4*

EKATO. THE BOOK

The contributions on the different aspects of mixing

technology were written by Ekato specialists. They
combine the knowledge and know how gathered
by Ekato from research, development, application,
construction, manufacturing and marketing.

Although the utmost care has been taken when

compiling and checking this publication, the publisher
is not responsible or liable for errors or omissions.
We reserve the right to make technical changes.

Reprints, translations or reproductions of this

publication, wholly or in part and in any form, both
electronic and mechanical, is not permitted without prior
written agreement from the publisher; this excludes
Publisher :
short citations for review purposes.
Ekato Holding GmbH
Postfach 10 70
Illustrations, descriptions, schematic diagrams and D -79010 Freiburg ( Germany )
other data serve exclusively for explanatory purposes
and presentation of the text. They may not be used as info @ ekato. com
a template or as a confirmation of tests. The publisher
www.ekato.com
disclaims all responsibility and liability for compliance
of the content of this publication with local, national or
Editor: Werner Himmelsbach
international laws, rules or regulations. Editorial services: Dr. Matthias Delbruck, Dossenheim
(Germany)
Translation: itl Institut fur technische Literatur AG,
All rights reserved.
Munchen ( Germany ), Gillian Scheibelein
© 2012 Ekato Holding GmbH
Concept and design:

3rd Edition
.
qu-int | marken | medien | kommunikation, Freiburg
( Germany)
ISBN No. 978- 3 - 00 - 038660 -2
Printed by: Druckerei Hofmann, Emmendingen ( Germany )
 a*

EKATO. THE BOOK

i

The conventional basic operations used in chemical, ther- The contributions to this third edition were written by
mal and mechanical process engineering, to which mixing specialists from the different sectors of the EKATO GROUP.
and blending technology also belongs, continue to be We are very grateful for their exceptional commitment.
the backbone of the process industry. New technologies,
e.g. micro process engineering, will supplement and en - We hope that this new edition will be received with the
hance them, but will certainly not substitute them. In fact, same enthusiasm as its two predecessors and that it will
quite the reverse: these " mature " basic operations are cur- contribute to the understanding of this important basic
rently experiencing a developmental spurt that is driven operation.
by industrial demands for greater profitability and resource
efficiency. This represents a challenge, not only for plant
operators and manufacturers of machines, apparatus and Schopfheim, June 2012
instruments, but also for academic research, which is once
again faced with a rewarding field to deepen our under-
standing of the physical and chemical processes taking
place inside mixing vessels. This applies to an even great-
er extent to future - oriented applications with new equip -
ment concepts, such as the manufacture of nanodisper-
sions, for which practical experience is still in its infancy. Dr Erich K. Todtenhaupt ' Dr Christian Watzelt
Furthermore, the underlying theoretical principles are still
not completely understood.

The third edition of the Handbook of Mixing Technology

first deals with the physical and rheological principles of
mixing technology and blending of solids as well as design
concepts for mixing and dispersing equipment. This is fol-
lowed by the mixing systems themselves, including descrip-
tions of the individual components, and progressing to the
full production plant. In addition to the classical design
methods, the discussions include methods based on com-
putational fluid dynamics and also numerical structural
analysis of all components. Topics covered rangefromthe
equipment concept to reliable and maintenance -friendly
operation. The third section gives examples of process
solutions that are being used in all sorts of different indus -
trial sectors. As a consequence, this Handbook of Mixing
Technology is aimed at students as well as process devel-
opers, plant engineers and production designers in the
process industry.

EKATO 3
PROGRESS
Nothing we do is exceptional; the reason we re so
successful is that we do quite ordinary things in an
extraordinary way.
— Ueli Praaer
 PERFECTION
EXCELLENCE \
Nowhere will you find nature practising . Her works are all masterpieces.
Johann Peter Hebei
 i

mm* gKf

SUSTAINABILITY
RESPONSIBILITY
Man is not a product of his environment — the environment is
a product of man.
Environmental protection is an opportunity, not a burden we
must bear.
— Helmut Sihler
 PROCESS ANALYSIS

RHEOLOGY P. 74 MIXING SOLIDS

DIMENSIONLESS NUMBERS P. 78 VACUUM CONTACT DRYING
HOMOGENIZING
SUSPENDING SOLIDS
DISPERSING GAS / LIQUID
DISPERSING LIQUID / LIQUID
HEATTRANSFER

EXPERIMENTAL METHODS
P. 92 NUMERICAL METHODS, CFD
P. 102 SCALING UP

P. 110 TRADITIONAL IMPELLERS

P. 114 HIGH - PERFORMANCE IMPELLERS
P. 120 SPECIAL DESIGNS
P. 126 WEAR PROTECTION
P. 130 PRINCIPLES OFAGITATOR DESIGN

\ \x P. 144 MECHANICAL SEALS

VNS^ — P. 168
P. 176
P. 182
NUMERICAL METHODS, FEA
DESIGN PRINCIPLES
IMPELLERS
P. 232
P. 248
P. 252
DETAIL ENGINEERING FOR SKIDS AND PLANTS
MATERIALS OF CONSTRUCTION
WELDING
P. 188 SHAFTS AND BEARINGS
P. 196 GEARBOXES
P. 202 ELECTRIC MOTORS, SPEED AND POWER CONTROL
P. 212 STIRRED TANK REACTORS AS FUNCTIONAL UNITS

S . 256 CONDITION - BASED MAINTENANCE

P. 266 WORLD - SCALE PLANTS P. 306 CRYSTALLIZATION

P. 272 POLYMERS P. 310 VISCOUS MEDIA
P. 280 HYDROMETALLURGY P. 312 OINTMENTS, CREAMS
P. 288 SPECIALTY CHEMICALS P. 318 NANOMATERIALS
P. 294 DRYING P. 324 SECOND GENERATION BIOETHANOL
P. 298 BIOTECHNOLOGY P. 328 LARGE VESSELS, STORAGE TANKS,
SIDE ENTRY AGITATORS

P. 333 AUTHORS
P. 334 NOMENCLATURE AND DIMENSIONLESS NUMBERS
P. 340 BIBLIOGRAPHY AND RELEVANT STANDARDS
P. 346 INDEX
 Only an integrated, case- by- case analysis of
the complex industrial requirements will provide
the optimum technical solution with regard to
investment and operating costs, energy and re -
source efficiency as well as environmental com-
The process industry is in a constant state of patibility. This integrated approach combines
change. Mass products in the chemical and met- a clear definition of the requirements, tried and
allurgical industries are produced in continuously tested experimental techniques, new numerical
operated " world- scale" plants with capacities methods and practical experience with respect
of more than a million tons per year. For such to design and scaling up to full - scale production.
huge tonnages, improving the process by a mere The correspondingly tailored components will
fraction of a percent can have a considerable eco - ensure that the equipment operates reliably and
nomic impact . Specialty chemicals have short profitably over its entire service life.
life cycles, and a fast-time -to - market provides
a decisive competitive advantage . A robust and
efficient mixing technology is essential to be able
to produce economically and profitably under
such boundary conditions. Individual substances
are finally combined into semi- finished goods or
consumer products. Manufacturing these prod-
ucts necessitates providing a high and consist-
ently reproducible quality. Thus the production
processes are extremely demanding and just as
varied as the products themselves.

The beginning of the 21st century marked a rad-

ical change: the spectrum of chemical intermedi-
ate products made from bio- based raw materials
is increasing daily, power stations fired with fossil
fuels are operated without CO2 emissions and the
generation, transport and storage of electrical en-
ergy from sustainable sources will revolutionise
the conventional energy sector. All these tech-
nologies require novel materials, catalysts and /
or storage media based on polymers or metals
or met l derivatives that require modern produc-
^
tion processes. Innovative process solutions will
ensure their successful implementation .
 PROCESS ANALYSIS PROCESS ANALYSIS
THE CUSTOMER
REQUIREMENTS AND ANALYSIS

- process description and workflow with respect to

formulation/ feeding, changes in the filling level and
material Process know-how
- material properties ( density, viscosity, etc.) - Process engineering,
- whether the design is for a target product or a range Basic operations
of products with different requirements ( mono -/multi- - Experience
- Process database
product systems) - Expert system
If the comprehensive problem analysis necessitates trials
Process plants can only operate profitably when all requirements and influencing factors have been or simulations, then the following must be carried out:
optimally tailored to each other. This also applies to the mixing processes and goes well beyond A
- definition of the primary mixing task Experimental Numerical methods:
designing the actual agitator. The basic engineering of the mixing vessel must account of the
investigations: - Computational
interaction between the mixing effect and all functional components in and on the vessel under - pre -screening for suitable mixing systems
- Model system fluid dynamics
the required operating conditions . Nowadays, the design of such mixing systems is based on pilot - specification of applicable scale - up/scale - down rules - Pilot plant/mini plant ( CFD)
trials, numerical calculations and expert knowledge . fortrials (see page 102) -Test at full-scale - Finite element
- the necessary numerical simulations, not only for the analysis ( FEA)

operating scale. A pilot- scale simulation also provides

a means to validate the numerical analysis.

REQUIREMENTS FOR MIXING SYSTEMS In the case of mixing technology, this means that stand- In order to fulfil the actual process function, the inte - Fig. 2 Instruments for designing mixing processes
ard solutions no longer suffice. On the one hand, impellers grated approach also requires that all system components
The customised design of any processing equipment have to be adapted and enhanced for individual basic are mutually optimised. These include: In addition to the pure functionality of these components,
starts with the specification of the desired goals and before mixing tasks or to meet the requirements of individual their strength and resistance to resonance must be checked
dimensioning the machinery and apparatus. In addition to reaction types or for the specific boundary conditions of - baffles because they are subjected to high hydraulic loads. It can
product quantities and quality, further factors that must be particular industries. These customised solutions have - heat exchangers thus be advantageous if the agitator manufacturer is re -
considered include the consumption of energy and materi- to be carefully tailored to the vessel geometry with all its - dip tubes sponsible for coordinating these interfaces, particularly if
als, operating aspects such as maintenance and operating functional fittings and also take account of the operating - inlet and outlet points the vessel and fittings do not conform to a standard.
costs of the equipment as well as the amount of cleaning requirements. This integrated approach applies to the
work in multi-product plants. These factors must be in- whole mixing system ratherthan to the impeller by itself.
cluded in the investment and operating costs because they For classical methods of manufacturing mass products,
greatly affect the economic viability of the plant and thus it also offers an opportunity to increase the economic
Requirements for the mixing process
Product characterisation Requirements for
ultimately the question of how profitable a product can be efficiency, particularly if economies of scale are to be to be vessels and components
placed on the market. exploited by increasing the size of the equipment. Com -
pletely new processes, for example those currently being
Homogenising Flow behaviour (Newtonian/ Reactor, preparation or Blend time, flow velocities, surface
used in hydrometallurgy, only become cost- effective by non-Newtonian), viscosity storage vessel entrainment, avoidance of air entrainment
- Impellers for individual basic Productivity using such an integrated approach. There are almost in -
mixing tasks Quality exhaustible possibilities in thefield of speciality chemicals Suspending Shape, size and size distribution Outlet position, Off-bottom or homogeneous suspension
- Customized solutions for an Energy (continuous process), start-up in the sediment
and formulated products because the interplay between of particles, solids concentration, wear protection
industrial sector or a particular Raw materials
liquid and solid densities
type of reaction Waste flow phenomena, the chemical and physical processing
- "Mixing systems", process-specific Wear steps and the apparatus provide a means to secure cost
matching of the shape and size Cleaning effectiveness. Dispersing Stoichiometric requirement or Self-aspirating or Course of the reaction and turnover, gassing
of vessels, agitators, and internal Maintenance gas excess of reactive gas, combined gassing, rate, superficial velocity, material transport
components Downtimes gas volume [p, T ), solubility, gas feed, coefficient , local/overall power input
kfa
ANALYSIS OFTHE REQUIREMENTS suspension of solid catalysts external gas circuit
Designing mixing systems during the basic engineering
phase requires compilation of the following basic informa - Dispersing Phase ratios, surface tensions, Avoidance of Localised specific power/turbulence,
tion, irrespective of the basic mixing tasks that determine liquids viscosities low-flow zones ' ' mixing/coagulation
the type of agitator:
- Equipment size ( economy of scale ) Costs, profit Heat transfer Viscosity, specific heat and Heat-exchange surface, Transferred heating power, heating and cooling
- Novel applications conductivity, risk of scaling, material and wall thickness time, local/global heat transfer coefficients,
- continuous or discontinuous operation
heating/cooling media of the vessel, heat-trans- surface temperatures on product side
- batch sizes or throughput rate/vessel dimensions ferring internal components
Fig. 1 This integrated approach to the development of agitator systems
increases the earnings potential. Fig. 3 Basic mixing tasks and influencing factors

22 23
 PROCESS ANALYSIS

Fig. 3 lists a few key characteristics and parameters de - classical rules of heat transfer calculation are used to design
pending on the type of basic mixing task. In practice, a com- heat exchangers such as vessel walls, tube bundles or coils. Process requirements,
bination of different mixing tasks is frequently required, But this requires knowledge of the agitated flow regimes plant concept
which leads to a more complex situation. Furthermore, the because they affect the inner heat transfer. The basic and
material properties under the actual process conditions
may be very difficult to measure and thus have to be esti -
detailed engineering steps that are required for designing
a fully functioning mixing system, e.g. a chemical reactor
I
Analysis of basic mixing task,
mated. For example, the viscosity of a reaction mixture or equipment to produce ointments, are described on Material properties apparatus concept, operating
under pressure if the atmospheric boiling point of the sol- page 212 and page 232. Recipes concept ( batch/semi-batch,
Mass/heat balances continuous )
vent has been exceeded and the viscosity cannot be meas -
ured with a standard rheometer. Experience from compa- The process design is followed by mechanical designing. 1
rable processes stored in expert systems may provide The latter starts with the selection of suitable materials of f }
support to the agitator manufacturer. construction on the basis of their resistance to corrosion Agitator design Basic eng. vessel/plant
and wear as well as their strength. -Type of impeller - Main dimensions/volume
In case of batch operations, the key influencing factors The hydraulic forces acting on the agitator, its natural Calculation - Number - Flow-directing internal components
must be considered with respect to how they develop frequencies and the resulting forces acting on the agitator Trials/CFD optional - Diameter, speed - Feed/discharge devices
- Power - Heat exchangers
over time. Many batch processes include a wide variety components, bearings and agitator supports as well as any - Peripheral devices/P&l diagram
of reaction stages and production steps such as precipi- tank internals such as baffles and heat exchangers are cal -
tation, crystallization, filtration or distillation, all of which culated on the basis of the process design (type of impeller,
interact and thus influence the mixing technology. diameter, speed, power ). These forces are then used to
dimension all components of the vessel and the agitator,
FROM DESIGN TO OPERATION particularly particularly the shaft size. For this design step,
Risk analysis/HAZOP
Designing an agitator generally starts with the above - it is imperative that the agitator is notoperated close to the
described process analysis (Fig. 4) . In the case of licensed critical speed ( page 130 ff.). If this is still not possible after
processes or mass - produced chemicals, either information altering the mechanical parameters ( e.g. different material f
from the licensor or experience, e.g. from a process data - of construction, use of a steady bearing, different masses), Dimensioning / agitator Detail eng. vessel/plant
Corrosion data - Materials of construction - Pressure vessel
base, can be used. However, if such information is not the process design must be adjusted correspondingly. This
Pressure, temperature - Drive, gearbox - Internal components
available, as is frequently the case for specialty chemicals, can be achieved, for example, by altering both the impeller Drive power/torque - Shaft size, bearing, - No resonance
individual or new processes or multi-purpose facilities,the diameter as well as the speed; however, such measures Hydraulic forces agitator flange - E/I&C
process requirements and material properties must be must not limit the suitability from the process engineering FE/modal analysis optional - Mechanical seal - Explosion protection
analysed to determine which basic mixing tasks will govern point of view. - Special requirements, finish
the design of the agitator. These basic mixing tasks will
then determine the appropriate type of impeller and the
number of stages. The impeller diameter, speed and pow-
Constructional details, e.g. a GMP- conformant design
that can be thoroughly cleaned and sterilized and which !
er can now be calculated on the basis of these process meets the respective requirements for surface finishes or Service/maintenance concept
requirements. getting the parts in and out of the vessel, must be specified
in advance because they affect the agitator construction I
If this information is insufficient to clearly identify the and thus the costs . Order processing, installation,
decisive basic mixing tasks or it is uncertain how the agita - start-up
tor affects the process result, trials with model fluids or the The agitator seal plays a vital role in the reliability of the
original product and/or computational fluid dynamics (CFD) equipment system. Pressure, temperature and mode of Fig. 4 Basic workflow for the process and mechanical design of a mixing system
are necessary to obtain an adequate knowledge base for operation govern the type of seal and sealing materials. Just
selecting an impeller and for specifying the diameter, speed as important is the seal liquid supply unit. The type and
and power. scope of its instrumentation and monitoring must also be The responsibility of the agitator manufacturer does not
specified by the operator and the manufacturer based on end with delivery of the machine or the complete system,
The vessel and its internals must be dimensioned in a joint risk analysis. it must also provide support to the customer during instal-
mutual agreement or parallel to designing the agitator. The equipment manufacturer must ensure not only that lation and commissioning. With its knowledge of the indi-
The vessel size depends on the required throughput and project processing is on schedule, but also that the agreed vidual boundary conditions, the manufacturer is also a part-
the process kinetics. The positioning of inlet and outlet quality assurance is transparent and that all project phases ner for planning and execution of measures relating to
points requires knowledge of the flow generated by the are completely documented. preventive maintenance and servicing over the entire ser -
agitator. This is the only way to avoid short- circuit flow vice life of the equipment.
paths during continuous operation and to achieve rapid
mixing of reactants down to the molecular level. Although

24 25

The processes taking place in a mixing or
blending vessel are very complex and are based
on the laws of fluid mechanics and bulk material
mechanics. The three- dimensional flow gener -
ated by the impellers interacts with the vessel
itself and its internal fittings. The classical laws
of fluid mechanics, particularly the conservation
laws for mass, energy and momentum as well as
the descriptions of internal friction, turbulence
phenomena and special rheological features are
sufficient to characterise and analyse the flow
dynamics.
RHEOLOGY
DIMENSIONLESS NUMBERS
HOMOGENIZING
SUSPENDING SOLIDS
DISPERSING GAS / LIQUID
DISPERSING LIQUID / LIQUID
The complex interplay between these factors HEAT TRANSFER
means that many of the individual steps are MIXING SOLIDS
difficult to understand. They are thus classified VACUUM CONTACT DRYING
systematically into the various basic mixing tasks,
and their interactions with the flow generated by
the agitator and with the individual chemical and
physical process steps are analysed. Knowledge
and understanding of the parameters relevant to
success then enable systematic planning of tests
or simulations and is the basis for subsequent
scaling up to production scale ,

