International Food Research Journal 21(5): 1911-1917 (2014)

Journal homepage: http://www.ifrj.upm.edu.my

Experimental characterization and modeling of thin-layer drying of mango

slices
Akoy, E. O. M.

Faculty of Environmental Sciences and Natural Resources, University of Al-Fashir, Sudan

Article history Abstract
Received: 28 January 2014 Convective air drying characteristics of mango slices at different drying temperatures (60C,
Received in revised form: 70C and 80C), at an air velocity of 0.5 m/s and for constant sample thickness (3 mm) were
25 March 2014
Accepted: 27 March 2014
investigated. Results indicated that drying took place in the falling rate period. Drying time
decreased considerably with increased drying temperature. Three mathematical models;
Keywords namely, Newton (Lewis), Henderson and Pabis, and Page were selected to describe and
compare the drying characteristics of mango slices. Comparisons were based on the coefficient
Mango of determination (R2), sum square error (SSE), root mean square error (RMSE) and reduced-chi
Mathematical models
square (2). Among the tested models, the Page model achieved the best fit. Moisture transfer
Thin-layer
Effective moisture
from mango slices was described by applying the Ficks diffusion model. Effective moisture
diffusivity diffusivity (Deff) values increased with increasing drying temperature and were found to range
Activation energy from 4.97 x 10-10 m2/s to 10.83 x 10-10 m2/s. The temperature dependence of the effective
diffusivity was described by the Arrhenius-type relationship and the activation energy for the
diffusion of the moisture associated with the mango slice was found to be 37.99 kJ/mol.
All Rights Reserved

Introduction two stages. The first stage happens at the surface of

the drying material at a constant drying rate and is
Mango (Mangifera indica L.) is one of the similar to the vaporization of water into the ambient.
tropical and subtropical fruit of great importance for The second stage drying process takes place with
both economical and nutritional point of view. It is decreasing drying rate (Midilli and Kucuk, 2003).
considered to be a good source of carbohydrates, When the drying process is controlled by the internal
vitamin C and very rich source of pro-vitamin A. In mass transfer, mainly in the falling rate period,
spite of its excellence, the perishable nature of this modeling of drying is carried out through diffusion
fruit and its short harvest season severely limit its equations based on Ficks second law.
utilization. Drying may be an interesting method in Drying is a complex thermal process in
order to prevent fresh fruit deterioration. which unsteady heat and moisture transfer occur
Drying is one of the most widely used primary simultaneously (Sahin and Dincer, 2005). From
methods of food preservation. The objective engineering point of view, it is important to develop a
drying is the removal of water to the level at which better understanding of the controlling parameters of
microbial spoilage and deterioration reactions are this complex process. Mathematical models of drying
greatly minimized (Akpinar and Bicer, 2004). It processes are used for designing new or improving
also provides longer shelf-life, smaller space for existing drying systems or even for the control of
storage and lighter weight for transportation (Ertekin the drying process. Many mathematical models have
and Yaldiz, 2004). Sun drying is the most common been proposed to describe the drying process, of them
method used to preserve agricultural products in thin-layer drying models have been widely in use.
tropical and subtropical countries. However, being Several thin-layer drying models available in
unprotected from rain, wind-borne dirt and dust, the literature for explaining drying characteristics
infestation by insects, rodents and other animal, of agricultural products. These models can be
products may be seriously degraded to the extent that categorized as theoretical, semi-empirical and
sometimes become inedible and the resulted loss of empirical. Moreover, the drying kinetics of food
food quality in the dried products may have adverse is a complex phenomenon and requires simple
economic effects on domestics and international representations to predict the drying behavior,
markets. Therefore, the drying process of agricultural and for optimizing the drying parameters. Many
products should be undertaken in closed equipment investigators have carried out mathematical modeling
(solar or industrial dryer) to improve the quality of and experimental studies on the thin-layer drying of
the final product. The drying process takes place in various vegetables and fruits. For example, potato

