Trends in Food Science & Technology 21 (2010) 383e398

Review

Computational fluid
dynamics (CFD) whey, instant coffee, milk, tea and soups, as well as health-
care and pharmaceutical products, such as vitamins, en-
zymes and bacteria. Normally, spray dryer comes at the
applications in spray end-point of the processing line, as it is an important step
to control the final product quality. It has some advantages

drying of food such as, rapid drying rates, a wide range of operating tem-
peratures and short residence times. In recent years, compu-
tational fluid dynamics (CFD) has been increasingly
products applied to food processing operations. In spray-drying op-
erations, CFD simulation tools are now often used because
measurements of air flow, temperature, particle size and hu-
Rinil Kuriakose and midity within the drying chamber are very difficult and ex-
pensive to obtain in large-scale dryer.
C. Anandharamakrishnan* CFD is a simulation tool, which uses powerful computers
in combination with applied mathematics to model fluid flow
Human Resource Development, Central Food situations and aid in the optimal design of industrial pro-
Technological Research Institute, Council for Scientific cesses. The method comprises solving equations for the con-
and Industrial Research (CSIR), Mysore 570 020, India servation of mass, momentum and energy, using numerical
(Tel.: D91 821 2514310; fax: D91 821 2517233; methods to give predictions of velocity, temperature and
e-mail: [email protected]) pressure profiles inside the system. It powerful graphics
can be used to show the flow behaviour of fluid with 3D im-
The application of computational fluid dynamics (CFD) in the ages (Anderson, 1984; Scott & Richardson, 1997).
area of spray drying of food ingredients is reviewed. The results The fluid motion was studied mathematically by
from previous studies have shown that CFD can be useful tool for Richardson in 1910 and later by Courant, Friedrichs, and
predicting the gas flow pattern and particle histories such as Lewy (1928). This basic study instigated the development
temperature, velocity, residence time and impact position. of powerful numerical techniques that have advanced the
Moreover, CFD can also be used for investigating different numerical description of all types of fluid flow (Shang,
chamber design on drying behaviour of food materials as well 2004). CFD activity emerged and gained prominence with
as scale-up studies. This paper provides the aireparticle interac- availability of computers in the early 1960s and it is applied
tions, different turbulence models, reference frames, reaction to all aspects of fluid dynamics (Parviz & John, 1997).
engineering model used for spray-drying applications along Since the early 1970s, commercial software packages (or
with the particle histories and different spray chamber configu- computer codes) became available, making CFD an impor-
rations. The recent developments in this research area such as tant component of engineering practice in industrial, de-
drying kinetics model, dropletedroplet interactions, unsteady fense and environmental organizations. By the 1990s,
state simulations and integration of population balance model advances in computing power produced a similar boom in
are discussed. In addition, the current limitations and future software development and solutions. Since then CFD has
scope for the potential research are also highlighted. been used extensively by the scientific community world-
wide. The reason for this interest is justified by the fact
that it is possible to model computationally the physical
fluid phenomena that are expensive and time consuming
Introduction
to measure directly. For more detailed historical perspective
Spray drying is a well established method for converting
the books by Roache (1976) and Tannehill, Anderson, and
liquid feed materials into a dry powder form. Spray drying
Pletcher (1997) are highly recommended. Today, CFD finds
is widely used to produce powdered food products such as
extensive usage in basic and applied research, in design of
engineering equipment and in calculation of environmental
* Corresponding author. and geophysical phenomena.
0924-2244/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tifs.2010.04.009
384 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398

Although the origin of CFD can be found in the automo- The objective of the current paper is to summarise the
tive, aerospace, and nuclear industries and also may other application of CFD in spray drying using different turbu-
processing industries, it is only in the recent years that lence models and reference frames with their feasibility
CFD has been applied to the food processing area (Scott, along with particle histories. Moreover, the recent develop-
1994). The ability of CFD to predict the performance of ments and future perspectives are also highlighted.
new designs or processes before they are ever manufactured
or implemented make them an integral part of engineering Spray drying
design and analysis (Schaldach, Berger, Razilov, & Berndt, Spray drying is the process of transforming a feed (solu-
2000). The general application of CFD to the food industry tion or suspension) from a fluid into a dried particulate form
was reviewed by many researchers (Anandharamakrishnan, by spraying the feed into a hot drying medium. Spray dry-
2003; Norton & Sun, 2006; Scott & Richardson, 1997; Sun, ing is a widely used industrial process for the continuous
2007; Xia & Sun, 2002). All the above reviews concluded production of dry powders with low moisture content
that CFD is a powerful and pervasive tool for process and (Anandharamakrishnan, Rielly, & Stapley, 2007; Charm,
product improvement in food processing sector. 1971; Masters, 1991). As shown in Fig. 1 spray drying in-
The main application of CFD includes baking process volves four stages of operation: (1) atomisation of liquid
(Mills, 1998e1999; Therdthai, Zhou, & Adamczak, 2003; feed into a spray chamber, (2) contact between the spray
Vries, Velthuis, & Koster, 1994), refrigerated display cabi- and the drying medium (3) moisture evaporation and (4)
nets (Cortella, Manzan, & Comini, 1998; Foster, 1996), ther- separation of dried products from air stream.
mal sterilisation (Abdul Ghani, Farid, Chen, & Richards,
1999a, 1999b, 2001; Datta & Teixeira, 1987), pasteurization Atomisation
of egg (Denys, Dewettinck, Pieters, 2005; Denys, Pieters, & Atomisation is a process, where the bulk-liquid breaks
Dewettinck, 2003, 2004), mixing (Sahu, Kumar, up into a large number of small droplets. The choice of at-
Patwardhan, & Joshi, 1999; Scott, 1977), and refrigeration omizer is most important in achieving economic production
(Davey & Pham, 1997, 2000; Hu & Sun, 1999, 2000; of high quality products (Fellows, 1998). The different
Moureh & Derens, 2000; Mariotti, Rech, & Romagnoni, types of atomizer are rotary atomizer, pressure nozzle at-
1995), spray-freezing (Anandharamakrishnan, Gimbun, omizer and two-fluid nozzle atomizer (Masters, 1991;
Stapley, & Rielly, 2010), heating and cooling processes Mujumdar, 1987).
(Wang & Sun, 2003), humidification of cold storage
(Verboven et al., 2008; Verboven, Nicolaı̈, Delele, Schenk, Spray-air contact
& Ramon, 2009) is also noteworthy. The list given above is During spray-air contact, droplets usually meet hot air in
non-exhaustive and for detailed review of CFD applications the spraying chamber either in co-current flow or counter-
to food processing, reader may refer elsewhere (Sun, 2007). current flow. The large surface area of the droplets leads
CFD modelling has widely been used for drying pro- to rapid evaporation rates, keeping the temperature of the
cess to study the simultaneous heat and mass transfer droplets at the wet-bulb temperature. In this period different
problems (Masters, 1994; Mathioulakis, Karathanos, & products exhibit different characteristics such as, expan-
Belessiotis, 1998; Mirade & Daudin, 2000). The non-line- sion, collapse, disintegration and irregular shape.
arity and complex process involved during drying was re-
viewed by Kerkhof (1994). During 1970s a number of Separation of dried products
simplified semi-empirical models was developed to under- The dry powder is collected at the base of the dryer and
stand drying of specific product in a specific spray dryer removed by a cyclone separator or screw conveyor or
(Paris, Ross, Dastur, & Morris, 1971; Place, Ridgway, & a pneumatic system with a cyclone separator. Other
Danckwerts, 1959). The use of computational fluid dy- methods for collecting the dry powder are bag filters and
namics in the area of spray drying was reviewed early
by Reay (1988) and Bahu (1992). Latter, Fletcher et al.
(2003) and Langrish and Fletcher (2001, 2003) compre- 1
Feed Liquid
hensively reviewed the applications of computational fluid Furnace
Air
dynamics in spray drying of food ingredients to predict the 2
flow patterns and temperature distributions of gas and Exhaust gas
droplets inside the drying chamber. More recently,
Langrish (2009) has discussed about the different levels 4

and scales of mathematical modelling that can be applied Spray-drying 3

Cyclone
chamber
to the spray-drying process. However, reference frames, separator
particle histories (temperature, velocity, residence time
and impact positions), reaction engineering approach, Product
and population balance modelling are overlooked in the
early reviews. Fig. 1. The process stages of spray drying.
 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398 385