The actual purpose of mixing is to achieve

a process result that is based on very diverse
physical and chemical phenomena that have their
own mechanisms and kinetics. Liquid, gaseous
or solid reaction partners must be dissolved
and homogenized down to the molecular level .
Crystallizing solids must be kept suspended and
.
their growth controlled The flow turbulence is
exploited to produce dispersions or stable emul-
sions. Most process steps require either the input
or removal of heat via the vessel wall or internal
internal heat exchangers or by evaporation or
condensation of process media .
 THE SCIENCE
FUNDAMENTALS
Rheology is the science of deformation and flow of materials. The rheology of a fluid is a funda -
mental property, knowledge of which is essential in order to understand and even predict the flow
in a mixing vessel . The basic mixing tasks and the power input by the impeller outside the turbu-
lent flow regime are influenced by the product ' s rheology.
DEFINITIONS
The experimental set-up shown in Fig. 5 can be used to
explain the viscosity. A fluid is applied to a firm substrate
and is then covered with a plate. A certain force F is re-
quired to slide the plate. The relationship between Fand
the area A of the respective fluid element is defined as
the shear stress r. For most fluids, force F only occurs
while the upper plate shown in Fig . 5 is moving with
. However, some materials exhibit elastic re -
velocity
storing forces. Given that the liquid in the set- up shown
in Fig. 5 is homogeneous and adheres to the substance
and the upper plate, i.e. vx(y = 0) = 0 and vx(y = h) = vp,
then
where y is the shear rate or the shear gradient.
If we now consideran infinitesimally small fluid element
in the experiment shown in Fig. 5 to be in the sheared
state, then for the stress state shown in Fig. 6, the normal
stresses acting on the fluid element are an and the
tangential stresses are xy and xik.
A v?
F \
! b
/ / / / / / /7777
dy
77777777777
u
y
Fig. 5 Shear test with definition of the basic parameters
28
RHEOLOGY RHEOLOGY
DIMENSIONLESS NUMBERS
HOMOGENIZING
SUSPENDING SOLIDS
DISPERSING GAS / LIQUID
DISPERSING LIQUID/ LIQUID
HEAT TRANSFER The relationship between the shear rate and the shear NON - NEWTONIAN FLUIDS
MIXING SOLIDS stress r is expressed by
In contrast to Newtonian media, the viscosity of non-
VACUUM CONTACT DRYING
x = t} - y (2 ) Newtonian media is usually strongly dependent on the
shear rate and/or the loading time. This leads to the flow
where r/ is the dynamic viscosity. anomalies shown in Fig. 8 and described in detail below.
The dynamic viscosity is mainly influenced by the tem - SHEAR THINNING
perature, the shear rate and the loading time. The influence By far the most common flow anomaly is shear thinning
of pressure can differ greatly in various liquids. However, ( also known as pseudo - plastic flow behaviour ) . In this
using the rule of thumb that a pressure change of case, the viscosity decreases with increasing shear rate,
Ap = 300 bar results in about the same change as i.e. the shear stress x is degressive with respect to the
AF = -1 K [96], the influence of pressure on the viscosity shear rate. In many cases, shear thinning fluids can be
can be neglected for the processes described in this book. described by the Ostwald - de Waele power law:
There are further dependencies for mixtures, e. g. on the
solids content of a suspension. x=K - ym (3)
In agitated vessels, it is possible to process materials
with viscosities below 1 to above 1 000 000 mPa - s There . where K is the consistency factor and m is the flow
is no standardised definition of viscosity ranges; how - index.
a
.
ever, they can be roughly divided into
Combining this power law with the Newtonian shear
xzy
y low viscosity:
medium viscosity:
Tj < 500 mPa s, -
TJ = 500 - 10 000 mPa s, -
stress equation gives
high viscosity: rj > 10 000 mPa s. - V =K - ym -1
(4 )
V
Fig. 9 shows some examples of viscosity curves for
Txz Ty
NEWTONIAN FLUIDS some shear thinning fluids, shown as straight lines in the
\og{ rj )Aag(y ) Ostwald - de Waele diagram. The gradient of
If the resistance forceFto plate movement, measured the straight line represents the degree of shear thinning,
y as the shear stress r, is proportional to the shear rate f , represented by the flow index m, for a particular material
the fluid is said to show Newtonian behaviour. For such group ( e. g. xanthan or CMC). The absolute value of the
fluids, the relationship generally represented by viscosity is described by the consistency factor. When
a flow curve, is a straight line that passes through zero. m = 1, Kcorresponds to the viscosity of the Newtonian
m The proportionality constant is known as the dynamic shear stress law at j = 1/s.
viscosity rj and is a material parameterthat is independ-
ent of the shear field in Newtonian fluids. DILATANCY
Ay
The corresponding flow curve and viscosity curve for Dilatancy is the opposite type of behaviour to shear
Fig. 6 Stress vectors acting on a fluid element Newtonian media are shown in Fig. 7. thinning, and is shown in Fig. 8 (Point 2 ) as a dashed line.
Although dilatant media are very rarely encountered, they
must be taken into special account when designing the
For the simple load case shown in Fig. 6, xxy = 2 = 0 agitator because they exhibit a progressive increase of
and
.
^
= = 0. In this case, the only shear stresses pre-
xxz xzx
sent are xyz = x~y For plate size A and force F, this gives
Flow curve Viscosity curve the shear stresses with the shear rate. It is even possible
to detect dilatancy in a glass beaker using the rod test:
F Xyz — Tzy — F/A . whereas a glass rod immersed vertically in the medium
T n and resting on the bottom slowly falls over when it is
released, the medium consolidates when the rod is pulled
up quickly so that it is possible to lift the beaker as well.
Y I
r| = dynamic viscosity n = T-
Fig. 7 Rheology of Newtonian fluids
29
 RHEOLOGY

Time-independent flow behaviour PLASTICITY Time - dependent flow behaviour is particularly impor-
Plasticity is a frequently occurring flow anomaly at low tant with respect to restarting the agitator after a standstill
5
.
1 Shear thinning 1. - Polymer dopes shear stresses. A plastic substance behaves like a solid at ( increased torques).
.
2 Dilatancy ) - Latex low shear stresses. The substance only starts to flow above
- Paper pulp 2
a limiting shear stress r0, which is termed the yield stress.
- Paints
n 2. - Rice starch in water Fluids are divided into Herschel-Bulkley and Bingham -type VISCOSITY MEASUREMENT
104
- Whipping cream fluids, depending on their flow behaviour above the yield
- Glazes stress. 5 The influence of the shear rate on the viscosity is meas-
- Sasil ured with a rheometer. The various available types operate
according to Ostwald- de Waele: Herschel - Bulkley fluids behave like shear thinning fluids
1
T =K • ym =K >'1
n - yn above the yield stress, i.e. the viscosity gradually de - 2 on different principles.
creases as the shear rate increases. Common practical These include rotational rheometers, falling ball
.
3 Herschel-Bulkley 3. - Lubricating greases examples are suspensions and gels. 103 viscometers and capillary rheometers.
- Gels In contrast, Bingham -type fluids behave analogously
5
n to Newtonian fluids above the yield stress, and the shear 1. ROTATIONAL RHEOMETER
stress increases proportionally to the shear rate. Thus as The requirement for a defined shear gradient y is almost
2
the shear rate increases, Bingham fluids behave increas - fulfilled with a rotational rheometer with coaxial cylinders
Y
ingly like Newtonian fluids. Typical examples of Bingham ( Fig. 11 right ) and a narrow annular gap. Rotation of
T =T +K - ym
T
0 +K - r media are drip -free paints, toothpaste and filled bitumen.
10 ' 2 5 102
the measuring body generates laminar flow within the
0 n= v 1
Y [s ]
’

annular gap filled with product. A constant rotational fre-

.
4 Bingham ( plastic ) 4. - Toothpaste ELASTOVISCOSITY
m- l
quency and a sufficiently small annular gap produce a de -
- Filled bitumen The fifth type of flow anomaly is elastoviscosity. Ostwald-de Waele: r|= K - y fined shear rate. For simplification purposes, a linear
n - Yeast dough
A " Maxwell fluid" is characterised by a combination of Cone. cw K Curve
relationship is assumed for the speed across the radius.
- Ketchup
- Wall paint viscous and elastic behaviour. At a constant shear rate, [ %] [Pa • sm ] [-] For a given product- contacted surface area on the cir-
Y they exhibit the viscosity curve given in Fig. 8 under
Xanthan
-
cumference of the cylinder 2% Ryh, the associated shear
Point 5. The curve shows an initial increase that corre - 1 13.5 0.12 1 stress r can be determined using a torque measuring
T=T
0 +VY
J +% sponds to the modulus of elasticity in shear G of the
^ value
3 40 0.12 2 device.
Hookean spring in the model and reaches a final 5 300 0.12 3 Pairs of ( y, T) values are determined at different rota -
that is determined by the characteristic viscosity ?/N CMC tional frequencies and are then plotted as a flow curve.
Time- dependent flow behaviour
of the Newtonian damping term. The intersection of the 0.5 0.6 0.78 4 An accurate statement on the viscosity as a function of
5. Elastoviscosity 5. - Long-chain initial tangent and the plateau value gives a further the shear rate can be obtained not only with a coaxial
polymers, descriptive parameter : the characteristic relaxation time Fig. 9 Viscosity curves of CMC and xanthan solutions cylinder rheometer, but also using the cone /plate princi -
n
e.g. polyacrylamide
?R = 7N/GH. Overall, a transition takes place from the ini- ple. Compared to the coaxial cylinder rheometer ( also
»WeiGenberg
effect « tial elastic behaviour represented by the spring to the THIXOTROPY AND RHE 0 PEXY known as a Couette rheometer ), the latter can be used to
t final viscous behaviour represented by damping. Thixotropy and rheopexy are two types of time- depend- measure products with significantly higher viscosities.
*R Typical examples of elastoviscous fluids such as poly - ent flow phenomena, see Points 6 and 7 in Fig. 8. In thix- In contrast to the cone/plate principle, the plate/plate prin -
YO + TIN [I exp( /y]
T= "
l
mer melts and solutions whose elongated molecular otropic substances, the viscosity decreases with time ciple (Fig. 11 left) uses a shear rate gradient overthe shear
tR = pN/GH characteristic
relaxation time structure causes elastic effects that are usually exhibited under a constant shear load ( stirring). After a rest period, gap. In this case, an exact correlation between the viscos-
_ only by elastic solids. This manifests itself in unusual flow the fluid regenerates itself and approaches its original ity and the shear rate is only possible using complicated
n = T = % !t1
‘ '
exP( tV] phenomena, e.g. the Weissenberg effect observed during viscosity ( with hysteresis effects). Rheopexy is the reverse correction methods ( e.g. Rabinowitsch -WeilJenberg cor-
.
6 Thixotropy 6. - Clay suspensions stirring in which the normal stresses force the medium process; however, it occurs comparatively rarely. rection or Schummer correction) ( [38], [95] ). However,
Shearing Regeneration the plate /plate principle has the advantage that it can ac -
. - Paints to creep up the rotating agitator shaft (see Fig. 10 ), where - Thixotropy generally occurs together with shear thin -
Y = const
n - Honey as a vortex is formed in Newtonian fluids as a result of ning and plasticity. These flow anomalies are due to the curately measure media with a solids fraction.
- Varnishes
centrifugal forces [95]. It is imperative that such phenom - formation of " molecular frameworks " caused by molecu-
t ena are taken into account in the constructive design of lar forces of attraction . Shear loading destroys this frame - 2. CAPILLARYVISCOMETER
mixing equipment. work structure and thus lowers the viscosity until, after According to Hagen -Poiseuille, the condition for laminar
.
7 Rheopexy 7. - Vanadium oxide a certain time, shear loading and the molecular forces are flow through apipe is met because the discharge nozzle
Shearing Regeneration
suspension in equilibrium thus resulting in an effective final viscos - of this viscometer is shaped like a capillary tube. The
- Soap lye
n - Gypsum /water
ity. These frameworks are gradually rebuilt during the respective viscosity can be determined from the known
suspension regeneration phase. geometric dimensions as well as the passage time of the
t volume of liquid.

Fig. 8 Flow behaviour of non-Newtonian fluids

30 31

Fig. 10 Mixing experiment with a vortex in a Newtonian fluid (left ) and
Weissenberg effect in a viscoelastic fluid Iright )
3 . FALLING BALL VISCOMETER
In this case, the measurement is based on Stokes ' Law
of Resistance for a ball surrounded by laminar flow in
a viscous medium. The time the ball takes to fall a defined
distance is used to determine the viscosity of the inves -
tigated fluid.
However, useful measurements are only obtained when
a rheometer is used correctly within the intended applica -
tion range. This means, for example, thatthe presence of
secondary flow phenomena such as Taylor vortices must
be avoided in a rotational rheometer with coaxial cylin-
ders. For the cone/plate principle, no suspensions may
be used due to central narrowing of the measuring gap.
Fig. 11 Rotational rheometer with plate/plate design (left) for fluids con-
taining solids and coaxial cylinders (right)
32
Furthermore, it must be taken into account that the
mechanical energy introduced during measurement is
converted into thermal energy and thus heats the sample.
The transformed energy can be calculated with the
relationship:
-
= T - j = T] Y
2 ^
£ (5)
This effect can lead to considerable temperature in -
creases at correspondingly high shear rates, even in
structurally shear thinning fluid.
In addition to the methods for measuring the true vis -
cosity shown in Fig. 11 , there are a number of methods
that are used for quality control purposes and which are
limited to the measurement of relative values.
Popular relative measuring methods that do not allow
direct conversion to the apparent viscosity at the impeller
rjs , include:
- the Brookfield viscometer
- viscosity cup according to DIN 53211 ( flow time is used
to determine the viscosity; primarily used for paints
and varnishes)
- the Engler viscometer according to DIN 51550 ( flow
time of the investigated liquid is compared to that of
the same quantity of distilled water at 20 °C; mainly
used in the petrochemical industry )
The Brookfield viscometer is widely used for quality
control. This system is based on the rotation principle
with a spindle rotating in a liquid that is assumed to be
an infinite body. Brookfield viscosities are usually sig -
nificantly higher than the true viscosity. Furthermore, the
measured viscosity cannot be assigned to an accurate
shear rate so that this measurement is not suitable for
generating flow curves and is thus very difficult to imple -
ment from the process engineering point of view.
RHEOLOGY
APPARENT VISCOSITY IN A MIXING VESSEL P
Ne = (8 )
3
pn d2
When agitating shear thinning liquids, for example, the
viscosity in the vicinity of the impeller has the lowest .
4 Determination of the Re number from the power
value and increases with increasing distance from the characteristic Re —f (Ne) of the respective impeller
impeller. In order to describe the viscosity experienced 5. Calculation of the apparent viscosity with
by the impeller, the apparent viscosity app can be defined
-
pndi
by modifying Equation ( 2):
r
^ RT (9)
?7 s =— 6. The resulting pair of values ( t ] s , ri) are used to obtain
Xapp IB) the agitator viscosity curve and, for geometric simi-
larity of vessel and impeller in the pilot and produc -
where j>app is the apparent shear rate at the impeller. tion scales, they can be employed for scaling up ( see
The apparent viscosity is used to determine the power also page 86 ) .
number Ne and and to calculate the power input of the
impeller in the laminar and transitional flow regime.
The best-known method of determining the apparent COMPUTATIONAL FLUID DYNAMICS ( CFD )
shear rate at the impeller was developed by Metzner and
Otto [84] , They proved thatthe shear rate is proportional If the product rheology is known, CFD ( see page 92 ff.)
to the agitator speed, i. e. can be used to determine how the shear rate or the vis -
cosity behaves. Fig. 12 shows an example of this for mix-
ing a shear thinning product using an Ekato Paravisc.
7 s = k M /O n 17)
Due to the high shear rate at the impeller, the viscosity
Approximate values of typical Metzner- Otto constants here is lower than in the more distant zones.
kMJO f°r common impellers are:
As already mentioned above, this affects the various
Marine - type propellers kMiO = 10 basic mixing tasks that that will be described described
in more detail in this chapter. Thus, the impeller operated
Flat blade disc turbines
Flelical ribbon impellers
^kMZO
M/0 = 12
= 30 close to the wall has a positive effect not only on the blend
Anchor impellers kMfO ~ time ( see page 38 ff.), but also induces good flow onto
the vessel wall. Lowering the viscosity close to the wall
After determining the apparent shear rate at the impel- also significantly improves heat transfer via the vessel
ler using Equation (7 ), the apparent viscosity can be read wall ( see page 68 ff.).
off the viscosity curve. The apparent values for the viscos -
ity and shear rate in the mixing vessel are made equal to
the real values from the viscosity measurement.
Strictly speaking, for most impellers, the Metzner- Otto
method using Equation (7 ) is only useful in the laminarflow
regime.
Viscosity i ]
If the Metzner - Otto constant is not known or if the flow
regime under consideration lies outside the valid range H 1000 Pa s -
of this method, the Rieger- Novak method [105] can be used
in the laminar flow regime and in the transition regime.
The Rieger-Novak method uses the following steps:
525 Pa s
1. Measurement of the shaft torque Mt - typically using
a pilot scale - and at several speeds
2. Calculation of the impeller power with P = 2 nnMt
3. Calculation of the power number Ne with
I 50 Pa s
Fig. 12 Viscosity distribution in a mixing vessel containing a shear thinning
fluid
33
 RHEOLOGY DIMENSIONLESS NUMBERS
THE SCIENCE
DIMENSIONSLESS NUMBERS
FUNDAMENTALS HOMOGENIZING
SUSPENDING SOLIDS
DISPERSING GAS/ LIQUID
DISPERSING LIQUID / LIQUID
HEATTRANSFER 1.
MIXING SOLIDS 103 C
M = [ d2 f t«] - [pr = La [
T b • M IT 2 )
~
•
VACUUM CONTACT DRYING (10 )
h
102
- T"

- M‘
Ss
From [77 ] = M - /T1 - T l , it follows that
~

(11 )
g 10
’ 2

.3 e« 1
-
6 1
Dimensionless numbers are widely used in physics and the engineering sciences. In mixing
technology, parameters determined by means of a dimensional analysis can make a major
contribution to describing the relevant influencing variables. The functional correlations, known as
10
°

