Phase3D’s real-time anomaly detection increases quality and reliability for US Air Force and NASA
June 26, 2024
Phase3D, based in Chicago, Illinois, USA, is reported to be the first company to correlate measurable metal powder bed fusion build anomalies to final part defects in real-time. The work was done for the United States Air Force and NASA, two leading AM producers for the aerospace industry, and validated with two materials on two different Laser Beam Powder Bed Fusion (PBF-LB) Additive Manufacturing machines.
Phase3D worked closely with the US Air Force and NASA to develop Fringe Research, an in-situ monitoring product which measures, in microns, every layer of a powder-based AM build. The system automatically identifies anomalies from the build process that lead to porosity, a major cause of part rejection of the US Air Force and NASA. Fringe Research can be used on most powder-based AM systems and employs structured light, a well-known and understood metrology technology, to create the measurements used for this correlation.
Fringe Research does not use artificial intelligence (AI) or machine learning (ML) to create measurements or identify anomalies.
Phase3D used Ti64 for the US Air Force and GRCop-42 for NASA to determine the effect of detectable build variation on part quality. The builds included an anomaly generator, which created realistic, geometry-based powder and melted anomalies, including hops and streaks patterns seen in most AM builds.
While measuring the build, Fringe Research automatically identified hops and streaks that were later correlated to porosity in the final part. Parts were inspected using CT scanning and correlation was done in Fringe Research and a commercial CT inspection software.
For Ti64 manufactured on an EOS M290 for the U.S. Air Force:
- 81% of Fringe Research identified anomalies correlated to part defects detected by CT
- 100% of Fringe Research identified anomalies ≥47um depressions correlated to defects detected by CT
For GRCop-64 on a Colibrium Additive (previously GE Additive) M2 for NASA:
- 83% of test specimen identified defects were correlated to layers with Fringe Research identified anomalies
- 100% of Fringe Research identified anomalies ≥42um depressions correlated to defects detected by CT
“Providing our customers a high correlation of measurable build anomalies to part defects is changing what is possible for AM,” shared Niall O’Dowd, Founder and CEO of Phase3D. “Our customers continue to request objective data that can identify part defects when they occur. With the data Fringe Research collects, we predict our aerospace customers will be able to increase machine throughput by more than 10% every year by stopping parts that will fail inspection early.”
Historically, Phase3D customers relied on visual images from the machine to either identify anomalies using AI/ML or an engineer to watch a video of the visual images or scroll through them. Fringe Research measures the entire build surface three times during every layer. The heightmaps and Fringe Research allow users the opportunity to identify how build anomalies impact part quality, including CT-identified defects, fatigue life, tensile strength, and more.
Phase3D will be showcasing the correlation data for the US Air Force and NASA during RAPID + TCT 2024 in Los Angeles, California from June 25-27, 2024.