*Corresponding author.
Email: [email protected]
Tel: +249912894598; Fax: +249731842111
1912 Akoy, E. O. M./IFRJ 21(5): 1911-1917

slices (Aghbashlo et al., 2009), onion slices (Arslan adjusted to the selected temperature for about half an
and zean, 2010), sweet cherry (Doymaz and Ismail, hour to reach thermal stabilization. Then the samples
2011) and banana (da Silva et al., 2013). However, were uniformly spread in a single layer of 3mm
there is limited information and research on drying thickness on a tray. Representative samples of sliced
kinetics of mango slices in the literature. Therefore, mango for moisture content determination were placed
the objectives of this study were: (a) to investigate in a circular wire mesh of 10 cm diameter and placed
the thin-layer drying characteristics of mango slices, onto the centre of the tray. For measuring the mass
(b) modeling of the thin-layer drying of mango slices of the sample at any time during experimentation,
by testing three drying models and (c) to estimate the circular wire mesh with sample was taken out of
the effective diffusivity coefficient and energy of the drying chamber and weighed on a digital balance
activation for mango fruit. and placed back into the drying chamber every 30
min during the drying process. The digital top pan
Materials and Methods balance (Sartorius, Goettingen, Germany) of 0.001 g
accuracy, was kept near to the drying unit and weight
Raw material measurement process took less than 10 seconds time.
Fresh mangoes, var. Kent, from Mali, were The drying process was stopped when the moisture
purchased at a local supermarket in Goettingen, content decreased to about 9 0.2% (w.b). All the
Germany and stored in a refrigerator at 4 0.5C. experiments were replicated three times at each
Prior to drying, samples were taken out of the drying temperature and the average values were used
refrigerator and left for 5 days for post-harvest for the drying characteristics of mango slices.
ripening at 25 2C and 50% relative humidity (Pott
et al., 2005). The fruits were then washed, manually Mathematical modeling of drying curves
peeled using a stainless steel knife, and sliced using The moisture ratio (MR) and drying rate of mango
an electric food-slicer (Krups variotronic, Germany) slices during drying experiments were calculated
to a thickness of 3 mm. using the following equations:

Drying experiments (3)

The drying experiments were performed in a
convective air oven (Heraeus: UT 6120, Germany) Where: MR is the dimensionless moisture ratio; M,
at temperature of 60, 70 and 80C. The oven is M0 and Me are the moisture content at any time, initial
consisted of heating unit, temperature control unit, moisture content and equilibrium moisture content,
drying chamber and centrifugal fan that has a fixed respectively. However, MR was simplified according
air velocity of 0.5 m/s. The average initial moisture to Pala et al. (1996) and Doymaz (2004) as:
content of the mango fruit was 82.5 0.4% (w.b.),
as determined using a precision air-oven method, at (4)
a temperature of 135C for 2 hours until constant
weight was reached, according to the standard method (5)
of AOAC ( 2000) and moisture content on wet basis
(w.b.) was calculated by the following equation: Where, Mt, and Mt+dt are the moisture content at t and
moisture content at t+dt (kg water /kg dry matter),
(1) respectively, t is drying time (hr).
The drying curves were fitted to three well-known
Where: thin layer drying models that are widely used in
MCwb = moisture content, percent, wet basis most food and biological materials; namely, Newton
Ww = weight of water, g (Lewis), Henderson and Pabis, and Page models.
Wd = weight of dry matter, g These models are generally derived by simplifying
Moisture content on wet basis was converted the general solution of Ficks second law. Henderson
to moisture content on dry basis by the following and Pabis model is the first term of a general series
equation: solution of Ficks second law. The model was used to
predict the drying characteristics of corn (Henderson
(2) and Pabis, 1961) and is expressed as follows:
Where: MCdb = moisture content, decimal, dry basis MR = a exp( kt ) (6)
Prior to starting the experiments, the oven was
Akoy, E. O. M./IFRJ 21(5): 1911-1917 1913

Newton (Lewis) model is a special case of the flow. A knowledge of effective moisture diffusivity is
Henderson and Pabis Model where the intercept is necessary for designing and modeling mass-transfer
unity and is used to describe the drying of barely processes such as dehydration, adsorption and
(Bruce, 1985) and grape seed (Roberts et al., 2008). desorption of moisture during storage. The drying
This model is expressed as: data in the falling rate period are usually analyzed by
Ficks diffusion equation (Crank, 1975).
MR = exp( kt ) (7) Ficks second equation of diffusion was used
to calculate effective moisture diffusivity of mango
Page model is an empirical modification of slices, considering a constant moisture diffusivity,
Newton (Lewis) model to overcome its shortcoming infinite slab geometry and uniform initial moisture
it was successfully used to describe the drying distribution as follows:
characteristics of some agricultural products (Singh
et al., 2006; Hassan-Beygi et al., 2009; Doymaz and
Ismail, 2011). This model is expressed as follows: (12)

