Modeling the Geochemical Impact of CO2 Leakage on Groundwater from an Aquifer with Variable Mineral Compositions

Developing a low-cost, easily implementable monitoring strategy for carbon storage reservoir leak detection is an important topic for geologic CO2 storage. Geochemistry-based methods for tracking aqueous chemistryincluding pH, species concentrations, and total dissolved solids (TDS)can provide early detection of well or reservoir integrity issues. In this study, reaction path modeling was performed to monitor the fluid chemistry changes as a response to trace, low, and high ranges of dissolved CO2 concentrations in the Santa Rosa aquifer, west Texas. This aquifer was chosen for study because it overlies an oil reservoir where CO2 was injected for enhanced oil recovery (EOR). To cover a wide range of sandstone compositions (i.e., beyond the major mineral components of quartz and kaolinite), small amounts of accessory minerals carbonate (calcite), feldspar (albite and anorthite), mica (annite and phlogopite) and chlorite (ripidolite) were turned on and off to form 18 different mineral composition scenarios. Each mineral composition was first set to equilibrium with the Santa Rosa groundwater, using average fluid data reported previously. Different concentration ranges of dissolved CO2 were then added to the equilibrated reservoir system. Mineral precipitation and fluid chemistry changes were modeled based on thermodynamic equilibrium for each sandstone scenario. For all the mineral composition scenarios, the pH and most aqueous species responded significantly to trace (<0.005 mol/kg) and low (<0.065 mol/kg) CO2(aq) concentrations and became more stable with higher (0.065-0.605 mol/kg) CO2(aq) concentrations. The presence of calcite in the sandstone helped buffer the pH impact of CO2(aq) addition to the system. For many sandstone scenarios, dolomite and calcite appeared as secondary mineral precipitates when CO2(aq) concentrations were trace and/or low. However, dolomite and calcite were no longer predicted to precipitate as CO2(aq) increased. The final mineral carbonate product, if there was any, was usually siderite. This modeling work is part of a larger simulation framework for CO2 leakage signal monitoring. The results of this geochemical model are being used to develop a Bayesian belief network that determines the probability of CO2 leakage based on groundwater chemistry measurements.