10 1
"
'
6 — 2.
« » 1 3c - 2.
•

characteristics, derived from similarity theory form the basis for extrapolating process parameters
determined at the model scale up to the production scale.
10 ° 101 102 103
Re
104 105 106
[P ] = [ d2 f - [ nf ipf = La T b [ M IT2
~

jC (12)
= La
2c
~

- T
-b
- Mc
1 Ekato Paravisc
Flat blade disc turbine, 6 blades
From [P] = M - L2 - T , it follows that
2 ~2
(13)
3 Pitch blade turbine, 4 blades h fd = &12S
Dimensional analysis and the theory of models have - The selected influencing quantities must be inde - 4 Intermig 07, two stages ^ ^
a common goal: to determinethecharacteristicdimension- pendent of each other. 5 Propeller, three blades e =l
less numbers in order to describe a physical process [98] , - The selected parameters must be primary parame - 6 Intermig 09, two stages 6=3
In his ^- theorem, Buckingham showed that the number of ters ( directly measureable), i.e. not derived from oth- a = 2 + 3c = 5 .
dimensionless numbers (^ variables) that are needed to er quantities. Fig 13. Power characteristics of various impellers
completely describe a particular physical process equals - Universal physical constants also belong in the list of Thus, according to Buckingham ' s 7T-theorem, two di-
the number of independent physical quantities involved relevant quantities. The aforementioned influencing quantities include the mensionless numbers result that are required to derive
minus the number of basic dimensions. - Preference should be given to selecting a geometric 3 fundamental physical quantities: length L , mass Mand the power input. The first number
Dimensionless quantities can be used to significantly parameter from the general layout as a characteristic time T, which gives the following dimensional matrix:
reduce the amount of experimental work that is necessary dimension of size.
p-n- dl = Re (14)
to determine the physical correlation of a particular pro - *\ = n
Physical quantities
cess. The judicious selection of parameters allows For example, to determine the power equation, the fol- gives the Reynolds number Re, which describes
the characteristic numberto be varied within wide limits lowing influencing quantities must be listed first: the flow regime.
n
so that the respective variables have a direct effect on the P
*
<2 P
dimensionless numbers. Applying this method, the char-
acteristic numbers required to design the process are
relatively simple to determine in the model scale. It also
Geometric:
Material property:
impeller diameter d2 [m]
densityp [kg/m3] ll
H
M X 0 0 l 1
The second number

TT2 =
p
= Ne
viscosity rj [kg/irrs] (15 )
provides a high flexibility of the experimental parameters. Kinematic: agitator speed n [s-1 ]
1.1>
3 V)
LL
L 2 1 0 -3 -1 p - n2 - d2
Another key advantage of process - relevant dimensionless Target parameter: power P [kg -m2/s3] CL
gives the well -known power equation used in mixing tech -
T - 3 0 0 -1
numbers lies in the unambiguous description of the phys - nology
ical phenomena under consideration, for example, to dif - The power input for single - phase liquids, i. e. without
ferentiate between laminar and turbulent flow using the a gas fraction, is thus defined by the following function: STEP 3: FORMATION 0 FTHE DIMENSIONLESS P - Ne -p tv ' d25 (16)
corresponding value of the Reynolds number. This allows NUMBERS
a deeper insight into thefundamental physical principles P =f ( n, d2, p, rj ) The dimensionless numbers are generated from the Ne is the Newton or power number. These two dimen -
of the examined processes. dimensional matrix . sionless numbers can be used with the experimentally
The most important steps in determining dimensionless STEP 2: CONVERSION OF THE MATRIX OF DIMENSIONS This results in « -r = 5 - 3 = 2 equations, where determined power characteristic JVie =f { Re ) to calculate
numbers by means of a dimensional analysis are [150]: INTO A MATRIX OF UNITS the power input ( see Fig. 13).
To facilitate subsequent interpretation of the dimension- n: number of influencing quantities n, d2 , p, t )
STEP 1: LISTING THE RELEVANT PHYSICAL QUANTITIES less numbers, a matrix that can be handled easily in the r : rank or fundamental physical quantities of the above ^
The chief difficulty here lies in accurately determining further procedure is generated. Irrespective of how it matrix. PHYSICAL IMPORTANCE OF DIMENSIONLESS
all the relevant influencing variables. From the point of is arranged, the same set of dimensionless numbers is NUMBERS
view of mixing technology, these are essentially the geo - always obtained; however, with different exponents. The Next, 3 of the 5 unknowns are selected and the dimen -
metric, material- based and kinematic parameters. When reader is referred to Zlokarnik for details on the use of sions of the remaining 2 are described by products of the Other dimensionless numbers can be determined in an
determining the influencing variables, particular attention dimension matrices [150] , exponents of the dimensions of these 3 variables. analogous manner for the various physical processes
must be paid to the following: involved in mixing technology [148] ,

34 35
 DIMENSIONLESS NUMBERS

'

Overthe years, various dimensionless parameters have

been derived and applied notonlytothe powerinput but Name Dimensionless number Physical significance Relevance to mixing technology
also to other problems, notably homogenizing and disper-
sion processes as well as the analysis of heat transfer [21]. &
Archimedes Ar =
V
S' P
2
-P
^ Buoyancy force
Frictional force
Settling phenomena

Convective flow
CHARACTERISTICS Bodenstein
A* Axial diffusivity
Back -mixing reactor

The graphical representation of dimensionless numbers Reaction rate

Damkohler Da = kr - cn ~
Reaction technology
are known as "characteristics". Particularly important with Convective mass transport rate
respect to mixing technology are the power characteristic
-
Ne = f( Re ), the blend time characteristic n 0 = f ( Re ) and the Euler Eu = IA
g
.
Dynamic pressure
Static pressure
Entrainment of gas
'
hs
heat transfer characteristicjVM ^/ e, Pr ). All characteristics
^
are used to describe defined properties and relationships. Froude Fr
n2
g
£% Centrifugal force
Gravitational force
Gassing, vortex formation
This is explained using the example of a power character-
istic: P Power input farce
Newton Ne = Agitator power
- -
p r? d2 Inertial force
The dimensionless number Ne , required for determining a d\ Total heat flux
NuBelt Nu Heat transfer
the power input by an agitator, depends on the impeller X Conductive heat flux
type, the geometric arrangement of the agitator in the
vessel and, as previously derived, on the Reynolds number. Peclet Pe =
n - dl - Cp - p = Re - Pr Convective heat transport
Heat transfer
X Conductive heat transport
The calculated Reynolds number can be used to determine
the Newton number Ne and thus calculate the power input. V Internal friction
Prandtl
V
Pr = -=-
X ^ Conductive heat flux
Ratio of physical properties

In accordance with the wide range of media to be mixed,

p - n - di Inertial force
which may have viscosities of less than 1 to more than Reynolds Re = Flow regime
n Internal frictional force
100000 mPa - s, there are two main types of flow condi-
tions: the laminarflow regime, in which only viscosity or Internal friction
Schmidt Sc = - Ratio of physical properties
D Diffusion rate
frictional forces are acting, and the turbulent flow regime,
in which only inertial forces are acting . The ratio between tA Convective mass transfer
Sherwood SH = Mass transfer
inertial and viscous forces, represented by the Reynolds Fig. 14 The occurrence of vortices can be described by the Froude number D Diffusive mass transport
number, is one of the most important dimensionless num - gA Reduction in pressure
Sm
ber in an agitator system. Smith Cavitation
K Local pressure change

Vortex shedding frequency x Length

All characteristics must be determined experimen - Strouhal KrJ±
u
~
Flow velocity
Vortex shedding frequency
(Karman vortices)
tally and, apply strictly only to the impeller configuration
tested. Further characteristics relevantto mixing technol- p - u2 - di Inertial force
Weber We = Dispersion processes
ogy are described in later chapters. <T Surface force

Gassing number
A Gas flow rate
Pumping rate Dimensionless volumetric flow rate

Homogenizing -
n 9 Number of impeller rotations Dimensionless blend time

Fig. 15 Dimensionless numbers relevant to mixing technology

36 37
 RHEOLOGY HOMOGENIZING
THE SCIENCE
DIMENSIONLESS NUMBERS
FUNDAMENTALS HOMOGENIZING
SUSPENDING SOLIDS
DISPERSING GAS/ LIQUID
DISPERSING LIQUID / LIQUID
HEAT TRANSFER During mixing of two components ( A and B ), there are
MIXING SOLIDS convective and diffusive mixing processes taking place mixing
in the vessel. A simplified representation of the mixing
VACUUM CONTACT DRYING
process is shown in Fig. 16. In general, there are three
mixing scales that are differentiated by the magnitude of
the characteristic mixing length:

- Macro -mixing takes place on the scale of the impeller

or vessel dimensions and is due to the flow primarily
The goal of every mixing process is to achieve the required homogeneity throughout the vessel . generated by the impeller. Macro - mixing causes glob - Fig. 16 Simplified representation of the mixing process due to convec-
During homogenizing, it is important to minimize or eliminate any concentration and /or temperature al mixing of the components in the vessel. The macro - tion ( mixing ) and diffusion
gradients arising as a result of e .g . adding products, mass transfers, chemical reactions as well mixing time is thus also synonymous with the blend
as input or removal of heat . This applies to both single- phase and multi- phase systems. time 6 because small unmixed zones cannot be regis - coefficient!) is used to estimate the diffusive blend time
tered by the aforementioned measuring probes and are 6 according to [129]:
also no longer visually perceptible. This corresponds
.
to states b) or c) in Fig. 16
0diff -
-
0.5 A
2

- Meso-mixing takes place on the scale between macro - 4 D

(20 )

To reduce the energy consumption and/or increase the
productivity it is necessary to clarify whether the blend
time is a limiting factor for the respective process. The
blend time of low-viscosity fluids is generally in the order
of minutes or less and is thus frequently non - critical. If a
fermenter has zones in which the pH value, temperature
or the nutrient concentration deviates from the ideal val-
ue, i.e. from the mean value, the organisms in such zones
produce less product or undesirable substances, which
has a negative effect on the productivity of the overall
process. Particularly in the case of fed- batch or continuous
processes, the influence of attaining homogeneity too
slowly is often underestimated. In continuously operated
mixing vessels, it is important that the ratio of residence
timer to the blend time 0Mdoes not drop below a minimum
value. This value lies in the range r/0M = 10-50 and depends
on the specific process requirements . Particularly in transitional and laminar flow regimes there
BLENDTIMEAND DEGREE OF HOMOGENEITY
The elimination of concentration or temperature gradi -
ents in the liquid phase is primarily characterised by the
.
blend time in the case of an agitated system This is the
time required to reach a defined degree of homogeneity
within the vessel. The degree of homogeneity is defined
by the time- dependent change in the concentration of a
tracer from CQ to c( t ), given as
Other degrees of homogeneity require different blend
times, given as
~ #95 '
( )
ta 1- jQo
M - fc)
'
(18)
MEASURING METHODS
A variety of methods are used to measure the degree
of homogeneity in mixing vessels. Measurements of tem-
perature, pH or conductivity using probes have the ad-
vantage of a high reproducibility. Strictly speaking, the
measured value only applies to the point of measurement.
are areas that are poorly mixed or where there are toroi -
dal vortices etc . that may not be correctly registered by
probes [46].
In this case, visual evaluation with discolouration meth-
ods is still a fast and robust method that can be used to
evaluate the entire contents of the vessel. A reliable com-
parison of blend times is only possible if both the degree
of homogeneity and the measuring method are evalu -
ated.
.
and micro -mixing The primary vortices generated by
the impellertransfertheir energy to ever smaller vorti-
ces so that mixing takes place on an increasingly small-
er scale. This still corresponds to states b) and c) in Fig.
16, which appear to be well mixed on the macroscopic
scale. Flowever, greater magnification reveals there is
still significant segregation of the two components A
and B.
- Micro -mixing takes place below the scale of the small -
est micro -vortex. The size of this scale is generally es-
timated using the Kolmogorov equation for isotropic
turbulence ( see Equation (19 ) and page 84). Below
this scale, mixing occurs exclusively via diffusion.
A homogeneous mixture ultimately results from dif -
fusive mixing at the molecular level. This corresponds
to states d) and e) in Fig. 16.
Unlike the representation in Fig. 16 in which convective
and diffusive mixing processes occur consecutively, these
processes take place simultaneously in real systems.
Flowever, the last mixing step until the mixture is com -
pletely homogenized is always diffusive transport.
The energy introduced into the mixing process is dissi-
pated at the level of the smallest micro -vortex and is com-
pletely converted into heat.
The micro -vortex size A is generally estimated using the
simplified Kolmogorov equation for isotropic turbulence
3 ,
1/ 4
is equivalent to the micro -mixing time. Typical diffu-
sion coefficients of 2 • 10-9 m2/s and A = 40 pm give a mi-
cro -mixing time 0.1 s.
Continuous mixers based on the rotor/stator principle
can be used to dissipate specific powers of up to
500 kW /kg in the gap zones, thus achieving micro - vortex
sizes of around 1 pm and diffusive blend times of < 1 ms.
MIXING MECHANISMS IN LAMINAR ZONES
In contrast to turbulent flow, purely laminar flow does
not exhibit any velocity fluctuations in the main direction
of flow and thus there is no mixing transverse to the in-
dividual flow layers. Consequently, for the transition from
state a ) to state b) shown in Fig. 16, in addition to the
deformation of streaks, as shown in Fig. 17, reorientation,
i.e. rearrangement and folding is necessary. This means
that the interfacial area between components A and B
continuously increases during the mixing process. Mixing
processes within the phase interfaces take place purely
by diffusion without fluctuating transverse flows.
The enlargement of the interfacial area can be illustrated
using streak deformation in the coaxial annular gap (Fig. 18).
The following empirical relationship applies here
S( t ) 1

cp - c( t )
MIXING MECHANISMS INTURBULENTZONES A=
V

£
(19)
Jl+ ( y - /)2 (21)

M {t ) = (17)
c0 coo i. e. the initial thickness 5 of the streak becomes smaller
Turbulent flow differs from laminar flow in that the veloc -
ity fluctuates in the main direction of flow. These velocity Below this scale, mixing occurs exclusively via molecular ^
over time t until a thickness S( t ) is reached, which depends
Blend times can only be compared for the same degree fluctuations cause differing degrees of swirling and thus lead diffusion. on the shear rate f . Given sufficient thinning and distribu -
of homogeneity, a typical value of which is x = 95 %. to mixing of different layers in a turbulent flow zone. In aqueous media and predominantly turbulent flow, tion, the remaining concentration gradient is subsequent-
the micro -vortex size is generally A < 40 pm. The diffusion ly eliminated by molecular diffusion.

38 39
 HOMOGENIZING

for low-viscosity media ( Fig. 19). The blade design of this

impeller is curved and has an upward lip on the tip that
prevents high - loss pressure - compensation vortices
around the blade tip. The laser light stripe image visual-
ises the flow conditions on the suction side of a single-
J stage impeller. In a multi- stage system, one stage push-
es the fluid into the suction region of the next. This results
in a bundled axial jet without pulsation, as shown in the
fluid dynamics simulation. The exponent in Equation
*
( 23) for this type of impeller lies significantly below the
otherwise typical lower limit of 1.5.
Fig. 20 clearly illustrates the effects of this design: the
blend time with the Ekato Isojet is half that of a conven-
tional axial flow impeller with a simple blade shape, even 3.0
though the diameter ratio is much smaller. The small di-
Fig. 17 Streak deformation ameter ratio meets the requirement of a higher speed so
2.2
that the agitator can be operated with a small torque but
at the same power. This reduces the manufacturing costs
HOMOGENIZING NEWTONIAN MEDIA IN of shafts, mechanical seals, bearings, and gearboxes . To 1.5
TURBULENTZONES achieve the same degree of homogenization in a continu-
ous reactor, and because nd = const., a conventional axial
0.8
In the turbulent flow regime, the blend times of New- flow impeller with the same diameter ratio must operate
tonian media for single - stage, axial pumping impellers at twice the speed and thus with eight times the power,
in a vessel with a height/diameter ratio of h /d ~ 1 can Traditional agitators, e.g. flat blade disc turbines or pitched 0.0
be calculated with the following relationship: ^ ^ blade turbines would be even more unfavourable with
-2
respect to blend time and torques [50].
n ' @M - 5,5 - h • Ne-1/3 Fig. 19 Ekato Isojet: Axial flow impeller with specially designed blade to suppress pressure compensation vortices. Laser light stripe image of the single
di

—
(22) impeller and flow lines in a tall vessel with h /dj = 3.5 and 5 impeller stages, velocity scale in m/s
Radial pumping impellers exhibit a slightly deviating ^
factor or exponent forthis diameter ratio ( [50], [83] ) . Thus, HOMOGENIZING NEWTONIAN MEDIA IN THE problems associated with applying the Reynolds number
irrespective of the impeller type, this gives the parameters TRANSITIONAL FLOW REGIME of the agitatorto homogenizing processes: strictly speak-
influencing the blend time as
For manufacturing formulated products, i.e. finished
(tjv (23)
consumer goods such as cosmetics, paints, varnishes,
adhesives, and sealants, the rate - determining step in
140
130
a batch process is often the absolute blend time because Hydrofoil
In tall vessels, the blend times are considerably longer,
rotating cylinder
these products have high demands with respect to the
120
110
,
d2/d = 0.6
so that a correction term for the aspect ratio has to be degree of homogeneity. Deviations have a direct effect 1
<e 100
added to Equation ( 22): on the product quality and are unacceptable. For medium
and high-viscosity products, the blend time can be con- *= 90
1
n - dM is ( = 1,5 ... 2,5)
*
The value of the exponent now depends not only on the
(24 )
siderable, particularly if there are poorly mixed zones in
the vessel [46], If the viscosity of the agitated medium
increases so that the Reynolds number drops below the
80
70
60

diameter ratio, but particularly on the type of impeller

critical value Rec
TT defined in Equation ( 25), the blend 50
time characteristic changes: 40
and the number of impeller stages. In contrast to Equa -
0.4 0.6 0.8 1 1.2 1.4 1.6
tions ( 22 ) and ( 23 ), this relationship reveals significant
differences in the suitability of individual impellers. The
° 1/3
Rec , TT - Ne
TT (25 )
Specific power input [ kW / m3]
requirement for high degrees of homogeneity, i.e. short The subscript " TT " stands for the changeover from tur-
vessel wall
blend times in large tall vessels, can be fulfilled with a type bulent to the transitional flow. This critical Reynolds num - Fig. 20 Comparison of blend times for axial flow impellers, V = 200 m3,
of impeller whose pumping efficiency has been optimised Fig. 18 Streak deformation in a coaxial annular gap ber is 10 to 100 times higher than the value at which the hj/dj = 3.5 and 5 impeller stages with the same bottom clearance and
spacing between stages [49]
power number Ne starts to increase. This illustrates the