MR = exp( kt n ) (8) Where:

8/ 2 = the shape factor and depends on the geometry
In the proposed models, a and n are the drying of the drying material (4/ 2 for a cylinder and 6/ 2 for
coefficients and k is the drying constant (hr-1). the sphere).
Deff = the effective diffusivity, ms
Statistical analysis L = half-thickness of slab, m
Non-linear regression analysis was used to n = positive integer
evaluate the parameters of the selected models. The For long drying times, the Eq. (12) can be
goodness of fit of the three selected drying models simplified as Eq. (13) by taking the first term of the
to the experimental data was determined using series solution and expressed in a logarithmic form as
four statistical parameters, namely; coefficient of follows (Doymaz, 2012):
determination (R2), sum square error (SSE), reduced
chi-square (2) and root mean square error (RMES). (13)
These parameters can be calculated by using the
following equations: The effective moisture diffusivity was obtained
by plotting the experimental data in terms of ln(MR)
1

N
SSE = ( MRexp,i MR pred ,i ) 2 (9) versus drying time (hr). From equation (13), a plot of
N i =1
ln(MR) versus time gives a straight line with a slope
of (k) in which:
(10)
(14)

1 The dependence of the effective moisture

N
RMSE = ( MRexp,i MR pred ,i ) 2 (11)
N i =1
diffusivity on temperature is generally described by
Where: the Arrhenius equation (Simal et al., 2005):
MRexp. = Experimental moisture ratio
MRpred. = Predicted moisture ratio (15)
N = Number of observations
n = Number of constants Where:
The higher R2 values and the lower 2, SSE and Do = the pre-exponential factor of the Arrhenius
RMSE values are goodness of fit (Sacilik et al., 2006; equation, m/s
Hassan-Beygi et al., 2009). Ea = activation energy, kJmol-1
R = universal gas constant, kJmol-1K-1
Determination of effective moisture diffusivity and T = absolute temperature, K
activation energy Eq. (15) can be rearranged into the form of Eq.
Effective moisture diffusivity describes all (16) as follows:
possible mechanisms of moisture movement within
Ea
the food, such as liquid diffusion, vapor diffusion, ln Deff = ln(D0 ) (16)
RT
surface diffusion, capillary flow and hydrodynamic
1914 Akoy, E. O. M./IFRJ 21(5): 1911-1917

A plot of lnDeff as a function of the reciprocal of Table.1. Values of the drying constants and drying
absolute temperature 1/T will produce a straight line coefficients of the selected models
Model Drying temperature (C) Drying constants Drying coefficients
with slope equal to (-Ea/R), from which the parameter Newton (Lewis) 60 k = 0.644hr-1 -

Ea can be estimated. 70
80
k = 0.864hr-1
k = 1.264hr-1
-
-
The activation energy (Ea) was calculated Henderson & Pabis 60 k = 0.624hr-1 a = 1.104
by plotting the natural logarithm of Deff versus 70
80
k = 0.853hr-1
k = 1.36hr -1
a = 0.963
a = 1.233
the reciprocal of the absolute temperature (Tabs). Page 60 k = 0.601hr-1 n = 1.071
Activation energy is a measure of the temperature 70
80
k = 0.851hr-1
k = 1.004hr-1
n = 1.069
n = 1.285
sensitivity of Deff and it is the energy needed to initiate k = drying constant (hr ); a = drying coefficient and n = drying coefficient
-1

the moisture diffusion within the mango slices. Table.2. Statistical results obtained from the selected thin
layer drying models
Results and Discussion Model T (C) R2 SSE RMSE 2
Newton (Lewis) 60 0.991 0.001297 0.02419 0.00139
70 0.995 0.000676 0.01533 0.00074

