You are on page 1of 49

Accepted Manuscript

Simulation of food drying processes by Computational Fluid Dynamics (CFD); recent


advances and approaches

Narges Malekjani, Seid Mahdi Jafari

PII: S0924-2244(17)30818-X
DOI: 10.1016/j.tifs.2018.06.006
Reference: TIFS 2247

To appear in: Trends in Food Science & Technology

Received Date: 30 December 2017


Revised Date: 11 June 2018
Accepted Date: 12 June 2018

Please cite this article as: Malekjani, N., Jafari, S.M., Simulation of food drying processes by
Computational Fluid Dynamics (CFD); recent advances and approaches, Trends in Food Science &
Technology (2018), doi: 10.1016/j.tifs.2018.06.006.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Simulation of food drying processes by Computational Fluid Dynamics (CFD); recent
advances and approaches

Narges Malekjani, Seid Mahdi Jafari

Graphical Abstract

PT
RI
Pre- processing Processing Post- processing

U SC
AN
• Problem thinking • Discretization of the • Numerical analysis
• Mesh generation mathematical of the results
• Choosing the equations • Visualization of the
M

computational model • Applying the results


boundary conditions
• Continuing iteration
D

until reaching
convergence
TE
EP

Main scheme of CFD analysis


C
AC
ACCEPTED MANUSCRIPT
Simulation of food drying processes by Computational Fluid Dynamics (CFD); recent
advances and approaches
Running title: CFD in food drying processes
Narjes Malekjani, Seid Mahdi Jafari*
Department of Food Materials and Process Design Engineering, Gorgan University of Agricultural Sciences
and Natural Resources, Gorgan, Iran
*Corresponding Details: Tel/Fax: +98 17 324 26 432. E-mails: smjafari@gau.ac.ir
Abstract

PT
Background

Understanding the mechanisms underlying the drying processes has a critical role in dehydration of

RI
food and agricultural products. Advanced computer modeling and simulation techniques can help in

SC
developing new dryers, modification of current systems, energy saving and process optimization.

Also the most important parameter during the drying food products is food quality (moisture

U
content, crack formation, case hardening, etc.) which can be enhanced through using appropriate
AN
modeling. Computational Fluid Dynamics (CFD) is a well- known modeling technique which has

received more attention in the food industry in the recent years. Hydrodynamics of fluid flow, heat
M

and mass transfer during drying can be predicted using CFD.

Scope and Approach


D

This article reviews fundamentals, merits and shortcomings of CFD in the drying process modeling
TE

with a special focus on dehydration of food products. Since the drying is a growing unit operation,

there is an emphasis on investigation of CFD utilization in modeling emerging drying processes of


EP

food products such as microwave assisted drying, infrared and superheated steam drying besides

conventional convective drying systems notably in recent 5 years.


C

Key Findings and Conclusions


AC

CFD has been considered as a promising method which could help developing the design of new

dryers, enhancing current dryers and the most important aspect of utilization of this method in the

food industry research and development is “food quality” improvement.

Keywords: Computational Fluid Dynamics (CFD); drying; simulation; control; optimization; food

quality.

1
ACCEPTED MANUSCRIPT
1. Introduction

Drying is a multidisciplinary unit operation which can be expressed as the heart of processing

operations in many industries (Dehnad, Jafari, & Afrasiabi, 2016). This process has versatile

utilizations in food, wood and paper, textile, pharmaceutical, chemical and biological material

production (Defraeye, 2014). In food processing, drying is also considered as one of the most

PT
crucial practices which is applied to different types of food products ranging from low moisture

agricultural crops (e.g. rice, corn, and wheat) to intermediate moisture (e.g. pasta, tea, and coffee)

RI
and high moisture (milk, fruits and vegetables) products (Jafari, Azizi, Mirzaei, & Dehnad, 2016;

SC
Jafari, Ghanbari, Ganje, & Dehnad, 2016; Azizi, et al., 2017). The main purpose of drying is

reducing the water content of the product in order to reach a safe level to preserve it from

U
deteriorative factors such as microbial, physical and chemical parameters while retaining its sensory
AN
and nutritional quality and declining the energy and time consumption and optimization of the

process throughput (Bahmani, Jafari, Shahidi, & Dehnad, 2016; Jafari, Ghalegi Ghalenoei, &
M

Dehnad, 2017). Drying process has a sophisticated nature. According to Defraeye (2014), drying is

a multicubed operation; multiphase means it involves different phases (solid food products, liquid
D

water and gas phase of drying air), multiscale ranging from dryer scale to cell walls of the drying
TE

product and multiphysics including simultaneous heat, mass and momentum transport in addition to
EP

many biochemical and chemical processes such as nutritional loss, enzymatic reactions, aroma and

color changes, fat oxidation and microbial degradation that are so critical for ensuring food quality
C

and safety (Defraeye, 2014). Drying is also an energy intensive process and includes 12-20% of the
AC

energy requirements of national industries in developed countries (Bardy, Hamdi, Havet, &

Rouaud, 2015) so it has always been a challenging issue in the food industry (Norton & Sun, 2006;

Strumiłło, 2006).

Investigation of the physicochemical changes occurring during drying in line with developing novel

strategies to optimize the energy consumption, utilization of renewable energy, recovering the heat

used for this process and application of environmentally friendly technologies are becoming more

2
ACCEPTED MANUSCRIPT
important in the current world and any achievement in this area is highly encouraged. These

developments need a lot of time and costly laboratory and experimental efforts. Mathematical

models and computer simulation tools are realistic and effective alternatives to experimental

practices (Hashemi Shahraki, Jafari , Mashkour, & Esmaeilzadeh, 2014; Jafari, Ganje, Dehnad, &

Ghanbari, 2016; Malekjani, Jafari , Rahmati, Esmaeelzadeh, & Mirzaee, 2013). These approaches

PT
can represent an excellent understanding of the transport phenomena during the drying operation

and enhancing the process control leading to drying optimization and improved food quality

RI
(Jamaleddine & Ray, 2010). The main external parameters which affect heat and mass transfer are

SC
ambient air or any drying fluid velocity, temperature and relative humidity; while density, porosity,

permeability, specific heat, mass diffusivity and thermal conductivity are some of the internal

U
parameters affecting the drying process. Simulation can act as virtual sensors of humidity, velocity
AN
and temperature and also internal parameters in inaccessible locations, it is sensitive to small

changes and there are no limitations in testing different and unusual drying conditions. Also, there
M

is no requirement for excessive laboratory spaces, skilled operators and additional maintenance.

Besides these advantages, there are some shortcomings in utilization of the computer modeling and
D

simulation methods in the field of food industry. Multicubed nature of drying process as stated
TE

before, lack of adequate data regarding the material properties especially physicochemical
EP

properties of food and agricultural products which are strongly moisture and temperature

dependent, low economic value of some agricultural products which makes the simulation
C

unworthy, and shortage of user-friendly computer software packages are some of these
AC

disadvantages (Defraeye, 2014).

Computational fluid dynamics (CFD) is a powerful and advanced numerical method to solve

governing partial differential equations (PDEs) of mass, momentum and energy conservation in

fluid flow and heat and mass transfer problems (Norton & Sun, 2010). It also could be noted as a

useful tool in food engineering problems. Visualization of the obtained simulation results for

pressure, velocity vectors, temperature distribution, and species concentration in the form of liquid

3
ACCEPTED MANUSCRIPT
or solid during thermal processes using attractive color figures and animations can help interpreting

of the occurring physical phenomena, thus enhancing the overall process and product quality

(Lemus-Mondaca, Vega-Gálvez, & Moraga, 2011). CFD was first used in the 1950s and since then,

it has been developed increasingly (Norton, Tiwari, & Sun, 2013). The most important motivation

for this study in the first part was to review fundamentals of CFD including the frameworks,

PT
numerical solution methods, introducing relevant software and analysis techniques. In the second

part of the study, some useful and highlighted examples of CFD utilization in the field of drying

RI
especially novel drying technologies are presented in order to introduce the capabilities of this

SC
method to researchers and students in the area of food engineering. As the previous literature of

CFD was mainly focused on drying methods such as spray drying and to some extent fluidized bed

U
drying, in this study it has been tried to show the ability of CFD in simulation of some novel drying
AN
strategies such as microwave assisted, superheated steam drying and infrared drying which are

growing quickly nowadays. Unfortunately, there is a shortage in literature in application of CFD in


M

these novel drying technologies but the authors tried to discuss some of the most important aspects

of CFD simulation for the mentioned methods. Further researches regarding the fundamentals of
D

CFD simulation and its application in drying of food products will be discussed in the following
TE

sections.
EP

2. Principles of CFD in food drying processes

There are some important PDEs which should be solved in order to determine heat, mass and
C

momentum transfer during drying processes. Modeling Newtonian fluid flow is performed using
AC

Navier- Stokes equations. Since drying process involves heat transfer and fluid properties are

usually temperature dependent, so energy equation is usually coupled with Navier- Stokes equation.

When a conjugated heat transfer is being studied, the continuity of thermal energy exchange across

the interface between fluid and solid should be maintained, so, heat transfer in solid body should be

taken into account in CFD simulation.

4
ACCEPTED MANUSCRIPT
Continuity equation or conservation law of mass (eq. 1) implies that there is an axact balance

between the incoming flow of mass in a fluid element with the mass leaving that element:

+ =0 (1)

where and t are density (kg/m3) and time (s), x is Cartesian coordinates (m), u is velocity

component (m/s) and i is Cartesian coordinate index.

PT
Conservation of momentum or Newton’s second law (eq. 2) states that there is a balance between

the rate change of linear momentum and sum of the external forces acting on the fluid element:

RI
+ = − + + + (2)

SC
where j is the Cartesian coordinate index, δ is Kronecker delta, is the dynamic viscosity (kg/ms)

and g is acceleration due to gravity (m/s2).

U
AN
Conservation of energy as the first law of thermodynamics (eq. 3) states that there is an equality

between the energy changing rate of a fluid element and added heat or work done on it:
M

+ − =! (3)
D

where is specific heat capacity (W/kgK), T is the temperature (K), is thermal conductivity
TE

(W/mK) and ! is thermal sink or source (W/m3) (Norton, Sun, Grant, Fallon, & Dodd, 2007).

Fourier equation (eq. 4) can be used to determine heat exchange in an isotropic solid as follows
EP

(Norton et al., 2013):

= +!
C

(4)
AC

Fick’s law of mass diffusion (eq. 5) is analogous to Fourier equation and is generally solved in

drying processes to describe moisture transfer:

"# "#
= $%&& (5)

where $%&& is the effective diffusion coefficient (m2/s) (Wang & Sun, 2003).

Although direct solving of Navier- Stokes equations is possible for laminar flow, solving the fluid

motion in the Kolmogorov microscales in turbulent flow is not possible computationally yet, so,
5
ACCEPTED MANUSCRIPT
Turbulence models should be solved besides Navier- Stokes equations in the case of turbulent

flow regime (which is generally encountered in food process modeling because of high flow rates

and complex geometry involved). Selecting an appropriate turbulence model is very critical and has

a direct effect on CFD results. Choosing from available turbulent models depends on the accuracy

and computational time (or cost) (Defraeye, 2014). Generally, turbulence models are categorized

PT
based on the governing equations used in them. There are two distinct groups of turbulence models:

(1) Reynolds averaged Navier- Stokes (RAN) models, and (2) the models which works on

RI
computing fluctuations e.g. large eddy simulations (LES).

SC
In the case of turbulence, with a statistical probability, it is supposed that process variables will be

in a certain range of values within the flow regime. So, RAN equations are used. The effect of

U
turbulence on flow field would be determined through time averaging using these equations. In
AN
other words, these equations are time averaged Navier- stokes equations of motion for fluid flow

(Norton & Sun, 2010). This averaging method eliminates the stochastic properties of the turbulent
M

flow regime and generates six additional stresses called Reynolds stresses which could be modeled

by different system of equations (for more information, please see Norton and Sun, 2010). Some
D

commonly used RAN turbulence models are (Table 1):


TE

- Standard ' − ( (' represents turbulence kinetic energy and ( for turbulence dissipation rate)

Renormalization Group or RNG ' − (


EP

- Realizable ' − (
C

- ' − ) model (' represents turbulence kinetic energy and ) for specific dissipation rate of
AC

kinetic energy)

- Reynolds Stress Model (RSM).

