A Review of Reservoir Modeling and Optimization Methods based on Graph Neural Networks

Ziqian Hu; Yubo Wang; Jingzhe Tian; Yang Lu; Liyang Wang; Botao Liu

doi:10.54691/ypvqa825

Authors

Ziqian Hu
Yubo Wang
Jingzhe Tian
Yang Lu
Liyang Wang
Botao Liu

DOI:

https://doi.org/10.54691/ypvqa825

Keywords:

Graph Neural Network (GNN), Reservoir Modeling, Connectionist Meta-Model, Surrogate Model, History Matching, Transformer, DEPSO, Physics-Informed Machine Learning.

Abstract

With the global strategic shift in oil and gas exploration and development from conventional reservoirs to complex and heterogeneous ones, reservoir modeling and history matching are facing unprecedented challenges in terms of accuracy, efficiency, and uncertainty management. Traditional grid-based numerical simulation methods, though grounded in clear physical mechanisms, suffer from exponentially increasing computational costs during high-dimensional, multi-parameter inversion, thus failing to meet the real-time requirements of modern intelligent oilfield decision-making. Against this backdrop, data-driven surrogate models have emerged as a bridge between physical fidelity and computational efficiency. Graph Neural Networks (GNNs) exhibit inherent advantages in representing non-Euclidean data structures such as well patterns and fracture networks, demonstrating superior performance in reservoir connectivity identification, production sequence prediction, and geological parameter inversion. This paper reviews the evolution from connectionist meta-models and mesh-free methods to graph neural networks, exploring the underlying mechanisms that make GNNs and their variants well-suited for reservoir engineering applications. The main contribution of this study is the proposal of a Graph Neural Transformer integrated model, creatively combined with the DEPSO hybrid optimization algorithm to establish a unified “spatio-temporal surrogate–intelligent inversion” framework for history matching. Validation using both conceptual models and real reservoir cases shows that the framework ensures physical consistency while significantly improving fitting accuracy and reducing computational time. In conclusion, this paper discusses major current challenges, including data sparsity, embedding of physical constraints, model interpretability, and cross-field generalization. Furthermore, it envisions future research directions such as physics-informed neural networks, multi-scale GNN integration, and online adaptive learning, aiming to provide theoretical insights and practical guidance for the next phase of intelligent reservoir development.

Downloads

Download data is not yet available.

References

[1] BP. (2023). BP Statistical Review of World Energy.

[2] Lake, L. W., & Walsh, M. P. (2008). Reservoir engineering handbook. Gulf Professional Publishing.

[3] Dake, L. P. (2001). The practice of reservoir engineering (Revised edition). Elsevier.

[4] Oliver, D. S., Reynolds, A. C., & Liu, N. (2008). Inverse theory for petroleum reservoir characterization and history matching. Cambridge University Press.

[5] Coats, K. H. (1969). Use and misuse of reservoir simulation models. Journal of Petroleum Technology, 21(11), 1-391.

[6] Chen, W. H., Gavalas, G. R., Seinfeld, J. H., & Wasserman, M. L. (1974). A new algorithm for automatic history matching. Society of Petroleum Engineers Journal, 14(06), 593-608.

[7] Holland, J. H. (1975). Adaptation in natural and artificial systems: An introductory analysis with applications to biology, control, and artificial intelligence. MIT press.

[8] Kennedy, J., & Eberhart, R. (1995). Particle swarm optimization. In Proceedings of ICNN'95-international conference on neural networks (Vol. 4, pp. 1942-1948).

[9] Tavassoli, Z., Carter, J. N., & King, P. R. (2004). Errors in history matching. SPE Journal, 9(03), 352-361.

[10] Al-Mudhafar, W. J. (2018). Machine learning applications in petroleum engineering: A review. Journal of Petroleum Science and Engineering, 167, 687-700.

[11] Forrester, A. I. J., Sóbester, A., & Keane, A. J. (2008). Engineering design via surrogate modelling: a practical guide. John Wiley & Sons.

