SPH-based modelling of the entire rock caving process: insights into failure mechanisms

Cave mining is a cost-effective method for extracting large, low-grade orebodies, but its success depends on the cavability of a rock mass, governed by complex interactions between fracturing, fragmentation, and material flow. Despite extensive research, two competing failure mechanisms dominate the literature: (1) a continuous damage profile ahead of the cave back, as interpreted in the Duplancic conceptual model, and (2) discrete parallel fracture banding observed in experimental and field studies. However, no continuum-based model has fully reproduced and explained these failure mechanisms. This study presents the first continuum-based numerical framework that captures the entire caving process and replicates experimentally observed failure mechanisms, both qualitatively and quantitatively. Using Smoothed Particle Hydrodynamics (SPH) coupled with an advanced damage-plasticity model, which accounts for rock failure under different loading conditions, ranging from compression-shear to tensile-shear failures and material flow, parallel fracture banding observed in physical experiments is successfully reproduced, challenging the assumption of a continuous damage profile. The simulation results indicate that parallel fracture banding arises due to the consistent undercutting span and the transition from compression-shear failure to tensile-shear failure during cave propagation. Additionally, based on our current simulation results, the dilation angle tends to play a decisive role in cave stability: a high dilation angle promotes stable arching, supporting the Duplancic model, whereas a low dilation angle facilitates progressive failure through discrete fracture bands. Further analysis reveals that tensile fracture energy, material randomness, and horizontal stress significantly influence fracture propagation and caving efficiency. Strong agreement between simulations and centrifuge test data validates the proposed framework as a robust tool for studying rock caving processes and provides critical insights into the underlying failure mechanisms.