Dependence of the effective diffusion coefcient of moisture with thickness

and temperature in convective drying of sliced materials. A study on slices

of banana, cassava and pumpkin
W.J.N. Fernando

, H.C. Low, A.L. Ahmad

School of Chemical Engineering, University Sains Malaysia, Nibong Tebal, 14300 Pulau Pinang, Malaysia
a r t i c l e i n f o
Article history:
Received 14 June 2010
Received in revised form 6 September 2010
Accepted 9 September 2010
Available online 17 September 2010
Keywords:
Convective drying
Effective diffusion coefcient
Drying rate
Slice thickness
a b s t r a c t
Several studies have indicated that effective diffusion coefcients of slices apparently vary with the thick-
ness of the samples. Eventhoughthe effective diffusioncoefcients have beenobserved to be dependent on
the square of the slice thickness, a theoretical explanation to this behavior is not available to date. A theo-
retical model is formulated herein, in order to correlate effective diffusion coefcient of moisture in the
slices with the slice thickness. Experiments are carried out for drying of slices of different thicknesses of
banana, cassava and pumpkin in order to evaluate the effective diffusion coefcients. The model is found
to describe the variation of the effective diffusion coefcient with slice thickness very satisfactorily. A pos-
sibilityof theestimationof the axial andradial diffusioncoefcients of moisture intheslices is alsooutlined.
2010 Elsevier Ltd. All rights reserved.
1. Introduction
In convective drying of wet materials, moisture diffuses from
the interior of the solid and reaches the outer surfaces. The mois-
ture reaching the outer surfaces thereafter evaporates into the air
due to convection (Sherwood, 1929, 1931, 1936; Gilliland and
Sherwood, 1933). Combination of several other transport mecha-
nisms including capillary ow, liquid diffusion, vapour diffusion,
viscous ow have also been shown to govern such moisture move-
ment (Valentas et al., 1997). However, diffusion is generally ac-
cepted as the principal mechanism of moisture transport in
drying processes (Jorge et al., 2002). Moisture diffusivity in mate-
rials such as starch can strongly be affected by the physical struc-
ture of the material, particularly the bulk porosity. Moisture
diffusivity in food materials such as fruits and vegetables can be
estimated by the analysis of respective drying rates.
General models for drying of slices assume that diffusion takes
place in one direction from inside to surface of the sample (Minh-
Hue and William, 2007). Thus, the diffusion from other direction is
generally assumed to be negligible.
The conventional equation (Tutuncu and Labuza, 1996) used in
order to estimate the effective diffusion coefcient of moisture
within drying samples is shown in Eq. (1). Eq. (1) is the widely used
equation for determining the effective diffusion coefcient (D
eff
).
Usually plots of lnM
R
(t) versus t were obtained in order to obtain
the gradient of the graph as p
2
D
eff
/4L
2
. The value of D
eff
was esti-
mated thereafter from experiments.
lnM
R
t ln 8=p
2
p
2
D
eff
=4L
2
t: 1
Tutuncu and Labuza (1996) studied the effective diffusion coef-
cients of two types of breakfast cereal. These effective diffusion
coefcients have been observed to be dependent on the square of
thickness of breakfast cereal.
Johnson et al. (1998) carried out drying of slabs banana of thick-
nesses varying from 5 to 20 mm at 65 C as the drying tempera-
ture. The effective diffusion coefcients have been found to be
5.12 10
10
and 1.80 10
9
m
2
/s for 7.2 and 17.8 mm respec-
tively. The results have clearly shown that slab thicknesses affect
the effective diffusion coefcients.
In an independent study, Aysun et al. (2002) have conrmed
that the above behavior of dependence of the effective diffusion
coefcients on slabs thicknesses. This study investigated the drying
rates of samples of thicknesses 0.71, 1.53, 2.20 and 2.86 mm
respectively of grape leather at a temperature 75 C. Effective dif-
fusion coefcients of 4.76 10
11
, 1.73 10
10
, 2.73 10
10
and
3.76 10
10
m
2
/s have been obtained for samples of thicknesses
of 0.71, 1.53, 2.20 and 2.86 mm respectively. The results also have
indicated that the effective diffusion coefcients increased with
increasing slab thicknesses.
Senol and Medeni (2002) have carried out drying experiments
for slabs of 1, 3 and 6 mm slab thickness of wheat-ouryogurt
mixture at 80 C. The dependence of the effective diffusion coef-
cient on the slab thickness has been conrmed. Here the effective
0260-8774/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfoodeng.2010.09.004

