Experimental and Numerical Model Studies of Frictional Instability Seismic Sources

Stick-slip frictional instability is widely regarded as a viable mechanism for crustal earthquakes, particularly because of the way that it can be incorporated into the notion of earthquakes as episodic unstable slip events along preexisting zones or planes of weakness represented by faults in the Earth. In this thesis, detailed laboratory observations of stick-slip events generated on a simulated fault provide a basis for extending a constitutive description of frictional sliding developed from quasistatic experiments to describe fault response under dynamic slip conditions. An appropriate fault constitutive relation is implemented into a numerical model of dynamic crack propagation. Test calculations reveal that a fully specified rateand fault state-dependent friction constitutive model is capable of producing effects which are consistent with earlier predictions of dynamic fault rupture using simpler slip weakening fault models, but differences from the slip weakening calculations arise because of some of the features of the state variable friction model. A possible means for extending the modeling to the scale of crustal earthquakes by assessing suitable length scales based on a fractal geometry approach is presented. Stick-slip shear failures have been generated on simulated faults of two different roughnesses and at different levels of applied normal stress a, between 0.6 and 4 MPa. The unique, large test sample size ( 2 m fault length ) used in these experiments provides the opportunity to investigate the shear instability over length scales which are not allowed by typical laboratory rock mechanics experiments, and the stick-slip event is clearly viewed as a crack-like failure which nucleates in one part of the fault and subsequently propagates over the entire fault surface. Rupture velocities are of the same order as the seismic wave speeds in the granite test sample. In addition, high-speed records of local shear stress and fault slip reveal details of the initial breakdown of fault frictional resistance or strength at the onset of stick slip resembling slip weakening. A critical displacement dQ associated with this slip weakening is identified and it is observed to be relatively insensitive to normal stress acting on the fault. This displacement parameter is larger for the rougher of the two prepared fault surfaces. Stick-slip stress drops and fracture energies increase with increasing normal stress, and fracture energies also increase for increased fault roughness through the dependence of critical displacement on roughness. The existence of a critical weakening displacement implies that the rupture characteristics of the stick-slip events are scale-dependent, based either on a critical crack length or, alternatively, a crack endzone size. Both the critical crack and crack endzone increase with increasing roughness and decrease with increasing normal stress. Rupture and slip velocities from both the smooth and the rough fault surfaces, reflect this scale-dependence such that when a rupture propagation distances are scaled by critical crack length, the data indicate that rupture velocity and fault