Comprehensive investigation and quantification of quality repeatability for laser powder bed fusion processed 316L stainless steel

• Main Author: Huang, De Jun
• Other Authors: Li Hua
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Manufacturing::Quality control; Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence; Engineering::Materials::Material testing and characterization
• Online Access: https://hdl.handle.net/10356/155240
• Institution: Nanyang Technological University

Laser Powder Bed Fusion (L-PBF) additive manufacturing has gained traction in several industries as a result of the recent progress in machine and material development. However, the repeatability of part quality remains as a bottleneck for large-scale implementations of this technology to make economic sense in the industry. This research aims to effectively investigate and quantify quality repeatability in L-PBF process in order to drive better adoption and make repeatability studies more accessible in the future. Three major studies were conducted to identify, understand and track the important L-PBF parameters beyond the classical ones to the process repeatability. In the first study, data mining and in-depth analyses were performed to simultaneously investigate the effects of production parameters to L-PBF repeatability. A combinational effect from the sample location and post-chamber pressure drop was identified to be highly correlated to the mechanical properties of printed samples. By analysing the printer design and sample fractography, post-chamber pressure drop was revealed to be closely related to the gas flow condition in the build chamber. In the second study, the print location and gas flow speed were stretched to the printer's limits in an experiment to understand their impacts to the printed material. It was found that the gas flow speed plays a primary role in porosity formation and the resultant tensile properties while the print location plays a secondary role in affecting the grain direction of the material. These observations can be attributed to the gas flow condition at different regions of the build chamber which affect the spatter removal rate and melt pool formation. In the third study, an in-situ spatter monitoring system was developed to quantify the gas flow-induced repeatability in L-PBF. Spatter images were analysed with deep learning to characterise the important spatter features that are indicative of the surrounding gas flow condition. A spatter scoring system was derived to express the desirability of a spatter which was proven to be a good representation of the resultant tensile properties of printed samples. Overall, the three studies presented in this thesis contribute to the understanding and quantification of quality repeatability in L-PBF. While the material analyses provided evidence of the gas flow- and location-related effects to the process repeatability, the data-driven approaches substantiated such understanding and provide new insight objectively in a big-data context.