Application of Machine Learning Algorithms for Enhanced Modeling of a Carbonate Reef Reservoir

Author details
  • Perm National Research Polytechnic University, Perm, Russian Federation
  • Perm National Research Polytechnic University, Perm, Russian Federation
  • Perm National Research Polytechnic University, Perm, Russian Federation
  • Perm National Research Polytechnic University, Perm, Russian Federation
  • Perm National Research Polytechnic University, Perm, Russian Federation
  • Perm National Research Polytechnic University, Perm, Russian Federation
  • Perm National Research Polytechnic University, Perm, Russian Federation

Abstract

The development and modeling of carbonate reservoirs with complex structure is an actual task. In conditions of high heterogeneity of reservoir properties due to the peculiarities of formation, frequent change of sedimentation cycles and the presence of diagenetic transformations, there is a high degree of uncertainty in the modeling process and, as a consequence, in the forecast of development indicators. Underestimation of the influence of diagenetic processes on changes in filtration-capacitance properties can have a critical impact on the organization and management of the waterflood system. When studying the geological structure of the Alpha field, intervals of highly permeable reservoirs (up to 18 Darcies) were identified. This study proposes an approach to identify such intervals in order to refine the dynamic model of the field based on machine learning methods. The paper compares the following algorithms: gradient bousting, random forest and support vector method. Based on the results of the study, the optimal algorithms were identified that allow predicting high permeability intervals with a high degree of accuracy. To improve model adaptation to the field development history, it is suggested to use a model trained on core and geophysical well survey data. To take into account the risks associated with highly permeable intervals, when drilling new wells, it is recommended to use a model trained only on geophysical well tests. In this paper, sensitivity analysis was performed when specifying properties for highly permeable intervals - absolute permeability and relative phase permeabilities. The permeability cube of the dynamic model was updated, the model was adapted and calculations on waterflood system optimization were performed. Based on the predictive analysis on the model with highly permeable intervals, a number of measures were proposed to optimize the development system to reduce the risk of water breakthrough in highly permeable intervals. According to the results of the forecast calculations, these measures will provide additional 750 thousand m3 of oil.

Keywords

carbonate reservoirwaterflood systemhigh-permeability intervalsmachine learningrandom forestgradient boustingsupport vector methodsensitivity analysis

Bibliography

  1. 1.Janjuhah H.T., Salim A.M.A., Ghosh D., Wahid A. Diagenetic process and their effect on reservoir quality in Miocene carbonate reservoir Offshore, Sarawak, Malaysia. In Proceedings of the International Conference on Integrated Petroleum Engineering and Geosciences. Singapore, 2016. DOI: 10.1007/978-981-10-3650-7_48 Scholar
  2. 2.Farzaneh S.A., Dehghan A.A., Lakzaie A. Using diagenetic processes in facies modeling of a carbonate reservoir. Energy Sources, Part A: Recovery, Utilization, and Environmental Effects, 2013, no. 35 (6), pp. 516-528. DOI: 10.1080/15567036.2010.504946 Scholar
  3. 3.Riabokon E., Turbakov M., Kozhevnikov E., Poplygin V., Guzev M. Study of the influence of nonlinear dynamic loads on elastic modulus of carbonate reservoir rocks. Energies, 2021, vol. 14, no. 24. DOI 10.3390/en14248559 Scholar
  4. 4.Huang Q., Liu W., Zhang Y., Shi S., Wang K. Progress of research on dolomitization and dolomite reservoir. Advances in Earth Science, 2015, no. 30 (5), 539 p. DOI: 10.11867/j.issn.1001-8166.2015.05.0539 Scholar
  5. 5.Xu Z.X., Li S.Y., Li B.F., Chen D.Q., Liu Z.Y., Li Z.M. A review of development methods and EOR technologies for carbonate reservoirs. Petroleum Science, 2020, no. 17, pp. 990-1013. DOI: 10.1007/s12182-020-00467-5. Scholar
  6. 6.Shcherbakov A.A., Khizhniak G.P., Galkin V.I. Otsenka effektivnosti meropriiatii po intensifikatsii dobychi nefti (na primere mestorozhdenii Solikamskoi depressii) [Effectiveness evaluation of oil production stimulation measures (on the example of the Solikamsk depression fields)]. Geologiia, geofizika i razrabotka neftianykh i gazovykh mestorozhdenii, 2019, no. 2, pp. 70-73. DOI: 10.30713/2413-5011-2019-2-70-73 Scholar
  7. 7.Erofeev A.A., Ponomareva I.N., Turbakov M.S. Otsenka uslovii primeneniia metodov obrabotki krivykh vosstanovleniia davleniia skvazhin v karbonatnykh kollektorakh [Evaluation of the application conditions for methods of processing well pressure recovery curves in carbonate reservoirs]. Inzhener-neftianik, 2011, no. 3, pp. 12-15. Scholar
  8. 8.Yang B., He J., Lyu D., Tang H., Zhang J., Li X., Zhao J. Production optimization for water flooding in fractured-vuggy carbonate reservoir-From laboratory physical model to reservoir operation. Journal of Petroleum Science and Engineering, 2020, no. 184, 106520 p. DOI: 10.1016/j.petrol.2019.106520 Scholar
  9. 9.Zhao M., Xin X., Yu G., Hu R., Gong Y. Mechanistic Study and Optimization of High Water Cut Phase Management Strategies in Fracture-Vuggy Carbonate Reservoirs with Bottom Water. Processes, 2023, no. 11, 3135 p. DOI: 10.3390/pr11113135 Scholar
  10. 10.Mahmood S., Salazar P., Zhao X., Pointing M., Sayed A. Waterflooding in giant carbonate reservoir; successes and challenges. In Proceedings of the Abu Dhabi International Petroleum Exhibition and Conference. Abu Dhabi, UAE, 2017. DOI: 10.2118/188532-MS Scholar
Download PDF

Article information

Published
24 February 2025
Issue date
24 February 2025
Issue
Vol. 25, No. 1 (2025)
Share