Earthquake nucleation, rupture and slip on rough laboratory faults

Abstract

Fault structural heterogeneity is commonly linked to the nucleation of earthquakes, and once propagating it is used to explain the frictional weakening which drives rupture. However, the current laboratory and modelling perspective of earthquake nucleation processes derives largely from investigations of homogenous materials which neglect many features of natural earthquake faults. This thesis addresses the role of geometric heterogeneity in the nucleation and propagation of earthquake rupture by means of triaxial experiments utilising laboratory simulated faults.
A range of roughness’s were applied to rock samples which were deformed at conditions spanning earthquake hypocentral conditions (30<  ≤200 MPa) to investigate the role of geometric heterogeneity (roughness) in earthquake nucleation. The stability and frictional behaviour of rough faults was mapped, revealing a complex interplay between roughness and normal stress conditions. Smooth faults are more unstable at low stress, whilst rough faults are general stable at low stress. Results are found to be in violation of current theories of frictional sliding, with normal stress acting to stabilise slip, and nucleation of rupture occurring in a rate-strengthening regime. A new microphysical model is thus developed, which matches observations, based on the interaction of flaws created by roughness and fracture energy considerations. Significantly this model has definable physical origins which are lacking from current theories of frictional instability.
Elastodynamics of fault slip during nanoearthquake propagation on rough faults is investigated. New experimental techniques are applied to obtain the coupled slip velocity and strength evolution in the nearfield of spontaneous earthquake ruptures. Results are well fit by analytical flash weakening models of high velocity frictional strength. Fracture energy scaling results and self-similarity of individual events supports the application of flash heating theory to explain the weakening expected during small earthquakes (M<5). These results are used to suggest that self-similar self-healing slip pulse models are the most appropriate model for use in seismological inversion, crucial for determining earthquake source parameters.
The role of roughness and normal stress on the frictional sliding is revisited using limestone. Experiments are performed on roughened bare surfaces, which were subsequently deformed in a direct shear configuration at a range of normal stress conditions (30<  ≤100 MPa). In all experiments sliding is stable, with the influence of roughness being less pronounced due to the rapid wear of experimental surfaces. Results show that frictional sliding can either be strongly velocity strengthening and accompanied by plastic deformation processes, or velocity neutral associated to brittle deformation processes depending on initial fault conditions. Result show the first experimental evidence linking the evolution of rate-and-state parameters to frictional wear widely observed on natural faults. It is therefore suggested that wear is a determining factor for the mechanical behaviour of natural faults.
In conclusion results show that the widely observed roughness of natural faults has important implications for the mechanics of faulting and earthquakes, in particular the frictional stability and microstructural evolution. These factors should be taken into greater consideration in future experimental and modelling studies.