I use a dynamic two-dimensional plane strain model of an elasto-plastic (Mohr-Coulomb) material with velocity-weakening friction to numerically simulate the emergence and propagation of shear fractures that accumulate slip intermittently during earthquakes. The model considers a 10 × 10 km block of homogeneous crust with a small weak inclusion, subjected to slow simple shear. Results show that deformation is distinctly bimodal, consisting of long periods (lasting ca. 50–200 years) where deformation is dominantly elastic separated by rapid sliding events lasting a few seconds during which time deformation is strongly localised on narrow plastic shear bands (faults). In terms of energy expenditure, the interseismic period is overwhelming dominated by accumulation of elastic strain energy whereas during earthquakes, most of the drop in elastic strain energy is consumed (dissipated) by plastic work. Model earthquakes have maximum slip rates of a few meters per second, involve a few meters of slip and result in drop of mean differential stress of several megapascals. Most earthquakes are complex events involving near-simultaneous rupture of multiple fault segments that propagate at supershear velocities (close to the P-wave speed) while they also induce a significant amount of more diffuse off-fault inelastic deformation. Earthquakes may involve formation of new faults and/or reshear on previously formed faults, which are favourably reactivated owing to their low pressure. Most individual faults establish much of their length early, whereas during subsequent events they continue to accumulate displacement at virtually constant length. As the confining pressure is increased, ruptures are more dynamic, which leads to increasingly widespread inelastic deformation. Overall, the models highlight the complex temporal and spatial interactions that exist between different faults within an array that communicate especially during earthquakes through static and dynamic changes in the stress field.
Read full abstract