First three models have similar forms and Standard ' − ( is the most commonly used turbulence

model due to a better conversion (Anandharamakrishnan, 2013). Comparison between RAN models

is presented in Table 1.

Table 1

6
ACCEPTED MANUSCRIPT
The second group of turbulence models, large eddy simulation (LES), are based on the fact that

highly anisotropic large turbulent eddies are dependent on geometry of flow domain and also mean

velocity gradients. The velocity field in this model is separated to a resolved part representing large

eddies and a sub-grid part which represent small scales (Norton & Sun, 2006). For more

information about LES models see Norton and Sun (2006).

PT
RAN models yield satisfactory results for simulation complex large systems with less cost

compared to some other methods. Using LES models is an accurate approach but it requires

RI
powerful computers and is time consuming and consequently expensive. Nowadays advanced

SC
methods like hybrid RANs-LES is utilized which have a higher accuracy to model drying processes

(Kuriakose & Anandharamakrishnan, 2010).

U
It should be noted that none of the mentioned turbulence models are complete and the prediction
AN
performed by these models is strongly dependent to the geometry of the system and flow condition.

Solving mentioned PDEs was a difficult problem in the past. The PDEs had to undergo
M

simplification processes in order to be solved by hand. Accurate determination of some local

phenomena such as fouling, cleaning, mixing, etc. was not possible using those simplified models.
D

Modern CFD approaches have overcome this problem (Norton & Sun, 2010). The main scheme of
TE

CFD analysis is presented in Fig. 1. Also, overall procedure of a reliable CFD model has been
EP

depicted in Fig. 2.

Fig. 1
C

Fig. 2
AC

First of all, it should be noted that simulation procedure may differ depending on the situation and

the information which the researcher wants to obtain in a special problem. Based on this idea, there

are two distinct modeling approaches named conjugated and non-conjugated models. If heat and

mass transfer in liquid or solid state food surface with variable or constant thermo-physical

properties is studied, it is called non-conjugated modeling; but, if solid or liquid food material is

7
ACCEPTED MANUSCRIPT
modeled along with its surrounding media (gas or liquid), it is named conjugated modeling (Lemus-

Mondaca et al., 2011).

As shown in Fig. 2, a critical determination in CFD analysis for multiphase systems is choosing

between Eulerian- Eulerian (EE), Eulerian- Lagrangian (EL) and volume of fluid (VOF) reference

frameworks to represent the governing equations. There are basic differences between EE, EL and

PT
VOF formulations (Fig. 3). To sum up, EL formulation is generally used for dense systems and

individual particles can be tracked in the flow domain. It is expensive and time consuming and

RI
advanced high speed computers are demanded to solve this formulation. EE is used for both dense

SC
and dilute systems and cannot predict the local behavior of the particles. VOF is applicable to two

or more immiscible fluids by solving a single set of momentum equations of each fluid and tracking

U
its volume fraction (Anandharamakrishnan, 2013). More information related to EE and EL
AN
formulations is provided by Jamaleddine and Ray (2010).

Fig. 3
M

One of the other most important factors in CFD analysis is mesh generation. Meshing technology

has improved extensively in recent years and utilization of tetrahedral, hexahedral hybrid and
D

polyhedral meshes has overcome the limitations of simple meshes (Norton et al., 2013).
TE

Details about discretization methods (FEM, FVM, and FDM) are not presented in this article since
EP

there are a lot of valuable articles discussing these methods. For more information about

discretization methods see Mallinson and Norris (2010).


C

There are many commercial CFD codes which can be used in different processes to model food
AC

drying processes. Some of the most important codes are ANSYS Fluent, ANSYS CFX,

OpenFOAM, COMSOL Multiphysics, STAR-CD, FIDAP, ADINA, CFD 2000, PHOENICS,

FLOW 3D, etc. (Bhutta et al., 2012). More details on CFD software packages and codes are

presented by Xia and Sun (2002).

It should be noted that the accuracy of any CFD simulation depends on many factors such as the

assumptions and simplification of the model which has been made, the empirical correlations used,

8
ACCEPTED MANUSCRIPT
simplification of boundary conditions and the geometry in order to reduce computation time and

existence of other physical mechanisms such as chemical kinetics.

Like any other modeling methods, CFD has some advantages and disadvantages for drying

simulations. Table 2 summarize some of these aspects (Norton et al., 2013; Xia & Sun, 2002)

Table 2

PT
Jamaleddine and Ray (2010) reviewed the application of CFD in some drying systems with a

special focus on gas-solid multiphase flows (Jamaleddine & Ray, 2010). Also, Pragati and sharma

RI
(2012) reviewed the application of CFD in food processing equipment design e.g. cleaning of food

SC
processing tanks, optimization of hygiene, drying, pasteurization, sterilization, mixing, baking,

refrigeration, heat exchangers and crystallization. They explained further research is demanded

U
regarding to control of the drying processes and reduction of energy costs. These authors concluded
AN
that CFD can be utilized for predicting gas flow patterns, particle histories (temperature, velocity,

residence time and impact position), designing drying chambers, scale up studies and air- particle
M

interactions successfully. They mentioned lacking spatial homogeneity of air velocity is a major

problem resulting in the instability of CFD data for dryers (Pragati & Sharma, 2012).
D

In a recent study, Norton et al. (2013) reviewed CFD modeling of some thermal food processes
TE

including sterilization, pasteurization of canned foods (food in pouches, intact egg, plate heat
EP

exchangers for milk and yoghurt processing), drying (fluidized bed drying, spray drying and forced

convection drying), cooking (natural and forced convection, commercial baking and microwave and
C

infrared ovens). They also had a short look on application of CFD in modeling some emerging
AC

technologies such as high pressure thermal processing and Ohmic heating (Norton et al., 2013).

Defraeye (2014) made a comprehensive review on computational methods for modeling

dehydration of porous materials specially food products including CFD. He reported that the

number of publications reflecting the application of CFD in drying technology has increased

significantly between 2002- 2014 (Defraeye, 2014).

9
ACCEPTED MANUSCRIPT
Although there are some valuable articles which have reviewed the studies of CFD modeling,

developing drying technologies and increasing the number of publications in this area made the

authors convinced to have a fresh look at novel achievements in this field, particularly the food

drying processes. The main goal of this review article is investigating the utilization of CFD in the

recent publications and also emerging drying technologies such as microwave assisted, infrared, and

PT
super-steam drying.

3. Classification of food dryers

RI
The drying technologies can be classified based on different aspects. Some researchers have

SC
categorized dryers as direct and indirect, batch and continuous, in gas or in vacuum, etc. One of the

best classification of dryers has been presented by Mujumdar (2008) who determined four

U
categories for drying facilities based on the strategy used for drying, the medium in which the
AN
material is being dried, the way that solids are handled and the heat input mode (Mujumdar, 2008).

Fig. 4 demonstrates briefly some of the dryers in this categorization. Nowadays more than 85% of
M

dryers in the industry are convective dryers which utilize hot air or combustion gases as the drying

medium (Moses, Norton, Alagusundaram, & Tiwari, 2014). The poor quality of the final dried food
D

products and low yields in these types of dryers is a critical issue. In order to find more efficient and
TE

economical drying methods, scientists have combined some of these different categories and
EP

developed new dryer designs (Moses et al., 2014). For example, microwave assisted convective

drying, ultrasound assisted drying or combining fluidized bed dryers with vacuum are some of these
C

novel techniques.
AC

Fig. 4

4. Application of CFD in convective drying systems

Convective dryers are the most frequent and popular equipment used in industrial applications. Hot

air or combustion gases might be used as heat transfer medium in these type of dryers (Moses et al.,

2014; Tzempelikos et al., 2015). The efficiency of convective dryers is dependent to several factors

such as drying air speed, temperature, relative humidity and flow uniformity (Chandramohan,

10
ACCEPTED MANUSCRIPT
2016). There are different types of convective dryers like tunnel, tray, fluidized bed, spray and solar

dryers. In the following sections, some of the recent studies concerning CFD modeling and

simulation of convective dryers in food processing are discussed.

4.1. CFD in spray dryers

Spray drying is a method for converting liquid materials into powder form. It has the advantages of

PT
high drying rates, broad range of drying temperatures and short drying periods (Jafari et al., 2017).

One of the most important problems in spray dryers is unsteady nature of flow which may cause

RI
wall deposition or overheating of the product. The 3D nature of the spray dryers makes it

SC
impossible to use empirical models for their describing. So, CFD is a useful tool for modeling this

type of dryers. Eulerian- Lagrangian approach has been used in CFD modeling of spray dryers in

U
most cases which can predict the motion path of particles and heat and mass transfer between the
AN
particles and drying air (Lo, 2005; Norton et al., 2013). Spray drying process is comprised of four

major parts including atomization, contact between the particles and drying air, evaporation of the
M

water and in the last part, separation of the dried powder from the drying air (cyclone). CFD has

been used for modeling all of these parts and it can predict velocity and temperature distribution,
D

particle size and humidity as well as several other factors in spray drying processes that effects the
TE

final product quality and dryer design (Anandharamakrishnan, 2013). Although there are a lot of
EP

advantages in CFD modeling of spray dryers, there are also some limitations because of

simultaneous existence of both solids and fluids and difficulty in predicting mass transfer within the
C

particles and necessity of applying sub-models (e.g. sub-models accounting for mass transfer,
AC

collision between particles, thermal reactions, stickiness and agglomeration) in order to take this

issue into account. Also, validation of CFD results in spray drying process is complicated since

local measurement of process parameters is to some extent impossible and expensive even in a

laboratory scale. Most of the results in literature have been validated using average outlet data. On

the other hand, the results are dependent on empirical models that are not completely reliable

(Mujumdar, Huang, & Chen, 2010).

11
ACCEPTED MANUSCRIPT
Kemp and Oakley (2002) stated that the main difficulties in CFD modeling of spray drying process

are:

- There is no similarity in drying hydrodynamics during process scaling

- Circulation of air inside drying chamber may move particles on different trajectories

- Heating story may change quality and morphology of the drying products

PT
- Determination of drying kinetics of material under drying process is difficult.

There have been many reviews on application of CDF in spray drying. For instance, Lo (2005) and

RI
Kuriakose and Anandharamakrishnan (2010) have reviewed the application of CFD in spray drying.

SC
Mujumdar et al. (2010) have also studied some advances in spray drying focusing on CFD

modeling. Jamaleddine and Ray (2010) and Norton et al. (2013) dedicated a part of their studies to

U
review the application of CFD in spray drying. Most of the CFD studies in the field of spray drying
AN
is concerned to co- current towers so, Abdullah et al. (2017) have recently reviewed fundamentals

of counter current spray dryers. As there are comprehensive reviews in the field of CFD modeling
M

of spray dryers, we refer the readers to these articles and only review some of the most recent

studies.
D

In most papers, CFD simulation has been performed to simulate hydrodynamics of bed or maximize
TE

drying thermal efficiency or lowering the overall cost of the spray drying process. Although these
EP

factors are important, the crucial factor in food drying is the product’s quality. Based on this

assumption Schmitz-Schug et al. (2016) studied the impact of spray drying conditions on lysine
C

loss. Reaction engineering approach (REA) which has been used for modeling drying kinetics was
AC

implemented in CFD analysis by STAR-CCM+ Software. The model representing kinetics of lysine

loss took temperature, moisture and physical state of lactose into account. The residence time and

properties of particles was solved using CFD and coupled with reaction kinetic model to predict

lysine loss. They suggested that this approach could be used in optimization of spray drying for

dairy powders in order to prevent lysine loss (Schmitz-Schug, Kulozik, & Foerst, 2016)

12
ACCEPTED MANUSCRIPT
Jaskulski et al (2017) performed a similar study and tried to investigate thermal inactivation of

whey proteins during spray drying of skim milk using ANSYS- Fluent CFD package. Particle

moisture, temperature profile and residence time was predicted by evaporation and particle

formation extended models and were utilized as input data in a quality model of inactivation

kinetics. Then, the resulted quality model was implemented into the CFD code and simulation was

PT
conducted and validated. Also, Differential Scanning Calorimetry (DSC) was used to measure whey

protein inactivation during drying experiments. A good agreement was shown between

RI
experimental and predicted whey activity that represented the successful development of the CFD

SC
code. So, this model can be suggested to predict the whey protein loss during spray drying of skim

milk (Jaskulski, Atuonwu, Tran, Stapley, & Tsotsas, 2017).

