You are on page 1of 34

Drying Technology

An International Journal

ISSN: 0737-3937 (Print) 1532-2300 (Online) Journal homepage: http://www.tandfonline.com/loi/ldrt20

Spray Dryer Modeling in Theory and Practice

Dr David E. Oakley

To cite this article: Dr David E. Oakley (2004) Spray Dryer Modeling in Theory and Practice,
Drying Technology, 22:6, 1371-1402, DOI: 10.1081/DRT-120038734

To link to this article: https://doi.org/10.1081/DRT-120038734

Published online: 06 Feb 2007.

Submit your article to this journal

Article views: 854

View related articles

Citing articles: 45 View citing articles

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=ldrt20
DRYING TECHNOLOGY
Vol. 22, No. 6, pp. 1371–1402, 2004

Spray Dryer Modeling in Theory and Practice

David E. Oakley*

AspenTech, Harwell International Business Centre, Didcot,


Oxfordshire, UK

ABSTRACT

This article considers the modeling of spray dryers at various levels


and the selection of the most appropriate level of detail for practical
situations. The following model levels are described: (1) Heat and
mass balances; (2) Equilibrium based models; (3) Rate based models;
(4) Computational fluid dynamic (CFD) models. The value of each
is discussed in relation to some typical problem scenarios. These
include preliminary process design; process improvement; and
troubleshooting operational and product quality problems. One
particular focus of this article is finding realistic models of the
performance characteristics of spray dryers which can be included
in process flow sheet simulations while not imposing excessive run
times and complexity.

*Correspondence: Dr David E. Oakley, AspenTech Ltd, The Gemini Building,


Fermi Avenue, Harwell International Business Centre, Didcot, Oxfordshire
OX11 0QR, UK; E-mail: David.Oakley@aspentech.com.

1371

DOI: 10.1081/DRT-120038734 0737-3937 (Print); 1532-2300 (Online)


Copyright & 2004 by Marcel Dekker, Inc. www.dekker.com
ORDER REPRINTS

1372 Oakley

Key Words: Spray dryers; Heat and mass balance; Equilibrium


model; Rate-based model; Computational fluid dynamics.

INTRODUCTION—ASPECTS OF A
SPRAY DRYER MODEL

Spray dryers are widely used unit operations because of their ability
to transform a liquid feed into dry spherical particles. This unique
combination of particle formation and drying, and the fact that the spray
drying can have a big impact on energy consumption and product
quality, imposes special demands on designers, operators and research-
ers. A closer examination of the physical processes taking place reveals
why spray dryers are more difficult to model than other dryer types and
why simple scale-up techniques cannot be used.

Atomization

The first important transformation is the formation of droplets by


atomization. The droplet size distribution produced will, in part,
determine the drying time required and also the chamber size needed to
enclose the spray without danger of wall deposits and, importantly, the
final particle size. Furthermore, spray drying commonly imposes
exceptionally tough atomizing conditions because of high feed solids
contents (to minimize the evaporative loads) and consequently high feed
viscosity and non-Newtonian behavior. Atomization is a complex
interaction of inertial, shearing, and surface tension forces. Simplified
theoretical models are available for atomization under idealized
conditions but are probably not appropriate for the conditions in
industrial spray dryers. Computational Fluid Dynamics cannot currently
handle the dynamics of droplet break-up under such conditions.
Currently, the best method of prediction for real spray dryers are
published empirical correlations relating drop-size distribution to the
operating parameters of the atomizer (e.g., Refs.[1,2]). However, care
should be taken because such correlations are very dependent on
atomizer geometry.

Gas Flow and Particle Motion

Once the droplet has formed and been ejected by the atomizer it
will move through the chamber under the influence of the turbulent gas
flow. This motion will determine the total residence time of the droplet,
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1373

the gas temperature and humidity conditions it encounters and whether


it strikes a chamber wall while still wet. The former considerations will
determine final moisture content of the formed particle and to some
extent its properties and morphology (see below). The latter consider-
ation could have serious implications for spray dryer operation if heavy
wall deposits were allowed to build up.
Up to the late 1970s methods for predicting turbulent gas flow in
spray dryer geometries were not available and models of particle motion
employed a simplified analytical representation of the gas flow pattern,
based either on experimental measurements or correlations for the
specific type of chamber. Models of this type are typified by Katta and
Gauvin[3,4] and Keey and Pham.[5] The flaw with such models is that
measurements or correlations of chamber aerodynamics are only valid
for specific geometries, and even in specific geometries flow patterns will
be influenced by inlet conditions.
The advent of Computational Fluid Dynamics in the early 1980s and
ability to predict flow patterns within spray chambers on a theoretical
basis brought a whole new generation of spray dryer models. The earliest
of these was by Crowe[6,7] who used the Particle-Source-in-Cell model to
calculate gas flow patterns and particle trajectories within a laboratory
spray dryer. Other work in this field has been widely reported, for
example.[8–12] These models were a significant advance—since, the
equations of fluid flow are solved theoretically, no restrictive assump-
tions about chamber geometry or inlet conditions were required. This
allows the investigation of novel chamber designs and inlet conditions
without the need to build physical prototypes.

Drying Behavior

Of central importance to the operation is the drying rate achieved as


the droplet moves through the chamber. At high moisture contents, when
the droplet still has a liquid surface, drying rates will be largely governed
by mass transfer through the boundary layer surrounding the droplet.
But, as drying continues and the droplet forms into a particle, drying
rates will be determined by moisture transport through the particle itself.
Moisture transport through a solid particle are highly material dependent
and there are a variety of theoretical models available to predict this,
ranging in complexity from scaling methods, such as the characteristic
drying curve concept, to diffusion models, through to fully mechanistic
models. Evidence reviewed by Keey[13] indicates that the characteristic
ORDER REPRINTS

1374 Oakley

drying curve works well for the particle sizes of interest in spray drying
(typically <500 micron). Langrish and Kockel[14] and Oakley and Bahu[9]
both used the characteristic drying curve concept in conjunction with
computation fluid dynamic (CFD) simulation of spray dryers. A number
of workers[15,16] have suggested diffusion-types models where an effective
diffusivity is fitted to experimental drying data. Others (e.g., Ref.[17])
have proposed receding interface models of slurry droplet drying.
Farid[18] describes a droplet drying model which includes the solution
of the conduction equation for the temperature distribution through
the droplet.
The simple approaches offer the advantages of ease-of-use and
calculation speed; while the more sophisticated drying models are more
rigorous, accurate over a wider range of conditions and give detailed
predictions of variations across a particle or droplet. Without exception
all models require experimental data either to provide model parameters
or drying curves for scaling. These concerns are common to all dryers,
but spray drying creates special difficulties because the droplet sizes
encountered in spray dryers (typically 20–200 micron) are far smaller
than can be handled in droplet drying kinetics measurements (>200
micron according to[21]). The need to scale-down to smaller droplets
inevitable creates uncertainty in the accuracy of drying kinetics.

