Evaluation of PFC2D Grain-Based Model for Simulation of Confinement-Dependent Rock Strength Degradation and Failure Processes

Estimation of rock mass strength has become more critical in recent years due to the increase in the number of mining and civil projects at greater depths (> 2 km). Empirical approaches for the estimation of rock mass strength (e.g., Hoek-Brown/GSI) are primarily based on experiences at shallow depths and low confinement problems (e.g., tunnel wall failure), and therefore may not be representative for the strength in highly confined rock masses (e.g., for the core of pillars or abutments at depth). In this study an attempt is made to investigate the strength of rock masses, using the discrete element code PFC2D. For this purpose, the recently developed PFC2D Grain-Based Model (GBM) was used to match the laboratory response of intact and granulated Wombeyan marble. The term “granulated” refers to a heat treated sample where the grains have been completely separated at their boundaries due to anisotropy and contrast of their thermo-elastic properties. This material is considered to represent an analogue for a randomly jointed rock mass. It is shown that the PFC2D-GBM calibrated to unconfined and confined intact marble strengths and then to the unconfined granulated marble strength, underestimates the strength of the confined granulated marble. This problem was resolved by increasing the grain boundary friction angle in the granulated model to account for micro-scale roughness of the grain boundaries as observed in the microscopic image of the granulated marble. The calibration methodology taken to obtain micro-properties for both intact and granulated marble as well as implications for the determination of rock mass strength at various confinement levels are discussed.