Perpustakaan Digital ITB

Predicting oil-surfactant-water system interfacial tension (IFT) may facilitate CEOR surfactant screening and injection. Traditional IFT prediction needs time-consuming and expensive laboratory studies. This study uses machine learning algorithms to estimate interfacial tension (IFT) values to optimize surfactant screening. This study will employ Orange and Python, machine learning tools. The orange program evaluates machine learning on existing data. Orange software's finest machine learning model is gradient boosting. After reviewing the literature, the Python-optimized XGBoost machine learning model was chosen. This study chose XGBoost since it has various advantages over gradient boosting. This study uses data from small lab experiments. This study developed a dependable and accurate model with a test data prediction R2 of 0.891 and MSE of 0.09 that can learn from limited data and predict IFT values. According to simulations, the hydrophobic number of surfactants and oil EACN most affect the IFT value in this system. These results show that machine learning can overcome data restrictions, properly forecast IFT, and identify the most important variables. This discovery is intended to help engineers improve surfactant screening and prepare the way for future IFT prediction research.