CO2 modeling in a deep saline aquifer: a predictive uncertainty analysis using design of experiment.

When field data are limited, stratigraphic models are used instead of detailed, fully heterogeneous models (FHM) to represent deep saline aquifers in numerical simulations of CO2 storage. This study evaluates parameter sensitivity and prediction uncertainty of three stratigraphic models of decreasing complexity (i.e., facies, layered, formation) against that of a FHM. For select simulation outcomes (i.e., CO2 mass profiles, gas plume shape, brine leakage), parameter sensitivity and associated prediction uncertainty are compared among the models, with the FHM serving as a reference. The analysis is conducted using the computationally efficient design of experiment (DoE) and response surface (RS) methodology. Results suggest that when a competent caprock exists (permeability <1 × 10(-4) mD), the facies and layered models are capable of capturing the most important sensitivity parameters of the FHM, that is, residual gas saturation, heterogeneity variance, and salinity. Using the important parameters identified by DoE, RS modeling then suggests that the same two models also capture the ranges of predictions in mobile gas, trapped gas, and brine leakage. The formation model is less accurate in capturing the sensitivity and prediction ranges of the FHM, although it is accurate in predicting brine leakage into the overlying formation.