A modified HLD-NAC equation of state to predict alkali-surfactant-oil-brine phase behavior

Reservoir crudes often contain acidic components (primarily naphthenic acids), which undergo neutralization to form soaps in the presence of alkali. The generated soaps perform synergistically with injected synthetic surfactants to mobilize waterflood residual oil in what is termed alkali-surfactant-polymer (ASP) flooding. The two main advantages of using alkali in enhanced oil recovery (EOR) are to lower cost by injecting less expensive synthetic surfactant and to reduce adsorption of the surfactant on the mineral surfaces. The addition of alkali, however, complicates the measurement and prediction of the microemulsion phase behavior that forms with acidic crudes. For a robust chemical flood design, a comprehensive understanding of the microemulsion phase behavior in such processes is critical. Chemical flooding simulators currently employ Hand's method to fit a limited amount of measured data, but that approach likely does not adequately predict the phase behavior outside of the range of the measured data. In this paper, we present a novel and practical alternative by employing a modified HLD-NAC equation of state that is predictive in nature. The HLD-NAC model uses the hydrophyllic-lypophillic difference (HLD) concept along with net and average curvature equations (NAC). The modified HLD-NAC model described by Ghosh and Johns (2014) to model surfactant-polymer (SP) phase behavior is extended here for ASP. We use an empirical equation to calculate the acid distribution coefficient from the molecular structure of the soap. Key HLD-NAC parameters like optimum salinities and optimum solubilization ratios are calculated from soap mole fraction weighted equations. The model is tuned to data from phase behavior experiments with real crudes to demonstrate the procedure. We also examine the ability of the new model to predict fish plots and activity charts that show the evolution of the three-phase region. The predictions of the model are in good agreement with measured data and all data is able to be matched.