Drying characteristics of mango slices Henderson &

80
60
0.979
0.991
0.00374
0.001614
0.03948
0.03043
0.00437
0.00186

The drying characteristics of mango slices Pa bis 70

80
0.996
0.972
0.001205
0.008498
0.02190
0.0525
0.001326
0.011897

are shown in Figures 1 and 2. The initial moisture Pa ge 60

70
0.994
0.998
0.000717
0.000261
0.019246
0.010784
0.000828
0.000287

content of mango slices before drying was about 82.5 80 0.999 3.61x10 -5 0.0051 5.05x10 -5
T = temperature (C); R2 = coefficient of determination; SSE = sum square
0.4% w.b. (mean std. deviation). As expected, the error; RMSE = root mean square error and 2 = reduced chi-square.

drying temperature had a significant effect on drying

characteristics of the mango slices. The moisture
content decreased continuously with time and an
increase in temperature resulted in reduced drying
time. The longest and shortest drying times were
recorded at 60C (7 hr) and 80C (3hr), respectively.
The time required to reduce the moisture content of
mango slices from 82.5 0.4% (w.b.) to a final 9
0.2% (w.b.) were 3, 5 and 7 hour at 80C,70C and
60C, respectively, as shown in Figure 1.
Figure 2 shows moisture ratio of the mango
slices plotted versus drying time. From the figure it Figure 1. Effect of drying temperature on the moisture
is clear that moisture ratio decreased considerably content of mango slice
with increasing drying time. The time required to
reduce the moisture ratio to any given level was
dependent on the drying temperature, being highest
at 60C and lowest at 80C. It was observed that
the main factor influencing drying kinetics was the
drying temperature, as noted in other studies (Belghit
et al., 2000; Koulia et al., 2002). Thus, a higher
drying temperature produced a higher drying rate and
consequently the moisture content decreased faster.
This is due to increase of air enthalpy to the mango
slices and subsequent acceleration of water migration
within the mango slices.
Figure 3 shows the effect of the three temperatures Figure 2. Effect of drying temperature on the moisture
ratio of mango slices.
on the drying rate of mango slices. From the figure it
can be observed that there is no constant rate drying Fitting of the drying models
period in the drying process of mango slices, and all Table 1 shows values of the drying constants
the drying process occurs in the falling rate period. and drying coefficients of the selected models.
This indicates that diffusion is the dominant physical From the table it is clear that drying constant (k) is
mechanisms governing moisture movement within temperature function. It is increased with increasing
the mango slices. Similar results have been reported drying temperature. The fitting of the three thin-layer
for the drying studies on raw mango slices (Goyal et drying models to experimental data were compared
al., 2006), apricots (Doymaz, 2004) and yacon slices in terms of the four statistical parameters; R2, 2,
(Shi et al., 2013). SSE and RMSE. The statistical analysis values are
Akoy, E. O. M./IFRJ 21(5): 1911-1917 1915

Figure 3. Drying rates versus the experimental moisture Figure 5. Effect of drying temperature on the effective
ratio of mango slices diffusivity of water in mango slices

Figure 4.Predicted MR versus Experimental MR by Page Figure 6. Arrhenius-type relationship between effective
model at 80C diffusivity and temperature.

