Efficient Model for Estimation of Dew Point Pressure in Gas Condensate Systems

Gas condensate reservoirs are a special class of gas reservoirs with temperatures which plot between the critical point temperature Tc and the cricondentherm Tcric on a Pressure–Temperature (P-T) phase diagram. The reservoirs’ produced streams are characterized by relatively, low C1 : C2 +ratio, implying compositions with significant fractions of high molecular weight (C2 +) hydrocarbons, called Natural Gas Liquids (NGLs). NGLs exist as gas at reservoir conditions of high pressures and temperatures but as liquids, called condensates (or distillates) at separator conditions, havinge higher market price than separator gas fractions. Condensation of NGLs can occur within reservoirs if reservoir pressures fall to the reservoir’s Dew Point Pressure (DPP) during production. Condensates formed within reservoirs are never produced owing to their less than critical saturation. Therefore, accurate estimation of DPP and ensuring production occurs above such estimates, is necessary for ensuring optimum productivity and profit. Correlations have proven valuable in providing quick DPP estimates, however, the several correlating parameters of common types, make them computationally intense, yet their associated errors remain immense. This study aims at presenting a simple correlation, with fewer correlating parameters that would guarantee better accuracy when compared to other popular models. A new correlation for DPP as a function of gas compositional analysis data and reservoir temperature was developed using method of multiple regression analysis. A total of 1,568 gas condensate datasets from 984 reservoirs obtained from literature with unrestricted geographic locations were used to develop the model. These data were divided in the ratio of 3:1:1 for training, cross validation and testing. The results obtained showed that the new correlation significantly outperforms other existing popular industry models and provides predictions with higher accuracy. This was verified in terms of highest correlation coefficient, lowest average absolute percentage error and lowest root mean square error