Journal of Food Engineering 156 (2015) 10–21

Contents lists available at ScienceDirect

Journal of Food Engineering

journal homepage: www.elsevier.com/locate/jfoodeng

Numerical modeling of heat and mass transfer during convective drying

of cylindrical quince slices
Dimitrios A. Tzempelikos a,⇑, Dimitris Mitrakos b, Alexandros P. Vouros c, Achilleas V. Bardakas d,
Andronikos E. Filios e, Dionissios P. Margaris a
a
Laboratory of Fluid Mechanics, Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
b
Greek Atomic Energy Commission, Greece
c
Laboratory of Fluid Mechanics and Turbomachinery, Department of Mechanical Engineering Educators, School of Pedagogical and Technological Education (ASPETE), Greece
d
Department of Mechanical Engineering Technology, Purdue University, USA
e
Department of Mechanical Engineering, Technological Education Institute of Piraeus, Greece

a r t i c l e i n f o a b s t r a c t

Article history: A numerical model for non-steady heat and mass transfer during convective drying of cylindrical quince
Received 28 November 2014 slices, with axis parallel to the air flow, is developed. The model is based on the numerical solution of the
Received in revised form 20 January 2015 coupled one-dimensional heat and mass transport equations, assuming moisture transport due to Fick’s
Accepted 25 January 2015
diffusion, with an effective moisture diffusion coefficient derived by fitting the analytical solution of the
Available online 7 February 2015
Fick’s law to experimentally derived drying curves, on the basis of an Arrhenius-type temperature depen-
dence. The necessary convective heat and mass transfer coefficients are obtained from CFD calculations of
Keywords:
the turbulent flow field around the slices using a commercial CFD package. A new correlation of the Nus-
CFD
Heat and mass transfer coefficients
selt number, as a function of Prandtl and Reynolds numbers is proposed for the specific geometric flow
Effective moisture diffusivity configuration. The model is validated against experimental data for different air stream velocities (1 and
Quince drying 2 m/s) and temperatures (40, 50 and 60 °C). The model was found to be robust, computationally efficient
and able to capture with sufficient accuracy the time evolution of the temperature and the moisture loss,
with a minimum need for experimental adjustment, and hence, is considered suitable from an engineer-
ing point of view.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction material and the external drying conditions, such as the tem-
perature, the velocity and the relative humidity of the hot air stream.
Several methods are employed to remove the moisture from A considerable number of experiments have been devoted to
organic or non-organic materials, in a variety of industrial applica- convective drying of agricultural products. Experiments have been
tions. In food industry, drying methods include solar, freeze, infra- mainly used to assess the drying rates and the total time needed for
red, microwave and convective drying. Among the various the removal of the moisture from products, for a range of drying
methods, convective drying is one of the most popular, used exten- conditions (Velić et al., 2004; Sacilik et al., 2006; Seiiedlou et al.,
sively to dehydrate foods for preservation, transportation and stor- 2010; Zlatanović et al., 2013; Babalis et al., 2004; Margaris et al.,
age purposes (Mujumdar, 2006). In convective drying, the required 2007; Doymaz, 2009; Erbay et al., 2010). Measurements of the spa-
heat is provided to the food by a hot air stream. Temperature tial variation of temperature (Wang et al., 1995), as well as of the
increase causes diffusion coefficient to get higher values resulting moisture content (Wang et al., 1995; Aregawi et al., 2013) have
to the increase of the drying rate. Drying process is governed by been also presented, providing further insight of the dominant phy-
the strong coupling between heat and mass transport, both inside sical mechanisms and detailed benchmark basis for the evaluation
the material and at the interface with the drying air, and is affected of the proposed mathematical drying models. Considering, in par-
by a number of parameters, including the properties of the food ticular, the spatial distribution of moisture inside the material,
the corresponding measurement is difficult to be obtained since it
often involves advanced radiography or tomography methods.
⇑ Corresponding author at: University of Patras, Department Mechanical Engi- These methods require specialized equipment and resources, which
neering and Aeronautics, Laboratory of Fluid Mechanics, University Campus, Rio can be very difficult even to be supplied or accessed, such as for
26504, Greece. Tel./fax: +30 2610997202.
E-mail address: [email protected] (D.A. Tzempelikos).

http://dx.doi.org/10.1016/j.jfoodeng.2015.01.017
0260-8774/Ó 2015 Elsevier Ltd. All rights reserved.
 D.A. Tzempelikos et al. / Journal of Food Engineering 156 (2015) 10–21 11

Nomenclature

A total area of product (m2) RH relative humidity of drying air (%)

C concentration of water vapor (kg/m3) t drying time (s)
c constant in Nusselt equation T absolute temperature (K)
cp product specific heat (J/kg K) u velocity (m/s)
d product diameter (m) V product volume (m3)
Deff effective moisture diffusivity (m2/s) X moisture content in dry basis (kg water/kg dry product)
D0 pre-exponential factor of the Arrhenius equation (m2/s) z axial coordinate (m)
Dva diffusivity of water vapor in air (m2/s)
DR drying rate (kg water/kg dry product s) Greek symbols
e exponent in Nusselt equation a thermal diffusivity (m2/s)
Ea activation energy (kJ/mol) q density (kg/m3)
h heat transfer coefficient (W/m2 K) qb bulk density of product = total mass/total volume
hm mass transfer coefficient (m/s) (kg/m3)
~
h weighted average heat transfer coefficient (W/m2 K) qs density of solid product = dry mass/total volume
~m
h weighted average mass transfer coefficient (m/s) (kg/m3)
hfg water latent heat of evaporation (J/kg)
k thermal conductivity (W/m K) Subscripts
L product thickness (m)
air drying air
Le Lewis number b back face
M moisture content in wet basis (kg water/kg product) d dry
n normal vector at the surface in initial
Nu Nusselt number
f front face
p pressure (Pa) fn final
pvs saturated water vapor pressure at surface (Pa) p product
Pr Prandtl number ps product surface
r radial coordinate (m)
s surface
R2 coefficient of determination 1 bulk
Re Reynolds number
Rg gas constant, (kJ/mol K)