40 41

-

ing, the Reynolds number of the agitator describes the
conditions arising from pressure and viscosity forces only
at the impeller, it does not apply to the entire contents of
the vessel . However, a Reynolds number, which
corresponds to a Fourier number, can be defined for the
vessel:
1 d -p
? HOMOGENIZING IN THE LAMINAR ZONE
Rev = —
Fo 0M rj (26 )
In the transitional flow regime, the blend time in New -
tonian liquids increases proportionally to the viscosity:
« - 0M x R e-l
(27)
Analogously to Equation ( 23) for the turbulent regime,
the following applies to the transition flow regime [97]
a
P
( '
di-2/3 (28)
Here, the diameter ratio and power input have a great -
er influence than in the turbulent flow regime. Paradoxi -
cally, in contrast to the turbulent flow regime, the blend
time is shorter in larger vessels in which Re takes higher
values, which, according to Equation ( 27 ), has a positive
effect. This effect of scale also explains why processes
in the production plant may show better results than those
on the laboratory or pilot scale.
H 0 M 0 GENIZING NON - NEWTONIAN MEDIA IN
THE TRANSITION FLOW REGIME
The above - mentioned relationships apply exclusively to
Newtonian media. However, liquids with increasing viscos -
ity are typically non- Newtonian. The different non- New -
tonian properties were discussed on page 28 ff, of which
the structural viscosity probably plays the greatest role in
practice . \
The blend time for mixing shear thinning fluids in the
transition flow regime can be determined analogously to
the methods used for Newtonian media. However, the
Newtonian viscosity in Equation ( 28 ) , which is used for
determining the blend time, must be replaced by the term
for the viscosity at the vessel wall
2/ 3
P fe) dx
-2 /3
The viscosity at the vessel wall tjyy can be determined
(29)
with various correlations that describe the relationship
42
HOMOGENIZING
between the viscosity at the wall and the torque introduced into pumping and dispersing, as implemented in a coaxial
by the agitator : .
system with a central dissolver disc ( see Fig 22 and
page 114 ff.).
?? W ocil/t (30)
SCALING UP
Process intensification often involves building larger
If a further increase in viscosity means that the Reynolds 4 reactors to increase the output for a given batch time or,
number drops below a second critical value for the tran- in continuous processes, for a given residence time.
sition zone to the laminar regime where For processes limited by the blend time, it is desirable to
scale up keeping the blend time, constant over the
CTL , scales. The aforementioned correlations have shown that
Rec, TL - CTL ^ 03 ' CTT (31) Fig. 22 Positive displacement impeller: Paravisc with inner baffle
Nem for batch processes with changing viscosities (left); coaxial system with the blend time depends on many factors and particularly
the blend time for impellers used at a distance from the Paravisc and centrally mounted dissolver discs to disperse high-viscosity on the flow regime. When scaling up processes taking
liquids (right)
wall, even for large diameter ratios d2 / dx , increases very place in a laminar or a transitional flow regime, the blend
over- proportionately with the decreasing Reynolds num - time can often be kept constant at acceptable power inputs
ber ( see Fig. 21). Product mixing in the wall, bottom and
surface regions is no longer sufficient and non-homoge-
nized toroidal vortices may develop. Homogeneity is not
transfer reasons. Although helical ribbon impellers pro -
vide efficient mixing under purely laminar flow conditions,
they are associated with drawbacks in industrial applica -
and impeller speeds. However, in turbulent flow regimes,
Equation ( 23) indicates a significant increase in the neces -
sary specific power. For example:
reached within a finite time. Anchor impellers, which are tions. Owing to their low pitch and the resulting large
still used in these viscosity ranges, do not induce vertical number of helical turns, the blades are generally not stiff Reactors used for precipitation or pearl polymerization
mixing and thus lead to completely unacceptable process enough. Therefore the blade has to be attached to the of polyolefins or PVC have at least tripled their volume V
results ([37], [36] ). shaft at several points along the length of the impeller in the last 30 years, and a further increase in reactor dimen-
using cross- bars. These bars and the small wall clearance sions can be expected in coming years. To keep the same
The only impellers that operate satisfactorily in this means that neither outer nor inner baffles can be used. level of homogeneity with respect to temperatures or con-
range are the so - called positive displacement impellers. Numerous batch processes pass through a wide range centrations, the blend time must be kept constant when
These include helical ribbon and Paravisc impellers that of viscosities, and low - viscosity work steps often require scaling up. For geometric similarity and = const.,
have to be adapted to the individual process requirements incorporation of liquid components or entrainment of this requires an increase in the specific power inputs with
with respect to their number of blades, blade angle, and powdery solids from the surface. In this case, helical rib- P /Vx V 2/3. The situation is complicated by the fact that
diameter ratio. Generally, a small wall clearance is required bon impellers are insufficient because they only make the the greater volumes are generally achieved by using
in order to avoid near- wall dead zones, but also for heat- liquid rotate. In contrast, the Ekato Paravisc with its steep - a higher aspect ratio ( hx /dx > 1). The reasons for this lie
ly angled blades can also be built for high torques without in the limited dimensions for road haulage of prefabri-
the need for intermediate cross -bars so that efficient inner cated components and the smaller footprint, which may
10000 baffles can be installed from the top right down to the be the decisive factor for capacity extensions in existing
bottom anchor, which shortens the mixing process, even plants with spatial limitations, in particular.
:
Intermig . at low viscosities.
\ pitched blade turbine
l
1000
Mixing of media with very different viscosities can also
I result in significantly longer blend times, as shown in
Fig. 21. The use of an internal baffle together with the
Paravisc \
100 Paravisc (Fig. 22 ) can significantly shorten the blend time
\ helical ribbon compared to a helical ribbon impeller.
impeller
10 Further problems may arise with the helical ribbon impel-
1 10 100 1000 10000 100000 ler if solids or non-miscible liquid components have to be
Re
Fig. 21 Blend time characteristics for the transition and laminar flow
dispersed in more viscous liquids. This requires impellers
generating high shear rates e.g. dissolver discs. These,
however, cannot globally mix the contents of the vessel,
which is necessary in order to distribute the dispersed
regimes material. This problem is solved by separating the functions
43
 RHEOLOGY SUSPENDING SOLIDS
THE SCIENCE
DIMENSIONLESS NUMBERS
FUNDAMENTALS HOMOGENIZING
SUSPENDING SOLIDS
DISPERSING GAS/ LIQUID
DISPERSING LIQUID / LIQUID
HEAT TRANSFER the bottom and the solids reach the surface of the liquid;
MIXING SOLIDS however, there may be localised areas with higher con-
centrations that can be up to 1.5 times the mean value.
VACUUM CONTACT DRYING
In a continuously operated cascade of mixing vessels,
much higher degrees of homogeneity are necessary to 1S
ensure that the material can flow without interference.
The density of the suspension at the surface of the liquid
tsa
must be similar to that close to the bottom.
a 10

£ 2
A suspending agitator can be used for various purposes . For example, an off - bottom suspension 1,
may be sufficient for washing processes or dissolving solids, whereas a continuously operated DESIGN FUNDAMENTALS So
cascade of stirred tanks e. g. for crystallisation or for ore leaching, may impose the highest
O 01
requirements regarding the homogeneity of the suspension . CHARACTERISING THE MATERIAL SYSTEM 0.1 1 10 100 1000
The material data to be taken into consideration in the
Equivalent particle diameter [pm]
design of suspending agitators are:
Fig. 24 Typical particle size distribution in a RRS diagram
- density of the pure liquid
STATES OF SUSPENSION liquid until a theoretically homogeneous suspension is - difference in density between the solid and the liquid
reached. - viscosity of the liquid 0.150 300
Depending on the power input or the selected impeller, - particle size 0.125 Settling velocity ws 250
there are different states of suspension or homogeneities The homogeneity of a suspension is generally charac - - solids concentration 5 0.100
of solids over the height of the liquid in the vessel. In an terised by means of the standard deviation a of the solids
£ SettiTng power 200

off -bottom or partial suspension (Fig. 23, left ), the particles £ 0.075
concentration along the vertical axis and is defined as These factors are all included in the hindered settling
I
150
°
are lifted and suspended by the surrounding flow; how- follows velocity wss of the particle swarm . > 0.050 100 I
ever, some of the solid remains on the bottom. A complete
suspension is obtained when all the solid has left the The settling velocity of a single particle ws can be cal -
Is 0.025 Hindered settling 50 05
2
cn
velocitywgs
bottom of the vessel i.e. Zwietering ' s [152] ls-criterion is culated using the methods given in the relevant literature. o.ooo

". ‘
<J = (32) 0
fulfilled. This means that the entire surface of the particle , ii » J The hindering effect on the settling process due to the
0 100 200 300 400 500
Particle size [ pm]
600 700 800 900

is accessible and thus available for mass transfer pro- presence of several particles is quantified by the following
cesses. A further increase in the agitator speed improves relation: Fig. 25 Settling velocities and settling power as a function of the
the degree of suspension over the entire height of the Achieving a standard deviation of 0.5 is sufficient to particle size
satisfy most suspending tasks: there are no deposits on wss = ws (i- CvT (33 )
In order to maintain a defined degree of uniformity in
According to the empirical approach of Zaki [104], the the suspension, the agitator must transfer sufficient pow-
Increase in speed or power
exponent m is a function of the particle Reynolds number er into the liquid to counteract the settling power. This
agitator power is always a multiple of the settling power.
.. dv ws - p (34 ) The additional power required to circulate the liquid can
7 be neglected for most technical applications.
V V It lies in the range from 0.5 < Rev < 1300 and is given by
V

J
Because the settling power and thus the agitator pow -
m = 4,375 - ,Re 0’0875
“

(35 ) er are proportional to the hindered settling velocity, choos -

ing the design particle size is very important .
and thus takes values between 4.65 and 2.33.
Granular materials always consist of many individual
ENERGETIC CONSIDERATIONS particles with different sizes and shapes owing to their
If it is assumed that all solid particles in the liquid are origin, e.g. grinding. The particle size distribution is usu-
already uniformly distributed and all simultaneously start ally a straight line in a RRS diagram. For atypical particle
to settle under the effect of gravity, they release a "settling size distribution, as shown in Fig. 24, the values shown in
power ", which can be quantified by the relation: Fig. 25 for the settling velocity and settling power for
Partial suspension Complete suspension a specified vessel volume are derived ( Ap = 2000 kg/m3).
( lifting off the bottom ) (off-bottom, 1-s criterion) Homogeneous suspension
Psettle= wss - Cv - A p - g - V (36 )
Fig. 23 Different types of suspended states as a function of the rotation speed

44 45
 SUSPENDING SOLIDS

If the design were to be based on the largest particle - extremely large versus extremely small particle sizes CLASSIFICATION OF SUSPENDING TASKS
size, this would give the 35 -fold settling or agitator pow - of the solid 30
20
er compared to the design based on an 80 % mesh passage - low versus high solids concentration Based on the hindered particle settling velocity as a dif -
forthe example in Fig. 25. Our own experience has shown ferentiating criterion, suspending tasks can be divided TT 10

that, depending on the uniformity characteristic (gradient

in the RRS diagram), a design particle size with a mesh
If we specify a single uniformity criterion and then try
to divide the suspension tasks into different requirement
into the following categories ( see Fig. 27):
I 5
3
passage of between 80 % and 90 % produces reliable classes, it quickly becomes obvious that the different 1. SIMPLE SUSPENDING TASKS 2
results. scale-up rules only apply to some of these tasks and that When the hindered steeling velocity results in laminar I2
1
the search for a single, all- purpose scale -up rule is doomed flow around the settling particles, the suspension task is t
to fail. said to be simple. i 0,5

SCALE - UP ISSUES 1 °' 3

No other basic mixing task is associated with as many

different scale - up criteria as suspending ( Fig. 26). The rea -
son forthis lies in the different evaluation states of suspen-
sion on the one hand, and the extremely wide range of
For typical suspending tasks is the scale up criterion
usually in between a constant tip speed and a constant
specific power input. A possible approach to scaling up
is given in [70].
2. DEMANDING SUSPENDING TASKS
According to the comprehensive investigations con-
ducted by Einenkel [23], this category includes the major-
ity of all suspending designs.
| A2
o
g 0,1
•

ir
0,05
10
’ 2 3 5 102 2 3 5
ffclpm ]
103
1
2
Particle size •\

process - and product-specific requirements on the other: 3 . DIFFICULT SUSPENDING TASKS

Ap kg/m 3
Simple suspension tasks
- off-bottom suspensions versus essentially uniform Tests are advisable to avoid over- designing and to find
i 50
solids distribution the appropriate scale-up rule for the commercial scale 2 100 Demanding suspension tasks
- solid with extremely narrow versus wide particle size equipment. 3 250
distribution
4 " UNIFORM " SUSPENSION IS NO LONGER POSSIBLE
4
5
6
500
1000
2000
D Difficult suspension tasks

2 2
In this category, off -bottom suspension of the solids is 7 4000
H » Uniform« suspensions
no longer achievable
1 Voigt, Mersmann, 1995

2 Pavluschenka, Kostin,
Matveev
dJd2 > 3 - 10 P/V x d

PIVccdj
^- 02
10
'
generally the best solution. Suspension of the solids up
.
to the surface of the liquid is no longer possible Only
dissolving or washing processes are achievable in this
fc = Correction factor for volumetric concentration
5
category. Fig. 27 Degree of difficulty for suspending tasks
3 Kneule, 1983 /< > 10 P / V = const.

3 Voigt, Mersmann, 1995

dJd2 < IQ
-3 P/ V = const.
I 3
pie shape is easier to manufacture, does not quite reach

3 Weisman, Efferding,
Zlokarnik, Judat dJd2 > 3 -ir
2
P/ V = const.
^- 2
COMPARISON OF AGITATION SYSTEMS the excellent values of the Ekato Isojet; however, it is almost
as good.
10
° CHOICE 0 FIMPELLER
> \ tf *) P/V - const. I Fig. 28 presents the results of a model study at the 1- m3
3 Kraume, Zehner, 1995 ReKf i
Fig. 29 shows a comparison of the flow field and flow
5 scale for a typical task for hydrometallurgical applications velocities for the pitched blade turbine and the Ekato Vis -
3 Geisler, Buurmann,
Mersmann, 1993 djd2 < 3 -l(T
5
P/ V = const.
!i 3
that compares the impellers coprop. This illustration demonstrates that the Viscoprop
generates higher flow velocities close to the bottom, thus

I
4 Langhans, Liepe, - 0,13
P / Vozd
Weissgraber, Einenkel ^ ^ - four -blade pitched blade turbine facilitating lifting of solids off the bottom. As a result, the
103—106 P/ Vazd m
- Ekato Viscoprop Viscoprop needs a lower power input for suspending the
1
~
5 Einenkel, 1979 Re =
DC 10 1
'
- Ekato Isojet solid.
w SS d
P/ Voc rf - 0,38
6 Einenkel, 1995 C
< 2103 °
10 2 3 5 10
' 2 3 5 102
The comparison criterion was reaching an identical den- FURTHER OPTIMISATION MEASURES
- 0,55 p
7 Zwietering P/ V <
* di Scale-up factor n = —2— sity quotient or an identical standard deviation at two In addition to selecting a suitable type of impeller, there
dl , M measuring points, one of which was close to the vessel are other process engineering methods to minimise pow-
P/ Vozd - 0 75
2 ,
dp/d2 > 3 - 10
“

8 Muller, Todtenhaupt
^ bottom and the other at the surface of the liquid. er input.
l
< 104 *) P / V oc dx
9 Kraume, Zehner, 1995 ReK [ u = const .
The simply shaped pitched blade turbine requires In analogy to improving surface entrainment by tilting
P/ V oc dx i
-
9 Geisler, Buurmann, the baffles in the upper part of the vessel, baffles can also
> 310 4 2.7 times more power and has to transfer 3.1 times more
dJd
"

Mersmann, 1993 2 u = const.

torque than the flow- optimised Ekato Isojet. The greater be tilted in the bottom part of the vessel for suspending
r V /3 torque necessitates larger gearboxes and shafts and thus tasks. Such optimisation intensifies the circulating flow
“ J]
A

Rercf = —
v
(
•

U
•% • d\ ' cv '
ss
higher investment costs. The Ekato Viscoprop, whose sim - that lifts the solids off the bottom ( [22], [2]).

Fig. 26 Scale- up of suspending applications according to various authors ([23], [24], [35], [70], [ 141], [151])

46 47
 SUSPENDING SOLIDS

Starting on page 212, various alternatives for product
draw - off points in continuously operated vessels are de-
scribed. If the power input required for suspending, and
thus the agitator size, exceeds a critical value, under cer-
tain circumstances it may be more expedient to use
a vessel with a dip tube or a base -mounted draw - off point
instead of an overflow vessel. However, the dip tube must
be operated so that the rising velocity in the tube is high
enough to transport even the largest particles out of the
vessel.
Particularly for applications requiring large tanks, flat-
bottomed vessels are used for both reasons of process
engineering and economy. The shape of such vessels is
unfavourable because there is a higher risk of deposits
accumulating in the corners and thus a significantly high -
er suspending power is required compared to a vessel
with a dished bottom. In this case, a relatively easily con-
structed fillet near the corners can be used to reduce the
required power.
Impeller Power / volume
[kW/m3]
Pitched blade turbine 0.51
Ekato Viscoprop 0.22
Ekato Isojet 0.19
Fig. 28 Comparison of the required power input:
pitched blade turbine versus Viscoprop and Isojet
SPECIAL SUSPENDING TASKS
HIGH VISCOSITY AND FLOW ANOMALIES
Suspensions with a larger solid fractions often exhibit
higher viscosities and non- Newtonian behaviour. This
can also be due to additional constituents ( e.g. fibrous
material). Fig. 30 shows an example of an ore suspension
with Herschel -type flow behaviour ( see page 28 ff .) .
In this type of behaviour, the settling velocities of the sol-
id particles are correspondingly low so that the primary
mixing task is no longer keeping the particles suspended,
and thus changes from "suspending " to "homogenizing".
Depending on the application, pilot-scale tests may be nec-
essary to check the design and to avoid over- designing.
from a purely axial flow impeller such as the Viscoprop,
Torque/ volume
[ Nm/m3] which requires the least suspending power in the ungassed
state, to an Isojet- B or a Combijet. Another very important
26.6 aspect is flooding of the impeller. At the flooding point,
the flow beneath the impeller breaks down completely
12.5
and the majority of the solid accumulates in the bottom
8.5 zone. The flooding point is influenced not only by the flow
velocities, but also by the type of impeller. As a conse-
quence of the more stable behaviour of the various impel-
lers under gassing conditions (Fig. 31), the flooding point
is correspondingly higher.
Fig. 30 High viscous ore suspension
GASSING
Another special case is suspending with simultaneous
gassing. The gas injected at the impeller changes the flow
field underneath the agitator and slows the flow velocities.
This diminishes the energy input at the bottom zone, which
is required to suspend the solid in the first place. Fig. 31
shows that different impellers behave differently under
gassing conditions. The ratio of the flow velocity with and
without gassing is plotted against the ratio of the power
input from the gas and from the impeller. Depending on
the gas flow rate, it can thus be advantageous to change

1.00

Combiiet
" mu ,,,,
!y“/1ww/w / ^
///
" ' HUt / / / / /,
III;
0.80
/ /
UHlUllli / ,,,,, nit / / / , E
Hum / ,/ , £
5& jjj lllUlin / o 0.60
s
BSs
imt!!IPnii
:r I8 0.40
"Widefoir type
“"
ttHHUUUUliiiim
0.20 ^itched blade turbine
I /
I
Is 0 0.25 0.50 0.75 1.00 1.25 1.50
CD
Power input by gas / Power input by impeller [-]

Fig. 31 Decreased axial flow velocity underneath the impeller under

Fig. 29 Comparison of the flow field for Ekato Viscoprop Ileft ) and a pitched blade turbine (right) gassing conditions

48 49
 THE SCIENCE RHEOLOGY DISPERSING GAS/LIQUID
DIMENSIONLESS NUMBERS
FUNDAMENTALS HOMOGENIZING
SUSPENDING SOLIDS
DISPERSING GAS / LIQUID
DISPERSING LIQUID/ LIQUID
HEAT TRANSFER
Thus, Fick 's First Law can be used
MIXING SOLIDS
dc
VACUUM CONTACT DRYING m - -D A (37)
dz
to calculate the mass flux.

