Grain fracture in the Level-Set Discrete Element Method: Can we distinguish between splitting, chipping, and fragmentation?

The fracture of individual grains can fundamentally change the mechanical load transmission, microstructure evolution, and energy dissipation of granular materials. Nonetheless, breakage models in the discrete element method (DEM) remain limited to simple failure criteria, being unable to distinguish between different fracture paths such as those created by splitting, chipping, and fragmentation. The bonded particle method (BPM) is a viable solution, but incurs a high computational cost due to having to simulate hundreds or thousands of sub-particles per grain. A comprehensive and efficient single-grain breakage model that can deal with different fracture paths and failure modes is currently missing. \nIn this work, we implement fracture in the level-set discrete element method (LS-DEM) in the open-source software YADE. First, we look at several optimisations to reduce the computational cost of LS-DEM and decrease the runtime by 22%. Second, we briefly review current breakage models and where they fail by simulating a series of well-known mechanical tests with established failure modes (e.g. Brazilian splitting and three-point bending). Third, we propose an improved breakage model to capture more realistic failure modes and fracture paths. A comparison of the new model using the aforementioned mechanical tests on single-grain fracture shows notable improvement over previous breakage models. \nThe improved breakage models will help to investigate the roles of chipping and fragmentation in the early production of fines in both natural processes and comminution. Future work will be aimed at these novel applications of the LS-DEM fracture model as well as the conservation and regularisation of fracture energy.

Apr 27, 2026 — Apr 30, 2026