A clumped particle model for rock

Abstract

Discrete element (particle) modeling is now often used to simulate the behavior of rock. While the method is attractive because it does not require the formulation of complex constitutive models, it requires extensive calibration with measured macro-scale results to determine the particle–contact parameters that will predict the macro-scale response. This calibration is usually carried out using the laboratory uniaxial compression test. The behavior of intact rock in compression is complex, resulting in a nonlinear failure envelope with high frictional resistance and a ratio of tensile to uniaxial compression strength of approximately 0.05–0.1. An extensive series of calibrations were carried out using a particle code to determine if the micro-parameters determined from uniaxial compression tests could be used to predict the intact rock response regardless of stress path, i.e., tension to triaxial compression. It is determined that the particle code cannot predict the failure envelope measured on laboratory samples, unless significant modifications are made. This study shows that adjusting the particle parameters had little effect on the macro-scale properties of compression and tension tests, but using a clumped-particle geometry improves the predictive capabilities of the particle code significantly. For both Lac du Bonnet granite and a weak synthetic rock, the particle code calibrated to uniaxial tests using the clumped-particle geometry predicts both the stress–strain behavior and the complete nonlinear failure envelope.