electrostatic precipitators (Fellows, 1998). The selection of diameter was used for the dispersed phase) then the Euler-
equipment depends on the operating conditions such as par- ian approach can be expensive. In addition there is some
ticle size, shape, bulk density and powder outlet position. uncertainty over the most appropriate Eulerian diffusion co-
efficients and heat transfer coefficients. Hence, the Eulerian
Types of spray dryer approach is best suited to flows with a narrow range of par-
The two main designs of spray dryer commonly used are ticle sizes where a high resolution of the particle properties
the short-form and tall-form driers. Tall-form designs are is not needed (Jakobsen, Sannaes, Grevskott, & Svendsen,
characterized by height-to-diameter aspect ratios of greater 1997; Mostafa & Mongia, 1987).
than 5:1. Short-form dryers, have height-to-diameter ratios In the EulerianeLagrangian particle tracking approach,
of around 2:1. The short-form dryers are the most widely the gas phase is modelled using the standard Eulerian ap-
used as they accommodate the comparatively flat spray proach described above and the spray is represented by
disk from a rotary atomizer. The flow patterns observed a number of discrete computational ‘particles’. Individual
in short-form dryers are more complex than those in tall- particles are tracked through the flow domain from their in-
form dryers, with many dryers having no plug-flow zone jection point until they escape the domain in a Lagrangian
and a wide range of gas residence times (Langrish & frame work (Nijdam, Guo, Fletcher, & Langrish, 2006).
Fletcher, 2001). The EulerianeLagrangian model has the advantage of be-
ing computationally cheaper than the EulerianeEulerian
Computational fluid dynamics method for a large range of particle sizes. It can also pro-
CFD is a numerical technique for the solution of the vide more details of the behaviour and residence times of
equations governing the flow of fluids inside defined flow individual particles and can potentially approximate mass
geometry. The flow of any fluid can be described using and heat transfer more accurately. On the other hand, the
the Navier’s stokes transport equations (Bird, Stewart, & approach can be expensive if a large number of particles
Lightfoot, 1960). These equations are derived by consider- have to be tracked and it is best when the dispersed phase
ing mass, momentum and energy balances in an element of does not exceed 10% by volume of the mixture in any re-
fluid, resulting in a set of partial differential equations. gion (Bakker, 2002).
They are completed by adding other algebraic equations In both the EulerianeEulerian and the Euler-
from thermodynamics such as the equation of state for den- ianeLagrangian methods, the exchange of momentum be-
sity and a constitutive equation to describe the rheology. tween particles and gas needs to be modelled. This
Detail descriptions of CFD and analysis of pre-processing, exchange can consist of several forces such as, drag, lift,
processing and post processing can be seen in standard text virtual mass, and wall forces. Mostafa and Mongia (1987)
books and published papers (Bakker, 2002; Fletcher, 2000; concluded that the Eulerian approach performs better than
Sun, 2007; Versteeg & Malalasekera, 1995; Xia & Sun, Lagrangian method. In contrast, Nijdam et al. (2006) found
2002). In recent years there was good progress in the devel- that both Eulerian and Lagrangian modelling approaches
opment of commercial CFD codes. Dombrowski, Foumeny, gave similar predictions for turbulent droplet dispersion
and Riza (1993) extensively reviewed the various CFD co- and agglomeration of sprays for a wide range of droplet
des with their applications and limitations. and gas flows. The two models were found to require sim-
ilar computing times for a steady axi-symmetric spray.
Reference frames However, the authors preferred the Lagrangian models be-
The two most general two-phase modelling frames are cause of their wider range of applicability.
the EulerianeEulerian and the EulerianeLagrangian
methods. In the EulerianeEulerian frame, the dispersed Air flow pattern
phase (droplets) are treated as a continuous (Eulerian) During spray drying the particle behaviour is dependent
phase, i.e. there are two Eulerian phases, one for the gas on the air flow pattern. Earlier studies were based on the
and another for droplets, which are interacting and inter- semi-empirical models for understanding of the spray-dry-
penetrating continua (Mostafa & Mongia, 1987). Each ing process. Inside the spray chamber there is presence of
computational cell contains certain fractions of gas and significant air flow instabilities due to the inlet swirl.
droplets, and the transport equations are written in such Hence, the effect of turbulence inside the spray chamber
a way that the volume fractions of gas and liquid sum to should be considered.
unity. If the computational cell consists of just a single
phase, the transport equations for the two phases revert to Turbulence models
the conventional single-phase system. The advantages of Most commercial CFD codes use turbulence models that
the EulerianeEulerian approach are that it is usually rela- are based on the splitting up of instantaneous quantities
tively cheap in terms of computational demands for one ad- into a time-averaged and a fluctuating part by a process
ditional set of equations and turbulence can be modelled known as Reynolds decomposition. Four turbulence
fairly simply. However, if a separate set of transport equa- models are commonly used for simulating sprays: (i) stan-
tions is solved for each particle size (single particle dard ke3 (k e turbulence kinetic energy and 3 e turbulence
386 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398

dissipation rate) (Launder & Spalding, 1972, 1974) (ii) k2

RNG ke3, (Yakhot & Orszag, 1986) (iii) realizable ke3, mt ¼ rCm ð3Þ
3
(Shih, Liou, Shabbir, & Zhu, 1995) (iv) Reynolds Stress
Model (RSM) (Launder, Reece, & Rodi, 1975). Three For calculating an approximate solution of fluid flow
models (standard, RNG, and realizable ke3) have similar equations, the equations have to be made discrete. For
forms, with transport equations for k and 3. The standard this, the flow domain is divided into number of control vol-
ke3 model focuses on the mechanisms that affect the turbu- umes. This is called a grid and at each grid cell approxi-
lent kinetic energy. It can be used over a wide range of tur- mate solutions for the NaviereStokes and the continuity
bulent flows due to its robustness and reasonable accuracy. equations are calculated.
In the realizable ke3 model, the term “realizable” means A model which can predict the effects of turbulence (dif-
that the model satisfies certain mathematical constraints ferential Reynolds stress model) under strong swirling condi-
on the normal stresses, consistent with the physics of turbu- tions was reported by Oakley and Bahu (1991). Although the
lent flows. It provides better prediction for flows involving model predictions showed good agreement with the experi-
rotation, boundary layers under strong adverse pressure mental data, the differential Reynolds stress model involved
gradients, separation, and recirculation (Anderson, 1984; a higher computational effort than the standard ke3 model.
Fletcher, 2000). Livesley et al. (1992) used the ke3 model in their work for
The RNG-based ke3 turbulence model is derived from the simulation of industrial spray dryers handling solutions
the instantaneous NaviereStokes equations, using a mathe- and slurries. They also found satisfactory agreement between
matical technique called “renormalization group” (RNG) the numerical solution and the experimental measurements.
methods. The analytical derivation results in a model with However, Huang, Kumar, and Mujumdar (2004a) showed
constants different from those in the standard ke3 model that RNG ke3 model prediction was better for swirling
and additional terms and functions in the transport equa- two-phase flow in the spray-drying chamber for drying of
tions for k and 3. The effect of swirl on turbulence is in- maltodextrin e sucrose solution compared to standard ke3,
cluded in the RNG model, which enhancing accuracy for realizable ke3 and Reynolds stress models.
swirling flows. Oakley (2004) used four different models such as (i)
The Reynolds Stress Model has the same general form heat and mass balances, (ii) equilibrium-based models,
as the instantaneous NaviereStokes equations, with the ve- (iii) rate-based models, and (iv) CFD models for modelling
locities and other solution variables ensemble-averaged (or of spray-drying operations. Author compared the above
time-averaged). The RSM is clearly superior for problems four models and described the relevance of each model
where anisotropy of turbulence has a dominant effect on for practical applications. The importance of applied gas
the mean flow (e.g. highly swirling flows) (Bakker, 2002). turbulence model, drying kinetics, effect of atomizing air
For turbulent flows, the standard ke3 model (k e turbulence and turbulent particle dispersion in the modelling of spray
kinetic energy and 3 e turbulence dissipation rate) is the dryers using liquid suspension of baker’s yeast as feed so-
most commonly used, because it converges considerably lution was studied by Li and Zbicinski (2005). They simu-
better than Reynolds stress model (RSM) (Bakker, 2002). lated a cocurrent spray dryer and determined the initial
The standard ke3 turbulence model is used commonly parameters of discrete and continuous phases experimen-
where there is no swirling flow. The transport of the turbu- tally to be used in the model. In a continuing work,
lence kinetic energy k and its dissipation rate 3 is given as Zbicinski and Li (2006) presented conditions for accurate
follows: CFD modelling during spray dying of maltodextrin solution
at different solid concentrations. They compared the simu-

   lation results with the experimental data and concluded that
v mt
ðrkÞ þ V$ðrk v Þ ¼ V$ mþ Vk þ Gk  r3 ð1Þ a maximum error of 20% will occur in the simulation re-
vt sk sults based on the discrete phase parameters. A brief review
of the air flow pattern studies in spray dryers was done by
and Southwell and Langrish (2000). Table 1 summarises the
works performed on the CFD simulation of air flow pattern

  
v mt 3 during spray drying with their findings.
ðr3Þ þ V$ðr3 v Þ ¼ V$ mþ V3 þ C13 ðGk Þ
vt s3 k
32 Atomisation
 C23 r ð2Þ The atomisation stage during spray drying is very impor-
k
tant since it affects the final particle size. A co-current
Gk is the generation of kinetic energy due to the mean spray-dryer fitted with pressure nozzle was investigated
velocity gradients. The quantities sk and s3 are the turbu- both in experiment and CFD simulation by Kieviet and
lent Prandtl numbers for k and 3, respectively and C13, Kerkhof (1996) to develop a theoretical model for the pre-
C23, are constant. The turbulent (or eddy) viscosity, mt is diction of final product quality. Good agreement was ob-
calculated from k and 3 as follows tained between the experimental data and the simulation.
 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398 387

Table 1. Spray drying e air flow pattern studies.

