Experimental Studies of Dynamic Fault Weakening Due to Thermal Pore-Fluid Pressurization

Abstract:

Thermal pressurization is a co-seismic weakening mechanism driven by the thermal expansion of native pore fluids, which leads to elevated pore pressures and significant co-seismic weakening. While thermal pressurization has been studied theoretically for many decades, and has been invoked in recent earthquake simulations, its activation in laboratory experiments has remained elusive. Several high-speed friction experiments yield indirect evidence for thermal pressurization, yet none have directly linked with existing theoretical models or the relevant physical parameters -- such as permeability, slip, and slip rate – that control the weakening rate. 

We are conducting thermal pressurization experiments on fluid-saturated, low-permeability rocks (primarily Fredrick diabase; also SAFOD gouge) at slip rates up to ~5 mm/s, with constant confining pressures in the range 21–149 MPa and initial pore pressures in the range 10–25 MPa. The impractically low permeability of the diabase, ~10^-23 m², is increased prior to the friction test by thermally cracking the samples, yielding measured permeabilities in the range 1.3*10^-18 to 6.1*10^-19 m². These permeabilites are high enough to allow sample saturation over one to several days, but also low enough to confine pore pressure rises during rapid sliding and allow thermal pressurization to occur. In recent experiments we also embed a thermocouple ~1–2 mm from the sliding surface, and use the resulting data to calibrate a finite element model that calculates the sliding surface temperature. 

One experiment revealed a rapid decay of shear stress by ~25% following a step-change in velocity from 10 μm/s to 4.8 mm/s. For the first 28 mm of slip the experimentally measured shear stress agrees closely with the theoretical solution for slip on a plane (Rice [2006]) with an inferred slip weakening distance of ~500 mm, which is in the range predicted by inserting laboratory determined rock and fluid properties into the formula for L^* from Rice [2006]. Deviations from the theoretical prediction occur at larger displacements since the experimental sample is not a semi-infinite half space, as assumed in the theoretical model, and heat is lost to the high-conductivity steel of the sample assembly. To our knowledge this is the best experimental validation of thermal pressurization to date.