food and bioproducts processing 8 7 ( 2 0 0 9 ) 96101

Contents lists available at ScienceDirect

Food and Bioproducts Processing

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

Mathematical modeling and simulation of shrunk

cylindrical materials drying kineticsApproximation
and application to banana
Bilel Hadrich, Nabil Kechaou
Groupe de Gnie des Procds Agroalimentaires de lUnit de Recherche en Mcanique des Fluides Applique et Modlisation, Ecole
Nationale dIngnieurs de Sfax, B.P 1173, 3038 Sfax, Tunisia

a b s t r a c t
In the present work, a diffusion-based model was adopted to represent the convective drying behavior of cylindrical
banana samples, taking into consideration the shrinkage along drying. The developed model simulated a signicant
number of situations resulting from the variations of some operating conditions. The temperatures tested were 50,
60, 67 and 70 C, the air velocities were 3, 4 and 4.5 m/s and the relative humidity range of the drying air was from
3.5 to 11.5%. The calculated drying curves were compared to the experimental ones in order to determine apparent
moisture diffusivity. An empirical equation was suggested, describing the apparent moisture diffusivity within the
banana versus product temperature and local moisture content. A good agreement was found between experimental
and calculated drying kinetics.
2008 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Keywords: Convective drying; Diffusive model; Moisture diffusivity; Shrinkage; Cylindrical banana

1.

Introduction

Drying is a complicated process involving simultaneous heat,

mass and momentum transfer phenomena; effective models
are necessary for process design, optimization, energy integration and control. The development of mathematical models to
describe drying processes has been dealt with in many studies for several decades. Undoubtedly, the literature about this
topic includes a limited number of empirical studies. However, the design of dryers is still a mixture of science and
practical experience. Thus, the prediction of Luikov that, by
1985, there would be no further need for empiricism in selecting optimum drying conditions represented an optimistic
perspective which, however, shows that the efforts must be
increased (Luikov, 1970). Presently, more and more sophisticated drying models are becoming available, but a major
question that is still asked concerns the importance of the
measurement or determination of such parameters used in
models as moisture diffusivity and density. The measurement
or estimation of the necessary parameters should be feasible
and practical for the general applicability of a drying model.

With the development of new preservation methods and

the easiness of obtaining fresh fruits from markets, drying is
currently regarded as a method for the diversication of products for consumers convenience. However, drying of shrunk
materials is particularly problematic. In fact, the process
involves simultaneous heat and mass transfers on the one
hand, and volume and surface changes on the other. Under
strict conditions, this results in a system of coupled nonlinear partial differential equations (Kechaou and Malej, 1994;
Zogzas et al., 1996; Hadrich et al., 2008). Since the governing
equations are generally nonlinear in structure, the numerical resolution is necessary. Understanding the heat and mass
transfers and shrinkage phenomena in the product will help
improving drying process, t the drying kinetics and determine the moisture diffusivity coefcient (Kechaou and Malej,
1994; Zogzas et al., 1996; Boudhrioua et al., 2003; Hadrich et al.,
2008).
Therefore, the objective of this study is to develop a diffusive model which uses the second Ficks law and takes into
account the mass and heat transfers as well as the shrinkage in a cylindrical banana sample. The model is used to t

Corresponding author. Fax: +216 74275595.

E-mail addresses: [email protected] (B. Hadrich), [email protected] (N. Kechaou).
Received 2 May 2007; Accepted 20 June 2008
0960-3085/$ see front matter 2008 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.fbp.2008.06.003

food and bioproducts processing 8 7 ( 2 0 0 9 ) 96101

2.

Materials and methods

2.1.

Raw material

97

Nomenclature
A
AH
D
hc
JD,w
Lv
m
Mw
r
R
RH
T
X

area surface of material (m2 )

absolute humidity (kg/kg dry air)
moisture diffusivity coefcient (m2 /s)
convective heat transfer coefcient (W/m2 K)
mass-transfer ux (mol/m2 s)
specic latent heat of vaporization (J/kg)
mass of humid sample (kg)
molar mass of water vapor (g/mol)
sample radius (m)
universal gas constant (8.31451) (J/mol K)
relative humidity (%)
temperature (K, C)
moisture content of sample (kg/kg d.b.)

