Numerical study of the effects of permeability heterogeneity on density-driven convective mixing during CO2 dissolution storage

Permanence and security of carbon dioxide (CO2) in geologic formations requires dissolution of CO2 into brine, which slightly increases the brine density. Previous studies have shown that this small increase in brine density induces convective currents, which greatly enhances the mixing efficiency and thus CO2 storage capacity and rate in the brine. Density-driven convection, in turn, is known to be largely dominated by permeability heterogeneity. This study explores the relationship between the process of density-driven convection and the permeability heterogeneity of an aquifer during CO2 dissolution storage, using high-resolution numerical simulations. While the porosity is kept constant, the heterogeneity of the aquifer is introduced through a spatially varying permeability field, characterized by the Dykstra-Parsons coefficient and the correlation length. Depending on the concentration profile of dissolved CO2, we classify the convective finger patterns as dispersive, preferential, and unbiased fingering. Our results indicate that the transition between unbiased and both preferential and dispersive fingering is mainly governed by the Dykstra-Parsons coefficient, whereas the transition between preferential and dispersive fingering is controlled by the permeability correlation length. Furthermore, we find that the CO2 dissolution flux at the top boundary will reach a time-independent steady state. Although this flux strongly correlates with permeability distribution, it generally increases with the permeability heterogeneity when the correlation length is less than the system size.