summarized in Table 2. with increasing drying air temperature. When samples

In all cases, the R2 values for the models were were dried at higher temperature, increasing heating
greater than 0.95, indicating a good fit (Doymaz and energy increases the activity of water molecules
Ismail, 2011). The R2 values varied between 0.972 leading to higher moisture diffusivities. The values
and 0.999, SSE values between 3.61 x 10-5 and 0.0085, of effective moisture diffusivity obtained from this
RMSE values between 0.0051 and 0.0395, and 2 study lie within the general range from 10-11 to 10-9
values between 5.05 x 10-5 and 0.0119. These values m2/s for food materials (Madamba et al., 1996). The
show that the three tested drying models predict thin values of the effective moisture diffusivity (Deff) are
layer drying process of mango slices adequately. consistent with the reported values of 2.27 to 4.97
Generally, Page model gave a higher R2 and lower x10-10 m2/s for the drying of apple in the temperature
SSE, RMSE and 2 values (Table 2). Thus, the Page range 40-60C (Sacilik et al., 2006), 2.62 to 4.97 x
model could be selected to represent the thin-layer 10-10 m2/s for the drying of raw mango (cv. Dasehari)
drying characteristics of mango slices. slices in the temperature range 55-60C (Goyal et
Figure 4 shows the plotting of the experimental al., 2006), 3.32 to 90.0 x 10-10 m2/s for berberis fruit
data with the predicted ones using Page model for at 50-70C (Aghbashlo et al., 2008) and 6.27 to 35.0
mango slices at 80C. The scatter diagram shows x10-10 m2/s for orange slices at 40-80C (Raffie et al.,
that the observations are clustered along the linear 2010).
regression line which means the adequacy of this
model in describing the drying characteristics of Activation energy
mango slices. Similar findings were reported by The activation energy (Ea) was found to be 37.99
Goyal et al. (2006) for raw mango slices, Doymaz kJ/mol (Figure 6.). The activation energy value
and Ibrahim (2011) for sweet cherry, Aghbashlo et obtained from this study lies within the general range
al. (2009) for potato slices and Doymaz (2012) for of 12.7 to 110 kJ/mol for various food materials
persimmon slices. (Zogzas et al., 1996). It is higher than activation
energies of 27.0 kJ/mol for kiwifruit drying in the
Effective moisture diffusivity temperature range 30-90C (Simal et al., 2005) and
The determined values of the effective moisture 30.0 kJ/mol for yacon drying (Shi et al., 2013), and
diffusivity (Deff) for the different temperatures are lower than the activation energies of 40.95 kJ/mol
shown in Figure 5. The diffusivity values were found for fig drying (Xanthopoulos et al., 2009) and 43.05-
to be 4.97 x 10-10, 6.79 x 10-10 and 10.83 x 10-10 m2/s at 49.17 kJ/mol for sweet cherry drying (Doymaz and
60, 70 and 80C, respectively. It is clear that effective Ismail, 2011). But similar to activation energies of
diffusivity values for mango slices increases greatly 30.46-43.26 kJ/mol in the temperature range 50-70C
1916 Akoy, E. O. M./IFRJ 21(5): 1911-1917

