Perpustakaan Digital ITB

Low Salinity Water Injection (LSWI) is an effective enhanced oil recovery (EOR) method, particularly in sandstone formations containing reactive clay minerals. However, LSWI may pose risks such as inefficiency, clay swelling, and formation damage if the injection design assumes the reservoir rock to be homogeneous, while in reality it exhibits lateral and vertical heterogeneity in mineral composition. Therefore, identifying the distribution and type of clay minerals is an essential step before selecting optimal injection zones. This study aims to identify clay types using well log data through a machine learning approach, estimate the Cation Exchange Capacity (CEC), and recommend optimal LSWI injection zones. The data were obtained from eight wells located in two structures (B and T) of Field S, South Sumatra. The well log parameters used include Gamma Ray (GR), Bulk Density (RHOB), Neutron Porosity (CNL), and Photoelectric Factor (PEF). Classification was carried out using Euclidean Distance, K-Means, and Proximity Similarity Classification (PSC) methods, based on clay mineral petrophysical references from the literature. The classification results indicate variations in clay distribution across wells and reservoir zones. Several wells such as B-4 and T-4 are dominated by smectite with high CEC values (>80 meq/100g), while wells T-1, T-3, and S-1 are dominated by illite and kaolinite with low to moderate CEC values. Zones containing stable clay are recommended for injection, while smectite-rich zones may still be utilized with preflush treatment. This approach enables efficient identification of reactive clay solely from well log data, without the need for core or XRD analysis, and can serve as a foundation for designing safer and more targeted LSWI strategies.