Optical process monitoring in Laser Powder Bed Fusion using a recoater-based line camera

Abstract

The multitude of influences on the Laser Powder Bed Fusion (LPBF) process and local deviations from the intended process conditions can lead to the occurrence of quality deviations. The cyclic nature of the process could enable in-situ process monitoring systems to detect those deviations and to reduce the need for post-process quality control or even prevent defects by utilizing the extracted data for closed-loop or intermittent control. Most existing approaches are focused on the thermal observation of the melt pool or use matrix cameras to view the solidified material from fixed positions. This work presents an approach for in-situ process monitoring by leveraging the process-inherent recoater movement to obtain highly resolving images of the complete process area with a recoater-based line camera. A monitoring system is designed and integrated into a LPBF-machine. The system is calibrated and benchmarked to obtain images with a width of 97.67 mm at a resolution of 5.97 µm/px and a recoater speed of up to 100 mm/s. Details as small as 12.40 µm can be identified according to a USAF1951 test and the focal depth is sufficient to encompass the whole process area (part surface & powder) of a typical LPBF process. Samples with varied conditions are produced and layer-wise images are acquired to evaluate the potential of the system under actual process conditions. Features relating to the brightness and periodicity of the images are defined and extracted from the image data. The relationships between the extracted features on the one hand and the laser power, scanning speed, layer thickness, shielding gas flow and laser defocus on the other hand are investigated and correlations are found. Overall, the findings indicate a high potential of the investigated system for the detection of deviations in the LPBF process and its future application for in-situ quality control. • A recoater-based line camera is integrated into an LPBF machine and benchmarked. • 12–14 µm small details are identified across 98 mm for a recoater speed of 100 mm/s. • The image data is correlated with variations of the process conditions.