PSI - Issue 7

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

63

6

Author name / Structural Integrity Procedia 00 (2017) 000–000

3.2. Defect Analysis The presence of defects commonly detected in samples produced via additive manufacturing and its study is crucial for understanding the behaviour of the parts produced and for predicting their mechanical performance (Romano et al. (2017)). As such, XCT measurements were executed to one representative sample per process parameters combination; these specimens were not mechanically tested. Subsequently, the non-destructive study of the defect population of the parts was performed, namely of voids and inclusions. The overall pore distribution in the same volume in each of the analysed specimens are presented in Fig. 4. These images represent a series of slices of the samples, and not the pores present on a single layer. The values of porosity percentage obtained with the associated software are also shown in Fig. 4, for each of the scanned samples identified by the parameter ID. Fig. 4 shows the presence of a large number of pores with a wide size distribution, which ranges from 70 µm to 800 µm for the samples produced with a layer thickness of 90 µm, and from 70 µm to 200 µm in the case of the samples manufactured with a layer thickness of 30 µm. The values obtained for the pore size depend on the resolution of the performed XCT scan (~18 µm) limiting the smallest detectable dimension, in this case 54 µm. On the other hand the maximum size could be impacted by the restriction with the defect analysis software in distinguish individual pores located in short distances of each other. This means that the largest defect identified by the software could be a cluster of pores. In addition, one may conclude that there is no preferred location of the pores over the sample diameter. However, it is evident that all the samples manufactured with a 30 µm layer thickness (ID 2, 7, 8, 9) showed a significantly lower number of pores, which was confirmed by the porosity values calculated (0.02%-0.04%). This observation supports the fatigue testing results, where the milled samples prepared using this combination of process parameters were considered to be run outs. With the correlation of the XCT results and the mechanical testing, one may concluded that the presence of large defects has a strong impact on the fatigue properties of the AlSi10Mg specimens, as reported in literature (Romano et al. (2017); Read et al. (2015); Greitemeier et al. (2016)). However, the effect of the presence of large defects is surpassed by the influence of the surface condition, as the fatigue behaviour of the net shaped samples manufactured with a 30 µm layer thickness (ID 8 and 9), is significantly poorer than the corresponding milled specimens. These results are in accordance with the literature, as reported by Brandl et al. (2012), Aboulkhair et al. (2016); Mower and Long (2016) and Greitemeier et al. (2016).

Fig. 4. Porosity distribution obtained via X-ray CT, represented by an overlapping image of the voids present in the same volume in each specimen. This image represents a series of slices of the samples, and not the pores present on a single layer.

3.3. Fracture surface analysis The fracture analysis was conducted on two samples per manufacturing condition, namely the ones that witnessed the lowest and the highest number of cycles to failure. Fig. 5 to Fig. 7 show the fracture analysis of the samples that survived the least number of cycles from the parameters in ID 6, 7 and 8, respectively. Figures (a) show the fatigue fracture surface and figures (b) highlight the defects which originated the failure in each specimen. Three different regions could be identified in the fatigue fracture surface of all the investigated samples of 11 conditions † ; (I) an initiation point, (II) a smoother fatigue crack propagation area (slow fracture), and (III) a fast fracture region, after the crack reached a critical size, leading to failure. All these stages are characteristic for a fracture surface of failure by fatigue (Aboulkhair et al. (2016); Engel and Klingele (1981)). The described regions are

† The fracture surface of samples from group ID 2 could not be investigated, as the tested samples did not fail under the test conditions (runouts).