Particle Morphology

A further aspect that needs to be considered are the changes that take
place to the droplet as it dries and is transformed to a particle. It is now
well known that the conditions a droplet experience will have a
significant influence on final particle properties such as size, density,
and attrition resistance. This is seen both through the impact of changes
in spray dryer operating parameters[19] and laboratory experiments on
single droplet drying (e.g., Refs.[20,21]). The latter have revealed the full
complexity of the size and morphology changes of droplets containing
solids as they are transformed to solid particles. From these studies it is
clear that the final particle morphology and therefore the properties of a
particle will depend on a range of factors including drying rate, drying
temperature, quantity of entrained or dissolved gas, degree of solubility
of solids and so on. These will also have an impact on drying time
requirements since the nature and integrity of the surface of the particle
will determine its drying kinetics. A further area of importance for
product quality, studied by King[22] and others[23] is the retention of
volatiles which relates to flavor and aroma loss in food products.
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1375

MODEL LAYERING APPLIED TO SPRAY DRYERS

It is clear that any detailed simulation of the spray drying process


would be exceedingly complex if it embraced all the aspects given in the
previous section. However, it is the contention here that in many typical
applications of spray dryer modeling this level of complexity is simply
unnecessary. Indeed such a level of complexity may be impractical
because of the computing resources and calculation time required. For
example a full computational fluid dynamics (CFD) model with a run-
time of several hours would not be appropriate for preliminary process
design but would be for the detailed investigation of problems associated
with the chamber aerodynamics. Kemp and Oakley,[24] described the
concept of model layering applied to general drying. In this article the
concept is adapted to the special case of spray drying. The basic concept
is that any item can be modeled at a number of different levels. The term
fidelity has been coined to describe the quality which determines the level
of a model. Models with high fidelity give more detailed predictions and
typically have longer calculation times (though they are not necessarily
more rigorous). The model levels described in this article for a spray dryer
are as follows:

1. Level 0 Heat and Mass Balances.


2. Level 1 Heat and Mass Balances with solid-vapor equilibrium.
3. Level 2A Rate-based with simplifying assumptions about
particle motion.
4. Level 2B Rate-based with simulation of gas flow and particle
motion (CFD).

These models are given in detail below and applied to real examples
in the section on the Practical Application of Spray Dryer Modeling.

Level 0—Energy and Mass Balance Model

Figure 1 shows schematics of co- and counter-current spray dryer


configurations with the liquid feed, inlet gas, exhaust gas, and exit solids
streams denoted by FEED, IN-GAS, EX-GAS, and EX-SOLID
respectively. Considering the simple case of single solid, gas, and
moisture components denoted S, G, and M respectively, the following
mass balances can be constructed
FFEED,S þ FIN-GAS,S ¼ FEX-GAS,S þ FEX-SOLID,S ð1Þ
ORDER REPRINTS

1376 Oakley

EX-GAS
FEED IN-GAS FEED

DRYER

DRYER

IN-GAS

EX-GAS

EX-SOLID EX-SOLID

Co-current Counter -current

Figure 1. Co- and counter-current spray dryer configurations.

FFEED,G þ FIN-GAS,G ¼ FEX-GAS,G þ FEX-SOLID,G ð2Þ

FFEED, M þ FIN-GAS, M ¼ FEX-GAS, M þ FEX-SOLID, M ð3Þ


where Fi, j is the mass flow (kg/s) of component j in stream i. Denoting the
humidity of gas in a stream j by Yj (kg/kg dry gas) and the moisture
content of solid as Xj (kg/kg dry gas). Equation (3) can be rewritten as
ðFFEED,S XFEED þ FFEED,G YFEED Þ
þ ðFIN-GAS,S XIN-GAS þ FIN-GAS,G YIN-GAS Þ
¼ ðFEX-GAS,S XEX-GAS þ FEX-GAS,G YEX-GAS Þ
þ ðFEX-SOLID,S XEX-SOLID þ FEX-SOLID,G YEX-SOLID Þ ð4Þ

An overall energy balance is


FFEED HFEED þ FIN-GAS HIN-GAS
¼ FEX-GAS HEX-GAS þ FEX-SOLID HEX-SOLID þ Q ð5Þ

where Hi is the enthalpy (J/kg) of stream i; Q is heat loss (W) from the
dryer.
For co-current dryers the EX-SOLID temperature is assumed to be
equal to the EX-GAS temperature; while for counter-current dryers the
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1377

EX-SOLID stream is assumed to be at the IN-GAS temperature. In


either case, if inlet stream conditions and the heat loss are completely
specified the above system of equations has at least one degree of freedom
since the enthalpy of the moisture component in the exit streams will
depend on its distribution between the solid and vapor phases. The
specification of the moisture content of the exit solids, XEX-SOLID (or
alternatively the exhaust gas humidity YEX-SOLID), is required before
the system can be solved.
Despite the obvious draw-back that it does not actually calculate the
moisture content(s) of the exit solids, this model does have a number of
benefits. Firstly it is a very simple method of predicting how the exhaust
gas conditions of the dryer depend on the required drying duty. It
requires no detailed knowledge of the dryer except specification of the
inlet streams and moisture contents of the exit solids plus simple thermal
properties (i.e., enthalpy of each component). The predictions for the
exhaust gas conditions can be compared with the saturation line for the
gas-moisture system in question (possibly using a psychometric chart) to
gain an insight into the feasibility of the operation.

Level 1—Equilibrium-Based Model

Here the Level 0 model is extended by using phase equilibrium


relationships in the form of desorption isotherms (i.e., curves of
equilibrium moisture content of moisture in the solid, Xeq, as a function
of the temperature and composition of the gas phase above the solid) to
predict the final moisture content. This means that, unlike the Level 0
model, the final solids moisture content is predicted rather than specified
by the user.
Experimentally measured desorption isotherms and correlations to
fit them are widely published in the open literature. For example,
Papadakis et al.[25] reviewed several correlations and recommended the
correlation
Xeq ¼ A exp½BT lnð1= Þ ð6Þ
where Xeq is the dry basis equilibrium moisture content in the solid (kg/kg
dry solid);  is the relative humidity of moisture in the surrounding gas
phase. A and B are empirical constants dependent on the solid and
moisture materials. Equation (6) is consistent with standard thermo-
dynamic principles of phase equilibrium. That is, the condition that at
thermodynamic equilibrium the chemical potential of any component is
ORDER REPRINTS

