PSI - Issue 38

Hugo Roirand et al. / Procedia Structural Integrity 38 (2022) 149–158 Author name / Structural Integrity Procedia 00 (2021) 000 – 000

156

8

This result should be nuanced because only pores near surface take part to the fatigue crack initiation process. Moreover, this size is also depending of the loaded volume size because there is an important scale effect for fatigue crack initiation for materials with defects (Andreau et al., 2019b; Khoukhi, 2021). The low fatigue resistance of SLM_Chess specimens correspond to their high number of large defects as it has been revealed by the tomographic analysis. It has been confirmed by SEM fracture surface analysis which revealed that crack has always initiates from lack of fusion closed to the surface (Figure 7). Moreover, the lacks of fusion observed in tomography have their maximum characteristic length oriented perpendicularly to the loading. This is the worst configuration for fatigue resistance because it increases the local stress concentrations and it reduces the area supporting the loading. Concerning Orlas specimens, the fatigue lifetime results surprisingly show that all samples tested broke around 400.000 cycles for maximum stresses covering the 140-260 MPa range. Considering Kitagawa approach, this seems to indicate that specimens loaded at low maximum stresses initiated on bigger lacks of fusion than those cycled at higher maximum stress. This has been checked by fracture analysis (in SEM) and no correlation between the crack initiating lack of fusion sizes and the maximum stress has been found. More fatigue tests must be done in order to further understand this behaviour. However, even if the regression has a sloping look, it is in accordance with the defect study in comparison with SLM specimens. Indeed, the fatigue lifetime is lower for Orlas specimen than meander strategies and higher than chessboard strategy. This correspond to the opposite evolution of defect diameter and density as it was expected (Andreau et al., 2019b; Garlea et al., 2019; Ronneberg et al., 2020).

Fig. 6. Fatigue resistance of specimens built with different scan strategies. S-log(N) graph.