Modeling of moisture diffusivity, activation energy and specific energy consumption of high moisture corn in a fixed and fluidized bed convective dryer

Thin layer drying characteristics of high moisture corn under fixed, semi fluidized and fluidized bed conditions with high initial moisture content (66.82% wb) in a laboratory fluidized bed convective dryer was studied at air temperatures of 50, 65, 80 and 95°C. In order to find a suitable drying curve, seven thin layer-drying models were fitted to the experimental data of moisture ratio. Among the applied mathematical models, Midilli et al. model was the best for drying behavior prediction in corn thin layer drying. This model presented high values for correlation coefficient (R). Fick ́s second law was used to compute moisture diffusivity with some simplifications. Computed values of moisture diffusivity varied at the boundary of 4.87 × 10 – 2.90 × 10 m s and 1.02 × 10 – 1.29 × 10 m s during the first and second drying falling-rate, respectively. Values of effective moisture diffusivity for corn were also increased as input air temperature was increased. Value of activation energy varied from a minimum of 18.57 to a maximum of 50.74 kJ mol from 50 to 95°C with drying conditions of fixed to fluidized bed. Specific energy consumption (SEC) for thin-drying of high moisture corn was found to be in the range of 0.33 × 10 – 1.52 × 10 kJ kg from 50 to 95°C with drying condition of fluidized and fixed bed, respectively. Increase in air temperature in each air velocity caused decrease in SEC value. These corn properties would be necessary to design the best dryer system and to determine the best point of drying process. Additional key words: drying; maize; Midilli et al. model; semi fluidized.


Introduction
The main goal in agricultural and food products drying is the reduction of their moisture content to a specific level, allowing safe storage over an extended period of time. Due to a longer storage life, product diversity, and a substantial volume reduction, fruits and vegetables drying is popular. Thin layer drying models are used to predict drying time for food and agricultural products and also to generalize the kinetics of drying process. Drying kinetics of products is greatly affected by air temperature, air velocity and material characteristics (Erenturk and Erenturk, 2007).
Corn (Zea mays L.) is one of the most important agricultural crops in Iran with 1,600,000 tons production in 2008 (FAOSTAT, 2008). Because of high moisture content, harvested corn is contaminated with molds after several days, which are harmful to human health. Drying corn in natural sun drying method takes time and corn could be contaminated by insects, dust, sand particles and molds. Drying this crop is therefore necessary to reduce the moisture fast and uniform. Thus, it is safe method for the production of food and agricultural crops. Also, precise prediction of drying time is crucial important to increase the dryer capacity and to reduce the energy consumption (Doymaz and Pala, 2002).
In Iran, milky corn is harvested for human consumption, but its high moisture content levels (about 70% d.b.) causes fast spoilage and growing molds. Reducing corn moisture content is a proper way to prevent these losses.
One of the most popular methods of drying materials with high moisture content is fluidized bed. Fluidization defined as suspending the grain particles in a fluid. When air flow is passed upward through grain bed at a low flow rate, a fixed bed will be obtained. With an increase in air flow rate, the grain bed is expanded to provide minimum fluidized bed (semi fluidized bed), bubbling fluidized bed and transportation, respectively. At the minimum fluidized bed, pressure drop is maximized and weight of the particles counterbalances the frictional force between particles. In a bubbling fluidized bed, gas bubbles disturb movement of the grain particles. In a transportation stage, pneumatic conveying of grain particles is occurred. Bubbling fluidized bed and transportation stage defined as fluidization state (Kunii and Levenspiel, 1991;Brooker et al., 1992). Foster et al. (1980) dried corn samples in two stages using a solar dryer, being samples successfully dried at the second stage.
Li and Morey (1984) studied the thin layer convective drying method in yellow dent corn. Results showed that drying process is affected by drying air flow rate, air temperature, air relative humidity and initial moisture content. Soponronnarit et al. (1997) studied the drying characteristics of corn in a laboratory fluidized bed dryer at 150, 170 and 200°C air temperatures. They reported that corn drying with high initial moisture content with air temperature at 170°C could be done without quality loss. Suárez et al. (1984) dried sweet corn samples with 3.2 to 4.4 kg water kg dry solid -1 initial moisture content. Results showed that the drying time of treated samples with ethyl oleate was about 2.1 to 2.8 times faster than those untreated. Some physical and thermal properties of food and agricultural products, such as moisture diffusion, heat and mass transfer, specific energy and activation energy consumption are important for a proper dryer design (Aghbashlo et al., 2008). Some researchers have studied activation energy and moisture diffusion in a thin layer drying of various agricultural and food products. These include hazelnuts (Ozdemir and Devres, 1999), grapes (Pahlavanzadeh et al., 2001), seedless grapes (Doymaz and Pala, 2002), potato slices (Akpinar et al., 2003), candle nuts (Tarigan et al., 2006), onion slices (Pathare andSharma, 2006), plums (Goyal et al., 2007), beriberi fruit (Aghbashlo et al., 2008), and milky mushroom (Arumuganathan et al., 2009). Although many information has been gathered about the activation energy and effective moisture diffusivity for various agricultural and food products, small number of reports are available on the activation energy and effective Modeling of drying characteristics of corn in a fixed and fluidized bed 29 moisture diffusivity for milky or high moisture content corn during fixed, semi fluidized and fluidized convective drying (Soponronnarit et al., 1997). Indices of effective moisture diffusivity and activation energy are necessary for designing, modeling and optimizing the mass transfer processes such as moisture adsorption or dehydration during storage.
The main objectives of this research were to determine the activation energy, effective moisture diffusivity and specific energy consumption of high moisture content corn during first and second falling-rate of fixed, semi fluidized and fluidized bed thin layer drying process and their dependence on factors such as input air temperature and air velocity.