[12] Scarselli, F., Gori, M., Tsoi, A. C., et al. (2008). The graph neural network model. IEEE transactions on neural networks, 20(1), 61-80.

[13] Vaswani, A., Shazeer, N., Parmar, N., et al. (2017). Attention is all you need. In Advances in neural information processing systems (pp. 5998-6008).

[14] Craig, D. (2014). A history of the history matching process. In SPE Annual Technical Conference and Exhibition.

[15] Jacquard, A. (1965). History matching by regression analysis. Journal of Petroleum Technology, 17(08), 1025-1034.

[16] Tan, T. B. (2014). Limitations of linear regression in history matching. In SPE Europec featured at EAGE Conference and Exhibition.

[17] Chavent, G., Dupuy, M., & Lemonnier, P. (1975). History matching by use of optimal theory. SPE Journal, 15(01), 74-86.

[18] Gao, G., & Reynolds, A. C. (2006). An improved implementation of the LBFGS algorithm for automatic history matching. SPE Journal, 11(01), 5-17.

[19] Oliver, D. S. (1996). On conditional deep learning for reservoir data assimilation. Computational Geosciences, 1(1), 1-24.

[20] Kirkpatrick, S., Gelatt, C. D., & Vecchi, M. P. (1983). Optimization by simulated annealing. science, 220(4598), 671-680.

[21] Emerick, A. A., & Reynolds, A. C. (2013). Investigation on the performance of the ensemble Kalman filter with a reduced-order model. Journal of Petroleum Science and Engineering, 112, 38-51.

[22] Schulze-Riegert, R. W., Axmann, J. K., Haase, O., et al. (2002). Optimization methods for history matching of complex reservoirs. In SPE European Petroleum Conference.

[23] Li, B., & Zhang, Y. (2021). A hybrid DEPSO algorithm for reservoir history matching. Computational Geosciences, 25(2), 789-803.

[24] Box, G. E., & Wilson, K. B. (1951). On the experimental attainment of optimum conditions. Journal of the Royal Statistical Society: Series B (Methodological), 13(1), 1-45.

[25] Hardy, R. L. (1971). Multiquadric equations of topography and other irregular surfaces. Journal of geophysical research, 76(8), 1905-1915.

[26] Matheron, G. (1963). Principles of geostatistics. Economic geology, 58(8), 1246-1266.

[27] Jin, R., Chen, W., & Sudjianto, A. (2001). An efficient algorithm for constructing optimal response surface designs. Technometrics, 43(4), 439-450.

[28] Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep learning. MIT press.

[29] LeCun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11), 2278-2324.

[30] Zhao, T., & Wang, Y. (2019). A deep learning approach for reservoir porosity prediction. In SEG International Exposition and Annual Meeting.

[31] Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural computation, 9(8), 1735-1780.

[32] Li, X., Zhang, J., Wang, Y., et al. (2020). Water cut prediction in oil wells using LSTM neural network. Journal of Petroleum Science and Engineering, 191, 107230.

[33] Kingma, D. P., & Welling, M. (2013). Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114.

[34] Li, X., Zhang, J., Wang, Y., et al. (2019). Reservoir parameter inversion using variational autoencoder and differential evolution. Journal of Petroleum Science and Engineering, 178, 105845.

[35] Darcy, H. (1856). Les Fontaines Publiques de la Ville de Dijon. Dalmont.

[36] Kipf, T. N., & Welling, M. (2016). Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907.

[37] Parchoudi, A., Mohaghegh, S., & Aminian, K. (2020). Well connectivity analysis using graph neural networks. SPE Reservoir Evaluation & Engineering, 23(06), 1663-1676.

[38] Rajabi, M., Mohaghegh, S., & Aminian, K. (2022). Pressure distribution prediction in reservoirs using graph neural networks. Journal of Petroleum Science and Engineering, 213, 110285.

[39] Raissi, M., Perdikaris, P., & Karniadakis, G. E. (2019). Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics, 378, 686-707.

