The effect of dilatancy and compaction on the stability and permeability of a fluid infiltrated fault

The role of fluids is a central issue in understanding fault slip, earthquake nucleation, and dynamic rupture propagation. We report on experiments in which simulated fault gouge was sheared under controlled pore pressure and true-triaxial stress conditions while permeability evolution was monitored via flow normal to the shear direction. Experiments were conducted in double direct shear within a pressure vessel at room temperature. Samples were jacketed and subjected to either constant pore pressure (Pp¬), or a differential pore pressure inducing flow across the gouge layer. Confining- and pore-pressures were maintained via high precision servo-controlled pressure intensifiers. Experiments involved slide-hold-slide tests and velocity step tests at effective normal stresses ranging from 2 to 30 MPa and Pp ranging from 1 to 4 MPa. Gouge layers were constructed using a precision leveling jig to be 4 mm thick prior to shear. Frictional contact area was 5 cm x 5 cm. Slide-hold-slide tests were conducted for hold times from 1 to 1000 seconds. Measurements of the increase in peak friction, relative to the value for steady sliding, increase logarithmically with time at rates from 0.003 to 0.006 per decade, similar to previous experiments. During hold periods the coefficient of sliding friction decreases logarithmically with increasing hold time and we observe compaction and creep within the gouge material. The decay in friction follows a trend of 0.0017 per decade for saturated fault gouge. Compaction and dilation of the layer is measured independently via pore fluid volume under drained conditions and with a DCDT monitoring layer thickness. These data show that holds expel water and slides imbibe water into the layer. Porosity changes follow the same pattern as friction, exhibiting a log-time decay in the rate of change. The volume of water expelled from the gouge layer also increases logarithmically at low effective normal stresses (<15 MPa), with rates between 0.006 to 0.01 cm3 per decade. Velocity stepping experiments were conducted by instantaneously increasing the load point velocity from 1 μm/s to values ranging from 3 to 100 μm/s, and back to 1 μm/s after steady state sliding was attained. We find that the friction rate parameter (a-b) ranges from positive (0.0003 to 0.004) to approximately velocity neutral. The critical slip distance (Dc) ranges from a few μm to a few 10's of microns. Upsteps in shearing velocity cause dilation as shown by increased water volume in the layer. The volume of water drawn into the gouge layer increases logarithmically with the magnitude of the velocity step at a rate of 0.02 cm3 per decade increase in load point velocity. Measurements of fluid flow normal to the layer during shear show that permeability normal to the layer varied systematically during application of shear load and reached a steady state value when frictional sliding began. When effective normal stress was 10 MPa permeability ranged from 1 to 10 x 10-16 m2, and at 20 MPa the permeability was 4 x 10^{-17 m2. Upon step increases in shear rate, permeability exhibited a transient increase followed by decay back to a new steady state value. We find that steady-state permeability increases by ~ 20-30% for a velocity upstep from 10 to 100 μm/s, at an effective normal stress of 10 MPa, which is consistent with our measurements of pore volume and layer thickness showing increased porosity and layer dilation.