Permeability prediction from well-logs by applying Hydraulic flow unit and Machine learning: a case study of Cuu Long basin, offshore Vietnam
Abstract
Relevance. This paper addresses the challenge of classifying and predicting hydraulic flow units in reservoirs, particularly in heterogeneous formations like the Cuu Long basin, offshore Vietnam. Integrating machine learning techniques offers a modern and efficient approach to reservoir characterization, which is crucial for improving permeability estimation and reducing uncertainty in reservoir modeling. Aim. To suggest a novel strategy to enhance the estimation of permeability and provide a more continuous geological profile for reservoir engineers. Objects. Core samples and well log data collected from 42 wells in the Cuu Long basin, offshore Vietnam. Methods. Machine learning techniques – both unsupervised and supervised were applied to predict permeability from well log data. Results. The study found that Fuzzy C-means clustering provided the optimal classification of hydraulic flow units, with 5 HUs being the best division for sedimentary reservoirs in the Cuu Long Basin. Among supervised learning models, Random Forest performed the best in clustering HUs, achieving an accuracy of 93% in training and 86% in testing, outperforming Boosted Tree, Artificial Neural Network, and Support Vector Machine. The integration of core and well log data improved hydraulic flow units classification and permeability prediction in heterogeneous reservoirs, providing a valuable tool for improved reservoir characterization and management in the Cuu Long basin and similar geological settings. For citation: Vu Hong Duong, Nguyen Thanh Tung, Nguyen Minh Hoa, Nguyen Danh Lam, Ha Quoc Dat, Nguyen Xuan Duy. Permeability prediction from well-logs by applying hydraulic flow unit and machine learning: a case study of Cuu Long basin, offshore Vietnam. Bulletin of the Tomsk Polytechnic University. Geo Assets Engineering, 2026, vol. 337, no. 8, pp. 211–226. http://doi.org/10.18799/24131830/2026/8/5049