New AI technology could help detect leaks in underground carbon storage

Innovative machine learning techniques could help detect leaks during underground carbon dioxide (CO2) sequestration, helping to reduce their environmental and economic impacts.

Academics from various universities have formed an international research partnership to implement artificial intelligence (AI) and machine learning to detect leaks in pipelines and strings of wells.

They will also use a new “digital twin” for leak detection during single-phase flow – crude oil or gas – and multi-phase flow during transport and injection of carbon dioxide into the underground storage site.

The idea was to create a virtual representation of a gas pipeline which is updated in real time via a network of sensors mounted and installed in the real gas pipelines.

The team will use computational fluid dynamics (CFD), in which AI simulates the flow of liquids and gases and hopes to be able to accurately predict the likelihood and location of leaks in single-phase and multi-phase flows. .

These techniques should more accurately predict the location, size, number, and direction of small chronic and larger leaks and ultimately take action by AI without requiring human intervention.

Academics from Teesside University’s School of Computing, Engineering and Digital Technologies are working with researchers from Texas A&M University in Qatar (TAMUQ), Qatar University, Texas A&M (USA), Birch Scientific (USA) and Rock-Oil Consulting of Canada for the project.

Dr Sina Rezaei Gomari, Lecturer in energy and environmental engineering said: “The University of Teesside is committed to research that uses new and disruptive technologies, processes and business models to forge a smarter and greener industrial economy.

“It is well documented how devastating pipeline leaks can be if not caught and dealt with quickly and efficiently.

“This research will examine how advanced computational techniques, including machine learning and digital pairing, can be applied to accurately predict where faults are occurring, without the need for remotely operated vehicles or aircraft to scan the pipeline, which can take time. and expensive.

“We will work with major oil and gas companies to ensure that this research can have real industrial applications.”

Bonny J. Streater