The spatial-temporal influence of grouped variables on pressure plume behavior at a geologic carbon storage project

Abstract

We explore the magnitude and lateral extent of the pressure plume produced by carbon dioxide injection over a range of six subsurface variables using an idealized layer cake reservoir model with open-flow boundaries. We group variables by inspection into a storage term S and a flow resistance term H. We explore the spatial-temporal distribution of the first-order sensitivity and Pearson correlation coefficients of S and H over 50 years of injection and 50 years post-injection. A global sensitivity analysis shows that H strongly influences pressure evolution at all radii throughout the entire 100-year project. For low values of H, the post-injection pressure plume is typified by rapidly falling pressures close to the wellbore and low amplitude pressure buildup further away. For large values of H, pressure buildup is larger and the rate of pressure equilibration during the post-injection period is slower. Using 200,000 realizations, a spatial-temporal risk map is developed for brine leakage based on Latin Hypercube Sampling of grouped variables and the permeability of hypothetical legacy wellbores placed throughout the area of pressure perturbation. Risk is concentrated spatially close to wellbores and decreases rapidly once injection ceases. The two grouped variables can be used to make quantitative predictions of reservoir behavior during injection and during the post-injection period.