EUROPEAN CONGRESS AND EXHIBITION ON ADVANCED MATERIALS AND PROCESSES - EUROMAT 2021

Among promising approaches to accelerate the discovery of novel alloys and processing routes, Integrated Computational Materials Engineering (ICME) strongly relies on coupling different modeling techniques, relevant to different length/time scales and/or different physics. While a broad range of models have been developed, key challenges remain in the efficient coupling between these different models.

Here, we present a framework for the simulation of microstructure formation during selective laser melting (SLM) of nickel-based superalloys. It couples three main components. Thermophysical data (e.g. phase diagram features, heat capacities, etc.), used as input to other models, is computed using the CalPhaD approach. Finite elements (FE) are used at the macroscale to compute the temperature field under different processing conditions. Finally, using the FE-calculated temperature field, phase-field (PF) modeling is applied at the scale of the melt pool to simulate microstructure formation.

Focusing on PF, well-converged quantitative simulations remain computationally challenging at that scale, even in 2D, which explains why state-of-the-art simulations typically focus on narrow regions of the melt pool. In this work, to enable simulations at the scale of the melt pool, we developed a multi-GPU parallel implementation, combined with various techniques to improve computational efficiency (e.g. nonlinear phase-field preconditioning and simplified polycrystalline representation).

While prospective pathways for the expansion of the current framework are multiple (e.g. related to multicomponent alloys or solute trapping), this first brick into the edifice of a full ICME solution for metal SLM already allows capturing the effect of different process conditions (e.g. laser power and scan velocity) upon microstructure characteristics (e.g. primary dendritic arm spacings and grain textures). First results of the proposed computational tool are compared to available experimental data for SLM of Inconel 718 and Hastelloy-X superalloys.