A novel way to determine number of spheres in clump-type particle-shape approximation in discrete-element modelling

摘要

Clump-type particle-shape approximation is often employed in discrete-element modelling (DEM). Previous studies have determined the number of spheres in a clump empirically. Herein, a novel approach is proposed to determine the number of spheres in a clump quantitatively on the basis of statistical optimisation using Akaike's information criterion (AIC). AIC quantifies the balance between the approximation accuracy of the clumps and the increase of the number of spheres in a clump. The proposed method is applied to determine a satisfactory number of spheres in a clump value for clumps of Toyoura sand particles. A series of clump-based isotropic-compression DEM simulations is conducted. When the number of spheres in a clump is higher than that determined by the proposed method, the simulated void ratio and the mean contact number converge to their respective values and agree with laboratory test data and X-ray computerised tomography measurements. Therefore, it is concluded that the proposed method is a valid one.