Meso-scale Grain Structure Prediction Models for Metallic Additive Manufacturing Processes

The emergence of additive manufacturing (AM) in recent decades marks a paradigm shift in metallic component fabrication. Parts are generated layer upon layer, by selectively melting and re-solidifying material with a high energy density heat source. The localised and intense nature of this heat source establishes high thermal gradients at the melt pool boundary, which promote columnar dominant modes of solidification. Nucleation in the bulk liquid is rare, making the columnar-to-equiaxed transition (CET) difficult, but not impossible to achieve. Since the material passes through multiple heating and cooling cycles, some of which will completely or partially melt the previously deposited layer(s), grain structure evolution must be tracked dynamically. In this work, an envelope cellular automata (CA) based solidification model is developed, which efficiently models the nucleation and competitive growth of large numbers of individual metallic crystals in response to thermal conditions. The stochastic and temperature dependant nature of nucleation events is accounted for, enabling the direct prediction of the CET. Crystallographic orientations of nuclei and grains in the substrate are set to be random, so that the grain structure that develops is solely a response to the thermal conditions of the process. The model was first verified against an extensively validated front tracking (FT) model for Al-Cu castings, and provided deeper insights into the effects of alloy composition and heat extraction rates on CET. AM-specific conditions were then applied, accounting for the near-rapid solidification conditions through the use of the KGT dendrite tip kinetics model, and a conduction-based melting model with a moving heat source. Simulations were extended to multiple layers, employing an analytical thermal model for computational efficiency, and keeping track of the grain structure as it was partially remelted over subsequent layers. The hallmarks of AM grain structures were successfully predicted, including epitaxial growth inwards from the edge of the melt pool, long columnar grains spanning multiple layers with a strong [001] texture orientation towards the build direction, and an aversion to CET. The model was then externally coupled to the thermo-fluid powder melting model of Parivendhan which accounts for many of the complex melt pool phenomena. The resulting grain structure predictions compared well with real-world AM micrographs produced under similar energy densities, demonstrating the capabilities and utility of the model.