Deploying AI for the methane challenge [Gas in Transition]

Artificial intelligence (AI) is revolutionising the energy industry’s ability to predict methane leakage.

It is a key frontier in the fight against climate change, as methane is 80 times more potent as a greenhouse gas than CO2. It also has a far shorter lifespan than CO2 in the atmosphere, meaning that action taken today will yield results faster.

The UN believes methane abatement is crucial to keep global temperature rises below 1.5 OC, as set out by international agreements. Its recent report on climate change stressed that methane should be the priority for emitters in the short term.

OGCI Climate Investments, the venture capital division of the Oil and Gas Climate Initiative (OGCI), is aware early-stage AI businesses can make a massive impact in this area. Last summer, OGCI wrote a series C cheque for Urbint, a leading predictive AI platform for infrastructure that deals with methane abatement. Urbint’s series C round landed $60mn of capital in all, from backers that also included National Grid Partners, Salesforce Ventures, American Electric Power, Energize Ventures, Energy Impact Partners and Blue Bear Capital.

The AI in Urbint's software automatically knows every subtype of surface infrastructure along the distribution grid. Using this information, the software can predict exactly where methane is likely to penetrate steel surface materials. This enables precautions and preventive rather than reactive maintenance to be undertaken, meaning that leaks can be avoided altogether, rather than fixed when they occur.

Speaking to NGW, Urbint’s vice president of product, David Preston, said the New York-based business had signed up 30 of North America’s major utilities and infrastructure companies as customers.

Preston compares the software to a “risk-based” prevention system, but using the benefits of AI to identify these risks more effectively. Not only can the technology be used to keep methane gases in the pipe where they’re supposed to be, but it can also be used for other infrastructure and excavation audits, such as keeping the pipeline temperatures or operational components in check.

The figures on methane abatement speak for themselves. In 2020 alone, Urbint’s technology prevented methane emissions equivalent to more than 60,00 metric tons of CO2 equivalent.

Preston said: “The unique advantage of this approach is rather than focusing on mitigation of emissions after the fact, is that we're making it possible to get ahead of accidental emissions, so we can keep methane gas in the pipe in the first place."

Threat detection

Urbint was founded in 2015. It has been expanding the market for its infrastructure decision support software through M&A activity, including the acquisition of distribution system monitoring company Opvantaek. Opvantek’s product was already being used by several of North America’s largest gas grid companies, for tasks such as cyber protection, risk management, pipe evaluations and asset investment forecasts.

Infrastructure threat detection is big business in North America, and the market opportunity is about to get even larger. Recently, the US legislature approved a $1.2 trillion, bi-partisan funding package for rebuilding the nation’s core systems – from roads to gas pipelines to clean water facilities – over the next ten years. It’s part of president Joe Biden’s pledge to rebuild America’s “crumbling" framework, expected to result in 1.5mn new jobs every year.

“All this work will require a lot of digging, meaning damage prevention is going to be important,” said Urbint’s director of marketing and communications Michael Dolmatch. “It also means companies focus more on resilience, and more on infrastructure renewal."

“In the meantime, given the several years it will take to update the US grid, we still have to keep the lights on, and the heat on in all of the existing infrastructure that still has to operate. Companies like us will have a major impact, and it's a lot of fun from a business perspective, seeing some great momentum from these trends."

National Grid, a power and gas grid operator active in the UK and the US, has been an Urbint shareholder since 2017. That strategic window gave National Grid access to Urbint's technology early in the game, and Urbint software is now helping to prevent damage to its underground pipelines in the northeastern US. The outcome must have been favourable, as National Grid has since recalled Urbint to its London headquarters, this time to explore how its worker safety AI software could be used to protect the Grid’s infrastructure personnel.

Preston admitted the initial contract was a massive undertaking, but it paid off.  He said: "In terms of getting set up with National Grid, there were certainly complexities, given the fact they operate across multiple jurisdictions across gas and electric."

"But our approach is to align data from [various infrastructure operators] into a standard Urbint data model, so we can best utilise and build AI models at scale, incorporating prior knowledge of both industry and cross-industry incidents.

“Our damage prevention applications helped National Grid reduce the damages on their system by 37%. That's a significant result in terms of underlying methane reduction, as well as underlying public safety and asset safety."

This year Urbint is scaling rapidly with the rollout of its worker safety AI software, its second product on the market. And it is still looking to win new deployments for its grid damage auditing tools. For Preston, having two products on the market represents a “critical milestone”, in terms of the startup’s growth trajectory.

Preston concluded: “Urbint’s growing in alignment with our anticipated revenue plans for 2022. We’ll take similar principles for damage reduction and methane reduction to help identify high-risk complex work that has a higher likelihood of causing an individual performing work on an asset to have an injury, and helping companies focus their investment and planning and controls to mitigate those incidents.

“That has been released earlier this year, and is probably a critical milestone as we’ve got two core applications or products in the marketplace.”