A K-value method, which is designed for fast computation, is presented. The method relates the K-value of each component with its boiling temperature, critical temperature and pressure, the mixture's pressure and convergence pressure, and overall compositional changes. Experimental data, from routine PVT tests of the mixture, are used to calibrate the K-value correlation.
A PVT simulator was developed to model the routine PVT laboratory tests. This simulator generates the K-value correlation for the mixture and employs a least squares-linear programming optimization routine to fine tune the correlation such that the simulated PVT behavior matches the actual one.
Nine reservoir fluid samples, four of them included carbon dioxide miscibility, were used to demonstrate and test the applicability of the new K-value method. The method has proven successful for predicting the phase behavior of hydrocarbon systems as well as the carbon dioxide-hydrocarbon mixtures.
The new method was compared with Peng-Robinson, Soave-Redlich-Kwong, and Schmidt-Wenzel equations of state which were embodied as options in the PVT model. The comparisons of the fluid samples have shown that the new method is reliable and is faster than the equations of state by a factor of seven to twenty. The method requires no iterations, less computer memory, less input data, and fewer tuning parameters than equation of state methods.