A new method for analyzing and monitoring data collected during powder bed fusion 3D printing allows for detection of flaws and anomalies early in the build process.

The in-situ monitoring setup developed by researchers from U.S. Oak Ridge National Laboratory (ORNL) and RTX Technology Research Center includes a high resolution visible light camera with a spatial resolution ranging from 50 microns to 75 microns, depending on the printing configuration. A near infrared camera supplies information on the thermal behavior experienced during the welding process of each 2D layer. A machine learning tool with cameras identifies anomalies and defects, and operators can add observed anomalies back into the training database for the machine learned tool to recognize in future builds.

The technology described in Additive Manufacturing offers a 90% detection rate while reducing the probability of false positives, which can lead to scrapping acceptable products. Tests were conducted with readily available 3D-printed designs and the camera setup, after which quality inspections were performed using X-ray computed tomography (CT).

The data was then aligned into a layered stack of images to inform the machine-learning algorithm. After flaws were initially identified by the algorithm using the CT scan images, the researchers annotated the rest based on visual cues in data collected during the printing process. Human feedback continues to train the software, so the algorithm recognizes flaws more accurately each time. Over time, this is expected to reduce the need for human involvement in manufacturing inspection.


To contact the author of this article, email shimmelstein@globalspec.com