Estimation of vapor–liquid equilibrium ratios of crude oil components: A comparative study

Abstract

The present paper deals with the proposal of a new correlation for vapor–liquid equilibrium ratio of Iranian crude oil components, using multivariable regression techniques. The database for this study was collected from different Iranian reservoir oil fields extracted from differential liberation tests. They were measured at temperature range of 150–292°F and pressures up to 4992psia. Compared to the most published empirical correlations, the number of coefficients used in the new correlation was decreased from 58 for the Almehaideb correlation to 21. The second objective of this work was to estimate the equilibrium ratio by artificial neural network models. The absolute average relative error for the whole database was estimated 17.93% for artificial neural network, 33.98% for the new correlation, 28.98% for Almehaideb correlation, and 69.73% for Whitson & Torp correlation. Furthermore, the accuracy of the models for calculating the bubble points of ten samples compared with experimental values. The results shows the absolute average relative error of the artificial neural network to predict the saturation pressures was 5.38% compared to 14.50% for the new correlation and 8.01% for the Peng–Robinson equation of state without tuning. The results clearly depicts that over a range of experimental condition, the artificial neural network predictions indicate better agreement with experimental data than classic thermodynamic models.