No.4(44),2025 PP.251-265

Neftyanaya provintsiya No. 4(44), 2025

A comprehensive approach to well interference modeling using physically-based graph neural networks

А.А. Gaysin, R.Kh. Nizaev
DOI: https://doi.org/10.25689/NP.2025.4.251-265

Abstract

For effective development of oil fields, it is necessary to correctly account for the interaction between injection and production wells. The interference coefficient is a key parameter characterizing the degree of water injection impact on oil production. Traditional methods for its determination: analytical calculations, field studies, and hydrodynamic modeling - have a number of limitations: from simplified physical assumptions to high computational complexity. In this regard, the application of machine learning methods, particularly graph neural networks (GNN), opens up new opportunities for more accurate and rapid determination of well interference, taking into account the complex structure of the development system.

Key words:

interference coefficient, graph neural network, algorithms, streamlines, machine learning, hydrodynamic modeling

References

1. Z. Zhu, Z. Lei, Z. Chen Waterflood Management Using Hybrid Approach of Streamline-Based Interwell Flux Information and Finite Volume Reservoir Simulation // Advances in Civil Engineering. – 2018. – No. 1. – 12 p.

2. V.F. Dyagilev, et al. Analysis of the tracer studies results: a study of AS1-3 formation of the Severo-Orekhovskoye oil field // Oil and Gas Studies. – 2018. – No. 4. – pp. 44-51. (in Russian)

3. C. Tian, R.N. Horne Inferring interwell connectivity using production data // SPE Annual Technical Conference and Exhibition, Sep 26-28, 2016, Dubai, UAE. – 2016. – 11 p. – SPE-181556-MS.

4. D.A. Martiushev, P.Iu. Iliushin Express assessment of the interaction between the production and injection wells in the tournasian-famennian deposits of Ozernoe field // Bulletin of PNRPU. Geology. Oil & Gas Engineering & Mining. – 2016 – Vol. 15, No. 18. – pp. 33-41.

5. Graph attention networks / P. Velickovic [et аl.] // ICLR 2018 : сonference paper. – 2018 – 12 p.

Authors

A.A. Gaysin, PhD Candidate, Chair Oil and Gas Fields Development and Operation, Almetyevsk State University of Technology – Higher Petroleum School, TatNIPIneft Institute – PJSC Tatneft
2, Lenina St., Almetyevsk, 423462, Russian Federation
E-mail: GaysinAA@tatneft.ru

R. Kh. Nizaev, Dr. Sc., Professor of Chair Oil and Gas Fields Development and Operation, Almetyevsk State University of Technology – Higher Petroleum School, TatNIPIneft Institute – PJSC Tatneft
2, Lenina St., Almetyevsk, 423462, Russian Federation
E-mail: nizaev@tatnipi.ru

For citation:

А.А. Gaysin, R.Kh. Nizaev Kompleksnyy podkhod k modelirovaniyu vzaimovliyaniya skvazhin s ispol'zovaniyem fizicheski obosnovannykh grafovykh neyronnykh setey [A comprehensive approach to well interference modeling using physically-based graph neural networks]. Neftyanaya Provintsiya, No. 4(44), 2025. pp. 251-265. DOI https://doi.org/10.25689/NP.2025.4.251-265. EDN ZWSOOR (in Russian)

© Non-governmental organization Volga-Kama Regional Division of the Russian Academy of Natural Science, 2015-2025 All the materials of the journal are available under the Creative Commons Attribution 4.0 License (https://creativecommons.org/licenses/by/4.0/)

A comprehensive approach to well interference modeling using physically-based graph neural networks А.А. Gaysin, R.Kh. Nizaev DOI: https://doi.org/10.25689/NP.2025.4.251-265

Key words: interference coefficient, graph neural network, algorithms, streamlines, machine learning, hydrodynamic modeling

A comprehensive approach to well interference modeling using physically-based graph neural networks

А.А. Gaysin, R.Kh. Nizaev
DOI: https://doi.org/10.25689/NP.2025.4.251-265

Key words:

interference coefficient, graph neural network, algorithms, streamlines, machine learning, hydrodynamic modeling