By integrating and using the simplification that the mass

transport resistance lies only on the liquid side, gives

Gassing of liquids is of key importance in the chemical and pharmaceutical industries as well as in
biotechnology. The aim is to achieve the highest possible mass transfer between the dispersed
gas and the liquid .
Gas / liquid applications can be divided into two basic
groups. The first group deals with technically pure gases,
such as hydrogen, chlorine, oxygen, ammonia, epoxide,
or synthesis gas. Typical reactions include the correspond-
ing hydrogenations, chlorinations, oxidations, aminations,
alkoxylations, and hydroformylations [42] , The second
group relates to processes in which the reactive gas is
a component of the gas mixture. This includes oxygen
in air and carbon dioxide in flue gas. Typical processes
are aerobic fermentation, oxidation in air ( e. g. synthesis
of terephthalic acid), precipitation of calcium carbonate,
and flue gas desulphurisation.
In addition to mixing vessels, which will be discussed
in detail below, various othertypes of apparatus are used,
e.g. injectors, packed columns, and bubble columns .
c c
Fig. 32 Two - film theory of Lewis and Whitman. The subscript "i" stands for interface. [75]
Since the reactive gas is generally consumed within the
liquid phase, this requires transfer of mass from the gas
to the liquid phase.
MASS TRANSFER COEFFICIENT
In analogy to heat transfer ( page 68 ), the resistance
to masstransfer can also be divided into individual resist-
ances. Processes occurring during mass transfer can be
simplified using the two film theory developed by Lewis
and Whitman [75] in which the two boundary layers of
the phases are regarded as steady state and non-variable
(Fig. 32 left ) . It is assumed that the concentration of the
exchanging gas on both sides of the phase interface cor-
responds to that of the equilibrium state.
Transport
C G= C G
cL
mass-transfer
resistance
Donor phase Receiving phase
m= A ( c* ~ c) (38 )
in which the ratios of the diffusion coefficient D and the
boundary layer thickness 5 are combined into the mass
transfer coefficient on the liquid side KL-
This simplification is admissible because diffusion in
the gas phase is generally several orders of magnitude
faster than in the liquid phase, so that the latter becomes
the rate - determining transport resistance ( Fig. 32 right).
The key parameters influencing the mass transfer coef -
ficient KL are the viscosity, which affects the boundary
layer thickness, the diffusion coefficient of the gas in the
liquid and the bubble size. According to Calderbank [10],
the following correlations apply for bubble sizes
z/g > 2.5 mm, respectively d& > 0.8 mm for smaller bubbles
Fig. 33 Flat blade disc turbine for gassing
> 2.5 mm: Sh = 0 ,42 - Scl / 2 - Ar 1' 3 (39)
chemical reaction) and thus the concentration in the liquid
afg > 0.8 mm: Sh = 0,31 - Sc1/ 2
Arm (40) phase is almost zero, the driving force for mass transport
rises on account of the increasing concentration differ-
Thus the mass transfer coefficient on the liquid side is ence c* - c. This effect can be used to accelerate reactions
determined by the physical properties of the media. These, that are limited by mass transfer because, in a series of
however, cannot be influenced by mixing technology. consecutive chemical reations the slowest step is always
rate determining. One of the methods used to assess
The influence of mixing technology is thus limited to whether there is indeed a mass mass transfer limitation
the bubble size, bubble residence time, gas content and is to apply the Hatta number. This describes the ratio
thus the interfacial area. between the reaction rate without mass transport limita -
The driving force behind masstransfer is the difference tion and the mass transfer rate, and is expressed as follows
in concentrations c* - c. This is particularly apparent with
gas mixtures containing an inert fraction, e.g. air. Mass
transfer causes a drop in the partial pressure of the reac -
i
Ha = — J — - k - (vc*
— )n - i D (41)
tive gas in the gas bubble. This has a direct effect on the
saturation concentration c* at the phase interface because,
kL V « +i ’
according to Henry ' s Law, the solubility is proportional Hatta numbers < 0.3 indicate reactions that are com-
to the partial pressure. For processes using technically paratively slow. The process rate cannot be influenced
pure gases, this is generally not a contributing factor, by improving mass transfer. For Ha values > 0.3, i. e. for
unless the process temperature is high enough to increase relatively moderate to fast reactions, the process can be
the vapour pressure of one of the components to such an accelerated by improving masstransfer.'
extent that the partial pressure of the process gas is The basic principles for scaling up gas/ liquid reactions
greatly reduced. These relationships are often exploited are described on page 102 ff. and methods of measuring
in practical situations by increasing the process pressure mass transfer on page 84 ff.
and thus the partial pressure of the gas. If the consump -
tion rate of dissolved gas is very fast ( e.g. due to a fast

50 EKATO 51

i
 DISPERSING GAS / LIQUID

1.2 Increasing speed with constant gas feed rate

1.0

0.8

% 0.6
£ 0.4 i+ = 0.5
Concave impeller

\
V
—
Ekato -Combijet

Ekato-Phasejet

Flat bladeldisc turbine

0.2

0.0
0.00 0.04 0.08 0.12 0.16

Gassing number Q Partially gassed Bubble column Start of bubbles Bubbles completely Recirculation of
bubble column, spreading underneath distributed gas bubbles
impeller flooded the agitator
Fig. 34 Drop in the power number for different impellers under gassing Fig. 35 Phasejet with rotating sparger and three Combijets
conditions Determined by gas flow rate Determined by the agitator

AGITATOR POWER Henzler [47 ] gives the following relationship for calculat- Fig. 36 Hydrodynamic conditions during gassing
ing the Ne value of a flat blade disc turbine in the gassed
Designing an agitator requires knowing the power num - state, Neg. - After passing the flooding zone, the gas bubbles are Unfortunately, there is no standard criterion in mixing
ber of a gassing impeller. Fig. 33 shows the complicated distributed over the entire cross - section of the vessel technology so that the measured values always depend
interaction between the flowing liquid and the gas being Ne0 + 18 4.6 Q125 above the agitator. The agitator acts as a gas distribu - on the measuring method. This must always be taken into
dispersed by a flat blade disc turbine. Depending on the Neg - Z (42) tor in a bubble column. account when comparing measurements by different
dA1-14
Q(
agitator speed and the gassing rates, blade - tip vortices - The gas bubbles are distributed throughout the authors.
1+ 136
with small bubbles form in the negative pressure zones 4 vessel.
with adjacent cavities and, in the flooded state, large gas NEQ: non- gassed Ne value. The expression is valid for Nienow [93] proposed the following flooding limit Qp
cushions. The cavities and gas cushions generated by Only small mass transfer intensities are possible with for flat blade disc turbines:
0.07
a standard flat blade disc turbine with six blades reduce
the flow resistance so that the power number of this impel-
ler drops steeply under gassing conditions, see Fig. 34.
Therefore, Smith developed a concave impeller whose
Fr
d2 3
d\
0.2 <
^
dx
< 0.42;
^
di
> 0.7; i?e > 104 (43) the first two states.
Once the agitator distributes the bubbles in the lower
part of the vessel, the mass transfer rate starts to increase.
Qp = 30 - Fr \
4
. 3.5
(45 )

blades are shaped like half -pipes. This results in a less
pronounced gassing-related power drop. The specially
contoured blades of the Ekato Phasejet ( Fig. 35) match the
gassing- induced cavities even better so that the drop in
This is intensified even further when the bubbles are dis - The flooding limit calculated with this equation is about
In a literature review with more than 40 references, tributed throughout the vessel and the recirculation rate 1/3 higher than Henzler ' s [47] method:
Midoux [86] evaluated the dimensional relationship of of the bubbles increases .
Michel and Miller [85], who derived a dimensional equa - /4 0.14
tion for the agitator power in the gassed state Pg
0.21 Fr 21'
^
1.3
• Fr1M d2
'

^
1.5
(46)
the power number is even smaller.
4 0.56
- n -4
0.45
(44 )
FLOODING (+- 2.04) (+-125)
7
The sparger ring can also be omitted in the Phasejet
without an additional power drop. In this case, fresh gas
is introduced through a simple feed pipe underneath the
Pg = 0.34 -
^
Good agreement was found between the values calcu-
At the latest by the time the aforementioned gas cavities
reach the next impeller blade, the gas bubble dispersion
Reducing the constant in Equation ( 45 ) to 20 results in
good agreement between the two relationships. Fig. 37
mechanism is disturbed and the impeller is flooded [145] , shows the flooding limit according to Equation ( 46) for
impeller. The possibility of dispensing without a sparger lated with Equations ( 42 ) and ( 44). However, the point at which this state is reached is not the flat blade disc turbine .
ring is of particular interest for processes with demanding easy to determine because it is very difficult to measure
cleanability requirements ( e.g. fermentation, pharmaceu- it directly. This is why indirect measuring techniques are The correlations between Q and Fr ( e.g. ( 45), ( 46)) can
tical production and precipitations during which the many HYDRODYNAMIC CONDITIONS DURING generally used. These include [145]: also be expressed as the ratio between the power trans-
small holes are susceptible to plugging) . GASSING ferred from the gas to the liquid
- visual assessment of the escape of large bubbles
The following states occur as the agitator speed is slowly - the power number starts to increase again P = Ap . g - vsg - V (47)
increased at a constant gas flow rate (Fig. 36): - the mechanical forces increase ( page 130)
- The agitator is flooded, and the bubble swarms rise in - change of flow in the vessel and the agitator power. This means that a flooded state
the centre of the vessel . - pumping efficiency occurs when the gas - induced flow dominates over the
- mass transfer efficiency agitator power.

52 53
 DISPERSING GAS / LIQUID

BUBBLE SIZE
The bubble size is governed by the hydrodynamic op -
erating state and the local turbulence events in the gas /
liquid mixture. The bubbles close to the impeller are gen-
erally small owing to the high local energy dissipation.
In coalescing systems, the bubble diameter can increase
creases. For otherwise identical conditions, the gas content Gassing with air leads to an oxygen partial pressure of
can be up to twice as high in a non - coalescing system. 0.21 bar, and in combination with the molecular weight of
If the mean bubble size and the gas content are known, and 32 g/mol for oxygen, this gives an oxygen saturation concen -
assuming that the gas bubbles are spherical, the specific sur- tration c * of
face area between the gas bubbles and the liquid can be de-
termined with the following relationship: c* = 1.1. 0.21. 32
^I = 7.4
^ (55)

by a factor of 6 to 7 in regions that are a little further away
from the impeller and where the ratio between the lo -
cally dissipated power and the mean power can drop to
values as low as 0.1.
However, adding substances that lower the surface ten-
sion, such as electrolytes, can essentially limit coalescence
and the resulting bubble size distribution in the liquid is
significantly narrower, see Fig. 38, [65],
GAS CONTENT
The volumetric fraction of gas dispersed by the impeller is
an important parameterthat can be used to determine the
icr1 2 5 ° 2 5 surface area between the gas and the liquid where mass
10 10
’ transfer can take place. The literature ( [10], [138], [147] ) reports
Fr = n2 d /g
2 - not only complex relationships that include material proper-
ties such as the surface tension, but also more simpler rela -
Flat blade disc turbine djd tionships. Parameters include the specific agitator power as
^ well as the superficial gas velocity.
hA
/dx = 1 0.42
A 6 - £g
a =—
V {^BK (5V
MASS TRANSFER
Because the parameters governing mass transfer influence
each other so that measuring the interfacial area is very
complicated and because the saturation concentration of
the gas in the liquid depends on the operating conditions,
the volumetric mass transfer is measured in most cases.
The volumetric mass transfer capability &Lacan be meas-
ured more easily and allows a quantitative evaluation of
mass transfer, irrespective of the gas solubility. There are
no correlations available that include all variables influenc -
ing mass transfer and which would enable calculations to
be made with sufficient accuracy without the need for tests.
Measurements with real systems on the pilot scale
( page 84 ff.), can be used to determine the relationship
If it is assumed that the oxygen concentration is 3 mg/I,
there is a concentration difference of 4.4 mg /I, and thus
mass transfer of
O
m = kLa - Ac = U 25 8 2 (56)
h - m3
is possible.
INFLUENCING FACTORS
A significantly higher mass transfer can be achieved if
coalescence is diminished by the presence of salts [77].
This is attributed to the smaller bubble size and the higher
gas content, which greatly increases the interfacial area.
However, adding small amounts of salts in the ppm range
.
can also cause the opposite effect For example, adding
defoaming agents (surfactants) reduces mass transfer by
more than 50 %. This is not directly due to the changed
bubble size but to adsorption onto the surface of the bub-
bles and the resulting additional interface layer that acts
an extra barrier to transport.

• - sx - t
^
H20/air 0.27 A comparison of the calculated and measured values shows kL a-c ) (52) Another important parameterthat influences mass trans -
1 0.2 that the available relationships can only be used to obtain fer is the viscosity. A higher viscosity diminishes mass
the magnitude of the gas content. in which the geometric arrangement and the type of mixing transfer and the bubble size distribution becomes very
Fig. 37 Influence of the diameter ratio on the flooding limit system have to be taken into account. Mixing influences the wide. Large bubbles rise quickly to the surface of the liquid
eg = VS kLa value mainly via the interfacial area a by generating the without providing a significant contribution to mass trans -
(48 ) smallest possible gas bubbles: port, whereas small bubbles with a longer residence time
become depleted if there is a significant fraction of inert
0.05 M
salt solution
0.10 M
salt solution
0.20 M
salt solution
In contrast, by determining the relationship dg /(^impreller ’^gas)’ ^gas ~ ^sg (53) gas with respect to reactive gas. In addition, the flooding
limit is shifted by elevated viscosities to much lower Q
0.6 0.6 0.6 £g — c - sy - vss (49) Van't Riet [139] evaluated 76 studies that had been per- values .
I formed on the pure water/air system and then proposed the
0.4
£“
0.4 0.4
real material systems can be extrapolated from model meas- following correlation:
1 0.2 urements with a high accuracy . SELF - ASPIRATING GASSING
&L a = 0.412 • £ 0' 40 ttgg.50
0.2 0.2
ELI
The following relationship applies to a standard geometry (54 )
o 0 and water: Mass transfer between an injected gas and the liquid
1 2 3 1 2 3 2 3 An overview with different sets of parameters is given in is generally incomplete, i.e. at least some of the gas enters
Bubble diameter d [mm ] £g = 0.65 - e
0'2
- Vgg.55 (50) Gezork [36]. the headspace of the vessel where it is no longer available
^ for mass transfer. Accumulation of gas in the headspace
close to the impeller In non - coalescing systems, the bubble size can be 10 times EXAMPLE CALCULATION increases the Vessel pressure so that, if no countermeas -
at a distance from the impeller smaller than in coalescing systems. The bubble rise velocity For a power input of 1 W/ kg and a superficial velocity ures are implemented, the supply of fresh gas has to be
is then significantly smaller and circulation of the bubbles of 0.03 m / s, the above equation gives a &La value of reduced to prevent the vessel pressure exceeding the
Fig. 38 Bubble size close to the impeller and further away; .
with the liquid dramatically increases This leads to greater 7.1 10-2 s-1. permissible value. The simplest method is to release gas
transition between coalescing/non- coalescing bubbles residence times and thus to a higher gas content. Expressed The Henry constant for oxygen in water at 30 ° C is from the headroom. This is frequently used for process -
qualitatively, this means that the influence of the specific 1.1 mmol/( l bar). es using gases with inert components (e.g. air ) because
power increases, whereas that of the superficial velocity de- the target gas is already depleted and recirculation is not

54 55

_
 DISPERSING GAS / LIQUID

worthwhile. However, for technically pure gases, there is complex - and thus more expensive - external circuits.
a great interest in recycling the gas back into the process The main application field for such systems includes hy - i

in order to reduce costs. drogenation ( e.g. production of aniline and sorbitol) and
all gas/ liquid reactions in which technically pure process
In older equipment, in particular, surface gassing is still gases are used.
used. Such vessels are equipped with an impeller installed oi
close to the surface of the liquid. This impeller mixes gas Similarto forced gassing, the quantity of gas, expressed 1
.
in the headspace back into the liquid Because the effi- by the superficial velocity, greatly affects mass 0.1
Ekato Gasjet
ciency of such systems is relatively low, many reactions transfer ( 52). Therefore, the goal is to achieve the highest I /
turbine 2
remain limited by mass transfer. This is a problem, par- possible gas aspiration rate for a given agitator power.
a
ticularly in view of the ever increasing size of reactors, Fig. 40 compares the aspiration rate of various types of
turbine 1
because as the volume Vincreases, the surface area of turbines.
the liquid A decreases according to
The pumping behaviour of the turbines is characterised 0.01
A J_ by the dimensionless gassing number 0.1 1 10
~
V dx (57)

Q= l Pressure number <

PGas
relative to the vessel diameter d \ . n-d \ (58)
Fig. 40 Aspiration rate of various self-aspirating turbines
Alternatively, a compressor or an external liquid circuit and the pressure number \p
can be used. The former feeds the gas from the headspace
2 - g - hs
back into the base of the vessel via an external pipe. Gas =
The latter solution uses a pump to transport process ^ n
2
- dl (59)
10000

fluid through an external circuit. An additional injector

operating like a water- jet pump sucks the gas out of the
headspace and feeds it back into the liquid. Both solutions 5000

require additional fittings, thus leading to greater invest-

ment costs, more maintenance, and larger spatial require -
Turbines 1 and 2 are closed multi- blade impellers with
an aspiration slot for the liquid.
1
ments.
The mass transfers achieved by variousturbines alone
2000
A cost- effective alternative is combined gassing (Fig. 39 ). without external gassing in a non - coalescing material
system are compared in Fig. 41. The Ekato Gasjet achieves
This concept has become firmly established for use good values owing to the high gas throughput rate with-
with technically pure gases such as H2, 02, CO, EO ( eth- out flooding phenomena. 1000
ylene oxide) etc. In the standard version, the fresh gas is 10
0 2 5
finely dispersed by the primary disperser Ekato Phasejet
P/V [kW/m3]
and some of it dissolves. The remainder rises into the
headspace where it is continuously recirculated into the
liquid via a self-aspirating turbine - the Ekato Gasjet. This
•Ekato Gasjet turbine 1 •turbine 2
affords high mass transfer rates as well as fast and com- Fig. 41 Mass transfer achieved with different gassing turbines
plete conversion of the gas.
The Gasjet - a self - aspirating gassing turbine - acts like
an internal compressor. Provided that the negative pres -
sure generated by the tip speed of the impeller blades is
high enough, the gassing impeller aspirates gas out of
the headspace of the vessel via holes in the flange coupling
of the hollow shaft and feeds it back into the liquid. The
gas is re -introduced into the zone with a high local pow-
er density so that small gas bubbles can form, thus pro -
moting mass transfer. This means that the mass transfer Fig. 39 Combined gassing with Phasejet (bottom ) and Gasjet (top );
efficiency of combined gassing is superior to the more a ) Agitator layout, b) Start of gassing as the agitator picks up speed,
c) Regular gassing
I