1378 Oakley

equal in all phases. In addition to temperature, pressure and composition,


desorption isotherms depend on the physical structure and chemistry of
the solid. A key factor is the binding energy between the moisture and
solid; the mechanism for which can vary from capillary suction in pores
to chemical bonding (as in hydrates for example). These can vary signif-
icantly with moisture content, leading to more complex curve shapes than
suggested by Eq. (6). Because of this complexity and dependency on
physical structure, in practice, desorption isotherms must be obtained by
experimental measurement. The desorption isotherms for the materials
concerned, can be combined with the heat and mass balance Eqs. (1)–(5)
above to form a solvable model for exit moisture content from the dryer.
The equilibrium model is simple with short calculation times and,
apart from specification of the conditions of the inlet streams, requires
only knowledge of the desorption isotherm of the material concerned.
Two potential flaws with this model are: the desorption isotherm will
normally have to be determined experimentally; and the assumption that
the exit solids will be in equilibrium with the surrounding gas. The model
could be made to accommodate nonattainment of equilibrium by the use
of an approach-to-equilibrium factor but this would have to be user-
specified and would give no insight to how the performance of the dryer
changes under changing operating conditions. If the latter assumption
does not hold, then the value of the equilibrium model is reduced. Figure
2 shows drying time predicted to approach equilibrium for various
droplet sizes containing solids. Clearly it is safest to use the equilibrium
model when particle sizes are small, drying kinetics fast and residence
times long. There is evidence to suggest that small particles, at least, come
close to attaining their equilibrium moisture content in a spray dryer.
Langrish and Kockel[26] who used CFD to simulate the spray drying of
milk powder, stated that the drying time of 80 micron particles is virtually
complete within 1 s—substantially shorter than the residence time in most
spray dryers. Results of an experimental study by Ozmen and Langrish[27]
also showed that the exit moisture content of solids from a spray dryer
are close to their equilibrium value (although in this case the final particle
size was only 30 micron).

Level 2A—Rate Based Model with Simplified Particle Motion

If particles in a spray dryer are not close to attaining equilibrium then


a model which takes account of the drying rate of the particles and their
residence time in the dryer is required for realistic prediction of exit solids
moisture content. Here the model objectives are to calculate the drying
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1379

10
300 micron
200 micron
100 micron
Xeq

1
0 10 20 30 t (s) 40 50 60 70
X (kg/kg dry basis)

0.1

Xeq

0.01

Figure 2. Drying times of water drops containing solids in air [Tg ¼100 C;
Yg ¼ 0.1 kg/kg; Xcrit ¼ 1 kg/kg db]. (View this art in color at www.dekker.com.)

rate of a droplet and its final moisture content but avoid the need to
calculate detailed gas flow patterns and particle trajectories which would
necessitate costly CFD simulations. Models of this type are descendants
of the early models described by Katta and Gauvin[3,4] and Keey and
Pham.[5] The model described here is an adapted version of the SPRY
models developed by the Separation Process Service (SPS) research in the
late 1980s and fully detailed in their reports.[28–30] Important aspects of
this are (1) A model of droplet-particle drying rates, and (2) A method of
estimating droplet-particle residence times in the spray dryer.

Droplet Drying Model

The moisture content of a particle at the exit of the dryer is calculated


by integrating the drying equation
dX
Md,S ¼N ð7Þ
dt
ORDER REPRINTS

1380 Oakley

over the particle’s residence time in the dryer. Where Md,S is the mass of
dry solids in the droplet-particle (kg); X is the moisture content of the
droplet-particle per mass of dry solids (kg/kg dry basis) and N is the
drying rate (kg s1). Solution of a differential equation is necessary since
the drying rate will change significantly with time. The drying rate of the
droplets containing solids are calculated using the Characteristic Curve
concept in conjunction with well established methods for calculation of
the evaporation rate from pure liquid droplets. If required, the droplet
size distribution leaving the atomizer can be represented by solving the
drying equation for more than one drop-size.

Drying of Droplets without Solids

In the case of droplets without solids, drying rates are determined by


mass transfer of moisture through the boundary layer surrounding the
droplet driven by the difference between the partial pressure of moisture
vapor at the surface of the droplet and in the bulk gas phase. The
droplet temperature equation needs to account for the heat input to the
droplet by convection and the heat absorbed by the liquid evaporation.
For moderate driving forces the equations are as follows:
Pure droplet mass

dMd
 ¼ N ¼ kAd ðPM ðTd Þ  PM,g Þ ð8Þ
dt

Pure droplet temperature

dTd dMd
Cpd Md ¼ hAd ðTg  Td Þ þ  ð9Þ
dt dt

where N is the evaporation rate (kg/s); Md is the mass of the droplet


(kg); k is the mass transfer coefficient based on a partial pressure driving
force (kg s1Pa1); Ad is the surface area of the droplet; PM is the
saturated vapor pressure of the moisture component; PM,g is the partial
pressure of moisture in the bulk gas phase. Cpd is the specific heat
capacity of the droplet (J kg1K1). The heat and mass transfer
coefficients, h and k, are determined using empirical correlations for
heat and mass transfer to droplets, for example those by Ranz and
Marshall.[31]
The heating-up time of droplets is very short and they quickly reach
a dynamic equilibrium temperature, known as the wet-bulb temperature
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1381

Twb. For the example of droplet drying given in Fig. 2 the time taken to
heat the 300 micron droplet from 0 C to the wet-bulb temperature is
about 0.1 s; for the 100 micron droplet the time is about 0.02 s. At the
dynamic equilibrium temperature the left-hand side of Eq. (9) is zero and
hence Twb is given by
dMd
hAd ðTg  Twb Þ ¼  ¼ :kAd ðPm ðTwb Þ  PM,g Þ ¼ :N^ ð10Þ
dt

Drying Rate of Droplets Containing Solids

The influence of solids on the droplet drying rate is determined using


the Characteristic Drying Curve concept described in the introduction.
The drying rate, N (kg/s), of a droplet containing solids is related to the
unhindered drying rate, N^ (kg/s), of a droplet without solids by

N ¼ f N^ ð11Þ

The unhindered drying rate is the drying rate of a pure liquid droplet
at its wet-bulb temperature given by Eq. (10). The Relative Drying Rate,
f, must be obtained from a measured drying curve. According to the
Characteristic Drying Curve Concept, for a given material, f is a unique
function of the characteristic moisture content

ðX  Xeq Þ
¼ ð12Þ
ðXcrit  Xeq Þ

Xeq is the equilibrium moisture content at the given conditions given by


Eq. (6) or similar. Xcrit is the moisture content at which the drying rate
becomes hindered, i.e., the moisture content at which the actual drying
becomes less than the unhindered rate of a pure liquid droplet under the
same conditions. By definition, above Xcrit, drying is unhindered and
f ¼ 1, below, Xcrit, f is between 0 and 1. Apart from these constraints, the
model itself places no restrictions on the shape of the f vs.  curve.
However, there are practical difficulties in obtaining a full drying curve
for a small droplet. To simplify matters further for many materials it is
acceptable to assume a linear falling rate (for example see Ref.[14]); in
other words the relative drying rate varies linearly between Xcrit and Xeq.
f ¼ ð13Þ
ORDER REPRINTS

1382 Oakley

Droplet-Particle Size Changes During Drying

To complete the model a relationship for calculating the droplet-


particle size is also required. The standard assumption here is that above
the particle forming moisture content, Xpart (kg/kg dry basis), the
droplet-particle has a free liquid surface and exhibits ideal shrinkage. In
other words, the volume reduction of the droplets is equal to the amount
of liquid that has been evaporated. Below Xpart the droplet-particle
behaves as a particle with fixed diameter, dpart.
 
part 1=3
X > Xpart dd ¼ dpart 1 þ ðX  Xpart Þ
l
X  Xpart dd ¼ dpart ð14Þ

where l and part are the densities (kg m3) of pure liquid moisture and
the dry particle respectively. Single droplet drying experiments[20,21] have
shown that droplet-particle sizes during drying can be more complex than
this, for example, with shrinking or puffing taking place after the particle
has formed. Furthermore, in some dryers (particularly counter-current)
agglomeration of particles may need to be considered. In principle there
is no restriction on the size changes that can be specified. However, in
practice such detail is rarely known. Whereas, parameters for the simple
model given by Eq. (5) can be determined by examining the density and
size distribution of the final dried particles.

