Phase behavior studies have been performed using three component, six Phase behavior studies have been performed using three component, six component and twenty-two component mixtures. These studies have been performed in such a way that the critical points for the mixtures are performed in such a way that the critical points for the mixtures are approached, either by increasing the temperature while keeping the composition constant or by changing the composition while keeping the temperature constant. It has been shown that regression on the data from these types of experiments gives a better Equation of State (EOS) match, and thus will allow a better simulation of slim tube displacements.
The phase behavior has been calculated using the Peng-Robinson EOS and a comparison between the calculated and experimental results is shown. A comparison of the quality of the phase match, especially the pressure temperature behavior, is made as a function of the number of regression variables. In all cases substantial improvement in the EOS's ability to simulate PT-x data was seen as the number of regression variables increased. As a result of the fact that a system with more components has more regression variables available, the twenty-two component system had the best match and the three component system the poorest match. The conclusion was that the two parameter EOS as it is normally used lacks flexibility to calculate phase behavior in the entire phase space. The phase matches were used to predict additional data at different phase matches were used to predict additional data at different temperatures and compositions to evaluate the ability of the EOS to extrapolate the phase match. The result was that the EOS did surprisingly well in some cases and very poor in other cases.
For many years the petroleum engineering, reservoir simulation community has been trying to use tables, equations, or correlations to provide vapor-liquid equilibria results for compositional reservoir models. Initial models used tables of K-values that were a function of pressure alone. These models did not allow the K-values to vary with composition. Thus, their applicability to reservoirs as the pressure declined and the composition changed was limited. Later models used various correlations, such as the convergence pressure correlation, to give K-values as a function of temperature, pressure and composition. This technique required the phase matching of PVT data to adjust the parameters of the correlation so it could calculate K-values at any point in the phase space. This phase matching process was time consuming and not straight forward nor did it necessarily produce a unique match. The quality of a phase match was often a function of the experience and persistence of the engineer performing the calculation. This technique did allow the K-values to change with composition, but the result could not be trusted to extrapolate phase behavior beyond the composition and pressure range of the initial data. This empirical correlation could not accurately predict the critical point nor the critical tie line if the phase matching process involved only PVT data far from the critical point. Thus it was unable to accurately simulate the multiple contact miscibility (MCM) process.
A few years ago Compositional Reservoir Simulators were introduced using Equations of State (EOS).