Reduced order models of transient CO2 and brine leakage along abandoned wellbores from geologic carbon sequestration reservoirs

We have developed reduced order models (ROMs) for CO2 and brine leakage rates along wellbores including abandoned wells at geologic CO2 storage sites using a Multivariate Adaptive Regression Splines (MARS) algorithm. The ROMs were developed for use within systems level performance assessment models such as Los Alamos National Laboratory's CO2-PENS model. The ROMs are used to compute leakage rates as a function of wellbore properties including effective permeability, depth as well as pressures and saturations in the reservoir where the wellbore intercepts the reservoir. The ROMs were created using results of complex, 3-D multi-phase numerical simulations of large-scale CO2 injection at a generic CO2 storage site with an abandoned wellbore. The generic site included not only the primary storage reservoir but also a groundwater aquifer and an intermediate permeable zone. Two sets of simulations were performed, one with and one without an abandoned wellbore in order to capture the effect of coupling between the storage reservoir and wellbore in a system level model where it is assumed that they are decoupled. Cross-validation against the complex, multi-phase numerical simulation results were used to evaluate the ability of the ROMs to reproduce numerical simulation results. Further, our ROM development approach effectively captures transient CO2 and brine leakage during and after CO2 injection as well as the effects of an intermediate permeable zone on leakage to a shallow groundwater aquifer and to the atmosphere. Ultimately, the ROM is a computationally efficient model that effectively captures many of the complex underlying processes taking place during CO2 and brine leakage along a wellbore at a geologic CO2 storage site.