Problem Model geometry Turbulence model Software Findings Authors

descriptions
Simulation of air 2D Standard ke3 and FLOW3D Non-swirling flow Oakley and Bahu
flow pattern with RSM spray chamber, the (1993)
experimental ke3 model gives
validation. good predictions of
gas velocity profiles,
whereas, for swirling
flows RSM model
gives better accurate
predictions.
Simulation of air 2D Standard ke3 FLOW3D Strongest oscillations Langrish et al. (1993)
flow pattern to find occur. Good
out the oscillations agreement between
in the flow field. hot-wire
anemometer velocity
measurements and
simulation results.
Effects of the air inlet 2D Standard ke3 CFX High swirl in the Langrish and
geometry and spray inlet air and large Zbicinski (1994)
cone angle on wall spray cone angle
deposition rates. gave the lowest wall
deposition rates in
both the experiments
and simulation.
Simulation of air 2D Standard ke3 CFX Good agreement Zbicinski (1995)
flow and particle between
trajectories in the measurements and
tall-form dryer with simulation results
experimental
validation.
Simulation of air 2D Standard ke3 CFX Model prediction Kieviet (1997)
flow pattern, agreed well with his
temperature, experimental
humidity, particle measurements of
trajectories and velocity, temperature
resistance time in and humidity.
a co-current spray
dryer fitted with
a pressure nozzle
Simulation studies 2D Standard ke3 CFX An increase in the Southwell, Langrish,
on the effects of amount of and Fletcher (1999)
increased turbulence evaporation resulted
in inlet air flow. directly from
enhanced inlet
turbulence.
Temperature and 2D Standard ke3 NIZO-DrySim The drying of Straatsma, Van
moisture content of droplets is Houwelingen,
the air with the influenced by Steenbergen, and De
trajectories of the particles surface to Jong (1999)
particles. surrounding air and
diffusion within the
particles.
Investigating the air 2D Standard ke3 Fluent The optimal Huang et al. (2003b)
flow pattern, chamber geometry
temperature, velocity will depends on the
and humidity profile feed properties,
at different spray atomizer type and
dryer chamber drop size distribution
configuration.

(continued on next page)

388 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398

Table 1 (continued)
Problem Model geometry Turbulence model Software Findings Authors
descriptions
Experimental and 3D RSM CFX 5.6 Comparison of with Langrish, Williams,
simulation studies of and without spray and Fletcher (2004)
inlet air swirl on the showed that the
stability of the flow introduction of spray
pattern in spray has significant effect
dryers. on the flow
behaviour. An
increase in swirl
angle changes the
internal flow pattern.
Simulation of a spray 3D Standard ke3, RNG Fluent 6.1 Realizable ke3 Huang et al. (2004a)
dryer with rotary ke3, realizable ke3 cannot be used to
atomizer. Kieviet’s and RSM simulate highly
(1997) spray dryer swirling two-phase
geometry was used. flow. RNG ke3
turbulent model
gives adequate
accuracy at
reasonable
computational time.
Simulation of spray 3D RNG ke3 Fluent More volume of Huang, Passos,
dryer fitted with drying chamber is Kumar, and
rotary atomizer. used by rotary Mujumdar (2005)
atomizer and
existence of strong
reverse flow just
beneath the rotating
disc.
Simulation of a spray 3D RNG ke3 Fluent Simulation results Huang et al. (2006)
dryer with pressure agreed well with
nozzle and rotary Kieviet (1997)
atomizer. Kieviet’s experimental results.
(1997) spray dryer
geometry was used.
Simulation of a spray 3D RANS Fluent Rotary atomizer has Ullum (2006)
dryer with rotary a big influence on
atomizer the flow pattern in
pilot scale spray
dryer, but its
influence decreases
with increase in size
of spray dryer.
Simulation of 3D Standard ke3 Fluent Good agreement Huang and
industrial scale spray with experimental Mujumdar (2007)
dryer with a new data. Off-design
drying kinetics performance of spray
model for a heat dryer was predicted
sensitive solution. to analyse the effect
of various operating
parameters on drying
performance
Evaluation of droplet 3D RNG ke3 Fluent The concept of Woo et al. (2008a)
drying models in particle rigidity
a spray dryer fitted prediction in a CFD
with rotary atomizer simulation was
using CFD explored and the
simulation effect of initial feed
moisture content on
the drying models
was also studied.
 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398 389

Table 1 (continued )
Problem Model geometry Turbulence model Software Findings Authors
descriptions
Modelling droplet 2D and 3D Standard ke3 Fluent 2D models can be Mezhericher
drying in a spray used for fast and et al. (2009)
dryer fitted with low-resource-
a pressure nozzle consumption
under steady and numerical
unsteady state. calculations with
some drawbacks. 3D
models can predict
the asymmetric flow
patterns and provide
actual 3D picture of
particle trajectories,
but require high
computing effort.
Simulation of 2D Standard ke3 Fluent Smaller spray cone Chen and Jin
industrial scale spray angle facilitates easy (2009a)
dryer attached with movement of
a Fluidized bed, particles to the
using Reaction fluidized bed. The
Engineering accuracy of REA
Approach (REA). model in predicting
the single droplet
drying kinetics was
also explained.

An ultrasonic nozzle spray dryer was studied numerically between the gas and the droplets. The phenomena of
by Huang, Kumar, and Mujumdar (2004b). Birchal, mass transfer from droplet to gas were coupling by evapo-
Huang, Mujumdar, and Passos (2006) simulated a spray ration, momentum exchange via drag and energy coupling
dryer fitted with a rotary atomizer for drying of milk emul- by heat transfer. Heat is transferred from the gas phase to
sion by using CFD and also by a model with simplified par- the droplets convectively and this leads to a decrease in
ticle motion. Authors also discussed the advantages and temperature of the gas this affects the viscosity and density
limitations of each model in the design and optimisation of the gas, which in turn may affect the gas flow field. This
of spray dryers. Effects of atomizers types (rotary disc then affects the droplet trajectories and the heat transfer rate
and pressure nozzle) on droplet behaviour was performed between the droplets and the gas (Crowe, Sharam, & Stock,
using CFD by Huang, Kumar, and Mujumdar (2006) for 1977). Hence, all three equations (mass, momentum and
spray drying of maltodextrin. They concluded that pressure energy) are interdependent and should be included in the
nozzle may lead to a high velocity variation in the center of gasedroplets interactions.
the chamber than the rotary atomizer. Moreover, large recir- Crowe et al. (1977) first proposed the particle source in
culation of droplets was also found during pressure nozzle cell (PSI-Cell) model. This is the basis for the discrete
atomisation. phase model (DPM). The EulerianeLagrangian frame
(DPM) provides better residence times of individual parti-
Particle histories cles with a large range of particle sizes. In the DPM, the
The understanding of particle histories such as, velocity, flow field is divided into a grid defining computational cells
temperature, residence time and the particle impact position around each grid point. Each computational cell is treated
are important to design and operating spray drying. More- as a control volume for the continuous phase (gas phase).
over, final product quality is depending on these particle The droplets are treated as sources of mass, momentum
histories. These particle histories can be tracked with the and energy inside the each control volume. The gas phase
help of CFD simulations. Many of the early models were is regarded as a continuum (Eulerian approach) and is de-
developed for gas temperature and velocity profiles but, scribed by first solving the gas flow field assuming no drop-
very few articles have been reported on particles histories. lets are present. Using this continuous phase flow field,
The importance of the particle histories and findings are droplet trajectories together with size and temperature his-
discussed in the following sub sections. tories along the trajectories are calculated. The mass,
momentum and energy source terms for each cell through-
Aireparticle interaction out the flow field is then determined. The source terms are
The primary problem in spray-drying modelling is the evaluated from the droplet equation and are integrated over
coupling of equations in mass, momentum and energy the time required to cross the length of the trajectory inside
390 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398

each control volume. The results are multiplied (scaled up) the effect of air inlet geometry and spray cone angle on
by the number flow rate of drops associated with this trajec- the deposition rate. They studied the particle impact posi-
tory (Crowe et al., 1977; Papadakis & King, 1988). tion and concluded that proper selection of chamber geom-
The gas flow field is solved again, incorporating these etry and cone angle will help in reducing the depositional
source terms and then new droplet trajectories and temper- problems and optimizing the performance of the system.
ature histories are calculated. This approach provides the In the CFD simulation combined Eulerian and Lagrang-
influence of the droplets on the gas velocity and tempera- ian model is used to obtain particle trajectories by solving
ture fields. The method proceeds iteratively calculating the force balance equation:
gas and particle velocity fields.

