INTEGRATION OF CONVENTIONAL WELL LOGS AND MACHINE LEARNING APPROACHES FOR FRACTURE TYPE PREDICTION: A CASE STUDY ON VOLCANIC RESERVOIRS IN INDONESIA

• Main Author: Arifinka Alhazmi, Enricho
• Format: Theses
• Language: Indonesia
• Subjects: Pertambangan dan operasi berkaitan
• Online Access: https://digilib.itb.ac.id/gdl/view/84949
• Institution: Institut Teknologi Bandung

Volcanic reservoirs are valuable targets in the oil and gas industry due to their potential for hydrocarbon storage. Secondary porosity plays a crucial role in storage capacity and fluid flow within volcanic reservoirs. However, predicting secondary porosity in volcanic reservoirs poses significant challenges due to the complex geological processes involved. Previous studies on volcanic reservoirs often exhibit significant complexity and heterogeneity in the distribution of secondary porosity. The proposed method attempts to train on information collected from conventional well logs to predict fracture category that indicates secondary porosity. This method uses the massive quantity of data available from well logs to locate and isolate porosity features, giving a more cost-effective and efficient alternative to formation micro imaging (FMI). In this study, the integration of advanced machine learning methods with conventional well logging data improved the accuracy of fracture category identification. This approach uses supervised machine learning and unsupervised machine learning techniques as a comparison to improve the accuracy of fracture category identification. Both supervised and unsupervised machine learning will be utilized to forecast fracture category in volcanic reservoirs. These proposed approaches are intended to provide important contributions to the understanding and characterization of volcanic reservoirs, as well as allow for more accurate decision making in the field of oil and gas.