A reactive transport model for geochemical mitigation of CO2 leaking into a confined aquifer

Long-term storage of anthropogenic CO2 in the subsurface generally assumes that caprock formations will serve as physical barriers to upward migration of CO2. However, as a precaution and to provide assurances to regulators and the public, monitoring is used detect any unexpected leakage from the storage reservoir. If a leak is found, the ability to rapidly deploy mitigation measures is needed. Here we use the TOUGHREACT code to develop a series of two-dimensional reactive transport simulations of the hydrogeochemical characteristics of a newly formed CO2 leak into an overlying aquifer. Using this model, we consider: (1) geochemical shifts in formation water indicative of a leak; (2) hydrodynamics of pumping wells in the vicinity of a leak; and (3) delivery of a sealant to a leak through an adjacent well bore. Our results demonstrate that characteristic shifts in pH and dissolved inorganic carbon can be detected in the aquifer prior to the breakthrough of supercritical CO2, and could offer a potential means of identifying small and newly formed leaks. Pumping water into the aquifer in the vicinity of the leak provides a hydrodynamic control that can temporarily mitigate the flux rate of CO2 and facilitate delivery of a sealant to the location of the caprock defect. Injection of a fluid-phase sealant through the pumping well is demonstrated by introduction of a silica-bearing alkaline flood, resulting in precipitation of amorphous silica in areas of neutral to acidic pH. Results show that a decrease in permeability of several orders of magnitude can be achieved with a high molar volume sealant, such that CO2 flux rate is decreased. However, individual simulation results are highly contingent upon both the properties of the sealant, the porosity-permeability relationship employed in the model, and the relative flux rates of CO2 and alkaline flood introduced into the aquifer. These conclusions highlight the need for both experimental data and controlled field tests to constrain modelling predictions. (C) 2014 The Authors. Published by Elsevier Ltd.