Crack growth and coalescence from discontinuities is the fundamental process that underlies the majority of rock mass failures. Still, however, numerical simulation of this process remains a formidable challenge. The main reasons include that tensile and shear fractures are often mixed in the cracking process, that the cracking patterns depend strongly on the configuration of preexisting discontinuities, and that the cracking process may take place at different scales. The purpose of this paper is to evaluate the capabilities of the recently developed double-phase-field model (Fei and Choo, 2021) for simulating mixed-mode fracture in laboratory- and field-scale rocks with discontinuities. Simulation results show that the double-phase-field model can well reproduce laboratory test data, in terms of not only qualitative mixed-mode cracking patterns but also quantitative stress-strain responses. The results further demonstrate that the model can simulate complex rock fracture processes at the field scale such as slope failure due to crack growth and coalescence from joints.
Crack growth and coalescence from discontinuities is a primary mechanism underlying the failure of rock masses. A myriad of studies has investigated the processes and consequences of cracking in rocks with discontinuities at the field and laboratory scales, bridging their observations across scales (see Einstein, 2021 and references therein). These studies have commonly found that rock cracking processes usually involve a combination of tensile (mode I) and shear (mode II) fractures.
Meanwhile, many types of numerical methods have been advanced to simulate crack growth and coalescence from rock discontinuities. These methods may be grouped into two categories: (i) macroscopic (continuum-based) methods, and (ii) microscopic (grain-based) methods. Macroscopic methods (e.g., Bobet and Einstein, 1998a; Shen and Stephansson, 1994; Wu and Wong, 2012) can be applied to both laboratory- and field-scale problems; however, they require one to carefully select criteria for crack nucleation and propagation and often involve complex algorithms for simulating the onset and growth of cracks. On the other hand, microscopic methods (e.g., Zhang and Wong, 2012, 2013; Zhang et al., 2019) can simulate fracturing processes based on first principles and without complex algorithms; however, their applicability is inherently limited to small-scale problems. Thus, although the microscopic methods are an indispensable tool to gain insights into the physics of rock cracking, the macroscopic methods remain useful for addressing engineering-scale problems in rock mechanics.