Experiment-based superposition thermal modeling of laser powder bed fusion

Parts experience significant local thermal variations during the Laser Powder Bed Fusion (LPBF) metal Additive Manufacturing (AM) process, providing a potential source of defects. Near real-time thermal predictions can enable better process planning and facilitate corrections on subsequent layers to enable the engineering of laser parameter and scan path combinations that avoid defect inducing scenarios. This paper considers an experiment-based Discrete Green’s Function (DGF) thermal model for temperature field prediction in LPBF. An analytical framework is developed and used to calculate an experimental DGF (i.e., powder bed’s single pulse temperature response) from spatiotemporal Short-Wave Infrared (SWIR) camera data. The extracted DGF is superimposed along a laser scan path to predict the future temperature history. Experimental results show the superposition model accurately predicts a rectangular layer’s temperature history (uncorrected for emissivity) with an 8 % average percent error. The model’s prediction of the temperature history and thermal features are shown to be consistent for various laser powers, laser exposure times, laser raster vector lengths, and scan path rotation angles. The superposition predictions slightly deviate from the experimental results where the laser corners in-layer, when high exposure times are used, and if there is scanning with short raster vectors. These deviations are attributed to evaporative cooling causing the experimental temperatures to saturate. There is the potential to reduce this error in future work by developing a higher dimensional DGF where the DGF functions explicitly account for those boundary conditions. Overall, the experiment-based DGF model demonstrates a strong potential for applications in feedforward correction of thermally driven LPBF process errors and baselining measurements from in-situ part qualification frameworks.