Quantifying and Reducing Uncertainty in Metal-Based Additive Manufacturing Laser Powder-Bed Fusion Processes

Abstract

Laser Powder-Bed Fusion processes capable of processing metallic materials are a set of relatively new and emerging Additive Manufacturing technologies that offer attractive potential and capabilities (e.g., design freedom, part consolidation and reduced material waste). Although they provide an exceptional advantage that cannot be matched by other traditional manufacturing processes, the path to widespread use of these new technologies still include some obstacles due to the limited understanding and intricate problems that the manufacturing process presents, such as low repeatability and low part quality compared to their conventional manufacturing counterparts. This dissertation presents one of the first applications of different formal tools and frameworks from a combination of scientific fields including Uncertainty Quantification, Statistics, Probability and Data Science, into different problems within Additive Manufacturing Laser Powder-Bed Fusion processes. Specifically, modeling techniques such as Gaussian Processes and generalized Polynomial Chaos Expansions are employed to optimize porosity in printed parts, calibrate and validate different computer simulation models, and identify processing regions for satisfactory manufacturing. Proper analysis of these techniques is undertaken and its validation is successfully presented such that informed and knowledgeable perspectives about the manufacturing process are gained to better understand it. In turn, these new insights and understanding translate into improvement and advancement of Additive Manufacturing, and contribute towards its further growth and consolidation as a competitive and qualified technology within the manufacturing industry.