Injection of CO2 into deep saline aquifers leads to a flow system of two fluid phases—injected CO2 and resident brine—that requires modeling tools to simulate migration of both fluids. While high-fidelity full three-dimensional (3D) multiphase flow models are available, their application for practical analysis remains challenging due to the high computational costs and the significant uncertainty of subsurface geological parameters. 

We present three alternative computationally efficient approaches: reduced-order, multiscale, and hybrid models. The reduced-order model assumes vertical pressure equilibrium (due to strong buoyancy of CO2) and a macroscopic sharp interface between CO2 and brine, which simplify the 3D two-phase flow problem into a 1D nonlinear advection-diffusion equation. Such simplifications allow us to solve the system analytically and perform detailed analysis of fluid flow behaviors and CO2 storage efficiency, as well as leakage risk assessment. Further, we present a multiscale model that relaxes the vertical equilibrium assumption while maintaining computational efficiency. Algorithms of this type fit naturally into a multiscale framework, and are able to produce similar results compared to full 3D models for both homogeneous and layered heterogeneous aquifers. Finally, we introduce a hybrid approach that couples the multiscale model with a full 3D model so that we can apply the 3D model locally to handle 3D heterogeneity while using the multiscale model in the (majority) rest of the domain. By combining the simplified and high-fidelity 3D models, the hybrid approach achieves both computational efficiency and accuracy. Overall, these new modeling approaches allow fast simulation, risk assessment, and uncertainty quantification for geological CO2 storage and other fluid injection applications in the subsurface.