A two-parameter damage-controlled strength model to predict the fracturing of brittle rock mass at great depth

The prediction of brittle failure under excavation in high-stress hard rock materials is always a challenge in rock mechanics and geological engineering in the past few decades. Existing researches have reflected that the rock failure is a progressive process involving crack initiation, propagation and coalescence. Nevertheless, a quantitative approach for characterizing the progressive failure of brittle rocks remains challenging and requires further research. In this study, we investigate the mechanical parameters variation of deep-buried granite sampled from Shuangjiangkou hydropower station considering the advancement of microcrack growth derived from an ordered series of cyclic loading and unloading experiments. The results indicate that the cohesion decreases while frictional strength becomes more dominant in the failure mechanism with damage accumulation. We adopt plastic shear strain to depict the damage degree quantitatively, and find the relation between cohesion and plastic strain can be well depicted by a two-parameter negative exponent function, while that between internal friction angle and plastic strain can be depicted by a two-parameter power function. Besides, tensile strength decreases to zero rapidly after cracks initiate. We then perform numerical simulation to investigate the deformation and stress conditions in rock mass under excavation of an underground powerhouse based on the proposed two-parameter damage-controlled strength model. Our results, which compare coupled shear and tensile failure, align well with real cases as well as the failure location and depth around the tunnel face. The stress and deformation process can then be accurately captured layer by layer under the excavation. This study provides a significant benchmark for predicting the progressive fracturing of brittle rock masses in deep tunnels, as well as for regulating safe and efficient site construction progress.