SVM modeling of the constant volume depletion (CVD) behavior of gas condensate reservoirs

Abstract

Knowledge of the phase behavior of condensate gas systems is important for predicting reservoir performance and future processing needs. In this study, new improved models based on support vector machine (SVM) algorithm are developed to calculate the amount of produced gas based on constant volume depletion (CVD) analysis of the well stream effluent at any depleted state in retrograde gas condensate systems. These methods are based on compositional analysis of more than 1150 compositions of condensate gases collected worldwide. The developed models can predict the cumulative produced gas (Gp) as a function of the gas composition, characteristics of plus fraction, temperature, and pressure. The prediction results exhibit good agreements when compared to the experimental values. Finally, a sensitivity analysis was performed to assess the impact of each correlating variable on the amount of produced gas. The developed models can be incorporated in petroleum engineering PVT simulation softwares for improved CVD modeling of gas condensate reservoirs.