Improving EOR Selection Efficiency with Novel EOR Screening Software: Applied Machine Learning

Abstract

With the decreasing rate of production, the application of Enhanced Oil Recovery (EOR) technology is one solution to increase the recovery of oil production from reservoirs. EOR is a method to increase production by influencing the interaction between fluids and reservoir rocks. Before an EOR method is implemented, it is necessary to conduct EOR screening so that the EOR method is most suitable for field conditions and the increase in oil production is achieved optimally. Conventionally, EOR screening methods are done manually matching field parameters with criteria sourced from Taber et al (1997) and from Al-Adasani et al (2010). This method has several drawbacks including taking a long time, high subjectivity, and qualitative. As a result, screening results using this method have poor accuracy, low effectiveness, and have high uncertainty. In this study, we proposed quantitative methods for EOR screening based on automation systems in software built with Machine Learning algorithms. This screening method is based on a static and probabilistic evaluation of the most suitable combination of 8 parameters of oil and reservoir characteristics and tested on 5 different fields based on Taber et al (1997) and Al-Adasani & Bai (2010) criteria. Based on the test results, the order of conformity rating of EOR method is obtained along with the evaluation of the score. Furthermore, a comparative analysis is conducted with the results of EOR screening manually and with the results of screening using other commercial software. The results of this study show that the proposed method can produce better output because the process is efficient in terms of computational time, more reliable results, and quantitative. This method is expected to be one of the solutions for the acceleration of the EOR program in support of the target of increasing national oil production.