Towards data-driven reprocessing: Multisensory process observation and hardness modeling in high-speed directed energy deposition

The transition towards circular factories requires robust and adaptable manufacturing technologies capable of reprocessing components with highly variable conditions while ensuring consistent quality. High-Speed Directed Energy Deposition with laser beam (HS DED-LB/M) represents a promising approach due to its high productivity, low heat input, and suitability for reprocessing high-value components. However, complex thermo-physical interactions and limited process observability constrain its reliable industrial deployment. In this study, the relationships between process parameters, in-situ multisensory signals, and resulting material properties are systematically investigated for HS DED-LB/M using AISI 4140. A coaxial two color pyrometer is employed to derive temperature-based intermediate feature variables (IFV), which are correlated with hardness. High-speed imaging is used to analyze process dynamics, and Partial Least Squares regression evaluates the predictive capability of process parameters, temperature-based descriptors, and geometry-based descriptors. The results demonstrate high relative densities (>99.77%) and small heat-affected zones (<40 μm), confirming the suitability of HS DED-LB/M for reprocessing. ... mehrHardness decreases with increasing laser power, decreasing processing speed, and increasing build height, highlighting the dominant role of global thermal effects. While temperature-based IFV moderately improve hardness prediction, their interpretability is limited by signal superposition effects caused by powder–laser interaction and an oscillating vapor plume. Build height emerges as the most influential descriptor, capturing cumulative thermal effects and significantly improving prediction accuracy. Overall, the results demonstrate that reliable process understanding and prediction in HS DED-LB/M require combining process parameters, in-situ observations, and global thermal history, supporting data-driven, sensor-based approaches for scalable reprocessing in circular factory environments.