Pore pressure change during nucleation and slip along experimental faults

Abstract

In order to clarify the main controlling factors influencing fluid pressure changes in fault zones during the seismic cycle, we conducted laboratory rock friction experiments where fluid pressure was monitored in situ during sequences of quasi-static loading followed by dynamic slip events. The simulated fault was a 30$^\circ$ saw-cut in a Westerly granite cylinder, saturated with water, tested under triaxial conditions. Pore pressure was held constant at the boundaries of the block, but the low hydraulic diffusivity of Westerly granite made the fault hydraulically disconnected from the boundaries. During quasi-static loading while the fault was locked, we observed pore pressure increases which we interpret as poroelastic closure of the fault. During dynamic slip events, pore pressure systematically dropped by amplitudes commensurate to the normal stress drop. A large contribution to the pore pressure drop is interpreted as poroelastic opening of the fault. Deviations from the poroelastic effects are observed: in small events, pore pressure dropped further than anticipated, indicating inelastic dilation. In a few large events, pore pressure dropped less than anticipated, which could be the sign of compaction or thermal pressurisation. Prior to macroscopic slip events, we detect systematic pore pressure decreases by up to around 1 MPa, correlated to the occurrence of inhomogeneous preslip along the fault. Slip nucleation, inferred by kinematic inversion of local strain gauge data, is linked to local slip magnitudes of the order of 1 to 10 $\mu$m, and appears to lead to inelastic dilation. A stability analysis of fault slip including dilatant and poroelastic effects shows that poroelastic coupling tends to compensate normal stress variations, leading to faults operating under mostly constant effective normal stress if conditions are undrained.