IODP Proceedings    Volume contents     Search

doi:10.2204/iodp.proc.314315316.219.2012

Experimental methods

We conducted experiments using a biaxial testing apparatus with servo-hydraulic control under true-triaxial stress conditions and controlled pore pressure (Fig. F3). Samples were sheared as 5.4 cm × 5.7 cm layers with thicknesses ranging from ~1 to 4 mm under load. Samples were jacketed, subjected to confining pressure, and saturated with 3.5 wt% NaCl brine as pore fluid. Effective normal stress was maintained at 25 MPa and includes the combined effects of confining pressure (Pc), externally applied normal load, and pore pressure (Pp). This value of effective stress was chosen for two main reasons: (1) it simulates effective stress conditions near the updip limit of the seismogenic zone (Moore and Saffer, 2001) and (2) it facilitates comparison with our previous experimental work using simulated clay-rich fault gouge (Ikari et al., 2009). The specimens were sheared between steel forcing blocks in the geometry shown in Figure F3. During shear, Pc was held constant at 6 MPa, pore pressure at the upstream end of the sample (Ppa) was held constant at 5 MPa, and the downstream pore pressure (Ppb) was set to a no-flow (undrained) condition in order to monitor pore pressure in the layer (e.g., Ikari et al., 2009).

In each experiment, shear was implemented as a displacement rate boundary condition (11 µm/s) at the gouge layer boundary. The resulting shear stress (τ) was measured and the coefficient of sliding friction (µ) was calculated by (Handin, 1969):

τ = µσn′ + c,

(1)

where

  • c = cohesion (assumed to be negligible) and
  • σn′ = effective normal stress computed using the average of the pore pressures at the drained and undrained boundaries.

After attainment of steady-state shear stress (typically at a shear strain of ~5), the velocity was increased step-wise in half-order of magnitude increments within the range 0.03–100 µm/s in order to measure friction constitutive parameters, the results of which are reported by Ikari and Saffer (2011). This was followed by a slide-hold-slide sequence performed to measure the frictional healing parameters reported here. Records of shear stress during shearing are shown in Figures F4 and F5. During each hold, the vertical ram was held stationary for prescribed hold times of 1, 3, 10, 30, 100, 300, or 1000 s, between which the sample was sheared for a boundary displacement of 500 µm at 10 µm/s. Upon reinitiating shear, friction reaches a peak value (µpeak) then declines to its prehold steady-state value (µss) (Fig. F6). We measured the change in friction (Δµ) after each hold:

Δµ = µpeak – µss.

(2)

The slope of Δµ versus the logarithm of hold time is the healing rate, here denoted as β (per decade, or tenfold time increase) (Marone, 1998). Here, we report β for hold times of 10–1000 s to avoid biasing due to nonlinearity at very low hold times. We were also able to make quasistatic compaction measurements (Δh) during the hold period (Fig. F6). We report compaction during hold periods as normal strain (ε):

ε = Δh/h, (3)

where h is the instantaneous layer thickness immediately prior to a hold. Compaction rates were calculated using the slope of ε versus the logarithm of hold time, also over the range 10–1000 s.

For samples from Site C0004, after shearing we cycled the applied normal stress in ~6 MPa steps, in order to independently measure sample compressibility (C) (MPa–1) (Fig. F7). From each step we measured the normal strain induced by a change in effective normal stress, which is the Young’s modulus (E) (MPa) (Jaeger et al., 2007):

E = Δσn′⁄ε. (4)

The compressibility is the inverse of the bulk modulus (K) (MPa), which is related to the Young’s modulus by Poisson’s ratio (v) (Gercek, 2007; Jaeger et al., 2007):

1/C = K = E/[3(1 – 2v)]. (5)

We did not explicitly measure K in these experiments; however, we assume a value of v = ~0.33, which is appropriate for claystones (Gercek, 2007). In this case, K = E and the compressibility may be calculated as the inverse of the Young’s modulus. In addition, we note that for a thin layer of frictional material, small perturbations in layer thickness occur at constant nominal contact area and thus changes in layer thickness are directly related to changes in volume. Under this assumption, K = E. For each sample, we fit multiple stress steps and the origin to compute the compressibility values.