Using Artificial Intelligence to Predict IPR for Vertical Oil Well in Solution Gas Derive Reservoirs: A New Approach

Abstract

Abstract Well Inflow Performance Relationship (IPR) has a wide range of applications in both applied and theoretical sciences, especially in the petroleum production engineering. An accurate prediction of well IPR is very important to determine the optimum production scheme, design production equipment, and artificial lift systems. For these reasons, there is a need for a quick and reliable method for predicting oil well IPR in solution gas drive reservoirs. In this paper, back propagation network (BPN) and fuzzy logic (FL) techniques are used to predict oil well IPR in solution gas drive reservoirs. The models were developed using 207 data points collected from unpublished sources. Statistical analysis was performed to define the more reliable and accurate techniques to predict the IPR. According to the results, the new fuzzy logic well IPR model outperformed the artificial neural networks (ANN) model and the most common empirical correlations. The average absolute error, least standard deviation and highest correlation coefficient were used to evaluate the models results. The proposed fuzzy logic well inflow performance relationship model achieved an average absolute error of 1.8 %, standard deviation of 2.9% and the correlation coefficient of 0.997. The developed technique will help the production and reservoir engineers to better manage the production operation without the need for any additional equipment. It will also reduce the overall operating cost and increase the revenue.