Corresponding author. Tel.: +60 17 424 1617; fax: +60 45941013.

E-mail address: [email protected] (W.J.N. Fernando).
Journal of Food Engineering 102 (2011) 310316
Contents lists available at ScienceDirect
Journal of Food Engineering
j our nal homepage: www. el sevi er . com/ l ocat e/ j f oodeng
diffusion coefcients of 0.90 10
10
, 3.30 10
10
and 5.00
10
10
m
2
/s have been obtained for slabs of thicknesses of 1, 3
and 6 mm respectively. The effective diffusion coefcients have
been shown to increase with increasing slab thicknesses.
Minh-Hue and William (2007) studied the inuence of slab
thickness on the kinetics of hot air drying of banana at 1m/s air
velocity and at a temperature of 70 C drying. The values of the
effective diffusion coefcients of banana changed from 7.87
10
10
to 2.27 10
9
m
2
/s when the thicknesses were changed
from 1 to 2 cm. The values of the effective diffusion coefcient
for slabs of 2 cm thickness has been observed to be nearly three
times greater than value obtained for slabs of 1 cm thickness. This
difference has been attributed to the deviation of the assumption
with the real situation. The assumption assumed that diffusion
took place in one direction and drying only at one side way
whereas in practise diffusion takes place in all direction and drying
takes place in all side of the slabs. It was also pointed out that the
assumption could only be valid for thin slabs since the edge effect
(i.e. moisture diffusion has from other sides of the slabs) could be
regarded to be negligible in thin samples. The necessity for
accounting for the diffusion of moisture arising due to diffusion
in three dimensions in drying of thick slabs has been pointed out
in their study. In a separate study, the relationship between thick-
ness and effective diffusion coefcient has been explained as aris-
ing due to contributory edge effect. Hardening effect and possible
existence of isolated air between the particles in the case of thick
slices could also contribute to such deviation (Minh-Hue and
William, 2007; Tutuncu and Labuza, 1996; Yan et al., 2008).
Nearlyall the above observations indicatethat the effective diffu-
sioncoefcients inslicedmaterials tendtohavea dependenceonthe
thickness of the samples. However, inthe theory of diffusion, the dif-
fusion coefcient need to be an parameter dependant on the pore
sizes and shapes, their interconnectivity and tortuosity, chemical
and physical structures of the material and the properties of the dif-
fusing medium. Therefore, there is a need for an explanation for its
dependence on the thickness is of very much importance in order
to have a correct insight of the diffusion process. Apart from the
empirical observations, theoretical explanations on variation of
effective diffusion coefcient with thickness are not available at
present. This study attempts to develop a theoretical model in order
tocorrelate the effective diffusioncoefcient withthe thickness. The
model is validated by using experimental data herein.
2. Materials and methods
2.1. Experiment rig
A schematic diagram of the pilot plant test dryer used in this
study is shown in Fig. 1. The dryer used is a conventional laboratory
convective tray dryer. The tray dryer has an insulated drying
chamber (2) (170.2 20.7 20.7 cm) in which trays (4) are lo-
cated at the middle of the chamber with opening door (5) which
is used for insertion or removal of the test samples. The trays are
hanged on an electronic balance (1) (Shimadzu UX6200H) which
is capable measuring weight of the sample and tray directly during
the experiment. The measurement range of the balance is 0.5
6200 g and accuracy 0.01 g. This dryer consists of heating ele-
ments (6) and a fan (8) capable of achieving air ow rates up to
235 ft
3
/min. Eight heating elements of capacity of 375 W each
are located on the upstream of the drying chamber. Blower fans
are installed upstream of heating elements and provide air to the
drying process. The hot air owing over the samples picks up
the moisture from the drying samples, and leaves at point (10) of the
downstream of dryer. The air velocity can be varied by regulating
the fan speed. It is measured in m/s using an anemometer (Turbo
meter Electronic Wind Speed Indicator) which is located at the
downstream of the drying chamber. A control panel (9) which is
located outside the dryer; controls the temperature (heater power)
and the fan speed.
2.2. Preliminary experiment
A preliminary experiment was initially carried out in order to
determine the initial moisture content of each selected materials.
The initial weight of the selected test samples (W
0
) and the nal
weight (W
f
) after overnight drying on the samples an oven at
