Rheology of a synthetic fault gouge simulated using smooth-joint contact model

Rate-and-state friction has provided a successful framework to study a range of fault slips, from stable to unstable. Under quasi-static loading, the transition between stable and unstable frictional sliding in 1D is thought to occur at the point where the loading stiffness, k, is equal to a rheology critical stiffness kc, expressed with normal stress (n), critical slip distance (Dc), and coefficients (b-a) that describe the rate-and-state direct and evolution effects: kc = n(b-a)/Dc. In this study, numerical simulations of rock deformation under triaxial compression were conducted using the Particle Flow Code (PFC) 3D, a discrete-element-method (DEM) code developed by Itasca Inc. The rock sample was a cylinder with 40mm in diameter and 107mm in length. Smooth-joint (SJ) contacts were deployed to simulate the gouge layer at a 60-degree dip angle. Preliminary tests showed that SJ contacts governed by Mohr-Coulomb failure criteria were best at reproducing ruptures along the fault. Further tests were conducted under various compression pressures. A series of fault slips, from stable to unstable, were observed from the simulations, which allowed for the estimation of the normalized critical fault stiffness despite the fact that the fault rheological characteristics were not simulated directly with rate-and-state friction. Although the model produced a range of fault slip, some results violated the established framework of rate-and-state friction. In particular, the synthetic fault slipped dynamically only when under lower normal stress on the fault. This inspired the development of a dynamic weakening algorithm applied onto SJ contact properties such as the cohesion and friction angle to correct the synthetic fault on its nominal friction behaviour. We present a dynamic weakening model that involved dropping contact cohesion locally around newly formed contact breakage. The new model was applied in the simulation of a previously-published experiment that used 150MPa confining pressure. The new model is shown to be successful at reproducing stick-slip instability from the experiment, and the fault nominal friction behaviour matches that of the rate-and-state friction.