RESULTS

Permeability

Permeability measurements were performed on sample 190-1174B-42R-1, 93–95 cm (vertical), under increasing isotropic pressure from effective pressures of 0.5–1.5 MPa over a period of 10 days (Fig. F2A). This sample was maintained at 0.5 MPa effective pressure for 8 days and absorbed 1262 mm3 water without significant total volume change, showing that the sample was not fully saturated at the beginning of the experiment. We calculated an initial water saturation of 86%. Measurements made on Sample 190-1174B-42R-1, 93–95 cm (vertical), allowed a comparison of steady-state flow and pressure decay methods. The two measurements performed using the flow method (1.2 x 10–18 m2 at an effective pressure of 0.6 MPa and 4.1 x 10–19 m2 at an effective pressure of 1.41 MPa) are at the upper limit of the range of measurements done using the pulse decay method (3.6–4.5 x 10–19 m2 at 0.5 MPa effective pressure and 1.6–3.7 x 10–19 m2 at 1.5 MPa effective pressure) but are compatible. The 1.2 ± 0.2 x 10–18 m2 permeability value at 0.6 MPa effective pressure we obtained using the flow method is close to the 2.3 x 10–18 m2 permeability measured using the same method on Sample 190-1174B-42R-3, 133–150 cm (538 meters below seafloor [mbsf]), at the same effective pressure (Gamage and Screaton, this volume).

Sample 190-1173A-55X-5, 135–150 cm (vertical), was fractured during a failure test (experiment 1) performed at a low confining effective stress (0.2 MPa). This fracturation induced a single order of magnitude permeability increase from 1–2 x 10–19 m2 to ~1.4 x 10–18 m2 (Fig. F2B, exp. 1). The sample was removed from the triaxial press to observe the fracture (Fig. F3). This same sample (called "previously fractured" in Table T2) was used 2 days later for new permeability measurements under increasing isotropic conditions from 0.2 to 2.4 MPa effective pressure for 6 days (experiment 2). The evolution of confining and pore pressures and the permeability measurements are shown in Figure F4. Permeability decreased by nearly one order of magnitude from ~10–18 to ~10–19 m2 from effective pressures of 0.5 to 1.5 MPa (Fig. F2B, exp. 2). Then the permeability remained nearly constant (~1.8 x 10–19 m2) with increasing effective pressure.

Sample 190-1173A-55X-5, 135–150 cm (horizontal), was loaded along an isotropic path to a high confining effective stress (2.5 MPa). Permeability was measured first under isotropic conditions and then under deviatoric stress before and after failure. Failure occurred at 5.0 MPa of effective axial stress and 1.5% of deformation. The deformation rate was 4.2 x 10–9 s–1 during the hydrostatic stage (14 days) and 1.9–3.2 x 10–8 s–1 during the deviatoric stage (16 days). Permeability decreased by a factor of two (from ~8 to ~4 x 10–19 m2) during isotropic loading as effective pressure increased from 1 to 2 MPa and then stayed roughly constant. Permeability decreased during failure and then recovered as the sample was deformed beyond the failure point. Photographs showing the fractures were taken after the experiments (Fig. F5).

All our permeability measurements performed in the triaxial cell are lower than measurements performed at a low effective stress (Taylor and Fisher, 1993; Gamage and Screaton, this volume), which range from 10–16 to 10–18 m2 for samples from ~600 mbsf. However, they are in the same range as values measured at <1–5 MPa effective confining stress (Byrne et al., 1993), ranging 1.3 x 10–20 to 1.3 x 10–18 m2. Byrne et al. (1993) also observed transient permeability variations with strain.

Overall results indicate permeability decreases with effective confining pressure as high as 1.5 MPa (Fig. F2A). This permeability reduction occurs with cumulative void decreases of 0.015–0.054 depending on the sample (Table T2). However, measurements at low effective pressure are too dispersed to yield a precise general relationship between pressure and permeability. When the effective pressure is increased as high as 2.5 Mpa, permeability is roughly constant (1–4 x 10–19 m2) but the void ratio continues to decrease, reaching a final cumulative void ratio decrease of 0.08–0.10. Our results also suggest that fracturing does not affect permeability when the effective stress is more than ~1.5 MPa. They also show the enhancement of permeability by fracturing when effective stress is low, followed by the permeability decrease with the increase in effective stress, interpreted as fracture closure.

