Assessing Fault Leakage Potential during Underground CO2 Storage in the Gulf of Mexico

With near 1 Gt/year CO₂ emissions from point sources (Ambrose et al., Environ. Geol., 2009), extensive oil-industry infrastructure, and prospective Miocene geology (Treviño and Meckel, eds., 2017), the Gulf of Mexico coast is an attractive target for pioneering large-scale underground CO₂ storage in the US. The occurrence of induced seismicity and associated potential CO₂ leakage has been identified as one of the most pressing questions surrounding the deployment of geologic CO₂ storage at the gigatonne scale (Zoback and Gorelick, PNAS, 2012; Juanes et al., PNAS, 2012). Here, we begin to develop quantitative models that address this question in realistic geologic settings. To this end, we developed a 2D geological model representative of the geology near Offshore Texas State Waters (OTSW) from public data. Then, we simulated CO₂ injection in prospective reservoirs to study fluid flow and pressure buildup, focusing on changes across and along faults.

Our computational model comprises a 45x8 km domain, meshed using unstructured triangular elements (Fig. 1a). We conducted two-phase flow simulations of the brine-CO₂ fluid system using MRST (Krogstad et al., SPE Reserv. Simul. Symp, 2015; Lie, MRST book, 2016). A fully-implicit solution strategy, tracking phase saturations, relative permeabilities and capillary pressures was employed; the automatic differentiation class within MRST was used to avoid analytical derivation and coding of the global Jacobian matrix.

We evaluated fluid flow and pressurization within the main fault and sealing layers as a result of different fluid and rock compressibilities, and injection rates. Preliminary results using a single-phase, compressible flow model suggest that fluid compressibility is the main factor driving pore pressure increase within the main fault, and that along-permeability dominates over across-permeability, with the model being less sensitive to porosity changes (Fig. 1b, c). As part of our next steps, we will couple the flow simulations with a geomechanics simulator to assess fault reactivation, and account for more realistic, dynamic hydraulic properties of faults.