FORT WORTH—Aviation is still “in the stone age” of using artificial intelligence (AI) in engine maintenance, but the use of its predictive capabilities can lead to major savings in how future fleets and their service are planned, panelists at Aviation Week’s Aero-Engines Americas conference agreed.
“We are at the beginning of AI in aviation,” Jason Reed, president of Digital Innovation Group at GA Telesis said at the event here Jan. 28. AI allows “big steps” in the airlines’ abilities to define the right level of buffers in their parts inventories, Delta TechOps VP of Operations Mike McBride added. If done right, and with improved turnaround times, airlines can “spend hundreds of millions less for the same,” he said.
But the adoption of AI is still facing major obstacles, mainly a culture change in companies to trust the benefits and to alleviate concerns that its use will lead to job losses. Data is still not widely available in a standardized digital form, making it hard and often impossible to share information. Also, the industry needs to figure out a new commercial model that determines, among other things, who gets the savings of AI applications. “That always comes up,” McBride said. “Savings are to be had on both sides. It’s a pretty valuable thing for the OEM to know what demand will be well ahead. It’s an ongoing discussion. It’s a card game right now.”
Reed said that GA Telesis has been into eight to nine years of data science and started with supply chain and maintenance modeling. “We are now taking it to the next level,” he said.
“Generative AI has helped us win bids, it has been transformation to have the right stock at the right location at the right moment,” Reed said. “It takes a long time but once you crack the code it drives a different level of procurement ability and ability to sell.” At GA Telesis, one “wow moment” was when a project using AI showed a $32 million benefit.
The leasing arm also benefits. “AI can give you a list of return conditions of leased aircraft in seconds and you get [many] more months of usage out of the aircraft,” Reed said. Transitions can take months, depending on how easily documentation is available and how much work has to be put into the aircraft.
Delta TechOps has been accelerating the use of AI on the maintenance side since the pandemic when a big part of the Delta Air Lines fleet was temporarily idled. “We started with predictive maintenance and scrap prediction modeling,” McBride said. “You have a lot of data as an operator.”
Delta built a proof of concept for an AI model around the Pratt & Whitney PW2000 engine maintenance program and compared it to the original non-AI based schedule. “The new model beat the old model by $20 million,” said McBride. Delta also extended the procurement horizon from six months to 18 months, being able to give suppliers advance notice much further out.
David Harper, fleet support director at GE Aerospace, highlighted that AI is a “great way of data sharing. Everyone stands to benefit.” The company is now bringing it into the workplace and is seeing 2,500 uses every day.
Related Content
