Abstract Scope |
Minimizing the percentage of porosity within fabricated samples has long been a primary optimization target in laser powder bed fusion (LPBF) and other AM processes. Due to improvements in system control and parameter development, parts are now consistently built with >99.5% density. However, in many situations, this baseline density metric does not adequately summarize process quality. In this work, we present an automated tool for porosity characterization and classification. The end-to-end processing pipeline ingests micrographs of sample cross-sections, identifies and indexes individual pores, extracts geometric properties for each flaw, and classifies each defect according to its formation mechanism using a trained machine learning model. We will also demonstrate the utility of the tool through three use cases involving Ti-6V-4Al: process mapping, evaluation of a novel scan strategy, and automated identification of rogue defects. |