As presented in Figure 1, the drying system consists of two units; 1) the shallow bed unit and 2) the heat (drying air) supply unit. Parts of the shallow bed unit include: collapsible support frame, center post, collapsible panel, side guard, roof, and tarpaulin base plenum. The heat (drying air) supply unit consists of parts including; furnace, two fans, chimney, attachment pin and a lock, pulley system, duct connector and plate heat exchangers. The dryer is developed to afford smallholder farmers the opportunity to dry their grains irrespective of location and can dry 3 batches per day. The dryer is designed to utilize agro residues such as maize cobs as its main heat source and can dry maize in batches of 500 kg,

lowering the moisture level from 18% - 20% to 12% - 14%. As shown in Figure 2, the biomass (maize cobs used in this study) is combusted in the combustion chamber (furnace). Heat generated is absorbed by the plate type heat exchangers which in turn heats the high velocity air generated by the intake fan with the aid of the engine connected by a v-belt. The heated air is blown into the plenum of the shallow bed unit through the connecting duct. Heated air in the plenum evaporates the moisture in the grains due to its low relative humidity.

The saturated air rises out into the atmosphere through the open vent in the shallow bed unit. The smoke generated in the combustion chamber is forced out by the exhaust fan through the chimney into the atmosphere. Maize is dried to the required moisture content if the process is continued for few hours with constant supply of heat depending on the initial moisture content of the grains.

“Obaatanpa”, a local white maize variety was obtained from the KNUST Agricultural Research Station farm for the drying experiment. The initial moisture content was determined using ISO 6540-1980 moisture content determination method for maize. 8 g of ground maize was dried in an oven at 130˚C for 4 h. Three replications were used for the determination of the initial moisture content and an initial average of 17.5% ± 0.86% wet basis was recorded.

Drying experiments were carried out using the AflaSTOP dryer (Figure 3) locally fabricated at the workshop of the Department of Agricultural and Biosystems Engineering, KNUST, Kumasi.

A total of 250 kg of “Obaatanpa” white shelled maize at initial MC of 17.5% (wb) was used for the drying experiment. The dryer powered by a 5-litre capacity diesel generator was allowed to run for at least 20 minutes before the commencement of drying for it to reach steady state condition. Drying air conditions

(temperature and relative humidity) were monitored at the plenum of the dryer using Kestrel data loggers positioned at 3 logging positions (Figure 4). Conditions of the ambient were also monitored during the drying process.

The MC of maize grains during the drying experiment was concurrently monitored hourly with a pre-calibrated moisture meter (Dicky-John mini GAC plus moisture analyser with accuracy of ±0.1%). Maize grains were sampled at different positions in the dryer and mixed thoroughly before the MC reading was taken and recorded. Three measurements were taken at each period and the means were calculated for analysis.

• Thermal Efficiency

According to [15], the thermal efficiency of biomass burners with heat exchangers can be determined as stated in Equation (1).

( % ) η t h = Q s Q a × 100 (1)

where,

Q s = M a i r × C a i r ( T a i r − T a m b ) (2)

Q a = M c c H v (3)

M a i r = V a i r × D a i r (4)

Q a = Heat available for drying (kJ/kg);

Q s = Heat supplied (kJ/h);

T a i r = Average heated air temperature (˚C);

T a m b = Average ambient air temperature (˚C);

C p a i r = Specific heat capacity of air (kJ/kg∙˚C);

M a i r = Mass flow rate of air (kg/h);

M c c = Biomass (maize cobs) consumed (kJ/kg;)

H v = Gross heating value of maize cob (kJ/kg);

V a i r = volume flow rate of air (m³/h);

D a i r = density of air (kg/m³).

• Drying System Efficiency

The drying system efficiency was determined from the formulae stated below. It is defined as the ratio of the heat used to evaporate moisture from the grains to the energy given to the drying air [16].

ŋ s y s = E u E f (5)

E u = M w × L v (6)

E f = M a × C p × ( T o u t − T d ) (7)

where

E f = Energy given to the drying air;

E u = Energy used to evaporate mass of water;

M a = mass of air (kg);

C p = specific heat capacity of air (kJ/kg∙˚C);

l v = latent heat of vaporization (kJ/kg);

m w = mass of water (kg);

T o u t = temperature of air going out after drying;

T d = drying air temperature.

