Subscribe Button 1

Enhancing Efficiency of Turbine Blade Manufacturing with Computer Vision

To lead the green energy revolution, Siemens Gamesa aims to boost capacity, but manual processes led to errors that slowed turbine blade production.

To accelerate manufacturing, Siemens Gamesa has partnered with IBM Services to create a machine learning (ML) solution on Microsoft Azure using a laser grid to show exactly where to place each fiberglass layer with pinpoint accuracy.

The new solution involves multiple technologies, including computer vision, ML, edge computing and the Internet of Things (IoT). By engaging experts from IBM Services to work with its own Digital Ventures LabsExternal Link (DVL), Siemens Gamesa gained the capabilities it needed to quickly lift its ideas off the drawing board and onto the factory floor.

Siemens Gamesa is now using its data-driven manufacturing solution on one of its production lines in Aalborg, Denmark, where technicians cast turbine blades from fiberglass.

Finn Mainstone, Senior Product Manager at Siemens Gamesa, explains: “Each turbine blade is custom designed by our engineers to precise specifications, and any defects during the manufacturing process can result in complex, costly and time-consuming corrections. To avoid this situation, our teams see a laser grid displayed on top of each blade that shows them exactly where to place each fiberglass layer. Crucially, they can now get instant alerts if the solution detects any errors or abnormalities in the surface of the blade.”

He continues: “Thanks to IoT-connected cameras in our factory and continuous analysis using machine learning models on the edge, all managed on Microsoft Azure, our technicians can place each blade layer with greater speed and accuracy. As a result, we are on course to reduce manufacturing error rates caused by misplaced material, which helps keep our production lines moving smoothly. In fact, once we roll the solution out globally, we will be better able to share best practices. This will reduce the learning curve for teams in our newly opened factories, such as Le Havre, France — enabling us to boost our throughput, accept more client orders, and bring the benefits of green energy to more people around the world.”

Defects Drag Down Productivity

The aerodynamic profile of turbine blades is crucial for efficient power generation, and building each blade involves highly skilled work. “Even though our blades for the newest SG 14-222DD turbine are 108 meters long, they are still built almost entirely by hand,” says Mainstone. “Because each blade is made to order, our teams are more like artisan craftspeople building furniture than workers on an assembly line. But as with any manual process, there is an ever-present risk of human error.”

Siemens Gamesa has a rigorous quality assurance process, and turbine blades are inspected and repaired during the final stages of manufacturing. For example, if a piece of fiberglass is placed incorrectly or laid on top of a foreign object, the affected section of blade is cut out and replaced – a rare but costly occurrence.

“Each time we rework a blade, it raises our costs, and limits the number of blades we can produce in each period,” continues Mainstone. “This additional pressure on our margins and throughput is a tough challenge in a highly competitive marketplace. Global demand for wind power is on the rise, and we knew that increasing our throughput would make it easier to capture these new opportunities and grow the business. To achieve our goal, we looked for a way to empower our technicians to work quickly with pinpoint precision.”

To build new digital capabilities that bring greater standardization and efficiency to its global activities, Siemens Gamesa formed an internal team of transformation specialists: the Digital Ventures Lab. One of the DVL’s first projects was a quality control system, which used a laser grid to show teams where to place fiberglass layers during production. However, the system could not detect defects in the manufacturing process and required significant and repetitive manual intervention to operate.

“We were confident we were on the right track by providing visual cues to our teams,” Mainstone recalls, “and we saw great potential to enhance our processes by augmenting the quality control system with intelligent automation.”

Melanie Beck, Senior Managing Consultant and Project Lead at IBM, continues, “The Siemens Gamesa team had an ambitious idea: mount an array of cameras above each manufacturing station, and validate the placement of each layer in real time using computer vision and ML models.”

Delivering Real-Time Feedback

By processing video on the Microsoft Azure IoT Edge platform, the company can apply advanced ML models to large amounts of unstructured data in real time and use its laser grid system to deliver feedback to factory teams. Because the new solution is built on Microsoft Azure, Siemens Gamesa gains the peace of mind that this mission-critical digital service is designed to run smoothly 24×7, thanks to robust high-availability cloud capabilities in line with Siemens Gamesa’s demanding corporate IT standards.

Building on the strong success of its pilot project for one production line in Aalborg, Siemens Gamesa is targeting a company-wide deployment of the new manufacturing solution. In the next phase of the project, Siemens Gamesa will extend the solution to cover all its manufacturing lines in Aalborg, its factory in Le Havre, France, and its factory in Hull, UK. Looking further ahead, the company is exploring the idea of implementing the solution in all its factories around the world.

“We expect a payback period of about two and a half years for our Azure-based production system,” says Kenneth Lee Kaser, Senior Vice President of Operations – Offshore at Siemens Gamesa. “And we expect the business case to get better and better as we add more functionality and see more secondary benefits.”

For more information: