Integrated Thermo-Hydro-Mechanical Workflow for Modeling CO2 Storage with Fault Activation Risk

Summary Carbon capture and storage is crucial in combating global climate change by effectively mitigating carbon emissions. Deep saline aquifers possess the largest identified storage capacity, making them the preferred storage location. However, when CO2 is injected underground, it tends to escape through interconnected fractures or reactivated faults toward the ground due to buoyancy. Thus, studying the impact of CO2 injection on faults and the feedback effect of fault activation on CO2 leakage is significant. In this work, we develop and validate a fully coupled thermo-hydro-mechanical (THM) model to simulate mutual interactions between CO2 injection, fault movement, and CO2 plume propagation. We further analyze the elastoplastic failure behavior of faults, and our simulation results show that fault activation leads to a dramatic change in permeability properties, which are highly heterogeneous and exhibit a strong dependence on plastic damage. In addition, the hydraulic behaviors of reservoirs are compared under closed (no-flow) and more generally open (hydrostatic pressure) boundaries. In a decades-long industrial-scale CO2 continuous injection scenario, the reservoir pressure disturbance under closed boundaries reaches 39%, whereas the pressure disturbance under open boundaries is less than 1%. This stark contrast underscores the paramount importance of boundary conditions and highlights the necessity to delve into whether effective fluid trapping zones can be formed in the vicinity of faults. Finally, we focus on optimizing decoupling strategies for tackling this multiphase, large-scale, and nonlinear THM problem. We compare six decoupling sequences exhaustively, and the test results reveal that the sequence in which the fields are solved significantly impacts the stability and convergence rate of the numerical model. Specifically, the sequence of first seepage, then temperature, and finally displacement fields yields the best stability and convergence and also maintains precision, offering a valuable reference for tackling such THM challenges efficiently.