PSI - Issue 68

J. Köckritz et al. / Procedia Structural Integrity 68 (2025) 962–968 J. Köckritz et al. / Structural Integrity Procedia 00 (2025) 000–000

966

5

in the measurement area, S V , were not significantly influenced by the surface treatment, with S V,AB = 69.8 µm and S V,TW = 60.93 µm, which does not allow conclusions due to the large variance of the data with an interquartile range IQR of IQR AB = 29.3 µm and IQR TW = 25.6 µm. This scatter originates in the limited measurement area at a specific position, which does not necessarily include the largest defect in the AM component. The surface treatment was therefore not able to penetrate deep enough to remove all surface errors. However, other surface treatments (see chapter 1) also primarily remove surface peaks. Finally, effectiveness of any surface treatment highly depends on the quality of the AB state – if an alloy is challenging to print, like Al2139, large printing errors will be more likely. Such are then more difficult to fully remove and significantly affect fatigue performance. The components show a complex failure behavior. First, a small crack on the upper side of the FWA initiates and grows to a maximum of 10 mm length in the area of high-pressure stresses. This “pressure crack” crack stagnates. Then, a crack on the lower edge of the FWA is initiated and, after a period of stable growth, shows sudden crack growth through the thin wall, reaching a crack length of up to 40 mm. Once this “tension crack” significantly decreased the load-carrying cross section, the pressure crack again grows alongside the tension crack, which finally leads to component failure. For this evaluation, the tension crack and the corresponding increase in displacement during the fatigue tests are selected as preliminary criteria for failure. The tension crack paths are highlighted in Fig. 5. The fatigue life for AB and TW components at their tested force amplitudes are displayed in Fig. 3. The transparent scatter band represents 10 % to 90 % chance of failure, the arrow signifies a runout at 2·10 6 cycles. Results show no benefit of the surface treatment for the fatigue life in the high cycle regime of the components, even though R a , R z and S a were significantly reduced by the surface treatment. The AB S-N curve overall shows, unexpectedly, a slightly lower inclination and lower scatter with a logarithmic standard deviation s log,AB = 0.156, compared to the TW components with s log,TW = 0.182. AM parts usually show a large scatter of fatigue life (Beretta et al. (2022)), so this s log are well in the expectable range (DIN50100 (2022)). The unchanged fatigue behavior after the surface treatment can be explained by the voids, which could not be removed by the TW. It has to be noted though, that all FWA perform significantly better than can be assumed from the previous fatigue analysis, sometimes reaching 20x higher lifetimes.

Fig. 3: Fatigue life of front wing attachments with two different surface treatments

These results show, that an improvement in R a , R z and S a does not always translate into an improvement in the fatigue life of AM parts. Such approaches can only be successful if the peaks and valleys are normally distributed across all the surface conditions investigated. For practical applications with a non-normally distributed roughness profile, as described here, S V might be better suited to assess a surface for its influence on fatigue life, as for example applied in Kahlin et al. (2020), provided a large enough measurement area around the critical areas is selected. When comparing the individual components of both AB and TW, no correlation to the projected fatigue life on a shared stress horizon can be observed for R a , R z and S a and only a weak correlation for S V , contrary to observations by Kahlin et al. (2020). This is because the measurement area was limited to a specific position on the components. While that allows a good screening of the general surface quality, the measurement area seldomly included one of the larger, crack inducing defects. On the FWA, crack location shifted greatly dependent on where the largest defect was located, sometimes initiating in an area of subcritical local stresses, see Fig. 5. If a larger measurement area were used and the location of the deepest notch tied to its local stress, the correlation between S V and fatigue life should increase, but