, it has been found that an amorphous co- cous streams. UF pretreatment results in a small in-
polymer of a certain perfluorinated dioxole monomer, crease in juice surface tension with a consequent
B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324 315

reduction in the tendency for membrane wet-out to


occur.

5.6. Process design

The most important design parameters of an OD


process for concentration of an aqueous feed are: (1) the
required plant capacity in daily volume of feed to be
concentrated; (2) the solute concentrations in the feed
and final concentrate; (3) the water vapour pressure/
concentration relationship for the feed stream; (4) the
water vapour pressure/concentration relationship for the
strip solution and (5) the intrinsic water vapour per-
meability of the OD membrane (Hogan et al., 1998). A
schematic for a simple juice-concentration unit is shown
in Fig. 4.
The process employs partial batch recycle on the feed
side and continuous countercurrent recycle plus evapo-
rative reconcentration of the brine strip. The contactors
can be arranged in a series, parallel, or series-parallel
array.
Recently, Nii, Jebson, and Cussler (2002) have built
and tested a flat-sheet membrane evaporator, made of
flat sheet of thin, non-porous membranes (supplied by
GKSS, Max-Plank Strasse, D-21502 Geesthact, Ger-
many) for removing water from dilute feed streams like
milk and orange juice. In the membrane evaporator, Fig. 5. Schematic diagram of (a) the membrane evaporator and (b) the
shown schematically in Fig. 5a, the membrane is flow channel (Nii et al., 2002).
mounted between two 0.2 cm gaskets which form flow
channels for liquid and vapour. Each 0.2 cm gasket
other forms of ‘‘membrane distillation’’. Such a mem-
defines a flow channel. In some cases, the flow channel
brane evaporator may offer advantages to conventional
was straight; in others, it had the tortuous shape shown
evaporator, including larger surface area per volume
in Fig. 5b. The top of the vapour channel is solid; the
and high selectivity. The membrane evaporator retains
bottom of the liquid channel is a piece of metal foil.
flavors effectively. Because it has an overall vapor phase
Steam condensing on the other side of the metal foil
mass transfer coefficient of about 1 cm/s, it is only 68%
supplies energy to evaporate water through the mem-
efficient: only about 0.68 kg water is evaporated per kg
brane. This vapour flows out of the apparatus; if de-
steam condensed. This efficiency should be over 95% for
sired, its flow may be enhanced with an inert gaseous
a membrane which is 10 times more permeable. This
sweep stream. Therefore, the energy for the water’s
efficiency must be improved for this system to be prac-
evaporation comes from steam channels next to the feed
tical. It seems possible by using a thinner, water selective
channels, so that the operation differs sharply from
membrane (Nii et al., 2002).
Although OD has been studied mainly in the labo-
single strength
ratory at present, concentration of fruit juices by OD
juice diluted brine have been successfully demonstrated at pilot facilities in
reconcentrated condenser Melbourne and Mildura, Australia. The Melbourne
osmotic distillation
membrane array
brine plant contains two 19.2 m2 Liqui-Cel modules and it can
concentrate 50 l/h of juice to 65–70Brix. The Mildura
condensate
to waste plant, designed by Vineland Concentrates (a subsidiary
of the Wingara Wine Group) and Celgard LLC, con-
evaporator
tains 22 4 in. · 28 in. Liqui-Cel modules (total interfacial
area 425 m2 ). It operates at 30–35 C and 2 atm. Color,
flavor and aroma retention are good, but module
cleaning is an issue (Hogan et al., 1998).
juice concentrate
The potential and suitability of OD for concentrating
Fig. 4. Typical OD system for juice concentration (Hogan et al., 1998). fruit juices have been well proved. The flux losses
316 B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324

Table 7
Advantages and disadvantages of MD and OD applied to the concentration of fruit juices
Advantages Disadvantages
Technical aspects
• Techniques suitable for heat-sensitive products • New technology that requires an evaluation at industrial level,
• A 60% of dry matter can be reached in the concentrated product flexibility to be evaluated
• Modularity, easy scale-up • Low evaporative capacity (3 l/m2 h) with a long time of treatment
• Possibility to treat solutions with high level of suspended solids • Necessity of an inactivation enzyme pre-treatment
• Possibility of using modules in series, the same unit can
concentrate several different products
• Low temperatures
• Low operating pressures
• No fouling problems, constant permeate flux in time
• New technologies based on the use of conventional
well-tested materials
Economical aspects
• Low investment cost • Production costs higher than the thermal evaporation
• High cost of membrane replacement

