The sealing and healing behaviour of faults filled with anhydrite gouge, by processes such as pressure solution, is of interest in relation both to the integrity of faults cutting geological storage systems sealed by anhydrite caprocks and to seismic events that may nucleate in anhydrite-bearing sequences, such as those present in the seismogenic zone beneath the Apennines. We have developed a detailed series of kinetic models for pressure solution in anhydrite fault gouge, allowing for dissolution, diffusion and precipitation control, to estimate the time scale on which such sealing and healing effects occur. We compare the models obtained with previously reported experimental data on compaction creep rates in simulated anhydrite fault gouge, tested under wet, upper crustal conditions. The results confirm earlier indications that compaction under these conditions likely occurs by diffusion-controlled pressure solution. Applying our most rigorous model for diffusion-controlled pressure solution, constrained by the fit to the experimental data, we infer that anhydrite fault sealing will occur in a few decades at most, which is rapid compared with both CO2 storage time scales and with the recurrence interval for seismicity in the Apennines.

No abstract

The frictional behavior of anhydrite-bearing faults is of interest to (a) the safety and effectiveness of CO 2 storage in anhydrite-capped reservoirs, (b) seismicity induced by hydrocarbon production, and (c) natural seismicity nucleated in evaporite formations. We performed direct shear experiments on simulated anhydrite fault gouges, at a range of temperatures (80-150°C) and sliding velocities (0.2-10 μm s À1 ), under a fixed effective normal stress of 25 MPa. Four types of experiments were conducted (1) dry experiments, (2) experiments pressurized with water, (3) dry experiments pressurized with CO 2 , and (4) wet experiments pressurized with CO 2 . Fluid pressures of 15 MPa were used when applied. Over the temperature range investigated water-saturated samples were found to be up to 15% frictionally weaker than dry equivalents. Wet samples containing CO 2 were also up to 15% weaker than CO 2 -free equivalents. Dry sample strength without CO 2 was independent of temperature, whereas wet samples without CO 2 strengthened 10% from 80 to 150°C. Samples containing CO 2 weakened by 4% (dry) and 10% (wet) from 80 to 150°C. Under the P-T conditions investigated, only dry anhydrite fault gouge showed velocity-weakening behavior above 120°C, required for faults to potentially generate earthquakes. Assuming natural fault gouges are wet in situ, seismicity is unlikely to nucleate in anhydrite-rich faults, though the presence of dolomite or (reaction-produced) calcite may change seismic potential. CO 2 penetration into wet anhydrite-rich faults may trigger slip on critically stressed faults due to the observed short-term CO 2 weakening effects (excluding possible formation of secondary minerals), but is not expected to influence slip stability.

Faults that crosscut subsurface CO 2 storage systems offer potential leakage pathways, especially if fault reactivation and dilation occur. After reactivation, however, newly formed fault gouge is expected to gradually compact and seal as a function of time. To estimate the time scale on which this occurs, the processes that control compaction must be understood. We performed uniaxial compaction experiments on simulated anhydrite fault gouge to investigate the deformation mechanisms that operate under postslip conditions in faulted anhydrite caprocks. This involved constant stress (5-12 MPa) and stress stepping experiments (5/7.5/10 MPa) performed at 80°C, under dry and wet conditions, on fault gouge samples prepared from crushed natural anhydrite sieved into different grain size fractions in the range 20-500 μm. Dry samples showed little to no compaction creep, whereas wet samples (i.e., flooded with presaturated CaSO 4 solution) showed rapid compaction. Our mechanical data and microstructural observations on wet samples suggest that for fine grain sizes (<50 μm) and low stresses, gouge compaction is controlled by diffusion-controlled pressure solution. With increasing grain size and stress, fluid-assisted subcritical microcracking becomes dominant. Pressurizing solution-flooded samples with CO 2 (15 MPa) led to no significant effect on compaction rates in fine-grained material, but it decreased compaction rates in coarse samples. Since fine grain sizes are expected in reactivated faults, we infer that pressure solution will dominate in anhydrite (cap)rocks, with extrapolation of our lab data to reservoir conditions suggesting sealing time scales of a few decades.

No abstract