Reactive Transport Modeling of Effects of Different (Physical, Chemical and Computational) Factors on the Convection Process during CO2 Geological Sequestration in Deep Saline Aquifers

Tuesday, 16 December 2014
Wei Zhang and Shuling Wang, National Geological Library of China, Beijing, China
Density-driven convective activity can significantly accelerate the transformation of the injected CO2 from supercritical or gas phase into liquid phase, thus improving the long-term storage security. It is well known that the convective instability is only caused by the heterogeneity of porous medium. However, we found that numerical perturbation resulted from the spatial and temporal discretization and the convergence criteria will affect the modeling results. Partial simulation results as follows: (1) the increase in vertical permeability (kv) has a strong effect on the convection process compared to the increasing horizontal permeability. The higher kv also can accelerate the downward migration of fingers and the CO2 dissolution; (2) increase in magnitude of medium (porosity and permeability) perturbation promotes the onset of convective activity, but does not affect the evolution of convection. However, for same magnitudes of the medium perturbation, the modification in permeability can lead to a more rapid onset of convective activity than that in porosity; and (3) the geochemical reactions and mineral compositions are important not only to the CO2 dissolution, but also to the dissolution of other species from mineral dissolution, which both can increase the water density. In additional, 3D and 2D modeling results indicated that their effect on the convection onset is similar. The CO2 dissolution rate for the higher-resolution 3D model is greatly slower than that for the higher-resolution 2D model, due to the increase in dimension for 3D model enhances the interaction of fingers. However, we found that once the bottom boundary becomes important, the decrease in CO2 dissolution rate for the 2D model is more significant than that for the 3D model. This is because the more significant interaction of fingers in the 3D model than that in the 2D model delays the downward propagation of fingers and then the impact of the bottom boundary.