Abstract
Accurate determination of the bubble point pressure is extremely important for several applications in oil industry. In reservoir engineering applications the bubble point pressure is an essential input for the reservoir simulation and reservoir management strategies. Also, in production engineering the bubble point pressure determines the type of the inflow performance relationship that describes the reservoir production performance. Accurate estimation and prediction of the bubble point pressure will eliminate the risk of producing in two phase region. Current correlations can be used to determine the bubble point pressure with high errors and this will lead to poor reservoir management. Artificial intelligent tools used in the previous studies did not disclose the models they developed and they stated the models as black box.
The objective of this research is to develop a new empirical correlation for bubble point pressure (BPP) prediction using artificial intelligent techniques (AI) such as; artificial neural network (ANN), adaptive neuro-fuzzy inference system (ANFIS), and supper vector machine (SVM). For the first time we changed the ANN model to a white box by extracting the weights and the biases from AI models and form a new mathematical equation for BPP prediction.
The results obtained showed that the ANN model yielded the highest correlation coefficient (0.988) and lowest average absolute error percent (7.5%) for BPP prediction as a function of the gas specific gravity, the solution gas oil ratio, the oil gravity, and the reservoir temperature as compared with ANFIS and SVM. The developed mathematical equation from the ANN model outperformed the previous AI models and the empirical correlations for BPP prediction. It can be used to predict the BPP with a high accuracy (the average absolute error (3.9%) and the coefficient of determination (R2) of 0.98)