U
Many CFD models have not been verified in the literature. So, data provided by such studies can
AN
hardly be used in a large scale. In a recent work by Lisboa et al (2018), an industrial scale spray

dryer was modeled. They used simplified basic models to estimate key parameters for spray drying
M

design. The model was validated for a wide range of food products at different temperature and

relative humidity levels. The authors suggested that such models could be used instead of
D

complicated CFD models in many applications (Lisboa, Duarte, & Cavalcanti-Mata, 2018).
TE

The authors think that although velocity and flow pattern during spray drying can be predicted
EP

easily in CFD packages, but it is necessary to conduct studies concerning drying process control and

energy and optimization in this era. Also, some simplifying assumptions like considering spatial
C

homogeneity which leads to unrealistic results should be avoided. Finally, since designing a spray
AC

dryer which produces high quality products is a very sophisticated process, it is thought that besides

aspects related to drying cost and efficiency, correlating quality kinetics of degradation of food

products to the transport phenomena occurring during the process may be the most holistic

approach in spray drying of valuable food and pharmaceutical products which is still lacking and

could be focused with more details.

13
ACCEPTED MANUSCRIPT
4.2. CFD in fluidized bed dryers (FBD)

Fluidization is a deep-rooted technique which has been applied successfully in many industrial

applications (Lettieri & Mazzei, 2009). FBDs can be used for the drying of wet particulate materials

that behave like a fluid in the drying chamber. High heat and mass transfer coefficients between

drying materials and drying air medium due to their high contact and gentle mixing leads to

PT
elevated drying rates compared to some other drying methods. However, there are some

shortcomings in this process e.g. scale-up issues, uneven fluidization and product quality. Solving

RI
these problems depends on gaining a comprehensive knowledge about the process and reliable

SC
mathematical correlations which can be used to predict, optimize and scale- up the process (Mortier

et al., 2011). CFD has been used extensively to explore the interactions between multiple phases

U
and prediction of different phenomena in this drying system (Lettieri & Mazzei, 2009; Malekjani,
AN
2017). In regard of existing different phases and chemical reactions which may be conducted in

such a process, the CFD modeling is a complex procedure. One of the most important difficulties in
M

CFD modeling of FBDs is modeling the turbulence behavior (because of 3D and transient nature).

Numerical modeling of fluidized beds is performed by coupled solving of mass, momentum and
D

energy equations with the equations describing interphase interactions and one of the main
TE

difficulties in CFD simulation of fluidized bed dryers is modeling multiphase interactions


EP

(Philippsen, Vilela, & Zen, 2015). As it was mentioned in the previous section, Eulerian- Eulerian

and Eulerian- Lagrangian approaches are used to simulate multiphase flows. Eulerian- Eulerian
C

approach is widely used for CFD modeling of the fluidized beds. One of the other important
AC

limitations in CFD modeling of fluidized bed dryers is very long computation times required for

modeling just a fraction of drying process (Szafran and Kmiec, 2005), although with the continuous

progress in computers and invention of powerful supercomputers, these calculations can now be

performed in a very short time.

Verification of the CFD models in FBDs has been conducted using X- ray imaging, optical fiber

and pressure probes, laser and 3D capacitance imaging methods (Mortier et al., 2011). Some of the

14
ACCEPTED MANUSCRIPT
applications of CFD in FBD include designing different parts of the dryers especially gas

distribution section and draft tube and presenting online models in order to control the drying

process (Norton et al., 2013).

Mortier et al. (2011) have reviewed modeling of FBD for wet granular materials. Lettieri and

Mazzei (2009) discussed some of the challenges on the CFD modeling of FBDs. Pan et al. (2016)

PT
reviewed the fundamentals and applications of three phase fluidized bed reactors. Jamaleddine and

Ray (2010) and Norton et al. (2013) dedicated a part of their works to review the studies concerning

RI
the application of CFD in modeling spouted and fluidized bed drying processes. In a recent study by

SC
Philippsen et al. (2015), they presented different methods of FBD modeling including CFD.

Spouted bed dryers are used instead of FBDs for drying larger, irregularly shaped, sticky and

U
heavy particles. In this method, there is a high speed injection of a gas flow which moves the solids
AN
to the center of the drying chamber until reaching the top of it (Anandharamakrishnan, 2013).

Nazghelichi et al. (2013) studied the hydrodynamics and heat transfer in a lab scale FBD for drying
M

carrot cubes using Fluent. Three levels of bed height, cube size and inlet air temperature were

investigated in order to evaluate the effects of each factor on energy optimization ratio. Also,
D

Taguchi technique was used to rank the mentioned factors. The results revealed that the cube size
TE

was the most significant effect among the others. The authors declared that their method could be
EP

applied in energy utilization optimization in FBD drying process. Their findings showed that as the

particle size is smaller, the bed is deeper and higher and the drying air temperature is higher, the
C

energy utilization is more (Nazghelichi, Jafari, Kianmehr, & Aghbashlo, 2013).


AC

Azmir et al. (2018) coupled CFD and discrete element method DEM method to describe heat and

mass transfer in a fluidized bed dryer containing spherical corn kernels. DEM method is used to

investigate the granular flow. Each grain is tracked using Newton’s laws of motion while all forces

due to gravity, particle- particle, particle- wall and electrostatic fields are considered. This method

helps in gaining more detailed information in particle scale. In this approach, the on-line data about

particle velocity, temperature, position and moisture content produced by DEM is used for solving

15
ACCEPTED MANUSCRIPT
CFD governing equations. Then the results of CFD analysis would be incorporated into the DEM.

These authors used 2D and 3D geometry for CFD and DEM analysis, respectively. An in- house

developed code was used for CFD-DEM analysis. They examined the effects of drying air

temperature and velocity on grain and air moisture content. The results showed that as the drying air

velocity and temperature rise, the drying rate increases. They described the corn kernel quality by

PT
standard deviation of its moisture distribution and reported that in higher air velocities and lower

drying air temperatures, the quality of drying material is superior (Azmir, Hou, & Yu, 2018).

RI
In Table 3, a brief overview of CFD application in the field of fluidized and spouted bed drying of food

SC
and agricultural products has been provided.

Table 3

U
4.3. CFD in solar dryers
AN
Sun drying is one of the oldest and cheapest methods of drying food and agricultural products but, it

is labor- intensive, long and has the risk of pollution of food with insects, molds, etc. So, solar
M

dryers have been developed because they are easier to control and protection of the product from

pollution is possible (Sanghi, Ambrose, & Maier, 2017). Solar dryers are categorized into four
D

major types: direct, indirect, mixed, and hybrid. CFD can be used to predict velocity and
TE

temperature distribution profiles in these dryers (Chauhan, Kumar, & Tekasakul, 2015). Chauhan et
EP

al. (2015) have reviewed the application of computer modeling in solar dryers. In two other recent

studies, Milczarek and Alleyne (2017) and Praka sh et al. (2017) reviewed mathematical modeling
C

of solar dryers.
AC

In a study by Romero et al. (2014), they investigated an indirect solar dryer for vanilla drying.

Variation of temperature inside the dryer was predicted using ANSYS- Fluent CFD package. Their

results showed a good agreement between experimental and predicted data at collector outlet while

inside the chamber, some deviations were found. The authors ascribed these variations to

approximation of heat transfer coefficient and suggested using a variable heat transfer coefficient as

a function of time during day for solar dryers (Romero, Cerezo, Garcia, & Sanchez, 2014).

16
ACCEPTED MANUSCRIPT
In another study, an indirect solar cabinet dryer was simulated using ANSYS- Fluent for drying

sliced tomatoes. The ambient local weather was used as the boundary condition. Temperature and

velocity distribution were predicted. Thermal efficiency of the dryer without any load, under half

load and full load was measured. The maximum and minimum temperatures in the dryer were

predicted to be 70 and 27 ℃, respectively. Thermal efficiency of the dryer was 21.5 %.

PT
(Tesfamariam, Bayray, Tesfay, & Hagos, 2015)

Solar cabinet drying process of corn has also been studied using CFD. Temperature, humidity and

RI
velocity profiles of the drying air was simulated and validated with experimental data. The major

SC
objective of this study was developing a predictive model based on local changes of the weather in

order to enhance drying efficiency. A 3D model was developed by solving continuity, Navier-

U
Stokes and energy equations using Ansys- Fluent package. Moisture transfer equation was solved
AN
independently. Thin layer drying models were also used to model the drying process of corn

kernels. The porosity of the flow fields was considered as the source terms in all of the governing
M

equations. Radiative transfer equation (RTE) was added as a source term to energy equation to

represent the radiative effects. The effect of turbulence was also evaluated using RNG ' − ( model.
D

The local weather data was used as boundary condition. The results showed that the model over-
TE

predicted both temperature and humidity to some extent. A stagnation which was seen in
EP

experimental humidity data was predicted by model accurately. This model was able to simulate the

drying process under overcast conditions and amount of removing the moisture was 32% less than
C

fair weather conditions (Sanghi et al., 2017).


AC

Recently, Orbegoso et al. (2017) performed numerical simulation of one- step and three- step solar

collectors for solar drying of cocoa bean. The aim of their study was simulating heat transfer due to

convection and radiation in order to determine the best configuration of the collectors regarding

outlet drying air temperature and drying thermal efficiency. Mass, momentum and energy equations

were solved coupled with discrete ordinate model (DOM) for discretizing RTE and standard ' − (

turbulence model in an Eulerian framework using Ansys- Fluent package. The results showed that

17
ACCEPTED MANUSCRIPT
the three- step collector located between the second and third channel in dryer was the most

thermally efficient configuration (67% more efficient than one- step). These authors showed the

importance of CFD modeling in design and optimization of energy in such type of dryers

(Orbegoso, Saavedra, Marcelo, & La Madrid, 2017).

Some limitations in CFD modeling of solar dryers include transient nature of mixed and natural

PT
convection of airflow within the dryer, variations and complexity of the boundary conditions which

is the local weather, shrinkage in drying material which necessities applying moving meshes,

RI
existing solar radiation besides long wave radiation, more sophisticated drying kinetics than forced

SC
convection systems because of partial rehydration (Prakash & Kumar, 2017).

5. CFD in freeze drying

U
Lyophilization or freeze drying is an excellent method for retaining original quality attributes of the
AN
dried products. This method preserves the biological activity of food components, flavors, colors,

and aroma but it is a complicated, costly and time-consuming process. In this method, the water
M

inside the food material is first frozen; then the created ice is sublimated (Nakagawa & Ochiai,

2015). This process can be implemented at atmospheric pressure or under vacuum (Li, Stawczyk, &
D

Zbicinski, 2007).
TE

There have been some studies on the modeling of freeze drying process for food products. Some
EP

researchers have focused on mathematical modeling of this process (Bubnovich, Quijada, & Reyes,

2009; Bubnovich, Reyes, Quijada, & Mahn, 2012; Nakagawa & Ochiai, 2015; Nam & Song, 2007),
C

and some others have emphasized on modeling the freeze dryer parts such as drying shelf (Cheng &
AC

Tsai, 2012), condensers (Petitti, Barresi, & Marchisio, 2013) or heat exchangers (Cheng, Tsai,

Cheng, & Chen, 2014). There are a few studies regarding CFD simulation of freeze drying of food

products (Coletto, 2015; Li et al., 2007).

Li et al. (2007) developed a two-dimensional model based on film sublimation and uniformly

reaching ice front (URIF) model using FLUENT 6.1 software. The model took into account phase

changes and water vapor diffusion inside the porous media. Species Transport model in FLUENT

18
ACCEPTED MANUSCRIPT
was used to simulate the transport of non-condensable species and to predict the rate of sublimation.

The drying samples (apple cubes) were divided into four zones including the frozen front (pure ice

plate), an ice- vapor interface, the dry zone (porous zone) and a gas phase. The frozen front zone

was treated as a wall where ice sublimation starts in and a UDF was used to determine the

sublimation rate. A source term was added to the adjacent walls in order to consider the effect of

PT
water vapor generation on the gas phase species distribution. Flow through the dry zone (porous

zone) was simulated using inbuilt CFD model for porous media. The results were validated with the

RI
experimental data of apple cubes and showed a reasonable agreement. Results showed the internal

SC
resistance of porous medium is dominant. Absorption and desorption of the samples enhanced

moisture transport (Li et al., 2007).