56 57
 RHEOLOGY DISPERSING LIQUID / LIQUID
THE SCIENCE
DIMENSIONLESS NUMBERS
FUNDAMENTALS HOMOGENIZING
SUSPENDING SOLIDS
DISPERSING GAS / LIQUID
DISPERSING LIQUID / LIQUID
HEAT TRANSFER
W 0
MIXING SOLIDS Emulsion-based products Emulsions as auxiliary or
reaction products
VACUUM CONTACT DRYING o o o
o_o o_o Herbicides Chemical industry
w o oo 0 o O
Cleaning agents Photographic industry
oil-in-water (O/W) ( OZW )-in-oil
( O/W/O )
Cosmetics Paper industry

Pharmaceuticals Wood industry

Liquid / liquid dispersions are mixtures of two or more non - miscible liquids . Such mixtures
ow
are prepared and used in many industrial applications. The chemical industry often employs Foodstuffs Mining, petrochemical industry
temporary mixtures of liquids to control extractions or reactions . In addition, dispersions with o o o
long -term stability are used in the pharmaceutical, cosmetics and food industries. Such stable 0.0 o o Paints Textile industry

liquid dispersions are known as emulsions . The following chapter discusses mechanisms of stable o o o oo Lubricants Mechanical engineering
emulsions .
water-in-oil (W/0) (W/0)-in-water Bitumen Road construction
( W/O /W )

Fuels Construction industry

Fig. 42 Types of emulsions
Fig. 43 Application fields of emulsions

LIQUID / LIQUID DISPERSIONS: EMULSIONS
DEFINITION
A dispersed system of two or more liquids that are nor-
mally immiscible is known as an emulsion.
One of the liquids in an emulsion is finely dispersed
(the dispersed phase) in a second liquid, which is assumed
to be continuous. This surrounding external liquid is thus
called the continuous phase of the emulsion.
TYPES OFEMULSIONS
An emulsion consists of a water- attracting ( hydrophilic)
liquid and an oil- and fat -attracting ( lipophilic) liquid [114].
These liquids are completely or partially immiscible.
If the oil phase is dispersed in the aqueous phase, the
oil phase in the resulting dispersion is present in the form
of small droplets. This is called an oil-in- water emulsion
( O/ W emulsion).
Conversely, the water may be dispersed in the oil or in
general terms, a hydrophilic liquid dispersed in a lipophilic
liquid. Thus the water is now the dispersed phase in this
water-in- oil or W/0 emulsion.
If an emulsion ( e.g. O/ W ) is then dispersed in a third
liquid ( e.g. oil), a multiple emulsion is obtained. These are
referred to as an oil-in-water-in-oil emulsion (O/ W/O emul-
sion ), a water -in - oil - in - water emulsion ( W/ O / W ) etc.
A detailed description of various emulsification process -
AREAS OF APPLICATION
Emulsions are used in numerous industrial processes
.
as auxiliary ingredients orfinished products Fig. 43 gives
.
an overview In the form of commercial products manu-
factured by the chemical, pharmaceutical and cosmetic
industries, they have become an essential part of our
daily life (Fig. 44).
PRINCIPLES OF MECHANICAL
EMULSIFICATION
The quality is affected by many factors, e.g. composi-
tion of a product, choice and purity of the raw materials,
or composition of the emulsion (fractions of dispersed
and continuous phases); however, one of the most im -
portant is the particle size distribution of the dispersed
phase. This means that the emulsification process should
always aim to achieve the smallest possible droplet size
for the dispersed phase together with a very narrow par-
ticle size distribution.
Owing to their thermodynamic instability, the finely dis-
persed droplets have a tendency to coalesce or to separate
out of the emulsion. This process is induced by various
phenomena so that when input of mechanical energy
ceases the emulsion always separates into the aqueous
and oily phases again (except in case of a phase inversion).
An overview of these phenomena is given in Fig. 45.
Emulsions have to be stabilised as quickly as possible.
This is achieved by additives, such as emulsifiers, whose
PROCESSES OCCURRING DURING MECHANICAL The general processes taking place during droplet
EMULSIFICATION breakup were described in detail by Karbstein [57] and
Fig. 46 illustrates the processes taking place during Stang [126], based on work by Walstra [142]. The forces
mechanical emulsification. maintaining the droplet 's shape result from surface ten-
sion. These forces can be characterised by the surface
A suitable process ( e.g. stirring) is used to produce a pre - tension resulting from capillary pressure.
emulsion with a continuous and a dispersed phase. The capillary pressure is calculated as:
Depending on the system, the emulsifiers are dissolved
either in the continuous or in the dispersed phase. The
2 (60)
Pc =a ~
mechanical energy input required to produce the pre - r
emulsion is comparatively small.
The ratio between the forces (stresses) deforming the
This pre - emulsion is then finely emulsified by means droplets and the forces maintaining the shape ( capillary
of a suitable dispersing technique (e.g. rotor/stator ), which pressure) is the Weber number We. The Weber number
generally requires a high energy input. This is associated is defined as the quotient of the shear stress r acting at
with various physical phenomena that deform the droplets the droplet interface and the capillary pressurepc of the
of the raw emulsion so strongly that they break up. undeformed droplet [126]:
If the finely dispersed emulsion is stabilised quickly
We = —
T r r-
enough, the resulting droplet size can be kept almost con- 2 - cr (61)
Pr
stant and the emulsion remains stable. However, if sta -
bilisation is too slow or insufficient, there is a risk of coa - In order to break up the droplets, the corresponding
lescence into larger droplets. deformation forces ( stresses) must act for a critical de-
A successful droplet breakup requires first overcoming .
formation time tfef crjt The critical deformation time es-
the surface tension forces in the droplets that maintain sentially depends on ' the viscosity of the dispersed phase
the droplet shape. Breakup is caused by flow-induced f /( j, the capillary pressurepc and the acting shear stresses
deformation effects that are transferred via the continu- r ( [126], [143]):
ous phase. These effects can be caused by both inertial
Pd
and shear forces ( turbulent emulsification) as well as by
extension forces ( laminar emulsification). If a droplet is
*def, crit -
T ~
Pr
(62)

es is given on pages 313 and 319. chemical and physical properties hinder coalescence of deformed beyond a critical value, it will break up [57 ].
the droplets in the emulsion .

58 59
 DISPERSING LIQUID/ LIQUID

Industry Types of emulsions Thus: According to Walstra [144], droplets in a turbulent flow
Examples Special requirements
can be influenced by the formation of vortices in two dif -
ferent ways. If the droplet diameter remains smaller than
Foodstuffs Mainly O/W emulsions Mayonnaises, sauces,
salad dressings,
Visual / sensory Gloss, mouth feel ^elam — ^elam crit , anc
^ hlef — ^def, crit (66 )
that of the vortex, the droplet follows the movement of
emulsions for candy bars only deformation and thus no droplet breakup and the vortex. In this case, mainly shear forces are acting on
Rheology Texture, flow properties, the droplet. Thus, analogous to deformation in a laminar
spreadability
(67) flow, the maximum stable droplet size can be estimated

Stability
^elam ^ ^elam , crit
^ ^def ^def
an < , crit
using the volumetric power density Py = P / Vas follows:
Physical, microbiological
deformation time exceeds the critical value, and droplet
Cosmetics/ O/WandW/O emulsions Creams, lotions, ointments, Visual Gloss, colour
breakup occurs.
^max KC - P y-0.5 .T?c 0.5
, -
(69 )
pharmaceutical suppositories, mascaras,
industry foundations, sunscreens DROPLET BREAKUP IN TURBULENT FLOW According to this equation, the droplet breakup effi -
Rheology Texture, flow properties Industrial emulsification processes are predominantly ciency increases with increasing viscosity of the continu-
Distribution of active
carried out in turbulent flow regimes. In contrast to lam - ous phase [126].
Homogeneity inar flow, there are several different forces that can cause
substances
droplet breakup under these conditions . If the droplet diameter is about the same size as the
Chemical industry O/W and W/O emulsions Paints, varnishes, resins, Homogeneity Distribution of pigments vortex, more local pressure fluctuations are acting on the
adhesives, plastics
As already mentioned, the dimensionless Reynolds droplet due to fluctuating velocities in the surrounding
number describes the flow regime: liquid. These pressure fluctuations induce inertial forces
Fig. 44 Examples of emulsions that deform the droplets and, if the critical deformation
Re =
- - _v d
v d p
(68) time is exceeded, breakup occurs.
Droplet breakup will therefore occur when the Weber DROPLET BREAKUP IN A LAMINAR FLOW il v
number exceeds a critical value or if the critical deforma - In laminar flow zones, the disruptive force acting on a
tion time is exceeded: droplet with diameter d is essentially the shear stress r. Mixer Emulsification equipment
It acts as a tangential stress against the capillary pressure,
We > Wecrit ; tie{ > tdef> crit (63) which is maintaining the shape. Thus the laminar Weber
number can be defined for Newtonian flow behaviour as mechanical energy mechanical energy
follows [126]:
x d
WeXam = 4 (7
, with x = TJC f - (64 )

Mechanisms causing physical instability

where rjc is the viscosity of the continuous phase and y is
continuous
the shear rate.
phase

cation
°o° % as &J8 The laminar critical Weber number WeXam crit describes
o „ o°| ^ the maximum droplet size dmax existing under the given
flow conditions: emulsifying
agent
r '
dmax
^lam, crit ~
4 er (65)
Phase inversion
The critical deformation time rdef crit can be calculated
Coalescence Ostwald ripening
using Equation ( 62). dispersed
phase

o
09 Oo o
oO
, C O
oO o ° o o
o °
Fig. 45 Causes of instability in emulsions
Fig. 46 Phenomena during mechanical emulsification [114]

60 61
 DISPERSING LIQUID/ LIQUID

Analogously to droplet breakup in a laminarflow, a tur - A detailed description of the aforementioned process is
bulent Weber number can be defined, which allows the given in [57 ]. 400
following statement:
The Sauter diameter is often used to define the maxi-
mum droplet size. The Sauter diameter describes the
Wetvxb > ^ eturb, crit and *
def > ?def, crit (70)
diameter of fictitious particles whose total volume and 300
deformation time exceeds the critical value, and droplet surface area are identical to those of the real particle
breakup occurs. swarm. The Sauter diameter is defined as:
t
The stresses z acting on the droplet are thus [142]:
d32 -
6 Vp
A (76 )
I2
2 200
T - P c ' iv' ) (71)
1
Methods such as laser diffraction or microscopy are
where v' is the fluctuating velocity of local fluid elements. generally used to characterise emulsions or to measure
The turbulent Weber number can thus be defined as particle sizes.
follows [126 ] : 150 Fig. 48 Fine emulsification of a raw emulsion generated in a vessel
DISPERSION TECHNOLOGYTO GENERATE EMULSIONS 0.2 0.5 1 2
Pciv' f - d PP - p - d 5/ 3
Wetaxh = —
Pc
T
4-o
C-
a ^ (72)
The appropriate emulsification technique must be se-
lected depending on the requirements for the emulsion
orthe product composition and properties. In addition to Impellers
Specific power Py [kW/m3] The theoretical number of passes of an emulsion
through the homogenizer is expressed as the BTO num -
ber ( batch turnover number ):
which is used to derive an expression for droplet breakup the requirements forthe product and the manufacturing
in a turbulent flow [132]: process, economic aspects also play a role. 1 Interprop exc. 0.40 V - tboUK /
2 Propeller 0.43 BTO =
In the majority of industrial processes, emulsification (78 )

d oc
n3/
5
is carried out exclusively in zones with turbulent flow.
3 Propeller
4 Pitched blade turbine
0.33
0.31
^ Batch
/5 (73) Thisappliesto both mixing technology as well as to emul - 5 Interprop 0.50 During continuous emulsification with rotor/stator sys -
sifying with a rotor/stator or in high - pressure homogeniz- tems, the emulsification efficiency is governed not only
50-I vessel with dished bottom, hjd = 1
In the equations formulated above, the shape - maintain - ers. W/O emulsion, 10 vol.% ^ by the mean power density Pv , but also by the mean
ing pressure is represented by the capillary pressure alone 4 baffles ( except for 1) residence time % of the emulsion in the dispersing gap
[132]. According to Arai et al. [6], these relations are only EMULSIFYING IN MIXING VESSELS [66] ,
valid for dispersed- phase viscosities ( rjd) up to 10 mPas, In the case of mixing processes, emulsification is Fig. 47 Dispersing efficiency of various impellers
for which only the value of the turbulent Weber number brought about by the power introduced into a limited Thus the droplet size decreases with increasing resi-
is critical for droplet deformation and breakup. At higher volume inside the vessel. This produces a coarse to fine- EMULSIFYING WITH R 0 T 0 R / STAT 0 R SYSTEMS dencetime of the emulsion in the dispersing gap because
values of the shape -maintaining forces become more ly dispersed emulsion, depending on the power input. A rotor/stator system can be used for emulsification the probability of the droplet being present in zones with
important owing to shear stress resulting from the inter- The attainable droplet diameter directly depends on both by homogenizing a previously formed raw emulsion high power density, i. e. in the presence of high- energy
nal flow induced in the droplet. Thus the Ohnesorge num - the power input and is calculated with the follow relation: ( see Fig. 48) and by direct mixing of dispersible and con- vortices, increases.
ber Oh must be taken into account as a second quantity tinuous phases in a rotor /stator. The pre - emulsion is
to describe deformation due to inertial forces in a turbu- d32 = C - Py (77) formed in the rotor/stator or just upstream of it. This is Smaller power densities are associated with a smaller
lent flow regime [124]: followed by formation of the fine emulsion in the shear number of the high - energy vortices that are necessary
The exponent b can take values between -0.4 for low gap of the rotor/stator. for breaking up droplets in turbulentflow. Thusforsmall
viscosities of the dispersed phase and -0.25 for high power densities, the residence time must be correspond-
^mL
3
>
Oh = 7d
'
-Pm (74 ) viscosities [57]. Ideally, rotor/stator homogenizers are flanged onto the ingly increased to achieve a sufficiently small droplet size.
(7
vessel so that the components can be added directly into
If the viscosity of the dispersed phase increases, the Fig. 47 shows the Sauter diameters obtained with the shear zone. The emulsion being homogenized is cir- Alternatively, the BTO number can be increased ( more
Ohnesorge number also increases. The maximum drop - a model emulsion using various impellers and different culated back into the vessel via the homogenizer. In this frequent recirculation through the homogenizer) to achieve
let diameter thus depends not only on the introduced specific powers. respect, the method is similar to continuous emulsification a larger number of passes through the zone with high
power density, but also on the viscosity of the dispersed in which the raw emulsion is fed to the homogenizer via power densities and thus to more frequent contact with
phase. According to Karbstein [57 ], the dependency of a feed and a discharge pipe. The only difference to con- high- energy vortices and a smaller droplet diameter.
the maximum droplet size at very high Ohnesorge num - tinuous emulsification is the possibility of passing the
bers is given as follows: emulsion through the homogenizer more than once . Both methods can be used to obtain a particular target
droplet size, depending on the power density.
4
< iax \ Oh —> eo KPc
V2
-4J4 - pfm (75 )

62 63

Instead of using the power density as the characteristic
comparative parameter for continuous emulsification,
Karbstein [57] recommends the volumetric energy input
Ey, which is the product of the mean power density and
the mean residence time:
^
EV = PV .TK (79 )
In analogy to the power density, the energy input per
volume unit is known as the energy density. If the condi-
tions for droplet breakup are fulfilled and if the residence
time influences the degree of breakup, the mean droplet
diameter can be determined as a function of the energy
density as follows [113]:
-h
dn - C - Ey (80)
EMULSIFYING WITH HIGH - PRESSURE HOMOGENIZERS
High-pressure homogenizers are used when the re-
quired droplet diameter lies below the minimum value
attainable with a rotor/stator homogenizer, for example.
The attainable droplet size may even be in the nanometre
range.
High - pressure homogenizers operate on different
principles. A distinction is made between homogenizing
systems with radial diffusors, nozzles and opposed jet
dispersers.
A special type of disperser is used in the opposed jet
technique, which is covered by the Ekato Nanomix patent.
The homogenizing pressure at constant viscosity is
adjusted using the flow rate of the medium being homog-
enized.
In the homogenizer ' s chamber, two opposing jets
of liquid hit each other and also experience a simultane -
ous drop in pressure. A description is given on page 318
The the energy density can be calculated either from
the power input /5 per volume flow For the pressure drop
Apyt across the homogenizer nozzle:
P ApyV
Ev J~-
V
= Apn
(81)
If a high- pressure homogenizer is used, a pre- emulsion
must always be formed before a fine emulsion can be
generated . ecules must occupy the surface of the resulting droplets
An overview of the achievable energy densities is given
in Fig. 49.
64
1 the material
/
2
g solids are generally in a powdered state, which poses
3 a ) Sintered or fused bridges
I /
-
CO
,4
Energy density Ey
1 Agitator
2 Rotor/stator homogenisers
3 High-performance homogenisers
4 Ekato Nanomix high-pressure homogeniser
5 Other high-pressure homogenisers
Fig. 49 Particle diameter as a function of the energy density
AUXILIARY MATERIALS USED IN EMULSIFYING
Auxiliary emulsifying materials use various mechanisms
to prevent physical instabilities in already generated
emulsions. In general, such auxiliary materials can be
subdivided into emulsifiers and stabilisers.
EMULSIFIERS
Emulsifiers belong to the group of materials known as
surfactants. Their molecular structure combines both
hydrophilic and lipophilic groups. Owing to their ambi-
philic character, the emulsifier molecules are able to absorb
onto interfaces as well as polar and non- polar surfaces.
According to Schubert [114], this ambiphilic character
leads to three primary properties:
1. Emulsifiers lower the surface tension between the dis -
persed and the continuous phases. This greatly
facilitates droplet disruption because the capillary
forces acting againstthe disruption are proportional to
the surface tension.
2. The arrangement of the emulsifier molecules on the sur-
face of the dispersed phase leads to steric and/or elec -
trostatic barriers between the droplets, thus inhibiting
droplet coalescence.
3. Immediately after droplet breakup, the emulsifier mol-
in order to stabilise them. However, the entire surface
of the droplet is not completely occupied at first. If two
droplets move towards each other, a higher interfacial
tension develops in the approach zone, causing the
DISPERSING LIQUID / LIQUID
SOLID / LIQUID DISPERSIONS
With bonding inside Without bonding inside
the material
Analogously to the dispersal of liquids, industrial pro -
A ) Solid bridges C) Attractive forces cesses also need to process solid components. Except
for large - scale industrial processes, e.g. ore leaching,
different requirements for mixing technology. The solids
e) van der Waals forces
5
have to be suspended ( see page 44), dissolved, wetted
/
CO
b) Crystallized solids,
electrical conductor
or simply incorporated into a sometimes more viscous
product . Not only non- reactive solids but often reactive
solids have to be mixed into liquids. It is important to
cured binders differentiate between solids that are chemically reactive
( dissolve with liberation of gas, exothermic /endothermic
B ) Liquid bridges
reactions, etc.) or whether they cause physical changes
oo
electrical insulator in the product (e.g. thickening agents and swelling agents).
f ) Electrostatic forces The respective mixing, blending or homogenizing tech-
c) Adsorbed layers niques have different duties to perform.
D) Interlocking
BREAK- UP OFAGGLOMERATES
In most cases, finely dispersed, powdery solids are not
d) Mobile liquid bridges
( capillarity) present as completely separate particles, i.e. they are not
primary particles. Solids usually contain agglomerates
Fig. 50 Adhesive mechanisms between solid particles caused by different effects. In the simplest case, the ag -
glomerates have formed due to liquid bridges and effec -
emulsifier film to move in the direction of the higher tive capillary forces have developed. When the solid is
interfacial tension, dragging the molecules of the con- incorporated into a liquid, these capillary forces are dis-
tinuous phase with it. This increases the pressure in the sipated and the agglomerate falls apart.
approach zone with simultaneous sorption of emulsi -
fier molecules at previously unwetted points. The entire However, there are also other agglomerating forces that
process is known as the Gibbs -Marangoni effect. cannot be overcome by simply mixing the solid into
a liquid. These include van - der -Waals and electrostatic
There are a variety of emulsifiers with different stabilis- forces. Agglomerates may also form as the result of in-
ing properties, and the selected emulsifier must have the terlocking (entanglement).
appropriate properties to meet the specific product re -
quirements. One of the key properties is the speed with They are thus groups of particles that are not the result
which the emulsifier can stabilise the dispersed phase. of direct bonding within the solid. An overview of the
adhesive mechanisms between the solid particles is giv-
STABILISERS en in Fig. 50 [115].
Stabilisers are macromolecules that increase the viscos -
ity. A higher viscosity hinders coalescence and sedimen- Similarto emulsification, deagglomeration is based on
tation phenomena so that physical stabilisation of the turbulent shear stresses. These must be all the higher the
.
emulsion can take place Examples of stabilisers include smaller the agglomerates or the higher the adhesive
hydrocolloids, such as starches and pectins, as well as forces are between the individual particles.
semi- or fully synthetic products, such as cellulose de-
rivatives or polyacrylates. The energy required for deagglomeration is introduced
with a suitable'impeller, dissolver disc or rotor /stator
system.
65
 DISPERSING LIQUID/ LIQUID

DISINTEGRATION OF SOLIDS WITH ROTOR / STATOR

SYSTEMS
If the primary particles or solid bridges between indi -
vidual particles ( see Fig. 50) have to be broken up, this is
known as comminution or grinding.