attributable to viscous polarization are sufficiently low esters) in fruit juices may have lower retentions in the
to render the OD process both technically and eco- retentate as they tend to be associated with the serum
nomically feasible for industrial production of high de- (Rao, Acree, Cooley, & Ennis, 1987). Therefore, the
gree Brix juice concentrate. A serious problem benefits including high quality product and lower energy
associated with commercial application of OD in con- consumption could be achieved with integrated mem-
centrated fruit juice processing is the management of the brane processes.
diluted brine strip. It is essential to re-use the brine Lawhon and Lusas (1987) described the use of UF to
several times before it is removed from the process. separate juice into the pulp and serum prior to heat
Corrosion and scaling makes expensive the regeneration treatment and help to minimize flavor loss and deteri-
of exhausted brines that can be hardly accomplished by oration. The serum containing almost all the flavor and
conventional evaporators. In addition to heat evapora- aroma components is concentrated by RO to levels
tion, the effective methods need to be developed for the above 42Brix. The retentate containing all the sus-
concentration of the osmotic brine. Thompson (1991) pended solids, pectins and the spoilage microorganisms
reported solar ponding or even reverse osmosis and is quickly heated to inactivate the spoilage microor-
pervaporation could be used to re-concentrate the di- ganisms, and then recombined with RO concentrate.
luted stripper. Petrotos and Lazarides (2001) suggested The reconstituted juice is claimed to have a quality close
the use of electrodialysis as the most suitable way to the of fresh juice. Based on the Lawhon and Lusas patent
regeneration of the NaCl brine. (1987), Cross (1989) described a commercial design of
Table 7 summarizes the advantages and the disad- UF–RO process for membrane concentration of orange
vantages of the MD and OD applications in fruit juices juice. After, it has been reported (Walker, 1990) that
production. Separasystems LP, a joint venture between FMC and
Du Pont, has developed a combined membrane process,
which is called FreshNote system, that could concen-
6. Integrated membrane processes trate orange juice to 60Brix while almost retaining the
fresh juice flavors. The process involves a plurality of
The potential for concentrating fruit juice by inte- two-stage RO system after MF or UF process, the first
grated membrane processes, particularly for industrial stage employing high-rejection membranes and the sec-
production of high quality concentrated juices, appears ond employing low-rejection membranes (Fig. 6). The
today very attractive. Because fruit juices such as orange orange juice is initially clarified using MF (i.e., 0.2 lm)
juice have high solids and pectin content, they create a and/or UF (i.e., 20 kDa). The permeate is fed to a high-
very viscous stream when directly submitted to con- rejection RO stage (i.e., three units in series, 10.2–13.5
centration by RO or OD, which results in a lower per- MPa, 10 C) equipped with polyamide hollow fiber
meate flux. Also, using a single-stage RO system it membranes having 98.5% salt rejection. The retentate
cannot reach concentrations larger than 25–30Brix due leaving the final high-rejection RO unit is fed to low-
to higher osmotic pressure limitation. If RO combines rejection RO. For example, there can be two low-rejec-
MF or UF being used for separating the suspended tion units in series, the first membrane having a 93% and
solids and pectins from juices, the viscosity decreases the second with a 97% salt rejection. The retentate of the
and the flux increases. Volatiles (alcohols, aldehydes, first having a sugar concentration of 50Brix is the feed
B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324 317

system can be obtained when compared with evapora-


UF HR RO LR RO tion and freeze concentration, respectively.
The theoretical basis for these processes mentioned
juice
above relies on the understanding that the effective os-
motic pressure to overcome is not equal to the osmotic
pressure of the retentate, but is rather determined by the
concentration difference existing across the membrane
(Girard & Fukumoto, 2000). The use of low-retention
Pasteurizer Concentrate membranes allows to reach high retentate concentration
values because part of the solute passes the membrane,
Fig. 6. Schematic diagram of the combined high-retention (HR)–low- thus reducing the effective osmotic pressure difference.
retention (LR) RO process for juice concentration (Walker, 1990). Low retention means that a large amount of solute is
lost in the permeate. The solute recovery can be ac-
complished by recycling the permeate stream to a high-
of the second unit. The permeate from the first, having a retention module, which also acts as a feed preconcen-
sugar concentration of 18Brix, and that of the second, trator. The key variable is the membrane retention, re-
with a sugar concentration of 40Brix, can be combined lated to the solute/solvent permeability ratio. Hollow
to form a 31Brix permeate which is recycled to the feed fibers are better suited than spiral wound modules for
of the first high-rejection unit. The retentate from the this process. Hollow fibers have lower concentration
second low-rejection unit having a sugar concentration polarization effects and lower fluxes are counterbalanced
of 63Brix is mixed with the pasteurized UF retentate to by larger packing density (Walker, 1990). Recovery ra-
make a blended 54Brix product. Finisher pulp, peel oil tios as large as 15–20% can be achieved for product
and/or essence can be added to produce a concentrated concentration of 55–60Brix. The membrane area nee-
product tailored to particular market needs. The re- ded to concentrate juices to 60Brix with a hollow fiber
constituted juice consistently scored higher in flavor system is about twice the area needed to concentrate up
than any thermally concentrated product, and is nearly to 20Brix with a conventional RO system, while the
indistinguishable from the freeze-concentrated and the energy requirement is approximately three times larger
feed juice. Although production costs are higher for the (Gostoli, Bandini, Di Franscesca, & Zardi, 1995).
product compared with the current blended concen- However, an accompanying disadvantage is the loss of
trates, this process is significantly cheaper than freeze important components of the juice. To overcome this
concentration (Cross, 1989). Also, this system has been problem, a fairly complex processing scheme must be
installed and operated in Japan in 1989 (Walker, 1990; adopted (Watanabe, Nabetani, & Nakajima, 1990).
Walker & Ferguson, 1990). This plant is designed to A combined membrane-evaporation process was ap-
process 7.5 m3 /h of single strength Satsuma mandarin plied to develop an improved method for concentrating
juice, yielding about 2 m3 /h of concentrate. orange juice (Johnson, 1993). An UF membrane was
This concept has also been implemented commer- used to separate the juice into pulpy suspended solids
cially for apple juice concentrate (Mans, 1992). Apple (retentate) and clarified juice (permeate). The mem-
juice is first submitted to centrifuges and then to tubular brane-clarified clear juice was then concentrated by a
polymeric membrane (0.2 lm) for complete clarification. conventional evaporative method using a TASTE
The system contained 51 RO hollow fiber modules (thermally accelerated short time evaporator) evapora-
(4-foot-long) piped in up to eight stages comprising tor. The pulpy retentate may be pasteurized by a heat
high-rejection and low-rejection units. The 58–60Brix exchanger. The clear juice concentrate from the com-
concentrate is pumped through a scraped surface heat bined membrane-evaporation technique can reach a
exchanger where it is cooled from 21 to 2 C followed by higher degree Brix (>80Brix) than traditional whole
bulk cold storage and a final aseptic packaging opera- juice concentrate processed in a TASTE evaporator
tion. Nabetani (1996) developed an integrated RO–NF under similar conditions (Hernandez, Chen, Johnson, &
membrane system for highly concentrated fruit juice. Carter, 1995). Higher degree Brix orange juice concen-
The feed juice is firstly concentrated to 30Brix with RO trates offer significant energy savings during storage and
membranes and the RO retentate is then concentrated to distribution due to reduced volume and increased mi-
above 45Brix using NF membranes. The NF permeate crobial stability at higher temperatures (Crandall &
is recycled to the feed of RO unit. This system is shown Graumlich, 1982). The resulting clear juice concentrated
to be suitable for concentrating various fruit juices, with can also be further utilized to formulate new fruit bev-
advantages of not only retaining the fresh juice flavor, erage juice blends. For example, it is ready to pour to
but also of minimizing energy consumption. For ex- yield six-fold concentrate at freezer temperatures about
ample, using this system to concentrate fruit juice from )15 C. In contrast, the traditional three-fold frozen
10 to 45Brix, energy savings of 1/8 and 1/5 by this juice concentrate becomes frozen at common domestic
318 B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324