Determination of drying conditions
In order to determine corn pressure drops and air flow velocities at the outlet, simultaneously, the fan speed was increased gradually using the inverter (Vincker VSD2) and the parameters were recorded. A differential digital manometer (Testo 505-P1) and a vane type digital anemometer were used for measuring static pressure loss and outlet air velocity, respectively. To obtain air pressure drop across the corn bed, at first, the total static pressure drop due to corn column and bed plate was measured. Then air pressure drop due to empty chamber was measured. In each experiment, the difference between total and bed plate static pressure drop gave the net static pressure drop of the bed material.
Maximum value of static pressure drop versus a specific air velocity in fluidization systems is defined as the minimum fluidization point or semi fluidized bed (Kunii and Levenspiel, 1991). Fluidization experiments were carried out in four replications with 50 g corn samples load. After determining the semi fluidized bed with air velocity about 1 m s -1 , one point before it (in f ixed bed domain) was selected as a f ixed bed condition with air velocity of 0.5 m s -1 and one point after it (in fixed bed domain) was selected as a fluidized bed condition with air velocity of 1.5 m s -1 and the drying experiments were conducted.

Experiments
Fresh milky corn was supplied from a local market in July 2009. The samples were stored in a refrigerator at 4 ± 1°C. Average environmental conditions of ambient air temperature and air relative humidity ranged between 30 to 36°C and 20 to 30%, respectively. Accuracy of thermometer Lutron TM-903 with sensor type k was ±(0.5% reading + 1°C) and accuracy for humidity meter Lutron TM-903 for relative humidity lower than 70% was ±3% RH and for upper than 70% was ± (3% reading + 1% RH). During the experiments, the ambient air temperature, air relative humidity and inlet and outlet temperatures of the dryer chamber were recorded. A fluidized bed dryer was used to carry out the drying experiments (Fig. 1). After 30 min: when the dryer conditions reached the steady state, about 50 ± 1 g corn were located on the tray of the dryer and drying process was started. Experiments were conducted at the drying air temperatures of 50, 65, 80 and 95°C, the drying conditions of fixed bed (0.5 m s -1 ), semi fluidized bed (1 m s -1 ) and fluidized bed (1.5 m s -1 ). Samples were weighed during the drying process using a digital balance with 0.01 g accuracy. The gravimetric method was used to determine the initial and final moisture contents of corn samples at 70°C during 24 h (ASAE, 2007). Drying process was carried out using samples with initial moisture content of about 66.8% (wb) and terminated when the moisture content decreased to about 5% (wb). In this study, the influence of the drying conditions on the effective moisture diffusivity, activation energy and specific energy consumption in thin-layer drying of high moisture corn are explained.

