Defect Detection Model Development for Large Scale Thermoplastic Printing

Large-format additive manufacturing (LFAM) offers several advantages, including high throughput, cost-effective pellet-fed extrusion, and the capability to produce large-scale structures. The main pain points of LFAM include start and stops during the printing process, warpage, long layer times that lead to bead freezing, and bead separation due to shrinkage. These issues can lead to overfill, underfill and buildup of material in different sections of a print. This can lead to hidden defects embedded within the printed layers, or even ultimate failure of the printed structure. This ensures these defects can only be identified through nondestructive testing (NDT) inspection methods after printing, which can be timely and costly. Aligned Vision work specializes in 2D projectors with visual inspection systems and machine learning. Traditionally system is used for composite layup and layup inspections. In this work we used the LFAM system at Oak Ridge National Laboratory to create defect rich samples. The Aligned Vision inspection system then performed in-situ monitoring of the print process after each part was printed. This in-situ vision inspection system was used to develop a layer-by-layer inspection model that looks for overfill, underfill, and the buildup of defects using only a camera-based vision system. This leads to the assurance of high-quality production components.