110 C were noted. The initial moisture fraction of each test sample
of material was estimated using Eq. (2):
M
0
W
0
W
f
=W
0
2
2.3. Sample preparation
Fresh banana, cassava and pumpkin were obtained from the lo-
cal supermarket. The slices of materials were prepared by peeling
the skin and then cut into slices of cylindrical shape of 0.5, 0.75,
1.0, 1.25 and 1.5 cm thickness. The thicknesses and the diameters
of the cylinders are measured by vernier caliper (Daisen) of the
range of 0150 mm and measurement accuracy of 0.05 mm. Sam-
ples of banana, cassava and pumpkin within a variation of 2% in
thickness and each diameter of 1.5, 2.0 or 2.5 cm were selected
for drying experiments described below.
2.4. Drying procedure
Approximately 510 selected samples of a selected species
were placed in a pre-weighed petri dish. The temperature was
set to a desired value (85, 90 or 95 C). Air velocity was set to
Nomenclature
a average radius of the cylindrical slice (m)
C moisture concentration (kg water/m
3
of sample) at a
location within the sample at the working temperature
C
0
initial moisture concentration (kg water/m
3
of sample)
C
e
equilibrium moisture concentration (kg water/m
3
of
sample)
C
*
moisture concentration ratio, given by (C C
e
)/C
0
C
e
)
D
eff
effective diffusion coefcient (m
2
/s)
D
r
lateral diffusion coefcient [in radial direction] (m
2
/s)
D
z
longitudinal diffusion coefcient [in axial direction]
(m
2
/s)
thickness of samples (m)
M(t) actual moisture content of a sample at time, t (kg water)
M(0) initial moisture content of a sample at t = 0 (kg water)
M (1) equilibrium moisture content of a sample at t ?1 (kg
water)
M
R
(t) M
0
moisture ratio at time, t Initial moisture fraction (initial
weight of moisture/weight of original sample)
r radial distance from the central axis of the slice (m)
t drying time (s)
W
0
initial weight of sample (kg)
W
f
nal weight of sample (kg)
Z distance of a point within slice along the axis of cylin-
drical slice from the bottom surface of slice (m)
W.J.N. Fernando et al. / Journal of Food Engineering 102 (2011) 310316 311
2 m/s at the exit of the fan throughout experiments. The system
was allowed to run for about half an hour without samples to reach
the steady state. A petri dish with samples was placed thereafter in
the tray dryer and the weight was recorded by mean of the bal-
ance. The door of the dryer was tightly closed in order to ensure
that there were no leakages of hot air through the door. Weights
of the samples and tray were weighed every 15 min for rst 1 h,
every 30 min for next 2 h and every 1 h subsequently. The nal
dry weight of sample (W
f
) in each petri dish was estimated using
Eq. (2) with the knowledge of previous evaluated value of M
0
(in
Section 2.2) and W
0
. Continuous data collection of each experiment
was suspended when the weights of samples reached 10% above
the nal dry weight. However, the drying process was allowed to
continue under the same conditions until the weight of the tray
did not show any change. At this point the experiment was com-
pletely terminated. The weight of the sample at this point was
noted and the equilibrium moisture content was thereafter evalu-
ated. Each experiment was repeated twice for reproducibility of
data. The experiments were further repeated if the average varia-
tion of moisture content during the experiment was observed to
be appreciable (more than 5%). The dry and wet bulb temperatures
of air entering the heater were noted regularly.
3. Theoretical
An attempt is made in this study in order to develop a theoret-
ical model for correlating the effective diffusion coefcients with
slice thicknesses. This model assumes an unsteady rate of diffusion
of moisture within a slice undergo during the drying process. This
process assumed that there is no shrinkage of the materials
throughout the process. The diffusion process is assumed to take
place in both axial and radial directions of the sliced materials
leading to moisture loss from all exposed surfaces. The material
of the slice is assumed to be cylindrical.
The overall distribution of moisture concentration with time at
a given position within the slice can be expressed in cylindrical
coordinates as (Crank, 1975):
@C=@t D
r
=rf@r @C=@r=@rg D
z
@
2
C=@z
2
3
If the slice is assumed to have an initial concentration of moisture
(C
0
) throughout, then an initial condition C = C
0
at t = 0 for all Z
and r could be formulated. Also at t ?1, C is assumed to reach
its equilibrium value C
e
. The slice is assumed not to undergo defor-
mation during the drying process.
If C
*
is assumed to denominate the factor (C C
e
)/(C
0
C
e
), then
Eq. (3) can be expressed as
@C

