Modeling the Mechanics of Dynamic Triggering of Earthquakes in Granular Fault Gouge

On faults where the static stress state is near but below the failure criterion, dynamic stress perturbations from passing seismic waves may initiate earthquakes. To study how these transient stresses can cause failure, we perform Discrete Element Method numerical simulations of shear failure in a layer of non-cohesive granular material. The granular material mimics crushed gouge in the fault core - the weakest part of the fault and the region most likely to initiate slip. The applied shear stress is slowly increased until failure (a slip event) to determine the frictional strength of the granular layer under non-perturbed conditions. Pre-failure states of the gouge layer are saved during the loading process and used as initial conditions in triggering experiments: one boundary of the gouge is subjected to a transient pulse in boundary stresses, to simulate a passing seismic waves. Various amplitudes and frequencies of the stress pulse are tested on layers at different static stress levels from 0.95 to 0.99 of the failure strength. In order to trigger an immediate failure, the pulse must increase the normal stress ratio to reach the previously measured frictional strength of the layer. However, we find that this is not a sufficient condition for immediate failure. The stress state must be above the frictional strength threshold long enough to allow for significant grain shifting (~ 0.1 grain diameters). This introduces a frequency dependence in addition to the amplitude dependence of the pulse. For low pre-stress levels, high frequency pulses do not remain above the threshold long enough and are incapable of causing immediate triggering. This frequency cutoff is directly proportional to the stress level, meaning lower frequencies can cause immediate triggering at a wider range of initial stress levels. Systems that are not immediately triggered may still experience "delayed triggering" - slip induced by the pulse after a time delay. This implies that transient boundary stresses can cause small permanent changes in the network of granular forces, which changes the ultimate failure strength of the layer. These experiments will help us better understand the conditions needed for seismic waves to trigger nearby earthquakes, and any possible changes such perturbations have on the stress state for future events.