Hard-to-recover hydrocarbon reserves in Western Siberia are mostly concentrated in reservoirs of complex structure, represented by a combination of massive conventional reservoirs with intervals of thinly bedded sandstones and clays, interbeds of sandstones and siltstones with increased content of dispersed clay, and reservoirs with partially carbonated void space. Reservoir delineation methods based solely on open porosity are limited to massive conventional reservoirs. Lowering the porosity threshold may result in the inclusion of clay intervals. In order to predict porosity in such sediments, it is advisable to use nuclear magnetic logging, which makes it possible to identify all interbeds in the section that have effective porosity, regardless of the type of reservoir. The limitation of the application of nuclear magnetic logging is the fact that this method is present in a small number of wells, usually not more than 3 % of the well stock. In this work, effective porosity was modeled using neural network models and prediction maps of additional effective thicknesses were constructed. The average effective thicknesses over the field area was about 5 m. In some areas, the additional reservoir thickness was determined to be 7-10 m, demonstrating the promising application of the proposed methodology to improve the efficiency of field development. The proposed interpretation method, based on modeled effective porosity from nuclear magnetic and neutron logging data, allows the identification of dense interbedded and partially thin bedded, highly clayey, low permeability pseudo-reservoirs. To evaluate candidate wells, a model for calculating well inflow after hydraulic fracturing from pseudo-reservoir intervals is proposed. The results obtained and the developed approach are planned to be used to study and interpret a complex geological section formed in the distal part of the sea shelf and deep-sea sedimentation environment.
References
1. Samuel A.L., Some studies in machine learning using the game of checkers, IBM Journal, 1959, July, pp. 210–229, DOI: https://doi.org/10.1147/rd.33.0210
2. Mitchell T.M., Machine learning, McGraw-Hill, 1997, 432 p.
3. Ghareb H., Elsakka A., Chaw Y.N., Artificial Neural Network (ANN) prediction of porosity and water saturation of shaly sandstone reservoirs, Proceedings of 2018 AAPG/EAGE/MGS Myanmar Oil and Gas Conference: A Global Oil and Gas Hotspot: Unleashing the Petroleum Systems Potential, 2018, DOI: https://doi.org/10.1306/51559nyein2019
4. Cvetković M., Velić J., Malvić T., Application of neural networks in petroleum reservoir lithology and saturation prediction, Geologia Croatica, 2009, V. 62, pp. 115-121, DOI: http://doi.org/10.4154/gc.2009.10
5. Kanaev I.S., Neyrosetevoe detektirovanie produktivnykh intervalov na primere ob»ekta BV10 Samotlorskogo neftegazokondensatnogo mestorozhdeniya (In Russ.), Neftyanaya provintsiya, 2019, no. 4(20), pp. 157-169, DOI: https://doi.org/10.25689/NP.2019.4.157-171
6. Mardi M., Nurozi H., Edalatkhah S., A water saturation prediction using artificial neural networks and an investigation on cementation factors and saturation exponent variations in an Iranian oil well, Petroleum Science and Technology, 2012, V. 30, pp. 42-434, DOI: http://doi.org/10.1080/10916460903452033
7. Liping Zhu, Hongqi Li, Zhongguo Yang, Chengyang Li, Yile Ao, Intelligent logging lithological interpretation with convolution neural networks, Petrophysics, 2018, V. 59, pp. 799-810, DOI: http://doi.org/10.30632/PJV59N6-2018a5
8. Murav’ev I.A., Application of machine learning algorithms to interpret well logging results in the context of the task of identifying terrigenous reservoirs (In Russ.), Vestnik TGU. Fiziko-matematicheskoe modelirovanie. Neft’ gaz energetika, 2019, V. 5, pp. 123-137.
9. Nadezhdin O., Efimov D., Minikeeva L., Markov A., Experience with using data analysis technologies in identification of lost production zones (In Russ.), SPE-191597-18RPTC-MS, 2018, DOI: http://doi.org/10.2118/191597-18RPTC-MS