=@t D
r
=rf@r @C

=@r=@rg D
z
@
2
C

=@z
2
4
A method of separation of variables in Eq. (4) is attempted in the
form:
C

f r; t z; t 5
Substituting Eq. (5) in Eq. (4) and separating the terms containing z
and r yields
@z; t=@t D
z
@
2
z; t=@z
2
6
and
@f r; t=@t D
r
=r @r @f r; t=@r=@r 7
Eqs. (6) and (7) are seen to be similar to the independent axial and
radial diffusion coefcient equation for a cylinder.
The actual moisture content, M(t) at time, t within the slice
could be expressed as
Mt
Z
z
0
Z
r
0
2prCdrdz 8
Eq. (8) could be employed together with Eq. (5) in order to obtain
Mt C
0
C
e
2
z
0
Z
a
0
f2pr f r; t z; tdrgdz C
e
pa
2
9
Application of Eq. (9) at t = 0 and C
*
= 1 yields
M0 C
0
pa
2
10
Application of Eq. (9) at t ? 1 and C
*
?0 yields:
M1 C
e
pa
2
11
M
R
(t) which is the moisture ratio at a given time, t can be dened as
M
R
t Mt M1=M0 M1 12
Combination of Eqs. (9)(11), gives an expression for M
R
(t) as
M
R
t F
R
t P

t 13
Fig. 1. Schematic diagram of pilot plant test dryer.
312 W.J.N. Fernando et al. / Journal of Food Engineering 102 (2011) 310316
where
F
R
t 1=pa
2

Z
r
0
f2prf r; tgdr 14
P

t 1=
Z

0
z; tdz 15
The initial and boundary conditions of Eq. (13) could be expressed
as M
R
(0) =1 and M
R
(1) =0 for all r and Z.
General solutions for F
R
(t) and P

(t) which are obtained based on

the solution of Eqs. (6) and (7) are available (Crank, 1975) as:
F
R
t
X
1
n1
4p=a
2
n
exp D
r
a
2
n
t=a
2
16
where a
n
values are the positive roots of the equation as shown:
J
0
a
n
0 17
where J
0
(x) is the Bessel function of zero order
The roots (a
n
) can be expressed as (Polyanin and Zaitsev, 2003):
a
n
2:4 3:13n 1 18
for n = 1, 2, . . ., with a maximum error 0.2% and
P

t
X
1
n0
8=2n 1
2
p
2
expD
z
2n 1
2
p
2
t=4
2
19
Thus, Eq. (13) could be expressed as:
M

