AbstractNickel-based superalloys are widely used in applications requiring resistance to high temperatures and high strain rates. Various additive manufacturing (AM) processes, such as Laser Metal Deposition (LMD), a Directed Energy Deposition (DED) process, can be used to produce these components. The quality of the components depends on the process parameters, so it is crucial to investigate the influence of each parameter and their combinations through extensive experimental campaigns. In this scenario, it would be very important to predict the mechanical properties of the produced components through the online monitoring of the process parameters using non-destructive techniques, such as thermography. The aim of this work was to explore the feasibility to predict the mechanical properties of Inconel 718 thin wallets around 10 mm produced by DED-LB, based on the extraction of suitable thermal features directly during the production. An experimental campaign analysed the effect of different process parameters (laser power, scan speed, powder flow rate, and energy density) on the mechanical properties achieved. All sample production was monitored with an infrared uncooled camera integrated with the laser head moving at the same scan speed. After the process, hardness measurements and tensile tests in both growth directions were carried out for each sample to evaluate the mechanical behaviour of the "as-built" coupons and the influence of selected process parameters. Macrographic analyses of the material structure were performed to determine the morphology of the passes and the degree of overlap between different passes and layers. Various thermal features and statistical models were considered to demonstrate the possibility of establishing a predictive model. The obtained results demonstrated the correlation between the hardness and the apparent temperature assuming a confidence level of 95%, and the possibility of predicting in this sense the final macrostructure and the mechanical behaviour of the printed material considering an empirical model with the R2 coefficient around 0.8.
Read full abstract