Tech Gains Show Up in Emissions Reductions, Midstream Efficiency

At a key natural gas station in Virginia, Williams Cos. replaced an aging fleet of engines with new turbines that decreased emissions and took up far less space.

At the same time, Enbridge was replacing hundreds of flow meters with newer models, delivering much better information to the company in real time.

They’re both part of the technological advance in the midstream sector, one that has been reducing emissions while improving efficiency. Better equipment and improved use of artificial intelligence (AI) are big parts of the gains.

Here’s how they’ve done it.

Williams

Station 165 in southern Virginia is part of the Transco pipeline network, the nation’s largest-volume interstate natural gas pipeline system. It’s also the endpoint of the Mountain Valley Pipeline, which carries natural gas from the Marcellus Shale.

The changes that came to the station began in 2018 with the Emissions Reduction Program, said Mark Gebbia, vice president for environmental, regulatory and permitting (ERP) for Williams. Phase 1 was completed in 2024; Phase 2 is scheduled through 2030.

“We’ve done a lot of work to put the program together,” he said. “It culminated in 2024.  We’ve replaced 112 compression units, 92 of those were done in 2024. We’re going to start seeing the results in emissions here as we move into this reporting cycle.”

Those 112 units delivered about 335,000 hp. By 2030, that will be up to 650,000 hp, Gebbia said.

The ERP started with sustainability reporting in the 2010s, according to Gebbia. The process identifies material issues to the business and begins corrective actions.

The company saw that emissions of nitrous oxides (NO_X) were rising on some of its lines, which were built in the 1940s and 1950s and carried gas south to north when they opened. The flow was reversed after the vast natural gas discoveries of the 2000s to carry gas to Gulf Coast ports for export.

The reverse flow led to an increase in NO_X emissions along the line, which wasn’t originally designed with reversal in mind.

“We’d done a lot of work to reduce emissions during that period and started seeing the trend reverse with reversed flow,” Gebbia said. “And so, we designed that first phase of the Transco emission reduction project to replace the old recip (reciprocating) compressors with modern turbines.”

It worked.

“It was amazing. We replaced 10 units, stand-alone recip engines, with one single block of turbine horsepower and NO_X emissions were typically reduced by about 98% from those old uncontrolled engines,” Gebbia said. “And then the methane reductions were pretty significant, too.”

The 10 recip units were designed to leak and vent methane. The new unit was designed to recapture the gas. Methane intensity, the amount released per unit of energy created, dropped by 60%.

The reductions also helped Williams get the state permits to expand.

“We were able to double the capacity of the station and reduce total NO_X emissions by 90-some-odd percent,” Gebbia said. “It helped streamline the air permitting process because we were now a minor source instead of a major source.”

The Williams project got its permits while two other nearby projects were denied.

The changes at Station 165 and throughout the system are also part of the bigger picture of the energy transition.

“Even from the CEO level here at Williams all the way down, there’s a lot of thoughts on what energy transition could be and can be,” Gebbia said. “At the end of the day, a lot of those are going to be more expensive and more difficult to build out than we thought they would be.

“The onus is on us to operate these assets where you’re just as efficient as possible keeping gas in the pipe, minimizing the energy it takes to move the gas,” he said. “It’s interesting when you go from ‘What’s the minimum we need to do to comply with regulations’ to ‘This is a material issue to the company. Go do something about it.’”

Enbridge

Enbridge began working with machine learning and AI seven years ago, said Bhushan Ivaturi, senior vice president and chief information officer. The systems help monitor assets and reach environmental objectives.

One application involved Enbridge’s traditional inline inspections, Ivaturi said.

“We’ve created a product called Trending Workbench that takes data from our inline inspections and from other modalities like non-destructive examination and brings those sources together,” he said. “The machine learning model is now able to help inform a much more precise allocation of where we want to go and look at things.”

Flow meters are one of the things. Enbridge has hundreds of flow meters in its system.

“We are able to predict the failure of these flow meters well in advance of any failure happening, and that gives us the flexibility in terms of time to do appropriate strategies,” Ivaturi said.

Enbridge also uses AI as part of its aerial survey program with multiple benefits.

“We’ve instrumented those flights with a UV LIDAR (laser imaging, detection and ranging), a smart camera that’s now capturing all the imagery on the right of way,” Ivaturi said. “We’ve created an AI that’s taking all of that data and it’s able to detect threats, such as some kind of vehicle that shouldn’t be there because we don’t have a permit to do any work on it or the vehicle doesn’t match the type of vehicle that was permitted.”

The AI can also detect the signatures of various hydrocarbons and help identify potential facility leaks, Ivaturi said.

Finding failure points helps reduce emissions, and so does general improvement in efficiency.

“We take a lot of the data in terms of how we run pumps and all the critical equipment that consumed that energy,” Ivaturi said. “Our control center then takes the AI’s recommendation to make decisions to optimize the use of energy. In that process, not only have we saved costs, but we’ve also reduced our GHG (greenhouse gases).”

Just on the liquid pipes where that system has been deployed, the annual savings is 16 million to 18 million kg of CO₂, Ivaturi said.

As the technology has developed, Enbridge has also had to consider when not to use AI. The company has a full accountability structure that considers policy, accountability and ethics.

AI deployment “goes through the checks and balances and the consideration is very rounded and it reflects our AI policy,” Ivaturi said. “There are some places where we can run AI in a more automated manner, so we can let it take actions. There are other cases that are around monitoring, controlling the pipe itself or critical equipment where we make sure that the human is in the loop.”