PVT Properties Prediction Using Artificial Neural Network

Phase behavior of hydrocarbons is a very complicated behavior that hydrocarbons show at the time of phase change or when they are in a particular phase. Without a good understanding of this behavior process design is almost impossible. Artificial neural network has been utilized very well for engineering applications during the last two decades. In this article one model are presented for prediction of Gas Solubility of Crude Oil using the Artificial Neural Networks. For this purpose five-layer neural networks has been designed and trained using 106 lab data. After the training step, nine experimental data were used for the model evaluation step and as a reliability check. The output of models for both the training and predicted data are compared with the empirical equations of Standing, Glaso and Marhoun. It has been concluded that Artificial Neural Network Approach has an excellent capability for this purpose compared to the conventional methods.