Том 2, № 2 (2020)

The performed analysis of the geological and geophysical data of the subsalt complex on the eastern side of the Caspian basin allows us to look with optimism at the exploration possibilities of new areas at great depths.
High prospects may be associated with organogenic constructions of the inner parts of the paleobasin. In addition, of particular interest is the search for lithological and stratigraphic types of reservoirs, which have not been properly studied.

The article describes a method for training a convolutional neural network for rock lithology recognition based on images of core material. High Resolution (Hi-Res) photos were used for training models. The principles of convolutional neural networks and their practical application in geology are considered. As an outcome of this work, the model of neural networks for recognizing rock lithology was created and applied in practice using a smartphone. It was established that many ML and DL technologies potentially can be applicable for oil and gas industry.

Due to reducing volumes of light oil reserves, oil companies are increasingly focusing on hard-to-recover reserves, in particular, deposits with high-viscosity oil. Shallow-lying oil deposits are mainly concentrated in the Cretaceous horizons, in the western region of the country, along the Caspian coast. One of them is a highly viscous oil reservoir, consisting of 3 Cretaceous horizons of the Moldabek Vostochny section of Kenbai field, where 54.5 million tons of oil are concentrated, of which 19.5 million tons are recoverable. The current produced reserves makes up 5% of the initial recoverable reserves, and the approved design oil recovery factor implies the full-scale application of thermal oil recovery enhancement methods. The aim of this work is to select the most suitable method of thermal impact and assess the prospects for implementation with the calculation of economic feasibility. Taking into account the geological features of the reservoir, the cyclic treatment of the wells was chosen as the optimal method for increasing oil recovery. According to the results of calculations on the hydrodynamic model, the starting production rates using the technology of steam-cyclic treatment of producing wells are 2–3 times higher than the production rates for “cold production”: the difference in accumulated
production over 5 years will be 20–30%.

The article describes a unified approach to the assessment of resources and business planning. As an example, the paper considers oil and gas resources of the Pechora Sea. Based on the specifics of the problem (the quantity and quality of the available data and their nature), the methods of multicriteria fuzzy clusterization are used. The assessment of the prospects for the development of oil and gas fields was carried out using 3 criteria, namely: availability of resources (natural and climatic conditions of the region); hydrocarbon resources and their degree of exploration; economic assessment of the development. The results of such an analysis are presented in the form of maps of the Pechora Sea area, illustrating the most important and accessible areas and the sequence of their development. Such a stepwise development based on minimizing the risk and maximizing the benefits can be the basis for the successful and troublefree development of the resources of the entire region.

Traditional approaches for justification the properties of reservoir oil, used in calculating reserves often lead to errors in estimation of reserves and the impossibility of high-quality modeling and forecasts of hydrocarbon production.
The article gives examples where the method of averaging the properties of reservoir oil has led to errors in estimating the reserves of reservoir oil and dissolved gas. To qualitatively substantiate the properties of reservoir oil, methods based on physical parameter correlations should be used. Article also provides criteria for dividing the fluid into regions, which must be taken into account when analyzing the data, as well as tools for checking the quality of the received parameters.