Seismic Versus Aseismic Slip and Maximum Induced Earthquake Magnitude in Models of Faults Stimulated by Fluid Injection

The assessment of earthquake hazard induced by fluid injection or withdrawal could be advanced by understanding what controls the maximum magnitude of induced seismicity (Mmax) and the conditions leading to aseismic instead of seismic slip. This is particularly critical for the viability of renewable energy extraction through engineered geothermal systems, which aim at enhancing permeability through controlled fault slip. Existing empirical relations and models for Mmax lack a link between rupture size and the characteristics of the triggering stress perturbation based on earthquake physics. We aim at filling this gap by extending results on the nucleation and arrest of dynamic rupture. We previously derived theoretical relations based on fracture mechanics between properties of overstressed nucleation regions (size, shape and overstress level), the ability of dynamic ruptures to either stop spontaneously or run away, and the final size of stopping ruptures. We verified these relations by comparison to 3D dynamic rupture simulations under slip-weakening friction and to laboratory experiments of frictional sliding nucleated by localized stresses. Here, we extend these results to the induced seismicity context by considering the effect of pressure perturbations resulting from fluid injection, evaluated by hydromechanical modeling. We address the following question: given the amplitude and spatial extent of a fluid pressure perturbation, background stress and fracture energy on a fault, does a nucleated rupture stop spontaneously at some distance from the pressure perturbation region or does it grow away until it reaches the limits of the fault? We present fracture mechanics predictions of the rupture arrest length in this context, and compare them to results of 3D dynamic rupture simulations. We also conduct a systematic study of the effect of localized fluid pressure perturbations on faults governed by rate-and-state friction. We investigate whether injection induces seismic or aseismic slip, depending on the timing of the perturbation relative to the earthquake cycle. Our previous slip-weakening and fracture mechanics analysis provides insight on the results of the rate-and-state models. Implications for the maximum magnitude of seismicity induced by fluid injection are discussed.