GE VERNOVA IS DEPLOYING AI-ENABLED MACHINES TO BOOST WIND TURBINE BLADE QUALITY

“The world needs wind,” said Vic Abate, CEO of GE Vernova’s Wind business, at the company’s recent Investor Day event. Indeed, wind power already plays an important role in worldwide energy production. However, to meet the world’s growing energy needs, and to foster decarbonization, wind power must increase the amount of electricity it supplies to the grid from 7% currently to 25%. “Think of an energy system where one out of four electrons comes from wind,” Abate added, “because this is a different world.”

With the sector’s total generation expected to increase at least sixfold by 2040, the world’s factory floors are projected to churn out hundreds of thousands of wind turbines, each one the product of a colossal manufacturing operation. It takes several days to cut, roll, and weld the steel sections that make up the turbine’s tubular tower, as well as mold the fiberglass of the nacelle, the box that houses the turbine’s generation equipment. That’s before you get to the three giant blades that catch the wind’s flow and spin its rotor to produce power.

A Labor of Love
“It’s a very labor-intensive process that goes into making these football-field-size combinations of carbon and glass,” says Veronica Barner, renewables director at GE Vernova’s Advanced Research Center. It takes around 2,000 labor hours to produce each blade — a kind of birthing process that involves scores of workers handcrafting fiberglass fabric and balsa wood into the shapes of enormous split pea pods, and then attaching tubes that suck out air and pump in gallons of molasses-like resin. Now GE Vernova, which boasts an installed fleet of approximately 56,000 wind turbines, is raising the bar on blade manufacturing. It is harnessing the power of robotics and artificial intelligence (AI) for inspections to help to ensure that the quality of each blade leaving the factory meets rigorous design specifications.

The inspection solution is complex, but it aims to keep GE Vernova’s turbines performing well into the future. Powerful algorithms scour each blade’s interior, looking for deviations before they are shipped out with a digital quality certificate that marks their high-tech digital vetting. The enhanced ability to catch deviations early in the process is a major boon for power producers because it reduces the likelihood of more serious issues once the turbine is up and the blades are spinning, which could result in costly downtime. Over the long term, this AI-enabled quality capability is expected to boost the lifetime of the critical components, beefing up the longevity of the turbines that are central to the world’s energy transition.

GE Vernova’s engineers are also employing AI to inspect the raw materials before molding and assembly, explains Barner. “The concept of using technology to enhance our ability to find potential issues at the right time is being applied to every critical step in the manufacturing process,” she says. “These capabilities will help to ensure that all blades that come out of the factory lines, no matter where they are in the world, have the same consistent quality.”

Needle in a Haystack
Congratulations, it’s a blade! After those 2,000 labor hours, a beautiful blade is born, weighing in at around 20 tons and measuring roughly 80 meters long. Size isn’t everything, but it’s important in the wind sector, where rotor size is directly correlated to energy production. “You’re putting three of these on and spinning them with the thrust that goes into a GE9X aircraft engine,” Barner says, referring to the most powerful aircraft engine in the commercial aviation industry.

The stresses and strains on the blades are massive, which ups the ante for manufacturers. Barner explains that even an anomaly consisting of a couple of millimeters of surface deviation can compromise the longevity of a blade. The risks have turned the manual inspection of wind turbine blades into one of the most meticulous jobs on the factory floor. “All blade suppliers have the same challenge — trying to find a needle in a haystack,” she says.

To solve the challenge, GE Vernova’s Advanced Research engineers working out of Niskayuna, New York, focused on the nature of blade manufacturing as a “repeatable” process. They decided the solution would consist of readily scalable technology that could accurately identify and record minuscule deviations in the blade, thereby enhancing the work of the company’s experienced human inspectors.

Crawling Inside
The engineers began by experimenting with state-of-the-art 360-degree digital cameras that could relay a high-quality video feed of the blade’s interior to the cloud. Back on the outside, remote engineers could extract images from the video feed, zooming in and out as needed, until they’d pieced together the entire surface, inside and out, into a high-definition montage. “That’s a huge accomplishment,” says Barner, “but it didn’t solve the entire problem.”

The next step was harnessing computer vision, the field of AI that trains computers to identify visual information in images. The human inspectors drew up a list of the deviations they looked for when scouring a wind turbine blade. They then put their AI system through blade boot camp. “We’re using the annotated images to train a series of AI algorithms that can analyze the images and autonomously flag potential anomalies,” says Barner.




It wasn’t long before their AI was up to speed, autonomously and accurately recognizing several of those deviations, the results of its training on tens of thousands of annotated images. Algorithms automatically log the flaws in a digital tool, which allows human technicians to review the findings and carry out any needed repairs before the blade is shipped to the turbine operator. “As we deploy this to the shop factories, the feedback loop becomes even richer,” says Barner. “In a matter of minutes, our teams on the ground are empowered with very critical, timely knowledge about the blade that they are working on bringing to life. “I think of it like a prenatal ultrasound,” she adds.

The final problem was access to the blade’s interior. To solve this problem, GE Vernova deployed a fleet of robotic remote-operated “crawlers” that could serve as their all-seeing eyes inside the components. While around 50% of the blade’s inner surface is off-limits to a human, the crawlers, which are about the size of a two-foot model car, can inch along its full length and scan the belly of the beast in just 30 minutes.



Embedding the Concept
Inspecting the blade at the end of the manufacturing process is just the tip of the proverbial iceberg. Barner says that the concept of using technology to find quality issues is traveling further up the production line.

“You don’t just get an ultrasound at the end of a pregnancy. There are a series of scans and tests that doctors rely on to provide the best care throughout,” she explains, extending the prenatal-care analogy. “Our vision is very similar. We are developing and deploying a series of technologies, leveraging AI, to certify the quality of blades in the most critical production steps.”

With this new AI blade inspection process in use, the blades for GE Vernova’s 154-meter rotors will be leaving factories bearing their freshly printed digital quality certificates, starting with the blades for the SunZia wind project in New Mexico.