Gazprom Neft: Specialists Taught Program to Recognize Rocks by Photo

Gazprom Neft has successfully tested a new technology based on machine learning and digital vision for studying rocks at fields in Western Siberia. The project proved the ability to accelerate laboratory analysis of a full-size core by 7 times to make a decision on its further research program. For this, cores are extracted from wells and “digitized” using special photography. The accumulated array of images is analyzed by an automatic system that recognizes lithological layers on the photo, determines their name, integrity and other characteristics.

To train the mathematical model, the Gazprom Neft Data Development and Monetization Center developed a lithological layer classifier – the principles according to which a particular element in a photo of a core cut should be correlated with certain characteristics of the geological rock. A dataset of more than 17,000 daylight and ultraviolet images of the core was prepared in partnership with the Faculty of Geology, Moscow State University. M.V. Lomonosov.

Together with other Gazprom Neft developments, the new system, according to preliminary estimates, will save about 85 million rubles annually on laboratory research.

“To optimize our work, we consider the entire range of technologies, including those used in other industries. By adapting them to our tasks, we automate routine processes, and each new digital data collection and analysis tool contributes to improving the efficiency of oil production in general. We are integrating machine learning into exploration to reduce uncertainties and, based on accurate information, make large-scale infrastructure decisions. ”

Alexey Vashkevich Director for Technological Development, Gazprom Neft

“Digitization of laboratory rock samples allows not only to significantly reduce the uncertainties in the asset development strategy (for example, by performing many simulations on the displacement of oil from oil samples using polymers of different chemical composition on the same sample), but also to speed up the decision-making process by several times, which is based on the results of laboratory research. This is a prime example of data-driven business process change, an approach that is embedded in a company’s digital transformation strategy. “

Andrey Belevtsev Director for Digital Transformation, Gazprom Neft