freezer temperatures of )5 to 15 C and it is inconve- Fresh juice


nient to pour and to mix with water (Chen, 1998).
Because clear orange juice concentrate produced by Retentate and
20˚C Enzyme recirculation
the combined membrane-evaporation technique is a new EMR
type of product and this highly concentrated (>80Brix) Clarified juice
clear orange juice resembles honey or sugar syrups,
which are usually stored and marketed at room tem- water 20-25˚C
perature, it is desired to evaluate the storage stability of RO Preconcentrated
this product at temperatures above freezing. This could Juice, 25˚Brix
feasibly replace conventional frozen storage of orange
juice concentrates. A preliminary study (Lee, Johnson, PV
Barros, & Chen, 1994) with this high-density clear or- 20˚C

ange juice concentrate of 80Brix indicated that major Dearomatized juice


changes in clear orange juice concentrate during storage Aroma
were due to chemical reactions rather than microbial Evaporation compounds
60-80˚C
deterioration. Recently, first-order rate constants for
non-enzymic browning and ascorbic acid loss of mem-
brane-clarified clear juice during conventional evapora- Concentrated Juice
(70-72˚ Brix)
tive processes at 70.3–97.6 C have been presented
(Johnson, Braddock, & Chen, 1995). Lee and Chen
(1998) studied rates of vitamin C loss and discoloration
in clear orange juice concentrate (80.2Brix) produced Aroma enriched
by the combined membrane-evaporation technique at concentrated juice
storage temperatures of 4, 14, and 24 C for 19 weeks.
Absorption at 420 nm and CIE L , a , b color para- Fig. 7. Integrated membrane process for the production of clarified

apple juice concentrate and apple juice aroma (Alvarez et al., 2000).
meters were determined to evaluate color changes due to
EMR ¼ enzymatic membrane reactor; RO ¼ reverse osmosis;
non-enzymic browning during storage. Calculated val- PV ¼ pervaporation.
ues of reaction rate constants indicated that vitamin C
degradation occurred slowly in clear orange juice con-
centrate at 4 C. Rates of vitamin C degradation ranged were tested in laboratory and pilot plant units. Pro-
from 4.79 · 104 to 3.13 · 102 mg/100 ml per week. The mising results were obtained with the membrane oper-
activation energy for vitamin C degradation was on the ations involved. It has been shown that a membrane
order of 34.3 kcal/g mol. Main color changes in stored reactor could be used for the production of clarified
clear orange juice concentrate in amber glass vials were apple juice with lower enzyme cost. The use of immo-
due to increases in b value and chroma. bilized enzyme resulted in better performance with
The coupling of UF and evaporation for concen- respect to fouling and cleaning. Permeate flow of 75–110
trating passionfruit juice has also been tested by Yu and l/h and concentration value of 25.5–26.6Brix were ob-
Chiang (1986). Pretreatment of the juice prior to UF tained in RO unit. The flux and NaCl rejection did not
with pectinase, centrifugation, and pasteurization could decrease significantly after three weeks of continuous
improve the UF flux of 50%. Permeate of passionfruit operation with apple juice. Rejection of aroma com-
juice from UF had a higher evaporation rate than juice pounds exceeded 90% for most compounds considered.
with solid due to improved heat transfer. The evapo- High overall enrichment factors in the range 100–1000
rated concentrate could be reblended with UF retentate and overall mass transfer coefficients in the range 5–500
to form a final concentrated product. Although the kg/m2 h were achieved in PV process using POMS-PEI,
process resulted in 20% flavor losses, the reconstituted POMS-PVDF and PDMS-PT1100 membranes. Fouling
juice and the fresh juice were not significantly different in did not give rise to any detectable problems and cleaning
overall acceptance. did not significantly affect the performance of the
An integrated membrane process for producing apple membranes. Organoleptic evaluation of the clarified and
juice and apple juice aroma concentrates has been pro- RO concentrate was excellent in terms of odour and

posed by Alvarez et al. (2000). The process scheme is flavor. The final products were more clear and brilliant
shown in Fig. 7. It involves the following operations: an than apple juice produced by conventional methods.
integrated enzyme membrane reactor to clarify the raw In order to have an economic process assessment, the
juice; reverse osmosis (RO) to preconcentrate the juice pilot plant units were assembled into an integrated unit
up to 25Brix; pervaporation (PV) to recover and con- and operated with raw apple juice. Process design basis
centrate aroma compounds; a final evaporation step to was to produce 2500 kg of apple juice per hour. Taking
concentrate apple juice up to 72Brix. These operations into account the periods when apples are available, 3500
B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324 319

Table 8

Capital investment and manufacturing costs for the conventional and integrated membrane process (Alvarez et al., 2000)
Item Cost (EURO)
Conventional process Membrane process
Investment
Apple storage 1.7 · 105 1.7 · 105
Washing and inspecting apples 0.1 · 105 0.1 · 105
Milling apples 0.4 · 105 0.1 · 105
Clarification 6.4 · 105 3.9 · 105
Preconcentration (RO) 1.3 · 105
Aroma compounds recovery and apple juice concentration 4.2 · 105 5.1 · 105
Total 13.73 · 105 12.48 · 105
Total capital investment 47.72 · 105 43.38 · 105
Operating costs
Raw material (apples) 810.0 EURO/ton conc. 810.0 EURO/ton conc.
Membranes 19.7 EURO/ton conc.
Others 99.9 EURO/ton conc. 50.2 EURO/ton conc.
Total variable cost 909.9 EURO/ton conc. 839.9 EURO/ton conc.
Labor, maintenance and other fixed costs 151.9 EURO/ton conc. 145.0 EURO/ton conc.
Total manufacturing costs 1068.8 EURO/ton conc. 984.9 EURO/ton conc.

