Crushing, Comminution and Fracture: Extreme Particle Deformation in Three-Dimensional Granular Aggregates

TL;DR

This paper introduces a high-fidelity 3D computational framework using peridynamics for simulating the deformation and fracture of granular aggregates, capturing both inter- and intra-particle behaviors without explicit crack tracking.

Contribution

The novel framework models deformable brittle grains with autonomous fracture evolution using a nonlocal continuum approach, validated through benchmark tests and large-scale simulations of irregular grain aggregates.

Findings

Convergence of bulk stress response under compression.

Feasibility of constructing representative volume elements (RVEs).

Excellent scalability on high-performance computing systems.

Abstract

We present a high-fidelity three dimensional computational framework for simulating the bulk mechanical behavior of granular aggregates composed of deformable brittle grains. Departing from classical discrete element methods (DEM), our approach captures both inter-particle and intra-particle deformation using a nonlocal continuum formulation based on peridynamics. Each grain is individually meshed from level-set representations, enabling accurate modeling of elastic response and autonomous fracture evolution without requiring explicit crack tracking or fragment reconstruction. We validate the method through benchmark simulations, including the Kalthoff-Winkler fracture test, crushing of hollow spheres, and compound impact-crushing scenarios. The framework is further applied to large aggregates of up to 1000 sand grains of irregular shapes reconstructed from three dimensional X-ray computed tomography. Simulations reveal convergence of bulk stress response under compression, suggesting the feasibility of constructing representative volume elements (RVEs) for multiscale modeling. Finally, we investigate the role of grain geometry and topology on the macroscopic strength of the aggregate, providing insight into microstructure-driven failure mechanisms. The framework exhibits excellent strong and weak scaling behavior, with simulations executed on up to 1600 cores, demonstrating its suitability for high-performance computing environments and large-scale modeling.