Numerical simulation of viscous fingering in CO₂ storage: Addressing the limitations of laboratory-derived relative permeabilities

Abstract

Gas injection in CCS is an unstable displacement process due to the viscosity contrast between injected gas and in-situ brine, a fact often overlooked in CO₂ storage modelling. Most studies estimating relative permeability (Relperms) for CCS rely on conventional methods. This paper applies a novel approach that explicitly considers viscous instability from injecting low-viscosity gas into more viscous brine. Based on our earlier studies (A. Beteta et al., 2024; Sorbie et al., 2020). viscous fingering is expected under viscous-dominated conditions. While strong capillary forces may suppress fingering at lab scale, the balance of viscous and capillary forces can shift at field scale, leading to pronounced gas fingering.

To model CO₂ injection, we used published CO₂/water Relperms from CCS experiments. Two approaches were compared: conventional Relperms estimation and an alternative viscous fingering-based method. Simulations using conventional CO₂/water Relperms showed no fingering patterns, whereas our earlier study indicated that core flood experiments with a water–CO₂ viscosity ratio of ∼55 may generate gas fingers. The second approach, based on maximum mobility, produced gas fingers and matched production and differential pressure equally well.

This apparent absence of fingering may suggest limitations in the conventional representation of flow dynamics under such conditions. Possible reasons why experimental relative permeabilities do not capture the expected “fingering behaviour” are discussed in this paper. The main difference between our proposed Relperms and lab derived Relperms is in the total mobility, with our Relperms having a higher total mobility at the shock front gas saturations (S_gf) of CO₂.