Dewpoint Pressure Model for Gas Condensate Reservoirs Based on Genetic Programming

Abstract

Successful prediction of the future performance of condensate reservoirs requires accurate values of dewpoint pressures. Although the dewpoint pressure can be measured experimentally from collected laboratory samples, these measurements are frequently not available. In these cases, dewpoint pressure is determined using empirical correlations or using an equation of state (EoS). This paper presents an application of genetic programming with the orthogonal least squares algorithm (GP-OLS) to generate a linear-in-parameters dewpoint pressure model represented by tree structures. The GP-OLS-based gas condensate reservoir dewpoint pressure model was generated as a function of reservoir fluid composition (in terms of mole fractions of methane through heptanes-plus, nitrogen, carbon dioxide, and hydrogen sulfide), molecular weight of the heptanes-plus fraction, and reservoir temperature. The new model was developed using experimental measurements of 245 gas condensate systems covering a wide range of gas properties and reservoir temperatures. A total of 135 gas condensate systems that had not been used in building the new model were used to test and validate the new model against the other early published correlations. The validity test shows that the new model has a lower average absolute relative error than other published correlations. Therefore, the new model can be considered an alternative method to estimate the dewpoint pressure when the experimental data are not available.