Dynamic estimates of extreme-case CO2 storage capacity for basin-scale heterogeneous systems under geological uncertainty

Abstract

Geological CO2 storage is expected to grow dramatically in the coming decades to meet global climate targets. Assessment of worldwide storage resources using static methods indicates significant theoretical potential for large-scale deployment. Dynamic capacity estimates are needed at the basin-scale that fully capture the impact of geological uncertainty and account for regional limits on pressure buildup. Accurate quantification of the risk of low or critically low capacity under extreme occurrences of heterogeneity will be increasingly important. There are significant challenges associated with efficient computation of low probability capacity within Monte Carlo frameworks at these scales. In this paper, we propose a workflow for uncertainty quantification that is able to efficiently estimate increasingly outer percentiles of dynamic capacity such as P1, P0.1, or even lower probability events. Our approach is based on the rare-event methodology that uses a subset simulation approach to concentrate sampling of the parameter space in the tail regions of the capacity distributions. This approach greatly speeds up uncertainty quantification for very small probabilities compared to standard Monte Carlo. We demonstrate the method by introducing a correlated heterogeneity field to a highly prospective basin-scale system that can support regional injection rates of 100 million tons annually. We find that the outer quantiles are more sensitive to the underlying geostatistical model compared to the median P50 capacity. This implies that for large-scale systems, well characterized heterogeneity is essential to identify the likelihood of very rare yet still relevant dynamic estimates of storage capacity.