Abstract:
Physical properties and pressure-volume-temperature (PVT) data of crude oil are necessary for various field applications, such as field development, production optimization, and the enhanced oil recovery process. In this work, crude oil data were gathered from publications for modeling correlations and artificial neural networks (ANN), which could be used to predict physical properties of crude oil, such as bubble point pressure, oil formation volume factor at bubble point. Solution gas oil ratio, and oil viscosity at pressure above bubble point. The data were divided into two sets. The first was used to develop and the second was used for testing the correlations and ANN models. The correlations were developed using a non-linear regression technique. For ANN development, different network architectures and transfer functions were used for developing the best ANN models. To ensure accuracy and applicability, the sets of data for testing were employed with the developed models. Moreover, the developed models were tested with other published correlations in terms of performance and accuracy using the data for testing. The results showed that the developed ANNs and correlations gave competitive performance compared with other published correlations under the data used in this work.
