An Equation-of-State Model To Predict Surfactant/Oil/Brine-Phase Behavior

Abstract

Summary Surfactant/polymer (SP) floods have significant potential to recover waterflood residual oil in shallow oil reservoirs. A thorough understanding of surfactant/oil/brine-phase behavior is critical to design SP-flood processes. Current practices involve repetitive laboratory experiments of dead crude at atmospheric pressure in a salinity scan that aims at finding an “optimum formulation” of chemicals for targeted oil reservoirs. Although considerable progress has been made in developing surfactants and polymers that increase the potential of a chemical enhanced-oil-recovery (EOR) project, very little progress has been made to predict phase behavior as a function of formulation variables such as pressure, temperature, and oil equivalent alkane carbon number (EACN). The empirical Hand (1930) plot is still used today to model the microemulsion-phase behavior with little predictive capability because these and other formulation variables change. Such models could lead to incorrect recovery predictions and improper SP-flood designs. In this research, we develop a new predictive-phase-behavior model and introduce a new factor β to account for pressure changes in the HLD equation. This new HLD equation is coupled with the net-average-curvature (NAC) model to predict phase volumes, solubilization ratios, and microemulsion-phase transitions (Winsor II–, Winsor III, and Winsor II+). The predictions of key parameters are compared with experimental data and are within relative errors of 4% (average 2.35%) for measured optimum salinities and 17% (average 10.55%) for optimum solubilization ratios. This paper is the first to use the HLD/NAC model to predict microemulsion-phase behavior for live crudes, including optimal solubilization ratio and the salinity width of the three-phase Winsor III region at different temperatures and pressures. Although the effect of pressure variations on microemulsion-phase behavior is generally thought to be small compared with temperature-induced changes, we show here that this is not necessarily the case. The predictive approach relies on tuning the model to limited experimental data (such as at atmospheric pressure) similar to what is performed for equation-of-state (EOS) modeling of miscible gasfloods. This new EOS-like model could significantly aid the design of chemical floods where key variables change dynamically, and in screening of potential candidate reservoirs for chemical EOR.