A line-based flash heating method for numerical modeling and prediction of directed energy deposition manufacturing process

The part-scale directed energy deposition (DED) process is often accompanied by sizeable thermal stress and deformation. It is of great significance to predict the residual deformation and stress of the large component produced by DED. In this paper, a line-based flash heating (LFH) method was proposed and used as the multi scale law in the thermo-mechanical model of the DED process. The building parameters of the DED process, such as bead dimensions, overlap, and scanning strategies, can be well considered. A series of experimental validation was carried out. Model sensitives and essential resolution, including length scaling factor and bead scaling factor, were identified and characterized. The whole processes of depositing two part-scale samples with different scanning strategies were simulated. Different scaling strategies, including the hybrid scaling strategy, were used in the simulation for the 60-layer model. The results between the simulation and experiments are in good agreement. The recommended critical values of length scaling factor and bead scaling factor are 312 and 0.2, respectively, when defining the critical value of prediction error as 10%. For the two part-scale models, the prediction error of the maximum residual deformation is less than 10% when using the bead scaling strategy with recommended scaling factors. Compared with the layer scaling technique, the hybrid scaling strategy improves the accuracy of prediction and controls the cost of calculation. This work provides a flexible multi-resolution framework for modeling the DED process and an important tool for further understanding the complex thermo-mechanical behaviors in the large DED component.

Automated summary

The study introduces a line-based flash heating method and incorporates it into the thermo-mechanical model of the DED process. Experimental validation was conducted, identifying model sensitivities and essential resolutions. It is found that the hybrid scaling strategy improves prediction accuracy and controls computational costs.