Evaluation of three-phase relative permeability models for WAG injection using water-wet and mixed-wet core flood experiments

Three-phase flow occurs during many important processes in oil reservoirs including tertiary gas and water-alternating-gas (WAG) injection. Accurate estimation of three-phase relative permeability (kr) is required to describe and predict the behaviour of these multi-phase flow conditions. The current approach in the oil industry for simulation of WAG experiments is to use two-phase relative permeability data to generate three-phase kr values by using correlations (e.g. Stone1, Baker, etc) available in commercial reservoir simulators.
In this paper, we evaluate the performances of the existing three-phase relative permeability models against experimental data. Two sets of WAG experiments were carried out by performing core flood experiments on a water-wet and a mixed wet core using low IFT oil/gas fluids. The results show that choosing inappropriate three-phase kr model in simulation of the WAG experiments can lead to large errors in prediction of fluids production and pressure. The predictions made by the existing models can be very variable and various models can predict vastly different 3-phase kr values from the same 2-phase data. While some models perform better than others, all of the 3-phase kr models examined in this study fail to predict the continued production of oil after the breakthrough of the gas which is one of the features of gas and WAG injection experiments at low gas-oil IFT (interfacial tension). The resulted three phase relative permeability obtained from WAG experiments revealed strong cyclic hysteresis for oil, water and gas relative permeabilities in both water-wet and mixed-wet systems.
Our results highlight some serious shortcomings of the existing hysteresis models. Comparison of the kr hysteresis observed for water-wet and mixed-wet WAG experiments demonstrates significant differences between the results obtained for the two wettability conditions which demonstrates the impact of wettability on three-phase flow during cyclic injection of water and gas and the importance of performing these experiments under representative wettability of the reservoir.
The results clearly demonstrate the need for improved models and methodologies for obtaining three-phase kr and also kr hysteresis especially for WAG injection.