Processing parameters, alloy composition and part geometry are all key components to metal additively manufactured microstructures. Even for a given alloy and manufacturing technique, local geometrical features of the printed part or component may be sufficient to cause local changes of solidification conditions (e.g,. temperature gradient and cooling rate), which in turn may lead to gradients of microstructures. The objective of our work is to develop a physics-based modeling approach to simulate and predict location-specific microstructural heterogeneities using a combination of geometry-aware fast thermal model and phase-field (PF) model of microstructure formation during (potentially far-from-equilibrium) solidification. The thermal model is based on the NRL Enriched Analytical Solution Model (NEASM) [1]. The PF modelling approach, applied in various regions of the part using thermal conditions provided by the NEASM, is based on the relaxation of the partition coefficient at the solid-liquid interface [2], which relaxes the equal diffusion potential (EDP) condition of conventional PF models. This PF model is developed in user-friendly Julia and parallelized in GPUs for performance improvement [3]. Combining these methods opens the way to the prediction of microstructure heterogeneities due to part geometry (e.g. in presence of corner, edges, etc.) in fusion-based additive manufacturing processes.

References

[1] J.C. Steuben et al. Additive Manufacturing, 2019, 25, 437-447.

[2] S.G. Kim et al. Computational Materials Science, 2021, 188, 110184

[3] J. Bezanson et al. SIAM review, 2017, 59.1, 65-98.