• Volume Flow Rate of Air (m³/h)

V = V a t (8)

where V, V_a and t are the Volume flow rate, quantity of air (m³), and the total drying time respectively.

• Mass Flow rate of Air (kg/s)

m a i r = ρ V (9)

where m_air, ρ and V are the mass flow rate of air (kg/s), density of air (kg/m³) and V volume flow rate of air (m³/h).

• Moisture Content

M C w b = M m − M d M m (10)

Moisture loss (%)

= Initial moisture content % ( M i w b ) – Final moisture content % ( M f w b ) (11)

where M C w b , M m and M d are moisture content (wet basis), initial and finial mass of grains.

• Moisture extraction rate (MER)

M E R = M w t (12)

M w = M m − M d (13)

where MER, M_w, t, M_m, and M_d are the moisture extraction rate, mass of moisture loss, drying time, initial and finial mass of grains.

• Rate of moisture loss (M_r)

M r = Δ M C t (14)

where, M_r, ΔMC and t represents rate of moisture loss, change in moisture content and time taken for the drying process respectively.

• Volume of fuel consumed (V_f)

(V_f) = Level of fuel in tank after drying period

× Length of fuel tank (22 cm) × breath of fuel tank (20 cm) (15)

where

level of fuel consumed for drying = Initial level of fuel – finial level of fuel (16)

where V_f = volume of fuel used.

• Thin layer mathematical modelling

Key parameters monitored were MC of maize (maize drying rate), temperature and relative humidity in the dryer and the ambient formed the basis for the thin layer drying kinetics modelling in the dryer using experimental models.

• Determination of moisture ratio (MR)

The moisture ratio during the thin-layer drying of maize in the dryer was calculated using Equation (17).

M R = M t − M e M o − M e (17)

where MR is the dimensionless moisture ratio, M_t is the moisture content (% db) at time t, and M_o and M_e are the initial and equilibrium moisture contents respectively, on dry weight basis. The values of M_e are relatively small compared to M_t and M_o, where error implied in the simplification is negligible [17]. The MR was therefore simplified as shown:

M R = M t M o (18)

The moisture ratio (MR) calculated from the MC of maize grains during the drying experiment were fitted to 10 common thin-layer drying models (Table 1) widely used to describe the drying characteristics of most food products. The mathematical drying models were used to assess and predict the moisture ratio of maize at an average drying air temperature of 50˚C. The best model was selected to describe the drying processes of maize in the dryer. The best fit model to the experimental data was selected using statistical parameters such as the coefficient of determination (R²) and the root mean square error (RMSE) to assess the goodness of the fit. The best fit model was that which resulted in higher R² and the lowest RMSE [18] [19].

The fitness of experimental data to the thin-layer drying models was evaluated as reported by [11] where statistical parameters such as the root mean square error (RMSE) and coefficient of determination (R²) were calculated using Equations 19 and 20 respectively. For better fitting procedures, [29] reported that higher R² and lower RMSE values should be obtained. Regression analysis was done using MATLAB (Version R2015a) statistical computer program.

R M S E = [ 1 N ∑ i = 1 N ( M R e x p , i − M R p r e , i ) 2 ] 1 2 (19)

R 2 = 1 − ∑ i = 1 N [ M R e x p , i − M R p r e , i ] ∑ k = i N [ M R p r e , i − ∑ k = 1 n M R p r e , i N ] (20)

where:

M R p r e , i is the predicted moisture ratio; M R e x p , i is the experimental moisture ratio; N is the number of observations/readings.

Figure 5 shows the temperature variation across the furnace and plenum compared to the ambient temperature. The plenum temperature varied between 38˚C at the start and peaking at 70˚C when the furnace combustion chamber was restocked with maize cobs. Overall an average plenum temperature of 52.8˚C was recorded while the average ambient temperature was 33.8˚C. The high temperature trend recorded in the plenum during the drying process was due to the high supply of heat energy from the combustion of maize cobs in the furnace of the AflaSTOP dryer. The recorded temperature in the dryer was high enough to dry the maize grains from the initial MC of 17.5% to 11.5% in 3 hours. According to [3], an average temperature of 50˚C is required for effective seed maize drying. The recorded temperature at the plenum of the dryer is suitable for drying seed maize as well.

