Streamline Simulation of CO2 Storage in Saline Aquifers

Abstract

AbstractInjecting CO2 in deep saline aquifers is one of many feasible options for mitigating climate change in the medium term. It is important, though, to evaluate the risk involved. Numerical simulations of CO2 injection may be performed to assess the likely migration paths. However, since aquifers may be extensive and highly heterogeneous, the time taken for conventional finite-difference simulations may be prohibitive. In this study, we investigated the application of streamline simulations as a means for speeding up simulations of CO2 injection. The simulations were carried out using a model that was constructed to represent a "generic" heterogeneous deep saline aquifer, using the geometry of the Tay Aquifer near the Forties Field in the North Sea. The main advantage of streamline simulation is its ability to process high-resolution grids efficiently and, in recent years, there has been a widespread increase in the use of streamline simulation in modelling oil reservoirs.The first part of this paper describes a sensitivity study on several streamline-specific parameters to examine their impact on the simulation of CO2 storage in saline aquifers. The differences between streamline and compositional models were then examined. The comparison between the streamline and compositional simulations was carried out for two different injection strategies, designated as follows:Scenario 1 – Injection of supercritical CO2 Scenario 2 – Injection of CO2 saturated brineThe results of the sensitivity analysis on the streamline parameters and the comparisons between the streamline and compositional models were analysed and contrasted. In particular, we found that in Scenario 1, where gravity effects are important, there were significant differences between the streamline and conventional simulation results. The streamline simulation produced a more diffuse plume. However, in Scenario 2, where density of the injected fluid was similar to the resident fluid, the simulation results were similar.