Strength of Calcite-Bearing Faults in Fluid Pressure-Controlled Experiments

Abstract:

Fluid pressure P_f  is a major factor controlling natural and induced seismicity. According to the Terzaghi's principle, the effective normal stress σ_eff = σ_n (1-α P_f) decreases with increasing P_f, with σ_n the normal stress and α the Biot coefficient. At α is usually attributed a value of 1, though it may vary with fluid composition and permeability of the fault zone.

Here we conducted 10 experiments on pre-cut (experimental fault) hollow-cylinders (50/30 mm ext/int diameter) of Carrara marble (99.9% calcite) installed in the rotary shear apparatus SHIVA. The hollow-cylinders were inserted in a fluid vessel filled with water (drained conditions) and subjected to σ_n of 15 MPa. After this, the P_f was raised to 5 MPa and the shear stress τ to 6.5 MPa, corresponding to an effective friction coefficient μ_eff = τ / σ_eff = 0.65 or close to the instability of the experimental fault; then P_f in the vessel was increased stepwise of ΔP_f = 0.5 MPa every 10 to 40 s to induce fault instability. The main instability was preceded by short-living slip bursts. Assuming α = 1, the increase in P_f induced the main instability for μ_eff = 1.3 in the case of stepwise intervals of duration ~10 s and for μ_eff = 0.7 in stepwise intervals of ~40 s: the time threshold t_d that resulted in the two different μ_eff (0.7 and 1.3) was ~30 s.

We interpret our experimental observations as due to the permeability decrease of the fault slipping zone which resulted in a smaller α. Wear and comminution during the slip bursts that preceded the main instability produced a low-permeability ultrafine calcite gouge layer in the experimental fault. The gouge layer isolated the fluid in the slipping zone from the fluid in the vessel. As a consequence, the ΔP_f increase imposed on the fluid in the vessel was delayed inside the experimental fault. For each ΔP_f increase, the P_f in the slipping zone could re-equilibrate with the P_f in the vessel provided that the duration of the stepwise increase was long (40 s) with respect to the fluid diffusion time. Given t_d ~30 s and the half width of the ring-shaped experimental fault l = 5 mm, the hydraulic diffusivity κ (= l²/t_d ) was ~ 10^-8m²s^-1, compatible with κ of a low permeability shale. We conclude that fault strength was controlled by slipping zone permeability and by the magnitude and duration of the fluid pressure increments.