U
In a more recent study, Coletto et al. (2015) aimed to simulate the entire process of freeze drying
AN
instead of a piece of product. They focused on atmospheric freeze drying (AFD) using wheat bran

as an adsorbent in a fluidized bed dryer to enhance the AFD process efficiency. ANSYS FLUENT
M

was used to simulate the process. Eulerian- Eulerian aproach for fluidized bed dryer and

hydrodynamics of the dryer was successfully simulated (Coletto, 2015).


D

6. CFD in novel drying technologies


TE

In order to achieve better results of energy efficiency and product quality, new dryer designs are
EP

emerging from the results of scientific research. Modern technologies such as microwave and

ultrasound are used to enhance existing drying systems for food products. Some other technologies
C

have been developed recently, e.g. superheated steam drying. Most of the researches in the field of
AC

drying are conducted on novel drying technologies so that the widespread and commercial use of

these systems are very limited due to lack of sufficient understanding of the process costs and

commercial design and large-scale applications (Moses et al., 2014). Therefore, CFD models can

provide new insights in developing and designing these systems in the industrial scale. In the

following subsections, some of the less common drying systems along with application of CFD in

modeling and simulation of these processes are discussed. More details are presented in Table 4.

19
ACCEPTED MANUSCRIPT
Table 4

6.1. CFD in microwave- assisted drying processes

Microwave drying refers to the utilization of electromagnetic radiations in order to produce a

volumetric heating pattern. This process causes a meaningful time and energy reduction and

improved product quality compared to conventional drying systems (Schubert & Regie, 2006).

PT
One of the main advantages of microwave and microwave-assisted drying is energy efficiency.

Energy saving in this type of dryers is mainly due to shorter drying times and higher dying rates.

RI
Another reason for energy efficiency of microwave-assisted drying is that, microwave energy is

SC
absorbed only by dielectric materials, so, energy loss through air or transfer to oven walls,

conveyors and other parts of the equipment is negligible which can result in significant energy

U
savings. For instance, 25-90% drying time reduction and 400-800 % increment in drying rate and
AN
32-71% energy saving has been reported during microwave-assisted drying in comparison with

convective drying techniques. This higher energy efficiency is more dominant specially in falling
M

rate drying period (Moses et al., 2014).

Application of microwave radiation combined with other drying techniques especially convective
D

drying helps to overcome some limitations such as non-homogenous heat distribution, unwanted
TE

food quality changes (textural deterioration, color, nutritional loss, etc.) and also, low penetration
EP

depth of microwave into the food products (Feng, Yin, & Tang, 2012; Malekjani, Emam-Djomeh,

Hashemabadi, & Askari; Moses et al., 2014). Microwave-assisted hot air drying is usually
C

employed at the final stages of the drying process. The most common applications of this
AC

technology include microwave-assisted vacuum drying, microwave-assisted fluidized bed drying

applied to granular food and agricultural products (notably spouted bed drying for heat sensitive

products), and microwave assisted vacuum drying (Rattanadecho & Makul, 2016).

Prediction of heat and mass transfer in microwave-assisted technologies is a challenging task due to

high dependency of dielectric constant (( , ) and dielectric loss factor (( " ) to the physical properties

20
ACCEPTED MANUSCRIPT
of drying material such as temperature, moisture content, porosity and changes in its chemical

composition (Feng et al., 2012).

There are various methods for the modeling of microwave-assisted drying process. It involves

simultaneous solving of the models which describe microwave heat generation and the models

which can predict temperature distribution inside the food product. Solving Maxwell’s equations

PT
(eqs. 6-9) which are four fundamental governing equations can describe microwave energy

absorption and pattern accurately. Although modeling microwave heating with Maxwell’s equations

RI
is highly accurate, it is sophisticated and additional data about dielectric properties of the food

SC
material is required.

1
∇×0 = − (6)

U
4
∇×2 = 3+ (7)
AN
∇. $ = 6 (8)

∇. 7 = 0
M

(9)

where E and H are electric (V/m) and magnetic field intensity (A/m), D and B are electric
D

displacement (N/V m) and magnetic induction (T) respectively. J is current flux (A/m2) and q is
TE

electric charge density.

Lambert’s law (eq. 10) is used for the determination of microwave energy and it is a straightforward
EP

formulation; however, it doesn’t ensure determining a comprehensive scheme of electromagnetic


C

field distribution (Liu, Wang, & Sakai, 2005; Yanniotis, 2007). So, it is possible to model
AC

microwave drying by solving governing heat and mass transfer equations while adding the

microwave volumetric heating as a source term with exponential dissipation (Lambert’s law) as an

alternative approach for solving complicated Maxwell’s equations. Many authors have applied

Lambert’s law in modeling of microwave drying (Arballo, Campañone, & Mascheroni, 2012;

Hemis, Choudhary, & Watson, 2012; Joardder, Kumar, & Karim, 2017; Kumar, Joardder, Farrell,

Millar, & Karim, 2016; Souraki & Mowla, 2008). Lambert‘s law has some limitations. First, the

size of the sample should be semi-infinite. Second, effects of standing wave are not considered, and
21
ACCEPTED MANUSCRIPT
finally, penetration of microwave energy is one-dimensional. In Lambert‘s law it is supposed that

incident energy is normal to the surface and energy dissipation has exponential manner.

8 9 = 8: ; <=> (10)

where P(x) is the amount of power dissipation at depth x (W), x is the depth from the samples

surface (m), P0 is the power at the surface (W), and ? is the attenuation constant (m-1) which is

PT
dependent to velocity of radiation (m/s), frequency (Hz) and loss tangent (Pitchai, 2011).

Ranjbaran and Zare (2012) performed a comprehensive CFD simulation of soybean drying in a

RI
microwave assisted fluidized bed dryer. They used a calorimetric method (Yang & Gunasekaran,

SC
2004; Zare & Ranjbaran, 2012) in order to determine the initial microwave power density and

proposed a modified equation to calculate microwave heat source. The model could predict the

U
effects of different levels of microwave power densities and initial drying air temperature on
AN
product moisture content, drying air temperature and absolute humidity with an appropriate

agreement between simulation and experimental data. They claimed that the small deviations of
M

simulation results from the experimental data could be related to laboratory measurement errors
D

occurred while determining the initial power density of microwave, air temperature, velocity and
TE

relative humidity (Ranjbaran & Zare, 2012).

Radio frequency (RF) module in Finite element software COMSOL Multiphysics is a useful and
EP

user-friendly tool in CFD modeling of microwave drying. Zhu et al. (2015) performed a more basic

study on microwave drying of spherically shaped food products (potato spheres). Instead of using
C

the experimental calorimetric method to determine the microwave power density, these authors
AC

coupled the solution of electromagnetic governing models with multiphase porous media models.

Maxwell’s electromagnetic field equations were solved to identify a more actual explanation of

electric field distribution inside the samples and microwave oven. They reported the use of less

sophisticated Lambert’s law in most cases can only present a qualitative explanation of microwave

penetration within the material and moisture content profiles so, they selected Maxwell’s equations.

Temperature distribution, point temperatures, moisture content and pressure changes in three

22
ACCEPTED MANUSCRIPT
different sizes of potato spheres and the oven were tracked during the simulation process (Zhu,

Gulati, Datta, & Huang, 2015). They found that different size samples had different responses to

microwave heating at the same drying conditions. The intermediate size spheres showed more heat

absorption in the microwave oven which caused their explosion during drying process while smaller

size particles experienced uniform temperature and had better quality. They also, reported that the

PT
model was very sensitive to mass transfer coefficients of surrounding air within the oven. The

model data was in a reasonable agreement with the experimental data (Gulati, Zhu, Datta, & Huang,

RI
2015).

SC
In another study, Joardder et al. (2017) used Lambert’s law instead of solving Maxwell’s equations

so that intermittent microwave convective drying with a multiphase porous media model was

U
simulated. The model was solved for liquid, gas and solid phases and considered shrinkage and
AN
porosity during drying; it was compared with the model without shrinkage. Temperature, moisture

content, porosity, and density was determined using CFD and apple slice was used for validating
M

the simulation process. A good agreement reported between the experimental and simulation data.

Taking the shrinkage into account helped to achieve more realistic explanation of heat and mass
D

transfer. The model provided a good insight of drying process and these researchers suggested it
TE

could be applied in process quality enhancement and optimization. The authors reported that some
EP

parameters e.g. the rate of evaporation, effective thermal conductivity, capillary diffusion and gas

pressure which cannot be predicted by simpler models, could be predicted using their developed
C

model. These parameters can help in some practical conditions instead of real time calculations
AC

(Joardder et al., 2017).

Although some efforts have been made in CFD modeling of microwave assisted drying but there

are still important challenges in this area. Variation of dielectric and thermal properties of food

products with temperature and frequency changes, lack of adequate data about dielectric properties

of food and agricultural products, scarcity of appropriate CFD packages including physics of

microwave process completely for solving the related equations in real situation, considering actual

23
ACCEPTED MANUSCRIPT
conditions in microwave ovens including the complex oven geometry are some of these issues

(Pitchai, 2011; Sosa-Morales, Valerio-Junco, López-Malo, & García, 2010).

6.2. CFD in infrared drying processes

Wavelengths between 0.75 and 1,000 µm in the electromagnetic spectrum is called infrared which

is used in modern food processing technologies (Moses et al., 2014; Riadh, Ahmad, Marhaban, &

PT
Soh, 2015). Infrared drying is a novel drying method which benefits from a higher rate of

dehydration, efficient conversion of electrical energy into heat, the selectivity of heating and higher

RI
speed of shutting down and starting up the drying processes. It also requires a small space, the

SC
control of the processing unit is easy and it has few installation and capital costs (Tsotsas &

Mujumdar, 2011).

U
One of the most critical aspects of infrared dryer utilization is energy saving which could be due to
AN
supplying the energy directly to the drying material without dispersing it into other objects. The

energy efficiency of infrared dryers is related to the absorption properties of drying material that
M

determines the characteristics of the dryer (Pawar & Pratape, 2015).

CFD modeling of infrared dryers has recently been conducted in some researches and it is still a
D

new concept. Wu et al. (2017) explained that there are two assumptions which can be made in
TE

modeling infrared drying based on infinity and zero penetration depth of infrared radiation. As the
EP

penetration depth of the infrared waves is about a few millimeters, for the products with a larger

size than several millimeters such as peanut, soybean, etc. it can be assumed as a boundary
C

condition for heat transfer. In the products with limited size like wheat, rice, etc. it can be
AC

considered as an internal heat source or a heat transfer boundary condition. These authors used both

concepts in order to study the influence of penetration depth of infrared radiation on drying rice

kernels. The radiation source term in heat transfer equation was assumed to exist in the model with

the penetration depth of infinity and was neglected in the model with zero penetration depth and the

radiation was treated as a boundary condition. The results showed that both models could accurately

predict drying curves and temperature during drying and the maximum predicted temperature

24
ACCEPTED MANUSCRIPT
difference was 1.5 ℃ between two models so both assumptions can be applied in CFD modeling of

infrared drying for small sized particles (Wu, Zhang, & Li, 2017).

In order to calculate infrared power source term, the penetration depth of IR waves should be

defined, which is the depth at which the absorbed infrared power declines to 1/e of its original value

at the surface of the material. It can be considered that 63.21 % [(1-(1/e)] of the power at the surface

PT
is the effective fraction of power in order to produce volumetric heat. So, the volumetric heat

generated per unit volume of the layer at any the particle temperature @ which it penetrates can

RI
be calculated using eq. 11:
L

SC
CDE,GHI J K<
8AB @ = N O <BO
M
(11)
P J BJ JMQ
O

where 8AB is the infrared power density (W/kg), 8AB,R

U
S is the initial infrared power (W) absorbed at

the surface of the particle, T@ is particle radius (m) and T@%U is internal penetration layer radius (m).
AN
8AB,R S is a function of particle surface temperature and can be calculated by using eq. 12.
M

N
V[ M X=YZ.K[ N < J X=YZ.K[ ]
8AB,R S @ = L]^M L L]^J (12)
X X
^M _M `M]J ^J _J
D

where (@ and (% are emissivity of particle and emitter, A is surface area and a%<@ is the view factor
TE

(part of radiation which strikes the particle surface from the emitter surface) (Nejadi & Nikbakht,

2017).
EP

Aktas et al. (2016) investigated a novel type of infrared dryer combined with a solar collector and a
C

heat recovery unit in order to decrease the high amount of energy required for infrared drying. They
AC

studied heat and mass transport with a three dimensional CFD model to investigate drying kinetics

of melon slice and flow behavior in the drying system. Velocity contours and temperature

distribution profiles were determined using CFD results. They reported that the performance of

solar-infrared dryer has been better with the utilization of the solar energy and heat recovery units.

