Artificial Intelligence Methods for Estimation of Formation Porosity from Well Logging Data

Document Type : Original Article

Authors

1 Birjand University of Technology, Birjand, Iran

2 Department of Petroleum Engineering, Kho.C., Islamic Azad University, Khomeinishahr, Iran; Stone Research Center, Kho.C., Islamic Azad University, Khomeinishahr, Iran

10.22077/jgm.2025.9727.1054

Abstract

Porosity is an essential rock property reflecting the capacity of the reservoir to store hydrocarbons, making it a critical parameter in the exploration and development of oil and gas resources. Well logging is a fundamental technique in the oil and gas industry for estimating the porosity of subsurface formations. Well logging methods do not directly measure porosity, but instead record physical parameters (e.g., bulk density, acoustic travel time, hydrogen index) that can be empirically or theoretically related to the porosity of the rock. The artificial intelligence (AI) methods of Backpropagation Neural Network (BPNN), General Regression Neural Network (GRNN), and Support Vector Machine (SVM) were adopted to estimate porosity from well-logging data points using MATLAB software. For this purpose, 70% of the data was used for training, and 30% of the data was used for testing these AI methods. The well logging dataset obtained from one of the Kaggle datasets belongs to the Peninsular Malaysia hydrocarbon field. The comparison of the porosity values predicted by the AI methods and the actual values indicated that the three AI methods predict porosity values with great accuracy (with R values all equal or very close to one). However, the BPNN method has a smaller error in estimating porosity compared to the other two AI methods, suggesting that BPNN outperforms the other two AI techniques in this study.

Keywords

Journal of Geomine
Volume 3, Issue 2
June 2025
Pages 126-139

Files

Share

How to cite

Statistics