Perpustakaan Digital ITB

The increasing global energy demand necessitates efficient and optimized oil and gas production methods. Enhanced Oil Recovery (EOR) techniques, particularly chemical EOR (CEOR), offer a promising solution by injecting chemicals into reservoirs to modify the properties of the reservoir and the oil, thereby enhancing recovery. A critical factor in CEOR is Interfacial Tension (IFT), which prevents the mixing of fluid molecules. This research focuses on analyzing variables that affect IFT prediction to develop a predictive model using the KNIME machine learning tool. The optimal scenario identified excludes the Oil Solution Ratio (C1) and includes variable phase behavior, achieving an R² value of 0.916 and a Mean Absolute Percentage Error (MAPE) of 0.096%. This highly accurate model can predict IFT based on the input variables used in its development. Consequently, future IFT predictions and analyses can be efficiently conducted with minimal cost and time by simply inputting the necessary variables. This advancement streamlines the process, making CEOR more accessible and cost-effective.