The range of droplet sizes produced by the atomizer is d u p 18m CD Re  rp  r g
¼ v  up þ g ð4Þ
represented by a number of discrete droplet sizes. Each ini- dt rp dp 24 rp
tial droplet size is associated with one trajectory, along
which the number flow rate of drops is constant, assuming where, v is the fluid phase velocity, up is the particle veloc-
that no coalescence or shattering occurs. Once the air ve- ity, rg is the density of the fluid and rp is the density of the
locities, temperatures, and humidity’s are postulated, the particle.
transport equations for the droplets of each size are inte- The particle force balance (equation of motion) includes
grated over time and position to yield droplet trajectories, discrete phase inertia, aerodynamic drag and gravity. The
velocities, sizes and temperatures. Calculations for droplets slip Reynolds number (Re) and drag coefficient (CD) are
of each initial size continue until the volatile fractions (e.g. given in the following equations.
water) in the droplets evaporate completely, exit the col-  
umn, or impact the column wall (Papadakis & King, 1988). rg dp  u p  v 
Re ¼ ð5Þ
The PSI-Cell model (Crowe et al., 1977) was used by m
Papadakis and King (1988) and they found good agreement
between the model and experimental results in a concurrent a2 a3
CD ¼ a 1 þ þ ð6Þ
spray dryer. O’Rourke and Wadt (1982) and Goldberg Re Re2
(1987) had further developed this model. The effect of swirl
on increase in residence time due to entrainment of parti- where, dp is the particle diameter, and a1, a2 and a3 are con-
cles and formation of central recirculation zone was ob- stants that apply to smooth spherical particles over several
served by Gupta, Lilley, and Syred (1984). However, all ranges of Reynolds number (Re) given by Morsi and
these studies are mostly simplified numerical models. Alexander (1972).
Oakley, Bahu, and Reay (1990) simulated the air flow in The velocity of particles relative to air velocity was used
a co-current cylinder-on-cone spray dryer using PSI-Cell in the trajectory calculations (equation (4)). Turbulent par-
model. The model’s predictions were compared to experi- ticle dispersion was included in this model as discrete eddy
mental measurements of the air flow using Laser Doppler concept (Langrish & Zbicinski, 1994). In this approach, the
Anemometry. Although that study did not include the influ- turbulent air flow pattern is assumed to be made up of a col-
ence of the spray on the air flow field, but both measure- lection of randomly directed eddies, each with its own life-
ments and simulations showed the presence of periodic time and size. Particles are injected into the flow domain at
oscillations in the size of the recirculation zones inside the nozzle point and envisaged to pass through these ran-
the drying chamber. The study also found that predictions dom eddies until they impact the wall or leave the flow do-
of the air flow patterns were sensitive to the values of tur- main through the product outlet.
bulence parameters (selected at the annular air inlet) of The heat and mass transfer between the particles and the
the standard ke3 turbulence model. Latter, Oakley and hot gas is derived following the motion of the particles.
Bahu (1993) validated the PSI-Cell model with experimen-   dmp
dTp
tal measurement. They also performed various works to as- mp cp ¼ hAp Tg  Tp þ hfg ð7Þ
dt dt
sess appropriate turbulence models and to test the validity
of the results against experimental data and concluded where, mp is the mass of the particle, cp is the particle heat
that CFD can be used for the determination of gas flow pat- capacity, Tp is the particle temperature, hfg is the latent heat,
tern, spray/gas mixing, particle temperature and moisture Ap is the surface area of the particle and h is the heat trans-
histories. Langrish, Oakley, Keey, Bahu, and Hutchinson fer coefficient
(1993) analysed low frequency oscillations in the flow field The heat transfer coefficient (h) is obtained from the
inside the drying chamber and their results are in agreement RanzeMarshall equation.
with the Oakley et al. (1990) observations. They found that
the ability to predict the onset of vortex breakdown is of hdp
Nu ¼ ¼ 2 þ 0:6ðRed Þ ðPrÞ ð8Þ
1=2 1=3
significant use in assessing the impact of the precessing kta
vortex core upon wall deposition, drying efficiency and
product quality. Langrish and Zbicinski (1994) studied where, Prandtl number (Pr) is defined as follows
 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398 391

cp m using different drying rate and turbulence models. They

Pr ¼ ð9Þ
kta found that the stochastic effect of turbulence flow has sig-
nificant effect on the particle path during the drying period.
where, dp is the particle diameter, kta is the thermal conduc- Anandharamakrishnan, Gimbun, Stapley, and Rielly (2008)
tivity of the fluid, m is the molecular viscosity of the fluid. studied the particle temperature in both short-form and tall-
The mass transfer rate (for evaporation) between the gas form spray dryer using CFD simulation for drying of whey
and the particles is calculated from the following equation. proteins. They found that due moisture evaporation of drop-
  lets, the temperature of droplets are high and is almost
dmp
¼ kc Ap Ys  Yg ð10Þ equal to the gas temperatures outside the core region. More-
dt over, the temperature of gas in the core spray region and the
where, Ys* is the saturation humidity, Yg is the gas humidity upper part of the chamber decreases due to the cooling ef-
and kc is the mass transfer coefficient and it can be obtained fects of evaporation. The particle nature was also affected
from Sherwood number by the outlet air temperature.

k c dp Particle residence time distribution (PRTD)

Sh ¼ ¼ 2 þ 0:6ðRed Þ ðScÞ ð11Þ
1=2 1=3
Di;m The particle residence time have a great impact on the
final powder quality and it also affect the product qualities
where, Di,m is the diffusion coefficient of water vapour in like solubility and bulk density. The residence time (RT) is
the gas phase and Sc is the Schmidt number, defined as divided into two parts namely, primary and secondary res-
follows idence times. The primary RT is calculated from the time
m taken for droplets leaving the nozzle to impact on the
Sc ¼ ð12Þ wall or leave at the outlet. The secondary residence time
rg Di;m
can be defined as the time taken for a particle to slide along
The values of vapour pressure, density, specific heat and the wall from the impact position to the exit.
diffusion coefficients can be obtained from various sources One of the earliest works to determine residence time dis-
like Perry and Chilton (1984), Incropera, Dewitt, Bergman, tribution directly in a co-current spray dryer was performed
and Lavine (2007), etc. by Pham and Keey (1977). Latter, Kieviet and Kerkhof
When the temperatures of the droplet has reached the (1995) determined the RTD of particles in a co-current spray
boiling point and while the mass of the droplet exceeds dryer during the drying of aqueous maltodextrin solutions.
the non-volatile fraction. The boiling rate model was ap- Kieviet (1997) observed that during spray drying of malto-
plied (Kuo, 1986) dextrin solution, the larger diameter particles have longer
 
  RT’s than smaller particles. He also found enormous differ-
d dp 4kta pffiffiffiffiffiffi c g T g  Tp ence between measured and predicted results due to particle
¼ 1 þ 0:23 Re ln 1 þ ð13Þ
dt rp c g dp hfg wall depositions and sliding movement. Ducept, Sionneau,
and Vasseur (2002) performed an experiment to determine
where, kta is the thermal conductivity of the gas and cg is the RTD of particles and validated with the CFD predictions
the heat capacity of the gas. in a superheated steam spray dryer. Authors concluded that
CFD is a valuable tool for studying particle size distribution
Particle temperature effects and flow rates. The residence time distribution of dif-
The particle temperature is very important in case heat ferent sized particles in a spray dryer was studied by Huang,
sensitive products. Aroma loss and thermal degradation of Kumar, and Mujumdar (2003a) and they found that different
heat sensitive materials is affected by the residence time droplets follow different trajectories in the drying chamber.
and the temperature of particles. Crowe et al. (1977) pre- Anandharamakrishnan et al. (2008) studied PRTD of whey
dicted that the smaller size particles have higher tempera- proteins in both short- and tall-form dryers and the residence
tures than the larger particles, because, the latter have time (Fig. 2) indicates most of the particles have very low RT
a smaller surface area to volume ratio and evaporate during spray drying (short form). It was observed that a bent
more slowly. Kieviet (1997) studied the air flow pattern, outlet pipe inside the chamber increases the gas and particle
temperature, humidity, particle trajectories and residence recirculation (Fig. 2); consequently, cold gas is mixed with
time in a 2D co-current spray dryer fitted with a pressure down-flowing hot inlet gas and dried particles will be ex-
nozzle using maltodextrin as feed solution. Author found posed to the high inlet gas temperatures. This recirculation
good agreement between the measured and modelled may lead to denaturation of proteins or inactivation of en-
data, and they concluded that the gradients in the center re- zymes (Anandharamakrishnan, 2008). Hence, bend outlet
gion of the drying chamber could be improved. The varia- pipe needs to be avoided inside the chamber for producing
tion of particle temperature and gas temperature along the high quality spray dried food products. Moreover, they
particle trajectories during spray drying of skim milk solu- found that large difference between the gas and particle res-
tion (Huang, Kumar, & Mujumdar, 2004c) was studied idence time. However, there is no direct measurement of
392 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398

main outlet aided by mechanical hammer operations.

They also found that in a short-form dryer large fraction
of the particles strike the conical part of spray-dryer cham-
ber, while in tall-form dryer the particle stroked the cylin-
drical part of the wall. In the both form of dryers they
found less impact on the ceiling, despite of the recirculation
of gas in this zone.