The drying efficiency with heat recovery system and without it at 50 °C were 33.4% and 22.6%,

25
ACCEPTED MANUSCRIPT
respectively. The experimental and theoretical findings were in a good agreement and such results

could be used in solar- infrared dryers design (Aktaş et al., 2016).

In another study, the previous authors studied heat and mass transfer characteristics of apricot in an

infrared dryer. One of the main assumptions in this study was neglecting the sample shrinkage.

VOF multiphase model was used to describe temporal and spatial changes of the fluid flow.

PT
Velocity and temperature distribution and water volume fraction in the chamber and near apricots

were determined using CFD results. The authors concluded that efficient CFD analysis could have a

RI
positive effect on designing dryer and product quality (Aktaş, Sözen, Amini, & Khanlari, 2017).

SC
6.3. CFD in super-heated steam drying processes

Super-heated steam drying is referred to a drying process which applies steam of higher temperature

U
than saturation temperature in a specific pressure instead of hot air. This process has the benefits of
AN
the high drying yields, reducing the risk of oxidation, small required space, reduced risk of

explosion and fire and no need of disinfection. The superheated steam passes over the drying
M

product and causes the moisture inside it to boil which is then moved from the interior parts into the

surrounding. This process can be accompanied by microwave, fluidized bed, infrared, or other
D

drying systems (Devahastin & Mujumdar, 2014; Moses et al., 2014; Mujumdar, 2014).
TE

Superheated steam drying is an energy efficient process. Utilization of superheated steam instead of
EP

hot air as a drying medium can reduce the energy consumption by recycling the exhausted vapor

energy (the latent heat of evaporation could be recycled during condensation of the exhausted
C

vapor) and avoiding emission of gas into surrounding medium by condensation. In contrast, in hot
AC

air drying systems, latent heat recovery in the released moist air is sophisticated and costly. It has

been shown in the literature that about 85% of input energy could be saved in pilot scale

superheated steam dryers. Also, because of high heat transfer coefficient of superheated steam, the

drying time is very short which leads to high drying rates and low drying costs and energy saving

(Romdhana, Bonazzi, & Esteban-Decloux, 2015).

26
ACCEPTED MANUSCRIPT
Superheated steam drying is a complicated process because of sophisticated heat and mass transfer,

interactions between vapor and solid phases and also particle-particle interactions. It is a multiphase

turbulence flow. Xiao et al. (2012) investigated superheated steam fluidized bed drying process for

rapeseeds as an isometric, spherical and constant size solid. They considered the system a two phase

vapor- solid mixture with volume fractions that obeyed eq. 13. The continuity equation in such a

PT
system is expressed as eqs. 14 and 15.

bc + bR = 1 (13)

RI
Vapor phase:

K
bc +S bc c e cS + bc h
= gRc

SC
c S f c cf (14)

Solid phase:

U
K
bR R +S S
bR R e RS + f
bR R Rf h
= gcR (15)
AN
The moisture fraction of drying particles (i , ) was predicted using eq. 16:

bR R i , + S
K
bR R e h , + gh
RS i + bR R Rf i = ∇. bR R $j,R ∇i
, ,
M

S f cR (16)

The momentum conservation equation are as follows:


D

Vapor phase:
TE

bc c kkkkl
c + ∇. bc c kkkkl
c kkkkl
c = −bc ∇8 + ∇. mnc + bc c
kkkkkkl
l+T cR + gRc
h kkkkkkl
Rc (17)
EP

Solid phase:

bR R kkkklR + ∇. bR R kkkklR kkkklR = −bR ∇8 − ∇8R + ∇. mnR + bR R


kkkkkkl
l+T Rc + gcR
h kkkkkkl
cR (18)
C

Energy conservation equation is as eqs. 19 and 20:


AC

Vapor phase:

oC
bc c 2c + ∇. bc c kkkkl2
c c = −bc + mnc : ∇kkkklc + qRc + gRc
h 2%c@ (19)

Solid phase:

oC
bR R 2R + ∇. bR R kkkkl2
R R = −bR + mnR : ∇kkkklR + qcR + gcR
h 2%c@ (20)

27
ACCEPTED MANUSCRIPT
One of the most important issues in superheated drying is distinct different periods during drying;

condensing and heating, constant drying and falling rate stage. These three stages were modeled

individually and three different models were developed for mass transfer in each stage. The heat

transfer coefficient in the first stage was calculated using McAdams equation (eq. 21) and mass

h was calculated using eqs. 22 and 23.


transfer rate between two phases (gcR

[xMyJ XzJy y < { ] |} ~•O| K‚

PT
uvG
ℎs = 1.13 × × • (21)
wG €| wG y < {

ƒGy
ghcR =

RI
(22)
xMyJ XzJy y < {

where:

SC
qRc = ℎs „ − c (23)

U
For the second period, empirical correlations for Nusselt Number was utilized (eq. 24) and the
AN
convective heat transfer rate was calculated (eq. 25).

u•y vG … G
ℎ= wG}
(24)
M

L L
† R = 2.0 + 0.74T;R } 8e O (25)
D

ƒ
ghRc = x yG (26)
MyJ
TE

where:

qcR = ℎ c − R (27)
EP

And finally for the falling rate period:

w"Š
C

uvG G 4M‹‹ "ŒI <"M• 4M‹‹ < ŒI


ghRc = −bR R w = wG}
exp −4‘ = wG}
(28)
AC

w G
qcR = ℎ c − R = bR R @R w + ghRc 2%c@ (29)

The inlet velocity and temperature of superheated steam were fixed. In the first stage of drying the

drying rate was negative because of initial condensation of the steam and the moisture content of

the drying samples increased. In constant drying stage, the simulation over-predicted the drying rate

so the predicted drying time in this period was also shorter. In the falling rate stage the predicted

temperature and drying rate were higher than the experimental results. These authors did not take
28
ACCEPTED MANUSCRIPT
into account evaporation from the material and only considered condensation of steam; so, there

were some variations between experimental and CFD results in the constant rate period in their

study (Xiao, Zhang, Wu, & Liu, 2012).

In another study, heat transfer in a rotary kiln operated with superheated steam was investigated.

Heat transfer model consisted of radiation exchange between superheated steam, refractory wall and

PT
the surface of solids, conductive heat transfer in refractive walls and the mass and energy balance

for drying material and the steam. Gas convection inside the chamber was also considered. A steady

RI
state model was solved and distribution of temperature in dryer walls, solids and steam phase was

SC
found. The CFD results were validated with the literature data of carrot drying in the same type of

dryer and a good agreement was observed (Sinhal, Ghoshdastidar, & Dasgupta, 2012).

U
In a more recent study, CFD modeling was conducted for superheated steam drying of spent grain
AN
pellet. Instead of utilization of classic drying models with transfer coefficients, Reynolds- Averaged

Navier-Stokes (RANS) equations were used. Ramachandran et al. (2017a) reported that mass
M

transfer prediction is complicated because a specified mass boundary condition does not exist and

evaporation is only dominated by heat transfer at the interface. Also, accurate determination of mass
D

transfer coefficient using the analogy between heat and mass transfer is not possible in the case of
TE

superheated steam drying. Superheated steam drying was divided into three stages:
EP

(1) primarily condensation of steam on the sample when it is in contact with superheated steam and

results in the condensation of superheated steam on the surface. In this stage, the temperature of the
C

sample is lower than the steam and the condensed steam would evaporate or absorbed by the sample
AC

depending on the hygroscopic behavior. The film condensation heat transfer has the most important

role in determining the heat flux and the empirical correlation for cylindrical objects was used to

determine heat transfer coefficient,

(2) condensed steam evaporation when the surface temperature of drying material is identical to

superheated steam temperature and evaporation is started and it is governed by heat transfer

coefficient between superheated steam and water on the surface of drying material,

29
ACCEPTED MANUSCRIPT
(3) drying phase when the solid temperature increases and moisture diffusion is the dominant

driving force for drying. Drying can be predicted using sensible heat transfer rate.

The validation of the numerical model with experimental data showed a good agreement with a

mean relative percentage error less than or equal to 10% which shows the capability of the CFD

model to be used for optimization and designing superheated steam dryers (Ramachandran,

PT
Akbarzadeh, Paliwal, & Cenkowski, 2017).

7- Conclusion

RI
As discussed in this article CFD can be used to predict fluid flow, heat and mass transfer, also

SC
product quality attributes in many drying systems even emerging drying technologies such as

microwave assisted, infrared and superheated steam drying. There has been a huge advancement in

U
application of CFD in drying of food and agricultural products but, lack of adequate knowledge
AN
about complex drying process, lack of appropriate information about physicochemical properties of

food material in literature and inefficient researches in industrial scale made CFD approximately
M

useless in industrial and commercial scale. Coupling CFD method with novel modeling techniques

such as reaction engineering approach, discrete element method, drying kinetics models made it
D

possible to use in process optimization regarding energy, time and the most important factor
TE

“quality”. The review shows there is few researches in drying optimization using CFD. Developing
EP

more user-friendly software packages and enhancing the processing capability of computers could

also extend CFD utilization in food drying. The validation problem of CFD results is also exists
C

because of complexity of gaining experimental data in some dryer designs and conditions. To
AC

conclude, in contrast to other fields such as mechanical or aerospace engineering, it should be noted

that, since a food material is a biological object, utilization of computational modeling and

simulation in Food Science and Technology cannot be successful without proper using of physical

experiments determining the biological and physicochemical characteristics of food materials.

Nomenclature
A Surface area (m2)
B Magnetic induction (T)
30
ACCEPTED MANUSCRIPT
Specific heat capacity (W/kgK)
@ Specific heat at constant pressure (J/kgK)
Deff Effective diffusion coefficient (m2/s)
D Electric displacement (N/V m)
E Electric field intensity (V/m)
a%<@ View factor
g Acceleration due to gravity (m/s2)
h Convective heat transfer coefficient (Wm2 K)
H Magnetic field intensity (A/m)
2%c@

PT
Latent heat of evaporation of water (J/kg)
J Current flux (A/m2)
gh Condensation rate or drying rate (kg/m3s)

RI
Nu Nusselt Number
P Pressure (Pa)
8AB Infrared power density (W/kg)
8AB,R

SC
S Initial infrared power (W) absorbed at the surface of the grain
P(x) Amount of power dissipation (W)
P0 Power at the surface (W)
8e Prandtl Number

U
q Electric charge density (C/m3)
Q Convective heat transfer rate (W/m2)
AN
r Radius (m)
R Drag force (N)
T@ Radius of particle (m)
M

T@%U Internal radius of the penetration layer (m)


T; Reynolds Number
! Thermal sink or source (W/m3)
D

t Time (s)
TE

T Temperature (K)
u Velocity component (m/s)
X Composition ratio by weight (%)
x Depth from the samples surface (m)
EP

i, Wet basis moisture content of the solid phase (kg/kg)


C

Greek symbol
b Volume fraction
AC

( Emissivity
(" Dielectric loss factor
(, Dielectric constant
δ Kronecker delta
? Attenuation constant (1/m)
Volume viscosity coefficient (Pa s); thermal conductivity (W/mK)
Dynamic viscosity (kg/ms)
Density (kg/m3)
m Reynolds stress tensor (N/m2)

Subscripts and Superscripts


31
ACCEPTED MANUSCRIPT
b Boil
cr Critical
e Emitter
eq Equilibrium
i,j Cartesian coordinate index
l Liquid
p Particle
s Solid
v Vapor
w Water

PT
x Cartesian coordinates (m)

RI
References
Abdullah, Z., Taip, F. S., Siti Mazlina, M. K., & Abdul Rahman, R. Z. (2017). Fundamental and

SC
empirical modelling of co-current spray drying process - a review.
Aktaş, M., Şevik, S., Amini, A., & Khanlari, A. (2016). Analysis of drying of melon in a solar-heat
recovery assisted infrared dryer. Solar Energy, 137, 500-515.

