Abstract Computational models are extremely helpful tools to assess and improve further understanding of process-structure-properties relationships in manufactured parts, especially through laser-based additive manufacturing (AM). For instance, when combined with experimental work, computational models have demonstrated how the melt pool’s morphology in laser-based AM processes such as laser powder bed fusion, are found to be intrinsically related to part’s microstructure and mechanical properties. In-situ data such x-ray imaging for melt pool visualization is often used for calibration in these models, allowing for higher accuracy and better predictability of physical phenomena present in AM processes, especially concerning melt pool depth and length. Less time-consuming and location-restricted techniques are needed to reach the plethora of materials, printing parameters, and respective combinations for data collection and computational model validation. Additionally, non-biased and automated image analysis tools are needed to process the in-situ results. This work provides a computer-aided image analysis tool that extracts melt pool and denudation zone information such as width, length, and the laser position, obtained from a custom system with in-situ high-speed imaging of laser scans that allows for top-view imaging of the melt pool dynamics and laser location. This software tool is based on wavelet transforms for the melt pool identification and provides unbiased investigation of imaging data that has the potential to serve as input for computational modeling calibration and future applications such as machine learning in decision-making of parameters ‘on-the-fly’.
Kublik et al. (Mon,) studied this question.