example the neutron sources required for neutron radiography of agricultural products using non-conjugated heat and mass
(Aregawi et al., 2013). transfer are summarized in Table 1. In non-conjugated models,
Mathematical modeling is a very important tool, as it con- the coupling of the heat equation with the equation of moisture
tributes to the better understanding of heat and mass transport, diffusion is often resolved (see the last column of Table 1), by intro-
while, when it is simply formulated, it is also a very useful tool ducing an additional term at the boundary conditions of the heat
in the improvement of the design and the control of the drying pro- equation to account for the heat consumed for water evaporation;
cess in the food industry. Two main categories of models have been an approach rather reasonable and preferable from physical con-
developed, with regard to the major transport mechanisms: in the siderations point of view (Datta, 2007a, 2007b; Zhang and Datta,
first, convection by Darcy flow and diffusion is considered, 2004). In these models, the heat and mass transfer at the air-pro-
accounting for capillary and pressure driven water flow and diffu- duct interface is simplified by applying the corresponding convec-
sion inside the drying material. In the second, Fick’s diffusion tive transfer coefficients (CTC) at the boundary conditions. Several
accounts for the water transport inside the product, by adopting approaches have been proposed in the literature for the derivation
an effective diffusion coefficient that is often obtained experimen- of the CTC involving: (i) constant values or correlations for simpli-
tally in a semi-empirical manner. An extensive analysis of the dif- fied flow configurations such as flow around plates, and cylinders,
ferent mathematical approaches used for drying modeling can be (ii) time-dependent coefficients based on a zero-dimensional heat
found in the works of Datta (2007a), Zhang and Datta (2004) and and mass balance, (iii) optimization and regression analysis of
Defraeye (2014). experimental and numerical data, and (iv) flow field simulations
Diffusion-based models can be further distinguished in two using computational fluid dynamics (CFD).
main categories, namely, conjugated and non-conjugated models The above approaches have been adopted in multi or one-di-
(Lemus-Mondaca et al., 2011; Defraeye, 2014). In conjugated mod- mensional models. In the multi-dimensional models of Hussain
els, the fully coupled unsteady transport equations for both the and Dincer (2003) and Oztop et al. (2008), constant, empirical val-
drying air and the product are solved (De Bonis et al., 2008; ues have been used. Heat transfer coefficient obtained by semi-em-
Curcio et al., 2008; Lamnatou et al., 2009, 2010; Marra et al., pirical equations for flow over flat plate in combination with mass
2010; Curcio, 2010; Halder et al., 2012; Sabarez, 2012; Kurnia transfer coefficient calculated on the basis of the heat and mass
et al., 2013). These models offer high accuracy and extended appli- transfer analogy (Chilton et al., 1934; Incropera et al., 2012) have
cability; however, the complexity of the non-linear mathematical been adopted in the model of Aversa et al. (2007). Lemus-
models makes the solution procedure complicated and computa- Mondaca et al. (2013) and Villa-Corrales et al. (2010) incorporated
tionally very demanding. In non-conjugated models, the calcula- time-averaged heat and mass transfer coefficients calculated by
tions of the heat and mass transfer inside the product are the solution of zero-dimensional heat and mass balance. CTC
performed independently from the flow field calculations. The gen- derived on the basis of CFD simulations have been used in the
eral aspects of the various models proposed for convective drying works of Kaya et al. (2006, 2007b, 2008), Chandra Mohan et al.
12 D.A. Tzempelikos et al. / Journal of Food Engineering 156 (2015) 10–21

Table 1
Summary of studies on convective drying of agricultural products using non-conjugated approach.

Authors Product Dimension HTCa MTCb Evaporation term in heat

equation
Ateeque et al. (2014) Rectangular potato 3D CFD-Tue HMTAc Yes
Lemus-Mondaca et al. Rectangular papaya 3D TDCd TDCd Yes
(2013)
Chandra Mohan and Rectangular apple 3D CFD-Laf HMTAc No
Talukdar (2010)
Esfahani et al. (2014) Rectangular apple 2D CFD-Laf HMTAc No
Villa-Corrales et al. (2010) Rectangular mango 2D TDCd TDCd Yes
Kaya et al. (2008) Rectangular Kiwi 2D CFD-Laf HMTAc No
Oztop and Akpinar (2008) Rectangular apple & 2D Constant Akpinar and Dincer (2005) Constant Akpinar et al. No
potato (2005)
Kaya et al. (2007b) Cylindrical broccoli & 2D CFD-Laf HMTAc No
banana
Aversa et al. (2007) Rectangular carrot 2D Empirical Perry et al. (1997) and Bird HMTAc Yes
et al. (1960)
Kaya et al. (2006) Rectangular apple 2D CFD-Laf HMTAc No
Hussain and Dincer (2003) Rectangular apple 2D Constant Dincer (1997) Constant Dincer (1997) No
Da Silva et al. (2014) Cylindrical banana 1D ORAg ORAg Yes
Barati and Esfahani (2012) Cylindrical carrot 1D Empirical Perry et al. (1997) HMTAc Yes
Da Silva et al. (2012) Cylindrical banana 1D – ORAg No
Barati and Esfahani (2011) Cylindrical mango 1D Empirical Perry et al. (1997) Empirical Janjai et al. Yes
(2008)
Ruiz-López et al. (2004) Cylindrical carrot 1D Constant Constant Yes
Wang and Brennan (1995) Rectangular potato 1D Empirical Earle (1983) Empirical Burnnett et al. Yes
(1962)
Maroulis et al. (1995) Potato 1D ORAg ORAg Yes
Present study Cylindrical quince 1D CFD-Tue HMTAc Yes
a
HTC: heat transfer coefficient.
b
MTC: mass transfer coefficient.
c
HMTA: heat and mass transfer analogy.
d
TDC: time-dependent coefficient.
e
CFD-Tu: CFD-turbulent flow.
f
CFD-La: CFD-laminar flow.
g
ORA: optimization and regression analysis.

(2010) and Esfahani et al. (2014). Considering, in particular the CFD significant contribution of the present work is the estimation of the
simulations, the air drying stream was considered laminar, except heat and mass transfer coefficients for the present flow configura-
in the recent model of Ateeque et al. (2014) where the SST k  x tion, which is scarcely considered in the literature. The convective
turbulence model (Menter, 1994) was used to analyze the flow heat transfer coefficient is obtained from steady-state CFD calcula-
field around the product. tions of the turbulent flow field around the slice, incorporating the
Regarding one-dimensional models, CTC related to the thermal SST-kx turbulence model. A new correlation for the heat transfer
and mass Biot numbers (Ratti et al., 1995; Cordova-Quiroz et al., coefficient, as a function of Prandtl and Reynolds numbers is pro-
1996), were used by Ruiz-López et al. (2004). In the work of posed for flow parallel to the axis of a cylindrical slice while the con-
Wang and Brennan (1995) semi-empirical correlation of the Nus- vective mass transfer coefficient is then derived using the heat and
selt number were used. The mass transfer coefficient was taken mass transfer analogy. New experimental data including the time
as a function of the Reynolds number and the water concentration variation of slice dimensions are presented. Experiments are per-
at the surface. Barati et al. (2011, 2012) used also a correlation for formed in a laboratory scale convective dryer (Tzempelikos et al.,
the heat transfer coefficient, while the mass transfer coefficient 2012, 2013) for a range of air stream velocities (1 and 2 m/s) and
was derived from heat and mass transfer analogy. Heat and mass temperatures (40, 50 and 60 °C). Numerical results are presented
transfer coefficients derived from optimization and regression ana- and compared with measurements and the drying heat and mass
lysis using numerical and experimental data have been also pre- transfer mechanisms are discussed in relation to theoretical or
sented (Maroulis et al., 1995; Da Silva et al., 2012, 2014). experimental findings in the literature.
In this work, a one-dimensional numerical model is developed
for the assessment of the non-steady heat and moisture transport 2. Modeling of heat and mass transfer in quince slices
within a cylindrical quince slice having its axis parallel to the air
flow. Convective drying of quince has been studied in the literature The mechanisms for moisture removal from the quince are
up to some extent (Kaya et al., 2007a; Tzempelikos et al., 2014; assumed to be (i) the transport of moisture through the product
Barroca et al., 2012). The coupled heat and mass transport equations material by liquid diffusion and (ii) the evaporation of water from
are numerically solved using the finite volume method, considering the surface. No phase change, i.e. evaporation, occurs inside the
a third kind (Robin) boundary condition to account for the evapora- quince, an assumption suitable for diffusion-based type model, in
tion term. The thermophysical properties of the matter are func- terms of mass conservation (Datta, 2007a; Zhang and Datta,
tions of the moisture content, based on relationships available in 2004). The driving force for water diffusion inside the quince is
the literature, while the effective moisture diffusion coefficient is the concentration gradient developed by the evaporation of water
estimated by fitting the analytical solution of the simplified Fick’s on the surface, due to the difference of the water vapor partial
diffusion equation (Crank, 1975; Kaya et al., 2007a) to experimental pressure between the surface and the air. Liquid water diffusion
data. A simple mechanistic approach is implemented to model pro- inside the product is assumed to follow Fick’s law. Hot air provides
duct shrinkage, based on the assumption of volume conservation. A heat energy to the quince, a part of which is consumed for the
 D.A. Tzempelikos et al. / Journal of Food Engineering 156 (2015) 10–21 13

evaporation of the water leaving the surface, while the remaining

is conducted through the material increasing its temperature.
Radiation effects are neglected. Fig. 1 provides a sketch of the com-
putational domain and the parameters involved in the model. Con-
sidering the above mechanisms and assumptions, the one
dimensional heat and mass transfer equations can be written as
follows (Maroulis et al., 1995; Ruiz-López et al., 2004; Wang and
Brennan, 1995; Balaban and Piggot, 1988; Tripathi et al., 1973):
 