Experimental and mathematical modeling
Fick's second law of diffusion with spherical coordinates was applied in this study. The assumptions in Fick's equation solution were: moisture migration in diffusion, negligible volume shrinkage, constant temperature and diffusion coefficients (Crank, 1975;Di Matteo et al., 2000): where MR = moisture ratio, decimal; M = moisture content at any time, kg water kg -1 dry matter ; M e = equilibrium moisture content, M 0 = initial moisture content, kg water kg -1 dry matter ; n = 1, 2, 3, ... the number of terms taken into consideration; t = drying time, s; D eff = effective moisture diffusivity, m 2 s -1 ; r = radius of kernel, m. The activation energy was calculated using an Arrhenius-type equation (López et al., 2000;Akpinar et al., 2003): where E a = activation energy, kJ mol -1 ; R = universal gas constant, 8.3143 kJ mol -1 K -1 ; T a = absolute air temperature, K; D 0 = pre-exponential factor of the Arrhenius equation, m 2 s -1 .
From Eq.
[6], the plot of ln(D eff ) versus 1/T a gives a straight slope of K 1 : Linear regression analysis method was used to fit the equation to the experimental data to obtain the coefficient of determination (R 2 ). Eq.
[2] can also be written in a more simplif ied form as: Eq.
[8] is known as single exponential equation. The empirical models were used as alternative approach to analyze thin layer drying. Some commonly used equations in thin layer drying studies are shown in Table 1. In order to select a suitable model describing the drying process of corn with high moisture content, drying curves were fitted with thin layer drying equations.
The values of M e are relatively small compared to Modeling of drying characteristics of corn in a fixed and fluidized bed 31 (1) inverter, (2) fan and electrical motor, (3) flange, (4) electrical heater, (5) mixing chamber, (6) pipe network, (7) diffuser, (8) input air temperature recorder, (9) drying chamber, (10) air velocity sensor, (11) output air temperature recorder, (12) temperature controller, (13) computer. [7] and all models in Table 1 can be reduced to: [9] Specific energy consumption (SEC, kJ kg -1 ) was calculated using the following equation (Zhang et al., 2002): where C Pv and C Pa = the specific heat capacity of vapor and air, respectively, 1004.16 and 1828.8 J kg -1°C-1 ; Q = inlet air to drying chamber, m 3 s -1 ; h a = absolute air humidity, kg vapor kg -1 dry air ; T in and T am = inlet air to drying chamber and ambient air temperatures, respectively,°C; V h = specific air volume, m 3 kg -1 ; t = total drying time, min -1 and m v = mass of removal water, kg.
To determine the drying kinetics, corn samples were dried in a laboratory thin layer dryer at 50, 65, 80 and 95°C. About 50 g corn sample was uniformly spread in a thin layer on perforated stainless steel tray for drying. Moisture loss was recorded by a digital balance (AND GF-6000, Japan). Drying process was continued untill there was no large difference between the subsequent moisture losses. The input air velocity passing through the corn sample was regulated at fixed, semi fluidized and fluidized bed conditions with air velocities of 0.5, 1 and 1.5 m s -1 , respectively. Experiments were conducted in three replications.
Non-linear regression analysis was done using MATLAB (version 7) software package. Correlation coefficient R 2 was one of the main criteria for selecting the best model. The goodness of fit was also determined using various statistical parameters such as reduced chi-square (χ 2 ) and root mean square error (RMSE) values. For quality f it, R 2 value should be higher and χ 2 and RMSE values should be lower (Togrul and Pehlivan, 2002;Demir et al., 2004;Erenturk et al., 2004). The parameters were calculated using the following expressions: [11] [12] [13] where MR exp,i = experimental moisture ratio of i th data; MR pre,i = predicted moisture ratio of i th data; N = number of observations; z = number of drying constants.

