Predicting CO2 Plume Evolution during Geologic Storage with Ensemble Kalman Filter

A major risk associated with geologic storage of CO2 is leakage from various pathways such as abandoned wells, faults and compromised seal layers. Extensive efforts must go into site characterization and selection to avoid regions with high leakage potential as well as post-injection monitoring activities to evaluate and predict the evolution of the injected CO2 plume for risk mitigating purposes. Monitoring systems also provide important data that can be used to infer the hydraulic properties of the storage aquifer. Here, we investigate the feasibility of estimating aquifer heterogeneity from monitoring measurements during CO2 injection with the Ensemble Kalman Filter (EnKF). We examine the performance of the EnKF in identifying heterogeneous aquifer permeability distributions, predicting the CO2 plume evolution and possible leakage from an abandoned well, and quantification of the associated uncertainties in each case. We consider a high-resolution three-dimensional aquifer model with extreme heterogeneity in both horizontal and vertical hydraulic conductivity. The extreme heterogeneity consists of preferential flow path and barriers that constrain the upward and lateral displacement of the CO2 plume in the aquifer, making it difficult to predict the movement of the CO2 plume. Using this model, we investigate the contribution of various monitoring data types, including pressure, saturation and time-lapse seismic data to estimation of aquifer hydraulic conductivity. The estimation results suggest that with a proper design of the filter, the EnKF can provide promising estimates of the aquifer hydraulic properties and long-term displacement and fate of the CO2 plume, which is critical for risk assessment and mitigation.