High speed imaging of molten pool free-surface hydrodynamics in laser-assisted directed energy deposition process clearly revealed a highly oscillatory and dynamic melt flow due to impinging powder particles. Surprisingly, most of the reported computational work exclude the injection of powder particles and rather adopt a homogeneous mass and energy addition approach, and therefore provides less accurate predictions. In this work, we develop a coupled multi-physics particle-scale approach utilizing the discrete element method for particle trajectory prediction, the computational fluid dynamics for free-surface thermo-fluidic modelling and the cellular automata method for grain growth evolution. In the model, the governing physical phenomena, such as laser-powder interaction, in-flight particle heating, phase change (melting, vaporization and solidification), free-surface evolution, molten pool hydrodynamics and impinging particles-melt interaction have been considered. Experiments for the deposition of Inconel-625 on an Inconel-625 substrate are carried out, and the model predictions are validated with the experimental measurements. For the first time, the predicted thermo-fluidic simulation results reveal highly oscillatory, chaotic and random melt flow attributed to the impinging powder particles. During the deposition, it is found that the role of the Marangoni convection is less significant as compared to the momentum imparted by the impinging powder particles in the melt pool. Using the simulated thermal undercooling data, cellular automata-based grain growth simulation predicts elongated columnar dendrites in the melt pool that grows epitaxially from the melt pool interface and stretches towards the centre. Using the Kurz-Fisher model, the effect of local thermodynamic solidification conditions on the size of dendritic microstructure is also described. The predicted melt pool geometry, temperature field and grain structure compare well with the experimental measurements.
Read full abstract