Predicting the PVT properties of Iran crude oil by Neural Network

Reservoir fluid properties are very important in material balance calculations, well testing, and reserve estimates. Ideally, those data should be obtained experimentally. Sometimes the results obtained from experimental tests are not reliable or accessible. In this study, the PVT properties are predicted by a new Artificial Neural Network (ANN) model using composition mole percent, solution gas oil ratio, bubble point pressure, reservoir pressure and temperature. The designed ANN used is from the Committee Machine type. These networks process their input using two parallel MLPs, and then recombine their results. The results obtained show that Committee Machines are dependable networks for prediction of PVT properties in reservoirs among the other ANNs and empirical correlations.