Impact of pore-scale three-phase flow for arbitrary wettability on reservoir-scale oil recovery

Three-phase flow in porous media is of great importance in enhanced oil recovery processes such as Water-Alternating-Gas (WAG) injection. In this work we use a new three-phase flow pore-network model, which fully captures the three-phase microscopic displacement processes, including accurate descriptions of oil films and layers, and which takes realistic 3D pore geometries extracted from pore-space images as an input. The pore-scale model is used to calculate physically robust relative permeability and capillary pressure functions for a range of realistic wettability scenarios. These three-phase flow functions have been used in a heterogeneous reservoir model with a range of equiprobable geological realisations of permeability and porosity distributions to demonstrate their impact on oil recovery after water, gas or WAG injection.Simulations at the pore-scale show that the residual oil saturations for a water-wet system are higher after water flooding compared to the corresponding residual oil saturations after gas injection. Residual oil saturations for oil-wet systems are significantly lower after both water and gas injection. This is due to the presence of stable oil wetting films, which provide hydraulic connectivity at low oil saturations.Simulations at the reservoir scale show that the lowest oil recovery for a water-wet case is obtained when a single fluid (gas or water) is injected; WAG injection yields significantly higher recovery factors. If the reservoir is oil-wet, however, pure gas injection results in the highest oil recovery. Overall, field-scale recovery predictions are very sensitive to the combination of both, different geological models and different three-phase relative permeability functions. Comparing recovery factors for network-derived and empirical relative permeability models demonstrates that the uncertainty in predicting oil recovery resulting from the choice of three-phase relative permeability models can be as large as the geological uncertainty present in a model that has been history matched using production data from a prolonged waterflood. This implies that the choice of three-phase flow functions directly affects the viability of WAG projects.