Real-time assessment of CO2 migration direction during geologic storage

Abstract

Abstract Minimizing the cost of large-scale geologic storage of CO 2 is a paramount concern, and consequently many aquifer storage projects may be implemented without a detailed characterization of the target formation. On the other hand, CO 2 migration beyond the volume designed for effective trapping is a paramount risk. Thus, inexpensive methods of monitoring the plume movement will be valuable for operators and regulators alike. Unanticipated heterogeneities within the target formation, whether high-permeability channels or low permeability barriers, are one of the most likely causes of migration beyond the design volume. We propose that routine measurements of injection rate and injection pressure in each well can be used to infer the existence of heterogeneities large enough to affect the plume path. We do not seek from these measurements a detailed spatial distribution of permeability in the formation, but merely an indication of features that affect the overall migration path. The advantage of this approach is that these measurements will be acquired routinely, frequently and cheaply in all projects, whereas methods yielding higher resolution (time-lapse seismic surveys, electromagnetic surveys, cross-well seismic, monitoring wells, etc.) are specialized and expensive. We have implemented this idea by combining (i) our previously developed research software (Pro-HMS) which carries out geologically consistent parameter estimation from injection and production data and (ii) a commercial compositional simulator (GEM from CMG) as a forward model which has been tuned to the full physics and phase behavior of the CO 2 /brine/rock system. In this paper we test the approach on model aquifers that exhibit permeability heterogeneity prescribed by a spatial correlation model. The permeability estimation process is performed within a fully probabilistic framework. We include the noise typical of pressure/rate data from real wells and find that signal of large heterogeneities can still be discerned.