Variation in the relative humidity (RH) of the ambient and plenum air is shown in Figure 6. The ambient had an average RH of 54.8% compared to the plenum average RH of 27.7%. The lower the RH of the air, the higher its ability to absorb moisture. The lower RH at the plenum of the dryer was due to the high temperature which created turbulent effect causing water molecules in the grains to evaporate. This phenomenon caused rapid moisture extraction in the grains as the water molecules were energized by the high drying air temperature to attain high kinetic energy to evaporate the moisture bound in the grains. The water molecules were easily absorbed carried away by the low RH of the surrounding air.

The moisture content (MC) of maize grains were monitored at hourly intervals by sampling grains at different positions in the shallow bed unit. The grains were mixed thoroughly before the measurement was done. The change in moisture content of maize with drying time in the AflaSTOP dryer is presented in Figure 7. It was observed that, moisture reduction in maize varied across the drying bed of the dryer with grains closer to the inlet of the heat source drying faster than grains further away from the heat source. However, on average, it was observed that, 3 hours of drying time was used to dry 250 kg of maize from a moisture content of 17.5% to an average moisture content of 11.5%. This resulted in a drying rate of about 2% moisture reduction per hour. In all, about 45 kg of maize cobs and 0.9 liters of diesel was used to evaporate 28.8 kg of water. Other performance indicators of the drying experiment is presented in Table 2.

The calculated thermal efficiency (29.6%) reveals the percentage of heat used to extract moisture from the grains. The drying system efficiency of 81.9% was high indicating the dryer is effective in removing moisture (29 kg) from the maize grains dried within the 3-hour period. Similar observations were made by [30] [31]. The study by [30] on the performance of a biomass fired dryer for copra drying reported a thermal and drying efficiency of 26% and 16% respectively. Work done by [31], on the performance of a biomass dryer for cashew nut drying also reported thermal efficiency and drying efficiency of 35% and 30.5% respectively.

Thin layer drying models presented in Table 1 were fitted to the experimental data. The values of coefficient of determination (R²) and RMSE for the selected models used to predict the drying behaviour of maize dried in the AflaSTOP dryer are presented in Table 3. The suitable model according to [11], is the model with the highest R² value and the lowest RMSE value. The Logistics model satisfactorily described the drying kinetics of maize dried in the AflaSTOP dryer.

From Table 3, the Logistics model recorded the highest R² value and the least RMSE values of 0.9902 and 0.04908 respectively. This makes it suitable for predicting the drying kinetics of the maize drying process in the AflaSTOP dryer. Drying of the maize grains occurred entirely in the falling rate period which involved moisture removal from the grains by the diffusion process where moisture from inside was transported towards the surface of the grains. During this period, the surface of a maize grain is not covered with a thin layer water, because the internal resistance to moisture transport has become greater than the external resistance. This mechanism of diffusion is best described by Fick’s second law of diffusion which assumes that the resistance to water diffusion in a product occurs in the outer layer of the product resulting in the continuous decrease in the drying rate during the course of drying [29]. Other studies have reported the selection of the Logistics model, a semi-theoretical drying model for predicting the drying behavior of some other crops during thin-lay drying experiments. The model has been successfully used to examine the drying kinetics of parsley [32], bamboo sheet [33], and rice [34].

The experimental and predicted data values using the Logistics model for moisture ratios of maize dried in the AflaSTOP dryer are shown in Figure 8. It shows the relationship between MR_pre, MR_exp and drying time. The overlapping data points show the closeness of the predicted and experimental values. It was observed that the Logistics model provided a very good conformity between the

experimental and the predicted moisture ratios and best describes the drying kinetics of maize using the AflaSTOP dryer.

Figure 9 presents the comparison of the MR_exp and MR_pre values of maize dried in the dryer. It was observed that, the predicted data banded around the ideal trend line indicating the suitability of the Logistic model in predicting the

drying behaviour of maize in the AflaSTOP dryer. The values obtained from the model were almost the same showing a uniform declination in moisture content from the start of the drying process to the end. The declination in moisture ratio can be attributed to high drying temperature, low relative humidity and high air velocity of the drying air.