Investigating the Influence of Pressure Distribution on Nucleation Size in Induced Seismic Events Using a Coupled Dynamic Reservoir Simulation with Adaptive Time-Step

Abstract

ABSTRACT Injection-induced seismicity poses a significant technical and socio-political risk to subsurface storage and resource engineering systems. A key challenge is to understand the nucleation process bridged between the dynamic rupture and flow-driven poromechanical deformation (quasi-static) and ensure the ability to capture the onset of ruptures. To address this challenge, a hybrid time-step controller was developed in this research. The controller integrates local discretization error and Coulomb friction and is capable of capturing pre-seismic triggering, co-seismic spontaneous rupture, and arrest. Importantly, the controller operates solely on the solution state, eliminating the need for theoretical indicators that may fail to capture transitions under heterogeneous pressure distribution. The controller was implemented using an A Posteriori approach within a mixed discretization scheme that combines the extended finite element method (XFEM) for poromechanics and the embedded discrete fracture model (EDFM) for multiphase flow, incorporating implicit Newmark scheme for inertial mechanics and a Lagrange Multiplier approach with a slip-weakening friction model to enforce fracture contact constraints. The results demonstrate that the distribution of the pore pressure profiles along the fault may have an effect on the nucleation size. Overall, this study provides insight into the nucleation process and offers a powerful tool for managing the risk of injection-induced seismicity in subsurface storage and resource engineering systems. INTRODUCTION Induced seismicity is a phenomenon that has become increasingly relevant as human activities such as fracking and wastewater disposal have been linked to earthquakes. One knowledge gap in the understanding of this process pertains to the critical size of a ruptured sliding patch along a fault patch beyond which, spontaneous sliding is to occur (producing considerable seismic events). An understanding of the nucleation process is essential to predicting and managing injection processes. Early laboratory studies of stick-slip shear failures suggest that the nucleation process involves two phases. The first stage entails a slow and longer-term propagation of slip. This may be followed by a sudden transition to the second phase with shorter periods of accelerated slip, where the sliding patch size and velocity accelerate very rapidly. The physical properties of the fault, such as its frictional behavior, and the normal stress acting on it are known to strongly influence this nucleation process (Dieterich, 1978; Okubo and Dieterich, 1984; Ohnaka and Shen, 1999). Reported theoretical and numerical efforts aim to quantitatively understand the nucleation process under idealized conditions. A particular focus has been on deriving an instability criterion for either slip-weakening or rate-and-state friction models (e.g., Campillo and Ionescu (1997); Favreau et al. (1999); Uenishi and Rice (2003); Rubin and Ampuero (2005); Ampuero and Rubin (2008); Kaneko and Ampuero (2011); Latour et al. (2013); Gvirtzman and Fineberg (2021)). According to the driving mechanisms behind the growth of the patch, the derivation of this instability criterion can be divided into two major classes: one is based on linear stability analysis where a stress criterion is applied because the yielding of the contact surface pre-dominates, and another is based on Griffth's criterion with energy consideration when the fracture mechanics takes over. However, the theoretical nucleation size assumes uniform stress distributions and only considers mechanics. Under realistic injection conditions, the fluid pressure field is spatially and temporally variable. This work explores the influence of such complexity on the nucleation process.