Chamber design and flow configuration

The effect of different chamber geometries (cylinder-on-
cone, lantern, hour-glass and pure cone) on the drying per-
formance and the particle residence time was studied by
Huang et al. (2003a). They suggested that it is possible to
change the chamber geometry for better utilization of the
dryer volume. Similarly, a parametric study of gas flow pat-
terns and drying performance was investigated by the same
research group (Huang, Kumar, & Mujumdar, 2003b). The
use of CFD to obtain uniform flow distribution in a plenum
chamber with a single, off-axis, inlet pipe was studied by
Southwell, Langrish, and Fletcher (2001). The flow distri-
bution was investigated and the result was confirmed by ex-
Fig. 2. Particle trajectories coloured by residence time (s)
perimental data for selected configurations. They concluded
(Anandharamakrishnan et al., 2008). that CFD can be used for several design alternatives for
overcoming poor flow distribution.
Apart from the conventional vertical design of the spray
primary RT available to confirm the predictions and this is an dryers the use of horizontal design is gaining importance
interesting challenge for future research. nowadays. Huang and Mujumdar (2006) investigated a sin-
gle-stage and two-stage horizontal spray dryer using CFD.
Particle impact positions They found that use of good chamber design and incorpo-
The knowledge of particle impact positions is important ration of fluidized bed will improve the drying perfor-
for the design and operation of spray dryers as it influences mance. In spray drying, the majority of works have been
the final product quality. In an earlier numerical study performed on co-current flow. The modelling of air flow
(Reay, 1988) has shown that most likely areas for wall de- patterns in a counter-current spray dryers (Harvie,
position are an annular area of the dryer roof and a region Langrish, & Fletcher, 2001), provided light into this area,
below the atomizer where large particles are likely to de- but has been lacked by experimental data for validation.
posit. Later, Kieviet (1997) investigated the interaction of However, Bayly, Jukes, Groombridge, and McNally
wall deposition with the residence time and the effect of (2004) has both measured and modelled air flow patterns
wall deposition on the product quality and yield during in a counter-current spray dryer using Reynolds Stress
spray drying of maltodextrin. Goula and Adamopoulos Model for the gas turbulence and they found good agree-
(2004) determined how operating conditions can influence ment with the measured results.
the fouling and residue accumulation of the equipment dur-
ing the drying process. Recently, Anandharamakrishnan Recent developments
et al. (2008) studied the particle impact position during dry- In the recent years, the application of Reaction Engineering
ing of whey proteins from the simulation data using in- Approach (REA), drying kinetics model, dropletedroplet in-
house post-processor. Fig. 3 aeb shows the top and front teractions, unsteady state modelling and population balance
cross-sectional views of the simulated results model for the simulation of spray dryers is increasing. In the
(Anandharamakrishnan et al., 2008). These figures indi- Reaction Engineering Approach, it assumes that evaporation
cates that a large fraction of the particles (50%) strike the is an activation process to overcome an energy barrier, while
conical part of the spray-dryer chamber (similar with the it is not the case for condensation or adsorption. REA is
earlier observation of Langrish & Zbicinski, 1994) and a two-way model as compared to the one-way characteristic
23% of particles hit the cylindrical part of the wall, but drying rate curve (CDRC) for predicting the single droplet dry-
only a small proportion (25%) of the particles come out ing. The basic of REA was described by Chen, Pirini, and
of the outlet pipe line (the intended destination). A very Ozilgen (2001) and Chen and Xie (1997). This method de-
small 2% of particles hit the ceiling despite the large vol- scribes the droplet drying trend, giving a detailed account of
ume of re-circulated gas, but particles hitting the cone the temperature changes that occurring within the droplet dur-
and/or cylindrical wall (73%) should slide down to the ing the drying period and it requires some experimental data to
 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398 393

Fig. 3. Particle impact positions (a) top view (b) front view (Anandharamakrishnan et al., 2008).

determine the model parameters. The REA model was used by (CDRC and REA) were also discussed and the recent appli-
Chen and Xie (1997) for the simulation of drying of thin layer cation of the latter model in the area of milk production for
food materials such as kiwifruit, silica gel, potato and apple sli- scale-up from lab to industrial scale was also noted.
ces. Moreover, Huang et al. (2004a) found that this approach Further, this REA based drying kinetics model can be in-
(REA) fits in well with the fluent commercial CFD code for corporated into the CFD programs in future years to deter-
spray drying. mine the wall deposition pattern, droplet residence time and
Chen and Lin (2004) compared the REA and CDRC for droplet trajectories during spray-drying operation.
drying of skim milk and whole milk droplets. The same Apart from the boiling-point evaporation model few works
modelling procedure was used by Chen and Lin (2005) to have been performed in determination of the drying kinetics
measure the process of single milk droplet drying under el- during spray drying. The experimental determination of
evated humidity conditions by REA. They found that the spray-drying kinetics was performed by Zbicinski,
REA method gave better predictions of the droplet drying Strumillo, and Delag (2002). They determined the spray-dry-
than the CDRC method. A REA model for drying of aque- ing kinetics as a function of atomisation ratio and drying agent
ous lactose droplet was developed by Lin and Chen (2006) temperature. Zbicinski, Piatkowski, and Prajs (2005) contin-
using the experimental data obtained by the improved glass ued the study and proved that based on the critical moisture
filament method. The REA method was used for modelling content of the material, the spray-drying kinetics could be de-
whey protein concentrate droplet drying (Chen & Patel, termined from the generalized drying curves. This lab scale
2008; Lin & Chen, 2007). They found that the model details can be used for scaling up of spray-drying process. Fur-
showed reasonably good agreement with the experimental ther, Woo, Daud, Talib, and Tasirin (2008b) analysed the effect
data. They concluded that the inlet air temperature, feed of wall surface properties on the deposition problem during
temperature and droplet size had a significant impact on spray drying using different drying kinetics. They concluded
the drying rate and the overall product quality. The drying that proper selection of dryer wall material will provide poten-
kinetics model based on REA is also found to be influenced tial alternatives for reducing the deposition problem. More re-
by the above parameters. A work was carried out to evalu- cently, Roustapour, Hosseinalipour, Ghobadian, Mohaghegh,
ate the different drying models, namely REA, CDRC linear and Azad (2009) performed a CFD study for the drying of
and CDRC modified in a spray dryer fitted with rotary disc lime juice. They determined the drying kinetics based on ex-
atomizer by Woo et al. (2008a). Authors found that the perimental results of moisture content variation along the
REA model can be used for prediction of particle surface length of chamber and numerically estimated residence time
moisture of wet particles with better results. Chen (2008) of droplets. Authors found that an increase in initial droplet di-
has reviewed the application of the REA model for ameter resulted in a lower particle residence time. A compar-
drying of thin layer materials and for solid droplets. The ison of two drying kinetics models namely, characteristic
similarities and differences between the two drying models drying curve (CDC) and REA was performed by Woo et al.,
394 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398

(2008c) and they found that the different responses of the REA solution. Moreover, this approach may give opportunity
and CDC to different initial feed moisture conditions due to the to perform the simulation of wet shells and puffing phe-
respective formulation of the driving forces in the two models. nomena during the period of drying for a better understand-
Further, the authors extended REA model to simulate the par- ing of the drying process.
ticle surface moisture.
More recently, dropletedroplet interactions during the Scope for future research
spray drying was performed by applying the transient Still there is a scope for future research in this area for
mode of calculations (Mezhericher, Levy, & Borde, optimisation of spray-drying process. The modelling of par-
2008). The droplet collisions influenced the temperature ticle interactions (by four-way coupling between
and humidity patterns while their effect on velocity was gaseparticle and particleeparticle), particle coalescence
less marked. They investigated both insulated and noninsu- and break up is a great challenge for the modellers. The
lated spray chambers and reported that the insulation of modelling of turbulence using the Detached Eddy Simula-
spray chamber will affect the air flow patterns and thereby tion (DES) is also gaining importance. The success of these
affecting the droplet trajectories. The probabilistic ap- models depends on the experimental validations. The re-
proach developed by O’Rourke (1981) was used by them duction of wall deposition (including the product nature)
for determining the dropletedroplet interactions. and thermal degradation by modifying the air inlet geome-
The unsteady state simulation of spray drying is new try and spray pattern also offers scope for future research.
area of research. Mezhericher, Levy, and Borde (2009) The prediction of the particle residence time (including
modelled droplet drying in spray-drying chamber by apply- sliding movement of particles in the secondary RT) is cur-
ing the unsteady mode of calculations. They performed rently lacking. Hence, there is scope for future research in
both 2D and 3D analysis and concluded that the 3D model the area to overcome problems like agglomeration, wall de-
predicts asymmetry of flow patterns in the spray chamber. position, particle residence time, thermal degradation of
However, 2D axi-symmetric model can be used for fast particles and aroma loss. Recently, Langrish (2007) also
and low-resource-consumption numerical calculations. observed the same. Thus modelling approach may leads
Similarly, Woo et al. (2008a) also observed the same in to the better productivity and high quality food products.
their unsteady state simulation. Chen and Jin (2009b) per-
formed transient 3D simulations in an industrial scale spray Conclusion
dryer (15 m tall and 10 m wide). They observed that the The objective of this review is to emphasis the impor-
particles make the central jet oscillate more non-linearly tance of CFD in the modelling of the spray dryers. There
and the frequency of oscillation decreases with the increas- is considerable growth in the development and application
ing feed rate. Recently, Woo et al. (2009) have performed of CFD in the area of spray-drying to predict the complex
unsteady state simulations of spray drying and investigated flow patterns. The EulerianeLagrangian model is suitable
the effect of chamber aspect ratio and operating conditions for the spray-drying operations as it has the advantage of
on flow stability. Authors observed that a large expansion being computationally cheaper than the EulerianeEulerian
ratio produces a more stable flow due to the limitations method for a large range of particle sizes. The particle his-
of the jet fluctuations by outer geometry constriction. tories prediction will give an idea regarding the drying be-
The modelling of spray dryers using population balance haviour based on particle temperature, the residence time
method is gaining importance nowadays because this model based on trajectories and the final product recovery based
accounts for the droplet growth, coalescence and break up on impact positions. The REA based drying kinetics model
during the spray-drying process. Nijdam, Guo, Fletcher, can be used to determine the temperature change within the
and Langrish (2004) modelled the particle agglomeration droplet during drying. However, more works needs to be
within the spray chamber using two different frameworks performed with population balance modelling. Moreover,
namely, Lagrangian and Eulerian. They validated their pre- some of the problems during spray drying of food products
diction using phase doppler anemometry (PDA) measure- have been highlighted. Now, with the increasing computing
ment and they found that Lagrangian approach is more power CFD will be a valuable tool for the food industry in
suitable for modelling of agglomeration of particles in answering complex fluid flow, heat and mass transfer phe-
terms of ease of implementation and range of applicability. nomena and aid in better design and process control of
However, authors concluded that more work needs to be spray dryers.
performed to refine the turbulence models for the Lagrang-
ian approach in order to account for the various particle tur- Acknowledgements
bulence phenomena and particleeparticle correlations. Authors wish to thank Dr. V. Prakash, Director, CFTRI for
Recently, Handscomb, Kraft, and Bayly (2009) included the support and encouragement in CFD simulation studies.
source term for droplet coalescence and break up. This ap- Anandharamakrishnan wishes to acknowledge Prof. C.D.
proach gives a better understanding of complexities of two- Rielly, Professor and Head, and Dr. Andy Stapley, Senior
phase flows in spray dryers. The model includes nucleation Lecturer, Chemical Engineering Department, Loughborough
and growth of suspended solids from an ideal binary University, UK for their help.
 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398 395

