Вестник нефтегазовой отрасли КазахстанаВестник нефтегазовой отрасли Казахстана2707-42262957-806XKMG Engineering10901310.54859/kjogi109013Research ArticleMachine learning-assisted prediction of polymer flooding performance in heterogeneous carbonate reservoirs of the Pre-Caspian Basin: a gradient boosting and physics-informed approachZhetpisAdil A.adiljetpeace@gmail.comhttps://orcid.org/0009-0000-0390-5436RyndinVladimir V.ryndin.v@teachers.tou.edu.kzhttps://orcid.org/0000-0002-4248-9516KenzhanovaMensulu M.mena-1997@list.ruhttps://orcid.org/0009-0004-4696-6908OmarbekovaInara K.inara_19@mail.ruhttps://orcid.org/0009-0004-4292-6765AigozhinaDinara G.dina7041976@mail.ruhttps://orcid.org/0000-0002-5712-5280AbdullinaGulnara G.gulnara_1277@mail.ruhttps://orcid.org/0000-0003-4493-1715Toraighyrov University300920268331410906202625082026Copyright © 2026, Zhetpis A.A., Ryndin V.V., Kenzhanova M.M., Omarbekova I.K., Aigozhina D.G., Abdullina G.G.2026<p><strong>Background: </strong>Polymer flooding remains one of the most technically viable enhanced oil recovery methods for mature fields in Kazakhstan’s Pre-Caspian Basin. Predicting its sweep efficiency in heterogeneous carbonate reservoirs is challenging due to complex pore-throat geometry, high-salinity formation waters, and pronounced permeability contrasts. Conventional reservoir simulation is computationally prohibitive for real-time operational decisions and probabilistic uncertainty analysis.</p> <p><strong>Aim: </strong>This study develops and validates a hybrid surrogate framework integrating gradient boosting regression with physics-informed constraints derived from polymer transport theory to predict incremental oil recovery factor and injectivity ratio for polymer flooding in carbonate formations of the Pre-Caspian Basin.</p> <p><strong>Materials and Methods: </strong>A dataset of 412 polymer flooding operations was compiled from published literature, Society of Petroleum Engineers (SPE) technical reports, and internal field data from three <strong>Kazakhstani assets: </strong>Tengizchevroil, Karachaganak Petroleum Operating, and the Uzen field operated by Ozenmunaigas. Eighteen reservoir and fluid parameters were used as input features. Gradient boosting regression was trained using a stratified 80/20 split with five-fold cross-validation. Physics-informed penalty terms derived from Darcy-scale polymer transport equations were embedded in the loss function to ensure physical consistency. Model performance was benchmarked against artificial neural networks, support vector regression, and full compositional simulation.</p> <p><strong>Results: </strong>The physics-informed gradient boosting model achieved a coefficient of determination R² = 0.924 and root mean square error of 2.31% on the held-out test set for recovery factor prediction, outperforming artificial neural networks (R² = 0.891) and support vector regression (R² = 0.857). The physics penalty reduced physically inconsistent predictions by 78%. SHapley Additive exPlanations (SHAP) analysis identified the permeability variation coefficient, polymer concentration, and formation water salinity as dominant predictors. Surrogate estimates agreed with full compositional simulation within ±1.8% absolute for the Uzen carbonate pilot.</p> <p><strong>Conclusions:</strong> The proposed surrogate model provides rapid and reliable predictions of polymer flooding performance suitable for field-scale integration, substantially reducing computational burden for enhanced oil recovery screening and optimization in heterogeneous carbonate reservoirs of Kazakhstan.