Nucleation and arrest of dynamic slip on a pressurized fault

Elevated pore pressure can lead to reactivation of slip on pre-existing fractures and faults when the static Coulomb failure is reached locally. As the pressurized region spreads diffusively, slip can accumulate quasi-statically (paced by the pore fluid diffusion) or dynamically. In this work, we consider a prestressed fault with a locally peaked, diffusively spreading pore pressure field to study (1) conditions leading to the escalation of slip and nucleation of dynamic rupture and (2) rupture run-out distance before it is arrested. Nucleation appears in this model when the fault friction decreases from its peak value with slip, while arrest of dynamic propagation is imminent on aseismic faults (i.e., such that prestress tau(b) is less than the residual fault strength tau(r) at ambient conditions). When fluid overpressure is a small-to-moderate fraction of the ambient value of normal effective stress (and prestress is large enough for fault slip to be activated by overpressure), dynamic rupture always nucleates, and the nucleation length increases with decreasing prestress practically independently of the overpressure value. Transition from the ultimately unstable (tau(b) > tau(r)) to the ultimately stable (tau(b) < tau(r)) fault loading is marked by a strong increase of the nucleation length (proportional to 1/(tau(b) - tau(r))(2)) as tau(b) approaches tau(r) from above. For aseismic faults (tau(b) < tau(r)), no dynamic rupture is nucleated at large fluid overpressures for all but the smallest values of prestress. The largest run-out distances of dynamic slip on aseismic faults correspond to overpressure/prestress just sufficient for slip activation. In such cases, the dynamically accumulated slip can lead to enhanced, dynamic fault weakening, resulting in a sustained dynamic rupture and generating a large earthquake. This is consistent with field observations when the largest injection-induced seismicity occurred after fluid injection ended.