BP Ventures has invested US$5 million in Houston-based Belmont Technology as part of its plans to utilise artificial intelligence in its upstream business.
BP will gain access to Sandy, Belmont Technology’s cloud-based geoscience platform.
Sandy works like an oil and gas Siri. It can process natural language requests from a user and provide appropriate data.
Sandy utilises a knowledge graph to connect related facts to each other, make its own deductions and present its findings.
A knowledge graph is more than a simple database – the information it contains is defined to make it meaningful and create relationships between facts. This means that the database can store more complex information than simple numbers. A reasoning engine is then able to make sense of the information and follow connections to create new information.
The graph structure is flexible and can accept new data – new connections are formed and old connections are updated. Oil companies produce more data than ever, as new monitors and sensors are installed to improve margins. Real time data is more common too. A knowledge graph has to accommodate a constant stream of data.
This combination of definition and connection is how Sandy can take real-world entities, understand them and present its own reasoning back to a user in natural language. If this sounds familiar, it is because knowledge graphs are what power Google Assistant and creates the information boxes that accompany Google searches.
BP group head of technology David Eyton stated in a press release: “This AI-based platform, which we’ve nicknamed Sandy, is expected to unlock critical data for our subsurface engineers at a much accelerated pace. Our experts will ask it questions about our reservoirs like, ‘What factors control production in the Chirag field?’ Sandy will then interpret our data, including mapping out many more scenarios than are currently constructed, helping us make faster, better informed upstream decisions.
By feeding Sandy with BP’s geology, geophysics, reservoir and historic project information, the company can use it to identify new connections and workflows, and create a knowledge graph of BP’s subsurface assets.
BP wants to use Sandy to accelerate project lifecycles, from exploration through to reservoir modelling, the technology is targeting a 90% time reduction in data collection, interpretation and simulation.