t
X
1
n0
4p=a
2
n
expD
r
a
2
n
t=a
2

X
1
n0
8=2n 1
2
p
2
expD
z
2n 1
2
p
2
t=4
2
20
In practice and in many previous studies (Andrieu and Stamatopoulos,
1986; Magee and Wilkinson, 1985; Vaccarezza and Chirife, 2006),
the radial diffusion of moisture had been assumed to be insigni-
cant in estimating the effective diffusion coefcient. These studies
have estimated the effective diffusion coefcient (D
eff
) by assum-
ing no lateral diffusion coefcient (D
r
= 0) resulting a simplied
equation:
M
R
t
X
1
n0
8=2n 1 p
2
expD
eff
2n 1
2
p
2
t=4
2
21
Eq. (1) which has been used to evaluate the effective diffusion coef-
cient, D
eff
is a simplied version of Eq. (21) with terms n P1 as-
sumed insignicant.
Eqs. (20) and (21) can be simplied to,
X
1
n0
"
8=2n 1 p
2
expD
eff
2n 1
2
p
2
t=4
2

X
1
n1
4=a
2
n
expD
r
a
2
n
t=a
2

#
0 22
Here it is assumed that D
eff
> D
z
because estimation of D
eff
assumes
no diffusion from the sides of the slice thereby leading to higher val-
ues of effective diffusion coefcients compared with actual diffu-
sion coefcients.
Therefore it can be observed that each exponential term of the
two series of Eq. (22) consist of a summation of exponential terms
of negative exponents,
Each such exponential term is of the form exp(wt). Exponen-
tial terms of the form exp(wt) is convergent and could be ex-
pressed (Bird, 2007) as a convergent series 1 (wt) + (wt)
2
/
2! (wt)
3
/3! + Hence the summation of exponent terms of Eq.
(22) could also be expressed as summation of a convergent series
as
T
0
T
1
t T
2
t
2
T
r
t
r
0 23
where T
r
is the multiplier of the term t
r
of the series.
It should be noted that due to convergence of the series,
jT
0
j > jT
1
tj > jT
2
t
2
j > jT
r
t
r
j 24
The term T
0
is evaluated as
T
0

X
1
n1
8=2n 1
2
p
2

X
1
n1
4=a
2
n
25
Theoretical value of the summation of each of the series tends to
unity as n ?1. Hence T
0
becomes zero.
The term T
1
which is the coefcients obtained by expansion of
each of the exponential series in Eq. (22) can be expressed as
T
1
2D
eff
D
z
p
2
=
2
eD
r
=a
2
26
Assuming the convergence of the series makes terms with higher
exponents of t P2 are less signicant compared with the term con-
taining t, an expression for Eq. (22) can be obtained as
2D
eff
D
z
p
2
=
2

