Bonded-Particle Modeling of Fracture and Flow

Discrete-element methods allow one to simulate the movement and mechanical interaction of hundreds of thousands of discrete particles. Bonded-particle models (Potyondy, 2015) represent a solid as a bonded collection of discrete particles, and provide a synthetic material whose mechanical behavior ranges from that of a solid material (such as rock, concrete or ceramics) when the bonds are intact to that of a granular material when the bonds have all broken. The macroscopic behavior of such models is an emergent property of the system that arises from a small set of microproperties for the particles, bonds and particle-particle interactions. These models support investigation of the relations between microstructure and material properties under both quasi-static and fully dynamic loading, and can be used to simulate any physical process whose physics can be described by the interaction of discrete particles.

After introducing the basic concepts of a bonded-particle model, examples of how such models are being applied to model rock fracture and material flow are presented. The first example models a rock-cut test (at the mm scale), during which a cylindrical cutter is moved across the rock surface while monitoring forces on the cutter and damage in the rock as shown in Fig. 1. The rock is a sandstone, with the particles and bonds representing grains and cement, respectively. The second example models cave mining (at the 10–100 m scale), during which the undercutting of a rock mass and subsequent draw of the collapsed material fragments the rock mass in an upwardly progressive fashion (Pierce, 2010). The size of rock fragments in the selfpropagating cave decreases as the cave matures, and is attributed to the attrition of fragments in the course of traveling from their origin to the draw point. PFC3D (Particle Flow Code in 3 Dimensions) was used to study both compression- and shearing-induced secondary fragmentation, and the results of these simulations, in combination with in situ data, were used to develop a rapid draw simulator (REBOP, Rapid Emulator Based on PFC) for cave mining to predict material drawdown and ore recovery. The 30-minute talk focuses on the first example.