The mechanical behavior of granular assemblies is strongly influenced by particle shape, yet many discrete element method (DEM) simulations rely on spheres or ellipsoids to reduce computational overhead and employ well-established stress-force-fabric relationships. Advanced imaging techniques such as X-ray micro-computed tomography (μCT) are increasingly capable of capturing high-resolution particle geometries but remain expensive and limited in availability. Superellipsoids capture key characteristics of natural particles – such as elongation, triaxiality, and varying surface curvature – through a relatively simple closed-form equation amenable to efficient DEM analysis. Herein, we investigate mixtures of arbitrarily shaped superellipsoids to better approximate the geometric diversity found in real grains and reference behaviour back to assemblies of monoshaped particles. Simulation results highlight that aspect ratio is the primary factor affecting compactivity (i.e., packing efficiency), shear strength, and fabric anisotropy. Spherical particles produce higher void ratios and lower shear strength, whereas more cubic shapes lead to denser packings and stronger force chains. Random distributions of shape parameters (sharpness and squareness) result in higher compactivity and shear strength, yet randomizing aspect ratios has an even more pronounced effect, particularly on fabric anisotropy. These findings underscore the significance of incorporating particle shape heterogeneity in DEM simulations, as an over-reliance on idealized spherical assumptions may underestimate both the compaction behavior and shear strength of granular materials.
Wen et al. (Mon,) studied this question.