Mathematical models
The drying time of corn samples at different temperatures derived for all bed conditions are presented in Figure 2. The average final moisture content of corn samples was about 5% (wb). According to the results drying air temperature played has an important role in drying. When the air temperature was increased, the drying time was reduced. This phenomenon is because of applying more energy rate to the bed material and increasing in draying rate. The results are similar to the earlier studies of drying garlic slices (Madamba et al., 1996), onion slices (Sarsavadia et al., 1999), R. Amiri Chayjan et al. / Span J Agric Res (2011) 9(1), 28-40  (Aghbashlo et al., 2009). With regard to the drying curves (Fig. 2), it is obvious that all drying process of high moisture corn took place in the first and second falling-rate period for the entire duration. In other words, each drying curve constructed from two lines, a straight line at first drying period and a decreasing line at the end of process. This claim was strongly approved in Figure 3. In this figure, each drying curve was clearly included two lines. The first and the second lines were shown the first drying and the second falling periods, respectively. Similar results have been observed in drying some agricultural products such as: onion (Rapusas and Driscoll, 1995), lettuce and cauliflower leaves (López et al., 2000), apricots (Doymaz, 2004), figs (Piga et al., 2004), peaches , plums , berberis fruit (Aghbashlo et al., 2008), mushroom (Arumuganathan et al., 2009) and carrot slices (Aghbashlo et al., 2009). In other word, moisture content of all these products has been high, but drying behavior of some was similar to a straight line. Corn drying in falling-rate period proved that the internal mass transfer occurred by diffusion.
The average moisture ratio of dried corn at different temperatures was verified using seven different empirical models to find out their suitability to describe the drying behavior. Non-linear regression analyses using MATLAB 7 (R 14) was employed for statistical modeling of drying curves through selecting the General Equations option from Curve Fitting toolbox 1.1. Correlation coefficient and other indices are summarized in Table 2. To select the best model for describing the drying curves, criterion of R 2 should have higher value and the others (and) lower values. All values of Page, Midilli et al., two term and Wang and Singh models were greater than 0.99. Table 2  tions are presented in Table 3. All predicted values of moisture ratio were plotted against experimental data for all temperatures and bed conditions as shown in Figure 4. The R 2 value of this curve proved that the prediction process has been carried out with high precision.

Computation of effective moisture diffusivity
Experiments of drying process were continued until the differential mass between the two weighing became less than 0.05 g. Figure 3 shows the variations of the Ln(MR) versus drying time (s) in different air velocity and temperature levels. These drying curves show that drying high moisture content corn was occurred in falling-rate period. In other words, drying force controlled the liquid diffusion in first and second falling-rate drying process, and drying curves are similar to two straight lines as the f irst and second falling-rate periods. Trend of plotted curves show that with increase in the temperature values, the slope of straight line was increased. Air velocity also affected the slope of D eff adversely; hence decrease in air velocity caused increase in D eff . Slope of D eff in second falling-rate was further. Values of D eff were determined using Eq. [4]. These values are shown in Table 4 for all levels of air velocities and temperatures. The maximum values of D eff during the first and second falling-rate of drying belonged to semi fluidized condition with air velocity of 1 m s -1 . Because of this condition the most effective contact between grain and air velocity was accrued. The maximum value of D eff for first and second fallingrate of drying was obtained as 2.90 × 10 -10 m 2 s -1 and 1.29 × 10 -9 m 2 s -1 , respectively, both were calculated at the air temperature of 95°C. The minimum values of D eff during the first falling-rate of drying (4.87 × 10 -11 m 2 s -1 ) belonged to fixed bed with air velocity of 0.5 m s -1 , and for second one (1.02 × 10 -10 m 2 s -1 ) belonged to fluidized bed, both at air temperature of 50°C.   Drying air temperature greatly affected the D eff values of high moisture corn. As it is observed in the Table 4, D eff value was increased as drying air temperature increased. A similar result regarding the effect of air drying temperature on moisture diffusivity during convective air drying has been found in apricots (Pala et al., 1996;Doymaz, 2004), peaches , plums , berberis fruit (Aghbashlo et al., 2008), mushroom (Arumuganathan et al., 2009) and carrot slices (Aghbashlo et al., 2009).

Effect of air condition on effective moisture diffusivity
Values of D eff were plotted against air temperature at different levels of air velocities as shown in Figure 5. Six power models were applied to fit the computed values of D eff in first and second falling period. Fitted models and related R 2 values are presented in Tables 5  and 6. Values of D eff at different levels of air tempera-ture are depicted in Figure 5. Results proved that the minimum value of D eff belonged to minimum value of air temperature. These findings also indicated that the influence of air velocity on increasing D eff at upper air temperatures was high. Drying air contact with corn kernels at semi fluidized conditions was most effective because of its highest values of D eff . Results also showed that at lower air temperatures, bed condition (air velocity levels) were not significantly different, as applying lower air velocities were even more effective. Because the drying process of corn was in the falling-rate period, mass transfer was therefore occurred by the use of diffusion phenomenon. In the diffusion mode, the effect of outer factors such as air temperature and air velocity on the mass transfer was not significant and only needed more time for transfer of moisture from the inner layer of grain to the surface.
Quadratic model type was fitted to calculated moisture diffusivity values. Applied quadratic models and related R 2 values for different air temperatures are presented in Table 6. The values of D eff were plotted against air temperature and air velocity as shown in Figure 5.

