Perpustakaan Digital ITB

Carbon Dioxide Enhanced Oil Recovery (CO2-EOR) offers the potential to increase oil recovery while supporting geological CO2 storage. This study evaluates the performance of integrated CO2-EOR and CO2 storage in a mature carbonate reservoir using compositional reservoir simulation and machine learning- based proxy models. A total of 101 simulation scenarios were generated by varying injection scheme, injection rate, total injected CO2, and Water Alternating Gas (WAG) ratio. Reservoir performance was assessed using Incremental Oil Recovery (IOR) and CO2 Retention. The results indicate that increasing injection rate improves IOR but reduces CO2 Retention due to earlier CO2 breakthrough, whereas increasing total injected CO2 primarily enhances storage performance. The influence of WAG ratio was relatively limited compared with continuous CO2 injection. Five machine learning algorithms were evaluated as proxy models, namely Ridge Regression, Random Forest, Extra Trees, Gradient Boosting, and Support Vector Regression (SVR). SVR achieved the best performance with cross-validation R2 values of 0.856 for IOR and 0.911 for CO2 Retention. The developed proxy model successfully enabled rapid scenario evaluation and identified continuous CO2 injection of 9,000 tons at 175 ton/day as the best balanced-performance scenario.