Practical application of GPU technologies for optimizing oil field development

UDK: 681.518.:622.276

DOI: 10.24887/0028-2448-2025-12-128-132

Key words: reservoir simulation, oil field development, Boundary Element Method (BEM), Graphics Processing Unit (GPU), parallel computing, CUDA technology, acceleration of hydrodynamic calculations

Authors: B.M. Latypov (Ufa State Petroleum Technological University, RF, Ufa); E.V. Yudin (Gazprom Neft Companу Group, RF, Saint Petersburg); V.A. Kupavykh (Ufa State Petroleum Technological University, RF, Ufa); Z.A. Bogdanov (NEDRA LLC, RF, Saint Petersburg); S.I. Kogakov (OIL AND GAS PRODUCTION TOOLS LLC, RF, Krasnogorsk); N.S. Markov (NEDRA LLC, RF, Saint Petersburg)

This paper presents the development and industrial validation of a Graphics Processing Unit (GPU) - accelerated reservoir simulation system based on the Boundary Element Method (BEM). Traditional CPU-centric numerical simulators approached their performance limits, motivating the transition to massively parallel computing architectures. The proposed three-layer system integrates a Python-based control layer, a C++ hybrid computational core, and a CUDA subsystem responsible for high-throughput parallel processing. The BEM formulation enables natural task decomposition: the contribution of each well, fracture or boundary segment is computed independently, enabling efficient distribution across thousands of GPU threads and minimizing memory-access overhead. The system was evaluated on four real-field reservoir models ranging from simple homogeneous structures to large, highly heterogeneous assets. Benchmarking demonstrated substantial performance gains: GPU acceleration achieved speed-up factors between 40 and 124 compared with a CPU-only implementation. Double-precision calculations maintained high numerical accuracy, with average deviations below 2–2,5 %. Single-precision mode, while slightly less accurate (3–5 % deviation), provided maximum performance suitable for rapid scenario screening and interactive workflows. The results exhibit nearly linear scalability and high efficiency even when the model size was increased by an order of magnitude. Pressure-field maps and well-pressure dynamics confirm the physical consistency of the model outputs. Overall, the study demonstrates that the proposed GPU-accelerated BEM framework enables near real-time reservoir simulation while preserving engineering-grade accuracy. The architecture is compatible with existing digital workflows and offers a practical pathway for integrating high-performance GPU computing into routine reservoir-engineering decision-making.

References

1. Bayat M., Killough J.E., An experimental study of GPU acceleration for reservoir simulation, SPE-163628-MS, 2013, DOI: https://doi.org/10.2118/163628-MS

2. Zhao L., Li S., Zhang C.S. et al., An improved multistage preconditioner on GPUs for compositional reservoir simulation, CCF Transactions on High Performance Computing, 2023, V. 5, no. 2, pp. 136–151, DOI: https://doi.org/10.1007/s42514-023-00136-0

3. Middya U., Manea A., Alhubail M. et al., A massively parallel reservoir simulator on the GPU architecture, SPE-203918-MS, 2021, DOI: https://doi.org/10.2118/203918-MS

4. Dogru A.H., Fung L.S.K., Al-Shaalan T.M. et al., From mega-cell to giga-cell reservoir simulation, SPE-116675-MS, 2008, DOI: https://doi.org/10.2118/116675-MS

5. Dogru A.H., Fung L.S.K., Middya U. et al., A next-generation parallel reservoir simulator for gi-ant reservoirs, SPE-119272-MS, 2009,

DOI: https://doi.org/10.2118/119272-MS

6. Cao H., Zaydullin R., Liao T. et al., Adding GPU acceleration to an industrial CPU-based simulator, development strategy and results, SPE-203936-MS, 2021,

DOI: https://doi.org/10.2118/203936-MS

7. Esler K., Gandham R., Patacchini L. et al., A GPU-based, industrial grade compositional reservoir simulator, SPE-203929-MS, 2021,

DOI: https://doi.org/10.2118/203929-MS

8. Khrapov S.S., Khoperskov A.V., Smoothed-particle hydrodynamics models: implementation features on GPUs, Communications in Computer and Information Science, 2018, V. 793, pp. 266–277, DOI: https://doi.org/10.1007/978-3-319-71255-0_21

9. Illarionov E., Temirchev P., Voloskov D. et al., End-to-end neural network approach to 3D reservoir simulation and adaptation, Journal of Petroleum Science and Engineering, 2022, V. 208, DOI: https://doi.org/10.1016/j.petrol.2021.109332

10. Temirchev P., Gubanova A., Kostoev R. et al., Reduced order reservoir simulation with neural-network based hybrid model, SPE-196864-MS, 2019,

DOI: https://doi.org/10.2118/196864-MS

11. Petrosyants M., Illarionov E., Koroteev D., Speeding up the reservoir simulation by real time prediction of the initial guess for the Newton-Raphson’s iterations, Computational Geosciences, 2024, V. 28, pp. 477–493, DOI: https://doi.org/10.1007/s10596-024-10284-z

12. Crespo A.C., Dominguez J.M., Barreiro A. et al., GPUs, a new tool of acceleration in CFD: efficiency and reliability on smoothed particle hydrodynamics methods, PLoS ONE, 2011, V. 6, no. 6, DOI: https://doi.org/10.1371/journal.pone.0020685

13. Feng C., Shen S., Liu H. et al., Toward cost-effective reservoir simulation solvers on GPUs, Advances in Applied Mathematics and Mechanics, 2016, V. 8, no. 6,

pp. 971–991, DOI: https://doi.org/10.4208/aamm.2014.m842

14. Temirchev P., Simonov M., Kostoev R. et al., Deep neural networks predicting oil movement in a development unit, Journal of Petroleum Science and Engineering, 2019, V. 180, pp. 1019–1029, DOI: https://doi.org/10.1016/j.petrol.2019.06.016

15. Takahashi T., Hamada T., GPU-accelerated boundary element method for Helmholtz’ equation in three dimensions, International Journal for Numerical Methods in Engineering, 2011, V. 80, no. 10, pp. 1295–1321, DOI: https://doi.org/10.1002/nme.2661

16. Yudin E.V., Gubanova A.E., Krasnov V.A., The method of express estimation of pore pressure map distribution in reservoirs with faults and wedging zones,

SPE-191582-18RPTC-MS, 2018, DOI: https://doi.org/10.2118/191582-18RPTC-MS

17. Yudin E.V., Modelirovanie fil’tratsii zhidkosti v neodnorodnykh sredakh dlya analiza i planirovaniya razrabotki neftyanykh mestorozhdeniy (Modeling of fluid filtration in heterogeneous environments for the analysis and planning of oilfield development): candidate of physical and mathematical sciences, Moscow, 2014.

18. Yudin E. V., Gubanova A.E., Krasnov V.A., Method for estimating the wells interference using field performance data (In Russ.), Neftyanoe khozyaystvo = Oil Industry, 2018, no. 8, pp. 64–69, DOI: https://doi.org/10.24887/0028-2448-2018-8-64-69