Heterogeneous oil reservoirs characterization using artificial intelligence techniques: Application to the Hassi Messaoud oil field in the Algerian-Saharan platform
{"title":"Heterogeneous oil reservoirs characterization using artificial intelligence techniques: Application to the Hassi Messaoud oil field in the Algerian-Saharan platform","authors":"Asma Kadri , Mohammed Said Benzagouta","doi":"10.1016/j.jappgeo.2025.105878","DOIUrl":null,"url":null,"abstract":"<div><div>This study explores the effectiveness of artificial intelligence (AI) techniques in characterizing heterogenous reservoirs, with a specific focus on the Hassi Messaoud oil field in southern Algeria, particularly its newly developed northern zone known for complex reservoir heterogeneity affecting oil extraction efficiency.</div><div>Three machine learning algorithms Support Vector Machine (SVM), Random Forest (RF), and Artificial Neural Networks (ANN) were applied to predict porosity from well log data. SVM outperformed RF and ANN, delivering the highest correlation coefficients (R<sup>2</sup>) and the lowest root mean squared errors (RMSE), thereby confirming its robustness in high-dimensional spaces and limited datasets. Beyond porosity prediction, this study addresses the critical task of permeability estimation, essential for optimizing reservoir development strategies. A Multi-Linear Regression (MLR) model was developed, achieving high predictive accuracy, particularly when lithological parameters such as clay content were incorporated. A key contribution of this work is the integration of SVM and MLR into a hybrid SVM–MLR model, which improved permeability prediction by leveraging the strengths of both methods: the nonlinear feature-handling capability of SVM and the interpretability of MLR. The SVM-generated correlation matrix facilitated the identification of dominant input features, enhancing the reliability of the permeability model.</div><div>The findings demonstrate that the integrated SVM–MLR approach provides a powerful and adaptable tool for reservoir characterization in heterogeneous environments. This AI-driven framework offers valuable support for data-informed decision-making, contributing to more efficient hydrocarbon recovery and improved reservoir management in geologically complex settings.</div></div>","PeriodicalId":54882,"journal":{"name":"Journal of Applied Geophysics","volume":"242 ","pages":"Article 105878"},"PeriodicalIF":2.1000,"publicationDate":"2025-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Applied Geophysics","FirstCategoryId":"89","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0926985125002599","RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"GEOSCIENCES, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0
Abstract
This study explores the effectiveness of artificial intelligence (AI) techniques in characterizing heterogenous reservoirs, with a specific focus on the Hassi Messaoud oil field in southern Algeria, particularly its newly developed northern zone known for complex reservoir heterogeneity affecting oil extraction efficiency.
Three machine learning algorithms Support Vector Machine (SVM), Random Forest (RF), and Artificial Neural Networks (ANN) were applied to predict porosity from well log data. SVM outperformed RF and ANN, delivering the highest correlation coefficients (R2) and the lowest root mean squared errors (RMSE), thereby confirming its robustness in high-dimensional spaces and limited datasets. Beyond porosity prediction, this study addresses the critical task of permeability estimation, essential for optimizing reservoir development strategies. A Multi-Linear Regression (MLR) model was developed, achieving high predictive accuracy, particularly when lithological parameters such as clay content were incorporated. A key contribution of this work is the integration of SVM and MLR into a hybrid SVM–MLR model, which improved permeability prediction by leveraging the strengths of both methods: the nonlinear feature-handling capability of SVM and the interpretability of MLR. The SVM-generated correlation matrix facilitated the identification of dominant input features, enhancing the reliability of the permeability model.
The findings demonstrate that the integrated SVM–MLR approach provides a powerful and adaptable tool for reservoir characterization in heterogeneous environments. This AI-driven framework offers valuable support for data-informed decision-making, contributing to more efficient hydrocarbon recovery and improved reservoir management in geologically complex settings.
期刊介绍:
The Journal of Applied Geophysics with its key objective of responding to pertinent and timely needs, places particular emphasis on methodological developments and innovative applications of geophysical techniques for addressing environmental, engineering, and hydrological problems. Related topical research in exploration geophysics and in soil and rock physics is also covered by the Journal of Applied Geophysics.