@ qb c p T @ @T
¼ kp ð1Þ
@t @z @z
 
@ qs X @ @X
¼ qs Deff ð2Þ
@t @z @z
where z 2 ½0; L is the axial coordinate (Fig. 1) and t is the drying
time.
Fig. 1. Flow configuration and one-dimensional numerical model and parameters
The following initial and boundary conditions are applied:
involved.
Tðz; t ¼ 0Þ ¼ T in and Xðz; t ¼ 0Þ ¼ X in ð3Þ
 
@T  ~ ðT 0  T Þ  h q D @T 

front face : kp  ¼ h ð4Þ
@z z¼0 The density of the solid material qs, defined by the ratio of the
f air fg s eff
@z z¼0
dry material mass to the total volume of the slice, is allowed to
  vary in the calculations, taking into account the volume shrinkage
@T  ~ ðT  T L Þh q D @X 

back face : kp  ¼h ð5Þ of the quince slice. A simple mechanistic approach is used to esti-
@z z¼L
b air fg s eff
@z z¼L
mate the slice volume variation, based on the assumption that the
 volume shrinkage at any time step can be approximated to be
@X  ~ ðC 0  C Þ
front face : qs Deff ¼h ð6Þ equal to the removed water volume in that time step (negligible
@z z¼0
m;f air
porosity and air mass in the quince slice). The removed water vol-
 ume is calculated from the moisture change during a time step,
@X  ~ ðC  C L Þ
back face : qs Deff ¼h ð7Þ which is directly derived by the solution of Eq. (2), while the
@z z¼L
m;b air
dry mass is directly derived from the known initial moisture con-
The vapor concentration at the surfaces is calculated from the tent and the total slice mass, which are both inputs for the
ideal gas law (Eqs. (8) and (9)), as a function of the water activity numerical model. The approach used for the estimation of volume
aw, using the formula proposed by Chen (1971) (Eq. (10)), where evolution during drying corresponds to a linear dependence of the
k1, k2 and k3 are constants obtained experimentally by Noshad product volume on the moisture, which have been recently used
et al. (2012): in the literature to fit experimental data for potato, with good
agreement (Chandra Mohan et al., 2014). The above assumption
p ðT 0 Þ
C 0 ¼ 2:1667  103  aw  v s ð8Þ alone was roughly evaluated for the present case, by comparing
T0 the calculated time evolution of the slice volume with measured
data.
p ðT L Þ
C L ¼ 2:1667  103  aw  v s ð9Þ Eqs. (1) and (2) are numerically discretized using the implicit
TL finite volume method (Patankar, 1980; Ferziger and Peric, 2002)
while the derived algebraic equations are solved using the Tridi-
aw ¼ exp½k1  k2 expðk3 MÞ ð10Þ
agonal Matrix Algorithm (Thomas algorithm). Internal iterations
The second term on the right hand side of Eqs. (4) and (5) are performed in each time step, in order to cope with the non-lin-
(boundary conditions for the heat equation) accounts for the heat earities associated with (i) the coupling of the heat and mass trans-
of evaporation of the water leaving the surface, where hfg (J/kg) port equations through the evaporation term in the boundary
is the water latent heat of evaporation, given by the following conditions, (ii) the thermophysical properties dependence on the
equation (ASHRAE, 2009): temperature and moisture and (iii) the dependence of the slice vol-
 0:3298 ume and density on the moisture. For the numerical solution of
647:3  T
hfg ¼ 2501:05  103  ð11Þ Eqs. (1) and (2), a spatial discretization using 101 grid points (cor-
647:3  273:15 responding to a grid size of about 0.06 mm) and a time step equal
The concentration of water vapor in air (Cair) is derived from the to 100 s are used, which provide spatial and temporal discretiza-
ideal gas law, according to the following equation: tion independency. For the internal iterations within each time
step, a convergence criterion equal to 106 is applied on the calcu-
RH pv s ðT air Þ lated average moisture, average temperature and density of the
C air ¼ 2:1667  103   ð12Þ
100 T air quince slice.
where pvs (Pa) is the water vapor pressure, given by (ASHRAE, The effective moisture diffusivity Deff is experimentally derived
2009): using the well-known slope method, by assuming an Arrhenius
" type equation (Meziane, 2011; Aghbashlo et al., 2009; Liu et al.,
5:8  103 2009). The thermal conductivity and specific heat (Table 2) were
pv s ¼ exp þ 1:391  4:864  102 T þ 4:176
T estimated as functions of the moisture content (Rao et al., 2005).
i The heat and mass transfer coefficients were derived from the
105 T 2  1:445  108 T 3 þ 6:545 lnðTÞ ð13Þ CFD modeling as described in Section 4.
14 D.A. Tzempelikos et al. / Journal of Food Engineering 156 (2015) 10–21

Table 2 The initial moisture content (Min) was measured by the oven-
Quince thermophysical properties, as a function of moisture content in wet basis, drying method at 105 °C for 24 h (AOAC, 1990) and found to be
M = X/(1 + X).
81.04% (w.b) or equivalently 4.274 kg water/kg dry (d.b.). The
Property Equation quince slices were placed having its axis parallel to the air-stream.
Thermal conductivity (W/m K) kp(M) = 0.148 + 0.493M Temperature was measured with a thermocouple Pt100 probe
Specific heat (J/kg K) cp(M) = (1.26 + 2.97M) ⁄ 1000 inserted in the quince slice. The LTCD unit was started 2 h before
each experiment, in order to achieve steady state conditions.
Experiments were performed for drying air temperature of 40, 50
and 60 °C and velocity of 1 and 2 m/s. The volumetric flow rate var-
3. Experimental facility and measurements
ied from 900 to 1800 m3/h, corresponding to a range of Reynolds
number from 2.64  104 to 5.27  104. The experimental condi-
The experimental dryer has been previously described in detail
tions are summarized in Table 3.
by Tzempelikos et al. (2012, 2013). Here, only a brief description is
The effective moisture diffusivity (Deff) is obtained from the
given. The layout of the Laboratory Thermal Convective Dryer
experimental moisture ratio data, using the slope method. Details
(LTCD) is shown in Fig. 2. The drying chamber has dimensions
of the method can be found in (Tzempelikos et al., 2014). A tem-
0.5 m  0.5 m  0.66 m. Insulation material (Frelen) of 10 mm
perature dependence of the effective diffusion coefficient is
thickness has been used to prevent heat losses. The whole system
assumed, following an Arrhenius type equation:
includes a centrifugal fan with backward curved blades, a diffuser
 
