MACHINE LEARNING APPROACH HELPS HIT 100% PREDICTION RATE A research team led by Tao Sun, associate professor of materials science and engineering at the University of Virginia, has made new discoveries that can expand additive manufacturing in aerospace and other industries that rely on strong metal parts. Their peer-reviewed paper, “Machine Learning Aided Real-Time Detection of Keyhole Pore Generation in Laser Powder Bed Fusion,” was published 6 January 2023 in Science Magazine and addresses the issue of detecting the forma- tion of keyhole pores, one of the major defects in a common additive manufacturing technique called laser powder bed fusion, or LPBF. Introduced in the 1990s, LPBF uses metal powder and lasers to 3D print metal parts. But porosity defects remain a challenge for fatigue-sensitive applications like aircraft wings. Some porosity is associated with deep and narrow vapor depressions, called keyholes. The formation and size of the keyhole is a func- tion of laser power and scanning velocity, as well as the material’s capacity to absorb laser energy. If the keyhole walls are stable, it enhances the surrounding material’s laser absorption and improves laser manufacturing efficiency. If, however, the walls are wobbly or collapse, the material solidifies around the keyhole, trapping the air pocket inside the newly formed layer of material. This makes the material more brittle and more likely to crack under environ- mental stress. Sun and his team, including materials science and engineering professor Anthony Rollett from Carnegie Mellon University and mechanical engi- neering professor Lianyi Chen from the University of Wisconsin-Madison, developed an approach to detect the exact moment when a keyhole pore forms during the printing process. “By integrating operando synchrotron X-ray imaging, near-infrared imaging, and machine learning, our approach can capture the unique thermal signature associated with keyhole pore generation with sub-millisecond temporal resolution and 100% prediction rate,” Sun said. In developing their real-time keyhole detection method, the researchers also advanced the way a state-of-the-art tool—operando synchrotron X-ray imaging—can be used. Utilizing machine learning, they additionally discovered two modes of keyhole oscillation. “Our findings not only advance additive manufac- turing research, but they can also practically serve to expand the commercial use of LPBF for metal parts manufacturing,” said Rollett, who is also the co-director of the Next Manufacturing Center at CMU. “Porosity in metal parts remains a major hurdle for wider adoption of the LPBF technique in some industries. Keyhole porosity is the most challenging defect type when it comes to real-time detec- tion using lab-scale sensors because it occurs stochastically beneath the surface,” Sun said. “Our approach provides a viable solution for high-fidelity, high-resolution detection of keyhole pore gener- ation that can be readily applied in many additive manufacturing scenarios.” The team’s research is funded by the Department of Energy’s Kansas City National Security Campus managed by Honeywell FM&T. NDT SOLUTIONS AND NDE LABS ANNOUNCE PARTNERSHIP NDT Solutions (New Richmond, WI) and NDE Labs (Benbrook, TX) have announced a strategic partner- ship to deliver a single point of contact for all nonde- structive engineering and testing services. With over 50 years of combined experience in nondestructive inspection, this alliance positions the two companies at the forefront of Industry 4.0. SCANNER Zhongshu Ren (left) and Tao Sun display the results of their research. Ren is the first author of the Science journal article. 10 M A T E R I A L S E V A L U A T I O N • J U L Y 2 0 2 3 2307 ME July dup.indd 10 6/19/23 3:41 PM PHOTO CREDIT: TOM COGILL FOR UVA ENGINEERING

Together, NDT Solutions and NDE Labs will provide a comprehensive range of nondestructive equipment, training, consulting, and testing services. Clients can now access solutions for their nondestructive inspection needs with just one phone call to either organization. NDT Solutions is a provider of equip- ment and engineering services designed to meet exacting customer testing requirements for aerospace and defense applications. NDE Labs specializes in nondestructive product and materials testing services for diverse industries, ranging from deep-sea oil and gas explo- ration, aerospace, and space. TPI, WINDSTAR CREATE DIGITAL TWIN FOR WIND BLADE MANUFACTURING TPI Composites Inc. (Scottsdale, AZ) collaborated with the Center for Wind Energy Science, Technology and Research (WindSTAR, Alexandria, VA), a National Science Foundation–funded Industry-University Cooperative Research Center (IUCRC), to design a composite manufacturing process based on a digital twin approach as released in the 2022 WindSTAR Annual Report. The project leveraged machine learning (ML), using big data to serve as the digital twin of the blade manufacturing process. This ML framework provides real-time feedback during fabrication, results in reduced defects, and enables more efficient production of wind blades versus the current high computational costs of the physics-based models. Stephen Nolet, senior director of Innovation and Technology for TPI, worked alongside student researchers and faculty from the University of Texas at Dallas, as well as technical experts from Olin Epoxy (Midland, MI) and Westlake Epoxy (Stafford, TX), to develop a framework for the digital twin of the vacuum-assisted resin infusion molding process. By applying an ML approach, the team achieved predictive accuracy of more than 95% with 100× faster computation than the physics-based simulations. “The primary value of utilizing [an] ML framework is leveraging historical results and data to inform current manufac- turing at a pace that significantly reduces defects from occurring in a real-time production environment,” Nolet explains. “Additionally, this technology allows users to create alternative manufacturing scenarios to increase production velocity in manufacturing operations while simultaneously reducing infusion-related problems.” In the coming year, the WindSTAR research team plans to focus on scaling the technology to larger components with greater manufacturing complexity. The work will apply tools taken from artificial intelligence to find patterns in historical data and predict outcomes on full-scale wind blade components including blade shells. INDUSTRYNEWS |SCANNER DATAFACTS |TRENDS IN MACHINE LEARNING FRAMEWORKS At the heart of artificial intelligence/machine learning developments are machine learning frameworks, a platform that allows developers to build and deploy machine learning models. The most popular frameworks have been Tensorflow, PyTorch, and Keras, with PyTorch recently emerging as the most popular. One more recent development is Google JAX, which provides a more lightweight functional programming environment. All of these tools are freely available to use (companies make money by providing specialized cloud services and deep learning model training using these frameworks). Source: Google Trends, 1 May 2016 to 25 May 2023. J U L Y 2 0 2 3 • M A T E R I A L S E V A L U A T I O N 11 2307 ME July dup.indd 11 6/19/23 3:41 PM