| Abstract Scope |
The morphology of metal feedstock powders used in prevalent metal additive techniques can directly affect powder feeder or powder bed flow, thus impacting part properties and ultimate part performance once printed. Common powder morphology characterization techniques, such as camera-based dynamic image analysis, are either limited by resolution, lack surface feature detection, assume only spherical particles, or measure derived parameters. Ultimately, these techniques struggle to identify powder parameters key to the performance of additively manufactured components, such as the presence of satellite particles and irregular particle shapes. To improve data collection of these parameters, this work compares dynamic image analysis morphology values of a given powder versus scanning-electron micrographs of the same powder, processed through a computer vision model, trained to identify these key parameters. |