tons of apple juice concentrate would be produced an- scheme is shown in Fig. 8. By allowing brine to flow
nually in the plant. As seen from Table 8, the total through the permeate side of the membrane in a
capital investment of the new integrated membrane counter-current mode to feed, the corresponding os-
process accounts for 14% less than that of the conven- motic pressure difference can be minimized or even re-
tional process. As compared with the conventional versed. High concentration of the feed could be achieved
process, apple consumption decreased by 5% when using at moderate operating pressures that keep high enough
the integrated membrane process, due to the corre- transmembrane permeate fluxes. Membrane areas to be
sponding increases in process yield. Total manufactur- required for high solute concentration could be reduced
ing costs decreased by 8%, mainly because less energy is
required to concentrate the juice. The profitability
analysis for the conventional and integrated membrane
Concentrated Concentrated
process was shown in Table 9. The integrated membrane
sugar out salt in
process also seemed to be more advantageous on the
basis of economics than the conventional one.
Coupling RO and osmotic dehydration process for
concentrating sucrose solution has recently been pro- Enricher
posed by Karode, Kulkarni, and Ghorapade (2000), in
which the feasibility of such a technique was investi-
(a) Lean sugar in Lean salt out
gated with both theory and experiments. The process

Table 9 Concentrated
Profitability analysis for the conventional and integrated membrane sugar out

processes (Alvarez et al., 2000)
Conventional Membrane Enricher
process
(1) Revenue (EURO/year) 5.469 · 106 5.469 · 106
Lean sugar in Concentrated
(2) Annual manufacture cost 3.716 · 106 3.432 · 106 Lean salt salt
(EURO/year) Water
(3) Gross profit (a–b) (EURO/ 1.753 · 106 2.037 · 106
year)
(4) Taxes (33%) (EURO/year) 0.758 · 106 0.672 · 106 Stripper
(5) Net profit (EURO/year) 1.175 · 106 1.365 · 106
(6) Gross margin (c/a) 0.32 0.37
(7) Return on investmenta (%) 24.6 33 (b)
(8) Payback timeb (years) 4.1 3.0
Fig. 8. Osmotic dehydration coupled reverse osmosis concentration
a
Means net profit/total investment  102 . process for concentrating fruit juice (a) enricher alone and (b) enricher-
b
Means total investment/net profit  102 . stripper configuration (Karode, Kulkarni, & Ghorapade, 2001).
320 B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324

by the process. They concluded that the process would centrated to 20–25Brix by RO, and the RO retentates
be limited by permeate-side concentration polarization are concentrated to above 60Brix by OD using the
and that to reduce contamination of the sucrose solution Liqui-Cel Extra-Flow module and Celgard micropor-
being concentrated, a very high rejection RO membrane ous PP hollow fibers. The proposed process employs
would be required. The same authors further investi- continuous countercurrent recycle plus evaporative
gated the concentration of sucrose solutions from 5 to reconcentration of the brine. The UF retentate could be
60Brix by coupled RO and osmotic dehydration with a submitted to a pasteurization process and then added to
hydrophobic high-rejection RO membrane in the en- the final OD concentrate.
richer and a high-rejection, thin-film composite RO The stability of the antioxidant activity (TAA) during
membrane in the stripper. A grapher method for esti- juice processing is of interest since they play an impor-
mating operating pressure limits in the enricher and the tant role in reducing the risk of free radical related ox-
stripper was discussed. The results showed that the type idative damage associated with a number of diseases.
of membrane used in the enricher, the operating pres- Changes in ascorbic acid, phenolics and total antioxi-
sure of the enricher and the stripper, the ratio of the dant activity during integrated membrane processing of
salt-solution elution, and the flow rate of the sucrose blood orange juice were evaluated (Galaverna et al.,
solution into the enricher affect the optimum salt con- 2002). The results showed that the content of these an-
centration that should be fed into the enricher to mini- tioxidant compounds besides anthocyanins and Vitamin
mize the total membrane area. For a specific set of C remained almost the same of the feed juice during the
operating pressures, the optimum salt concentration proposed integrated membrane process (Table 10). As
reduced as the ratio of flow rates increases. A configu- compared with the fresh blood orange juice, the contents
ration has proposed that would enable the concentra- of anthocyanins and Vitamin C decreased by 15–20% in
tion of a sucrose solution from 5% to 60% using a the concentrate. Only 15% reduction of TAA was found
natural source of brine (seawater, 3.5% salt) under a in the blood orange juice concentrate. The concentrate
limited maximum operating pressure of 54.4 bar (800 retains its bright red color and its pleasant flavor.
psi). The proposed configuration may reduce operating Successful application of integrated membrane oper-
and installation costs due to the management of flow ations in fruit juice concentration has been developed by
rate and pressure requirements. The seawater can be the Australian company (Wingara Wine Group, Mel-
eventually discharged back to the sea. Therefore, the bourne). A hybrid pilot plant where UF/RO and OD are
integrated RO and osmotic dehydration system can be integrated has been realized. It consists of UF and RO
used to concentrate fruit juices. pretreatment stages, an OD unit and a single-stage brine
The coupling of UF to MD has also examined for evaporator. A first-stage UF treatment for removal of
concentration of kiwifruit and orange juice (Jiao, Ca- suspended solids and colloids is also desirable to mini-
labro et al., 1992). More recently, Cassano et al. (2002, mize possible fouling of either RO or OD units. This
2003) proposed an integrated membrane processes for plant concentrates fresh juices up to 65–70Brix and has
the production of high quality fruit juice concentrate a capacity of 50 l/h. Being athermal, OD allows con-
such as blood orange and kiwifruit juice. As shown in centrating without product deterioration or loss of fla-
Fig. 9, UF process firstly separates fruit juices into the vors (Hogan et al., 1998).
serum and the pulp. The clarified juices are pre-con- Hogan et al. (1998) reported a total process cost of
OD concentration on the order of $1.00/l of concentrate.
From one liter of fresh juice, it is possible to achieve
pulp pasteurizer
about 200 ml of 70Brix concentrate. The value of this
concentrate is between $2.50 and $7.50/l. From these
fruit data, the economical benefits of the integrated mem-
juice
UF RO
water brane process seem evident.
(10-11 ˚Brix)
Table 11 summarizes the main advantages and dis-
evaporator advantages of membrane concentration techniques dis-
preconcentrated diluted concentrated
cussed with respect to the evaporation process. The low
juice
(25-26 ˚Brix)
brine brine (CaCl2 ) relatively permeation fluxes attainable with direct os-
mosis, OD and MD (when operating at low tempera-
tures) with respect to those achievable with the RO
OD process would appear to make these processes eco-
nomically uncompetitive. When, however, the product
concentrated
juice to be concentrated is a solution containing macrosolutes
(63-65 ˚Brix)
or colloids sensitive to shear or thermal degradation
Fig. 9. Integrated membrane process for the production of concen- there are serious limitations to the degree of concen-
trated fruit juices (Cassano et al., 2003). tration achievable via RO without significant product
B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324 321