Similar to emulsification, a certain amount of energy

must be applied to the solid particles in order to break
them apart. Initially, this involves overcoming the inter-
molecular forces as well as the attractive forces along the
broken edges. The necessary energy is known as the in-
terfacial energy.

In addition to the interfacial energy, there are other

energies that have to be applied to induce the processes
associated with disintegration [115]:

- permanent deformation in the tiny area around the

crack tip,
- structural changes,
- generation of elastic waves in the fragments (kinetic
energy),
- generation of heat .

The sum of these energies is known as the fracture

energy.

Numerous methods have been reported in the literature

to calculate and derive the fracture energy so that such
processes will not be discussed in detail here .

Grinding of primary particles with a rotor/stator system

is only possible to a very limited extent. Furthermore, the
rapid wear caused by very hard or abrasive media reduc -
es the cost - efficiency of the equipment. With softer or
very brittle materials, on the other hand, a rotor /stator
system can give very good results. However, the final par-
ticle size is very limited due to the dispersing gap of a rotor/
stator. Under ideal conditions and high tip speeds of the
rotor, the particle size of such materials can be reduced
to approx. 40-60 pm.

66 KATQ EKATO 67
 RHEOLOGY HEAT TRANSFER
THE SCIENCE
DIMENSIONLESS NUMBERS
FUNDAMENTALS HOMOGENIZING
SUSPENDING SOLIDS
DISPERSING GAS / LIQUID
DISPERSING LIQUID / LIQUID
HEAT TRANSFER
Agitated side Heating/cooling side Medium Temperature « , [ W/( m2K )l
MIXING SOLIDS
VACUUM CONTACT DRYING Steam 100 - 200 °C 8000 .. . 10 000

Water 100 °C 8000

Water 20 °C 4200

“ i
- Sw,e Marlotherm S 300 °C 1340

In many mixing processes, transfer of heat is one of the main tasks to be fulfilled by cooling / Marlotherm S 20 °C 200

heating elements in conjunction with the impeller and the vessel geometry. Typical objectives Fig. 52 Typical temperature profile for heat transfer through a vessel Fig. 53 Typical values of the external heat transfer coefficient at
include keeping the product temperature constant during an exothermic or endothermic reaction wall
or attaining a certain product temperature within a specified time by heating or cooling.
Heat-transfer medium Typical application range
account . For a well-homogenized product, Ut = 9pr The 90 .
term gdiv includes all other heat sources and sinks, for Steam 150 °C ( 5 bar ) ... 200 °C ( 16 bar )
example heat introduced or removed by gassing or by
Thermo-oil -15 . . . 350 °C
heat losses into the surroundings via the reactor surface.
In many cases, optimisation of heat transfer in temper -
Cooling water 15 . .. 30 °C
ature - sensitive processes has a positive effect on the P
The heat flux Qw is generally introduced/removed via
product quality. Shortening of the heating or cooling times a heating/cooling medium that isflowing through a double Cooling brine -5 ... -30 °C
can significantly improve the cost- effectiveness of the jacket or external half pipe coils, etc. gwcan be calculated
process. These are only two reasons for giving special Low-temperature brine -70 °C
with
consideration to heat transfer. Qdiv Fig. 54 Typical application ranges of different heat- transfer media
Gw - k - A - ( 9pr - $med) (83)
From the point of view of mixing technology, the heat The heating/cooling medium - side ( external ) heat trans -
transfer resistance on the internal side of the vessel wall Qy where £ isthe heat transfer coefficient, is the area avail- fer coefficient aecan take on a wide range of values depend-
can sometimes be significantly improved by choosing a able for heat transfer and Spr - &med is the characteristic
^ ing on the material properties of the medium and the flow
suitable impeller. To minimise the overall heat transfer temperature difference between the product side and the velocity (see Fig. 53).
resistance, the resistances through the vessel wall and heating /cooling medium. The choice of heat-transfer medium can also signifi-
on the external surface of the vessel must also be con -
sidered. The equations to calculate the individual resist-
ances can be used to estimate what the optimisation
potential is for heattransferand whether additional heat-
|- |
. If the temperature of the heat transfer medium in the
heating/cooling jacket changes between the inlet and
outlet, i.e. $med in 9med, out then
)
*'
cantly affect the achievable heating /cooling jacket tem-
perature Smed (see Fig. 54) .
Fleat transfer through the vessel wall must take account
of the thermal conductivity of the vessel material and
exchange surfaces have to be provided in order to meet
the process requirements.
(.9pr -
9med
i
;
jjj ) ( 9pr
- i -
9med
<
; out ) the wall thickness, which can vary from s = 2 mm up to
% — ^med - - 9yr —
l
(84 ) s > 40 mm, depending on the size and pressure resistance
of the vessel.
i

Fig. 51 shows the heat sources and sinks that can influ-
% 'Yr.e'. out
ence the product temperature 3pr of a mixing process. The heat transfer coefficient k is composed of the prod - The internal heat transfer coefficient ct, can be deter-
Qojin uct-side ( internal ) («j) and coolant- side ( external ) (cte) heat mined using the Nusselt equation
The heat balance for a well-agitated vessel gives transfer coefficients as well as the resistance of the ves - 0.14
Fig. 51 Heat flows in a continuously operated vessel sel wall ( wall thickness s, thermal conductivity X ) (Fig. 52). (86 )
( mPr ' cp,?T + mV cp,v )
'
At
= P + QVL + GW - GE 1821
With respect to the heat sources and sinks, P is the Neglecting the radius of the vessel wall (i. e., the wall The constant c is influenced by flow along the heat - trans -
+GFeed Gpr
“ “
Gdiv power input of the agitator into the product which is com - thickness is very small with respect to the vessel diam- fer surfaces. This means its value depends not only on
pletely dissipated into heat, QR is the heat generated by eter d\ ), gives the type of impeller system and the geometric design of
Flere it is assumed that the mixing vessel with mass my chemical reactions, Qw is the heat flux introduced / the reactor, buf also on the type of heat- transfer surfaces
1
and specific heat cp y has the same temperature as the prod- removed by the heating /cooling elements and QE is the k= (85) ( e.g. smooth vessel wall or internal pipe* coil ) and must
uct &pr. In a well- filled vessel, typically the product of the heat removed through evaporation of product compo - 1jae + s.fX + lj(ai
therefore be determined correspondingly. The parameters
mass and the product ' s specific heat mPrcp Pr » mwcp v,
> ;
.
nents In the case of a continuously operated reactor, the where 1/ae, \ ta\ and s!X are the resistances to thermal a and b essentially depend on the flow regime: for most
so that the term mycp y can be neglected. heat fluxes due to the product mass fluxes in and out of conduction externally, internally and through the vessel impellers, a = 2/3 and b - 1/3 underturbulent flow condi-
the reactor, /Mjn respectively mout must also betaken into wall, respectively . tions, whereas a = b = 1/3 under laminar flow.

68 69

it

For non - Newtonian liquids, the Newtonian viscosity term
in the Nusselt equation must be replaced by the apparent
viscosity or the viscosity at the wall in Re, Pr and the viscos-
.
ity ratio qfyyj The viscosity at the vessel wall is derived from
the distribution of shear stresses on the heat transfer sur-
faces.
For heating/cooling processes, mixing vessels are often
equipped with half pipe coils that are welded onto the
exterior of the vessel. Heat transfer through such pipe
coils is influenced by factors such as the geometry and
the flow regimes. A simple estimation for the achievable
external heat transfer coefficient ae in these pipe coils
can be obtained from
Nu = ae 4 - Prl / 3
'
^-med
= 0.024 - Re? °* (87)
where dp is the pipe diameter and ReP = vdpplri is the
Reynolds number for flow through a pipe. For non - cir-
cular pipes, the hydraulic diameter <iH = AAlU \ s used,
where A is the cross - sectional area and Uis the wetted
circumference.
STEADY- STATE HEATTRANSFER
If the product temperature is to be kept constant,
i.e. dr?pr/d t = 0, the term on the left- hand side of Equation
( 82) is omitted. Atypical application example is maintain-
ing a constant product temperature in a continuously
operated reactor during an exothermic chemical reaction.
Furthermore, if the power introduced by the impeller /’,
power losses due to evaporation Qp and other heat sourc -
es /sinks 2div are so small that they can be neglected,
Equation ( 82 ) can be simplified to:
0W - QR +
~
Pr (cp, Pr % ”
cp,m & in ) (88 )
NON - STEADY STATE HEAT TRANSFER
If the product has to be heated or cooled and provided
that
- the heat capacity of the vessel is negligible with respect
to that of the product,
- the heat transfer coefficient k is independent of time t
or can be replaced by a suitable mean value and
- apart from heat transfer through the vessel wall, all
other heat sources/sinks are independent of time t
and can thus be consolidated into a constant as
follows
70
HEAT TRANSFER
OPTIMISATION OF HEATTRANSFER HEATTRANSFER COEFFICIENT k
Smisc - P + QR + Ql
- + ftliv
* /(/) (89)
The previously described cases illustrate that heat trans-
Equation ( 85) can be used to estimate the impact chang -
es in the individual heat transfer resistances have on the
then Equation ( 82) can be simplified to fer in an agitated vessel depends on many factors. In ad - heat transfer coefficient k. Numerically, improving the larg -
dition to the effects of the impeller, various factors relating est resistance provides the greatest potential for increasing
dQ
wPr • CP , Pr =k A (,9pr -,9med ) + <2misc (90) to the vessel ( geometry, weight, material properties), to the k value. If the limiting heat transfer resistance occurs
the product ( density p, thermal conductivity X , viscosity on the product side, it should be first checked whether
Three cases can be distinguished for integration of curve 77 - f ( j ) or specific heat cp ) and to the design of the another impeller could achieve sufficient improvement.
Equation ( 90) [181]: heat exchanger, including choice of heating/cooling me -
dium, its inlet temperature and mass flow all influence the Whereas in the complete turbulent flow regime differ -
1. Electric heating is used to keep the heat input con- heat flux £7W that can be transferred via heat- exchange ences between the individual impeller configurations are
stant with respect to time. Integration of ( 90 ) gives surfaces. still relatively minor for a given specific power input
Pv = P/ V, the choice of impeller becomes increasingly
,9 (/) = ,9 (r = 0) + Qw + 6tnisc t (91) Many of the parameters are often selected on the basis important with increasing product viscosity or decreasing
cp, Pr
'
2. The energy input is carried out at a constant heating
of process requirements other than heat transfer. Never -
theless, even after all the aforementioned parameters
Re number. This requires knowledge of constant c in Equa -
tion (86 ) as well as the performance characteristics of the
medium temperature, e.g. in a steam - heated vessel have been specified, it is expedient to optimise heat trans - impeller. In general, however, the choice of impeller be -
jacket. Then ( 90) gives fer on the basis of the specified process data. This is par - comes progressively critical as both the viscosity and the
ticularly relevant to cooling processes because the impel- non-Newtonian flow behaviour increase. Central impeller
ler power input P completely dissipates into heat and thus systems with a diameter ratio of d2ld\ ~ 0.7 can be used
S(«)
“ »»e4 in (92) counteracts the cooling process. underthese conditions . For predominantly laminarflow,
however, sufficient product flow at the vessel wall is no
k-A The impeller speed has to be optimised because longer possible with a central impeller system. In this
t
+ P ~ n2 ... w 3 and gw ~ «1/3... «2/3. There is an optimum case, impellers with a small wall clearance can ensure
impeller speed n0pt at which |.P + 2w | has a maximum that the product is flowing close to the wall, thus avoiding
value, i.e. the maximum heat flux can be removed and or reducing product layers near the wall that limit heat
3.If the temperature of the heat transfer medium above which the cooling time increases again. To shorten transfer [37 ], Optimisation of the impeller becomes in -
changes, e.g. over the length of the half pipe coil, in - batch times and also save energy, it is therefore advisable creasingly important here .
tegration of ( 90) gives that cooling processes are not operated with n > «opt.
The dependency range of the parameters P and £?won n In addition, the Ekato Paravisc can be fitted with flexible
& (/ ) = rned, in 6misc depends on both the flow regime ( laminar/turbulent ) and wall scrapers on its vertical bars. These scrapers are uni-
^ K*
(93 )
on the product ' s flow properties. The associated values formly pressed by the product flow against the vessel
must be known in order to calculate «opt . Otherwise, af- wall thus compensating any eccentricity of the vessel.
+ fs( f= o) - smed in ~ fimiscl . Q
[ mPfcp, Pr / ter designing the agitator system using assumptions for The CFD simulation in Fig. 55 visualises how the bound-
)
K* the product properties, a final optimisation should be ary layer is removed from the wall so that it passes around
-k - A carried out on the existing process. the scraper and the bar. In case of a cooling process, the
where K* = mmed • cmed l- e
^ d cmed
According to Equation ( 83 ), there are three approaches
cold liquid is transported into the region of vertical flow
and is gradually replaced by hotter medium from the cen-
to increasing the heat flow : tre of the vessel. Wall scrapers can reach up to a 10-fold
If (2misc = then for case 3 improvement of the heat transfer [ 55].
increasing the heat transfer coefficient k,
TEMPERATURE DIFFERENCE dpr
0 ^
5( - med, in
(94 ) increasing the temperature difference t?pr
enlarging the heat transfer area A.
-
tfmed or ^
med
Raising the temperature difference t?pr- t?med , if achiev -
k- A able, provides a huge potential to increase (?w . In this
1- e "feed ‘ined '4
’ If heat removal via the heat exchangers is insufficient, case, consideration should be given to using alternative
mpr -^
C pr it may be necessary to use reflux cooling based on solvent heat transfer media ( see Fig. 54) that allow higher or
+ (5 (r = 0) - 5med in ) - e evaporation and external condensation. lower temperatures in the heating /coofing jacket . How-
ever, if the heat transfer medium is changed, it must be
checked that changes in ae do not have negative conse -
quences forgw .
71

Fig. 55 Flow along the wall in the laminar flow regime: improvement of heat transfer using a scraper
HEAT- EXCHANGE SURFACES
There are a number of ways to enlarge the heat exchange
surfaces. If a heating/cooling jacket is used, the first step
is to check to what extent the effective area can be ex-
tended, and this should also include the vessel bottom.
By selecting a more tall vessel, a significant enlargement
of the heat exchange area can be obtained for a given
filling volume; however, the choice of geometry is usu -
ally limited or governed by other aspects.
If enlarging the effective surface area with only a heat-
ing/cooling jacket is insufficient, the next step is to check
whether internal heat exchange surfaces can be used.
Various methods of installing internal heat exchange sur-
faces are shown in Fig. 56. In this way, the specific heat -
exchange area can be increased by 1-5 m 2/m3, or by more
than 20 m 2/m3 for plate heat exchangers.
72
HEAT TRANSFER
<
c -s c n>
)
& &
( ( ) ) r i
(
?S c
c
)
)
( ( ) )
c c )
c )
c ) )
. c ) c
( c ; C
: C )
)
x.
0
Fig. 56 Additional internal heat exchangers: different layouts of tube bundle heat exchangers and plate heat exchangers
SCALING UP
When scaling up a heat transfer process with a given
geometric similarity, i.e. A ~ d\ and V ~ d ] , the specific
heat transfer area decreases with A/ V ~ df1 Therefore, .
when scaling up from laboratory scale to production scale,
it should be checked whether the respective heat transfer
requirements can still be fulfilled. As discussed above,
this is particularly important for cooling processes. A heat
exchange process that is satisfactory on the laboratory
scale may no longer be technically possible on the pro -
duction scale. In order to identify this situation in good
time, it is particularly important to carry out laboratory-
scale tests under conditions that are capable of being
scaled up.
As an alternative to enlarging the internal heat-exchange
area, external heat exchangers can be used through which
the product is pumped. This provides a means to greatly
increase the heat exchange area. However, in case of cool-
ing processes, account must betaken of the fact that me -
chanical pumping power is converted into thermal en -
ergy that also has to be removed.
REFLUX COOLING
In case of strongly exothermic reactions or very large
vessels, the aforementioned method of heat removal is
usually insufficient. If the product properties are appropri-
ate, a further method to cool the product is to evaporate
a product component. The vapour is then recovered by
external condensation and fed back into the vessel.
Depending on the heat of evaporation, this method pro -
vides considerable potentials for heat removal.
73

