Automated system for selecting the type of relative phase permeability functions under conditions of insufficient core data
UDC: 622.276.031.011.433
DOI: 10.33285/0207-2351-2022-9(645)-38-44
Authors:
IGNATOVSKY SERGEY I.1,
RYAZANOVA ELENA N.1,
AGUPOV MIKHAIL A.2,
VISHNYAKOV ALEXEY YU.3
1 LUKOIL-Engineering Limited PermNIPIneft Branch Office in Perm, Volgograd, Russia
2 LUKOIL-Engineering, Perm, Russia
3 LUKOIL Mid-East Limited, Basra, Iraq
Keywords: engineering programming, relative phase permeabilities, oil field, hydrodynamic modeling, machine learning, mathematical models, machine learning in the oil and gas industry, core studies
Annotation:
The article discusses the use of machine learning to generate relative phase permeabilities (RPhP) for using them in hydrodynamic models. This approach makes it possible to obtain the form of RPhP functions based on the prepared database of core experiments. Algorithms of mathematical models are described and a software module has been developed with the help of which it becomes possible to automatically obtain the form of relative permeability curves for objects based on the available field data (reservoir type, oil saturation coefficient, porosity, permeability, oil and water viscosities, sample depth, pressure and temperature). Two mathematical models ("Multivariate Linear Regression" and "Random Forest") implemented in Python high-level programming language are considered. The control sample presents the results of the methods convergence with real laboratory studies. The method was tested by applying a retrospective analysis on several objects and conclusions were drawn about the approach applicability.
Bibliography:
1. Aziz Kh. Matematicheskoe modelirovanie plastovykh sistem. – M.–Izhevsk: In-t komp’yuternykh issledovaniy, 2004. – 416 s.
2. Geologicheskoe i gidrotermodinamicheskoe modelirovanie mestorozhdeniy nefti i gaza / R.M. Ter-Sarkisov, V.M. Maksimov, K.S. Basniev [i dr.]. – M.–Izhevsk: In-t komp’yuternykh issledovaniy, 2012. – 452 s.
3. Demidenko E.Z. Lineynaya i nelineynaya regressiya. – M.: Finansy i statistika, 1981. – 302 s.
4. Balashov V.A. Tsifrovoy kern. Modelirovanie mikrotecheniy v porovom prostranstve porod-kollektorov // Neftegaz.RU. – 2019. – № 7. – S. 83–85.
5. Tsifrovoy kern. Kompleksirovanie dannykh petrograficheskikh issledovaniy karbonatnykh porod s rezul’tatami izucheniya kerna / S.A. Idrisova, M.A. Tugarova, E.V. Stremichev, B.V. Belozerov // PRONEFT’’. Professional’no o nefti. – 2018. – № 2(8). – S. 36–41.
6. Breiman L. Random Forests // Machine Learning J. – 2001. – Vol. 45, no. 1. – P. 5–32. – DOI:10.1023/A:1010933404324
Back to issue