Abstract
We present a multifidelity Multilevel Monte Carlo (MLMC) method for efficient uncertainty quantification in CO $$_2$$ 2 storage applications. By combining full-physics reservoir models with reduced-order vertical equilibrium (VE) models, the approach significantly reduces computational cost while maintaining accuracy. VE models, which assume vertical equilibrium and model only lateral flow, enable rapid simulation of complex, layered reservoirs. We demonstrate the method on realistic case studies, including the Johansen and Sleipner sites, showing that MLMC with VE can significantly reduce simulation time compared to standard Monte Carlo methods. This makes large-scale ensemble simulations feasible for applications such as storage capacity estimation and improved uncertainty estimates to be used in value-of-information (VOI) monitoring design.