Identification of a representative dataset for long-term monitoring at the Weyburn CO2-injection enhanced oil recovery site, Saskatchewan, Canada

Abstract

Leakage at geological carbon storage (GCS) sites, driven by increased system pressure and higher CO2 saturations, represents a key risk to secure containment of injected CO2. For long-term GCS monitoring, it is critical to determine a level of information needed to minimize leakage risks while keeping costs under control. This study demonstrates a goal-oriented, retrospective design concept called minimum data set requirement (MDR) for Weyburn Midale Project (WMP), a commercial-scale, CO2-injection enhanced oil recovery (EOR) site in Canada. More than a decade of research at the WMP site has led to an extensive collection of site characterization data, a situation that is unlikely to be true for many other GCS projects around the world. By screening existing data retrospectively, our MDR identification process seeks to establish a level of data needed to define a sufficient reservoir model for guiding post-EOR monitoring, under user-defined performance metrics. Our starting point is an existing history-matched WMP reservoir model and three datasets consisting of logs from hundreds of wells and a seismic survey. An iterative approach is taken to systematically and gradually reduce the level of information used in parameterizing a geological model, from which conditional stochastic realizations of model properties are generated and simplified reservoir models are developed. Results show that the minimum dataset for predicting CO2 migration is based on the heterogeneity and anisotropy of selected parameters of the field. For WMP, about 80% of the 403 wells can be eliminated without having a detrimental impact on the simulated pressure field.