Development and Experimental Validation of a Multi-Sphere Discrete Element Method Model for Sesame Seeds

Accurate representation of particle shape is essential for reliable discrete element method (DEM) simulations of small, irregularly shaped seeds such as sesame, yet the effect of multi-sphere resolution on predictive accuracy remains poorly quantified. This study characterised the physical properties of sesame seeds and evaluated five multi-sphere particle models (9, 11, 15, 19 and 24 spheres) constructed in Ansys Rocky using the filling-sphere method. The contact parameters were transferred unchanged from a previous sesame study, and the measured seed–seed and seed–glass static friction and restitution coefficients agreed with the adopted values to within 4%. Each model was validated against experimental piling tests (angle of repose and bulk density; five repetitions), and the 19- and 24-sphere models were further validated by a hopper discharge test with 150 g of seed (three repetitions). The seeds had a mean length, width, and thickness of 3.12, 1.88, and 0.90 mm, respectively, and an experimental angle of repose of 20.74°. As the number of spheres increased from 9 to 24, the error in the simulated angle of repose decreased from 24.06% to 1.16%, and the error in bulk density from 5.67% to 0.12%. The 24-sphere model was the only model whose angle of repose did not differ significantly from the experimental value (p = 0.057), and it predicted the discharge time to within 1.70%, compared with 5.66% for the 19-sphere model, while requiring 2.9 times the run time of the 9-sphere model. Because its clump volume was within 0.8% of the measured seed volume, it achieved this agreement with four fewer spheres per particle than the 28-sphere model previously reported for sesame. The 24-sphere model is recommended for this seed lot under the tested conditions and provides a validated particle model for DEM simulation of sesame seed handling and planting equipment.