U
Aktaş, M., Sözen, A., Amini, A., & Khanlari, A. (2017). Experimental analysis and CFD simulation
of infrared apricot dryer with heat recovery. Drying Technology, 35(6), 766-783.
AN
Anandharamakrishnan, C. (2013). Computational fluid dynamics applications in food processing
Computational Fluid Dynamics Applications in Food Processing (pp. 1-9): Springer.
Arballo, J., Campañone, L., & Mascheroni, R. (2012). Modeling of microwave drying of fruits. Part
M

II: Effect of osmotic pretreatment on the microwave dehydration process. Drying


Technology, 30(4), 404-415.
D

Azizi, D., Jafari Seid, M., Mirzaei, H., Dehnad, D., (2017). The Influence of Refractance Window
Drying on Qualitative Properties of Kiwifruit Slices. International Journal of Food
TE

Engineering 13(2), 1-9.


Azmir, J., Hou, Q., & Yu, A. (2018). Discrete particle simulation of food grain drying in a fluidised
bed. Powder Technology, 323, 238-249.
EP

Bahmani, A., Jafari, S. M., Shahidi, S.-A., & Dehnad, D. (2016). Mass Transfer Kinetics of
Eggplant during Osmotic Dehydration by Neural Networks. Journal of Food Processing
and Preservation, 40(5), 815-827.
C

Bardy, E., Hamdi, M., Havet, M., & Rouaud, O. (2015). Transient exergetic efficiency and moisture
loss analysis of forced convection drying with and without electrohydrodynamic
AC

enhancement. Energy, 89, 519-527.


Bhutta, M. M. A., Hayat, N., Bashir, M. H., Khan, A. R., Ahmad, K. N., & Khan, S. (2012). CFD
applications in various heat exchangers design: A review. Applied Thermal Engineering, 32,
1-12.
Bubnovich, V., Quijada, E., & Reyes, A. (2009). Computer Simulation of Atmospheric Freeze
Drying of Carrot Slices in a Fluidized Bed. Numerical Heat Transfer, Part A: Applications,
56(2), 170-191.
Bubnovich, V., Reyes, A., Quijada, E., & Mahn, A. (2012). Numerical simulation of lyophilization
of carrot slices at atmospheric pressure in a fluidized bed. Journal of Food Engineering,
109(4), 659-667.

32
ACCEPTED MANUSCRIPT
Chandramohan, V. (2016). Experimental Analysis and Simultaneous Heat and Moisture Transfer
with Coupled CFD Model for Convective Drying of Moist Object. International Journal for
Computational Methods in Engineering Science and Mechanics, 17(1), 59-71.
Chauhan, P. S., Kumar, A., & Tekasakul, P. (2015). Applications of software in solar drying
systems: A review. Renewable and Sustainable Energy Reviews, 51, 1326-1337.
Cheng, C., Tsai, S.-M., Cheng, H.-P., & Chen, C.-H. (2014). Analysis for heat transfer
enhancement of helical and electrical heating tube heat exchangers in vacuum freeze-drying
plant. International Communications in Heat and Mass Transfer, 58, 111-117.
Cheng, H. P., & Tsai, S. M. (2012). Analysis of CFD heat transfer of vacuum freeze-drying shelf.

PT
HEFAT 2012.
Coletto, M. M. (2015). Atmospheric freeze drying of food in fluidized beds-Practical aspects and
CFD simulation. Politecnico di Torino.

RI
da Silva, F. R. G. B., de Souza, M., da Costa, A. M. d. S., de Matos Jorge, L. M., & Paraíso, P. R.
(2012). Experimental and numerical analysis of soybean meal drying in fluidized bed.

SC
Powder Technology, 229, 61-70.
Defraeye, T. (2014). Advanced computational modelling for drying processes – A review. Applied
Energy, 131(Supplement C), 323-344.

U
Dehnad, D., Jafari, S. M., & Afrasiabi, M. (2016). Influence of drying on functional properties of
food biopolymers: From traditional to novel dehydration techniques. Trends in Food
AN
Science & Technology, 57, Part A, 116-131.
Devahastin, S., & Mujumdar, A. S. (2014). Superheated steam drying of foods and biomaterials.
Modern drying technology, 5, 57-84.
M

Farid, M. M. (2010). Mathematical modeling of food processing: CRC Press.


Feng, H., Yin, Y., & Tang, J. (2012). Microwave drying of food and agricultural materials: basics
D

and heat and mass transfer modeling. Food Engineering Reviews, 4(2), 89-106.
Gulati, T., Zhu, H., Datta, A. K., & Huang, K. (2015). Microwave drying of spheres: Coupled
TE

electromagnetics-multiphase transport modeling with experimentation. Part II: Model


validation and simulation results. Food and Bioproducts Processing, 96, 326-337.
Hashemi Shahraki, M., Jafari , S. M., Mashkour, M., & Esmaeilzadeh, E. (2014). Optimization of
EP

Closed-Cycle Fluidized Bed Drying of Sesame Seeds Using Response Surface Methodology
and Genetic Algorithms International Journal of Food Engineering (Vol. 10, pp. 167).
Hemis, M., Choudhary, R., & Watson, D. G. (2012). A coupled mathematical model for
C

simultaneous microwave and convective drying of wheat seeds. Biosystems Engineering,


112(3), 202-209.
AC

Jafari, S. M., Azizi, D., Mirzaei, H., & Dehnad, D. (2016). Comparing Quality Characteristics of
Oven-Dried and Refractance Window-Dried Kiwifruits. Journal of Food Processing and
Preservation, 40(3), 362-372.
Jafari, S. M., Ganje, M., Dehnad, D., & Ghanbari, V. (2016). Mathematical, Fuzzy Logic and
Artificial Neural Network Modeling Techniques to Predict Drying Kinetics of Onion.
Journal of Food Processing and Preservation, 40(2), 329-339.
Jafari, S. M., Ghalegi Ghalenoei, M., & Dehnad, D. (2017). Influence of spray drying on water
solubility index, apparent density, and anthocyanin content of pomegranate juice powder.
Powder Technology, 311, 59-65.

33
ACCEPTED MANUSCRIPT
Jafari, S. M., Ghanbari, V., Ganje, M., & Dehnad, D. (2016). Modeling the Drying Kinetics of
Green Bell Pepper in a Heat Pump Assisted Fluidized Bed Dryer. Journal of Food Quality,
39(2), 98-108.
Jamaleddine, T. J., & Ray, M. B. (2010). Application of Computational Fluid Dynamics for
Simulation of Drying Processes: A Review. Drying Technology, 28(2), 120-154.
Jaskulski, M., Atuonwu, J. C., Tran, T. T. H., Stapley, A. G. F., & Tsotsas, E. (2017). Predictive
CFD modeling of whey protein denaturation in skim milk spray drying powder production.
Advanced Powder Technology, 28(12), 3140-3147.
Joardder, M. U. H., Kumar, C., & Karim, M. A. (2017). Multiphase transfer model for intermittent

PT
microwave-convective drying of food: Considering shrinkage and pore evolution.
International Journal of Multiphase Flow, 95(Supplement C), 101-119.
Kemp, I. C., & Oakley, D. E. (2002). Modelling of particulate drying in theory and practice. Drying

RI
Technology, 20(9), 1699-1750.
Khan, F. A., Fischer, C., & Straatman, A. G. (2015). Numerical model for non-equilibrium heat and

SC
mass exchange in conjugate fluid/solid/porous domains with application to evaporative
cooling and drying. International Journal of Heat and Mass Transfer, 80, 513-528.
Kumar, C., Joardder, M., Farrell, T., Millar, G. J., & Karim, M. (2016). Mathematical model for

U
intermittent microwave convective drying of food materials. Drying Technology, 34(8), 962-
973.
AN
Kuriakose, R., & Anandharamakrishnan, C. (2010). Computational fluid dynamics (CFD)
applications in spray drying of food products. Trends in Food Science & Technology, 21(8),
383-398.
M

Lemus-Mondaca, R. A., Vega-Gálvez, A., & Moraga, N. O. (2011). Computational simulation and
developments applied to food thermal processing. Food Engineering Reviews, 3(3-4), 121-
135.
D

Lettieri, P., & Mazzei, L. (2009). Challenges and Issues on the CFD Modeling of Fluidized Beds: A
Review. The Journal of Computational Multiphase Flows, 1(2), 83-131.
TE

Li, S., Stawczyk, J., & Zbicinski, I. (2007). CFD Model of Apple Atmospheric Freeze Drying at
Low Temperature. Drying Technology, 25(7-8), 1331-1339.
EP

Lisboa, H. M., Duarte, M. E., & Cavalcanti-Mata, M. E. (2017). Modeling of food drying processes
in industrial spray dryers. Food and Bioproducts Processing.
Lisboa, H. M., Duarte, M. E., & Cavalcanti-Mata, M. E. (2018). Modeling of food drying processes
C

in industrial spray dryers. Food and Bioproducts Processing, 107(Supplement C), 49-60.
Liu, C. M., Wang, Q. Z., & Sakai, N. (2005). Power and temperature distribution during microwave
AC

thawing, simulated by using Maxwell's equations and Lambert's law. International journal
of food science & technology, 40(1), 9-21.
Lo, S. (2005). Application of computational fluid dynamics to spray drying. Le Lait, 85(4-5), 353-
359.
Malekjani, N. (2017). Heat and mass transfer modelling of microwave assisted fluidized bed drying
of hazelnut and evaluation of quality attributes of dried product. (doctoral dissertation),
University of Tehran.
Malekjani, N., Emam-Djomeh, Z., Hashemabadi, S. H., & Askari, G. R. (2018). Modeling Thin
Layer Drying Kinetics, Moisture Diffusivity and Activation Energy of Hazelnuts during
Microwave-Convective Drying. International Journal of Food Engineering.

34
ACCEPTED MANUSCRIPT
Malekjani, N., Jafari , S. M., Rahmati, M. H., Esmaeelzadeh, E., & Mirzaee, H. (2013). Evaluation
of Thin-Layer Drying Models and Artificial Neural Networks for Describing Drying
Kinetics of Canola Seed in a Heat Pump Assisted Fluidized Bed Dryer International Journal
of Food Engineering (Vol. 9, pp. 375).
Mallinson, G. D., & Norris, S. E. (2010). Fundamentals of Computational Fluid Dynamics: Boca
Raton, Florida, CRC Press/Taylor & Francis Group.
Mortier, S., De Beer, T., Gernaey, K. V., Remon, J. P., Vervaet, C., & Nopens, I. (2011).
Mechanistic modelling of fluidized bed drying processes of wet porous granules: A review.
European Journal of Pharmaceutics and Biopharmaceutics, 79(2), 205-225.

PT
Moses, J., Norton, T., Alagusundaram, K., & Tiwari, B. (2014). Novel drying techniques for the
food industry. Food Engineering Reviews, 6(3), 43-55.
Mujumdar, A. S. (2008). Guide to Industrial Drying: Principles, Equipment and New

RI
Developments.
Mujumdar, A. S. (2014). Handbook of industrial drying: CRC press.

SC
Mujumdar, A. S., Huang, L.-X., & Chen, X. D. (2010). An overview of the recent advances in
spray-drying. Dairy Science & Technology, 90(2-3), 211-224.
Nakagawa, K., & Ochiai, T. (2015). A mathematical model of multi-dimensional freeze-drying for

U
food products. Journal of Food Engineering, 161(Supplement C), 55-67.
AN
Nam, J. H., & Song, C. S. (2007). Numerical simulation of conjugate heat and mass transfer during
multi-dimensional freeze drying of slab-shaped food products. International Journal of Heat
and Mass Transfer, 50(23), 4891-4900.
Nazghelichi, T., Jafari, A., Kianmehr, M. H., & Aghbashlo, M. (2013). CFD Simulation and
M

Optimization of Factors Affecting the Performance of a Fluidized Bed Dryer. Iranian


Journal of Chemistry and Chemical Engineering (IJCCE), 32(4), 81-92.
D

Nejadi, J., & Nikbakht, A. M. (2017). Numerical Simulation of Corn Drying in a Hybrid Fluidized
Bed‐Infrared Dryer. Journal of Food Process Engineering, 40(2), e12373.
TE

Norton, T., & Sun, D.-W. (2006). Computational fluid dynamics (CFD)–an effective and efficient
design and analysis tool for the food industry: a review. Trends in Food Science &
Technology, 17(11), 600-620.
EP

Norton, T., & Sun, D.-W. (2010). CFD: an innovative and effective design tool for the food
industry Food Engineering Interfaces (pp. 45-68): Springer.
Norton, T., Sun, D.-W., Grant, J., Fallon, R., & Dodd, V. (2007). Applications of computational
C

fluid dynamics (CFD) in the modelling and design of ventilation systems in the agricultural
industry: A review. Bioresource Technology, 98(12), 2386-2414.
AC

Norton, T., Tiwari, B., & Sun, D.-W. (2013). Computational fluid dynamics in the design and
analysis of thermal processes: A review of recent advances. Critical reviews in food science
and nutrition, 53(3), 251-275.
Orbegoso, E. M., Saavedra, R., Marcelo, D., & La Madrid, R. (2017). Numerical characterisation of
one-step and three-step solar air heating collectors used for cocoa bean solar drying. J
Environ Manage, 203(Pt 3), 1080-1094.
Pawar, S. B., & Pratape, V. M. (2015). Fundamentals of Infrared Heating and Its Application in
Drying of Food Materials: A Review. Journal of Food Process Engineering, 40(1), e12308.
Petitti, M., Barresi, A. A., & Marchisio, D. L. (2013). CFD modelling of condensers for freeze-
drying processes. Sadhana, 38(6), 1219-1239.

