Vol. 337 No. 4 (2026)
DOI https://doi.org/10.18799/24131830/2026/4/5689
Creation of an algorithm for probabilistic forecasting of development indicators of the BV8 oil and gas condensate field using machine learning based on historical data on the operation of production wells
Relevance. Related to forecasting the parameters of oil field development based on historical data. One of the effective technological solutions in this area is the modification of displacement characteristic models through the use of machine learning algorithms based on Bayesian probability theory in combination with stochastic Monte Carlo modeling. The use of digital technologies in the mining sector has increased dramatically in recent years, which will help reduce production costs and allow for the efficient development of previously recognized unprofitable subsoil areas. Aim. To develop a machine learning-based algorithm capable of predicting key oil field development parameters based on historical data with an accuracy exceeding those values in currently used methods. Objects. Set of historical data on the operation of production wells of a large oil and gas field, including daily fluid flow rate, changes in water cut and average weighted reservoir pressure in the zone under consideration. Methods. General scientific methods (analysis, generalization, synthesis, classification) and specific scientific ones (mathematical modeling, software modeling). The set and combination of these methods are adequate to the goals and objectives, object and subject of research of this work. Results. The paper considers an algorithm for constructing calculations based on integral displacement characteristics using wells no. 1133, 1165, 1221 and 1328 with different water cuts as an example. The best method for all wells of any water cut category is the G.S. Kambarov and A.M. Pirverdyan group of methods, since its relative error has the smallest range – from 1.2 to 5.6%. Next, the theoretical foundations and mechanism of the future program based on Bayesian networks with Markov chains are formed. The created program in the Python programming language using the BAMT library made it possible to reduce the relative forecast error to 0.63%. The scientific novelty of the work lies in the development of a unique algorithm for forecasting oil production at a well, which includes the use of currently developing machine learning. The use of the proposed algorithm will reduce the time for assessing the accumulated indicators and minimize the calculation error.
For citation: Arutyunyan A.S. Creation of an algorithm for probabilistic forecasting the development indicators of the BV8 oil and gas condensate field using machine learning based on historical data on the operation of production wells. Bulletin of the Tomsk Polytechnic University. Geo Assets Engineering, 2026, vol. 337, no. 4, pp. 152-167.
Keywords:
forecast of oil field development indicators based on historical data, analysis of the production decline curve using the Arps equation, integral models of displacement characteristics, forecast algorithm using models of displacement characteristics, relative errors of integral displacement characteristics, Monte-Carlo modeling with the construction of Markov chains, algorithm for forecasting development indicators
References:
1. Bovkonyuk P.A., Savenok O.V. Development of a program algorithm for the automated selection of oil and gas analogue fields. Oil Engineer, 2023, no. 3, pp. 20–24. (In Russ.)
2. Koshelev A.T., Solovyova V.N., Orlova I.O., Datsenko E.N. The problem of prospecting for fields-analogues and the method applied for its salvation. Construction of oil and gas wells on land and at sea, 2014, no. 7, pp. 10–12. (In Russ.)
3. Nigmatullin F.N., Ponomarev A.I. On solving problematic issues when designing the development of new gas-oil and oil-gas deposits in conditions of insufficient data. Oil industry, 2024, no. 6, pp. 73–77. (In Russ.) DOI: 10.24887/0028-2448-2024-6-73-77
4. Shumko V.S., Mamchistova E.I., Kuzovlev S.S. Estimating recoverable oil reserves using integral displacement characteristics based on probabilistic methodology. News of higher educational institutions. Oil and gas, 2021, no. 2, pp. 78–88. (In Russ.) DOI: 10.31660/0445-0108-2021-2-78-88
5. Paryani M., Ahmadi M., Awoleke O., Hanks C. Decline curve analysis: a comparative study of proposed models using improved residual functions. Journal of petroleum & environmental biotechnology, 2018, vol. 9, Iss. 1, pp. 1–8. DОI: 10.4172/2157-7463. 1000362
6. Spivey J.P., Gatens J.M., Semmelbeck M.E., Lee W.J. Integral type curves for advanced decline curve analysis. SPE Mid-Continent Gas Symposium. Amarillo, Texas, 13–14 April, 1992. pp. 91. DOI: 10.2118/24301-MS
7. Ma X., Liu Zh. Predicting the oil production using the novel multivariate nonlinear model based on Arps decline model and kernel method. Neural Computing and Applications, 2016, vol. 29, Iss. 2, pp. 579–591. DOI: 10.1007/s00521-016-2721-x
8. Maurenza F., Yasutra A., Tungkup I.L. Production forecasting using Arps decline curve model with the effect of artificial lift installation. Scientific Contributions Oil and Gas, 2023, vol. 46, Iss. 1, pp. 9–18. DOI: 10.29017/scog.46.1.1310
9. Petuhov A.V., Roshchin P.V. Addendum to Arps law for estimating recoverable oil reserves in conventional and unconventional reservoirs. Bulletin of Eurasian Science, 2023, vol. 15, no. 4, serial number: 28. (In Russ.)
