Advanced Flaw Detection for Additive Manufacturing

Oak Ridge National Laboratory (ORNL), Oak Ridge, Tennessee, researchers have reportedly improved flaw detection in laser beam powder bed fusion (PBF-LB) metal parts through a method that combines post-production inspection of the manufactured part with data gathered from sensors during the build process.

The combined data teaches a machine-learning algorithm to identify flaws in the product. “We can detect flaw sizes of about half a millimeter — about the thickness of a business card – 90% of the time,” shared ORNL researcher Luke Scime. “We’re the first to put a number value on the level of confidence possible for in-situ flaw detection.”

Scalable in-situ non-destructive evaluation of AM components using process monitoring, sensor fusion, and machine learning are the results of the process they developed. This process reportedly achieved a 90% detection rate and reduced the likelihood of false positives, thereby preventing the unnecessary discarding of good products.

Next, the ORNL team intends to train the deep learning algorithm to improve differentiation between types of irregularities and to categorize flaws with multiple causes.