35
ACCEPTED MANUSCRIPT
Philippsen, C. G., Vilela, A. C. F., & Zen, L. D. (2015). Fluidized bed modeling applied to the
analysis of processes: review and state of the art. Journal of Materials Research and
Technology, 4(2), 208-216.
Pitchai, K. (2011). Electromagnetic and Heat Transfer Modeling of Microwave Heating in
Domestic Ovens. (Master of Science), University of Nebraska - Lincoln.
Pragati, K., & Sharma, H. (2012). Concept of computational fluid dynamics (CFD) and its
applications in food processing equipment design. Journal of Food Processing and
Technology, 3(1).
Prakash, O., & Kumar, A. (2017). Solar Drying Technology: Concept, Design, Testing, Modeling,

PT
Economics, and Environment: Springer.
Ramachandran, R. P., Akbarzadeh, M., Paliwal, J., & Cenkowski, S. (2017). Three-dimensional
CFD modelling of superheated steam drying of a single distillers’ spent grain pellet. Journal

RI
of Food Engineering, 212, 121-135.
Ranjbaran, M., & Zare, D. (2012). CFD modeling of microwave-assisted fluidized bed drying of

SC
moist particles using two-fluid model. Drying Technology, 30(4), 362-376.
Rattanadecho, P., & Makul, N. (2016). Microwave-assisted drying: A review of the state-of-the-art.
Drying Technology, 34(1), 1-38.

U
Riadh, M. H., Ahmad, S. A. B., Marhaban, M. H., & Soh, A. C. (2015). Infrared heating in food
drying: An overview. Drying Technology, 33(3), 322-335.
AN
Romdhana, H., Bonazzi, C., & Esteban-Decloux, M. (2015). Superheated Steam Drying: An
Overview of Pilot and Industrial Dryers with a Focus on Energy Efficiency. Drying
Technology, 33(10), 1255-1274.
M

Romero, V., Cerezo, E., Garcia, M., & Sanchez, M. (2014). Simulation and validation of vanilla
drying process in an indirect solar dryer prototype using CFD Fluent program. Energy
Procedia, 57, 1651-1658.
D

Sanghi, A., Ambrose, R. K., & Maier, D. (2017). CFD simulation of corn drying in a natural
TE

convection solar dryer. Drying Technology, 1-12.


Schmitz-Schug, I., Kulozik, U., & Foerst, P. (2016). Modeling spray drying of dairy products –
Impact of drying kinetics, reaction kinetics and spray drying conditions on lysine loss.
EP

Chemical Engineering Science, 141, 315-329.


Schubert, H., & Regie, M. (2006). Novel and Traditional Microwave Applications in the Food
Industry. In M. Willert-Porada (Ed.), Advances in Microwave and Radio Frequency
C

Processing: Report from the 8th International Conference on Microwave and High
Frequency Heating held in Bayreuth, Germany, September 3–7, 2001 (pp. 259-270). Berlin,
AC

Heidelberg: Springer Berlin Heidelberg.


Sinhal, K., Ghoshdastidar, P., & Dasgupta, B. (2012). Computer Simulation of Drying of Food
Products With Superheated Steam in a Rotary Kiln. Journal of Thermal Science and
Engineering Applications, 4(1), 011009.
Sobieski, W. (2008). Numerical analysis of sensitivity of Eulerian multiphase model for a spouted-
bed grain dryer. Drying Technology, 26(12), 1438-1456.
Sobieski, W. (2010). Selected aspects of developing a simulation model of a spouted bed grain
dryer based on the Eulerian multiphase model. Drying Technology, 28(12), 1331-1343.
Sosa-Morales, M., Valerio-Junco, L., López-Malo, A., & García, H. (2010). Dielectric properties of
foods: Reported data in the 21st Century and their potential applications. LWT-Food Science
and Technology, 43(8), 1169-1179.
36
ACCEPTED MANUSCRIPT
Souraki, B. A., & Mowla, D. (2008). Experimental and theoretical investigation of drying behaviour
of garlic in an inert medium fluidized bed assisted by microwave. Journal of Food
Engineering, 88(4), 438-449.
Strumiłło, C. (2006). Perspectives on developments in drying. Drying Technology, 24(9), 1059-
1068.
Szafran, R. G., & Kmiec, A. (2004). CFD modeling of heat and mass transfer in a spouted bed
dryer. Industrial & engineering chemistry research, 43(4), 1113-1124.
Szafran, R. G., & Kmiec, A. (2005). Point-by-Point Solution Procedure for the Computational Fluid
Dynamics Modeling of Long-Time Batch Drying. Industrial & engineering chemistry

PT
research, 44(20), 7892-7898.
Tesfamariam, D. A., Bayray, M., Tesfay, M., & Hagos, F. Y. (2015). Modeling and Experiment of
solar crop dryer for rural application. Journal of Chemical and Pharmaceutical Sciences

RI
ISSN, 974, 2115.
Tsotsas, E., & Mujumdar, A. S. (2011). Modern Drying Technology, Volume 3: Product Quality

SC
and Formulation (Vol. 1): John Wiley & Sons.
Tzempelikos, D. A., Mitrakos, D., Vouros, A. P., Bardakas, A. V., Filios, A. E., & Margaris, D. P.
(2015). Numerical modeling of heat and mass transfer during convective drying of

U
cylindrical quince slices. Journal of Food Engineering, 156(Supplement C), 10-21.
Wang, L., & Sun, D.-W. (2003). Recent developments in numerical modelling of heating and
AN
cooling processes in the food industry—a review. Trends in Food Science & Technology,
14(10), 408-423.
Wu, J., Zhang, H., & Li, F. (2017). A study on drying models and internal stresses of the rice kernel
M

during infrared drying. Drying Technology, 35(6), 680-688.


Xia, B., & Sun, D.-W. (2002). Applications of computational fluid dynamics (CFD) in the food
industry: a review. Computers and Electronics in Agriculture, 34(1), 5-24.
D

Xiao, Z., Zhang, F., Wu, N., & Liu, X. (2012). CFD modeling and simulation of superheated steam
TE

fluidized bed drying process. Paper presented at the International Conference on Computer
and Computing Technologies in Agriculture.
Yang, H., & Gunasekaran, S. (2004). Comparison of temperature distribution in model food
EP

cylinders based on Maxwell's equations and Lambert's law during pulsed microwave
heating. Journal of Food Engineering, 64(4), 445-453.
Yanniotis, S. (2007). Heat Transfer in Food Processing: Recent Developments and Applications
C

(Vol. 21): WIT Press.


Zare, D., & Ranjbaran, M. (2012). Simulation and validation of microwave-assisted fluidized bed
AC

drying of soybeans. Drying Technology, 30(3), 236-247.


Zhu, H., Gulati, T., Datta, A. K., & Huang, K. (2015). Microwave drying of spheres: Coupled
electromagnetics-multiphase transport modeling with experimentation. Part I: Model
development and experimental methodology. Food and Bioproducts Processing, 96, 314-
325.

37
ACCEPTED MANUSCRIPT
Table 1. Comparison between Reynolds averaged Navier- Stokes (RANs) models (Farid, 2010; Kuriakose &
Anandharamakrishnan, 2010; Norton & Sun, 2006, 2010; Norton, Tiwari, & Sun, 2013)
Model Description Advantages Disadvantages
Good convergence

Basics of this model are transport Robust


Inadequate in some cases due
equations accounting for , Economical to assumptions and empiricism
Standard k-ε turbulence kinetic energy and , which model is based on (the
turbulence dissipation rate Reasonably accurate most important assumption
made in this model is assuming
Satisfactory results in processes

PT
an equilibrium condition for
without adverse pressure gradient
turbulence)
Suitable for external flow around
Not highly accurate in complex
complex geometries

RI
flow regimes, geometry and
Frequently used in drying severe pressure gradients
processes with high Reynolds
number

SC
Similar to Standard k-ε but includes Independency from empiricism and Some limitations because of
some additional expressions for taking anisotropy of complex flows assuming eddy viscosity as
RNG k-ε development of dissipation rates and being isotropic
into account

U
some constants different from
Better results for recirculating Require high power computers
Standard k-ε model constants.
flows
AN
Some convergence difficulties

Instead of using constant Empirical


turbulence model constant (like
Standard k-ε model), this term is a
M

function of turbulence properties Same benefits as RNG but works


Realizable k- Some limitations because of
and mean better in more complex flows such
ε assuming eddy viscosity as
Flow. So, it satisfies certain as jet impingement
being isotropic
D

mathematical constraints
on the Reynolds stress tensor that
are consistent with the physics of
TE

turbulent flows. A new model for


the rate of dissipation is also used.
Convergence is more difficult than
Basics of this model are transport k-ε
EP

equations accounting for ,


Works better for boundary layers
Standard turbulence kinetic energy and
under adverse pressure gradients
− model specific dissipation rate ( rate of
dissipation per unit turbulent Suitable for flows with convoluted
C

kinetic) curvatures and internal flows


Frequently used in drying
AC

processes

This model includes transport Most complete model physically Long computational time and
Reynolds equations for the Reynolds stresses, powerful memory
Suitable in the case of adverse
Stress Model a transport equation for the
pressure gradient involved Poor convergence was reported
(RSM) turbulence energy dissipation rate
and three transport equations for the Suitable for complex flow behavior in many literatures
turbulent fluxes of each scalar
property.

Table 2. Advantages and disadvantages of CFD modeling in food drying processes


1
Advantages
ACCEPTED MANUSCRIPT
Disadvantages
Providing detailed understanding of heat, mass and Very small time steps are needed in some cases
momentum transfer in the drying system because of different time scale of fluid flow, heat,
mass and scalar transport resulting in long
computation times
Declining scale up problems for drying systems Incapability of online controlling the thermal
processes
Working as virtual sensors in drying systems to Lack of adequate data about physicochemical
improve final product quality properties of food materials during drying

PT
Simulating unusual conditions such as hot Shrinkage during drying which makes mesh
temperatures or dangerous environment generation more complex

RI
U SC
AN
M
D
TE
C EP
AC

Table 3. Some of the CFD studies in the field of fluidized and spouted bed drying of food and agricultural
products
2
Case description
ACCEPTED Software
Modeling methods
MANUSCRIPT Findings Authors

Spouted bed drying Eulerian multiphase model was Fluent - Importance of inlet air (Sobieski,
of grains at used to describe the velocity 2008)
different gas inlet hydrodynamics of the bed and - Importance of selecting
velocities, grain grain height in the bed with suitable turbulence models
densities and different model parameters - Good agreement between
diameters (turbulent models, initial and experimental and simulation
maximum coefficient of results
packing)
Spouted bed drying Various important features of Fluent - Every factor has its own (Sobieski,

PT
simulation such as selecting effect on simulation results 2010)
between 2D and 3D space, so, the function and practical
geometry, structured or implication should be
unstructured grid, configuration examined carefully to

RI
of multiphase model, the improve computational
turbulence model and the results
sensitivity of the model to phase

SC
and flow parameter changes.