Greek symbols

solid coordinate in Lagrangean space

density (kg/m3 )


gradient


divergence
Subscripts
0
initial
a
air
eff
effective
eq
equilibrium
s
solid phase
surf
at surface
v
vapor phase
w
liquid phase

Fresh bananas (Cavendish variety) were obtained from a local

market. The fruits were selected according to the required
perfect cylindrical form. The length of the perfectly cylindrical part in each fruit (50 1 mm) was selected for air drying
experiments. The initial radius of banana was 14 1 mm.

2.2.

Moisture content was determined by drying the samples during 24 h in an oven at 105 C (AOAC, 1996). Initial moisture
content of banana ranged between 3.04 and 3.26 kg/kg d.b.

2.3.

Drying kinetics

2.3.1.

Equipment

The experimental apparatus shown in Fig. 1 is a pilot air dryer,

which was designed to perform hot air drying experiments
under the following conditions: air temperature ranging from
25 to 80 C, air velocity between 0.3 to 5 m s1 and relative
humidity ranging from 4 to 90%. In the drying chamber, a
single banana sample was laid on a perforated tray placed
over an analytical balance (METTLER) with a precision of
0.01 g. Periodic weighing was performed throughout the drying experiments at constant time intervals (10 s).

2.3.2.

the experimental drying kinetics and to predict the moisture

content evolution within the material during drying. It was
based on some simple assumptions on conservation equations and on the appropriate initial and boundary conditions.
The shrinkage of the material was also included by establishing an empirical relationship between material displacement
and moisture loss. The determination of moisture diffusivity
in banana is made by comparison between experimental and
predicted drying curves.

Moisture content

Experiments

Six drying experiments were run at different air conditions.

Table 1 shows the experimental conditions of drying air, the
products characteristics and the number of the corresponding drying experiment. Tests 5 and 6 were reserved for the
validation of the established model.
Drying air temperature, relative humidity and velocity
ranges are: 5070 C, 3.611.5% and 34.5 m/s, respectively.

2.4.

Modeling of transfer phenomena

2.4.1.

Assumptions of the model

The product is assumed to be composed of a continuous liquid

element (water) present within a second solid element (dry
matter). The following assumptions are made:

Fig. 1 Experimental drying apparatus.

98

food and bioproducts processing 8 7 ( 2 0 0 9 ) 96101

Table 1 Banana air drying experiments: air conditions and product characteristics
Test no.

Ta ( C)
Th ( C)
RH (%)
AH (g/kg dry air)
va (m/s)
X0 (kg/kg d.b.)
Xeq (kg/kg d.b.)

50
23
7.6
6.120
3
3.10
0.196

50
24
9.5
7.636
4
3.26
0.199

67
27
3.6
6.052
3
3.04
0.138

70
29
5.7
9.690
3
3.06
0.151

50
25
11.4
9.219
4.5
3.25
0.210

60
26
6.4
7.595
3
3.22
0.178

The initial distribution of moisture content and temperature of the banana are uniform.
The water vaporization takes place only at the surface.
During drying, the temperature in the banana sample is
considered to be uniform at a xed drying time.
The shrinkage is not negligible.
Physical and thermal proprieties are assumed to be functions of local moisture content and temperature.
The moisture transport phenomenon within the banana
during drying can be described by the second Ficks law.
The transfer is considered only as radial.

2.4.2.

 X
t


 s X
+v

 (s D X)

=

(1)

In order to take shrinkage of the material into account, Eq. (1)

has to be written within a reference movement at the velocity
of the solid phase (Lagrangean system of coordinates). In this
reference, the spatial coordinate is denoted as . Finally, Eq. (1)
becomes:
s

 DX 
Dt

 (s D X)

=

(2)

Taking into consideration that s =  w = /(X + 1) (Jomaa,

1991; Talla et al., 2004), Eq. (2) becomes:
s

 DX 
Dt

= div

 D 

X

(3)

1+X

Changing the coordinates system requires knowing the deformation ratio (s /s0 ). For the ideal deformation case where
shrinkage is linear versus moisture content, this ratio could
be written as (Kechaou and Malej, 1994):
s
s0

= (1 + 1.4855 X)

(4)

This equation is also used by Lima et al. (2002) and Talla et

al. (2004) for ellipsoidal and slice conguration, respectively.
The authors used the same ranges of moisture content and
temperature as those investigated in this work.
Finally, Eq. (3) becomes:

 DX 
Dt

 a + a X
1
2

Mass balance equation

The equation of mass conservation (Hadrich et al., 2008) could

be written as follows:
s

In Eq. (5), D(X, T) is introduced as a parametric model depending on local moisture content and temperature of the product.
The temperature dependence versus apparent moisture diffusivity could be described by the ARRHENIUS relationship
(Zogzas et al., 1996; Kechaou, 2001; Hadrich et al., 2008). However, the inuence of moisture content in the estimated values
of apparent moisture diffusivity has not yet been formulated
into a generally accepted model.
Several equations as those presented by Zogzas et al. (1996),
Kechaou (2001) and Fernandes and Rodrigues (2007) were
tested. Finally, expression (7) (Kechaou, 2001) that best ts the
experimental values was chosen:

1 


 Deff (X, T)

 X  


(5)

with
Deff (X, T) =

D(X, T)
(1 + 1.4855 X)

 r 2


(6)

D(X, T) = a0 exp

 

a3 + a4 X

a5 + a6 exp(a7 X)
RT

exp


(7)

where ai (0 i 7) are the parameters to identify.

2.4.3.

Heat balance equation

The heat balance equation could be represented according to

the hypothesis:
m Cp

dT
= A hc (Ta T) Lv A JD,w, surf
dt

(8)

with JD,w,surf = km (Mw /(RTa ))(Pv,surf Pva ): the mass ux at the

product surface and Cp = ((1412 + 4186X)/(1 + X)): the specic
heat of banana (Kechaou, 2001).

2.4.4.

Initial and boundary conditions

Initially, local moisture content and temperature of banana

are uniform for any : at t = 0, T = T0 and X = X0 .
At the products center, all gradients are equal to zero: for
t > 0 and for  = 0, (X/) = 0.
At the surface ( =  s ), mass ux density is given
 by Ficks
law: for t > 0 and  = s JD,w,surf = s0 Deff (X, T) r X


2.5.

Numerical resolution

The system of equations describing mass transport consists of a partial differential Eq. (5) associated with one
initial and two boundary conditions. The eld 0   s is
divided into N equal subdivisions. In this case, the obtained
system consisted of N + 1 coupled nonlinear ordinary differential equations where the unknown variable was the
local moisture content. To solve the equations system, the
implicit nite difference method and the ode45 Matlab
solver (based on Runge-Kutta method) have been used
with an assumed dependence of apparent moisture diffusivity. The heat balance, where the unknown variable is
the temperature of the material, was associated with the
initial condition. At each time span, the value of X was

99

food and bioproducts processing 8 7 ( 2 0 0 9 ) 96101

calculated at each node and the global value of T was

obtained.
The average moisture content at the instant t was calculated by:
t =
X

X(i , t) dV

(9)

where V is the volume of the sample (m3 ).

2.6.

Parameters used in the model

The model is based on a set of physical and equilibrium proprieties and transfer coefcients:
Equilibrium moisture content (Xeq ) was the calculated equilibrium moisture content determined by using the GAB
model to t the desorption isotherms (Kechaou and Malej,
1999).
Convective heat transfer coefcient was calculated for an
air ow established around an isolated cylinder by using a
relationship to estimate the value of the Nusselt (Whitaker,
1972):
Nu =

hc d
= (0.4 Re1/2 + 0.06 Re1/2 ) Pr2/5


The
mass-transfer
km = (hc /(a Cpa ))

coefcient

is

for Re < 105

expressed

by:

The inputs to the mathematical models are: physical properties of drying air, initial sample dimension and material
shrinkage coefcient.

2.7.

Statistical analysis

The parameters ai of Eq. (7) were determined by minimizing

the root mean square (RMS) differences between experimental and predicted drying curves. The RMS was calculated as
follows:


RMS = 100

n
((Xi,exp
i=1

Xi,cal )/Xi,exp )

n1

Results and analysis

3.1.

Drying kinetics

A constant-rate period was not observed in any of the drying

experiments. Therefore, the entire drying process for banana
occurs in the range of the falling-rate period. Moisture content and drying rate decrease continuously with diminishing
drying time. This shows that diffusion is the dominant physical mechanism governing moisture movement in the product.
The results were generally in agreement with some previous ndings on banana drying (Kechaou and Malej, 1994;
Kechaou, 2001; Gbaha et al., 2007; Fernandes and Rodrigues,
2007).