Particle Residence Time—The Chamber Coefficient

A key part of the SPRY model is the use of an empirical factor, ", the
chamber coefficient, to relate the droplet-particle residence time,  d with
the mean gas residence time,  g.
d ¼ "g ð15Þ
The mean gas residence time in the chamber is straight-forward to
calculate from the volume of the chamber, Vchamber (m3) and the
volumetric flow rate of gas, V_ (m3/s) at exhaust conditions
Vchamber
g ¼ ð16Þ
V_ g
This approach to calculating the particle residence time avoids the
need to directly calculate particle trajectories which would require full
calculation of gas flow patterns in the chamber. " is an empirical constant
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1383

which must be determined by fitting it to the performance of an


operational spray dryer or the predictions of a full simulation—i.e., using
a computational fluid dynamics model to calculate particle trajectories.
Clearly the usefulness of this model depends on whether " varies in a
predictable way such that its value at one set of conditions can be used to
determine dryer performance at another set of conditions.
The Separation Processes Service [SPS], have performed an extensive
series of measurements to determine how " varied over a wide range
of conditions.[28–30] This includes both pilot scale and full scale spray
dryers, with a variety of test materials. This work drew the following
conclusions: " is a function of the features of the dryer, such as the choice
of atomizer, and the method of introducing the drying gas into the
chamber. However, " for a specific dryer varies little with operating
conditions. An important conclusion from the above work is that,
although the SPRY type model cannot be used as a method of design;
once the chamber coefficient has been determined for a spray dryer, it can
be used to predict the influence of changes in operating conditions on
dryer performance such as increasing gas flow-rate through the dryer.
Another example quoted is investigating the impact of converting a spray
dryer from once-through to recycle operation with a consequently large
increase in inlet gas humidity.
Its worth making a few further comments about the evaluation of ".
Preferably it will be fitted to existing operational data. Failing that, for a
new design for example, it could be determined using computational fluid
dynamics to calculate droplet-particle trajectories and their residence
times. In practice there are always uncertainties in modeling spray dryers
because of difficulties of measuring droplet drying kinetics at realistic
conditions and other factors. If " is fitted to dryer performance then its
value will factor-in these inaccuracies, making it a robust method of
investigating the affect of changes in operating conditions.

Level 2B—Rate-Based with Simulation of


Gas Flow and Particle Motion (CFD)

The limitation of the Level 2A class of model is that the motion of


the particles through the chamber is not actually modeled but fitted
empirically. This is fine if you have a source of data (in other words, an
operational dryer) to which the model can be fitted. But, such an
approach cannot be used to investigate new or modified designs where
changes to the particle motion through the chamber are a concern. In the
Level 2B model the gas flow patterns and trajectories of droplet-particles
ORDER REPRINTS

1384 Oakley

are calculated on a theoretical basis using computational fluid dynamics


(CFD) to get detailed predictions of the motion of particles, their
residence time in the chamber and the drying conditions they encounter.
It is not the intention here to give a detailed account of the CFD
modeling of spray dryers which is described adequately elsewhere. But it
is worth summarizing the main points

1. A 3 dimensional grid is constructed which conforms to the


geometry of the chamber.
2. The gas flow field in the chamber is calculated by solving the
mass, momentum conservation equations, together with con-
servations equations for energy and turbulence quantities for
each cell of the 3-dimensional grid.
3. Once the gas flow field has been solved, particle trajectories can
be calculated by solving their equations of motion taking into
account drag and gravity forces. Particle trajectories could be
calculated using time-averaged gas conditions; or a statistical
model could be used to account for the influence of turbulent
fluctuations on the motion of the particle. The latter is more
realistic but significantly increases the computing time.
4. In parallel to 3, the heat and mass transfer from droplets and
drying rates are calculated using an appropriate drying model.
The characteristic drying curve (as in Eqs. (8)–(14) above)
has been favored by a number of workers[9,10,14] but more
sophisticated drying models could be used if the higher level of
detail is warranted e.g., if modeling of variations in moisture
content and temperature across a droplet was considered
necessary.
5. The droplets-particles have a coupling affect on the velocity,
temperature, and humidity fields of the gas. This is accounted for
by adding source terms for the coupling back into the gas
equations and repeating step 2. Steps 2, 3, 4, and 5 are repeated
until convergence.

Experimental validation of the gas flow patterns and particle


trajectories predicted by these models was an important aspect of getting
their widespread acceptance. Noninvasive measurement of gas and
particle motion within an enclosed spray chamber are by no means
straightforward and have normally been achieved by either laser or hot
wire anemometry.[9,11] Most published data is on lab-scale spray dryers,
data on industrial scale spray dryers[10] are limited because of the
practical difficulties and propriety nature of such measurements.
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1385

The demands of CFD modeling can depend heavily on dryer


configuration and operation. For obvious reasons, chambers with single
atomizers and simple axisymetric geometries are more straightforward to
deal with than multiple nozzle configurations or complex geometries. The
author can vouch for this having been involved in a time-consuming but
ultimately successful attempt to model flows in a dryer fitted with over
100 nozzles! Such considerations are less important nowadays with better
handling of complex geomtries and improvements in computer speed.
Counter-current spray dryers (see Fig. 1), typically used in detergent
manufacture, have been found more difficult to simulate satisfactorily
compared with the more common co-current spray dryers. This is partly
because the internals of counter-current dryers are more complex with
multiple nozzles arranged over several layers; but mainly because
agglomeration is such an important feature of such dryers. Evidence
for agglomeration can be seen by observing the large size and granular
nature of particles collected at the bottom of the chamber.
Agglomeration significantly impacts drying times and other particle
behaviors but currently cannot be modeled in an acceptable way.
Although the method of dealing with droplet motion and drying in
CFD spray drying models have not changed radically in recent years,
there have been significant improvements to the efficiency, accuracy, and
robustness of the underlying models for the gas phase: the use of semi-
automatic mesh generation with unstructured meshes allows rapid
generation of good quality meshes which can resolve larges changes
of flow rates in the inlet region; improved turbulence models and
convergence for the strong swirling flow frequently found in spray dryers.
The general speed increase in computers together with the use of parallel
computing has increased the speed of solution. However, CFD is not
currently able to predict complete spray break-up processes taking place
during atomization to give the particle size distribution.
The chief advantages of using CFD are that there are no restrictions
on the chamber geometry that can be investigated. Unlike the Level 2A
model there is no need to fit an empirical model to the particle motion.
Furthermore, there is a very high level of detail in the predictions which
include gas velocities, temperature, histories in every part of the flow
field; and full histories of particle moisture content, temperature, velocity,
and position. The chief disadvantages are associated with the high level
of effort to construct the grid and long calculation times. The main use of
computational fluid dynamics models would be the investigation of
problems associated with the aerodynamics or chamber design which
could not be investigated by other models.
ORDER REPRINTS

1386 Oakley

PRACTICAL APPLICATION OF
SPRAY DRYER MODELING

In this section the practical use of the model layering concept and the
spray dryer models described above is demonstrated by reference to a real
industrial spray drying operation.

