Simulation of carbon dioxide injection at the FutureGen2.0 site: Class VI permit model and local sensitivity analysis

Numerical simulation was conducted to evaluate supercritical CO2 (scCO2) storage in the Mt. Simon sandstone at the FutureGen 2.0 site in Morgan County, Illinois. The simulations illustrate the recent modeling effort conducted to successfully obtain a Class VI permit under the Underground Injection Control Program of the U.S. Environmental Protection Agency. The scCO2 was injected into a highly stratified reservoir, distributed among four lateral wells. Results show that each of the multiple reservoir layers with low permeability traps a fraction of the scCO2 beneath it so that the injected scCO2 does not reach the top of the reservoir by the end of the simulation period. Hence, all of the injected scCO2 is considered to be safely trapped by the stratified structure within the reservoir. A local sensitivity analysis (LSA) was conducted for injectivity and plume size at the end of the 20 year injection period. The LSA indicates that the imposed initial conditions and the hydraulic properties of the injection layer contribute the most to the sensitivity. Overall, the simulation outputs are very sensitive to only a small fraction of the inputs. However, the parameters that are important for controlling CO2 injectivity are not the same as those controlling the plume area. The three most sensitive inputs for injectivity are the horizontal permeability and the residual aqueous saturation of the injection layer, and the initial fracture-pressure gradient. For the plume area, the most sensitive inputs are the horizontal permeability of the injection layer and the residual aqueous saturation of the two adjacent layers above the injection layer. The advantages of requiring just a single set of simulation results, scalability to the proper parameter errors, and straightforward calculation of the composite sensitivities make the proposed analysis attractive for guiding site characterization, injection well design, estimating Area of Review uncertainty and, and monitoring network design.