with a 6.7° divergence angle to adjust LTCD length with minimum Ea
Deff ¼ D0 exp  ð14Þ
pressure losses, a heat exchanger connected to boiler-hot water Rg T
system, a spraying system that can feed the air stream with water
mist and a set of dumpers which allow flow recirculation or new, where D0 is the pre-exponential factor of the Arrhenius equation (in
fresh air to enter into the dryer. A metal tray of 78% permeability m2/s); Ea is the activation energy (in kJ/mol); Rg is the gas constant
is used to align the quince slices in the drying chamber. The design (in kJ/mol K); T is the absolute temperature, resulting to the
of the whole facility is modular, so that the drying chamber can be equation:
vertically or horizontally aligned. For the experiments presented  
4431:713
here, the vertical orientation was adopted. Five A class tolerance Deff ¼ 5:492  104 exp  ð15Þ
T
Pt 100 probes were used for temperature measurements for the
product and the drying air. Three humidity transmitters were used The correlation of this equation with the experimental data is
for measuring the relative humidity of the drying air. In order to shown in Fig. 3, while the values of the effective diffusion coeffi-
measure the air drying velocity at the inlet of the drying cient for the different experimental cases are given in Table 4.
chamber, a pitot tube rake was connected to differential pressure
transmitter. 4. Computational fluid dynamics simulations
A system of four load cells and analogue transmitter connected
to a PC with a DAQ device through different modules provides the Steady-state CFD simulations were performed to estimate the
option for automated weight loss measurement and data acquisi- heat and mass transfer coefficients. Flow and temperature fields
tion. The average velocity is adjusted by weighting twelve dynamic inside the drying chamber of the LTCD unit were calculated using
pressure measurements using three pitot tube rakes – each one the commercial CFD code FluentÒ (Fluent Inc., 2006). In the initial
consisting of four pitot tubes – and static pressure measurements stage of drying, an unsteady transitional variation of the flow and
at the wall of the duct system, prior to the inlet of the drying cham- temperature field around the slice is expected, until steady state
ber. Additional measurements of the weight and the dimensions of conditions are reached. However, this transitional period is very
samples were taken at selected time steps, out of the dryer. A cus- short and there is no practical benefit rather than increased com-
tom-made virtual instrumentation application was developed in putational cost (Ateeque et al., 2014). For the present geometry,
LabviewÒ in order to operate and control the LTCD. due to the symmetry, the chamber and the cylindrical quince slice

Fig. 2. Schematic diagram of the LTCD unit (curved arrows indicate the flow direction when dryer is in recirculation operation).
 D.A. Tzempelikos et al. / Journal of Food Engineering 156 (2015) 10–21 15

Table 3
Conditions for the experimental measurements.

Tair (°C) uair (m/s) RH (%) d (m)  103 Lin (m)  103 Lfn (m)  103 Xfn (d.b.) win (kg)  103 wd (kg)  103 Tin (°C)
40 1.0 19.54 92.50 9.20 2.50 0.198 61.74 11.71 16.40
50 1.0 12.24 82.80 10.60 2.79 0.583 56.66 10.75 17.70
60 1.0 6.77 89.00 10.58 2.14 0.253 65.43 12.40 20.88
40 2.0 17.14 81.80 9.20 2.06 0.371 48.25 9.15 17.12
50 2.0 9.00 85.00 10.04 2.70 0.150 56.48 10.71 20.65
60 2.0 2.93 87.01 9.50 2.05 0.180 56.28 10.67 24.51

Fig. 3. Arrhenius-type relationship between effective moisture diffusivity and

temperature.

Table 4
Effective moisture diffusivity coefficient (Deff) . Fig. 4. Computational domain.
2 2
Tair (°C) uair (m/s) Deff (m /s) R
40 1.0 3.82343  1010 0.9921 surfaces of the slice, indicating a sufficient grid resolution in accor-
50 1.0 6.01866  1010 0.9901
dance with the turbulence model used.
60 1.0 8.98063  1010 0.9889
40 2.0 3.88313  1010 0.9954
50 2.0 6.60007  1010 0.9894 4.1. Estimation of heat and mass transfer coefficients
60 2.0 9.02179  1010 0.9865

The local convection heat transfer coefficient h can be deter-

mined from the calculated air temperature field, using the balance
between conductive and convective heat transfer:
are modeled as a 2-D axis-symmetric configuration, as shown in  
Fig. 4 taking into account only the half of the chamber and quince. @T
kair ¼ hðT ps  T 1 Þ ð16Þ
Measured velocity and temperature at the inlet and measured @n s
pressure and temperature at the outlet were imposed as boundary
where n, is the normal vector at the surface s of the product and T1
conditions (Fig. 4).
is the drying air bulk temperature. After the convective heat trans-
The flow around the quince slice is considered as a flow around
fer coefficient is determined, the convective mass transfer coeffi-
a bluff body. It is well known that for high Reynolds numbers, tur-
cient hm (m/s) is calculated using the analogy between the
bulence modeling is needed to assess the fluctuations of the flow
thermal and concentration boundary layers (Chilton and Colburn,
and thermal fields (Murakami, 1993; Defraeye et al., 2010;
1934; Incropera et al., 2012), as commonly made in the literature:
Ateeque et al., 2014). In the present work, the SST kx turbulence !
model is used to account for turbulence (Menter, 1994). This model Dv a Le1=3
has been used in the previous works of Defraeye et al. (2010) and hm ¼ h ð17Þ
kair
Ateeque et al. (2014), in order to reproduce with sufficient accura-
cy experimental data for external flows. where Le, is the Lewis number representing a measure of the rela-
CFD simulations were performed for all cases listed in Table 3. A tive thermal and concentration boundary layer thickness:
structured mesh, with appropriate refinement around the quince
aair
slice is used, in order to capture reattachment and wall fluxes Le ¼ ; ð18Þ
Dv a
towards the quince surface. The mesh consisted of 715 and 218
grid points in the axial and in radial direction, respectively. The aair is the thermal diffusivity of the air and Dva is the diffusivity
wall characteristic length (y+) was of the order of 1 for all of the of water vapor in air.
16 D.A. Tzempelikos et al. / Journal of Food Engineering 156 (2015) 10–21

The calculated spatial distribution of the convective heat trans-

fer coefficients exhibits a similar profile in all experimental cases.
For this reason, results are presented here for a single experimental
case. In Figs. 5 and 6 the spatial distribution of the heat transfer
coefficient at the two sides of the slice are shown, for the case of
40 °C and 2 m/s. The values of heat and mass transfer coefficients
are in general higher at the front face of the quince. At the front
face, the values are larger at the center of the slice and they
decrease moving radially towards the edge. At the back face, the
heat and mass transfer coefficients present an inverse spatial dis-
tribution: the maximum value is located at the edge of the slice,
while the minimum near the center.
In Tables 5 and 6, the calculated convective heat and mass
transfer coefficients, in terms of their minimum, maximum and
average values in both sides of the slice are summarized for all
experimental cases. The coefficients were found to be practically
independent from the air temperature, but they depict a strong
dependency on the air velocity. In particular, the average heat
and mass transfer coefficients increase for more than 50%, as the
velocity increases from 1 m/s to 2 m/s. This result indicates the
relative importance of the air drying velocity on the drying pro- Fig. 5. Calculated heat transfer coefficient at the front and the back face of the
cess: higher air velocity result to higher heat and mass transfer quince slice (T = 40 °C, u = 2 m/s).