</p>Big polymer floodingmachine learninggradient boostingenhanced oil recoverycarbonate reservoirsPre-Caspian BasinSHAP analysisphysics-informed machine learningpermeability heterogeneityполимерлі суландырумашиналық оқытуградиенттік бустингмұнай өндіруді арттырукарбонатты қабаттарКаспий маңы ойпатыSHAP-талдауыфизикалық-ақпараттандырылған машиналық оқытуөткізгіштіктің біркелкісіздігіполимерное заводнениемашинное обучениеградиентный бустингповышение нефтеотдачикарбонатные коллекторыПрикаспийский бассейнSHAP-анализфизически-информированное машинное обучениенеоднородность проницаемости1.Zhanserkeyeva A.A., Kassenov A.K. Prospect evaluation based on integrated petroleum system analysis: Block E case study, south-eastern edge of Precaspian Basin (Kazakhstan) // J Pet Explor Prod Technol. 2022. Vol. 12. P. 2625–2642. doi: https://doi.org/10.1007/s13202-022-01466-52.Akhmetov R., Seitkali D., Bekova G. Water flooding performance analysis of mature Kazakhstani carbonate fields // J Pet Sci Eng. 2022. Vol. 208. doi: 10.1016/j.petrol.2021.109632.3.Sagyndikov M.S., Seright R.S., Tuyakov N. An Unconventional Approach to Model a Polymer Flood in the Kalamkas Oilfield // SPE Improved Oil Recovery Conference; Apr 25–29, 2022; Virtual. Available from: onepetro.org/SPEIOR/proceedings-abstract/22IOR/22IOR/D021S017R001/483984.4.Lake L.W., Johns R.T., Rossen W.R., Pope G.A. Fundamentals of Enhanced Oil Recovery. Tulsa : Society of Petroleum Engineers, 2014. 270 p.5.Sorbie K.S. Polymer-Improved Oil Recovery. Glasgow: Blackie Academic & Professional; 1991. 359 p.6.Wang D, Cheng J, Yang Q, et al. Viscous-Elastic Polymer Can Increase Microscale Displacement Efficiency in Cores // SPE Annual Technical Conference and Exhibition; Oct 1–4, 2000; Dallas, Texas, USA. Available from: onepetro.org/SPEATCE/proceedings-abstract/00ATCE/00ATCE/SPE-63227-MS/132143.7.Sheng J. Enhanced Oil Recovery Field Case Studies. Amsterdam : Elsevier, 2013. 685 p.8.Christie M.A., Blunt M.J. Tenth SPE comparative solution project: A comparison of upscaling techniques // SPE Reserv Eval Eng. 2001. Vol. 4, Issue 4. P. 308–317. doi: 10.2118/72469-PA.9.Al-Dousari MM, Garrouch AA. An artificial neural network model for predicting the recovery performance of surfactant polymer floods // J Pet Sci Eng. 2013. Vol. 109. P. 51–62. doi: 10.1016/j.petrol.2013.08.012.10.Mogensen K., Masalmeh S. A review of EOR techniques for carbonate reservoirs in challenging reservoir conditions // J Pet Sci Eng. 2020. Vol. 195. doi: 10.1016/j.petrol.2020.107889.11.Chen T., Guestrin C. XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining; Aug 13–17, 2016; San Francisco, California, USA. Available from: dl.acm.org/doi/proceedings/10.1145/2939672.12.Lundberg S.M., Lee S.I. A unified approach to interpreting model predictions // Adv Neural Inf Process Syst. 2017. Vol. 30. P. 4765–4774. doi: 10.48550/arXiv.1705.07874.13.Zhu D., Bai B., Hou J. Polymer Gel Systems for Water Management in High-Temperature Petroleum Reservoirs: A Chemical Review // Energy & Fuels. 2017. Vol. 31, Issue 12. P. 13063–13087. doi: 10.1021/acs.energyfuels.7b02897.14.Raissi M., Perdikaris P., Karniadakis G.E. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations // J Comput Phys. 2019. Vol. 378. P. 686–707. doi: 10.1016/j.jcp.2018.10.045.15.Karniadakis G.E., Kevrekidis I.G., Lu L., et al. Physics-informed machine learning // Nat Rev Phys. 2021.Vol. 3. P. 422–440. doi: 10.1038/s42254-021-00314-5.16.Dake LP. Fundamentals of Reservoir Engineering. Amsterdam : Elsevier, 1978. 498 p.17.Dykstra H., Parsons R.L. The prediction of oil recovery by waterflood. In: Secondary Recovery of Oil in the United States. <math><msup><mn>2</mn><mi>nd</mi></msup></math> ed. Washington, DC : API, 1950. P. 160–174.18.Stiles W.E. Use of permeability distribution in water flood calculations // J Pet Technol. 1949. Vol. 1, Issue 01. P. 9–13. doi: 10.2118/949009-G.19.Goudarzi A., Delshad M., Sepehrnoori K. A chemical EOR benchmark study of different reservoir simulators // Comput Geosci. 2016.Vol. 94. P. 96–109. doi: 10.1016/j.cageo.2016.06.013.20.Shakeel M., Sagandykova D., Mukhtarov A., et al. Maximizing oil recovery: Innovative chemical EOR solutions for residual oil mobilization in Kazakhstan’s waterflooded sandstone oilfield // Heliyon. 2024. Vol. 10, Issue 7. doi: 10.1016/j.heliyon.2024.e28915.