[This article belongs to Volume - 54, Issue - 04]
Gongcheng Kexue Yu Jishu/Advanced Engineering Science
Journal ID : AES-08-06-2022-201

Title : A Comparative Study Between Experimental And Computerized Observations By Generating An Optimal Surfactant-Polymer Design Using AI Intelligence
Omar Khaled, Philopateer Mehanni, Youssef Hablas, Attia M. Attia,

Abstract : Designing an optimal surfactant polymer design is very crucial and requires numerous laboratory tests that require infinite time, higher financial statements, and probably lack accuracy. Consequently, AI techniques have taken over, proving excellent accuracy levels with less time and lower cost. This paperwork is objectified to eliminate any laboratory work when it comes to generating an optimal surfactant-polymer design with the freedom to do multiple adjustments across the factors affecting the process and their responses in minutes and with financial ease. Consequently, a comparative study is done throughout this paper in order to have an optimum prediction for either factor affecting EOR recovery or their outcomes. However, this paper has traced the footprints of the previous experimental study by considering the critical-micelle-concentration of 1.25wt.% is the optimum alongside depending on the laboratory observations at evaluating the surfactant, the values of each factor, and its respective result under specific conditions i.e.: high-salinity. Nonetheless, the computerized observations have slightly deviated when it comes to the polymer concentration which altered everything afterward analytically and statistically, however it is realistic. The Design-Expert software was used to initiate this comparative study, it has shown promising outcomes as another XG concentration was predicted at 1871.61 ppm rather than the former 2000 ppm. This caused a difference between both the previous and present study. For instance, experimentally IFT chosen was 3.1 mN/m, contact angles chosen were 50.1° and 60.52°, viscosity chosen was 11.78 cp and the oil recovery factor percentage was 53% after SP flooding. Oppositely, the AI design shows an IFT of 3.0072 mN/m, contact angles of 49.75° and 59.53°, a viscosity of 11.99 cp, and an oil recovery factor of 51.41%. The model has shown high accuracy with the observation of R^2 reaching 0.9988 and standard deviation as low as 0.1121. The success of the proposed AI model would be a plot twist if applied to Egyptian oil fields due to the minor role of EOR in such regions.