Fluidized bed Eulerian- Eulerian approach MFIX - Drying conducted in falling (da Silva, de
drying of soybean with incorporation of mass rate period and diffusion is Souza, da
meal at different transfer model the most important Costa, de

U
temperatures, controlling parameter. Matos Jorge,
velocities and bed Symalal O’Brien drag model - Temperature had a & Paraíso,
was selected as the best model
AN
heights significant effect on drying 2012)
but, the effects of velocity
and bed height were not
significant
M

Circulating An Eulerian model with kinetic Ansys - Pressure drop declined along (Priya,
fluidized bed drying theory of granular flow Fluent the higher length Pradeep, &
of some seeds were - Deviation between the Saravanan,
studied for experimental and predicted 2017)
D

evaluation of the data were satisfactory and in


influence of axial the range of 7- 9.5 %.
pressure variations
TE

with gas flow rate


and solid
circulation rate.
C EP
AC

3
ACCEPTED MANUSCRIPT

Table 4. Recent CFD studies in the field of novel drying technologies for the food and agricultural products
Type of model Geometry and Discretization Physical properties Boundary condition Mesh type Other considerations
Case description Modeling methods dimensions method Software Findings Authors

Microwave assisted drying


• Two-fluid Eulerian-Eulerian • Conjugated • 2D axis-symmetry FVM Dielectric constant Velocity inlet Structured tetrahedral • Neglecting temperature
CFD simulation of model and kinetic theory of geometry (half of a and dielectric loss Pressure outlet mesh gradient inside the • Small deviations of
spherical moist
granular flow was used cylindrical column factor as a function particles and shrinkage • GAMBIT 6.3.26 simulation results from the
particles (soybeans)
• UDF for calculation of representing the drying of temperature and • FLUENT 6.3.26 experimental data could be
in a microwave chamber was selected moisture content; related to laboratory
microwave power
assisted fluidized bed measurement errors
• Calorimetric method for initial as computational thermal conductivity (Ranjbaran & Zare, 2012)
dryer in order to domain) of drying material as occurred while
microwave power density

PT
study the behavior of a function of determining the initial
gas-solid flow determination and a modified
moisture content; power density of
equation
thermodynamic microwave, air
temperature, velocity and
• standard − model with properties of air as a
relative humidity
some modifications was used to function of

RI
model turbulence temperature

• Maxwell’s electromagnetic • Conjugated • 3D geometry • FEM • Dielectric • Surrounding air • Tetrahedral mesh Shrinkage was not
Microwave drying of • COMSOL • Intermediate sized
field equations and porous media (microwave cavity and constant, dielectric temperature and elements considered

SC
spherically shaped Multiphysics Maxwell’s spheres showed more
models were solved spherical sample with loss factor , specific relative humidity was
food products equations were solved microwave absorption and
• A source term was added to the diameters of 0.06, heat and thermal used as boundary
exploded during drying
by the RF module, mass (Zhu, Gulati, Datta, &
energy equation in order to 0.03, 0.012 m) conductivity as a condition for
function of transport models. transfer by Transport of • Uniform and low Huang, 2015)
represent the effect of the
Diluted Species and temperature drying in
microwave from the temperature • Forced convection
Transport of smaller size particles

U
electromagnetic model. was considered as
Concentrated module, • Capillary diffusion
• Darcy’s law was used boundary condition
heat transfer by Heat dominates in smaller sized
assuming flow in a porous media for energy equation
transfer in Fluids samples
containing lower permeability

AN
module and momentum
instead of solving standard equation by Darcy’s
Navier-Stocks equations
law

M
Intermittent • Lambert’s law was used to • Non- conjugated • 2D axis-symmetry • FEM • The dialectic • Forced convection • Triangular mesh - • Arbitrary Lagrangian- • Intermittency of
microwave calculate microwave energy geometry for a properties were and evaporation at with a homogenous Eulerian framework for microwave result in
convective drying generation within the drying cylindrical sample with considered as a sample surface element size moving mesh and Non- uniform moisture and
considering shrinkage products. diameter of 40 mm and function of moisture linear Solid Mechanics temperature distribution
• A good agreement

D
and porosity was thickness of 10 mm content. module for large
(Joardder, Kumar, & Karim,
compared with the • Thermo-physical deformation were used reported between the
2017)
model without properties as a in.COMSOL experimental and
shrinkage for apple function of liquid • Mass, energy and simulation data.

TE
slice drying and gas volume momentum governing • Taking the shrinkage
fraction in the equations for liquid into account helped to
sample water and vapor and air achieve more realistic
inside the sample were explanation of heat and
solved using Transport mass transfer.
EP
of Diluted Species, Heat
Transfer in Fluids and
Darcy’s Law modules,
respectively.
Infrared drying
C

• Non- conjugated • One-eighth of the • FEM • Thermal • Convection and • Tetrahedral mesh • Rice kernels were
The influence of • Fick’s second law of diffusion • COMSOL • Both models could
kernel geometry conductivity, evaporation at the elements assumed to be composed
penetration depth of and heat transfer equation was Multiphysics accurately predict drying
(ellipsoid) was selected density and specific surface as boundary of three different parts;
AC

infrared radiation on solved. curves and temperature


heat as a function of condition for energy husk, bran and the
temperature and • The radiation source term in during drying (Wu, Zhang, & Li, 2017)
moisture content equation and endosperm.
moisture content of heat transfer equation was • maximum predicted
equilibrium between
rice kernels during assumed to exist in the model temperature difference
moisture transferred
drying with the penetration depth of was 1.5 ℃ between two
to the surface and the
infinity and was neglected in the models
water which leaves
model with zero penetration
the samples through
depth and the radiation was
surrounding drying
treated as a boundary condition.
air

4
ACCEPTED MANUSCRIPT

• Drying simulation was • Conjugated • 3D geometry of solar • FVM • Effective moisture • Heat flux to the • Tetrahedral mesh • The flow was assumed
conducted at surface collector (80*50*11 diffusivity and mass horizontal part of the elements steady state
temperatures of 50 and 60℃ and cm) and drying transfer coefficient collector and
the air velocity of 0.5 m/s. chamber (114*64*37 of food sample were convection heat in
• Steady state condition cm) calculated using the contact region of
• Drying air was considered as CFD ambient air and
an ideal gas mixture. insulated walls
• Energy equation, −
A novel type of standard wall functions • The performance of
• ANSYS FLUENT,
infrared dryer turbulence model and Mixture solar-infrared dryer has
DESIGN MODELER
combined with a multiphase model was selected been better with the
and MESH programs (Aktaş, Şevik, Amini, &
solar collector and a and both air the solar collector utilization of the solar
were used for solving Khanlari, 2016)
heat recovery unit in and drying chamber was energy and heat recovery

PT
the models, generating
order to decrease the simulated units.
the geometry and the
high amount of • Fluids in the system were the • The experimental and
mesh, respectively.
energy required for removed vapor from the melon theoretical findings were
infrared drying of slices and drying air. Air mass in a good agreement
melon slice flow in dryer inlet, the rate of

RI
heat radiation and convection
heat transfer were introduced as
boundary conditions.
• Velocity contours and
temperature distribution profiles

SC
were determined using CFD
results

• Validation experiments were • Conjugated • 3D geometry • FVM • Heat and mass • Vapor injection rate • Tetrahedral mesh • One of the main
conducted at air velocity levels • Six infrared lamps transfer coefficients was applied to drying (mesh density near the assumptions in this study • ANSYS FLUENT • CFD helped obtaining
Velocity and
of 0.5 and 0.25 m/s and fixed and a tray (40*80 cm) were calculated sample surfaces as a IR lamps and apricot was neglecting the sample (VOF multiphase uniform air velocity and (Aktaş, Sözen, Amini, &

U
temperature
apricot surface temperatures of containing half apricots based on the results boundary condition was higher than other shrinkage. model), DESIGN optimization of drying Khanlari, 2017)
distribution and water
60 and 65℃. were modeled. to avoid simulating vacant positions in • Steady state flow MODELER and MESH efficiency
volume fraction in
• Calculating Reynold’s Number • Drying chamber the latent heat drying chamber) • k-ɛ turbulent model was software were used to

AN
the chamber and near
revealed laminar flow inside the (114×64×64 cm) calculations during applied solve the equations,
apricots were
chamber so the Nusselt Number diffusion process creating geometry and
determined in an
and heat transfer coefficient mesh.
infrared dryer for
were easily calculated using
apricots
relevant relationships.
Diffusivity of vapor into air was

M
also calculated using the analogy
between heat and mass transfer.
• − turbulence model was
used

D
Superheated steam drying
• Conjugated • Axisymmetric • FVM - - • Not specified • Drying particle size was • FLUENT
• These authors did not

TE
• Two-phase flow equations for cylindrical drying considered constant during
Superheated steam take into account
an axis-symmetrical cylindrical chamber (120*250 drying
fluidized bed drying evaporation from the
drying chamber was solved mm)
process for rapeseeds material and only
• Heat and mass transfer and
considered condensation
drag models were incorporated (Xiao, Zhang, Wu, & Liu,
of steam; so, there were
into the software using a UDF. 2012)
EP
some variations between
• The inlet velocity and experimental and CFD
temperature of superheated
results in the constant rate
steam were fixed.
period in their study

• Conjugated • 3D geometry ( a • FVM • Thermo-physical • Varying boundary • Tetrahedral mesh


C

Superheated steam • Reynolds- Averaged Navier- - • SIMPLEC algorithm • Good agreement with a
drying to study Stocks (RANS) equations were quarter of a pellet properties and condition according with increased density in ANSYS CFX mean relative percentage (Ramachandran,
temporal and spatial used model (12.7*35.5 mm) and moisture diffusivity to the stages of near the walls program error less than or equal to Akbarzadeh, Paliwal, &
drying chamber as a function of drying
AC

temperature and • − turbulence model was 10% Cenkowski, 2017)


moisture changes of a used (272*356 mm) ) temperature and •
single pellet moisture content

5
ACCEPTED MANUSCRIPT

Pre- processing Processing Post- processing

PT
• Problem thinking • Discretization of the • Numerical analysis

RI
• Mesh generation mathematical of the results
• Choosing the equations • Visualization of the
computational model • Applying the results

SC
boundary conditions
• Continuing iteration
until reaching
convergence

U
AN
Fig. 1. Main scheme of CFD analysis
M
D
TE
C EP
AC

1
ACCEPTED MANUSCRIPT

2- Defining initial and boundary


1- Defining the physical problem
conditions

3- Choosing between eulerian-eulerian


4- Considering governing

PT
(EE), eulerian-lagrangian (EL) and volume
equations of conservation called
of fluid (VOF) reference frameworks in
navier-stocks equations
multiphase systems

RI
SC
5- Implementation of additional transport
equations for turbulence, granular flow, 6- Selecting between 2D or 3D
solid settling or granular temperature in dimensions. Reducing the dimension (if
some cases; see Norton et al. (2013) for possible) can reduce the complexity of

U
more details about governing equations in the model.
thermal processing.
AN
M

8- Discretization of the partial differential


7- Dividing computational domain into equations using finite element (FE), finite
cells called “mesh”. difference (FD) or finite volume (FV)
D

methods
TE

9- Solving the resulting algebraic equations


10- Checking the quality of the mesh
for each mesh by means of linear equation
EP

and mesh independency of the


solvers including gaussian elimination or
results.
gauss- seidel method.
C
AC

11- Checking the residual errors existing between consecutive


iterations and if it is high, modification of the model input or mesh
refinement is required until reaching convergence

Fig. 2. Fundamentals of CFD analysis

2
ACCEPTED MANUSCRIPT

PT
RI
SC
Fig. 3. (a) Eulerian–Eulerian, (b) Eulerian–Lagrangian approach illustration

U
AN
M
D
TE
C EP
AC

3
ACCEPTED MANUSCRIPT

Determinant factor
for classification of
food dryers

Heat transfer Method of solid


Drying strategy Heat input mode
medium handling

PT
RI
SC
- Vacuum drying
- Spray drying - Hot air drying - Fixed bed drying
- Conductive

U
- Freeze drying - Superheated steam - Fluidized bed
drying - Convective
- Intermittent drying drying
AN
- Tray drying - Radiative
- Pulse combustion - Solar drying
- Drum drying - Microwave
drying - Inert medium
drying - Spouted bed - Ultrasound
- Impinging stream
drying - Supercritical fluid drying' - High electric field
M

- Cyclic pressure drying - Jet- zone drying - Infrared


drying - Vibrating bed
drying
D
TE
EP

Fig. 4. Different aspects for classification of food dryers


C
AC

4
ACCEPTED MANUSCRIPT
Highlights

- A comprehensive knowledge of heat, mass and momentum transfer is critical in CFD.

- CFD modeling could be useful to enhance drying process design and development.

- CFD could enhance quality by correlating the results to degradation kinetics.

- Developing novel drying technologies with the aid of CFD is possible.

PT
RI
U SC
AN
M
D
TE
C EP
AC

You might also like