Frictional and sealing behavior of simulated anhydrite fault gouge : Effects of CO2 and implications for fault stability and caprock integrity

To limit climate change, humanity needs to limit atmospheric CO2 concentrations, hence reduce CO2 emissions. An attractive option to do this involves capture at industrial sources followed by storage in depleted oil and gas reservoirs. In such reservoir systems, faults cutting the topseal are considered one of the most likely leakage pathways, especially in case of fault reactivation. Anhydrite is a common caprock in many hydrocarbon fields, globally and in the Netherlands. Faults crosscutting anhydrite caprock contain anhydrite-rich, damage or frictional-wear material, termed “fault gouge”. To assess the likelihood of leakage and/or fault reactivation, it is important to have a detailed understanding of the frictional and healing/sealing behavior of fault gouge, including any effects of CO2. Here, I present the results of experiments on simulated anhydrite fault gouge aimed at understanding 1) its compaction creep behavior and potential for self-sealing, and 2) its frictional behavior and hence the potential for fault reactivation and seismicity . This understanding is also key to evaluating seismic risk in tectonically active, anhydrite-bearing terrains, such as the Italian Apennines. All experiments were performed under pressure -temperature conditions representative for CO2 storage conditions and upper crustal tectonics (80-150°C; effective stresses £ 25 MPa), under dry and wet conditions, with and without CO2. A limited number of experiments were also performed on anhydrite-dolomite gouges. The results obtained show that compaction creep of wet anhydrite fault gouge is strongly dependent on grain size and stress and involves stress-driven dissolution and precipitation (pressure solution). Extrapolation of the results using kinetic models for this process indicates that compaction-sealing of wet, newly-formed anhydrite fault gouge should occur within decades. Hence, if faults are reactivated during CO2 injection, self-sealing should be rapid. Friction coefficients obtained fall for anhydrite lay between 0.5 and 0.7, with the lowest values applying for samples containing water and CO2. A roughly 15% strength reduction is seen in wet samples upon CO2-exposure, which should be considered when determining allowable pressures and rates for CO2 injection at storage sites. The velocity-dependence of frictional strength indicated self-stabilizing (velocity strengthening) behavior and hence little seismogenic potential for wet anhydrite, dolomite or mixed gouges under CO2 storage conditions, with or without CO2. Results obtained from slide-hold-slide friction experiments on anhydrite showed significant static restrengthening of vacuum-dry samples and samples containing dry CO2. These effects were strongly enhanced in wet samples, without or with CO2, even in those samples that contain only traces of water. Extrapolation of the results to the conditions pertaining at seismogenic depths in the Apennines implies that post-seismic fault restrengthening will take place within tens of days, so that the much longer recurrence times of destructive (M≥5) earthquakes in the Apennines are likely controlled by tectonic loading rate