References Chen, X. D., Pirini, W., & Ozilgen, M. (2001). The reaction engi-
neering approach to modeling drying of thin layer of pulped kiwi
Abdul Ghani, A. G., Farid, M. M., Chen, X. D., & Richards, P. (1999a). fruit flesh under conditions of small biot numbers. Chemical
Numerical simulation of natural convection heating of canned Engineering and Progress, 40, 165e181.
food by computational fluid dynamics. Journal of Food Engineer- Chen, X. D., & Xie, G. Z. (1997). Fingerprints of the drying behavior of
ing, 41(1), 55e64. particulate or thin layer food materials established using a reaction
Abdul Ghani, A. G., Farid, M. M., Chen, X. D., & Richards, P. (1999b). engineering model. Transactions of the Institution of Chemical
An investigation of deactivation of bacteria in a canned liquid food Engineers, Part C, 75, 213e222.
during sterilisation using computational fluid dynamics (CFD). Cortella, G., Manzan, M., & Comini, G. (1998). Computation of air
Journal of Food Engineering, 42(4), 207e214. velocity and temperature distributions in open display cabinets. In:
Abdul Ghani, A. G., Farid, M. M., Chen, X. D., & Richards, P. (2001). Advances in the refrigeration systems, food technologies and cold
Thermal sterilisation of canned food in a 3-D pouch using computa- chain (pp. 617e625). Paris, France: International Institute of
tional fluid dynamics. Journal of Food Engineering, 48(2), 147e156. Refrigeration.
Anandharamakrishnan, C. (2003). Computational fluid dynamics Courant, R., Friedrichs, K., & Lewy, H. (1928). Die partiellen differ-
(CFD) e applications for the food industry. Indian Food Industry, enzengleichungen der mathematischen Physik. Mathematische
22(6), 62e68. Annalen (Historical Archive), 100, 32e74.
Anandharamakrishnan, C. (2008). Experimental and computational Crowe, C. T., Sharam, M. P., & Stock, D. E. (1977). The particle source
fluid dynamics studies on spray-freeze-drying and spray-drying of in cell (PSI-Cell) model for gas-droplet flows. Journal of Fluid En-
proteins. Ph.D. thesis. UK: Loughborough University. gineering, 9, 325e332.
Anandharamakrishnan, C., Gimbun, J., Stapley, A. G. F., & Rielly, C. D. Datta, A. K., & Teixeira, A. A. (1987). Numerical modelling of natural
(2008). A study of particle histories during spray drying using com- convection heating in canned liquid foods. Transactions of the
putational fluid dynamic simulations. In 16th International Drying ASAE, 30(5), 1542e1551.
Symposium (IDS 2008), Hyderabad, India 9e12 November 2008. Davey, L. M., & Pham, Q. T. (1997). Predicting the dynamic product
Anandharamakrishnan, C., Gimbun, J., Stapley, A. G. F., & Rielly, C. D. heat load and weight loss during beef chilling using a multi-region
(2010). Application of computational fluid dynamic (CFD) simu- finite difference approach. International Journal of Refrigeration,
lations to spray-freezing operations. Drying Technology, 28, 20(7), 470e482.
94e102. Davey, L. M., & Pham, Q. T. (2000). A multi-layered two-dimensional
Anandharamakrishnan, C., Rielly, C. D., & Stapley, A. G. F. (2007). finite element model to calculate dynamic product heat load and
Effects of process variables on the denaturation of whey proteins weight loss during beef chilling. International Journal of Refriger-
during spray-drying. Drying Technology, 25, 799e807. ation, 23(6), 444e456.
Anderson, J. D. (1984). Computational fluid dynamics e The basics Denys, S., Dewettinck, K., & Pieters, J. G. (2005). CFD analysis for
with applications. New York: McGraw-Hill Inc. process impact assessment during thermal pasteurisation of intact
Bahu, R. E. (1992). Spray drying d maturity or opportunities. In eggs. Journal of Food Protection, 68, 366e374.
A. S. Mujumdar (Ed.), Drying ’92. Proc. eighth international drying Denys, S., Pieters, J. G., & Dewettinck, K. (2003). Combined CFD and
symposium IDS ’92, Montreal, Canada (pp. 74e91). Amsterdam: experimental approach for determination of the surface heat
Elsevier. transfer coefficient during thermal processing of eggs. Journal of
Bakker, A. (2002). Computational fluid mixing. Lebanon, NH, USA: Food Science, 68, 943e951.
Fluent Inc.. Denys, S., Pieters, J. G., & Dewettinck, K. (2004). Computational
Bayly, A. E., Jukes, P., Groombridge, M., & McNally, C. (2004). Airflow fluid dynamics analysis of combined conductive and convective
patterns in a counter-current spray drying tower e simulation and heat transfer in model eggs. Journal of Food Engineering, 63,
measurement. In:Proceedings of the 14th international drying 281e290.
symposium (IDS 2004), São Paulo, Brazil, 22e25 August 2004, Vol Dombrowski, N., Foumeny, E. A., & Riza, A. (1993). Know the CFD
B (pp. 775e778). codes. Chemical Engineering Progress, 89(9), 46e48.
Birchal, V. S., Huang, L., Mujumdar, A. S., & Passos, M. L. (2006). Spray Ducept, F., Sionneau, M., & Vasseur, J. (2002). Superheated steam
dryers e modeling and simulation. Drying Technology, 24, 359e371. dryer - simulations and experiments on product drying. Chemical
Bird, B., Stewart, W. E., & Lightfoot, E. N. (1960). Transport phe- Engineering Journal, 86, 75e83.
nomena. New York: John Wiley & Sons. Fellows, P. J. (1998). Food processing technology-principles and
Charm, S. E. (1971). The fundamentals of food engineering. Westport, practice. Cambridge: Woodhead Publishing Limited.
CT: The AVI Publishing Company. Fletcher, A. J. (2000). Computational techniques for fluid dynamics,
Chen, X. D. (2008). The basics of a reaction engineering approach to (2nd ed.). New York: Springer-Verlag.
modeling air-drying of small droplets or thin-layer materials. Fletcher, D., Guo, B., Harvie, D., Langrish, T., Nijdam, J., &
Drying Technology, 26, 627e639. Williams, J. (2003). What is important in the simulation of spray
Chen, X. D., & Jin, Y. (2009a). Numerical study of the drying process of dryer performance and how do current CFD models perform. In:
different sized particles in an industrial e scale spray dryer. Drying 3rd International Conference on CFD in the minerals and pro-
Technology, 27, 371e381. cess industries. Melbourne, Australia: CSIRO. 10e12 December
Chen, X. D., & Jin, Y. (2009b). A three-dimensional numerical study of 2004.
the gas/particle interactions in an industrial-scale spray dryer for Foster, A. (1996). The costs involved in modelling retail display cabi-
milk powder production. Drying Technology, 27, 1018e1027. nets. In: Paper presented at the Seminar on Achievements with
Chen, X. D., & Lin, S. X. Q. (2004). Air drying of milk droplet under CFD in the food industry. Leatherhead, UK: Leatherhead Food
constant and time dependent conditions. AIChE Journal, 51(6), Research Association.
1790e1799. Goldberg, J. E. (1987). Prediction of spray dryer performance. Ph.D.
Chen, X. D., & Lin, S. X. Q. (2005). Prediction of air-drying of milk thesis. Oxford, UK: University of Oxford.
droplet under relatively high humidity using the reaction engi- Goula, A. M., & Adamopoulos, K. G. (2004). Influence of spray drying
neering approach. Drying Technology, 23, 1395e1406. conditions on residue accumulation e simulation using CFD.
Chen, X. D., & Patel, K. C. (2008). Sensitivity analysis of the reaction Drying Technology, 22, 1107e1128.
engineering approach to modeling spray drying of whey proteins Gupta, A. K., Lilley, D. G., & Syred, N. (1984). Swirl flows. Turnbridge
concentrate. Drying Technology, 26, 1334e1343. Wells: Abacus.
396 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398

