Laser Powder Bed Fusion (LPBF) of a Complex Aerospace Component: Effects of Layerwise Process Control on Part Quality

Abstract

This research concerns the laser powder bed fusion (LPBF) metal additive manufacturing (AM) process. The goal is to optimize critical processing parameters for a complex LPBF part. Currently, LPBF process qualification relies on empirical parameter experiments and rigorous materials characterization. Once applied to end-use, complex geometries, this process can take over 2 years of engineering effort and more than $4 million dollars. Further, empirically-determined process parameters often do not scale to complex parts without expensive supporting structures required to maintain a stable manufacturing process. During the LPBF process, the part is subjected to continual heating and cooling cycles. This thermal history governs every aspect of the part's quality, from its microstructure, defects, geometric accuracy, residual stresses, and ultimately, determines its functional properties. Hence, controlling the thermal history is critical for obtaining high quality LPBF parts. Accordingly, the objective of this work is to evaluate part quality, thermal history, and support structure optimization of process-controlled parts to compare against their nominally printed counterparts. This approach leverages a feedforward thermal history control algorithm known as DynamicPrint for LPBF of a complex aerospace component (GE bracket) made from Inconel 718 material. This approach, which is a form of model predictive feedforward control, uses a physics-based model to autonomously adjust processing parameters layer-by-layer before the part is printed to avoid heat build-up and thermal history excursions during printing. This research tests and affirms the hypothesis that controlling the thermal history via layerwise modulation of laser power and velocity improves part quality. This thesis evaluates and compares the quality of thermal history-controlled LPBF parts against parts printed with empirically optimized, constant process parameters. Compared to uncontrolled parts, process-controlled parts show improved dimensional accuracy and reduced thermal-induced warping. Specifically, compared to parts produced under fixed, nominal conditions, layerwise control eliminated distortion defects and reduced necessary support structure volume by 45%. This work demonstrates a scalable approach for applying layerwise thermal history process control to a complex geometry, where a key challenge in practical LPBF applications lies in dynamically modifying process parameters to accommodate complex geometries. Notably, this method enables rapid, autonomous optimization of process conditions in LPBF – critical to accelerated part qualification – and reduces process development time from months to weeks, while fabricating better quality parts that cost less.