Perpustakaan Digital ITB

Reservoir characterization is a process to explain the quantitative and qualitative characteristic of a reservoir using all existing data. By understanding the reservoir, we can maintain and optimize its lifetime performance, its ability to store and produce hydrocarbons (Sukmono, 2002). This study proposes integrated reservoir characterization workflow to predict lithology, porosity, and permeability using machine learning algorithms. The study conducted using wireline logs and core measurement data from Baturaja (BRF) and Talang Akar (TAF) formation, with a total of 379 core samples available from 14 wells. Machine learning is a field of study that focuses on understanding and building computational methods that are able to learn and improve with experience (Mitchell, 1997). Machine learning techniques can be applied as a quick, cost-effective solution to estimate reservoir parameters from complex coupled relations to indirect measurements (Andersen, et al., 2022). Models are undergone training using several machine learning algorithms and optimized by applying hyperparameter tuning. The best model result gives promising performance for predicting unknown properties in uncored interval of wells.