Data-driven Simulation of a Hypothetical Oil Field and Comparison of Results with Numerical Simulation

Document Type : Research Paper

Authors

Faculty of Chemical and Petroleum Engineering, Sharif university of Technology, Tehran, Iran

10.22078/pr.2025.5524.3456

Abstract

Data-driven reservoir simulation is a novel approach in reservoir modeling that complements or even replaces traditional numerical modeling methods. This method, also known as top-down modeling, utilizes artificial intelligence, specifically deep learning, and due to the nature of these tools, it requires data obtained from field measurements in the oil and gas industry, both for wells and reservoirs. Traditional numerical methods perform the simulation process using numerical modeling and the current understanding of the physics governing fluid flow in porous media. In this research, efforts have been made to model a hypothetical field using PETREL software, and then use the output data from this software as input for the desired top-down modeling using the Python programming language. After constructing three data-driven models for average reservoir pressure, average water saturation in the reservoir, and gas production from each well, the performance of the constructed models will finally be evaluated using machine learning. Moreover, the outputs of the data-driven model can include all that is expected in numerical models; however, in this research, only the mentioned outputs have been considered. In other words, the goal is for the data-driven model to understand the physics governing fluid flow in the porous medium through the measured data. In addition, during this process, history matching has been performed using the production data from the first two years of the field. In addition, the R2 score for the average reservoir pressure model, the average field water saturation, and gas production has been calculated as 0.9802, 0.97, and 0.987, respectively. Ultimately, a forecast for the third year will also be made, and the results obtained from the data-driven models will be compared with the results obtained by the numerical simulator.

Keywords

Main Subjects

References

[1]. Ansari, A. (2023). Reservoir simulation of the volve oil field using AI-based top-down modeling approach (Doctoral dissertation, West Virginia University). doi.org/10.33915/etd.11970.##
[2]. Sayarpour, M., Zuluaga, E., Kabir, C., & Lake, L. W. (2009). The use of capacitance–resistance models for rapid estimation of waterflood performance and optimization. Journal of Petroleum Science and Engineering, 69(3–4), 227–238. https://doi.org/10.1016/j.petrol.2009.09.006.##
[3]. “Equinor, Disclosing all Volve data,” 2018. [Online]. Available: https://www.equinor.com/news/archive/14jun2018-disclosing-volve-data.##
[4]. Mohaghegh, S. D. (2011). Reservoir simulation and modeling based on pattern recognition. All Days. doi.org/10.2118/143179-ms.##
[5]. Faisal, A., & Mohaghegh, S. D. (2016). A data-driven smart proxy model for a comprehensive reservoir simulation.
[6]. Schuld, M., & Petruccione, F. (2018). Supervised learning with quantum computers. In Quantum science and technology. doi.org/10.1007/978-3-319-96424-9.##
[7]. Donald J. Ford, P. (2011). Training industry. Retrieved from How the Brain Learns: trainingindustry.com/articles/content-development/how-the-brain-learns/.##
[8]. Shahkarami, A., Mohaghegh, S. D., Gholami, V., & Haghighat, S. A. (2014). Artificial intelligence (AI) assiste d history matching. All Days. doi.org/10.2118/169507-ms.##
[9]. Haykin, S. S. (2010). Neural networks and learning machines. ci.nii.ac.jp/ncid/BB00465945.##
[10]. Hernandez, A. (2016). Model calibration with neural networks. SSRN Electronic Journal. doi.org/10.2139/ssrn.2812140.##
[11]. Mohaghegh, S. D. (2017b). Shale Analytics. In Springer eBooks. doi.org/10.1007/978-3-319-48753-3.##
[12]. Mohaghegh, S. D. (2017). Data-Driven reservoir modeling. In Society of Petroleum EngineersRichardson, Texas, USA eBooks. doi.org/10.2118/9781613995600.##
[13]. Mohaghegh, S. D., Gaskari, R. ., Maysami, M. ., & Khazaeni, Y. (2014). Data-Driven reservoir management of a giant mature oilfield in the Middle East. All Days. doi.org/10.2118/170660-ms.##
[14]. Haifi, A. H. M. A. (2019). Confirmation of data-driven reservoir modeling using numerical reservoir simulation. In Partial Fulfillment of the Requirements for the Degree of Master of Science in Petroleum and Natural Gas Engineering doi.org/10.33915/etd.3835.##
[15]. Gomez, Y., Khazaeni, Y., Mohaghegh, S. D., & Gaskari, R. (2009). Top-Down intelligent reservoir modeling (TDIRM). All Days. doi.org/10.2118/124204-ms.##
[16]. Haghighat, S. A., Mohaghegh, S. D., Gholami, V., & Moreno, D. (2014). Production analysis of a Niobrara field using intelligent Top-Down modeling. All Days. doi.org/10.2118/169573-ms.##
[17]. Maysami, M., Gaskari, R., & Mohaghegh, S. D. (2013, September). Data driven analytics in powder river basin, WY. In SPE Annual Technical Conference and Exhibition? (p. D031S033R002). SPE.##
[18]. Alatrach, Y., Saputelli, L., Narayanan, R., Mohan, R., Alklih, M. Y., & Rubio, E. (2019, November). Data-driven vs. traditional reservoir numerical models: A case study comparison of applicability, practicality and performance. In Abu Dhabi International Petroleum Exhibition and Conference (p. D021S030R002). SPE.##
[19]. Zargari, S. & Mohaghegh, S. D. (2010). Field development Strategies for bakken shale Formation. All Days. doi.org/10.2118/139032-ms.##
[20]. Mohaghegh, S. D., Gruic, O., Zargari, S., Dahaghi, A. K., & Bromhal, G. S. (2012). Top-down, intelligent reservoir modelling of oil and gas producing shale reservoirs: case studies. International Journal of Oil Gas and Coal Technology, 5(1), 3. doi.org/10.1504/ijogct.2012.044175.##
[21]. Martinez, Y. A. (2020). Top-Down model development using data generated from a complex numerical reservoir simulation with water injection. In Partial Fulfillment of the Requirements for the Degree of Master of Science in Petroleum and Natural Gas Engineerin doi.org/10.33915/etd.7571.##
Journal of Petroleum Research
Volume 35, 1404-3 - Serial Number 142
September and October 2025
Pages 44-57

Files

Share

How to cite

Statistics