PSI - Issue 57

David Mellé et al. / Procedia Structural Integrity 57 (2024) 61–72 David Melle´ / Structural Integrity Procedia 00 (2023) 000–000

66

6

Table 3. Sizes of critical surface features for each initiation mechanism on 30 min etched surfaces.

Min size µ m

Mean size µ m

Max size µ m

Surface valleys

91 38 95

-

195 190 292

Gas pores

85

Lack-of-fusion

165

them. The sizes of both lack-of-fusion and surface valleys are significantly decreased compared to the net-shape surfaces sizes for the same mechanisms. A Kitagawa-Takahashi diagram was drawn with these critical surface micro-geometric features and the El-Haddad and Topper (El Haddad et al. (1979)) model was identified. These points and this prediction, including the material strength variation range that enclose the critical points are shown in Figure 5.

Fig. 5. Kitagawa-Takahashi diagram of critical features, measured on SEM images, for net-shape and 30 min chemically etched coupons.

Considering the populations of surface micro-geometric features an in-house python script was used to isolate the surface valleys and surface connected porosities on the scans of the 30 min etched surfaces. This script was not run on net-shape surfaces as they were too rough and chaotic. The algorithm was not able to isolate the di ff erent features for this condition due to the fact that they are all connected to each other making it nearly impossible to spatialy delimitate them. Figure 6 gives an example of a scan of a 30 min etched surface with the delimited surface micro geometric features. The critical feature was determined in this population and SEM images based measurements were compared to the di ff erent metrics ( √ area , depth) extracted from the surface interferometry based measurements. It is easily understandable that for surface connected porosities with overhang areas the SEM measurements were signif icantly more accurate than the values determined from surface scans. In the following sections, when the full surface features population needs to be represented, the coupons for which these two di ff erent measurements methodologies are consistent will be the only one shown (9 coupons among the 16 valid experimantal points availables). As mentionned in subsection 2.5, several measured parameters are extracted from the surface scans. The first one is Murakami’s √ area parameter. Maximal depth and the stress concentration factor SCF are also determined. The