Modeling of hierarchical solidification microstructures in metal additive manufacturing: Challenges and opportunities

Metal-based additive manufacturing (AM) processes often produce parts with improved properties compared to conventional manufacturing and metal working routes. However, currently, only a few alloys can be reliably additively manufactured as the vast majority of the alloys in use today are not explicitly designed for this manufacturing route. This is because the highly non-equilibrium nature of melting and rapid solidification phenomena during AM leads to undesirable microstructures with complex growth morphologies and unpredictable microstructural inhomogeneities including solidification defects, leading to unwanted variability in final material properties. In this context, the review article discusses the underlying physical mechanisms of microstructure and associated defects formation during ultrarapid cooling rates typical of AM in order to suggest approaches to minimize and control microstructural heterogeneities for improved printability and microstructure robustness (and hence properties). In particular, the physical effects of cooling rates and alloy parameters on rapidly moving complex solid-liquid interface shapes and the nucleation behavior during non-steady thermal conditions in heterogeneous liquid during AM must be well-understood to control the solidification microstructure and grain morphology. Suitable integration of physics-rich macroscale melt-pool, microstructure, and atomic-scale nucleation models (but benchmarked by experimental measurements) could quantitatively simulate the above hierarchical AM solidification problems that extend across multiple length scales and associated chemical heterogeneities. To address the tremendous computational expense of the above solidification problems toward large part-scale or full-melt-pool simulation, exascale computing hardware and software has been leveraged as a part of the exascale computing project. Further, the AM solidification simulations would guide parameter-microstructure optimization via data-driven modeling and, ultimately, alloy and processing development to suit various envisaged applications in the AM community.