Search

In reservoir engineering calculations, PVT properties are critical. There are two methods for calculating different PVT properties, empirical correlations, and machine learning techniques. The development of neural networks has paved the way for data mining modeling processes to be a key player in the oil and gas industry. However, there are several defects and limits in the neural network modeling structures because they were created with a specific collection of reservoir fluid properties. As a result, various artificial intelligence approaches to predicting PVT qualities became necessary. This paper uses Support Vector Machines (SVM) to formulate new oil properties correlations for Iraqi oil reservoir fluids. In a comparative study, the efficiency of support vector machines regression is compared with existing empirical correlations. The results show that support vector machines outperform some published correlations in accuracy and reliability, which is a promising future for support vector machine modeling. [ABSTRACT FROM AUTHOR]