SHALE GAS FACIES CHARACTERIZATION WITH MACHINE LEARNING

• Main Author: Novis Saputri, Tiara
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/76500
• Institution: Institut Teknologi Bandung

This research examines the analysis of mineral content in shale rocks in the Central Sumatra and South Sumatra Basins using X-ray Diffraction (XRD) and facies classification using machine learning methods such as K-Means clustering and Support Vector Machine (SVM) algorithms. Two fields, field A (Kel. Pematang) from Central Sumatra and field B (Lahat) from South Sumatra were the source of shale rock samples. XRD results revealed the dominance of clay minerals in the two fields with different percentages: field A has smectite (11.9%), illite (25.9%), kaolinite (12.6%), and chlorite (11.7%), while field B has smectite (1%), illite (26.7%), kaolinite (16%), and chlorite (12%). The distribution of minerals in the samples identified using ternary plot mineral diagrams shows both fields are dominated by non-swelling clay minerals (illite, kaolinite, and chlorite). The KMeans clustering algorithm in this research successfully separated the samples into two clusters according to the set k value, but the categorization accuracy of the field of origin still needs to be improved. The Support Vector Machine (SVM) algorithm successfully classified the data with a high level of accuracy rates of 0.83 and 1.00, although it is less effective on identical data. This research provides further understanding of the characteristics and mineral composition of shale rock as well as the application of machine learning in facies classification.