The paper presents a method for characterizing well sections using classification and clustering methods based on geological, geomechanical, and petrophysical parameters. The results obtained can assist in predicting development indicators in unexplored drilling areas of oil field and aid in selecting technologies for geological and technical activities. The study focuses on eight formations of Achimov deposits found in wells of the Priobskoye field. These formations have a complex mineral composition, low filtration capacity, and poor reservoir connectivity. Based on geophysical and laboratory core material studies, the authors calculated the filtration and capacity properties, brittleness index, and mineral macro components. These values helped to determine the integral characteristics of each well, including the thickness of the collector intervals and average filtration and capacity properties. The geological marking is determined based on expert assessment of the sedimentary facies. Long short-term memory (LSTM) neural networks were used to solve the classification problem. The machine learning methods identified most of the classes that correspond to the boundaries of the facies zones identified by the expert. Unsupervised k-means clustering was performed using integral characteristics of the section with weights. This allows for predicting facies of sedimentation areas. The Achimov formation reservoirs were classified based on well data from the Priobskoye oil field, taking into account their distinct geological and production characteristics.
References
1. Kalmykov G.A., Metodika opredeleniya mineral'no-komponentnogo sostava terrigennykh porod v razrezakh neftegazovykh skvazhin po dannym kompleksa GIS, vklyuchayushchego spektrometricheskiy GK (Methodology for determining the mineral-component composition of terrigenous rocks in oil and gas well sections based on data from a complex of geophysical well studies, including spectrometric gamma-ray logging): thesis of candidate of technical science, Moscow, 2001.
2. Nadezhdin O.V., Elkibaeva G.G., Shagimardanova L.R. et al., Peculiarities of building volume mineralogical model for rocks with complex component composition (In Russ.), Neftyanoe khozyaystvo = Oil Industry, 2020, no. 5, pp. 36-40, DOI: http://doi.org/10.24887/0028-2448-2020-5-36-41
3. Paatero P., Tapper U., Positive matrix factorization: A non-negative factor model with optimal utilization of error estimates of data values, Environmetrics, 1994, V. 5,
no. 2, pp. 111–126, DOI: https://doi.org/10.1002/ENV.3170050203
4. Merkulov V.P., Posysoev A.A., Otsenka plastovykh svoystv i operativnyy analiz karotazhnykh diagramm (Evaluation of reservoir properties and operational analysis of well logs), Tomsk: Publ. of TPU, 2004, 176 p.
5. Hochreiter S., Schmidhuber J., Long short-term memory, Neural computation, 1997, no. 9, pp. 1735-1780, DOI: https://doi.org/10.1162/neco.1997.9.8.1735
6. Gorban A.N., Zinovyev A.Y., Principal graphs and manifolds, In: Handbook of research on machine learning applications and trends: Algorithms, methods and techniques, 2009, pp. 28-59.
7. Vorontsov K.V., Lektsii po algoritmam klasterizatsii i mnogomernogo shkalirovaniya (Lectures on clustering and multidimensional scaling algorithms), 2007.
URL: htpp: // www.ccas.ru/voron/download/Clustering.pdf
8. Sergeychev A.V., Toropov K.V., Antonov M.S. et al., Automated intelligent assistant in the selection of well placement when developing hard-to-recover reserves (In Russ.), Neftyanoe khozyaystvo = Oil Industry, 2020, no. 10, pp. 76-81, DOI: https://doi.org/10.24887/0028-2448-2020-10-76-81
