Application of Intelligent Models for Prediction of Solution Gas Oil Ratio

Accurate calculation of PVT properties is a basic requirement for petroleum engineering computations like reservoir simulation, material balance, and well-test. Experimental tests of PVT are time-consuming and costly. Therefore, prediction models for PVT properties such as bubble point pressure, dew point pressure and solution gas oil ratio have been developed using regression models. In this study, new intelligent models for solution gas oil ratio were developed using more than 1100 experimental data series. Two robust intelligent tools, namely adaptive neuro-fuzzy inference system (ANFIS) and artificial neural network were used for development of the models. Precise comparison of the developed models with other published correlation showed that the developed models are superior to all other correlations. Comparison of the ANFIS model and ANN model showed that, ANFIS model is more accurate than ANN model and is the best model for calculation of solution gas oil ratio

Keywords:

solution gas oil ratio , crude oil , adaptive neuro-fuzzy inference system , artificial neural network