Quantifying uncertainties of CO2-oil relative permeability and capillary pressure curves through automatic history matching core flood data

Relative permeability and capillary pressure curves are essential data for predicting multiphase flow in porous media. Measuring these properties in the lab is very challenging and involves various issues. For example, core preparation, establishing initial water saturation, and determining residual oil saturation after a gas flood experiment need special attention and precise measuring devices. Further complications are expected during a CO2 injection, due to the combined effects of molecular diffusion, swelling, and vaporization processes alongside the displacement process. In this situation, determining relative permeability curves, which is a pure displacement concept, needs a more comprehensive inversion routine that can discriminate the non-displacement effects from the displacement recovery mechanism. In this paper, through multiple simulation case studies, it is demonstrated that overlooking the compositional effect results in underestimating residual oil saturation. Then, a new numerical inversion routine is introduced by coupling a compositional core flood model and a nonlinear regression routine that enables estimating relative permeability and capillary pressure curves simultaneously. This integration enables the generation of rather ensemble curves for identifying major uncertainties and quantifying the respective variabilities. Finally, it is demonstrated that initial water saturation and residual oil saturation can compensate for each other whenever they are searched by a regression routine simultaneously. On the other hand, constraining initial water saturation would result in more confident residual oil saturation accordingly. The results of this research can provide a basis for future CO2 injection studies dealing with more complex multiphase fluid flow in porous media.