4D
r
=a
2
0 27
Eq. (27) can therefore be re-arranged as
D
eff
D
z
2D
r
=p
2
a
2

2
28
From the model Eq. (28), it can be observed that the plot of D
eff
ver-
sus
2
needs to be a straight line with intercept D
z
and gradient 2D
r
/
a
2
p
2
. Hence, in a plot of D
eff
versus
2
, a linear relationship will dem-
onstrate the validity of the model equation with D
z
as the intercept.
The gradient presents a value 2D
r
/a
2
p
2
from which the radial diffu-
sion coefcient (D
r
) could be estimated.
4. Results and discussion
4.1. Drying curves
Figs. 24 represent the plots of M
R
versus t obtained fromexper-
imental data obtained for banana, cassava and pumpkin respec-
tively at temperature 85 C. The patterns of variations of M
R
versus t observed in the experiments are compatible with the
typical variations of M
R
versus t obtained by other researchers
(Johnson et al., 1998; Karim and Hawlader, 2005; Doymaz, 2007;
Minh-Hue and William, 2007; Nawirska et al., 2009). Concerns
were extended in this study to assess the possible causes of errors
in the experimental data. The variations of air humidity during the
experiment, the precision errors in the measurement of the weight
of the samples and the control of operating temperature were
given thought. The variations of humidity were monitored by reg-
ular monitoring of the wet bulb and dry bulb temperatures of
incoming air before entering the heater. The variation of the wet
bulb and dry bulb temperatures were observed to be 30 0.8 C
and 25 0.6 C. The extreme values of specic humidities were
therefore evaluated to be 0.016309 and 0.018943 kg moisture per
kg of moist air respectively. The respective partial pressures of
moisture in the air were evaluated to be 2.66 and 3.09 kPa respec-
tively. Taking into consideration the standard saturation vapour
pressures of moisture at 85, 90 and 95 C as 57.8, 70.11 and
84.53 kPa respectively, it can be observed that the variation of
the partial pressure driving force for the drying operation were
54.93 0.21, 67.24 0.21, 81.66 0.21 kPa for the respective
temperatures. Therefore it can be observed that the expected
variance of the drying rates due to the humidity uctuations is be-
tween 0.25% and 0.4% from the mean values for the experiments
conducted which could be regarded as not very signicant.
W.J.N. Fernando et al. / Journal of Food Engineering 102 (2011) 310316 313
4.2. Effective diffusion coefcient (D
eff
)
Table 1 shows the values of effective diffusion coefcients
(D
eff
) of moisture for banana, cassava and pumpkin for different
temperatures (85, 90 and 95 C) and slice thicknesses (0.5, 1.0
and 1.5 cm) evaluated using Eq. (1). It can be observed that D
eff
increased with increasing of temperature and slice thickness in
conformity with previous researchers (Johnson et al., 1998; Aysun
et al., 2002; Senol and Medeni, 2002; Doymaz, 2007; Minh-Hue
and William, 2007). The D
eff
values are found to be within the gen-
eral range of 10
11
10
9
m
2
/s for agricultural products (Madamba
et al., 1996). The values of the effective diffusion coefcients for
drying of banana obtained by previous researchers ranged from
5.12 10
10
to 1.80 10
9
m
2
/s (Johnson et al., 1998) and 7.87
10
10
to 2.27 10
9
m
2
/s at 6570 C (Minh-Hue and William,
2007). Also values of the effective diffusion coefcients ranging
from 3.88 10
10
to 9.38 10
10
m
2
/s (Doymaz, 2007) have been
obtained for drying of the pumpkin at 5060 C. Comparison of
Fig. 2. Plot of moisture ratio, M
R
versus time, t for banana at temperature 85 C.
Fig. 3. Plot of moisture ratio, M
R
versus time, t for cassava at temperature 85 C.
Fig. 4. Plot of moisture ratio, M
R
versus time, t for pumpkin at temperature 85 C.
314 W.J.N. Fernando et al. / Journal of Food Engineering 102 (2011) 310316
these values with the diffusion coefcients obtained for drying at
8595 C in this experiment showed compatibility with the previ-