Table 10
Contents of ascorbic acid, flavonoids, hydroxycinnamic acids and TAA in blood orange juice as affected by membrane processes (Galaverna et al.,
2002)
Samples Total Ascorbic Total Hesperi- Narirutin Sinapic Caffeic Ferulic p-coumaric TAA
soluble acid (mg/l) anthocya- din (mg/l) (mg/l) acid (mg/l) acid (mg/l) acid (mg/l) acid (mg/l) (mMtrolox)
solids nins (mg/l)
(Brix)
Fresh juice 12.6 701.0 ± 2.3 56.3 ± 0.2 45.1 ± 0.1 50.7 ± 0.1 6.6 ± 0.1 6.8 ± 0.1 51.3 ± 0.1 33.5 ± 0.1 8.65
UF 12.4 636.5 ± 3.1 55.0 ± 0.2 45.5 ± 0.1 50.8 ± 0.1 6.6 ± 0.1 6.8 ± 0.1 51.7 ± 0.1 34.9 ± 0.1 8.21
permeate
UF 13.0 624.9 ± 2.5 55.3 ± 0.2 42.1 ± 0.1 49.9 ± 0.1 6.1 ± 0.1 7.0 ± 0.1 53.9 ± 0.1 33.9 ± 0.1 8.29
retentate
RO 21.4 610.6 ± 2.0 44.7 ± 0.2 46.6 ± 0.1 50.2 ± 0.1 6.0 ± 0.1 7.4 ± 0.1 51.1 ± 0.1 34.3 ± 0.1 7.47
concentrate
OD 60.6 594.2 ± 3.1 43.3 ± 0.2 46.3 ± 0.1 48.7 ± 0.1 5.6 ± 0.1 7.6 ± 0.1 51.0 ± 0.1 33.5 ± 0.1 7.33
concentrate

Table 11
Key factors of conventional evaporation and membrane concentration techniques
Process Maximum Product Evaporation Possibility of Operating Capital Energy Maturity of
achievable quality rate or flux treating different cost investment consumption technology
concentration products with
(Brix) the same
installation
Evaporation 80 Very poor 200–300 l/h No Moderate Moderate Very high Developed
Reverse 25–30 Very good 5–10 l/m2 h No High High High Developed
osmosis
Direct 50 Good 1–5 l/m2 h Yes High High Low Developing
osmosis
Membrane 60–70 Good 1–10 l/m2 h Yes High Moderate Low Developing
distillation
Osmotic 60–70 Very good 1–3 l/m2 h Yes High Moderate Low Developing
distillation

deterioration or costly process equipment. For these 7. Conclusions


applications OD has important advantages.
OD is inherently more costly for concentration of The potential advantages of membrane concentration
aqueous mixtures than such conventional process as techniques over conventional evaporation for concen-
evaporation or RO. However its capacity to remove trating fruit juice have been successfully demonstrated,
selectively water from other low molecular weight sol- including improved product quality, easily scaled up and
utes and, thus, to yield concentrates of superior quality lower energy consumption, but they are generally limited
is an additional advantage of the OD process. On the by the fouling and lack of longer durability of mem-
other hand, RO, where it can be carried out at relatively branes. Although today fruit juice concentration by
low pressure, is the cheapest system for water removal membranes may be more expensive than evaporation,
for feeds of low solute concentration. Losses in volatile with the enlargement of the world’s fruit juice market
and non-volatile flavours, fragrances and other essential and the request of product quality, commercial appli-
microsolutes can be minimized by limiting the fraction cations of membrane processes in concentrated fruit
of water removed. Acceptable concentrates can be pro- juice processing, especially integrated membrane system,
duced economically in this manner. Therefore, an inte- will expand in the near future. However, in order to gain
grated process involving preconcentration of the feed by a foothold in the juice industry, studies on developments
RO followed by further concentration by OD or MD of new membranes which are both highly selective and
should yield a high-solids product concentrate of quality permeable, or robust and stable in long-term application
comparable to that achieved by OD alone but at sig- for juice processing, and improvements of process engi-
nificant reduction in processing cost (Hogan et al., neering including module design and process design and
1998). Direct osmosis hydrophilic membranes are optimization need to be carried out in detail. It can be
cheaper than MD and OD membranes; they have a anticipated that the use of membrane processes will bring
longer life cycle if compared with hydrophobic mem- great changes in the juice industry in the future, with the
branes used in MD and OD. development of membrane science and technology.
322 B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324

References Chou, F., Wiley, R. C., & Schlimme, D. V. (1991). Reverse osmosis
and flavor retention in apple juice concentration. Journal of Food
Ali, F., Dornier, M., Duquenoy, A., & Reynes, M. (2002). Transfer of Science, 56(2), 484–487.
volatiles through PTFE membrane during osmotic distillation. In Chua, H. T., Rao, M. A., Acree, T. E., & Cunningham, D. G. (1987).
Proceedings of the 2002 international congress on membrane and Reverse osmosis concentration of apple juice: flux and flavor
membrane processes, Toulouse, France. retention by cellulose acetate and polyamide membranes. Journal