for persimmon slices drying (Doymaz, 2012). Crank, J. 1975. The mathematics of diffusion. Claredon
Press, Oxford, UK.
Conclusions da Siva, W.P., e Silva, C., Gama, F. and Gomes, J. 2013.
Mathematical models to describe thin- layer drying
Drying curves were greatly affected by the drying and to determine drying rate of whole bananas. Journal
of the Saudi Society of Agricultural Sciences, htt://
temperature. Increased in drying temperature caused
dx.doi.org/10.1016/j.jssas.2013.01.003.
a decrease in the drying time. Drying of mango slices Doymaz, I. and Ismail, O. 2011. Drying characteristics of
occurred in the falling rate period, which indicates sweet cherry. Food and Bioproducts Processing 89:
that moisture removal from the product was governed 31-38.
by internal diffusion phenomenon. According to Doymaz, I. 2004. Drying kinetics of white mulberry.
statistical analysis applied to the three drying models, Journal of Food Engineering 61(3): 341-346.
Page model was found to be the most suitable model Doymaz, I. 2012. Evaluation of some thin-layer drying
for describing the thin-layer drying characteristics of models of persimmon slices (Diospyros kaki L.).
mango slices. The effective diffusivity coefficients Energy Conservation and Management 56: 199-205.
increased with increasing drying temperature, which Ertekin, C. and Yaldiz, O. 2004. Drying of eggplant and
selection of a suitable thin-layer drying model. Journal
ranged from 4.97 x 10-10 to10.83 x 10-10 m2/s over the
of Food Engineering 63: 349-359.
temperature range (60 to 80C). The activation energy Goyal, R . K .; Kingsly, A. R. P.; Manikantan, M. R. and
for the mango slices was estimated to be 37.99 kJ/ Ilyas, S. M. 2006. Thin-layer drying kinetics of raw
mol. mango slices. Biosystems Enginerring 95(1): 43-49.
Hassan-Beygi, S.R., Aghbashlo, M., Kinamehr, M.H.
Acknowledgements and Massad, J. 2009. Drying characteristics of
walnut (Juglan regia L.) during convection drying.
The author would like to thank the DAAD International Agrophysics 23: 129-135.
(German Academic Exchange Service) for the 6- Henderson, S. M. and Pabis, S. 1961. Grain drying theory
month scholarship grant (A/05/19861) which enabled I: temperature effect on drying coefficient. Journal of
him to conduct this experiment at the Institute of Agricultural Engineering Research 6(3): 169-174.
Kouhila, M.; Kechaou, N.; Otmani, M.; Fliyou, M. and
Agricultural Engineering, University of Goettingen,
Lahsasni, S. 2002. Experimental study of sorption
Germany. isotherms and drying kinetics of Moroccan Eucalyptus
Globulus. Drying Technology 20(10): 2027-2039.
References Madamba, P.S.; Driscoll, R.H. and Buckle, K.A. 1996.
The thin layer drying characteristics of garlic slices.
Aghbashlo, , M., Kianmehr, M.H. and Arabhosseini, A. Journal of Food Engineering 26: 113-130.
2009. Modeling of thin-layer drying of potato slices in Midilli, A. and Kucuk, H. 2003. Mathematical modeling
length of continuous band dryer. Energy Conservation of thin layer drying of pistachio by using solar energy.
and Management 50: 1348-1355. Energy Conservation and Management 44: 1111-
Aghbashlo, M., Kianmehr, H. and Samimi-Akhijahani, H. 1122.
2008. Influence of drying conditions on the effective Pala, M.; Mahmutoglu, T. and Saygi, B. 1996. Effects of
moisture diffusivity, energy of activation and energy pre-treatments on the quality of open-air and solar
consumption during the thin-layer drying of berberis dried products. Nahrung/Food 40: 137-141.
fruit (Berberidaceae). Energy Conservation and Pott, I.; Neidhart, S.; Muhlbauer, W. and Carle, R. 2005.
Management 49: 2865-2871. Quality improvement of non-sulphited mango slices
Akpinar, E. K. and Bicer, Y. 2004. Modelling of the drying by drying at high temperatures. Innovative Food
of eggplants in thin-layers. International Journal of Science and Emerging Technologies 6: 412-419.
Food Science and Technology 39: 1-9. Rafiee, S., Sharifi, M., Keyhoni, A., Omid, M., Jafari,
AOAC. 2000. Official Method of Analysis of AOAC A. and Mohtasebi, S.S. 2010. Modeling effective
International.17th Ed.; Association of official moisture diffusivity of orange slices (Thompson Cv.).
Analytical Chemist: Horwitz, USA. International Journal of Food Properties 13: 32-40.
Arslan, D. and zcan, M.M. 2010. Study the effect of sun, Roberts, J.S., Kidd,D.R. and Padilla-Zakour,O.2008.
oven and microwave drying on quality of onion slices. Drying kinetics of apple seeds. Journal of Food
LWT-Food Science and Technology 43: 1121-1227. Engineering 89: 460-465.
Belghit, A.; Kouhila, M. and Boutaleb, B. C. 2000. Sacilik, K.; Keskin, R. and Elicin, A. K. 2006. Mathematical
Experimental study of drying kinetics by forced modelling of solar tunnel drying of thin layer organic
convection of aromatic plants. Energy Conservation tomato. Journal of Food Engineering 73: 231-238.
and Management 44(12): 1303-1321. Sahin, A. Z. and Dincer, I. 2005. Prediction of drying times
Bruce, D.M. 1985. Exposed layer barely drying, three for irregular shaped multi-dimensional moist solids.
models fitted to new data up to 15oC. Journal of Journal of Food Engineering 71: 119-126.
Agricultural Engineering Research 32: 337-347. Shi, Q.,Zheng, Y. and Zhao, Y. 2013. Mathematical
Akoy, E. O. M./IFRJ 21(5): 1911-1917 1917

modeling on thin-layer heat pump drying of

yacon (Smallanthus sonchifolius) slices. Energy
Conservation and Mangement 71: 208-216.
Simal, S.; Femenia, A.; Garau, M.C. and Rossello, C.
2005. Use of exponential, page and diffusional models
to simulate the drying kinetics of kiwi fruit. Journal of
food engineering 66(3): 323-328.
Singh, S., Raina, G.S., Bawa, A.S. and Saxena, D.G. 2006.
Effect of pretreatments on drying and rehydration
kinetics and colour of sweet potato slices. Drying
Technology 24: 1487-1497.
Xanthopoulos, G., Yanniotis, S. and Lamberinos, Gr. 2009.
Water diffusivity and drying kinetics of air drying of
figs. Drying Technology 27(3): 502-512.
Zogzas, N.P., Maroulis, Z.B. and Marinos-Kouris, D. 1996.
Moisture diffusivity data compilation in foodstuffs.
Drying Technology 14: 2225-2253.