3.2.

Identication of the apparent moisture diffusivity

The parameters of Eq. (7) identied for all experimental conditions by using the model are shown in experimental (11) as
follows:

 1 + 22 X 

D(X, T) = 6.95 106 exp

 

1 + 14 X

exp

22 103 + 17.5 103 exp(1.5 X)

RT


(11)

(10)

where Xi,exp and Xi,cal were the experimental and the predicted
average moisture contents respectively, and n was the number of experimental points. The computation was preformed
with experimental kinetics by using average moisture content
(X = f(t)).

3.

Fig. 2 Variation of banana moisture content versus drying

time at various air conditions.

Fig. 2 shows changes in the moisture content of banana during

hot air drying at different conditions. The moisture content
decreases gradually while the drying time elapsed, exhibiting a smooth downward curve. The air temperatures have
a major effect on the drying kinetics (Fig. 2) (Kechaou and
Malej, 1994; Lahsasni et al., 2004; Boudhrioua et al., 2003;
Hadrich et al., 2008). A drying rate was calculated from moisture content data. Fig. 3 shows the variation of the drying
rate versus the moisture content of the banana samples. For
all experiments, the drying rate decreased continuously from
the initial moisture content to approximately 0.1 kg/kg d.b h.

The identied apparent moisture diffusivity could be used for

the air temperatures ranging from 50 to 70 C and for the initial
moisture content ranging from 3.04 to 3.26 kg/kg d.b.
Table 2 shows the minimum values of RMS for the rst
four conditions of the experiment. It can be observed that the
suggested model shows a good tting of experimental drying
kinetics (RMS values are less than 10%).

Fig. 3 Variation of banana drying rate versus average

moisture content at various air conditions.

100

food and bioproducts processing 8 7 ( 2 0 0 9 ) 96101

Table 2 RMS differences between predicted and

experimental drying curves
Number of the experiment
1
2
3
4

RMS (%)
2.85
5.46
8.24
8.57

Fig. 6 Variations of predicted and experimental average

moisture content versus time obtained at Ta = 60 C,
RH = 6.4%, va = 3 m/s (Test 6).
values obtained in both cases were less than 5%. The suggested model was validated and was largely satisfactory for
overall experimental measurements: 50 Ta 70 C, 3, 4 and
4.5 m/s and 3.5 RH 11.5%.
Fig. 4 Variations of predicted and experimental average
moisture content versus time obtained at Ta = 50 C,
RH = 7.6% and va = 3 m/s (Test 1).
Fig. 4 represents a comparison between the average values
of calculated banana and the experimental moisture content as function of drying time obtained for Test 1 (Ta = 50 C,
RH = 7.6%, va = 3 m/s). The observed differences are due to
some simplications made during model development such
as separating mass from heat transfer and the negligence of
axial transfer.

3.3.

Validation of the suggested model

The model validation was carried out through a comparison of

predicted average moisture values with those obtained experimentally in Tests 5 and 6 (Figs. 5 and 6, respectively). The RMS

Fig. 5 Variations of predicted and experimental average

moisture content versus time obtained at Ta = 50 C,
RH = 11.4% and va = 4.5 m/s (Test 5).

3.4.

Prediction of moisture distribution

Fig. 7 shows the banana moisture distribution as a function

of effective radius and drying time predicted for drying at
Ta = 50 C, RH = 7.6% and va = 3 m/s. Moisture content evolution shows a falling parabolic curve during the drying and,
after a certain time, it was ending at the equilibrium moisture
content within the sample.
Every curve of Fig. 7 refers to the moisture distribution
from the bottom to the surface of the sample after the drying
time indicated on the curve. It can be seen that the moisture content at the sample surface decreases sharply with
the increase of the drying time. The moisture content gradients from the surface to the bottom decrease when drying
time increases. The corresponding moisture content near the
bottom decreases steadily with time.

Fig. 7 Predicted moisture distributions in banana sample

as a function of both position and drying time obtained at
Ta = 51 C, RH = 7.6%, va = 3 m/s.

food and bioproducts processing 8 7 ( 2 0 0 9 ) 96101

The sample radius decreases during drying. This result

indicates the sample shrinkage considered in the established
model.