ous published data.
4.3. Diffusion coefcient, D
r
and D
z
Figs. 57 present the plots of effective diffusion coefcients
(D
eff
) versus
2
at three different temperatures for banana, cassava
and pumpkin respectively. As expected, the plots were seen to
yield straight lines with satisfactory correlation coefcients (R
2
)
between 0.975 and 0.982 for all plots. From the gures, the diffu-
sion coefcient, D
z
and D
r
was estimated from the intercept and
gradient of the plots respectively.
Table 1
Effective diffusion coefcient, D
eff
for three materials.
Materials Thickness (cm) Effective diffusion coefcient, D
eff
10
9
(m
2
/s)
Temperature (C)
85 90 95
Banana 0.5 4.3137 4.3370 4.4193
1.0 4.3199 4.3449 4.4271
1.5 4.3263 4.3529 4.4369
Cassava 0.5 5.0869 5.1272 5.1905
1.0 5.0941 5.1361 5.2007
1.5 5.1015 5.1450 5.2110
Pumpkin 0.5 4.4792 4.5034 4.5907
1.0 4.5320 4.5554 4.6450
1.5 4.5849 4.6095 4.6979
Fig. 5. Plot of effective diffusion coefcient (D
eff
) versus
2
for banana at temperature 85, 90 and 95 C.
Fig. 6. Plot of effective diffusion coefcient (D
eff
) versus
2
for cassava at temperature 85, 90 and 95 C.
Fig. 7. Plot of effective diffusion coefcient (D
eff
) versus
2
for pumpkin at temperature 85, 90 and 95 C.
W.J.N. Fernando et al. / Journal of Food Engineering 102 (2011) 310316 315
Table 2 presents the diffusion coefcients, D
z
and D
r
obtained
for banana, cassava and pumpkin at temperature 85, 90 and
95 C respectively. It can be seen that values of D
z
have been higher
than the values of D
r
for the three materials indicating an-isotropic
behavior of the materials tested. This has to be expected particu-
larly in agriculture products where internal textures could be dif-
ferent from the longitudinal direction of growth (axial direction
of slice herein) compared with lateral (radial direction of slice
herein) direction of growth (Boudhrioua et al., 2002; Chen and
Ramaswamy, 2002; Gonalves et al., 2005).
5. Conclusion
Experiments for drying of slices of banana, cassava and pump-
kin of thicknesses 0.51.5 cm were carried out under three drying
temperatures (85, 90 and 95 C). The drying rates and effective dif-
fusion coefcient, D
eff
were studied in this study. A theoretical
model was developed in order to correlate the D
eff
with slice thick-
ness . It was observed that the model enabled the estimation of
the diffusion coefcients, D
z
and D
r
from experimental data. The
D
z
and D
r
were estimated using effective diffusion coefcient, D
eff
based on a plot against value of
2
. It was also observed that the
values of D
z
and D
r
increased with increasing of the temperature
as expected. The values of the axial diffusion coefcient (D
z
) were
found 4.31E09, 4.34E09 and 4.42E09 m
2
/s for banana, 5.09E09,
5.13E09 and 5.19E09 m
2
/s for cassava and 4.47E09, 4.50E09
and 4.59E09 m
2
/s for pumpkin for 85, 90 and 95 C respectively.
The respective values of the radial diffusion coefcient (D
r
) were
found 0.69E10, 0.86E10 and 0.52E10 m
2
/s for banana, 0.80E10,
0.97E10 and 1.12E10 m
2
/s for cassava and 5.76E10, 5.68E10
and 5.60E10 m
2
/s for pumpkin respectively. It was observed
that the value of D
z
were higher than D
r
for the materials tested
apparently due to an-isotropic behavior of the materials.
Acknowledgements
The authors wish to acknowledge the Ministry of Science, Tech-
nology and Innovation (MOSTI) for the support extended through
an e-Science grant for this project. Special thanks to the RCMO of
the Universiti Sains Malaysia (USM) for the advice and encourage-
ment given in carrying out the grant during the tenure. Apprecia-
tion is also extended to the USM for the award of the USM
Fellowship to Ms. Low Hua Chin in order to carry out the research.
References
Andrieu, J., Stamatopoulos, A.-A., 1986. Durum wheat pasta drying kinetics.
Lebensmittel Wissenschaft und Technologie 19, 448456.
