Parametric sensitivity analysis for CO2 geosequestration

CO2 sequestration (CCS) in geological formations appears to be a viable technology for large-scale storage of CO2 to mitigate the impacts of climate change. Simulations of the behaviour of sequestration systems using mathematical models play a significant role in risk estimation and consequently in decision-making processes. Uncertainty arises in the application of models because of physical complexities, simplifying assumptions, and parameter variability. A sensitivity analysis comparing the influence of different model parameters on predicted CO2 plume evolution uncertainty is presented. Both the role of the parameter in the model and the parameter uncertainty are included in the measure of sensitivity to distinguish between parameters with equal influence in the model output but having different degrees of intrinsic uncertainty. The sensitivity of the plume interface location, the maximum breakthrough distance of CO2, and the moment of inertia of the CO2 plume with respect to intrinsic physical system parameters and parameters introduced in constitutive relationships is investigated in a hypothetical site. parameters are used, each with a probability density obtained from measurements of the Nisku Aquifer, targeted for CO2 injection in Alberta, Canada. Our sensitivity analysis shows that formation porosity, residual brine saturation, and entry capillary pressure are the most influential parameters in the uncertainty of plume evolution. Thus, the influence of constitutive relationship parameters in plume evolution uncertainty is as high as that of the physical characteristics of the system. Simulation outputs where insensitive to CO2 viscosity and the exponents in the Brooks-Corey model for capillary pressure and relative permeability; therefore, it is possible to reduce the dimensionality of future uncertainty and risk analyses, based on the model presented, by neglecting the uncertainty of these parameters and assuming them to be deterministic. While results are specific to the synthetic problem presented in the article, the methodology is general and applicable to other CO2 geosequestration sites.