Bulk Compressibility Measurements

The bulk compressibilities of the samples were estimated using the volume of pore fluid expelled from the sample (V) in response to an increase of the effective confining pressure in isotropic stress conditions and before the yield point. The compressibility is calculated as

= 1/V x (V/P),

where V = the volume of the sample. Results show a value of 1.1 ± 0.2 x 10–8 Pa–1 for Sample 190-1174B-42R-1, 93–95 cm; 1.5 ± 0.2 x 10–8 MPa–1 for Sample 190-1173A-55X-5, 135–150 cm, in the vertical direction; and 1.4 ± 0.5 x 10–8 MPa–1 for the same sample in the horizontal direction. These compressibility values correspond to the elastic deformation of the sample. Compressibilities determined on the unloading curve were in the same range, confirming that the compressibilities were measured in the elastic domain of the samples.

These estimations can be compared with values deduced from the stiffness modulus derived from consolidation tests realized after Leg 131 (Moran et al., 1993). The compressibility is the inverse of the stiffness modulus. Before the yield point, Sample 131-808C-23R-3 (514 mbsf) showed a stiffness modulus of 50–250 MPa, corresponding to a compressibility of 4.0 x 10–9 to 2.0 x 10–8 Pa–1. These different estimations remain in the same range.

Friction Coefficients

Friction coefficients were calculated for vertically and horizontally oriented Samples 190-1173A-55X-5, 135–150 cm (horizontal), and the same sample in the vertical direction. These samples were subjected to drained axial compression in the triaxial cell and developed slickenlined faultlike fractures during failure (Figs. F3, F5). The shear and normal stresses resolved on a rupture plane inclined at an angle of from the vertical are as follows (Jaeger, 1959):

= [(13)/2] x sin(2), and
n = {[(1 + 3)/2] – [(13)/2]} x cos(2),

where

1 = effective axial stress (axial stress – pore pressure) and
3 = effective confining pressure.

As axial strain increases, axial stress reaches a peak stress corresponding to the failure of the sample. Then the stress reaches a plateau lower than the peak stress, corresponding to an axial strain increase at constant stresses. This plateau is interpreted as steady-state sliding on the previously formed fracture. The friction coefficient, µ, is calculated at this plateau (Byerlee, 1978):

µ = /n.

Three friction coefficients were calculated on two samples (Table T3). Sample 190-1173A-55X-5 (horizontal) presented a friction coefficient of 0.40. The previously fractured Sample 190-1173A-55X-5 (vertical) presented a strain-stress curve with a plateau at 4.9 MPa axial effective stress, interrupted by a 45% drop in axial stress, followed by a new axial stress buildup to a second plateau at 5.2 MPa. The sudden drop in axial stress could be the result of oversliding (unstable sliding) on the fracture. The friction coefficients calculated at these two plateaus are 0.37 and 0.40. These friction coefficients are consistent with measurements performed (usually at much higher effective pressure on the order of 100 MPa) on pure Ca smectite gouge during velocity-stepping direct shear experiments (Saffer et al., 2001) or on a gouge of a saturated mixture of ~60% quartz and ~40% montmorillonite (Logan and Ranenzahn, 1987) during steady-state sliding and on saturated illite (Morrow et al., 1992). Kopf and Brown (2003) measured friction coefficients of ~0.14 for purified smectite and ~0.25 for illite under a range of effective pressures of 1–30 MPa. They estimated, based on ring shear experiments and mineralogical composition, the friction coefficient on Nankai décollement to be 0.30–0.32. These values, lower than our measurements, were performed using elevated displacement. The values of Kopf and Brown (2003) may be representative of well-developed faults like the décollement, whereas our measured friction coefficients may be representative of fractures with shorter sliding, and therefore closer to initial failure conditions.

NEXT