coefficients, which in turn induce higher heat and moisture fluxes

at the surface and therefore speed up the overall drying process.
The weighted average heat and mass transfer coefficients at the
front and back side of the slice, which are necessary for the one-di-
mensional model of heat and mass transfer inside the quince, are
calculated from the local values according to the following
equation:
Z
~¼1
h hdA ð19Þ
A

Z
~ ¼1
h hm dA ð20Þ
m
A
The Nusselt number is typically expressed as a function of Pr
and Rep numbers of the flow around the slice:

Nu ¼ cReep Pr 1=3 ð21Þ

In order to obtain more data for the fitting of Eq. (21) and also to
extend the range for the application of the proposed correlation,
additional CFD simulations were performed for two additional air
velocities, namely 0.5 m/s and 1.5 m/s (for 40 °C, 50 °C and
Fig. 6. Calculated mass transfer coefficient at the front and the back face of the
60 °C). The results at both sides of the slice for all the test cases quince slice (T = 40 °C, u = 2 m/s).
are shown in Fig. 7. The figure includes also the fitting equations.
The values of c and e in Eq. (21) for the front and the back face
of the cylindrical slice and presented in Table 7. Table 5
Maximum, minimum and weighted average heat transfer coefficients at different
velocities and temperatures.
5. Results and discussions Faces Tair (°C) Heat transfer coefficient, h (W/m2 K)
Air velocity, 1.0 m/s Air velocity, 2.0 m/s
The approach used in the model for calculating the change of
the volume of the slice during drying is roughly evaluated by com- Min. Max. Average Min. Max. Average

paring the calculated slice volume with the measured data. The Front 40 27.58 75.96 37.86 46.21 98.53 58.75
measured data of the slice volume were obtained by direct mea- Front 50 27.50 76.18 37.91 44.53 99.08 57.01
Front 60 24.29 69.96 34.25 43.81 100.87 56.38
surements of the slice thickness and diameter and assuming a per-
Back 40 14.43 24.12 18.84 21.85 41.18 31.00
fect cylindrical geometry. In Fig. 8, the calculated slice volume, as a Back 50 14.48 24.08 18.86 21.18 39.43 29.66
function of time is shown, in comparison with measured data for Back 60 13.20 21.54 17.05 20.94 38.63 29.08
the experimental case of 40 °C and 2 m/s. A fine agreement is
observed between the calculated and the measured data for the
two thirds of the drying time. At the last stages of drying, sig- is increasingly inaccurate, in particular in the last period of drying,
nificant deviations are observed between the model predictions where the slice is highly deformed.
and measurements, which can be attributed to the approach used In order to take further into account in the calculations the
for measuring the volume which is made geometrically, based on shrinkage of the quince, the thickness of the slice was estimated
the assumption of a perfect cylinder. This assumption, however, as the average of the initial and the final thickness of the product.
 D.A. Tzempelikos et al. / Journal of Food Engineering 156 (2015) 10–21 17

Table 6 Table 7
Maximum, minimum and weighted average mass transfer coefficients at different Correlation constants at opposite faces of cylindrical quince slice for air.
velocities and temperatures.
Faces Nu ¼ c Reep Pr 1=3 R2
Faces Tair (°C) Mass transfer coefficient, hm (m/s)
Constant (c) Exponent (e)
Air velocity, 1.0 m/s Air velocity, 2.0 m/s
Front 0.472 0.657 0.994
Min. Max. Average Min. Max. Average Back 0.150 0.710 0.993
Front 40 0.0270 0.0745 0.0371 0.0453 0.0966 0.0576
Front 50 0.0265 0.0734 0.0383 0.0429 0.0954 0.0576
Front 60 0.0230 0.0662 0.0356 0.0415 0.0955 0.0586
Back 40 0.0141 0.0236 0.0185 0.0214 0.0404 0.0304 diffusion equation to the experimental drying curves (slope
Back 50 0.0139 0.0232 0.0190 0.0204 0.0380 0.0299 method). However, the analytical solution used for the post-ex-
Back 60 0.0125 0.0204 0.0177 0.0198 0.0366 0.0302 perimental processing is a result of a series of approximations
(e.g. isothermal conditions, negligible surface resistance, constant
properties), attributing ‘artificially’ the effect of a number of
The initial thickness is known, while the final can be approximated neglected parameters to the diffusion coefficient itself (see also
a priori, based on the assumption that the dimensions of the slice Ruiz-López et al., 2004).
vary only in the axial direction, i.e. only the thickness of the slice These approximations are not employed in the present numeri-
changes during the drying process, while the diameter remains cal model. It is, thus, reasonable that the diffusion coefficient
constant. The final thickness can be estimated from the ratio of derived by fitting the analytical solution to the measured data
the final slice volume (which is approximated to be the initial vol- needs additional matching, in order to be used in the present more
ume minus the volume of the initial water content) over the area of detailed model. In order to achieve higher accuracy between
the quince slice, which is assumed to remain constant during the experimental data and predictions a non-linear regression analysis,
drying. Using this approach, the final thickness of the slice is calcu- using an Arrhenius type effective diffusion equation, was also per-
lated to be about 20% of the initial thickness, a value close to the formed. However, preliminary result, obtained using the SQP
measured data that range between 20% and 27% of the initial thick- (Sequential Quadratic Programming) and Interior Point algorithms
ness (Table 3). (available in constrained non-linear optimization toolbox of
In Fig. 9, the spatially-averaged moisture content and the tem- MATLABÒ) for the minimization of the Chi-square function (see also
perature at the center of the quince slice are plotted as functions of Da Silva et al. (2012)) showed only small improvement of the calcu-
time, in comparison with the measured data. For the present mod- lated results (with a significant increase of the computational cost:
el, the effective diffusion coefficient given by Eq. (15) was correct- from less than a minute per run to about 1/2 hour). The limitation
ed by a factor equal to 0.40, for all the cases examined. The on the improvement, even when optimization is used, may be
correction of the value of the diffusion coefficient was necessary, attributed to the formula used for the estimation of the diffusion
due to the different level of approximations inherent in the coefficient. In order to further improve the accuracy of the model,
approach used for the experimental derivation of the diffusion more complicated forms of the diffusion coefficient equation
coefficient, in comparison with the present numerical model. It is should be employed, by introducing in addition to the temperature,
commonly assumed that the effective diffusion coefficient a dependence also on the moisture content (Ateeque et al., 2014;
describes the combined effects of air and liquid transport inside Islam et al., 2003; Datta, 2007b) in the non-linear regression analy-
the drying material. So, its value cannot be derived by the mass sis (as also discussed in Ruiz-López et al. (2004)).
transport theory (Datta, 2007a, 2007b; Zhang and Datta, 2004); From the comparison between the experimental and the
instead, it can only be derived by a semi-empirical method, based numerical results shown in Fig. 9, it can be inferred that the pre-
on the post-processing of the available experimental data. In the sent model is able to predict with acceptable accuracy without
approach used in Section 3, the derivation of the diffusion coeffi- the need for any further adjustment or experimental fitting, the
cient is based on the fitting of the analytical solution of the Fick’s effect of the different experimental drying conditions, i.e. the

Fig. 7. Correlations for the Nusselt number at opposite faces of cylindrical quince
slice. Fig. 8. Time evolution of the volume of the product (T = 40 °C, u = 2 m/s).
18 D.A. Tzempelikos et al. / Journal of Food Engineering 156 (2015) 10–21

(a) (b)

(c) (d)

Fig. 9. Calculated moisture (a and c) and temperature (b and d) as a function of time, in comparison with measurements, during the drying process for the conducted
experiments.

