Reservoir simulation is a valuable tool for assessing the potential success of enhanced recovery processes. Current chemical flooding reservoir simulators, however, use Hand's model to describe surfactant-oil-brine systems even though Hand's model is not predictive, and can fit only a limited data set. Hand's model requires the tuning of multiple empirical parameters using experimental data that usually consist of salinity scans at constant reservoir temperature and atmospheric pressure. Given experimental data supporting the change in microemulsion phase behavior with key formulation properties (e.g. temperature, pressure, salinity, EACN, and overall composition), there is a need for an improved model that can capture changes in these relevant parameters at the reservoir scale. The recent EOS proposed for microemulsion phase behavior (Ghosh and Johns 2014, 2016), which is based partially on the hydrophyllic-lypophyllic difference and net average curvature model (HLD-NAC, Acosta et al. 2003), has been supported by numerous experimental data and provides a more mechanistic phase behavior model than the Hand's model. In this paper, the EOS model with the extension to two-phase regions is incorporated for the first time into the chemical flooding simulators, UTCHEM, and our new in-house simulator PennSim. Hand's model is only used for comparison purposes, and is no longer needed even for flash calculations in the type II- and type II+ regions. The results show excellent agreement between UTCHEM and PennSim both in composition space and for composition/saturation profiles. Further, the HLD-NAC based EOS model and Hand's models are fitted to the same experimental data and the results of these simulations are nearly identical when variations of salinity, pressure and temperature are small. For large gradients, the results of the physics-based EOS deviates from Hand's model, and shows it is critical to incorporate these gradients in recovery predictions at large scale.