Scoping and Preliminary Process Design

In preliminary process design the main objective will be to obtain


estimates of flow-rates and heat inputs into the process for approximate
equipment sizing. Normally an aim will be to identify an envelope of
feasible operating conditions rather than precise predictions. For this
purpose the model should be quick and simple to use so that it is practical
to rapidly investigate various operational scenarios before fixing on
more specific conditions for detailed investigation. Also inputs into the
calculation should be limited because detailed input parameters of both
the equipment and material may not be available. The higher level models
(Level 2) can therefore be ruled out because of the detailed input
parameters and long calculation times required.
Heat and Mass Balances combined with a saturation curve for the
vapor–liquid system of interest, i.e., the Level 0 model, provide a
convenient method of analysis in such a situation. Exit solids moisture
contents is an input into the Level 0 heat and mass balance model.
However, by fixing this to a target value, conclusions can be drawn
from the proximity of the exhaust conditions to the saturation line.
Table 1 gives overall specification for the example process. Milk
powder has been chosen because of the availability of data on dryer
operation and material properties but the principles apply to any solid
wetted with any solvent. A co-current spray dryer in a once-through
system (Fig. 1) is used; the target moisture content is fixed at 0.04 kg/kg
of dry solids. The situation is shown in Fig. 3 in the form of a Mollier
chart. Inlet condition and estimated exhaust conditions are shown on
the plot. A small allowance (5% of total heat input) has been made for
heat loss.
Any exhaust conditions which approaches too closely or even cross
the dew point curve ( ¼ 1) will mean incomplete drying of the solid. A
practical measure of closeness to saturation is the approach-to-dew point
i.e., the gas temperature minus its dew point. Although purely empirical
this has proven a practical method of setting realistic exhaust conditions.
Of course, the acceptable closeness to saturation for a particular drying
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1387

Table 1. Sample process—basic specification.

Product Milk powder


Moisture Water
Feed moisture content 1.2 (kg/kg dry solids)
Product moisture content 0.04 (kg/kg dry solids)
Production rate 0.77 (kg/s dry solids)
Inlet gas Air
Inlet gas temperature 215 C

Mollier Chart for Air/water at 101.3 kPa


280 300 320 340 360 380 400 420

260
220
240
200
220 Dryer inlet, (215 C, 0.0003 kg/kg)
180
200

160
Enthalpy kJ/kg dry gas

180

Gas Temperature (C)


160 140
Ψ = 0.001
Enthalpy (kJ/kg)

140
120
120
100
100
Dryer exhaust, (88.89 C, 0.04568 kg/kg)
Ψ = 0.01 80
80
Approach-to-dew point 60
60 Xeq=0.04 kg/kg db = 50oC
Ψ = 0.1 40
40
dew point
20 Heater inlet Ψ = 1.0 20

0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05 0.055 0.06 0.065
Gas humidity Gas humidity
kg moisture/kg dry gas

Figure 3. Mollier chart for sample process. (View this art in color at
www.dekker.com.)

operation will depend on drying kinetics and residence times. Historical


plant data would be an acceptable method of estimating a realistic value
for the approach-to-dew-point. A review of spray dryer operation for this
material indicates approach-to-dew points will be in the range 30–80 C
and based on this an estimated of 50 C is used.
If the equilibrium moisture content curve (or desorption isotherm)
for the material is available then using the equilibrium model it is possible
ORDER REPRINTS

1388 Oakley

to be more definitive about the exit solids moisture content and allowable
approaches to saturation. As outlined in the previous section, the
equilibrium model will give the minimum exit solids moisture content
achievable under a given set of conditions. Furthermore, under certain
conditions (i.e., small particle sizes, long residence times, fast drying
kinetics) the exit moisture content achieved will be close to the equi-
librium moisture content. From Ozmen and Langrish[27] the equilibrium
moisture content, Xeq, is fitted by Eq. (6) with A ¼ 0.1499 kg/kg and
B ¼ 2.306  103 K1. Figure 3 includes a curve of the necessary
conditions to obtain an equilibrium moisture content of 0.04 (kg/kg
db) given these parameters. This gives additional confidence that the
target moisture content is achievable under the conditions specified
because there is a safety margin between the Xeq line and the exhaust
conditions.

Investigation of Process Change Scenarios

Here the objective is to use a simulation to investigate the affect of


changing process conditions on an existing process. This could be to
investigate increases in production rate, to improve energy efficiency or
to investigate the impact of changing feed conditions. The predictive
requirements are more demanding than above because the model must be
capable of predicting output from the dryer and how this is influenced
by changing conditions. Prediction of the exit moisture content is
particularly important because this is normally the main target
specification for a drying operation. Furthermore, it may be necessary
to consider the spray dryer as part of an overall process rather than in
isolation because of its impact on other unit operations. A good example
is a spray dryer within a closed recycle loop with solvent recovery.
Changing the gas flow rate, say, would not only affect spray dryer
performance but also heater and condenser loads, pressure drops and the
operation of particle collection equipment.
Normally, the impact of changing flow-rates on the process as a
whole would need to be considered and this could be done using a process
simulation package. The scenarios described below have been investi-
gated with Aspen PlusTM in conjunction with models described above.
The requirement for a predictive model capability (i.e., it must be capable
of predicting moisture content) means that the Level 0 alone could not be
used for the purpose. If there was confidence that equilibrium was
reached then the equilibrium-based model could be used but normally a
rate-based model which took account of variations in residence time
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1389

Table 2. Sample process—operating data.

Feed moisture content 1.2 (kg/kg dry solids)


Feed rate 1.7 (kg/s total)
Product moisture content 0.04 (kg/kg dry solids)
Product particle D32 100 mm
Inlet gas temperature 215 C
Inlet gas flow rate 18.9 kg/s
Chamber volume 343 m3
Fitted chamber coefficient 0.13

would be required. An additional factor to consider is that models with


long run-times (e.g., CFD based models) would generally be impractical
to use because they limit the opportunity to explore a wide range of
operating conditions. Also a single process simulation may need (if
included in a recycle loop for example) to repeat a given unit operation
model many times. This last requirement rules out the use of CFD based
models with run times of several hours. However, using the Level 2A
rate-based model is possible since if the dryer is already in operation there
is plenty of operational data to which the empirical chamber coefficient
can be fitted.
Table 2 shows relevant parameters for the drying process under
investigation. This is a real operational co-current dryer and the chamber
coefficient has been fitted to historical plant data and the equilibrium
moisture content curve given above for milk powder.