Computation of activation energy
Values of Ln(D eff ) were plotted against 1/T as shown in Figure 6 for first and second falling period. Activa-tion energy (E a ) was obtained using Eq. [6]. Computed values of E a for different levels of air velocities and related R 2 values are presented in Table 7. In general, E a for food and agricultural crops lies in domain of 12.7-110 kJ mol -1 (Aghbashlo et al., 2008). Minimum and maximum values of E a for figs have been reported 30.8 and 48.47, respectively (Babalis and Belessiotis, 2004). Minimum and maximum values of E a for high moisture corn varied from 18.57 to 26.19 kJ mol -1 in first falling period and from 22.96 to 41.69 kJ mol -1 in second falling period for all air velocity levels. Two forms of water existence in fruits include surface and chemical absorptions. As most of the water in high moisture corn in first falling-rate period is in the form of surface absorption, little energy is required to exhaust water and undesirable change in chemical properties is negligible in this period (Aghbashlo et al., 2008). If proper dryer with suitable air velocity and temperature is selected for corn drying, damages should be decreased. Activation energy of high moisture corn is Modeling of drying characteristics of corn in a fixed and fluidized bed 37
Values of E a were plotted against air velocity as shown in Figure 7. Quadratic models were fitted to data set and values of R 2 were presented. Maximum value of E a was occurred when air velocity laid in the range of 1-1.5 m s -1 (Fig. 7). Activation energy was increased when air velocities were above 1 m s -1 . Similar result has been obtained by Demirel and Turhan (2003) about lesser activation energy requirement for banana slices during high air temperature drying. Two order equations are fitted to the calculated data of E a versus air velocity as follows: (FFP) [14] (SFP) [15]

Computation of specific energy consumption
During the experiments, the specific energy consumption (SEC) for removing 1 kg moisture content from material by the use of an electrical heater and energy requirements for drying 1 kg of fresh high moisture corn were calculated for each experiment using Eq. [10]. Computed values of SEC is shown in Figure 8. It was observed that the SEC increased as drying air temperature was decreased. Increasing air velocity affected intensively causing an increase in SEC. Maximum value of SEC 1.52 × 10 6 (kJ kg -1 ) obtained at air velocity of 1.5 m s -1 with drying air temperature of 50°C. The minimum value of SEC needed 0.33 × 10 6 (kJ kg -1 ) while air velocity and drying air temperatures were 0.5 m s -1 and 90°C, respectively. Results proved that increasing in drying time affect SEC inversely. In other words, each factor caused an increase in drying time, also caused an increase in energy consumption. With increasing in air velocity, effective contact between air and corn kernels was reduced and SEC was therefore increased. Similar results have been obtained for paddy (Khoshtaghaza et al., 2007) and berberis fruit (Aghbashlo et al., 2008).

Conclusions
Results showed that the Midilli et al. model was the best for prediction of high moisture corn drying ki-   netics. Maximum value of D eff during corn drying was obtained in semi fluidized bed condition (air velocity of 1 m s -1 ) and the air temperature of 95°C. Minimum value of D eff was obtained in fixed bed condition (air velocity of 0.5 m s -1 ) and the air temperature of 50°C. values of D eff for corn were also increased as input air temperature increased. Minimum and maximum values of E a for dried corn were 18.57 and 41.69 kJ mol -1 , respectively. In fixed bed condition, activation energy had the minimum value. Maximum value of E a was calculated in semi fluidized bed condition. Maximum value of SEC for corn drying obtained in fluidized bed condition with drying air temperature of 50°C. Minimum value of SEC was also obtained in fixed bed condition with drying air temperature of 95°C. Increase in air temperature in each air velocity caused a decrease in SEC value.