«Geology, geophysics and development of oil and gas fields»

Application of machine learning methods for porosity assessment of carbonate reservoir rocks

UDC: 550.8.013:622.276.031.011.431.2
DOI: 10.33285/2413-5011-2023-11(383)-56-61

Authors:

¹ Perm National Research Polytechnic University, Perm, Russia

Keywords: machine learning, well logging, porosity, forecasting methods, random forest, gradient boosting, linear regression

Annotation:

Carbonate reservoirs are characterized by a complex structure, cyclic sedimentation, uneven development of secondary processes, which causes their high heterogeneity of properties. The standard approach to determining reservoir properties along a wellbore based on well logging methods does not always allow a reliable forecast. The article presents an approach based on the use of machine learning methods for predicting the porosity of a carbonate reservoir. Rock porosity in this case is determined not by one well logging method, but by a complex of different methods. The study identified the parameters that have the greatest impact on the porosity parameter. Based on these parameters, porosity was predicted by various machine learning methods that solve regression problems. Metrics for assessing the algorithms quality are calculated and compared with each other. Recommendations and conclusions about the possibility and rationality of applying these models in practice are presented.

Bibliography:

1. Izuchenie neodnorodnosti i anizotropii pronitsaemosti slozhno postroennogo karbonatnogo rezervuara / S.V. Galkin, A.A. Kochnev, S.N. Krivoshchekov [i dr.] // Geologiya, geofizika i razrabotka neftyanykh i gazovykh mestorozhdeniy. – 2023. – № 3(375). – S. 25–32. – DOI: 10.33285/2413-5011-2023-3(375)-25-32

2. Utochnenie kuba pronitsaemosti geologo-gidrodinamicheskikh modeley v usloviyakh razlichnogo ob"ema iskhodnykh dannykh / N.D. Kozyrev [i dr.] // Geologiya, geofizika i razrabotka neftyanykh i gazovykh mestorozhdeniy. – 2021. – № 5. – S. 24–29. – DOI: 10.33285/2413-5011-2021-5(353)-24-29

3. Shiverskiy G.V., Krivoshchekov S.N. Perspektivy primeneniya metodov iskusstvennogo intellekta v neftegazovoy geologii // Master’s Journal. – 2022. – № 2. – S. 57–67. – EDN MXLKEW

4. Primenenie metodov mashinnogo obucheniya dlya poiska propushchennykh produktivnykh intervalov i prognoza "skrytykh" petrofizicheskikh svoystv / Yu.D. Kantemirov [i dr.] // Rosgeologiya. V poiskakh novykh otkrytiy. – 2019. – S. 24–26.

5. Ispol’zovanie neyrosetevykh algoritmov pri prognoze petrofizicheskikh svoystv tonkosloistogo razreza po dannym seysmorazvedki i GIS (na primere achimovskikh otlozheniy mestorozhdeniya v YaNAO) / R.B. Yanevits [i dr.] // Geologiya, geofizika i razrabotka neftyanykh i gazovykh mestorozhdeniy. – M.: OAO "VNIIOENG", 2017. – № 7. – S. 40–54.

6. Feng R., Grana D., Balling N. Variational inference in Bayesian neural network for well-log prediction // Geophysics. – 2021. – Vol. 86. – № 3. – Pp. M91-M99. – DOI: 10.1190/geo2020-0609.1

7. The implementation of machine learning in lithofacies classification using multi well logs data / S. Saroji [et al.] // Aceh International Journal of Science and Technology. – 2021. – Vol. 10. – № 1. – Pp. 9–17. – DOI: 10. 9-17. 10.13170/aijst.10.1.18749

8. Well Log data augmentation influence at accuracy of machine learning interpretation / A. Semenikhin [et al.] // EAGE/AAPG Digital Subsurface for Asia Pacific Conference. – European Association of Geoscientists & Engineers. – 2020. – № 1. – Pp. 1–3. – DOI: 10.3997/2214-4609.202075029

9. Rastorguev M.N. Ispol’zovanie diskriminantnogo analiza dlya interpretatsii dannykh gazovogo karotazha na primere Pavlovskogo neftyanogo mestorozhdeniya // Nedropol’zovanie. – 2019. – T. 19. – № 1. – S. 39–55.

10. Shcherbenev A.V. Ispol’zovanie veroyatnostno-statisticheskikh metodov dlya deleniya porod na pronitsaemuyu i nepronitsaemuyu chasti (na primere terrigennykh otlozheniy vizeyskogo yarusa Sof’inskogo mestorozhdeniya) // Nedropol’zovanie. – 2017. – T. 16. – № 1. – S. 14–22.

11. Restoration of seismic data using inpainting and edge connect / M. Radosavljevic [et al.] // SPE Russian Petroleum Technology Conference. – SPE, 2021. – DOI: 10.2118/206523-MS

12. Application of Artificial Intelligence Methods for Predicting Water Saturation from New Seismic Attributes / C.H. Sambo [et al.] // Offshore Technology Conference Asia. – OTC, 2018. – DOI: 10.4043/28221-MS

13. Application of machine learning for lithology-on-bit prediction using drilling data in real-time / T. Zhekenov [et al.] // SPE Russian Petroleum Technology Conference. – SPE, 2021. – DOI: 10.2118/206622-MS

14. An adaptive data driven model for characterizing rock properties from drilling data / H. Zhou [et al.] // 2011 IEEE International Conference on Robotics and Automation. – IEEE, 2011. – Pp. 1909–1915.

15. Scikit-learn: Mashinnoe obuchenie na Phyton. 1.1. Linear Models. 1.1.1. Ordinary Least Squares [Elektronnyy resurs]. – URL: https://scikit-learn.org/stable/modules/ linear_model.html#linear-models (data obrashcheniya: 31.03.2023)

16. K metodike otsenki perspektiv neftegazonosnosti Solikamskoy depressii po kharakteristikam lokal’nykh struktur / V.I. Galkin, I.A. Kozlova, A.V. Rastegaev [i dr.] // Neftepromyslovoe delo. – M.: OAO "VNIIOENG", 2010. – № 7. – S. 12–17.

17. Scikit-learn: Mashinnoe obuchenie na Phyton. 1.1. Linear Models. 1.1.3. Lasso [Elektronnyy resurs]. – URL: https://scikit-learn.org/stable/modules/linear_model.html#lasso (data obrashcheniya: 31.03.2023)

18. Scikit-learn: Mashinnoe obuchenie na Phyton. sklearn. ensemble. Gradient Boosting Regressor [Elektronnyy resurs]. – URL: https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.GradientBoostingRegressor.html (data obrashcheniya: 31.03.2023).

19. Scikit-learn: Mashinnoe obuchenie na Phyton. sklearn. ensemble. Random Forest Regressor [Elektronnyy resurs]. – URL: https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestRegressor.html#sklear... (data obrashcheniya: 31.03.2023).

20. Koskov V.N. Teoreticheskie osnovy distsipliny "Geofizicheskie issledovaniya skvazhin" i metodika vypolneniya kvalifikatsionnykh rabot: ucheb.-metod. posobie. – Perm’: Izd-vo Permskogo nats. issled. politekhnich. un-ta, 2016. – 121 s.