According to Anna Dubovik, Head of Advanced Analytics at Russian oil giant Gazprom Neft, the company began introductory market research on digital technologies in 2017, evaluating offerings and identifying resources. In 2019, it began proactively implementing digital tools, focusing on exploration and specifically on seismic analysis. The reason, Dubovik says, is the high cost of a mistake during the first stages of exploration. Error in this step affects every step that follows. Gazprom Neft began using AI algorithms to identify exploration horizons in its seismic data analysis.
“A year ago, geophysicists viewed AI as a research tool for promising solutions. However, we are now rolling out the AI technology across our fields that are either at the development or exploration stage and have seen outstanding results. During the coronavirus lockdown people became more inclined to use digital transformation tools. This was a perfect opportunity to switch to AI that does not fear the virus and does not need a vaccine or a sick leave. It simply does what is required”, she says.
Anna Dubovik says AI is dramatically changing how the company identifies drilling horizons, or potential targets. “The speed and accuracy are incredible,” she says. In five days, one expert with expensive proprietary software identified drilling in 50% of the surveyed area with 25% traceability, however AI found horizons in 95% of the area with close to 100% traceability, which shows possible geological targets exactly.
Not only is Gazprom Neft improving its own accuracy, the company is sharing what it has accomplished with the industry, posting its technology on the GitHub development platform. “Everybody who is trying to adopt data science in the industry can just take our libraries and the data they have inside and adapt them to make the digital transformation and AI applications faster in their company,” she explains. This move also encourages collaboration. “We are hoping the tools are improved through open sourcing. Everyone who wants to contribute can contribute to it.”
Dubovik is optimistic about the potential results: “Companies in oil and gas tend not to share such data openly, and very few have entered this niche of open source and collaborating. Based on the examples that we see in high tech, however, this is definitely the direction to go”