THE SCIENCE
FUNDAMENTALS
One of the key factors when mixing solids, particularly bulk materials, is to determine the degree
of homogeneity. Statistical principles can be used to find an effective method of determining the
degree of homogeneity in practical applications. The scale- up of blend times from the laboratory
to full production is also of great importance .
IMPORTANCE IN PROCESS ENGINEERING
Mixing of solids is a fundamental step in many manu-
facturing methods and is used throughout the entire range
of process industries. Examples include mixing of active
substances and excipients in the pharmaceutical industry
or blending muesli ingredients in the food industry. At
least one component has to be distributed within a second,
thus reducing inhomogeneities with respect to composi-
tion, material properties and/or temperature. It follows
from this requirement that a mixing process is in fact
a transport process. The solid components may be present
as powders, granulates, etc., and are grouped together
under the term " bulk materials". The primary objective is
to achieve the highest possible degree of homogeneity
within the shortest possible blend time using the least
possible energy.
Tests are generally unavoidable owing to the fact that
bulk materials cannot be described using a single physi-
cal parameter, such as the viscosity in the case of liquids,
and because there are many influencing factors ( bulk
density, solid density, particle shape, porosity, moisture,
particle size distribution, unconfined yield strength, ad-
hesive forces, chemical composition, temperature, flowa -
bility index, etc.). Test results from a geometrically simi-
lar pilot unit are thus commonly used as the basis for
scaling up to the larger production scale. Most tests are
carried out in a vertical apparatus with a rotating agitator.
Alternative mixing methods for bulk materials include
metering transport from stockpiles or silos. The following
discussions aim to describe feasible methods of determin-
ing the blend time on the production scale.
74
RHEOLOGY
DIMENSIONLESS NUMBERS
HOMOGENIZING
SUSPENDING SOLIDS
DISPERSING GAS / LIQUID
DISPERSING LIQUID/ LIQUID
HEAT TRANSFER
MIXING SOLIDS
VACUUM CONTACT DRYING
DEFINITION OFTHE DEGREE OF HOMOGENEITY
In general, blending of bulk materials achieves only an
approximation of an ideal mixture, even after an infinite
blend time, because an equilibrium between homogeniz-
ing and segregation mechanisms ( e.g. due to agglom-
eration, flotation or percolation) develops during blending.
This means that the first step is to define the target
degree of homogeneity for each process. The time until
this homogeneity is obtained must be determined on the
laboratory scale. These findings are used as the starting
point to determine the blend time forthe production scale.
Experiments to determine the blend time often make use
of a tracer component. This method is particularly com -
mon when quantification of the concentration is compli -
cated. Alternatively, the bulk density or the temperature
profile can be analysed. In general, the analysis method
is selected on the basis of the requirements for the bulk
material and its properties. For example, an analysis of
the colour intensity is useful for blending colouring pig-
ments. The tracer component must have a measurable
property that differentiates it from the mixture. It should
also be chemically inert and easily detectable. Further-
more, the particle properties of the tracer component
should correspond to those of the mixed material to pre-
vent segregation, which would falsify the experimental
results. Methods for analysing tracer components are
divided into direct methods ( e. g. counting, sieving or
chemical dissolution) and indirect methods (e.g. radioactive
detection, dyeing plus photometric evaluation or determi-
nation of the conductivity).
The sample size also plays a key role in the experimen-
tal determination of the degree of homogeneity. The sam-
ple weight of bulk materials must be much larger than
that of a single particle because, in an extreme case,
a sample having the weight of a single particle would
falsely imply complete segregation. A useful reference
value is the later package size ( e. g. a bag of cement in the
building material industry or the tablet size for pharma -
ceutical applications). The ratio between sample size and
total amount should lie within the following range [110 ]:
1Q-4 < ^sample < 1Q-1 (95 )
mtot
In addition, accurate determination of the degree of
homogeneity requires an appropriate number of samples,
representative sampling points as well as a reproducible
sampling methodology. Based on statistical methods, the
dimensionless degree of homogeneity Mg is defined as
follows:
Mg = 1
,2
(96)
max
With respect to graphic representation, it is expedient
to use the variance instead of the degree of homogene-
ity as the ordinate because the end point of the mixing
trial can be read off a logarithmic scale more easily (Fig.
61). The sample variance s2, also known as the empirical
variance, for a number /Vof samples is expressed by:
(97)
where xi is the measured percentage mass fraction of
tracer component in the zth sample and p is the actual
.
fraction of this component present /* as well as the sam -
ple mass must remain constant during the mixing tests
to ensure that the measurements and the resulting vari-
ances are reproducible. The sample variance comprises
the variances associated with mixing and measuring. In
case of a completely segregated system of two compo -
nents, the maximum variance tr2max for the total quantity
of bulk material can be calculated with [74]:
O-P) (98 )
° max = Pq = P '
Where p and q are the relative fractions of the two
components being blended. In two - component systems,
p = \ - q.
Because a full analysis of the bulk material is impracti-
cal, the completely segregated state is related to the sam -
MIXING SOLIDS
pie mass. The maximum variance with respect to the
sample mass <r2max sample can be calculated with the fol-
lowing equation [127 ]:
i i
p q
max,sample
^sample
P'
D=0
\ D+ q \
D= 0
mq ( D ) dD
(99)
The integral terms reflect the influence of the respective
particle size distribution of the two components (a bulk
material is almost never monodisperse). This involves
integrating the measured distributions (profile of the par-
ticles with mesh passage D ) over the size intervals to
obtain a parameter for the width of the particle size dis -
tribution . Spherical particles are assumed in order to
simplify calculation of the particle mass within the indi-
vidual size classes.
Consequently, the results of the investigation should
be extended by specifying a confidence interval that de-
scribes the probability of the true value lying within cer-
tain limits. In general, the number of degrees of freedom
increases with the number of samples so that the curve
of the distribution function approaches that of a normal
distribution. If stochastic homogeneity is reached, the
sample variance will vary around the value of the homo -
geneous random mixture with a certain probability. Thus
the stop criterion can be selected as the time at which the
value of the sample variance lies within the confidence
interval.
METHODOLOGY FOR DETERMINING THE BLEND
TIME
Since determination of the blend time based on the bulk
density is generally too inaccurate and determination
based on temperature differences is not feasible for con-
structive reasons, the following section describes
the methodological approach to analyse the blend time
using quartz sand (Si02) and sodium sulphite (Na 2S03)
as an example system.
Quartz sand ( relative mass fraction: p ) is cheap, free-
flowing, easy to dispose of and has a poor electrical con-
ductivity. Sodium sulphite ( relative mass fraction: q) is
also cheap, free - flowing, easy to dispose of and not
physiologically harmful; however, it has a very good elec -
trical conductivity, which means it is a suitable tracer
component for analysing the homogeneity. A constant
sample weight is dissolved in demineralised water, and
the conductivity of the solution is measured. The sample
75
 MIXING SOLIDS

Addition point of sodium sulphite

1.0E-02 ,
SM with 99 rprp
1.0E-03 ^ VSM with 66 rpm
?1. E-04
0
VSM with 33 rpm
2 1.0E 05-

> 1.0E-06
limit of variance '
1.0E-07

Fig. 57 Microscope image of sodium sulphite 1.0E-08 -

Fig. 58 Microscope image of quartz sand 0 0.5 1.0 1.5 2.0 2.5

mass was defined as 5.0 g. This value lies just above the
minimum sample quantity for the investigated combina -
tion of bulk materials. To simulate the concentration of
active substance commonly used in the pharmaceutical
industry, the sodium sulphite fraction was defined as 1 %.
Both bulk materials differ only slightly with respect to
their particle shape and bulk density; however the parti-
cle size distribution of quartz sand is narrower (see Fig.
59) . This enables simulation of unfavourable blending
properties that reflect reality more effectively. The parti-
cle size distribution of both bulk materials has to be ana
- with respect to the fully segregated state;
lysed in order to calculate the completely segregated state
with segregated state related to the sample weight. The
variability is then determined by integrating the curves.
Therefore, a sieve analysis was carried out with both in-
vestigated bulk materials.
For the integration, the mean particle diameter within
each mesh range was used to determine the particle
weight. Both bulk materials had a solid density of ap
- dom mixture. This involves adding up the individual errors
Comparison of particle size analysis of quartz sand/sodium sulphite
Blend time [ min]
proximately 2640 kg/m3. Integration gives the following
variabilities for quartz sand ( p) and sodium sulphite (q ): Fig. 61 Variance plot of Ekato So lid mix VSM at different rotation speeds
1 (materials: quartz sand/sodium sulphite )
>
£ =0
J mp (D) dD = 1.66 - 10 ~ 5
g
the blend time for a geometrically similar apparatus and
i
an identical product. For a given tip speed of 2 m/s, the
J
0= 0
mq (Z)) dD = 4.89 - 10~5 g Solids -Paravisc of series VSM 2500 rotates at 23 rpm. This
approach gives a blend time of just under 4 minutes for
the axial- pumping mixer Ekato Solidmix VSM 2500.
This gives the following variance for the sample mass Fig. 60 Design of the Ekato Solidmix VSM
0.99 - 0.01 In all experiments, quartz sand was filled into the vessel,
•(0.99 1.66 0.01 • 4.89
=
5.0 g + ) 10-5 g then sodium sulphite was added as a tracer component
(102 ) by pouring it onto the surface half - way between the shaft
= 9.62 - 10-8 and the wall. The blending process was started and then
interrupted at defined intervals so that samples could be
To obtain a meaningful variance plot, it must be checked taken from 6 different locations. The conductivity meas-
whether the measuring variance ( variance of the measur -
urements of the samples and calculation of the sample
ing error) is small with respect to the homogeneous ran - variance were carried out afterwards. Once the confidence
interval was reached, determination of the blend time
of the selected measuring method. Errors may arise from under the given boundary conditions was regarded as
the mass determination, dilution of the sample using complete. Blending tests were carried out at three differ-

S i.o
a volumetric flask and pipette and from the conductivity
measurement. These give a total sum of 2.61 1Q-8 forthe
ent shaft speeds .
Q 0.9
measuring variance. Thus the measuring variance lies
- The dimensionless blend time n -0 was determined us-
i sodium sulphite ing the aforementioned variance plot for the investigated
0.8 well below that of the fully segregated state, which con
- geometry and the tested materials. The point of intersec -
a 0.7 firms the suitability of the applied measuring procedure.
quartz sand tion of the variance curve with the variance limit was read
-
i °-
6
! The confidence interval in this case requires that the meas- off the variance/blend time diagram and multiplied by the

i
° 04
5
uring value has a 95 % probability of lying within this
range. For 5 degrees of freedom (6 withdrawn samples),
shaft speed. The results gave dimensionless blend times
fi 0.3 of between 81 and 87 revolutions. These values lie close
| 0.2 -
this gives a variance limit of 2.13 10-7. Once the theoreti-
cal suitability of the measuring procedure has been prov
to each other, which is typical of axial - pumping impellers
= 0.1 - ( see page 38 ff. ). Therefore, it can be assumed that the
en and the boundary conditions have been defined, investigated parameters are also constant for scaling up
0 0.1 0.2
experimental determination of the homogeneity curve to the larger production scale. It is thus easy to calculate
0.3 0.4 0.5 0.6 0.7
can begin. The blending experiments were carried out in
Upper interval limit [ mm]
a conical 40 - litre vessel with a cylindrical body
and an
Ekato Solids-Paravisc. The results were extrapolated to
Fig. 59 Particle size distribution of the components being mixed a mixer of the VSM2500 series.

76
77

. a
 THE SCIENCE RHEOLOG Y
VACUUM CONTACT DRYING
DIMENSIO NLESS NUMBERS
FUNDAMENTALS HOMOGE NIZING
SUSPEND ING SOLIDS
DISPERS ING GAS/ LIQUID
DISPERS ING LIQUID / LIQUID
HEAT TRANSFE R
Criteria Emptying Required
MIXING SOLIDS Systems maintenance work
Free-flowing
bulk materials
Pasty Product-contacted
products components
VACUUM CONTACT DRYING
Horizontal system + +

Conical screw system + +

Tumbling system o O

Ekato VSM + + o
Mixers with a vertical design are a well- proven choice for thermal drying of + +
wet solids under
a vacuum . In addition to the theoretical principles, the procedure for
scaling up from the Ekato VPM + + o + +
laboratory to the production scale is of particular interest with respect to
the drying time. Fig. 62 Matrix for dynamic systems for mixing and drying solids

IMPORTA NCE FOR THE PROCESS INDUSTRY
Thermal drying of wet solids is a combined heating and
mass -transfer process that aims to achieve a dry end
product by removing moisture. For energetic reasons,
thermal drying is often used downstream of mechanical
equipment for solid / liquid separation e.g. a centrifuge or
a filter press. In contrast to mechanical moisture remov-
al, thermal drying involves transforming the liquid into
a vapour. In the vacuum contact drying process, the boil -
ing point of the liquid is lowered by generating a vacuum
inside the dryer. This means that the final moisture content
in the product can be adjusted to the desired value using
lower temperatures than under atmospheric pressure
conditions. The thermal energy is introduced via heated
surfaces, and an agitator constantly circulates the mate -
rial coming into contact with these surfaces, thus accel -
erating the drying process. In a static system, such as an
evacuated tray cabinet, the product is not circulated, which
results in long drying times and an inhomogeneous end
product. For certain products, the only way to achieve
the required final quality is gentle drying in a dynamic
vacuum contact drying system. An alternative method
for removing moisture even more gently is freeze - drying,
which is used to dry herbs, for example. In this case, the
moisture is not evaporated, it is removed by sublimation,
and this requires very high vacuum. Established dynam-
ic systems for vacuum contact drying are horizontal,
vertical, tumbling and Nauta ( vertical screw) systems.
An overview of common systems is given in Fig, 62.
A vertical design offers clear advantages because emp -
tying residual solid out of the apparatus is more effective
owing to the more favourable geometry, the drying times
can be shortened due to the high heat transfer coefficients
on the product side and no seals or drive components are
120
30
necessary in areas having product contact. Vacuum contact
drying in a vertical apparatus is discussed below. jacket temperature
100
25
THEORET ICAL BACKGRO UND
80 20
It is generally expedient to carry out tests of vacuum
contact drying on the laboratory scale using conditions
e f
1
analogous to those of the targeted production scale so
that the obtained data can be used for scaling up. This is
i 60
IE
15 -
due to the large number of parameters influencing the i
.
resulting drying time Vacuum contact drying involves 40
product temperature
10
a combination of heat- and mass-transfer processes, which
is why drying is divided into two phases. Initially, the moisture
process is limited by heat transfer, whereas mass transfer 20 5
becomes the limiting factor at the end of drying. This is
shown by the temperature and moisture curves in Fig. 63,
which were measured at a constant vessel pressure. 0 0
0 5 10 15 20 25 30 35 40 45 50 55 60
The first phase, in which the product temperature de -
Time [min]
pends on the applied vessel pressure, is known as the first
drying phase. The drying rate is limited by heat transfer Fig. 63 Typical temperature and moisture curves at a constant pressure
because there is sufficient moisture on the particle sur -
faces that can evaporate freely. The saturation pressure of
.
temperature Moisture is first removed from the macro - - small fraction of micro-capillaries inside the particle
the solvent dominates during this phase. This dependency capillaries and then from the micro - capillaries.
structure
can be seen in the curve of Fig. 63 up to approximately the
The following influencing variables promote high dry- - generation of turbulence by circulation of the bulk
36th minute. Subsequently, the product temperature in- ing rates or a short drying time:
material
creases although the vessel pressure remains constant . - large temperature difference between the product
- injection of gas into the bulk material
After a short transitional phase, the second drying phase and the heating medium (vacuum and/or temperature
starts. In contrast to the first, it is limited by mass transfer.
of the heating medium) The integration of additional heat- exchange surfaces
The saturation vapour pressure no longer dominates at - large surface areas for heat exchange
and a solvent recovery system that generates a particu-
the particle surface owing to the ever greater resistances - high moisture content at the surface
larly high vacuum and thus largertemperature differences,
to the diffusion of solvent towards the surface. The drying
- small particle sizes ( for short diffusion paths to the .
generally lead to much higher costs The above list dem -
rate decreases due to the moisture bound in the particles particle surface)
onstrates that optimisation of the drying process can be
and the thermal energy now starts to increase the product
primarily achieved by efficient circulation of the bulk mate-

78
79

free water in
macrocapillaries
the bulk material with free water
microcapillaries
water bound with bound water
in particles
Fig. 64 Mechanisms for moisture release from a bulk material
rial during the drying process because the remaining in-
fluencing variables remain essentially unchanged. Par-
ticular attention must be paid to ensure that the agitator
in a vertical vacuum contact dryer provides a high degree
of axial mixing. This accelerates heat transfer to the prod-
uct because it promotes free convection within the bulk
material. Another option is to inject a carrier gas near the
end of drying (to remove bound moisture), thus accelerat-
ing the process. In this case, mass transfer can be improved
by changing the partial pressure conditions. Equation (103)
shows an approach for extrapolating laboratory results to
the production scale [133], Drying times obtained at the
model scale are determined with the original product un-
der similar boundary conditions. The resulting drying time
tM for the model scale, in which the wet material is dried
to the desired final moisture content, is thus used as the
basis for calculating the drying time for the production
scale tP :
A
V /M Ai9jy[ 4iix, P *
> mix
h ~ tw
( y )p lW ’
M 003)
This approach assumes that the first drying phase is
rate - determining for the overall drying process. The sec -
ond drying phase is limited by the diffusion rate and thus
needs the same time in the model and in the production
scale for identical drying parameters. This approach also
takes account of different surface area -to-volume ratios
in the model and production scales. Additional heating
of the impeller leads to large temperature differences at
high relative velocities between the particle and agitator,
which increases the product side heat transfer coefficients
80
"
VACUUM CONTACT DRYING
compared to heating only the vessel wall. Model-scale As described on page 38 ff, an approach with a con-
experiments with a heated impeller have shown that the stant dimensionless blend time ( a - 6 = const.) can be
above equation is only a worthwhile basis for scale - up if .
assumed for axial- pumping mixing systems The rotation
.
the agitator is heated at the production scale as well If this speeds from the two experiments can then be inserted
fact is not taken into consideration, the calculated drying into the logarithmic blend time term of the denominator.
times for the production apparatus may be too short. The Differences in the surface area -to -volume ratio as well as
same applies to the use of a chopper to break up lumps the temperature differences between the heating medium
that slow down the process towards the end of drying. and the product can be taken into account here, although
The chopper enables the desired particle size distribution they should be as small as possible. In general, mixing
to be reached and can also significantly shorten the dry- exponents lie in the range between 0.1 ( minor influence)
ing time by shortening the diffusion paths for the solvent. to 0.5 ( major influence of the speed on the drying time).
Both the heated agitator as well as the chopper raise the
costs of the apparatus, and this must be weighed against
the process engineering benefits.
The second term of Equation (103) can be used to take
into accountthe mean temperature differences between
the product and the heating medium. If the vessel pres-
sure is to be changed for the production scale, the Antoine
Equation (104) can be used to calculate the boiling tem -
perature for the first drying phase:
B
9b= -C 004 )
A - log p
The constants in this unit equation A, B and Care solvent
specific . The different heat transfer coefficients between
the model and production scales are taken into account
in the last term. Since the heat transfer coefficient for the
production scale is unknown, the calculation can be car-
ried out based on the blend time [109 ]:
a oc
1 i ^
mix
rmix 005)
The above approximation assumes that the product side
heat transfer coefficient is limiting for heat transfer in
both the model and production scales. To isolate the influ-
ence of the blend time on the drying time, two drying
procedures must be carried out under nearly identical
conditions, but with different agitator speeds. The fol-
lowing equation is then used to calculate the mixing
exponent £m;x :
log
(F), A9j h
(4 A!h h
kmix ~
fmix,1
006)
log
4rix, 2
81