The dryer is located at Menengai GDC station, and utilizes thermal energy from Menengai well 12 (MW-12) and has a maximum handling capacity of 6 tones [1]. Figure 1 shows the main operational parts which include the heat supply system, dryer, storage bin, wet bin, and elevator motors.

The elevator motor loads the wet grains to the dryer, through the dryer chute. Brine is supplied to a heat exchanger at a temperature of 90˚C and leaves at 76˚C [2]. Hot geothermal water provides the thermal energy that heats fresh water from room temperature to 78˚C. The hot water is then supplied to the radiator at a temperature of between 78˚C and 55˚C [2]. Figure 2 below shows the heat exchange system between the brine and fresh water.

A suction pump then blows ambient air through the hot water tubing to the dryer. The supplied hot air enters the grain buffer section in the drier through evenly spaced openings across the four compartments. After attaining the desired moisture content, the dried grains are loaded to the storage bin.

This study entailed both ANSYS Fluent simulation and physical experiments to validate the model. Physical experiments were conducted to determine the extent of uneven grain drying and hot air distribution. A calibrated Moisture Analyzer, Serial No.0202777, superpoint, manufactured by Supertech Agroline Ltd, Denmark, was used to measure the moisture content (wet basis) in each compartment. Validated temperature-humidity meters, UT333, manufactured by Uni-Trend Technology, China, were used for measuring the temperature and relative humidity of air at the inlet and exhaust sections.

The first experiment involved drying 2.7 tons of maize with initial moisture content of 21.1%. The grains were filled up to half of the third compartment. Grain drying parameters such as the hot air supply rate, temperature distribution, relative humidity, time, and grain moisture content were monitored throughout the experiment. The rate of air supply to the dryer was derived analytically based on the suction pump power ratings and the cross sectional area of the hot air inlet. The hot air was supplied at an average speed of 1 m/s and at a temperature of 50˚C. Grain drying began at 1:40 pm and moisture contents were measured after every 20 minutes. Due to uneven drying, grain recirculation began after two hours. The mixed grains were then sampled after every 20 minutes to measure the moisture content. The data was manually collected from the moisture analyzer and temperature-humidity meters. They were then tabulated and analyzed in Microsoft excel to determine the trends.

The second experiment was conducted to determine temperature distribution across the dryer without loading grains. Temperature magnitude in all the four compartments and exhaust sections were recorded after every 10 minutes. The goal was to study air distribution in the dryer without the load and also to provide validation data.

Detailed dimensions of the geothermal grain dryer were collected from physical measurements and then modeled in Autodesk Inventor. The study only focused on the dryer section and did not include the wet and storage bin as the goal was to model air flow in the grain buffer section. Figure 3 shows parts of the dryer geometry that was modeled in Autodesk Inventor. It has the air inlet and outlet sections and four compartments in the grain buffer section.

The grain dryer has four compartments with evenly spaced fins that guide air flow through the dryer. Figure 4 shows the dimensions of the dryer, including the grain buffer section. The fins provide sufficient space for air to penetrate through the grains. Consequently, the design allows sufficient interaction between the grains and hot air, thereby improving drying efficiency.

The 3D CAD geometry of the dryer was developed in Autodesk investor. It was then imported to ANSYS Fluent for flow modeling. Air flow through the dryer was modelled without incorporating the grains to determine the velocity, temperature, and turbulence kinetic energy distribution across the dryer. Flow distribution was modeled under steady state conditions as mean flow parameters at the inlet and outlet compartments did not significantly change with time. The velocity, temperature, and turbulence kinetic energy distribution across the dryer were recorded in vectors and contours.

Navier-strokes equations were used to solve the fluid flow problem in the ANSYS Fluent. They included three momentum, energy, and mass conservation shown in Equations (1) to (6) [18].

x-momentum, u = ρ D u D t = δ ( ρ u ) δ t + d i v ( ρ u U ) = δ p δ x + d i v ( μ g r a d u ) + S M x (1)

y-momentum, v = ρ D v D t = δ ( ρ v ) δ t + d i v ( ρ v U ) = δ p δ x + d i v ( μ g r a d v ) + S M y (2)

z-momentum, w = ρ D w D t = δ ( ρ w ) δ t + d i v ( ρ w U ) = δ p δ x + d i v ( μ g r a d w ) + S M z (3)

Internal Energy, i = δ ( ρ i ) δ t + d i v ( ρ i U ) = − p d i v U + d i v ( k g r a d T ) + ∅ + S i (4)

Mass Conservation, δ ( ρ ∅ ) δ t + d i v ( ρ ∅ u ) = ρ D ∅ D t (5)

The general transport equation solved through finite volume Computational Fluid Dynamics (CFD) program is;

δ ( ρ ∅ ) δ t + d i v ( ρ ∅ U ) = d i v ( γ g r a d ∅ ) + S ∅ (6)

where, ρ = Density, U = Velocity term, γ = Diffusion constant, S = Source term, µ = Viscous stress constant, k = Grain drying constant, p = Pressure, ∅ = Viscous dissipation term that describes conservation of mechanical energy to heat [18].

Given that the hot air was unevenly distributed across the grain buffer section,

design modifications were conducted to determine its effect on flow distribution. Hot air flow-guides with filleted edges were introduced at the inlet section. The guides ensured that hot air was evenly distributed across the four compartments in the dryer. The filleted edges also enhanced fluid transition to the grain buffer section. A 500 m fillet reduced turbulence in the fourth compartment. Also, 110 mm fillets were put in each of the other three flow-guides in compartments 3, 2, and 1. The sizing criteria were based on the need to equally divide the incoming hot air into the four compartments. Therefore, the flow guides were equally spaced at a distance of 137.5 mm. All the flow-guides began from the base of the first compartment and stretched to the end of the air inlet sections of each compartments. Table 2 indicates the sizing parameters for the improved dryer. Figure 5 shows the modifictions that were incoporated in the dryer to improve hot air flow distribution across the dryer. Figure 5(a) and Figure 5(b) shows the dimensions of the current and improved dryer respectively.

