An energy-conserving polyhedron–sphere discrete element method for bearing capacity evaluation of stone-filled embankments

Polyhedral particles provide an accurate geometric representation of irregular gravel in discrete element simulations, but their application is often limited by the high computational cost associated with complex contact detection. To overcome this limitation, this study develops an energy-conserving polyhedron-sphere discrete element method in which polyhedral particles are combined with spherical elements within a unified framework. A contact algorithm based on energy-conserving contact theory is established for polyhedron-sphere interactions, reducing contact detection to interactions between surface triangles of polyhedral and spherical elements. The stability and robustness of the proposed contact formulation are verified through collision tests, followed by packing simulations that demonstrate a substantial improvement in computational efficiency. This method is subsequently applied to the numerical simulation of the California Bearing Ratio test to evaluate the bearing capacity of stone-filled embankments. Numerical results show close agreement with corresponding experimental results, confirming the reliability of the model in reproducing macroscopic bearing behavior. A parametric analysis is further conducted to examine the effects of coarse aggregate content and particle friction on bearing capacity. The results indicate that increases in both parameters enhance particle interlocking and load transfer efficiency, leading to a marked improvement in bearing capacity. These results demonstrate that the presented polyhedron-sphere discrete element method provides an efficient and reliable numerical framework for the analysis of granular materials with complex particle shapes.