Handscomb, C. S., Kraft, M., & Bayly, A. E. (2009). A new model for Langrish, T. A. G. (2007). New engineered particles from spray dryers:
the drying of droplets containing suspended solids. Chemical En- research needs in spray drying. Drying Technology, 25, 981e993.
gineering Science, 64, 628e637. Langrish, T. A. G. (2009). Multi-scale mathematical modelling of spray
Harvie, D. J. E., Langrish, T. A. G., & Fletcher, D. F. (2001). Numerical dryers. Journal of Food Engineering, 93, 218e228.
simulations of gas flow patterns within a tall-form spray dryer. Langrish, T. A. G., & Fletcher, D. F. (2001). Spray drying of food in-
Transactions of the Institution of Chemical Engineers, 79(A), gredients and applications of CFD in spray drying. Chemical
235e248. Engineering and Processing, 40, 345e354.
Hu, Z., & Sun, D. W. (1999). The temperature distribution of cooked Langrish, T. A. G., & Fletcher, D. F. (2003). Prospects for modelling and
meat joints in an air-blast chiller during cooling process: CFD design of spray dryers in the 21st century. Drying Technology, 21,
simulation and experimental verification. In Paper presented at the 197e215.
20th International Congress of Refrigeration, Sydney, Australia. Langrish, T. A. G., Oakley, D. E., Keey, R. B., Bahu, R. E., &
Hu, Z., & Sun, D. W. (2000). Simulation of heat and mass transfer for Hutchinson, C. A. (1993). Time-dependent flow patterns in spray
vacuum cooling of cooked meats by using computational fluid dy- dryers. Transactions of the Institution of Chemical Engineers, Part
namics code. In Paper presented at the 8th International Congress on A, 71, 355e360.
Engineering and Food, Paper no. O-130, Puebla, Mexico. Langrish, T. A. G., Williams, J., & Fletcher, D. F. (2004). Simulation of
Huang, L. X., Kumar, K., & Mujumdar, A. S. (2003a). Use of compu- the effects of swirl on gas flow patterns in a pilot-scale spray dryer.
tational fluid dynamics to evaluate alternative spray chamber Transactions of the Institution of Chemical Engineers, Part A, 82
configurations. Drying Technology, 21, 385e412. (A7), 821e833.
Huang, L. X., Kumar, K., & Mujumdar, A. S. (2003b). A parametric Langrish, T. A. G., & Zbicinski, I. (1994). The effects of air inlet ge-
study of the gas flow patterns and drying performance of co-current ometry and spray cone angle on the wall deposition rates in spray
spray dryer: results of a computational fluid dynamics study. Drying dryer. Transactions of the Institution of Chemical Engineers, Part A,
Technology, 21(6), 957e978. 72, 420e430.
Huang, L. X., Kumar, K., & Mujumdar, A. S. (2004a). Simulation of Launder, B. E., Reece, G. J., & Rodi, W. (1975). Progress in the de-
a spray dryer fitted with a rotary disk atomizer using a three-di- velopment of a reynoldsstress turbulence closure. Journal of Fluid
mensional computational fluid dynamic model. Drying Tech- Mechanics, 68, 537e566.
nology, 22(6), 1489e1515. Launder, B. E., & Spalding, D. B. (1972). Lectures in mathematical
Huang, L. X., Kumar, K., & Mujumdar, A. S. (2004b). Simulation of models of turbulence. London, UK: Academic Press.
spray evaporation using pressure and ultrasonic atomizer e Launder, B. E., & Spalding, D. B. (1974). The numerical computation of
a comparative analysis. Russia Tambo State Technical University, turbulent flows. Computer Methods in Applied Mechanics and
Transactions, (English Version), 10(1A), 83e100. Engineering, 3, 269e289.
Huang, L. X., Kumar, K., & Mujumdar, A. S. (2004c). Computational Li, X., & Zbicinski, I. (2005). A sensitivity study on CFD modeling of co-
fluid dynamic simulation of droplet drying in a spray dryer. In:- current spray-drying process. Drying Technology, 23(8), 1681e1691.
Proceedings of the 14th international drying symposium (IDS Lin, S. X. Q., & Chen, X. D. (2006). A model for drying of an aqueous
2004), São Paulo, Brazil, 22e25 August 2004, Vol A lactose droplet using the reaction engineering approach. Drying
(pp. 326e332). Technology, 24, 1329e1334.
Huang, L. X., Kumar, K., & Mujumdar, A. S. (2006). A comparative Lin, S. X. Q., & Chen, X. D. (2007). The reaction engineering approach
study of a spray dryer with rotary disc atomizer and pressure nozzle to modelling the cream and whey protein concentrate droplet
using computational fluid dynamic simulations. Chemical drying. Chemical Engineering and Processing, 46, 437e443.
Engineering and Processing, 45, 461e470. Livesley, D. M., Oakley, D. E., Gillespie, R. F., Elhaus, B.,
Huang, L. X., & Mujumdar, A. S. (2006). Numerical study of two-stage Ranpuria, C. K., Taylor, T., et al. (1992). Development and vali-
horizontal spray dryers using computational fluid dynamics. dation of a computational model for spray-gas mixing in spray
Drying Technology, 24, 727e733. dryers. In A. S. Mujumdar (Ed.), Proc. eighth international drying
Huang, L. X., & Mujumdar, A. S. (2007). Simulation of an industrial symposium IDS ’92, Montreal, Canada, drying ’92 (pp. 407e416).
spray dryer and prediction of off-design performance. Drying Amsterdam: Elsevier.
Technology, 25, 703e714. Mariotti, M., Rech, G., & Romagnoni, P. (1995). Numerical study of air
Huang, L. X., Passos, M. L., Kumar, K., & Mujumdar, A. S. (2005). A distribution in a refrigerated room. In:Proceedings of international
three dimensional simulation of a spray dryer fitted with a rotary congress of refrigeration, Vol 2 (pp. 98e105), Den Hague, The
atomizer. Drying Technology, 23, 1859e1873. Netherlands.
Incropera, F. P., Dewitt, D. P., Bergman, T. L., & Lavine, A. S. (2007). Masters, K. (1991). Spray drying. Essex: Longman Scientific & Tech-
Fundamentals of heat and mass transfer. New York: John Wiley & nical/John Wiley & Sons Inc.
Sons. Masters, K. (1994). Scale-up of spray dryers. Drying Technology, 12
Jakobsen, H. A., Sannaes, B. H., Grevskott, S., & Svendsen, H. F. (1e2), 235e257.
(1997). Modelling of vertical bubble-driven flows. Industrial Mathioulakis, E., Karathanos, V. T., & Belessiotis, V. G. (1998). Simu-
Engineering Chemistry Research, 36, 4052e4074. lation of air movement in a dryer by computational fluid dynamics:
Kerkhof, P. J. A. M. (1994). The role of theoretical and mathematical application for the drying of fruits. Journal of Food Engineering, 36
modelling in scale-up. Drying Technology, 12(1e2), 1e46. (2), 183e200.
Kieviet, F. G. (1997). Modeling quality in spray drying. Ph.D. thesis. Mezhericher, M., Levy, A., & Borde, I. (2008). Dropletedroplet inter-
Netherlands: Endinhoven University of Technology. actions in spray drying by using 2D computational fluid dynamics.
Kieviet, F. G., & Kerkhof, P. J. A. M. (1995). Measurements of particle Drying Technology, 26, 265e282.
residence time distributions in a co-current spray dryer. Drying Mezhericher, M., Levy, A., & Borde, I. (2009). Modeling of droplet
Technology, 13(5e7), 1241e1248. drying in spray chambers using 2D and 3D computational fluid
Kieviet, F. G., & Kerkhof, P. J. A. M. (1996). In: A. S. Mujumdar (Ed.), dynamics. Drying Technology, 27, 359e370.
Using computational fluid dynamics to model product quality in Mills, D. (1998e1999). Development and validation of a preliminary
spray drying: Air flow temperature and humidity patterns. model for optimisation of baking ovens. The Food and Packaging
Drying’96, Vol. A (pp. 259e266), Krakow, Poland. Cooperative Research Centre annual report (1998e1999), Australia.
Kuo, K. K. Y. (1986). Principles of combustion. New York: John Wiley Mirade, P. S., & Daudin, J. D. (2000). A numerical study of the airflow
and Sons. patterns in a sausage dryer. Drying Technology, 18(1e2), 81e97.
 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398 397