4.

Discussion and conclusion

Fundamental equations for the liquid diffusion in cylindrical

banana samples, considering shrinkage effect, were developed. By investigating the drying curves obtained by modeling
and experimenting, an empirical equation for the apparent
moisture diffusivity coefcient of banana has been presented
as a function of temperature and banana moisture content.
From the results presented, it can be concluded that:
The drying kinetics of cylindrical banana, presenting a
shrinkage phenomenon, can be described by an isothermal
model with moisture diffusivity coefcient presenting an
ARRHENIUS-type dependent on temperature.
The variation of banana apparent moisture diffusivity
coefcient versus moisture content and temperature is suggested for drying conditions ranging from 51 to 70 C for
temperature and from 3.04 to 3.26 kg/kg d.b for initial moisture content.
This model accounts for the changes of moisture distribution in the cylindrical banana.
The model can be adapted, with small modications, to
describe the drying process of other cylindrical food products in a tunnel dryer with the appropriate properties
(variable or not), and with other boundary conditions.

Acknowledgments
The authors would like to thank Mr. Ali AMRI, MA in
English Theoretical and Applied Linguistics, and his association English Polisher for the exhaustive proofreading of the
papers English.

References
Association of Ofcial Analyst Chemists (AOAC), 1996, Ofcial
method of analysis 952.08.
Boudhrioua, N., Bonazzi, C. and Daudin, J.D., 2003, Estimation of
moisture diffusivity in gelatinstarch gels using

101

time-dependent concentrationdistance curves at constant

temperature. Food Chemistry, 82(1): 139149.
Fernandes, F.A.N. and Rodrigues, S., 2007, Ultrasound as
pre-treatment for drying of fruits: dehydration of banana.
Journal of Food Engineering, 82: 261267.
Gbaha, P., Yobouet Andoh, H., Kouassi Saraka, J., Kamnan Koua,
B. and Tour, S., 2007, Experimental investigation of a solar
dryer with natural convective heat ow. Renewable Energy, 32:
18171829.
Hadrich, B., Boudhrioua, N. and Kechaou, N., 2008, Drying of
Tunisian Sardine (Sardinella aurita)experimental study and
three-dimensional transfer modeling of dying kinetics.
Journal of Food Engineering, 84(1): 92100.
Jomaa, W., 1991, Schage de matriaux fortement dformables:
prise en compte de la vitesse de retrait, Thse de doctorat de
lUniversit de Bordeaux I, France.
Kechaou, N. and Malej, M., 1994, Evaluation of diffusion
coefcient in the case of banana drying, in Drying
Technology94, vol. A, Mujumdar, A.S. (ed) , pp.
841848.
Kechaou, N. and Malej, M., 1999, Desorption isotherms of
imported Bananaapplication of the GAB theory. Drying
Technology, 13(6): 12011213.
Kechaou, N., 2001, Etude thorique et exprimentale du
processus de schage des produits agroalimentaires, Thse de
doctorat dtat es-Sciences-Facult des Sciences de Tunis,
Tunisie.
Lahsasni, S., Kouhila, M., Mahrouz, M., Idlimam, A. and Jmali, A.,
2004, Thin layer convective solar drying and mathematical
modelling of prickly pear peel (Opuntia cus indica). Energy, 29:
211224.
Lima, A.G.B., Queiroz, M.R. and Nebra, S.A., 2002, Simultaneous
moisture transport and shrinkage during drying solids with
ellipsoidal conguration. Chemical Engineering Journal, 86:
8385.
Luikov, A.V., 1970, Prognosis of the development of science of
drying capillary-porous colloidal materials. International
Chemical Engineering, 10: 599604.
Talla, A., Puiggali, J.R., Jomaa, W. and Jannot, Y., 2004, Shrinkage
and density evolution during drying of tropical fruits:
application to banana. Journal of Food Engineering, 64(1):
103110.
Whitaker, S., (1972). Fundamental Principals of Heat Transfer.
(Pergamon Press Inc, New York).
Zogzas, N.P., Maroulis, Z.B. and Marinos-Kouris, D., 1996,
Moisture diffusivity data compilation in foodstuffs. Drying
Technology, 14(10): 22252253.