Alvarez, 
V., Alvarez, 
S., Riera, F. A., & Alvarez, R. (1997). Permeate of Food Process Engineering, 9(3), 231–245.
flux prediction in apple juice concentration by reverse osmosis. Courel, M., Dornier, M., Herry, J. M., Rios, G. M., & Reynes, M.
Journal of Membrane Science, 127(1), 25–34. (2000). Effect of operating conditions on water transport during the

Alvarez, 
S., Riera, F. A., Alvarez, R., & Coca, J. (1998). Permeation of concentration of sucrose solutions by osmotic distillation. Journal
apple aroma compounds in reverse osmosis. Separation Purification of Membrane Science, 170(2), 281–289.
Technology, 14(1–3), 209–220. Crandall, P. G., & Graumlich, T. R. (1982). Storage stability and

Alvarez, 
S., Riera, F. A., Alvarez, R., Coca, J., Cuperus, F. P., Bouwer, quality of high brix orange concentrate. Proceedings of Florida
S., Boswinkel, G., van Gemert, R. W., Veldsink, J. W., Giorno, L., State Horticultural Society, 95, 198–210.
Donato, L., Todisco, S., Drioli, E., Olsson, J., Tr€agardh, G., Gaeta, Cross, S. (1989). Membrane concentration of orange juice. Proceedings
S. N., & Panyor, L. (2000). A new integrated membrane process for of Florida State Horticultural Society, 102, 146–152.
producing clarified apple juice and apple juice aroma concentrate. Cuperus, F. P. (1998). Membrane processes in agro-food: state of the
Journal of Food Engineering, 46(2), 109–125. art and new opportunities. Separation Purification Technology,

Alvarez, S., Riera, F. A., Alvarez, R., & Coca, J. (2001). Prediction of 14(1–3), 233–239.
flux and aroma compounds rejection in a reverse osmosis concen- Curcio, E., Barbieri, G., & Drioli, E. (2000). Operazioni di distillazione
tration of apple juice model solutions. Industrial Engineering a membrana nella concentrazione dei succhi di frutta. Industrie
Chemistry Research, 40(22), 4925–4934. delle Bevande, XXIX(aprile), 113–121.
Bailey, A. F. G., Barbe, A. M., Hogan, P. A., Johnson, R. A., & Deblay, P. (1995). Process for at least partial dehydration of an
Sheng, J. (2000). The effect of ultrafiltration on the subsequent aqueous composition and devices for implementing the process. US
concentration of grape juice by osmotic distillation. Journal of Patent 5,382,365, 26 January.
Membrane Science, 164(1–2), 195–204. Drioli, E., Calabro, V., & Wu, Y. (1987). Membrane distillation in the
Bandini, S., Gostoli, C., & Sarti, G. C. (1991). Role of mass and heat treatment of aqueous solutions. Journal of Membrane Science,
transfer in membrane distillation process. Desalination, 81, 91–106. 33(3), 277–284.
Barbe, A. M., Bartley, J. P., Jacobs, A. L., & Johnson, R. A. (1998). Drioli, E., Jiao, B., & Calabr
o, V. (1992). The preliminary study on the
Retention of volatile organic flavour/fragrance components in the concentration of orange juice by membrane distillation. In
concentration of liquid foods by osmotic distillation. Journal of Proceedings of VII international citrus congress, Acireale, Italy.
Membrane Science, 145(1), 67–75. Drioli, E., & Wu, Y. (1985). Membrane distillation: an experimental
Beaudry, E. G., & Lampi, K. A. (1990a). Membrane technology for study. Desalination, 53(1–3), 339–346.
direct osmosis concentration of fruit juices. Food Technology, 44(6), Durham, R. J., & Nguyen, M. H. (1994). Hydrophobic membrane
121. evaluation and cleaning for osmotic distillation of tomato puree.
Beaudry, E. G., & Lampi, K. A. (1990b). Osmosis concentration of Journal of Membrane Science, 87(1–2), 181–189.
fruit juices. Fluessiges Obst, 57(10), 652–656, 663–664. Franken, A. C. M., Mulder, M. H. V., & Smolders, C. A. (1990).
Bolin, H. R., & Salunke, D. K. (1971). Physicochemical and volatile Pervaporation process using a thermal gradient as the driving
flavor changes occurring in fruit juices during concentration and force. Journal of Membrane Science, 53, 127–141.
foam-mat drying. Journal of Food Science, 36(4), 665–668. Fukutani, K., & Ogawa, H. (1983a). A comparison of membrane’s
Bowden, R. P., & Isaacs, A. R. (1989). Concentration of pineapple suitability and effect of operating pressure for juice concentration
juice by reverse osmosis. Food Australia, 41(7), 850–851. by reverse osmosis. Nippon Shokuhin Kogyo Gakkaishi, 30(11),
Bowser, J. J. (2001). Osmotic distillation process. US Patent 6,299,777, 636–641.
9 October. Fukutani, K., & Ogawa, H. (1983b). Juice concentration by RO and
Braddock, R. J., Nikdel, S., & Nagy, S. (1988). Composition of some membrane properties relating to permeability of juice components.
organic and inorganic compounds in reverse osmosis-concentrated Nippon Shokuhin Kogyo Gakkaishi, 30(12), 709–715.
citrus juices. Journal of Food Science, 53(2), 508–512. Gadea, A. (1987). Reverse osmosis of orange juice. In Proceedings of
Calabr o, V., Jiao, B., & Drioli, E. (1994). Theoretical and experimental the international fruit juice congress, Orlando, USA.
study on membrane distillation in the concentration of orange Galaverna, G., Cassano, A., Di Silvestro, G., Sforza, S., Cagnasso, P.,
juice. Industrial Engineering Chemistry Research, 33(7), 1803– Drioli, E., & Marchelli, R. (2002). Variation of the antioxidant
1808. activity in fruit juices during membrane-based clarification and
Cassano, A., Jiao, B., & Drioli, E. (2002). Integrated membrane concentration processes. In Proceedings of the 1st workshop Italy–
process for production of concentrated kiwifruit juice. In Proceed- China, Cetraro, Cosenza, Italy.
ings of 1st workshop Italy–China on state of research and Girard, B., & Fukumoto, L. (2000). Membrane processing of fruit
applications of membrane operations for a sustainable growth. juices and beverages: a review. Critical Review in Biotechnology,
Cetraro (CS), Italy. 20(2), 109–175.
Cassano, A., Drioli, E., Galaverna, G., Marchelli, R., Di Silvestro, G., Gostoli, C., Bandini, S., Di Franscesca, R., & Zardi, G. (1995).
& Cagnasso, P. (2003). Clarification and concentration of citrus Concentrating fruit juices by reverse osmosis. The low retention-
and carrot juice by integrated membrane process. Journal of Food high retention method. Fruit Processing, 6, 417–421.
Engineering, 57, 153–163. Gostoli, C. (1998). Gas membrane extraction: a new technique for the
Chen, C. S., Shaw, P. E., & Parish, M. E. (1993). Orange and tangerine production of high quality juices. Fruit Processing, 9, 417–421.
juices. In S. Nagy, C. S. Chen, & P. E. Shaw (Eds.), Fruit juice Gostoli, C. (1999). Thermal effects in osmotic distillation. Journal of
processing technology (pp. 110–165). Auburndale, FL: AgScience. Membrane Science, 163(1), 75–91.
Chen, C. S. (1998). Methods for producing ready to pour frozen Hernandez, E., Chen, C. S., Johnson, J., & Carter, R. D. (1995).
concentrated clarified fruit juice, fruit juice produced therefrom, Viscosity changes in orange juice after ultrafiltration and evapo-
and high solids fruit product. US Patent 5,756,141, 26 May. ration. Journal of Food Engineering, 25(3), 387–396.
B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324 323

