Comparative Assessment of Physics-Based Computational Models on the NIST Benchmark Study of Molten Pool Dimensions and Microstructure for Selective Laser Melting of Inconel 625

This work presents a comparative assessment of different computation models with considering varying degrees of physics for the challenges within the National Institute of Standards and Technology (NIST) Additive Manufacturing benchmark problem AMB2018-02. The melt pool geometry, cooling rates, and dendritic microstructure of the single laser scan tracks on bare Inconel 625 plates are predicted by three types of computational models, namely the high fidelity welding model, fluid model, and conduction model for two cases without and with the formation of keyholes. The molten pool geometry in terms of its depth, width, and length as well as the cooling rates at the surface is used for comparing simulated results of various approaches against the NIST experimental results from the two testbeds, which are referred to as the additive manufacturing metrology test bed and commercial build machine (CBM) cases. A comparison of the spatial distribution of cooling rates is also presented to illustrate the importance of using a high fidelity welding model. The thermal gradient and the growth rate of the solid-to-liquid interface are used to predict the primary dendrite arm spacing. It is identified that the high fidelity welding model played a pivotal role in achieving accurate predictions of the CBM cases. The CBM cases with a higher laser energy density resulted in keyhole formation, which led to a high aspect ratio of the molten pool shape. Neglecting the keyhole model leads to large under-predictions of the molten pool depth. Additionally, the correct primary dendrite arm spacing prediction of the CBM cases is only possible with the keyhole model.