Process Change: Increasing Production Rate

A common requirement is to increase the production rate of the


dryer by increasing feed rate. Figure 4 shows the impact predicted by
both the equilibrium and rate-based model of increasing feed rate without
changing any other input conditions for this dryer. The equilibrium
model predicts an increase in exit moisture contents because the increased
evaporative load decreases the exhaust temperature and increases its
humidity. The rate-based model indicates a further increase in exit
moisture content due to drying rate effects. If the drying gas flow rate was
increased in proportion to the increase in feed rate such that the exhaust
temperature and humidity are the same, then the equilibrium moisture
content will predict the same exit moisture content. This is where the
equilibrium model starts to fail because it indicates there is no limit to the
ORDER REPRINTS

1390 Oakley

0.1
Rate-based, Dry gas flow=18.9 kg/s
Equilibrium, Dry gas flow=18.9 kg/s
0.09
Rate-based, Dry gas flow=25 kg/s
Equilibrium, Dry gas flow=25 kg/s
0.08 Operating point
Exit solids Moisture (kg/kg dry solids)

0.07

0.06

0.05

0.04

0.03

0.02

0.01

0
1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3
Total Feed Rate (kg/s)

Figure 4. Predicted exit moisture content against feed rate for sample process.
(View this art in color at www.dekker.com.)

feed rate that can be handled by increasing the gas flow in this way. This
is not the case for the rate-based model because increasing the gas flow
rate reduces residence times in the dryer. If the drying gas flow is
increased too far there will be insufficient residence time for the particles
to dry. However, Fig. 4 does show that according to the rate-based
model, in this case, the feed rate can be increased while maintaining the
exit moisture content by increasing the proportion of drying gas. If the
drying gas flow rate is increased to 25 kg/s, then a feed rate of just under
2.1 kg/s can be handled whilst maintaining the target moisture content of
0.04 kg/kg db.
This demonstrates the feasibility of increasing the throughput in
energy-balance and drying rate terms. But other factors may need to be
considered before implementing such a scheme. Increasing the through-
put in this way will increase the specific energy consumption which may
be unacceptable. Will the pressure drop through the system be
significantly increased and can this be handled by the existing fan? Will
the atomizing system be able to deal with the increased feed or will a
significantly larger droplet-particle size be produced? With rotary
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1391

atomizers for example, the atomizer wheel speed can be increased to


maintain a constant drop-size. Finally, will increasing the gas flow
through the dryer disturb the gas flow patterns? In our experience minor
variations in gas flow will not significantly affect flow patterns (and hence
the chamber coefficient) but this could be tested with a one-off CFD
simulation if required.

Process Change: Operating with Gas Recycle

The operator wants to investigate whether it is feasible to operate


the dryer operation given above with gas recycle as an energy saving
measure. A simplified flow-sheet is given in Fig. 5. Because part of the
drying gas is recycled, the inlet gas humidity will be higher leading to
longer drying times. But the drying gas flow can be increased slightly to
compensate for this. Table 3 shows predictions of the rate-based model
(Level 2) for this example obtained by adjusting the inlet gas flow rate to
maintain an exit solids moisture content of 0.04 kg/kg dry basis. In this
case the energy savings by recirculation are relatively small but if these
were seriously considered, then a more detailed simulation of the process
would be in order. This would take into account aspects such: pressure
drop through the loop and fan duty; and particle collection efficiency and
fines build-up in the recycle. The affect of increased humidity and
different drying conditions on product quality would also need to be
considered.

Investigation of Spray Dryer Aerodynamics


and Particle Trajectories

Figure 6 shows a number of possible causes of poor spray dryer


operation associated with the chamber aerodynamics and the trajectories
of droplets and particles. These include the following: heavy wall deposits
caused by the impingement of particles on the chamber wall while still
wet; by-passing in the chamber leading to excessively short droplet-
particle residence times and wet solids at the exit. Naturally, only models
that predict gas flow patterns are capable of looking at these types of
problems—which rules out everything apart from CFD. In this case
run-time considerations are not so important because the spray dryer
chamber can be considered in isolation.
Figure 7 shows a real example where 2 chamber sizes (to scale) were
being considered for a spray drying application. Both were co-current
1392

HEATER
B10

FEED DRY-GAS INLET MAKE-UP

SPRY-DRY
ORDER

RECYCLE
REPRINTS

TAKE-OFF

DRY-OUT2
DRY-OUT1
EXHAUST
PURGE

PRODUCT

Figure 5. Spray dryer with recycle. (View this art in color at www.dekker.com.)
Oakley
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1393

Table 3. Predicted energy savings by recycling.

Recycle Drying gas flow Heater duty


(%) (kg/s) (MW)

0 18.7 3.68
25 19.5 3.57
50 20.5 3.41
75 22.5 3.24

atomizer
hot air
inlet
c a a) Wall/roof impingement
b) Insufficient residence time
c) Particle over-heating

particle
collection

Figure 6. Problems associated with poor chamber aerodynamics. (View this art
in color at www.dekker.com.)

spray dryers fitted with rotary atomizers. Computational fluid dynamics


simulations showed that, in the case of the smaller chamber, 187 mm
particles would strike the chamber wall while still wet. This would lead to
very heavy wall deposits and severe operating problems. For this reason
the larger chamber was adopted.
An instructive example of the importance of the aerodynamics of the
chamber and how these can influenced by variations in inlet conditions is
given in Figs. 8a and 8b. In this case the spray dryer, fitted with a 2-fluid
nozzle, had operational problems due to blocking of the exit duct by a
build up of product. One explanation put forward for this was incomplete
drying in the chamber leading to partially wet particles sticking to the
ORDER REPRINTS

1394 Oakley

143 µm

187 µm 109 µm

109 µm 187 µm

143 µm

Figure 7. Particle trajectories in 2 co-current spray chambers designs. (View this


art in color at www.dekker.com.)

walls of the exit duct. CFD simulation of the chamber revealed that part
of the problem was due to the combination of a 2-fluid nozzle with
narrow cone and angle and inlet gas with no swirl (Fig. 8a). This led to
poor utilization of the spray chamber with the spray concentrated along
the central axis of the chamber and short particle residence times. This
short residence time was insufficient for complete drying resulting in
partially wet product. A strategy to solve the problem, tested by CFD
simulation, was that introducing some swirl to the inlet gas would
increase particle residence times by reducing gas velocities on the central
axis of the chamber (Fig. 8b), thereby increasing drying time to
sufficiently dry product at the exit. It may be noted that in Fig. 8b
some particles strike the chamber wall. This probably was not a problem
because they are dry by this stage.
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1395

Particle
Gas velocity
trajectories
vectors

Gas velocity Particle


vectors trajectories

Figure 8. (a) Spray dryer with zero swirl. (b) Spray dryer operating with swirl.
(View this art in color at www.dekker.com.)
ORDER REPRINTS