The geometry of the fluid domain for the current and improved dryer is shown in Figure 6(a) and Figure 6(b) below, respectively. Figure 7 is the CFD model in the ANSYS Fluent domain.

Mesh sensitivity was conducted for the current dryer at 100 mm, 75 mm, 60 mm, and 50 mm. The convergence tolerance reduced with the decrease in mesh size. From 75 mm mesh size, there was insignificant change in temperature and velocity contours. However, computation time significantly increased as the mesh was reduced from 60 mm to 50 mm. The increase in computational time is attributed to the rise in the total number of nodes and elements as the mesh size reduced.

Figure 8(a) is the velocity and temperature contours as the mesh size was reduced from 100 mm to 75 mm. As the mesh size reduced, the convergence tolerance decreased from 2.899022e−04 to 6.329312e−06. The maximum temperature remained 49.85˚C even with the reduction in mesh size. However, the minimum temperature reduced from 38.16˚C to 40.12˚C while the maximum velocity declined from 1.2 to 1.1 m/s.

The temperature distribution contours remained relativly uncanged from 75 mm to 50 mm Mesh size. There was a slight change in velocity contours from 100 mm to 75 mm mesh size. However, the contours remained relatively unchanged as the mesh size reduced from 60 mm to 50 mm. Given that the temperature and velocity distributions were unchanged between 60 to 50 mm as shown in Figure 8(b), a further reduction in mesh size was not necessary. Therefore, a mesh size of 50 mm was considered for the simulation model.

As the mesh size reduced from 100 mm to 75 mm, the maximum velocity slightl increased from 1.36 to 1.37 m/s as shown in Figure 9(a). The maximum temperature remained unchanged at 49.85.

The temperature and velocity contours were gradually refined with the reduction in mesh from 100 mm to 50 mm. The convergence tolerance also reduced with the decrease in mesh size. However, the temperature and velocity contours did not significantly change as the mesh was reduced from 60 mm to 50 mm, Figure 9(b). Therefore, a mesh size of 50 mm was also considered for the improved model.

Velocity magnitudes were sampled at the inlet, center, and outlet sections of the grain buffer section for 100 mm, 75 mm, 60 mm, and 50 mm mesh sizes. The results displayed in Table 3 and Figures 10(a)-(d), indicates that the velocity distribution did not significantly change as the mesh was reduced from 75 mm to 60 mm and then 50 mm.

Based on the results, the maximum velocity deviation from 75 mm to 50 mm mesh was 0.05 m/s. The velocity distribution for 75 mm, 60 mm, and 50 mm were nearly the same. However, the time taken to solve the energy equations significantly increased with further reduction in mesh size. Therefore, a mesh size of 50 mm was considered for the CFD model.

The experimental results confirm the existence of uneven grain drying in the current geothermal dryer. Grains near the hot air inlet dry at a considerably higher rate that those near the exhaust section. Further, grains in the first compartment near the hot air inlet dried at a faster rate than those in the second and third compartments. The temperature at the inlet section ranged between 51˚C and 50˚C and 44˚C to 41˚C in the exhaust section. Table 4 shows the

temperature distribution along with moisture content in the three compartments occupied with grains.

After two hours of drying, the moisture content in compartment 1 near the inlet section was 15.6% and 17.8%, wet basis in the exhaust section. Wettest grains had a moisture content of 19.5% and were in the third compartment, near the exhaust section. Grain recirculation was initiated to ensure uniform drying. However, the rate of moisture reduction during recirculation was considerably low. After 2 hours 30 minutes of simultaneous grain drying and recirculation, the moisture content reduced to 16.5%. Therefore, it took 4 hours 30 minutes to reduce the grain moisture content from 21.1% to 16.5% as shown in Figure 11 below.

Based on the data, temperature distribution varies slightly across the dryer. The hottest sections are located near the hot air inlet. Conversely, the sections near the exhaust have the lowest temperature. The drop in temperature is attributed to losses along the walls of the dryer.

The no-load experiment affirms that the air flow is statistically steady as the average flow features do not vary with time. Table 5 shows recorded temperature distribution at different points of the dryer. The temperature magnitudes at the inlet, exhaust, and 4 compartments did not significantly change with time. The temperature in the inlet compartments is considerably higher than that in the exhaust section.

The model was validated using no-load experimental results obtained from the current geothermal grain dryer in Menengai. The temperature distribution at the

inlet and outlet sections obtained from experiments was compared with the Ansys simulation results. Hot air entered the dryer at 50˚C, and at a velocity of 1 m/s. The temperatures at the inlet and exhaust sections of the compartments were taken at steady state conditions. The results are shown in Table 6.

Figure 12 compares the experimental temperature distribution with that of the simulation model. The average percentage error obtained by comparing the experiment with simulation results is −0.272. The small percentage error shows that the Ansys simulation model is a good representation of the geothermal grain dryer. The obtained temperature distribution can be corrected by adding 0.272 to the displayed values.

Table 7 and Table 8 below shows the velocity and temperature magnitudes sampled at the inlet, center, and outlet sections for both current and improved dryer. The average values are also indicated to quantify the percentage increase in magnitude.