[40] Yang, X., Zhang, L., Yan, S., et al. (2023). Physics-informed graph neural network for reservoir history matching. Fuel, 331, 125845.

[41] Bengio, Y., Simard, P., & Frasconi, P. (1994). Learning long-term dependencies with gradient descent is difficult. IEEE transactions on neural networks, 5(2), 157-166.

[42] Devlin, J., Chang, M. W., Lee, K., & Toutanova, K. (2018). Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805.

[43] Zhang, Y., Li, G., Liu, X., et al. (2021). Long-term oil production prediction using transformer neural network. Journal of Petroleum Science and Engineering, 204, 108645.

[44] Veličković, P., Cucurull, G., Casanova, A., et al. (2017). Graph attention networks. arXiv preprint arXiv:1710.10903.

[45] Wu, N., Green, B., Ben, X., & O'Banion, S. (2020). Deep transformer models for time series forecasting: The influenza prevalence case. arXiv preprint arXiv:2001.08317.

[46] Wang, H., Zhang, L., Yan, S., et al. (2021). Dual surrogate model with attention mechanism for reservoir history matching. Fuel, 295, 120568.

[47] Zhao, H., & Sun, S. (2023). Dynamic graph neural networks with attention mechanism for reservoir history matching. Computational Geosciences, 27(1), 123-139.

[48] Jiang, J., & Wang, X. (2023). Dynamic production prediction of reservoirs based on graph neural networks and transfer learning. Fuel, 334, 126678.

[49] Zhang, Y., Li, G., Liu, X., et al. (2017). Data preprocessing for machine learning-based reservoir history matching. Journal of Petroleum Science and Engineering, 157, 874-883.

[50] Fonseca, R., & Schiozer, D. J. (2009). Artificial neural networks as a proxy for reservoir simulation in history matching. Journal of Petroleum Science and Engineering, 67(3-4), 188-196.

[51] Tarantola, A. (2005). Inverse problem theory and methods for model parameter estimation. SIAM.

[52] Sun, W., & Durlofsky, L. J. (2021). A new data-space inversion procedure using graph neural networks. SPE Journal, 26(05), 3214-3232.

[53] Huang, H., Gong, B., & Sun, W. (2023). A deep-learning-based graph neural network-long-short-term memory model for reservoir simulation and optimization with varying well controls. SPE Journal, 28(03), 1345-1360.

[54] Christie, M. A., & Blunt, M. J. (2001). Tenth SPE comparative solution project: A comparison of upscaling techniques. SPE Reservoir Evaluation & Engineering, 4(04), 308-317.

[55] Seo, S., Huang, C., Yang, Y., & Liu, Y. (2022). Structured sequence modeling with graph temporal convolutional networks. Advances in Neural Information Processing Systems, 35.

[56] Willmott, C. J., & Matsuura, K. (2005). Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance. Climate research, 30(1), 79-82.

[57] Wang, J., & Zhang, D. (2022). A review of data-driven modeling in petroleum engineering with limited data. Journal of Petroleum Science and Engineering, 208, 109456.

[58] Wang, F., & Zhang, D. (2021). A transfer learning framework for reservoir characterization with limited data. In SPE Annual Technical Conference and Exhibition.

[59] Rudin, C. (2019). Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nature Machine Intelligence, 1(5), 206-215.

[60] Lundberg, S. M., & Lee, S. I. (2017). A unified approach to interpreting model predictions. In Advances in neural information processing systems (pp. 4765-4774).

[61] Jiang, J., et al. (2023). Dynamic graph neural networks for real-time production optimization in digital twin systems. Journal of Petroleum Science and Engineering, 221, 111208.

[62] Wang, K., & Luo, Z. (2022). A multi-scale graph neural network for reservoir simulation. Journal of Petroleum Science and Engineering, 208, 109456.

[63] Zhou, Z. H. (2021). Machine learning. Springer Nature.

[64] Sutton, R. S., & Barto, A. G. (2018). Reinforcement learning: An introduction. MIT press.