POWER CONTROL AND OPTIMIZATION: Proceedings of the 7th Global Conference on Power Control and Optimization
978-983-44483-63
Prediction of Porosity and Permeability of Oil and Gas Reservoirs Using Support Vector Machines and Artificial Neural Networks: A Comparative Study
Ewenla A., Anifowose F., Akanbi L., Oluwatope O. A., Aderounmu G. A
PCO Conf-Proc 2008 (2013), - PDF
Abstract. There has been a persistent quest for improved prediction of the properties of oil and gas reservoir. The performance of existing techniques has been acceptable but needs to be improved upon. Petroleum reservoir modeling has evolved from the use of empirical and statistical tools to the embrace of computational intelligence techniques especially Artificial Neural Networks (ANN). ANN has been used extensively in petroleum engineering applications but has a lot of limitations. This paper presents a comparative study of the application of ANN and Support Vector Machines (SVM) in the prediction of porosity and permeability of petroleum reservoirs. SVM has been reported in literature to be efficient, easy to train and resistant to overfitting. The petroleum industry has not adequately benefitted from the excellent generalization capability of SVM. Datasets from different petroleum reservoirs were used in this study to develop, train and evaluate the comparative performance of ANN and SVM models. The results showed that the SVM models performed better, in terms of higher correlation coefficients, lower root mean squared errors and less execution time. Hence, we present SVM as a possible alternative to ANN, especially, in the prediction of petroleum reservoir properties.
© 2013 PCO based on American Institute of Physics