Microsoft research boss says AI is helping tackle issues with carbon capture

A leading researcher at Microsoft said artificial intelligence (AI) could help to tackle some of the issues with carbon capture and storage (CCS).

Ranveer Chandra, managing director of research for industry at the tech giant, said his team has been collaborating with different stakeholders on how AI can be used to identify the right areas for storage as well as streamline how suppliers inject carbon into the ground and reduce associated costs.

It comes after Prime Minister Rishi Sunak announced two further transport and storage projects in the Humber and Aberdeenshire in July on top of the UK’s two existing carbon capture projects in Merseyside and Teesside.

CCS involves capturing carbon dioxide from industrial facilities, such as oil refineries and chemical plants before compressing it into a liquid, transporting it and injecting it into deep geological formations to be stored.

The process has been touted as an important climate solution for almost 20 years and progress in its development is gathering pace as private and public investment increases.

The intended use in the transition to net zero would be to capture carbon from industries that will be harder to decarbonise like cement, iron and steel.

But the debates and issues facing CCS have also been well-documented.

The process is incredibly energy-intensive, racks up high costs and storage carries significant risks, like leaks that could contaminate groundwater and soil.

Opponents have also argued that the process only captures a fraction of the total emissions from the lifecycle of oil and gas production and investments in CCS divert funding from lower-cost decarbonisation efforts while incentivising polluting industries to continue.

Speaking to PA Mediapoint, Mr Chandra said CCS is “not a silver bullet” but that it is “one part of the climate solution problem”.

“It will need to work with other technologies, with other investments around climate tech but that said, CCS is one of the one of the promising technologies as well,” he said.

His team have been working on ways AI can help solve some of the challenges around CCS so it can be a more feasible tool in the future of energy as well as helping hard-to-abate industries decarbonise.

Mr Chandra said potential storage sites and injection methods have so far been mapped out using traditional numerical simulations but these are slow and create a “big bottleneck in trying to evaluate some of these CCS technologies”.

He believes AI can help to model scenarios that answer questions like where to inject the carbon, how much can be stored, how safe is the site and whether it can be done cost-efficiently.

“A lot of those kinds of questions are the ones where artificial intelligence is much better positioned to answer,” he said.

“For example, one of the questions we use artificial intelligence for is to identify the best reservoirs to store CO2 in and the way we do that is we simulated a digital version and that digital version has things like temperature, the pressure, gas, location – all of that stuff.

“And then traditionally, people would have used numerical simulations to help answer that using numerical solution.

“But these simulations take a really long time and so the kinds of things we’re looking at is how do you optimise that, how do you run it much faster.”

Other issues include where to place the injection well, which influences how much CO2 gets stored through the various trapping mechanisms, and whether CO2 can leak from the storage site.

Mr Chandra said using AI means they have gotten more than 1,500 times the speed up of running the same questions that have usually been modelled through traditional simulations.

He also said they can simulate many more scenarios and optimise the locations of injection wells so the given CCS site can safely store more CO2.

All of this is helping to build a “digital trend”, which can support the analysis of uncertainties around CCS, he explained.

“With that, you can make better decisions and you can also add a lot of transparency in the system.”

Further down the line, Mr Chandra said a data-driven approach should become standard across the industry to help establish the best methods, sites and scenarios while flagging those that are not performing as well.

In terms of costs, Mr Chandra said there is an upfront price tag to implementing AI but it can ultimately reduce the high costs by decreasing the turnaround time of the simulations.

“If you’re investing in CCS, using AI will significantly improve the efficiency of any investment you make,” he said.

The work his team are doing is funded by private-public partnerships like the Northern Lights project in Norway, which is an investment from the Norwegian Government and several energy companies.

“Given where the costs are right now, it would require a multi-stakeholder approach,” he said.

“It’s technology innovation, it’s business innovation, it’s policy innovation. – I think everything needs to come together.”

He added that they are also taking a “very very collaborative approach” by open-sourcing their work and working with institutions like Stanford University, Georgia Tech and Imperial College London.

“We don’t want to keep this closed,” he said. “The climate problem is so immediate at this point, we are totally committed to making our research open and available, and to work with other stakeholders in this space,” he said.

“So we’re working with all the stakeholders out there to try and push the boundaries of how technology can help make a promising approach like CCS successful.”

Don’t forget to subscribe to our weekly email, to stay ahead with the latest AI news, analysis and insights.