1396 Oakley

Investigation of Product Quality Issues

It is well known both from observations of spray dryer operation and


single droplet drying experiments that a range of spray dried product
quality factors depend on the conditions experiences by the particles in
the spray chamber. Such factors include: density, attrition resistance,
retention of aroma and flavor (e.g., in food products), and other active
materials (e.g., sodium tripolyphosphate in detergents) and particle
morphology. It should also be noted that spray dried materials are
often heat sensitive. Single droplet drying experiments (e.g., Refs.[20,21])
allow these dependencies to be isolated and examined under laboratory
conditions. The detailed predictions of a CFD model, ideally in
conjunction with single droplet drying experiments, provide a powerful
tool for the examining product quality issues. In this instance use of a
more sophisticated drying model which resolves variations in tempera-
ture and concentration across a particle may be warranted (e.g., Ref.[18]).
Figure 9 shows particle temperature histories when drying in a
co-current dryer. In the early stages of drying the droplet-particle
normally quickly reaches and remains at the wet bulb temperature (see
Eq. (10)), which is significantly less than the gas temperature. In the latter
stages the evaporation rate is much slower and the particle temperature
approaches that of the surrounding gas. In a cocurrent dryer, however,
the gas temperature experienced by the dry particles should be low since
the gas is cooled as it passes through the chamber, thus protecting heat
sensitive materials from over-heating. Trajectory c in Fig. 6 and
temperature-history b in Fig. 9 indicate how this can go wrong. A
dried particle recirculating back to the inlet will experience significantly
higher temperatures than expected leading to poor product quality.

CONCLUSIONS

Spray dryer models, ranging in complexity from heat and mass


balances to computational fluid dynamics (CFD), have been investigated
for practical relevance to spray drying problems. This is outlined in
Table 4. The most appropriate model level to use will depend on the level
of detail and accuracy required as follows:

. Heat and mass balance models are useful for estimating


evaporative and heating loads, exhaust conditions and flows in
preliminary process design but do not predict exit moisture
content. They cannot predict variations in performance with
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1397

Inlet gas temperature


b) Over-heating due to recirculation
Particle Temperature,

Exhaust gas temperature

Hindered drying
a) Ideal in co-current flow

Wet-bulb
temperature

Unhindered drying

0
Time

Figure 9. Typical temperature-histories in a co-current spray dryer. (View this


art in color at www.dekker.com.)

operating conditions and are therefore unsuitable for process


simulation and optimization.
. Equilibrium based models, given (experimentally determined)
desorption isotherms for the solid of interest, predict exit solid
moisture contents at equilibrium. The assumption of equilibrium
may be acceptable for small particles dried in chambers with long
residence times in which case they are valid to use in process
simulation. In other cases these models give lower limits to the
exit moisture content attainable.
. Rate-based models account for the finite drying rates and predict
variations in exit moisture with dryer operating conditions and
residence times. As such they are appropriate for process
simulation (i.e., simulating the impact of changing operating
conditions on process performance) but require (experimental)
data on the drying kinetics of the material.
. There are two approaches to the rate-based models. In the
simplest, gas flows and particle motion are not predicted but
fitted to the known behavior of the system. Once the model
has been fitted to a given spray dryer, it is suitable for process
1398

Table 4. Comparison of spray dryer models.

Normal Approximate
Level Model type Normal inputs outputs run timea Application

0 Heat and Mass Inlet streams (flow, tem Exhaust gas < 1s Scoping, preliminary
Balances perature, composition) conditions design
Exit moisture content Heat requirements
1 Equilibrium-based Inlet streams As (0) + <1 s Scoping, preliminary
Desorption isotherm Exit moisture at design,
ORDER

equilibrium Process simulation


2A Rate-based with As (1) + Exit moisture v time 5-60 s Overall process
simplified Chamber volume Variations in simulation,
particle motion Chamber coefficient performance Process optimization
Droplet-Particle sizes
REPRINTS

Drying kinetics
2B Rate-based with full As (2A) + As (2A) + >5h Detailed designs
simulation of particle Detailed chamber geometry Gas flow field Investigation of
motion Detailed gas inlet Gas temperature field aerodynamics
conditions Particle trajectories Investigation of
Detailed spray inlet Particle temperature- product quality
conditions history
a
Run times are approximate and for illustrative purposes only. Based on 1800 MHz PC.
Oakley
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1399

simulation. It is therefore good for examining variations in


operating conditions of an existing process. However, since it
requires prior knowledge of particle motion in the chamber (i.e.,
through operating data), it is not appropriate for investigating gas
flows and particle motion in new or modified chamber designs.
. In the more complex rate-based model, computational fluid
mechanics is used to calculate gas flow and particle motion in the
chamber. However, long calculation times mean the model is
probably unsuitable for routine process simulation. There are no
alternatives when there is no prior knowledge of particle motion
and it is therefore an invaluable tool for investigating particle
motion in a chamber and new or modified chamber designs.

ACKNOWLEDGMENT

The author acknowledges the advice of Dr. Ian Jones of ANSYS


CFX on recent advances in computational fluid dynamics.

NOTATION

Ad Surface area of droplet (m2)


CPd Specific heat capacity of droplet (J/kg/K)
dd Droplet diameter (m)
f Relative drying rate ()
F Mass flow (kg/s)
Fi,j Mass flow of component j in stream i (kg/s)
g Acceleration due to gravity, approximately 9.81 m/s2
(m/s2)
H Enthalpy (J/kg)
h Heat transfer coefficient (W/m2 K)
kP Mass transfer coefficient (partial pressure driving force)
(kg m2 Pa1 s)
Md Mass of droplet (kg)
Md,s Mass of solids in droplet (kg)
N Drying rate (kg s1)
N^ Unhindered drying rate (kg s1)
PM,g Partial pressure of moisture in gas phase (Pa)

PM,g Saturated vapor pressure of moisture (Pa)
Q Heat loss (W)
ORDER REPRINTS

1400 Oakley

T Temperature (C)
t Time (s)
Vchamber Volume of chamber (m3)
V_ Volumetric flow rate (m3 s1)
X Solids moisture content (dry basis) (kg/kg dry basis)
Xeq Equilibrium moisture content (kg/kg dry basis)
Xcrit Critical moisture content (kg/kg dry basis)
Xpart Particle forming moisture content (kg/kg dry basis)
Y Absolute humidity of gas (kg/kg dry basis)

Greek Letters

 Latent heat of evaporation (J/kg)


 Relative humidity ()
 Characteristic moisture content ()
 Density (kg/m3)
 Residence time (s)

Subscripts and Superscripts

d Droplet
G Dry gas component
g Gas phase
S Dry solid component
M Moisture component
FEED Feed stream
IN-GAS Inlet gas stream
EX-GAS Exhaust stream
EX-SOLID Exit solids stream
crit At critical moisture content
wb Wet bulb conditions

REFERENCES

1. Friedman, S.J.; Gluckert, F.A.; Marshall, W.R. Centrifugal disk


atomization. Chemical Engineering Progress 1952, 48 (4), 181.
2. Herring, J.R.; Marshall, W.R. Performance of vaned-disk atomizers.
AIChE. J. 1955.
ORDER REPRINTS

