Modeling, Uncertainty Quantification and Sensitivity Analysis of Subsurface Fluid Migration in the Above Zone Monitoring Interval of a Geologic Carbon Storage

Abstract:

Carbon dioxide (CO₂) storage into geological formations has significant potential for mitigating anthropogenic CO₂ emissions. An increasing emphasis on the commercialization and implementation of this approach to store CO₂ has led to the investigation of the physical processes involved and to the development of system-wide mathematical models for the evaluation of potential geologic storage sites and the risk associated with them. The sub-system components under investigation include the storage reservoir, caprock seals, and the above zone monitoring interval, or AZMI, to name a few. Diffusive leakage of CO₂ through the caprock seal to overlying formations may occur due to its intrinsic permeability and/or the presence of natural/induced fractures. This results in a potential risk to environmental receptors such as underground sources of drinking water. In some instances, leaking CO₂ also has the potential to reach the ground surface and result in atmospheric impacts. In this work, fluid (i.e., CO₂ and brine) flow above the caprock, in the region designated as the AZMI, is modeled for a leakage event of a typical geologic storage system with different possible boundary scenarios. An analytical and approximate solution for radial migration of fluids in the AZMI with continuous inflow of fluids from the reservoir through the caprock has been developed. In its present form, the AZMI model predicts the spatial changes in pressure – gas saturations over time in a layer immediately above the caprock. The modeling is performed for a benchmark case and the data-driven approach of arbitrary Polynomial Chaos (aPC) Expansion is used to quantify the uncertainty of the model outputs based on the uncertainty of model input parameters such as porosity, permeability, formation thickness, and residual brine saturation. The recently developed aPC approach performs stochastic model reduction and approximates the models by a polynomial-based response surface. Finally, a global sensitivity analysis was performed with Sobol indices based on the aPC technique to determine the relative importance of these input parameters on the model output space.