Abstract

Metal laser powder bed fusion (LPBF) is an additive manufacturing (AM) technology applicable for manufacturing of structurally loaded components. As with many industrial applications, ensuring the safe use of the 3D printed component during service life is a strong requirement. As such, durability analysis is performed to calculate the fatigue life for the given load conditions and to validate the design. However, conventional durability analysis does not consider the local material properties and defects resulting from the AM process and as such cannot predict accurately the fatigue performance of the 3D printed part. With LPBF the component is built layer by layer, having a laser scan and weld the metal bead by bead. This process leads to varying local material properties resulting from the AM induced local conditions like microstructure, porosity content and surface roughness. To predict accurately the fatigue life of a 3D printed metal component one needs to characterize the impact of the above-mentioned local conditions on the fatigue property of the material and to efficiently consider these local fatigue properties in the part scale durability calculation. This poses a rather significant challenge as there are endless combinations of local conditions and post treatments (for example surface and heat treatments). Performing a solely test based characterization would lead to a very large and very expensive test campaign. The solution to this problem proposed by the authors relies on a Machine Learning based material model able to predict the effect of any combination of AM induced local condition on the fatigue property of the material. Combined with an enhanced durability solver the solution can efficiently consider the AM induced local properties on the part scale and provide a more accurate fatigue analysis. The authors acknowledge SIM and VLAIO (Flanders, Belgium) for funding the “M3-FATAM” (HBC.2016.0446) project (M3 research program).