Artificial intelligence–driven prediction on oil formation volume factor for Niger delta region

Oil formation volume factor (OFVF) is very important reservoir fluid property used in the reservoir engineering calculations either directly or indirectly. It indicates the change in the volume of produced oil from the reservoir to surface conditions. The industry standard is to measure this parameter in the laboratory using reservoir samples but the procedures of acquiring these experimental data are very expensive, and time consuming, hence the application of correlations and artificial intelligent. Before now, some empirical correlations existed that determines the oil formation volume factor however, they show high error due to the applied assumptions, and their specification to operate only under a particular range of data. In this study, Super Learner (SL) and Deep Neural Network (DNN) artificial intelligent algorithms were trained to forecast oil formation volume factor at different reservoir condition using Niger Delta reservoir fluid. A total number of 1147 data set was obtained from PVT report from Niger-Delta and validated, out of which, 70% (803) were used to train the models, 15% (172) for cross validation and 15% (172) for test. Quantitative and qualitative statistical analysis were carried out to compare the performance and reliability of the new developed machine learning models with some selected oil formation volume factor empirical correlations. The Super Leaner (SL) model predicted better than Deep Neural Network (DNN) and some of the selected empirical models with the best Rank of 0.05, average absolute percent relative error of 0.054and a correlation coefficient of 0.989. The findings from this research can be applied for estimation of the volumetric properties of hydrocarbon reservoir fluids without the need for conducting routine laboratory analyses.