Herron, J. R., Beaudry, E. G., Jochums, C. E., & Medina, L. E. (1994). Medina, B. G., & Garcia, A. (1988). Concentration of orange juice by
Osmotic concentration apparatus and method for direct osmotic reverse osmosis. Journal Food of Process Engineering, 10(3), 217–
concentration of fruit juice. US Patent 5,281,430. January 25. 230.
Hogan, P. A., Canning, R. P., Peterson, P., Johnson, R. A., & Mengual, J. I., Zarate, J. M., Pena, L., & Velazquez, A. (1993).
Michaels, A. S. (1998). A new option: osmotic distillation. Osmotic distillation through porous hydrophobic membranes.
Chemical Engineering Progress, 7, 49–61. Journal of Membrane Science, 82(1–2), 129–140.
Jariel, O., Reynes, M., Courel, M., Durand, N., & Dornier, M. (1996). Merson, R. L., & Morgan, A. I. (1968). Juice concentration by reverse
Comparaison de quelques techniques de concentration des jus de osmosis. Food Technology, 22(5), 631–634.
fruits. Fruits, 51(6), 437–450. Merson, R. L., Paredes, G., & Hosaka, D. B. (1980). Concentrating
Jiao, B., Molinari, R., Calabr o, V., & Drioli, E. (1992). Application of fruit juices by reverse osmosis. In Ultrafiltration membranes and
membrane operations in concentrated citrus juice processing. Agro- applications (p. 405). New York: Plenum Press.
Industry Hi-tech, 3(1), 19–27. Michaels, A. S. (1998). Methods and apparatus for osmotic distilla-
Jiao, B., Calabr o, V., & Drioli, E. (1992). Concentration of orange and tion. US Patent 5,824,223, 20 October.
kiwifruit juice by integrated ultrafiltration and membrane distilla- Michaels, A. S. (1999). Osmotic distillation process using a membrane
tion. In Proceedings of the 1992 international membrane science laminate. US Patent 5,938,928, 17 August.
and technology congress, Sydney, Australia. Milleville, H. (1990). Direct osmotic concentrates juices at low
Johnson, R. A., Valks, R. H., & Lefebvre, M. S. (1989). Osmotic temperature. Food Processing, 51(1), 70–71.
distillation––a low temperature concentration technique. Austra- Nabetani, H. (1996). Development of a membrane system for highly
lian Journal of Biotechnology, 3(3), 206–207, 217. concentrated fruit juice. Journal of Membrane (Japanese), 21(2),
Johnson, J. R. (1993). Technical and economical feasibility of a 102–108.
nonconventional method for concentrating orange juice. Ph.D. Nii, S., Jebson, R. S., & Cussler, E. L. (2002). Membrane evaporators.
Thesis, University of Florida. Journal of Membrane Science, 201(1–2), 149–159.
Johnson, J. R., Braddock, R. J., & Chen, C. S. (1995). Kinetics of ascor- Palmieri, L., Dalla Rosa, M., Dall’Aglio, G., & Carpi, G. (1990). Dalla
bic acid loss and nonenzymatic browning in orange juice serum: ex- production of kiwifruit concentrate by reverse osmosis process.
perimental rate constants. Journal of Food Science, 60(3), 502–505. Acta Horticulturae, 282, 435–439.
Karode, S. K., Kulkarni, S. S., & Ghorapade, M. S. (2000). Osmotic Paulson, D. J., Wilson, R. L., & Spatz, D. D. (1985). Reverse osmosis
dehydration coupled reverse osmosis concentration: steady-state and ultrafiltration applied to the processing of fruit juice. In S.
model and assessment. Journal of Membrane Science, 164(1–2), Sourirajan, & T. Matsuura (Eds.), Reverse osmosis and Ultrafiltra-
277–288. tion (pp. 325–344). ACS Symposium Series: Washington, DC.
Karode, S. K., Kulkarni, S. S., & Ghorapade, M. S. (2001). Coupling Pepper, D. (1990). RO for improved products in the food and chemical
reverse osmosis and osmotic dehydration: further investigation. industries and water treatment. Desalination, 77(1), 55–71.
Separation Science and Technology, 36(4), 3091–3103. Pepper, D., Orahard, A. C. J., & Merry, A. J. (1985). Concentration of
Kimura, S., Nakao, S. I., & Shimatani, S. (1987). Transport tomato juice and other fruit juice by reverse osmosis. Desalination,
phenomena in membrane distillation. Journal of Membrane Sci- 53(1–3), 157–166.
ence, 33(3), 285–298. Peri, C., Battisti, P., & Setti, D. (1973). Solute transport and
K€oseoglu, S. S., Lawhon, J. T., & Lusas, E. W. (1990). Use of mem- permeability characteristics of reverse osmosis membranes. Leben-
branes in citrus juice processing. Food Technology, 44(12), 90–97. smittel-Wissenschaft und-Technologie, 6(4), 127–132.
Kunz, W., Benhabiles, A., & Ben-Aim, R. (1996). Osmotic evaporation Petrotos, K. B., Quantick, P. C., & Petropakis, H. (1998). A study of
through macroporous hydrophobic membranes: a survey of the direct osmotic concentration of tomato juice in tubular
current research and applications. Journal of Membrane Science, membrane module configuration. I. The effect of certain basic
121(1), 25–36. process parameters on the process performance. Journal of Mem-
Lagan a, F., Barbieri, G., & Drioli, E. (2000). Direct contact membrane brane Science, 150(1), 99–110.
distillation: modelling and concentration experiments. Journal of Petrotos, K. B., Quantick, P. C., & Petropakis, H. (1999). Direct
Membrane Science, 166(1), 1–11. osmotic concentration of tomato juice in tubular membrane
Lawhon, J. T., & Lusas, E. W. (1987). Method of producing sterile and module configuration. II. The effect of using clarified tomato juice
concentrated juices with improved flavor and reduced acid. US on the process performance. Journal of Membrane Science, 160(2),
Patent 4,643,902, 7 September. 171–177.
Lee, H. S., & Chen, C. S. (1998). Rates of vitamin C loss and Petrotos, K. B., & Lazarides, H. N. (2001). Osmotic concentration of
discoloration in clear orange juice concentrate during storage at liquid foods. Journal of Food Engineering, 49(2-3), 201–206.
temperatures of 4–24 C. Journal of Agricultural and Food Popper, K., Camirand, W. M., Nury, F., & Stanley, W. L. (1966).
Chemistry, 46(11), 4723–4727. Dialyzer concentrates beverages. Food Engineering, 38(4), 102–
Lee, H. S., Johnson, J., Barros, S., & Chen, C. S. (1994). Chemical 104.
changes and nonenzymic browning in 80Brix of concentrated Rao, M. A., Acree, T. E., Cooley, H. J., & Ennis, R. W. (1987).
orange serum at 5 and 24 C. In Proceedings of the 1994 annual Clarification of apple juice by hollow fiber ultrafiltration : fluxes
meeting of institute of food technologists, Chicago, USA. and the odor-active volatiles. Journal of Food Science, 52(2), 375–
Lefebvre, M. S. M. (1988). Method of performing osmotic distillation. 378.
US Patent 4,781,837, 1 November. Schofield, R. W., Fane, A. G., & Fell, C. J. D. (1987). Heat and mass
Lefebvre, M. S. M. (1992). Osmotic distillation process and semiper- transfer in membrane distillation. Journal of Membrane Science,
meable barriers therefore. US Patent 5,098,566, 24 March. 33(3), 299–313.
Mans, J. (1992). The champagne of apple juice concentrates. Prepared Schofield, R. W., Fane, A. G., & Fell, C. J. D. (1990). Gas and vapor
Foods, 161, 70. transport through microporous membranes. Journal of Membrane
Mansouri, J., & Fane, A. G. (1998). Membrane development for Science, 53(1), 159–171.
processing of oily feeds in IMD. In Proceedings of the workshop on Scott, W. C., & Veldhuis, M. K. (1966). Rapid estimation of
membrane distillaition, osmotic distillation and membrane cont- recoverable oil in citrus juices by bromate titration. Journal of the
actors, Cetraro (CS), Italy. Association of Official Analytical Chemists, 49, 628–633.
Mansouri, J., & Fane, A. G. (1999). Osmotic distillation of oily feeds. Shaw, P. E., Lebrun, M., Dornier, M., Ducamp, M. N., Courel, M., &
Journal of Membrane Science, 153(1), 103–120. Reynes, M. (2001). Evaluation of concentrated orange and
324 B. Jiao et al. / Journal of Food Engineering 63 (2004) 303–324

