PSI - Issue 7

Ana D. Brandão et al. / Procedia Structural Integrity 7 (2017) 58–66 Author name / Structural Integrity Procedia 00 (2017) 000–000

64

7

approximately marked in Fig. 5a, Fig. 6a and Fig. 7a. The extension of the flat region on the fracture surface was larger for the samples that survived a higher number of cycles, as the crack propagation in fatigue is an incremental process. After this stage, a catastrophic failure is observed. In addition, none of the samples showed evidence of ductile fatigue fracture behaviour, as the presence of dimples on the fracture surface was not observed. Similar observations were reported by Brandl et al. (2012) for fatigue fracture surfaces of as-built specimens and by Aboulkhair et al. (2016) for samples subjected to higher stress levels. In the case of the milled specimens (see Fig. 5), it was observed that the failure started in areas of lack of fusion. These defects were also detected as the origin of failure in net shaped specimens. However, in these samples a combined effect between lack of fusion and a crack initiation on the surface was often observed (see Fig. 6 and Fig. 7). It is assumed that this was caused by the relatively high surface roughness of the net shaped condition. In addition, a significant scatter was observed on the fatigue data, especially for the milled specimens. To better understand this observation, the size and location of the defect that caused the fatigue failure of the milled specimens was investigated. The defect size ranged from 143 μm (ID 3), to 674 μm (ID 5), with distances to the surface of up to 1430 μm (ID 5). This variation in the defect characteristics may explain the scatter perceived in the acquired data. It was observed that the largest defect visible on a fracture surface did not necessarily lead to the crack initiation. In some samples minor defects which were located closer to the surface were the origin of the crack which lead to failure. This shows the relevance of the size / location binary, as it is expected for a failure in fatigue. Taking into account all these observations, the characterization of the defect population of an AM part is considered to be critical in the design of its fatigue requirements. Nevertheless other defect parameters should be included in future characterizations, such as shape and volume of voids and pores.

Fig. 5. SEM images of a vertically printed sample, with a 90 µm layer thickness, milled (ID 6 of Table 1), showing an overview of the fracture surface (a)), and the defect which caused the failure during the investigation (b)). This sample survived the lowest number of cycles from group ID 6.

Fig. 6. SEM images of a vertically printed sample, with a 30 µm layer thickness, net shaped, vibratory polished, and built without contour parameter (ID 7 of Table 1), showing an overview of the fracture surface (a)), and the defect which caused the failure during the investigation (b)). This sample survived the lowest number of cycles from ID 7 .