dependence on the air stream temperature and velocity, provided and the effective diffusivity increases. In addition, higher air velo-
an appropriate single Arrhenius type diffusion coefficient relation cities are associated with higher heat transfer coefficients (Table 5),
and the appropriate heat transfer coefficient. The deviations resulting in turn to faster increase of temperature and thus to high-
observed between the calculated and measured data, can be also er diffusivity and higher moisture removal rate. As it is discussed
attributed to other uncertainties, as for example in relation to later in this paper, the variation of the mass transfer coefficient
the change of the slice geometry and shape during drying that and the boundary conditions of the moisture equation have negli-
can produce variations of the heat and mass transfer coefficients gible influence on the drying process. After the initial period, the
as the drying process evolves in time, an effect that has not been pattern inverses and the drying rate is lower for higher air tem-
taken into account in the present CFD simulations. peratures and velocities, since the higher drying rate at the first
As mentioned previously, the results in Fig. 9 indicate that period leads faster to such a low moisture content that is more dif-
water removal is faster for higher drying air temperatures and ficult to be removed. This behavior is also confirmed by experi-
velocities. This can also be illustrated in Fig. 10, where the time ments (Tzempelikos et al., 2015).
evolution of the drying rate is presented. In this figure, it is shown The drying rate curves depict the characteristic shape of the so-
that up to a critical time where the larger amount of water has called falling rate period. The drying rate initially decreases with a
been removed, higher drying air temperatures and velocities are high rate within a relatively narrow period of time followed by a
associated with higher drying rates. This behavior is due solely to more extended period which is characterized by a slower decrease
the strong dependence of the moisture effective diffusivity on tem- of the drying rate. It is widely accepted that in the falling rate peri-
perature. For higher temperature differences between air stream od the drying process is governed by the internal water transport
and product, the heat flux to the interior of the product is higher by diffusion (diffusion-controlled), while the external conditions
 D.A. Tzempelikos et al. / Journal of Food Engineering 156 (2015) 10–21 19

Fig. 10. Calculated drying rate for the different experimental cases.

have practically no influence (Barbosa-Canobas et al., 1996). In the

falling rate period, only a very small amount of water exist at the
Fig. 12. Calculated heat fluxes and temperature, as a function of time during drying
surface of the product because the moisture content at the surface (T = 40 °C, u = 2 m/s).
is evaporated very fast, due to the low ‘resistance’ from the quince
material there, while at the same time, the internal moisture transfer coefficient at the back and the front face of the slice, the
migrates (to the surface) due to the concentration gradients. This moisture profile inside the material exhibits a symmetrical shape,
is also illustrated in Fig. 11a, where the calculated spatial distribu- indicating the negligible effect of the external conditions. Similar
tion of moisture profile inside the product at various moments dur- conclusion have been reached and in the work of (Aregawi et al.,
ing drying is presented. As it is shown, a moisture profile is 2013) for apples, where a different modeling approach was used.
developed in the interior of the product, while close to the surface Fig. 11b shows the spatial distribution of the temperature along
boundary the moisture drops almost immediately creating high the thickness of the quince slice, only for a short period (up to
spatial gradients. This behavior has been also confirmed about 1/2 hour) after the beginning of drying, as predicted by the
experimentally for other fruits (Aregawi et al., 2013). The diffusion model. The temperature profile presents only small gradients along
of the internal water towards the surface, however, is much slower the quince slice, which vanish quickly. This particular behavior is
due to the high ‘internal mass resistance’, resulting to a consider- principally attributed to the low heat transfer Biot number, which
ably slower progress of the water removal (Ateeque et al., 2014; range between 0.22 and 0.48. These values are of the same order of
Chandra Mohan and Talukdar, 2010). The existence of strong spa- magnitude compared to the limit of ‘thermally-thin’ materials, set
tial gradients of moisture inside the product is also compatible to the order of 0.1 (Incropera et al., 2007).
with the high values of the mass transfer Biot number (of the order In Fig. 12, the time evolution of the energy balance during dry-
of 106), which is much higher than the corresponding limit, of the ing is depicted, in terms of the calculated heat fluxes, namely the
order of 1000 (Parti, 1993). Despite the different values of the mass convective heat flux (energy provided from the drying air to the

(a) (b)

Fig. 11. Moisture (a) and temperature (b) profiles inside the quince slice, for different times during drying (T = 40 °C, u = 2 m/s).
20 D.A. Tzempelikos et al. / Journal of Food Engineering 156 (2015) 10–21