Spray Dryer Modeling in Theory and Practice 1401

3. Katta, S.; Gauvin, W.H. Some fundamental aspects of spray drying.


AIChE. J. 1975, 21 (1).
4. Gauvin, W.H.; Katta, S. Basic concepts of spray dryer design.
AIChE. J. 1976, 22 (4), 713.
5. Keey, R.B.; Pham, Q.T. Behaviour of spray dryers with nozzle
atomizers. Chem. Engr. 1976, 311, 516.
6. Crowe, C.T. Modelling spray-air contact in spray drying systems. In
Advances in Drying; Mujumdar, A.S., Ed.; Hemisphere: New York,
USA, 1980; Vol. 1, Chapter 3.
7. Crowe, C.T. Droplet-gas interaction in counter-current spray
dryers. Drying Technology 1983, 1 (1).
8. Papadakis, S.E.; King, C.J. Air temperature and humidity profiles
in spray drying. 1. Features predicted by particle-source-in-cell
method and 2. Experimental measurements. Ind. Eng. Chem. Res.
1988, 27, 2111–2123.
9. Oakley, D.E.; Bahu, R.E Computational Modeling of Spray Dryers;
Computers & Chemical Engineering 1992, Volume 17, Supplement,
Proceedings of the European Symposium on Computer Aided
Process Engineering—2 (ESCAPE2), Toulouse.
10. Livesley, D.M.; Oakley, D.E.; Gillespie, R.F.; Elhaus, B.; Ranpuria,
C.K.; Taylor, T.; Wood, W.; Yeoman, M.L. Development and
validation of a computational model for spray-gas mixing in spray
dryers. In Drying ’92; Mujumdar, A.S., Ed.; Elsevier: Amsterdam,
1992; 407–416.
11. Kievet, F.G.; Kerkhof, P.J.A.M. Using Computational Fluid
Dynamics to Model Product Quality in Spray Drying: Air Flow;
Temperature and Humidity Patterns in Drying ’96, Proceedings of
the 10th International Drying Symposium (IDS ’96); Mujumdar,
A.S., Ed.; Krakow: Poland, 1996; 259–266.
12. Langrish, T.A.G.; Zbincinski, I. The effect of air inlet geometry and
spray cone angle on the wall deposition rate in spray dryers. Trans.
IChemE 1994, 72 (A), 420–430.
13. Keey, R.B. Drying of Loose and Particulate Materials; Hemisphere:
New York, 1992.
14. Langrish, T.A.G.; Kockel, T.K. The assessment of a characteristic
drying curve for milk powder for use in computational fluid
dynamics modeling. Chemical Engineering Journal 2001, 84, 69–74.
15. Wijlhuizen, A.E.; Kerkhof, P.J.A.M.; Bruin, S. Theoretical study of
the inactivation of phosphate during spray drying of skim milk.
Chemical Engineering Science 1979, 34, 651–660.
16. Stevenson, M.J.; Chen, X.D.; Fletcher, D. Modelling the Drying of
Milk; CHEMECA 98: Queensland, Australia, 1998.
ORDER REPRINTS

1402 Oakley

17. Cheong, H.W.; Jeffreys, G.V.; Mumford, C.J. A receding interface


model for the drying of slurry droplets. AIChE Journal 1986, 32,
1334–1346.
18. Farid, M. A new approach to modeling of single droplet drying.
Chemical Engineering Science 2003, 58, 2985–2993.
19. Masters, K. Spray Drying Handbook, 5th Ed.; Longman Scientific &
Technical: England, 1991.
20. Charlesworth, D.H.; Marshall, W.R. AIChE J. 1960, 6, 9.
21. Walton, D.E.; Mumford, C.J. The morphology of spray-dried
particles. Trans. IChemE 1999, 77 (Part A).
22. King, C.J. Spray drying: retention of volatile compounds revisited.
Drying Technology 1995, 13 (5–7), 1221–1240.
23. Coumans, W.J.; Kerkhof, P.J.A.M.; Bruin, S. Theoretical and
practical aspects of aroma retention in spray drying and freeze
drying. Drying Technology 1994, 12 (1–2), 99–149.
24. Kemp, I.C.; Oakley, D.E. Modelling of particulate drying in theory
and practice. Drying Technology 2002, 20 (9), 1699–1750.
25. Papadakis, S.E.; Bahu, R.E.; McKenzie, K.A.; Kemp, I.C.
Correlations for the equilibrium moisture content of solids.
Drying Technology 1993, 11 (3), 543–553.
26. Langrish, T.A.G.; Kockel, T.K. The assessment of a characteristic
drying curve for milk powder for use in computational fluid
dynamic modeling. Chemical Engineering Journal 2001, 84, 69–74.
27. Ozmen, L.; Langrish, T.A.G. A study of the limitations to spray
dryer outlet performance. Drying Technology 2003, 21 (5), 895–917.
28. Huber, R.A. Spray Dryer Design Programs (SPRY1 and SPRY2),
Research Report RR17; Separation Processes Service (SPS) now in
The Process Manual, Aspentech, Harwell, UK, 1982.
29. Robinson, R.A.; Huber, R.A. Spray Dryer Experiments 1983–85
and Testing of the SPRY1 Computer Code, Research Report RR39;
Separation Processes Service (SPS) now in The Process Manual,
Aspentech, Harwell, UK, 1989.
30. Robinson, R.A.; Huber, R.A. Spray Dryer Experiments 1986–88
and Testing of the SPRY1 Computer Code, Research Report RR63;
Separation Processes Service (SPS) now in The Process Manual,
Aspentech, Harwell, UK, 1989.
31. Ranz, W.E.; Marshall, W.R. Evaporation from droplets. Chem.
Eng. Prog. 1952, 48 (3–4).
Request Permission or Order Reprints Instantly!

Interested in copying and sharing this article? In most cases, U.S. Copyright
Law requires that you get permission from the article’s rightsholder before
using copyrighted content.

All information and materials found in this article, including but not limited
to text, trademarks, patents, logos, graphics and images (the "Materials"), are
the copyrighted works and other forms of intellectual property of Marcel
Dekker, Inc., or its licensors. All rights not expressly granted are reserved.

Get permission to lawfully reproduce and distribute the Materials or order


reprints quickly and painlessly. Simply click on the "Request Permission/
Order Reprints" link below and follow the instructions. Visit the
U.S. Copyright Office for information on Fair Use limitations of U.S.
copyright law. Please refer to The Association of American Publishers’
(AAP) website for guidelines on Fair Use in the Classroom.

The Materials are for your personal use only and cannot be reformatted,
reposted, resold or distributed by electronic means or otherwise without
permission from Marcel Dekker, Inc. Marcel Dekker, Inc. grants you the
limited right to display the Materials only on your personal computer or
personal wireless device, and to copy and download single copies of such
Materials provided that any copyright, trademark or other notice appearing
on such Materials is also retained by, displayed, copied or downloaded as
part of the Materials and is not removed or obscured, and provided you do
not edit, modify, alter or enhance the Materials. Please refer to our Website
User Agreement for more details.

Request Permission/Order Reprints

Reprints of this article can also be ordered at


http://www.dekker.com/servlet/product/DOI/101081DRT120038734

You might also like