passionfruit juices prepared by osmotic evaporation. Lebensmittel- Walker, J. B. (1990). Membrane process for the production of superior
Wissen und-Technologie, 34(2), 60–65. quality fruit juice concentrate. In Proceedings of the 1990 interna-
Sheng, J., Johnson, R. A., & Lefebvre, M. S. (1991). Mass and heat tional congress on membrane and membrane processes, Chicago,
transfer mechanism in the osmotic distillation process. Desalina- USA.
tion, 80(2–3), 113–121. Walker, J. B., & Ferguson, R. R. (1990). Process to make Juice
Sheu, M. J., & Wiley, R. C. (1983). Preconcentration of apple juice by products with improved flavor. US Patent 4,933,197, 7 June.
reverse osmosis. Journal of Food Science, 48(2), 422–429. Watanabe, A., Nabetani, H., & Nakajima, M. (1990). Development of
Sheu, M. J., & Wiley, R. C. (1984). Influence of reverse osmosis on multi-stage RO combined system for high concentration of apple
sugar retention in apple juice concentration. Journal of Food juice. In Proceedings of the 1990 international congress on
Science, 49(1), 304–305. membrane and membrane processes, Chicago, USA.
Thompson, D. (1991). The application of osmotic distillation for the Wrolstad, R. E., McDaniel, M. R., Durst, R., Michaels, N., Lampi, K.
wine industry. The Australian Grapegrowers Winemakers, 11, 11– A., & Beaudry, E. G. (1993). Composition and sensory character-
14. ization of red raspberry juice concentrated by direct-osmosis or
Vaillant, F., Jeanton, E., Dornier, M., O’Brien, G. M., Reynes, M., & evaporation. Journal of Food Science, 58(3), 633–636.
Decloux, M. (2001). Concentration of passion fruit juice on an Yu, Z. R., & Chiang, B. H. (1986). Passionfruit juice concentration by
industrial pilot scale using osmotic evaporation. Journal of Food ultrafiltration and evaporation. Journal of Food Science, 51(6),
Engineering, 47(3), 195–202. 1501–1505.

You might also like