Abstract
The French Alternative Energies and Atomic Energy Commission (CEA) is currently estimating the industrial potential of Additive Manufacturing (AM) processes. Particularly, the construction of stainless steel components, for nuclear applications, using the Laser Beam Melting (LBM) technology is evaluated. Anisotropic mechanical behaviour is frequently observed in parts manufactured by LBM. This is directly linked to the component’s grain structure’s characteristics, which themselves are dependant on the LBM process’ parameters. In this context, the formation of grain structures during single line 316L stainless steel samples has been investigated. We focus on the different phenomena identified during solidification. To this end, an approach combining experiment, numerical modelling and simulations is adopted. Two experimental studies were performed. The first consisted in building single-layer tracks and melt-runs on an instrumented LBM bench, that enabled the observation of the formation and evolution of the melt pool during the process. The second consisted in building geometries composed of multi-layer melt-runs, for the study of grain behaviour undergoing fusion-solidification cycles. The metallographic characterisation of the produced samples along with high-speed camera recordings during the tests allowed for the determination of grain characteristics and melt-pool sizes and geometries. Moreover, several process parameter combinations were used, and so the relations between operating parameters, heat transfer mechanisms and generated grain structures are studied. Additionally a numerical model of grain structure formation was implemented and developed, incorporating the main physical phenomena identified throughout our experimental investigations. The numerical model is based on a three-dimensional “CAFE” model, which couples Cellular Automata (CA) and Finite Element (FE) simulations. The thermal model defined is purely conductive and was calibrated using instrumented experiments, and validated by comparing the size and shape of the numerical and experimental molten zones. Finally, numerical grain characteristics resulting from the simulations are compared to the experimental ones, and the numerical results are discussed as a function of the different implemented growth models and tested process parameters.