Quantification of CO2 trapping and storage capacity in the subsurface: Uncertainty due to solubility models

The purpose of this chapter is to examine how different solubility algorithms provide different estimates of storage capacity assessments and lead to different assessments of CO 2 trapping mechanisms. Secure storage of carbon in deep saline aquifers requires CO 2 ―brine―rock reactions to convert injected CO 2 into dissolved species and solid carbonate minerals. Effective characterization of these reactions on the capacity and security of storage requires accurate representations of CO 2 solubility in brine. Several widely used solubility models and the geochemical reaction simulator Geochemist's Workbench© (GWB) were compared. These models incorporate various fugacity coefficients, interacting parameters, and corrections for nonideal behavior of the mixtures (H 2 O―CO 2 ―salt). The solubility models of Duan and Sun [2003] and Spycher and Pruess [2005] agree well with experimental data both in pure water and in saline solutions. The model of Enick and Klara [ 1990] also produces results in agreement with experimental data if the fugacity coefficient is calculated based on Duan and Sun [2003]. The radius of formation necessary to store 3.3 x 10 11 kg of CO 2 (equivalent to 30 years of CO 2 emissions from a 1000-MW coal-fired power plant) for the 60-m thick Rose Run Sandstone ranges from 6 to 28 km, depending on the solubility model used. Predictions of silicate mineral dissolution and the precipitation of CO 2 trapping carbonate minerals also depend considerably on the choice of solubility model. The choice of solubility model has tremendous impact on sequestration evaluations, especially: predictions of the volume of a formation required for specific amounts of CO 2 , assessments of hydrodynamic, mineral, and solubility trapping mechanisms, and forecasts of density-driven flow patterns. Complementary to this study, the next chapter in this volume explores how simulations of flow and transport processes are impacted by choice of solubility model and other equation-of-state components.