10. Arps J.J. Analysis of decline curves. Transactions of the AIME, 1945, vol. 160 (01), pp. 228–247. DOI: 10.2118/945228-g
11. Nazarov S.N., Sipachev N.V. Methodology for forecasting technological indicators at a late stage of oil deposit development. News of higher educational institutions. Oil and gas, 1972, no. 10, pp. 41–45. (In Russ.)
12. Sipachev N.V., Posevich A.G. On the characteristics of oil displacement by water. News of higher educational institutions. Oil and gas, 1980, no. 12, pp. 26–32. (In Russ.)
13. Gaisin D.K., Timashev E.M. Evaluation of recoverable reserves under conditions of oil displacement by water at a late stage of development. Proceedings of BashNIPIneft, 1985, iss. 73. (In Russ.)
14. Maksimov M.I. Method for calculating recoverable oil reserves at the final stage of exploitation of oil formations under conditions of oil displacement by water. Geology of oil and gas, 1959, no. 3, pp. 42–47. (In Russ.)
15. Abyzbaev I.I. Features of development of oil deposits at a late stage. Oil industry, 1978, no. 9, pp. 27–30. (In Russ.)
16. Sazonov B.F. Methodology for forecasting oil recovery in designing the development of oil fields. Proceedings of Giprovostokneft, 1969, Iss. 12, pp. 104–109. (In Russ.)
17. Pirverdyan A.M., Nikitin P.I., Listengarten L.B., Danelyan M.G. On the forecast of oil and associated water production during the development of layered heterogeneous reservoirs. Azerbaijan Oil Industry, 1970, no. 11, pp. 19–22. (In Russ.)
18. Kambarov G.S., Almamedov D.G., Makhmudova T.Yu. On the determination of the initial recoverable reserve of an oil field. Azerbaijan Oil Industry, 1974, no. 3, pp. 22–24. (In Russ.)
19. Kazakov A.A. Forecasting field development indicators based on the characteristics of oil displacement by water. Oil Field Business, 1976, no. 8, pp. 5–7. (In Russ.)
20. Zhang R., Jia H. Production performance forecasting method based on multivariate time series and vector autoregressive machine learning model for waterflooding reservoirs. Petroleum Exploration and Development, 2021, vol. 48, Iss. 1, pp. 201–211. DOI: 10.1016/s1876-3804(21)60016-2
21. Khamehchi E., Bemani A. Prediction of pressure in different two-phase flow conditions: Machine learning applications. Measurement: Journal of the International Measurement Confederation, 2021, vol. 173, pp. 108665. DOI: 10.1016/j.measurement.2020.108665
22. Pal M. On application of machine learning method for history matching and forecasting of times series data from hydrocarbon recovery process using water flooding. Petroleum Science and Technology, 2021, vol. 39, pp. 519–549. DOI: 10.1080/10916466.2021.1918712
23. Dachanuwattana S., Jin J., Zuloaga-Molero P., Li X., Xu Y., Sepehrnoori K., Miao J. Application of proxy-based MCMC and EDFM to history match a Vaca Muerta shale oil well. Fuel, 2018, vol. 220, pp. 490–502. DOI: 10.1016/j.fuel.2018.02.018
24. Mayet A.M., Alizadeh S.M., Nurgalieva K.S., Hanus R., Nazemi E., Narozhnyy I.M. Extraction of time-domain characteristics and selection of effective features using correlation analysis to increase the accuracy of petroleum fluid monitoring systems. Energies, 2022, vol. 15, Iss. 6, serial number: 1986. DOI: 10.3390/en15061986
25. Fursov I., Christie M., Lord G. Applying kriging proxies for Markov chain Monte Carlo in reservoir simulation. Computational Geosciences, 2020, vol. 24, Iss. 4, pp. 1725–1746. DOI: 10.1007/s10596-020-09968-z
26. Babeshko L.O. Methodical aspects of generalized diagnostics methods in vector autoregressive models: Jarque–Bera tests. Modern Mathematics and Concepts of Innovative Mathematical Education, 2020, vol. 7, Iss. 1, pp. 320–328. (In Russ.)
27. Upendra S. Residual analysis in simple linear regression: a practical application with the Jarque–Bera test. International Scientific Journal of Engineering and Management, 2023, pp. 1–12. DOI: 10.55041/ijsrem26596