quince slice), the water evaporation heat flux (energy consumed ASHRAE, 2009. ASHRAE Handbook-Fundamentals. American Society of Heating,
Refrigiration and Air-Conditioning Engineers Inc., Atlanta.
for the evaporation of the water at the surface) and the conductive
Ateeque, M., Mishra, R.K., Chandra Mohan, V.P., Talukdar, P., 2014. Numerical
flux (energy conducted in the interior of the quince heating the modeling of convective drying of food with spatially dependent transfer
material). These fluxes correspond to the three terms in the bound- coefficient in a turbulent flow field. Int. J. Therm. Sci. 78, 145–157.
ary conditions of the heat equation (Eqs. (4) and (5)). The tem- Aversa, M., Curcio, S., Calabrò, V., Iorio, G., 2007. An analysis of the transport
phenomena occurring during food drying process. J. Food Eng. 78 (3), 922–932.
perature evolution is also shown. As it is shown from this figure, Babalis, S.J., Belessiotis, V.G., 2004. Influence of the drying conditions on the drying
the energy available for the heating of the material is only a very constants and moisture diffusivity during the thin-layer drying of figs. J. Food
small portion (one order of magnitude lower) of the total energy Eng. 65 (3), 449–458.
Balaban, M., Piggot, G.M., 1988. Mathematical model of simultaneous heat and mass
provided to the product by the air stream. The latter is practically transfer in food with dimensional changes and variable transport parameters. J.
consumed for the water evaporation. At the first short period of Food Sci. 53 (3), 935–939.
drying, however, even this small portion is adequate to heat the Barati, E., Esfahani, J.a., 2011. A new solution approach for simultaneous heat and
mass transfer during convective drying of mango. J. Food Eng. 102 (4), 302–309.
material, causing a rapid increase of its temperature. The increase Barati, E., Esfahani, J.A., 2012. Mathematical simulation of convective drying:
of the temperature results to the decrease of the convective energy Spatially distributed temperature and moisture in carrot slab. Int. J. Therm. Sci.
provided to the material or equivalently to a decrease of the energy 56, 86–94.
Barbosa-Canobas, G.V., Vega-Mercado, H., 1996. Dehydration of Foods. Chapman &
available for further water evaporation and material heating. After Hall.
this short period and up to the end of the process, almost the entire Barroca, M.J., Guiné, R.P.F., 2012. Study of Drying Kinetics of Quince. In: Proceedings
energy is consumed for water evaporation, resulting to a very slow (Electronic) da 362 International Conference of Agricultural Engineering CIGR-
AgEng2012.
temperature increase and to the ‘‘plateau’’ depicted in the
Bird, R.B., Stewart, W.E., Lightfoot, E.N., 1960. Transport Phenomena. John Wiley &
temperature evolution curve. Sons, London, UK.
Burnnett, C.O., Myers, J.E., 1962. Momentum, Heat and Mass Transfer. McGraw Hill,
New York.
6. Conclusions Chandra Mohan, V.P., Talukdar, P., 2010. Three dimensional numerical modeling of
simultaneous heat and moisture transfer in a moist object subjected to
convective drying. Int. J. Heat Mass Transfer 53 (21–22), 4638–4650.
A 1-D unsteady heat and mass transfer numerical model was Chandra Mohan, V.P., Talukdar, P., 2014. Experimental studies for convective drying
developed to study the transient behavior of temperature and mois- of potato. Heat Transfer Eng. 35 (14–15), 1288–1297.
ture content during drying of cylindrical quince slices. The heat Chen, C.S., 1971. Equilibrium moisture curves for biological materials. Trans. ASABE
14 (5), 924–926.
transfer inside the quince is considered to be by conduction while
Chilton, T.H., Colburn, A.P., 1934. Mass transfer (absorption) coefficients prediction
the moisture transfer is considered to be governed solely by liquid from data on heat transfer and fluid friction. Ind. Eng. Chem. 26 (11), 1183–
diffusion. Evaporation is considered to take place only at the sur- 1187.
Cordova-Quiroz, A.V., Ruiz-Cabrera, M.A., Garcfa-Alvarado, M.A., 1996. Analytical
faces of the quince slice. The averaged heat and mass transfer coef-
solution of mass transfer equation with interfacial resistance in food drying.
ficients required for the model are calculated from CFD simulations. Drying Technol. 14 (7–8), 1815–1826.
The latter are performed for a steady turbulent flow in a two-dimen- Crank, J., 1975. The Mathematics of Diffusion. second ed.. Oxford Sciences
sional axisymmetric computational domain, incorporating the SST Publication.
Curcio, S., 2010. A multiphase model to analyze transport phenomena in food
kx turbulence model for the prediction of the external flow and drying processes. Drying Technol. 28 (6), 773–785.
temperature fields around the product. A new correlation is derived Curcio, S., Aversa, M., Calabrò, V., Iorio, G., 2008. Simulation of food drying: FEM
for the heat convective transfer coefficient, as function of Reynolds analysis and experimental validation. J. Food Eng. 87 (4), 541–553.
Da Silva, W.P., e Silva, C.M.D.P.S., Farias, V.S.O., Gomes, J.P., 2012. Diffusion models
and Prandtl numbers, for the particular flow configuration, namely to describe the drying process of peeled bananas: optimization and simulation.
air flow parallel to the axis of cylindrical slices, which is scarcely Drying Technol. 30 (2), 164–174.
considered in the literature. The effective diffusivity is estimated Da Silva, W.P., Silva, C.M.D.P.S., Gama, F.J.A., e Silva, C.M.D.P.S., 2014. Estimation of
thermo-physical properties of products with cylindrical shape during drying:
experimentally using a typical method that involves a simplified the coupling between mass and heat. J. Food Eng. 141, 65–73.
analytical solution of diffusion inside the product. Datta, A.K., 2007a. Porous media approaches to studying simultaneous heat and mass
In diffusion-based model the various complex moisture trans- transfer in food processes. I: problem formulations. J. Food Eng. 80 (1), 80–95.
Datta, A.K., 2007b. Porous media approaches to studying simultaneous heat and
port mechanisms in the product, which still are not fully resolved mass transfer in food processes. II: property data and representative results. J.
in drying theory and modeling, are surrogated by a simple Fick’s Food Eng. 80 (1), 96–110.
diffusion transport model. Due to the approximations of the slope De Bonis, M.V., Ruocco, G., 2008. A generalized conjugate model for forced
convection drying based on an evaporative kinetics. J. Food Eng. 89 (2), 232–240.
method that is used for the experimental derivation of the effective
Defraeye, T., 2014. Advanced computational modelling for drying processes – a
diffusivity, its value in the model was corrected by a factor of 0.4. review. Appl. Energy 131, 323–344.
Comparisons with experimentally acquired data were performed Defraeye, T., Blocken, B., Carmeliet, J., 2010. CFD analysis of convective heat transfer
for a range of flow and thermal conditions demonstrating that at the surfaces of a cube immersed in a turbulent boundary layer. Int. J. Heat
Mass Transfer 53 (1–3), 297–308.
the present, diffusion-based model is able to describe the strong Dincer, I., 1997. Heat Transfer in Food Cooling Applications. Taylor & Francis,
heat and mass transfer coupling and capture the time evolution Washington, DC.
Doymaz, I.,_ 2009. An experimental study on drying of green apples. Drying Technol.
of moisture content and temperature with minimum experimental
27 (3), 478–485.
adjustment. The model is robust, computationally efficient (trivial Earle, R.L., 1983. Unit Operations in Food Engineering. Pergamon Press, Oxford.
for a modern PC), and so it is considered suitable for analyzing the Erbay, Z., Icier, F., 2010. A review of thin layer drying of foods: theory, modeling, and
drying process, from an engineering point of view. experimental results. Crit. Rev. Food Sci. Nutr. 50 (5), 441–464.
Esfahani, J.A., Majdi, H., Barati, E., 2014. Analytical two-dimensional analysis of the
transport phenomena occurring during convective drying: apple slices. J. Food
References Eng. 123, 87–93.
Ferziger, J.H., Perić, M., 2002. Computational methods for fluid dynamics, 3rd ed.
Springer-Verlag, Berlin.
Aghbashlo, M., Kianmehr, M.H., Arabhosseini, A., 2009. Modeling of thin-layer
Fluent Inc., 2006. FLUENT 6.3 User’s Guide.
drying of potato slices in length of continuous band dryer. Energy Convers.
Halder, A., Datta, A.K., 2012. Surface heat and mass transfer coefficients for
Manage. 50 (5), 1348–1355.
multiphase porous media transport models with rapid evaporation. Food
Akpinar, E.K., Dincer, I., 2005. Application of moisture transfer models to solids
Bioprod. Process. 90 (3), 475–490.
drying. Proc. Inst. Mech. Eng., Part A: J. Power Energy 219 (3), 235–244.
Hussain, M.M., Dincer, I., 2003. Numerical simulation of two-dimensional heat and
AOAC, 1990. Official Methods of Analysis, 15th ed. Association of Official Analytical
moisture transfer during drying of a rectangular object. Numer. Heat Transfer,
Chemists, Arlington, VA.
Part A: Appl. 43 (8), 867–878.
Aregawi, W., Defraeye, T., Saneinejad, S., Vontobel, P., Lehmann, E., Carmeliet, J.,
Incropera, F.P., DeWitt, D.P., Bergman, T.L., Lavine, A.S., 2007. In: Incropera, F.P.,
Verboven, P., Derome, D., Nicolaï, B.M., 2013. Understanding forced convective
Incropera, F.P.F.O.H.A.M.T. (Eds.), Fundamentals of Heat and Mass Transfer. John
drying of apple tissue: Combining neutron radiography and numerical
Wiley & Sons.
modelling. Innovative Food Sci. Emerging Technol.
 D.A. Tzempelikos et al. / Journal of Food Engineering 156 (2015) 10–21 21