Aysun, M., Sevim, K., Medeni, M., 2002. Hot air and sun drying of grape leather
(pestil). Journal of Food Engineering 54, 8188.
Bird, J., 2007. Engineering Mathematics, fth ed. Elsevier Ltd., USA. pp. 104105.
Boudhrioua, N., Michon, C., Cuvelier, G., Bonazzi, C., 2002. Inuence of ripeness and
air temperature on changes in banana texture during drying. Journal of Food
Engineering 55, 115121.
Chen, C.R., Ramaswamy, H.S., 2002. Color and texture change kinetics in ripening
bananas. Lebensmittel-Wissenschaftund-Technologie 35, 415419.
Crank, J., 1975. The Mathematics of Diffusion. Oxford University Press Inc.. pp. 71
77.
Gilliland, E.R., Sherwood, T.K., 1933. The drying of solids VI. Industrial and
Engineering Chemistry 25, 11341136.
Gonalves, E.M., Brazo, R., Pinheiro, J., Abreu, M., Silva, C.L.M., Moldo-Martins, M.,
2005. Inuence of maturity stage on texture, pectin composition and
microstructure of pumpkin. In: 2nd Mercosur Congress on Chemical
Engineering, 4th Mercosur Congress on Process Systems Engineering.
Doymaz, I., 2007. The kinetics of forced convective air-drying of pumpkin slices.
Journal of Food Engineering 79, 243248.
Karim, M.A., Hawlader, M.N.A., 2005. Drying characteristics of banana: theoretical
modeling and experimental validation. Journal of Food Engineering 70, 3545.
Johnson, P.-N.T., Brennan, J.G., Addo-Yobo, F.Y., 1998. Air-drying characteristics of
plantain (Muss AAB). Journal of Food Engineering 31, 233242.
Jorge, W., Gustavo, V.B., Jos, M.A. (Eds.), 2002. Engineering and Food for the 21st
Century. Copyright Clearance Centre (CRC) Press, pp. 1720.
Madamba, P.S., Driscoll, R.H., Buckle, K.A., 1996. The thin-layer drying
characteristics of garlic slices. Journal of Food Engineering 29, 7597.
Magee, T.R.A., Wilkinson, C.P.D., 1985. Air drying of apple slices. Irish Journal of
Food Science Technology 9, 115122.
Minh-Hue, N., William, E.P., 2007. Air-drying of banana: inuence of experimental
parameters, slab thickness, banana maturity and harvesting season. Journal of
Food Engineering 79, 200207.
Nawirska, A., Figiel, A., Kucharska, A.Z., Sok-e towska, A., Biesiada, A., 2009.
Drying kinetics and quality parameters of pumpkin slices dehydrated using
different methods. Journal of Food Engineering 94, 1420.
Polyanin, A.D., Zaitsev, V.F., 2003. Handbook of Exact Solutions for Ordinary
Differential Equations, second ed. Chapman Hall, Washington. p. 750.
Senol, I., Medeni, M., 2002. Effect of cooking on the drying behaviour of tarhana
dough, a wheat ouryoghurt mixture. Journal of Food Engineering 54, 119
123.
Sherwood, T.K., 1929. The drying of solids I. Industrial and Engineering Chemistry
21, 1214.
Sherwood, T.K., 1931. Application of theoretical diffusion equations to the drying of
solids. Transactions of the American Institute of Chemical Engineers 27, 190
200.
Sherwood, T.K., 1936. The air drying of solids. Transactions of the American Institute
of Chemical Engineers 32, 150168.
Tutuncu, M.A., Labuza, T.P., 1996. Effect of geometry on the effective moisture
transfer diffusion coefcient. Journal of Food Engineering 30, 433447.
Vaccarezza, L.M., Chirife, J., 2006. On the application of Ficks law for the kinetic
analysis of air drying of foods. Journal of Food Science 43, 236238.
Valentas, K.J., Rotstein, E., Singh, R.P., 1997. Handbook of Food Engineering Practice.
Copyright Clearance Centre (CRC) Press. p. 129.
Yan, Z., Sousa-Gallagher, M.J., Oliveira, F.A.R., 2008. Shrinkage and porosity of
banana, pineapple and mango slices during air-drying. Journal of Food
Engineering 84, 430440.
Table 2
Diffusion coefcient, D
r
and D
z
for three materials
Materials Temperature (C) D
z
10
9
(m
2
/s) D
r
10
10
(m
2
/s)
Banana 85 4.3130 0.6872
90 4.3362 0.8621
95 4.4186 0.9380
Cassava 85 5.0861 1.4155
90 5.1263 1.7264
95 5.1894 1.9894
Pumpkin 85 4.4739 0.1600
90 4.4975 0.1604
95 4.5863 0.1607
316 W.J.N. Fernando et al. / Journal of Food Engineering 102 (2011) 310316