Investigating brittle rock failure and associated seismicity using laboratory experiments and numerical simulations

Rock failure process is a complex phenomenon that involves elastic and plastic deformation, microscopic cracking, macroscopic fracturing, and frictional slipping of fractures. Understanding this complex behaviour has been the focus of a significant amount of research. In this work, the combined finite-discrete element method (FDEM) was first employed to study (1) the influence of rock discontinuities on hydraulic fracturing and associated seismicity and (2) the influence of in-situ stress on seismic behaviour. Simulated seismic events were analyzed using post-processing tools including frequency-magnitude distribution (b-value), spatial fractal dimension (D-value), seismic rate, and fracture clustering. These simulations demonstrated that at the local scale, fractures tended to propagate following the rock mass discontinuities; while at reservoir scale, they developed in the direction parallel to the maximum in-situ stress. Moreover, seismic signature (i.e., b-value, D-value, and seismic rate) can help to distinguish different phases of the failure process. The FDEM modelling technique and developed analysis tools were then coupled with laboratory experiments to further investigate the different phases of the progressive rock failure process. Firstly, a uniaxial compression experiment, monitored using a time-lapse ultrasonic tomography method, was carried out and reproduced by the numerical model. Using this combination of technologies, the entire deformation and failure processes were studied at macroscopic and microscopic scales. The results not only illustrated the rock failure and seismic behaviours at different stress levels, but also suggested several precursory behaviours indicating the catastrophic failure of the rock. Secondly, rotary shear experiments were conducted using a newly developed rock physics experimental apparatus ERDÎź-T) that was paired with X-ray micro-computed tomography (ÎźCT). This combination of technologies has significant advantages over conventional rotary shear experiments since it allowed for the direct observation of how two rough surfaces interact and deform without perturbing the experimental conditions. Some intriguing observations were made pertaining to key areas of the study of fault evolution, making possible for a more comprehensive interpretation of the frictional sliding behaviour. Lastly, a carefully calibrated FDEM model that was built based on the rotary experiment was utilized to investigate facets that the experiment was not able to resolve, for example, the time-continuous stress condition and the seismic activity on the shear surface. The model reproduced the mechanical behaviour observed in the laboratory experiment, shedding light on the understanding of fault evolution.Ph.D