Incropera, F.P., DeWitt, D.P., Bergman, T.L., Lavine, A.S., 2012. Principles of Heat and Noshad, M., Shahidi, F., Mohebbi, M., Ali Mortazavi, S., Mortazavi, S.A., 2012.
Mass Transfer, seventh ed. John Wiley & Sons. Desorption isotherms and thermodynamic properties of fresh and osmotic-
Islam, M.R., Ho, J.C., Mujumdar, A.S., 2003. Convective drying with time-varying ultrasonic dehydrated quinces. J. Food Process. Preserv. 37 (5), 381–390.
heat input: simulation results. Drying Technol. 21 (7), 1333–1356. Oztop, H.F., Akpinar, E.K., 2008. Numerical and experimental analysis of moisture
Janjai, S., Lamlert, N., Intawee, P., Mahayothee, B., Haewsungcharern, M., Bala, B.K., transfer for convective drying of some products. Int. Commun. Heat Mass
Müller, J., 2008. Finite element simulation of drying of mango. Biosyst. Eng. 99 Transfer 35 (2), 169–177.
(4), 523–531. Parti, M., 1993. Selection of mathematical models for drying grain in thin-layers. J.
Kaya, A., Aydin, O., Dincer, I., 2006. Numerical modeling of heat and mass transfer Agric. Eng. Res. 54 (4), 339–352.
during forced convection drying of rectangular moist objects. Int. J. Heat Mass Patankar, S.V., 1980. Numerical Heat Transfer and Fluid Flow Hemisphere. CRC Press.
Transfer 49 (17–18), 3094–3103. Perry, R.H., Green, D.W., Maloney, J.O., 1997. Perry’s Chemical Engineers Handbook,
Kaya, A., Aydin, O., Demirtas, C., Akgün, M., 2007a. An experimental study on the 7th ed. McGraw Hill, USA.
drying kinetics of quince. Desalination 212 (1–3), 328–343. Rao, M.A., Rizvi, S.S.H., Datta, A.K., 2005. Engineering Properties of Foods. Taylor &
Kaya, A., Aydin, O., Dincer, I., 2007b. Numerical modeling of forced-convection Francis, Boca Raton.
drying of cylindrical moist objects. Numer. Heat Transfer, Part A 51 (9), Ratti, C., Crapiste, G.H., 1995. Determination of heat transfer coefficients during
843–854. drying of foodstuffs. J. Food Process Eng. 18, 41–53.
Kaya, A., Aydin, O., Dincer, I., 2008. Experimental and numerical investigation of Ruiz-López, I.I., Córdova, A.V., Rodrı́guez-Jimenes, G.C., Garcı´a-Alvarado, M.F., 2004.
heat and mass transfer during drying of Hayward kiwi fruits (Actinidia Deliciosa Moisture and temperature evolution during food drying: effect of variable
Planch). J. Food Eng. 88 (3), 323–330. properties. J. Food Eng. 63 (1), 117–124.
Kurnia, J.C., Sasmito, A.P., Tong, W., Mujumdar, A.S., 2013. Energy-efficient thermal Sabarez, H.T., 2012. Computational modelling of the transport phenomena
drying using impinging-jets with time-varying heat input – a computational occurring during convective drying of prunes. J. Food Eng. 111 (2), 279–288.
study. J. Food Eng. 114 (2), 269–277. Sacilik, K., Elicin, A.K., 2006. The thin layer drying characteristics of organic apple
Lamnatou, C., Papanicolaou, E., Belessiotis, V.G., Kyriakis, N., 2009. Conjugate heat slices. J. Food Eng. 73 (3), 281–289.
and mass transfer from a drying rectangular cylinder in confined air flow. Seiiedlou, S., Ghasemzadeh, H.R., Hamdami, N., Talati, F., Moghaddam, M., 2010.
Numer. Heat Transfer, Part A 56 (5), 379–405. Convective drying of apple: mathematical modeling and determination of some
Lamnatou, C., Papanicolaou, E., Belessiotis, V.G., Kyriakis, N., 2010. Finite-volume quality parameters. Int. J. Agric. Biol. 12, 171–178.
modelling of heat and mass transfer during convective drying of porous bodies Tripathi, G., Shukla, K.N., Pandey, R.N., 1973. An integral equation approach to heat
– non-conjugate and conjugate formulations involving the aerodynamic effects. and mass transfer problem in an infinite cylinder. Int. J. Heat Mass Transfer 16
Renew. Energy 35 (7), 1391–1402. (5), 985–990.
Lemus-Mondaca, R.A., Vega-Gálvez, A., Moraga, N.O., 2011. Computational Tzempelikos, D.A., Vouros, A.P., Bardakas, A.V., Filios, A.E., Margaris, D.P., 2012.
simulation and developments applied to food thermal processing. Food Eng. Analysis of air velocity distribution in a laboratory batch-type tray air dryer by
Rev. 3 (3–4), 121–135. computational fluid dynamics. Int. J. Math. Comput. Simul. 6 (5), 413–421.
Lemus-Mondaca, R.A., Zambra, C.E., Vega-Gálvez, A., Moraga, N.O., 2013. Coupled Tzempelikos, D.A., Vouros, A.P., Bardakas, A.V., Filios, A.E., Margaris, D.P., 2013.
3D heat and mass transfer model for numerical analysis of drying process in Design, construction and evaluation of a new laboratory convective dryer using
papaya slices. J. Food Eng. 116 (1), 109–117. CFD. Int. J. Mech. 7 (4), 425–434.
Liu, X., Qiu, Z., Wang, L., Cheng, Y., Qu, H., Chen, Y., 2009. Mathematical modeling for Tzempelikos, D.A., Vouros, A.P., Bardakas, A.V., Filios, A.E., Margaris, D.P., 2014. Case
thin layer vacuum belt drying of Panax notoginseng extract. Energy Convers. studies on the effect of the air drying conditions on the convective drying of
Manage. 50 (4), 928–932. quinces. Case Stud. Therm. Eng. 3, 79–85.
Margaris, D.P., Ghiaus, A.G., 2007. Experimental study of hot air dehydration of Tzempelikos, D.A., Vouros, A.P., Bardakas, A.V., Filios, A.E., Margaris, D.P., 2015.
sultana grapes. J. Food Eng. 79 (4), 1115–1121. Experimental study on convective drying of quince slices and evaluation of
Maroulis, Z.B., Kiranoudis, C.T., Marinos-Kouris, D., 1995. Heat and mass transfer thin-layer drying models. Eng. Agric., Environ. Food.
modeling in air drying of foods. J. Food Eng. 26 (1), 113–130. Velić, D., Planinić, M., Tomas, S., Bilić, M., 2004. Influence of airflow velocity on
Marra, F., De Bonis, M.V., Ruocco, G., 2010. Combined microwaves and convection kinetics of convection apple drying. J. Food Eng. 64 (1), 97–102.
heating: a conjugate approach. J. Food Eng. 97 (1), 31–39. Villa-Corrales, L., Flores-Prieto, J.J., Xamán-Villaseñor, J.P., García-Hernández, E.,
Menter, F.R., 1994. Two-equation eddy-viscosity turbulence models for engineering 2010. Numerical and experimental analysis of heat and moisture transfer
applications. AIAA J. 32 (8), 1598–1605. during drying of Ataulfo mango. J. Food Eng. 98 (2), 198–206.
Meziane, S., 2011. Drying kinetics of olive pomace in a fluidized bed dryer. Energy Wang, N., Brennan, J.G., 1995. A mathematical model of simultaneous heat and
Convers. Manage. 52 (3), 1644–1649. moisture transfer during drying of potato. J. Food Eng. 24, 47–60.
Mujumdar, A.S., 2006. Book Review: Handbook of Industrial Drying, third ed. CRC Zhang, J., Datta, A.K., 2004. Some considerations in modeling of moisture transport
Press. in heating of hygroscopic materials. Drying Technol. 22 (8), 1983–2008.
Murakami, S., 1993. Comparison of various turbulence models applied to a bluff Zlatanović, I., Komatina, M., Antonijević, D., 2013. Low-temperature convective
body. J. Wind Eng. Ind. Aerodyn. 46–47, 21–36. drying of apple cubes. Appl. Therm. Eng. 53 (1), 114–123.