Surfactant-based enhanced oil recovery methods have been a promising technique for the last several decades due to the surfactant's ability to mobilize previously trapped oil by significantly reducing capillary forces at the pore-scale. However, the field-implementation of these techniques have been challenged by the high cost of chemicals, which makes the margin of error for the deployment of such methods increasingly narrow. Some commonly recognized issues are surfactant adsorption, surfactant partitioning to the excess phases, thermal and physical degradation, and scale-representative phase behavior. Recent contributions to the petroleum engineering literature have used the hydrophilic-lipophilic difference net-average curvature (HLD-NAC) model to develop a phase behavior EoS to fit experimental data and predict phase behavior away from tuned data. The model currently assumes spherical micelles, which may yield errors in the bicontinuous region where micelles transition into cylindrical and planar shapes. In this paper, we introduce a new empirical phase behavior model based on chemical potentials and HLD that eliminates NAC so that spherical micelles are no longer assumed. The model is able to describe physical two-phase regions, and is shown to represent accurately experimental data at fixed composition and changing HLD (e.g. A salinity scan) as well as compositional data at fixed HLD. Further, the model is extended to account for surfactant partitioning into the excess phases. The model is benchmarked against experimental data, showing excellent fits for a wide variety of experiments, and is compared to the HLDNAC EoS model for reference.