Failure of rock under impact loadings involves complex micro-fracturing and progressive damage. Strength increase and splitting failure have been observed during dynamic tests of rock materials. However, the failure mechanism still remains unclear. In this work, based on laboratory tests, numerical simulations with the particle flow code(PFC) were carried out to reproduce the micro-fracturing process of granite specimens. Shear and tensile cracks were both recorded to investigate the failure mode of rocks under different loading conditions. At the same time, a dynamic damage model based on the Weibull distribution was established to predict the deformation and degradation behavior of specimens. It is found that micro-cracks play important roles in controlling the dynamic deformation and failure process of rock under impact loadings. The sharp increase in the number of cracks may be the reason for the strength increase of rock under high strain rates. Tensile cracks tend to be the key reason for splitting failure of specimens. Numerical simulation of crack propagation by PFC can give vivid description of the failure process. However, it is not enough for evaluation of material degradation. The dynamic damage model is able to predict the stress-strain relationship of specimens reasonably well, and can be used to explain the degradation of specimens under impact loadings at macro-scale. Crack and damage can describe material degradation at different scales and can be used together to reveal the failure mechanism of rocks.
To reveal stress distribution and crack propagation of Brazilian discs under impact loads, dynamic tests were conducted with SHPB (split Hopkinson pressure bar) device. Stress states of specimens were monitored with strain gauges on specimen surface and SHPB bars. The failure process of specimen was recorded by ultra speed camera FASTCAM SA1.1 (675 000 fps). Stress histories from strain gauges offer comprehensive information to evaluate the stress equilibrium of specimen in time and space. When a slowly rising load (with loading rates less than 1 200 N/s for d 50 mm bar) is applied, there is usually good stress equilibrium in specimen. The stress distribution after equilibrium is similar to its static counterpart. And the first crack initiates at the disc center and propagates along the load direction. But with the front of incident wave becoming steep, it is hard for specimens to get to stress equilibrium. The first crack may appear anywhere on the specimen together with multiple randomly distributed secondary cracks. For a valid dynamic Brazil test with stress equilibrium, the specimen will break into two halves neatly. While for tests with stress disequilibrium, missing strap may be found when broken halves of specimens are put together. For those specimens broken up neatly at center but having missing wedges at the loading areas, it is usually subjected to local buckling from SHPB bars.