Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Improving the accuracy of terrigenous reservoir porosity modeling based on machine learning methods
The purpose of this paper is to improve the prediction of porosity in Visean terrigenous sediments to increase the accuracy of digital static model of an oil field in Perm Krai. The study proposes an approach for porosity prediction based on machine learning methods, which can compensate for the shortcomings of traditional methods based on geophysical well logging data. Technical limitations of well logging and high geological heterogeneity often interfere with obtaining reliable porosity distribution data. The study uses such algorithms as Random Forest, Gradient Boosting, Support Vector method and Adaptive Boosted Decision Trees for porosity determination based on a set of geophysical methods. The developed model, trained on a specially created database using radioactive, electric and acoustic logs, model was implemented for real reservoir. Implementation of the model allowed to significantly refine the static model of the field and adjust the reserves estimation. The economic effect is achieved by reducing the cost of additional research and improving the efficiency of reservoir management. The proposed methodology has been successfully tested and can be used for other fields in the south of the Perm region, which opens up prospects for improving the efficiency of oil field development in the region.
Improving the accuracy of terrigenous reservoir porosity modeling based on machine learning methods
The purpose of this paper is to improve the prediction of porosity in Visean terrigenous sediments to increase the accuracy of digital static model of an oil field in Perm Krai. The study proposes an approach for porosity prediction based on machine learning methods, which can compensate for the shortcomings of traditional methods based on geophysical well logging data. Technical limitations of well logging and high geological heterogeneity often interfere with obtaining reliable porosity distribution data. The study uses such algorithms as Random Forest, Gradient Boosting, Support Vector method and Adaptive Boosted Decision Trees for porosity determination based on a set of geophysical methods. The developed model, trained on a specially created database using radioactive, electric and acoustic logs, model was implemented for real reservoir. Implementation of the model allowed to significantly refine the static model of the field and adjust the reserves estimation. The economic effect is achieved by reducing the cost of additional research and improving the efficiency of reservoir management. The proposed methodology has been successfully tested and can be used for other fields in the south of the Perm region, which opens up prospects for improving the efficiency of oil field development in the region.
Improving the accuracy of terrigenous reservoir porosity modeling based on machine learning methods
Krivoshchekov, S. N. (Autor:in) / Shiverskii, G. V. (Autor:in) / Kochnev, A. A. (Autor:in) / Yuzhakov, A. L. (Autor:in)
Geosystem Engineering ; 28 ; 96-111
04.03.2025
16 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Application of Extreme Learning Machine to Reservoir Proxy Modeling
| Springer Verlag | 2022
Application of Extreme Learning Machine to Reservoir Proxy Modeling
| Online Contents | 2022
Assessment of Different Machine Learning Methods for Reservoir Outflow Forecasting
| DOAJ | 2023