Use of machine learning technology to model the distribution of lithotypes in the Permo-Carboniferous oil deposit of the Usinskoye field

Permo-Carboniferous oil deposit of the Usinskoye field is characterized by an extremely complex type void space with intense cross-sectional distribution cavernous and fractured rock. In this study, for production site, process 3D geological modeling has been implemented. At first stage, it provided automated identification reservoir volumes comparing data core well logging surveys; at second rock lithotypes according to Dunham classification performed on basis comparison thin sections examination data. A large array factual information enables use machine learning technology Levenberg – Marquardt neural network apparatus toward achievement our research goals. The prediction algorithms lithotype using methods obtained training samples are applied wells without sampling. implemented approach enabled complementing model about permeability porosity, taking into account structural features identified lithotypes. For field, volumetric zoning different established. Taking based algorithms, density openness fractures were determined, fracture in volume was calculated. general, during implementation, errors remained within 3-5 %, which suggests reliability predictive solutions. results incorporated existing digital under study.