Influence of defects and as-built surface roughness on fatigue properties of additively manufactured Alloy 718

2018 
Abstract Electron beam melting (EBM) and Selective Laser Melting (SLM) are powder bed based additive manufacturing (AM) processes. These, relatively new, processes offer advantages such as near net shaping, manufacturing complex geometries with a design space that was previously not accessible with conventional manufacturing processes, part consolidation to reduce number of assemblies, shorter time to market etc. The aerospace and gas turbine industries have shown interest in the EBM and the SLM processes to enable topology-optimized designs, parts with lattice structures and part consolidation. However, to realize such advantages, factors affecting the mechanical properties must be well understood – especially the fatigue properties. In the context of fatigue performance, apart from the effect of different phases in the material, the effect of defects in terms of both the amount and distribution and the effect of “rough” as-built surface must be studied in detail. Fatigue properties of Alloy 718, a Ni-Fe based superalloy widely used in the aerospace engines is investigated in this study. Four point bending fatigue tests have been performed at 20 Hz in room temperature at different stress ranges to compare the performance of the EBM and the SLM material to the wrought material. The experiment aims to assess the differences in fatigue properties between the two powder bed AM processes as well as assess the effect of two post-treatment methods namely – machining and hot isostatic pressing (HIP). Fractography and metallography have been performed to explain the observed properties. Both HIPing and machining improve the fatigue performance; however, a large scatter is observed for machined specimens. Fatigue properties of SLM material approach that of wrought material while in EBM material defects severely affect the fatigue life.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    38
    References
    48
    Citations
    NaN
    KQI
    []