Morsi, S. A., & Alexander, A. J. (1972). An investigation of particle Sahu, A. K., Kumar, P., Patwardhan, A. W., & Joshi, J. B. (1999). CFD
trajectories in two-phase flow systems. Journal of Fluid Mechanics, modelling and mixing in stirred tanks. Chemical Engineering Sci-
55(2), 193e208. ence, 54(13e14), 2285e2293.
Mostafa, A. A., & Mongia, H. C. (1987). On the modeling of turbulent Schaldach, G., Berger, L., Razilov, I., & Berndt, H. (2000). Computer
evaporating sprays: Eulerian versus Lagrangian approach. Interna- simulation for fundamental studies and optimisation of ICP spray
tional Journal of Heat and Mass Transfer, 30(12), 2583e2593. chambers. ISAS (Institute of Spectrochemistry and Applied Spec-
Moureh, J., & Derens, E. (2000). Numerical modelling of the temper- troscopy) Current Research Reports, Berlin, Germany.
ature increase in frozen food packaged in pallets in the distribution Scott, G. M. (1977). Simulation of the flow of non-newtonian foods using
chain. International Journal of Refrigeration, 23(7), 540e552. computational fluid dynamics. Campden & Chorleywood Food Re-
Mujumdar, A. S. (1987). Handbook of industrial drying. New York: search Association R&D Report No. 34, Chipping Campden, UK.
Marcel Dekker. Scott, G. M. (1994). Computational fluid dynamics for the food in-
Nijdam, J. J., Guo, B., Fletcher, D. F., & Langrish, T. A. G. (2004). dustry. Food Technology International, Europe, 49e51.
Challenges of simulating droplet coalescence within a spray. Scott, G. M., & Richardson, P. (1997). The application of computa-
Drying Technology, 22(6), 1463e1488. tional fluid dynamics in the food industry. Trends in Food Science
Nijdam, J. J., Guo, B., Fletcher, D. F., & Langrish, T. A. G. (2006). and Technology, 8(4), 119e124.
Lagrangian and Eulerian models for simulating turbulent dispersion Shang, J. S. (2004). Three decades of accomplishments in computational
and coalescence of droplets within a spray. Applied Mathematical fluid dynamics. Progress in Aerospace Sciences, 40, 173e197.
Modelling, 30, 1196e1211. Shih, T. H., Liou, W. W., Shabbir, A., & Zhu, J. (1995). A new ke3 eddy-
Norton, T., & Sun, D. W. (2006). Computational fluid dynamics (CFD) e viscosity model for high Reynolds number turbulent flows e model
an effective and efficient design and analysis tool for the food industry: development and validation. Computers Fluids, 24(3), 227e238.
a review. Trends in Food Science and Technology, 17, 600e620. Southwell, D. B., & Langrish, T. A. G. (2000). Observations of flow
Oakley, D. E. (2004). Spray drying modeling in theory and practice. patterns in a spray dryer. Drying Technology, 18(3), 661e685.
Drying Technology, 22(6), 1371e1402. Southwell, D. B., Langrish, T. A. G., & Fletcher, D. F. (1999). Process
Oakley, D. E., & Bahu, R. E. (1991). Spray/gas mixing behaviour within intensification in spray dryer by turbulent enhancement. Transac-
spray dryers. In A. S. Mujumdar, & I. Filkova (Eds.), Drying ’91 tions of the Institution of Chemical Engineers, Part A, 77, 189e205.
(pp. 303e313). Amsterdam: Elsevier. Southwell, D. B., Langrish, T. A. G., & Fletcher, D. F. (2001). Use of
Oakley, D. E., & Bahu, R. E. (1993). Computational modelling of spray computational fluid dynamics techniques to assess design alter-
dryers. Computers and Chemical Engineers, 17, 493e498. natives for the plenum chamber of a small spray dryer. Drying
Oakley, D. E., Bahu, R. E., & Reay, D. (1990). The aerodynamics of co- Technology, 19(2), 257e268.
current spray dryers. In A. S. Mujumdar, & M. Roques (Eds.), Drying Straatsma, J., Van Houwelingen, G., Steenbergen, A. E., & De Jong, P.
’89 (pp. 359e365). New York: Hemisphere. (1999). Spray drying of food products: 1. Simulation model. Journal
O’Rourke, P. J. (1981). Collective drop effects in vaporizing liquid of Food Engineering, 42, 67e72.
sprays. Ph.D. thesis. NJ: Princeton University. Sun, D. W. (2007). Computational fluid dynamics in food processing.
O’Rourke, P. J., & Wadt, W. R. (1982). A two-dimensional, two-phase Boca Raton: CRC Press/Taylor and Francis Group.
numerical model for spray dryers. Los Alamos, NM: Los Alamos Tannehill, J. C., Anderson, D. A., & Pletcher, R. H. (1997). Computa-
National Laboratory. Report LA-9423-MS. tional fluid mechanics and heat transfer, (2nd ed.). Philadelphia:
Papadakis, S. E., & King, C. J. (1988). Air temperature and humidity profiles Taylor & Francis.
in spray drying. 1. Features predicted by the particle source in cell Therdthai, N., Zhou, W., & Adamczak, T. (2003). Two-dimensional
model. Industrial Engineering Chemistry Research, 27, 2111e2116. CFD modelling and simulation of an industrial continuous bread
Paris, J. R., Ross, P. N., Dastur, S. P., & Morris, R. L. (1971). Modelling baking oven. Journal of Food Engineering, 60, 211e217.
of the air flow pattern in a counter current spray-drying tower. In- Ullum, T. (2006). Simulation of a spray dryer with a rotary atomizer: the
dustrial and Engineering Chemistry Process Design and Develop- appearance of vortex breakdown. In I. Farkas, & A. S. Mujumdar
ment, 10(2), 157e164. (Eds.), Drying 2006 e Proceedings of the 15th international drying
Parviz, M., & John, K. (1997). Tackling turbulence with supercom- symposium (pp. 251e257). Budapest: Hungary.
puters. Scientific American, 1, 276. Verboven, P., Nicolaı̈, B., Delele, M., Tijskens, E., Atalay, Y., Ho, Q., et
Perry, R. H., & Chilton, C. H. (1984). Perry’s chemical engineers al. (2008). Combined discrete element and CFD modelling of air-
handbook. London: McGraw-Hill. flow through random stacking of horticultural products in vented
Pham, Q. T., & Keey, R. B. (1977). Some experiments on the residence- boxes. Journal of Food Engineering, 89(1), 33e41.
time distribution of droplets in a cocurrently worked spray cham- Verboven, P., Nicolaı̈, B., Delele, M., Schenk, A., & Ramon, H. (2009).
ber. Canadian Journal of Chemical Engineering, 55(4), 466e470. Evaluation of a chicory root cold store humidification system using
Place, G., Ridgway, K., & Danckwerts, P. V. (1959). Investigation of air- computational fluid dynamics. Journal of Food Engineering, 94(1),
flow in a spray dryer by tracer and model techniques. Transactions 110e121.
of the Institution of Chemical Engineers, 37(A), 268e276. Versteeg, H. K., & Malalasekera, W. (1995). An introduction to com-
Reay, D. (1988). Fluid flow, residence time simulation and energy ef- putational fluid dynamics. Essex, England: Pearson Education Ltd.
ficiency in industrial dryers. In M. Roques (Ed.), Proc. sixth inter- Vries, U. D., Velthuis, H., & Koster, K. (1994). Baking ovens and
national drying symposium IDS, unknown, Versailles, France product quality: a computer model. Food Science and Technology
(pp. KL.1eKL.8). Today, 9(4), 232e234.
Richardson, L. F. (1910). The approximate arithmetical solution by finite Wang, L., & Sun, D. W. (2003). Recent developments in numerical
differences of physical problems involving differential equations, with modelling of heating and cooling processes in the food industry
an application to the stresses in a masonry dam. Philosophical e a review. Trends in Food Science and Technology, 14,
Transactions of the Royal Society of London, Series A, 210, 307e357. 408e423.
Roache, P. J. (1976). Computational fluid dynamics. Albuquerque, Woo, M. W., Daud, W. R. W., Mujumdar, A. S., Talib, M. Z. M.,
NM: Hermosa Publishers. Wu, Z. H., & Tasirin, S. M. (2008a). CFD evaluation of droplet
Roustapour, O. R., Hosseinalipour, M., Ghobadian, B., Mohaghegh, F., drying models in a spray dryer fitted with a rotary atomizer. Drying
& Azad, N. M. (2009). A proposed numericaleexperimental Technology, 26, 1180e1198.
method for drying kinetics in a spray dryer. Journal of Food Engi- Woo, M. W., Daud, W. R. W., Mujumdar, A. S., Talib, M. Z. M.,
neering, 90, 20e26. Wu, Z. H., & Tasirin, S. M. (2008c). Comparative study of droplet
398 R. Kuriakose, C. Anandharamakrishnan / Trends in Food Science & Technology 21 (2010) 383e398

drying models for CFD modelling. Chemical Engineering Research Yakhot, V., & Orszag, S. A. (1986). Renormalization group analysis of
and Design, 86(9), 1038e1048. turbulence: i. basic theory. Journal of Scientific Computing, 1, 1e51.
Woo, M. W., Daud, W. R. W., Mujumdar, A. S., Talib, M. Z. M., Zbicinski, I. (1995). Development and experimental verification of
Wu, Z. H., & Tasirin, S. M. (2009). Non-swirling steady and tran- momentum, heat and mass transfer model in spray drying. The
sient flow simulations in short-form spray dryers. Chemical Product Chemical Engineering Journal, 58, 123e133.
and Process Modeling, 4(1). Article 20. Zbicinski, I., & Li, X. (2006). Conditions for accurate CFD modeling of
Woo, M. W., Daud, W. R. W., Talib, M. Z. M., & Tasirin, S. M. (2008b). spray drying process. Drying Technology, 24(9), 1109e1114.
Effect of wall surface properties at different drying kinetics on the Zbicinski, I., Piatkowski, M., & Prajs, W. (2005). Determination of spray
deposition problem in spray drying. Drying Technology, 26, 15e26. drying kinetics in a small scale. Drying Technology, 23, 1751e1759.
Xia, B., & Sun, D. W. (2002). The application of computational fluid Zbicinski, I., Strumillo, C., & Delag, A. (2002). Drying kinetics and
dynamics (CFD) in the food industry: a review. Computers and particle residence time in spray drying. Drying Technology, 20(9),
Electronics in Agriculture, 34, 5e24. 1751e1768.