SR-DEM Recovers Amontons-Coulomb Friction from Explicit Asperity Contacts in 2D Granular Simulations
SR-DEM demonstrates that explicit asperity resolution can recover standard friction laws yet expose additional many-body behaviors missed by current DEM. The approach trades computational cost for physics fidelity and uses neural acceleration to remain practical. Evidence is limited to 2D simulations; 3D validation and experimental benchmarks are required next.
The framework trades a simple Amontons-Coulomb law for direct modeling of surface asperities whose collective frictionless collisions generate emergent friction. An FE2-style algorithm plus a FiLM neural surrogate, both GPU-parallelized, offsets the added cost of resolving topography at each contact. In 2D tests the method reproduced classic two-body friction coefficients from minimal roughness representations while exposing angle-of-repose values that deviate from single-parameter DEM predictions under identical complexity. Applications cited include suspension rheology design and metamaterial mechanics where many-body surface effects dominate.
Pan et al.: Within 18 months a 3D SR-DEM implementation will match experimental repose angles within 5 percent for at least two granular materials where standard DEM deviates by more than 15 percent.
Sources (2)
- [1]Primary Source(https://arxiv.org/abs/2610.33363)
- [2]Supporting Source(https://doi.org/10.1016/j.cma.2023.116432)