Development of a new approach using an artificial neural network for estimating oil formation volume factor at bubble point pressure of Egyptian crude oil

Understanding the phase behavior and volumetric changes of reservoir fluids throughout the extent of reservoir lifetime are crucially required for effective-commercial oil recoveries. Ideally, reservoir fluid properties are experimentally measured by laboratory experiments known as PVT tests nonetheless, these tests are prohibitive, time-consuming, and required to restrict sampling and transporting procedures. For these discernible reasons, several modeling approaches have been developed. By reviewing the literature, one crucial obstacle that encounters field applicability of most extant models is the selection of input variables. Moreover, a great percentage of extant models employ the results of lengthy experimental tests such as differential gas solubility or even the sample’s chemical composition. Replicability of models’ results using different datasets is also one of the main challenges when employing AI models. Frequently, inadequate descriptions for AI models have been provided in many studies which limits their utility. The inadequate description includes the analysis of ANN model weights and biases besides, the final mathematical model. In this study, a rigorous ANN model with its mathematical model has been implemented to predict oil formation volume factors based on 600 compiled datasets from Egyptian oilfields.

A detailed comparison between widely used empirical correlations and the proposed new ANN model is provided in this study. Statistical and graphical analysis depicted the outstanding performance of the new model with R² = 0.974, ARE = −0.0017%, and AARE = 2.13%. The ANN model provides remarkable sustainable performance compared to local Egyptian empirical correlations and all the other global models.