Simulation and Experimental Validation of the Operating Parameters Influencing Groundnut Sheller Performance

Groundnut pod shelling is an important post-harvest process, the locally produced shellers have low shelling efficiency, high kernel damage and poor energy utilization. As a consequence, smallholder groundnut farmers rely on manual shelling, which is laborious and time-consuming. Additionally, the experimental development and testing of shellers also require repeated fabrication and physical trials, thereby increasing development time and cost. This study evaluated applicability of Discrete Element Method (DEM) to simulate performance of a rasp bar groundnut sheller for Ndovu variety. A three factor, three level Box-Behnken Design examined the influence of feed rate (100 -500 kg h^-1), concave clearance (10-0.025 m) and drum speed (2.36-4.71 m s^-1) on shelling efficiency, kernel damage and throughput capacity per unit power requirement. Statistical validation was performed using paired sample Student’s t-test, regression analysis, and residual errors analysis. The results indicated shelling efficiency and kernel damage were strongly influenced by drum speed, where a higher drum speed from 2.3 m s^-1 to 4.71 m s^-1 improved the rate of shelling (87.8% to 92.0%) and a higher kernel damage (3.76% to 7.79%). Throughput capacity per unit power requirement increased from 501 to 738.86 kg h ^-1 kWh ^-1 as the feed rate increased from 100 to 500 kg h^-1. An increase in concave clearance from 0.01 m to 0.025 m reduced shelling efficiency, 99.12% to 86.02 and kernel damage from 7.6% to 1.6%. The DEM simulation model reproduced the trend from experimental responses. Shelling efficiency was predicted with high accuracy with R²>0.9, MAPE<2% and no significant statistical difference. Kernel damage models predicted responses with R²>0.95 and paired Student’s t-test p>0.05 but a MAPE>10% indicating that it underestimated damage in highest speeds and lowest clearance. Throughput capacity per unit power model yielded significant differences and high residual errors with actual values indicating inability to account for machine power losses from transmission and friction. Overall, DEM simulation was effective as a pre fabrication tool with adjustments required to reflect energy losses under the tested operating conditions.