PSI - Issue 53

Benjamin Möller et al. / Procedia Structural Integrity 53 (2024) 190–202 Author name / Structural Integrity Procedia 00 (2023) 000–000

200

11

(a)

(b)

Fig. 13. (a) FE simulation result for shape variations (smoothing); (b) Effect of the smoothing parameter on the local stress.

As an example, in Fig. 13a the results of the three types of analyses for one specific hollow structure specimen are shown. By looking at the peak stress values, it becomes clear that the small waves and the macro-geometric shape variations have very similar effects on the stress concentration. From this result, it is already possible to point out that the improvement of the WAAM process must equally focus on the elimination of both these defects. However, prognosis of the fracture location is also possible based on the simulation of the smoothed model, so that macro

geometric shape variations play a major role on the fatigue fracture. Further details on the FE study will be given by Taliani (2023).

3.3. Stress-based S-N curves The S - N curves for von Mises equivalent stresses of specimens modelled with ideal flat, smoothed and “real” scanned surface versus experimentally derived cycles to failure are shown in Fig. 12. Each series has again been evaluated using the Maximum Likelihood Estimation, Störzel and Baumgartner (2021), resulting in a knee point at N k = 800,00 cycles to failure. The number of results included in the evaluation for the series with smoothed and “real” scanned is slightly reduced, since useable scans were not available for the first specimens. Furthermore, the slope after the knee point was again set to k * = 22.0, as it recommended for aluminum / welded joints by Sonsino (2007). With these assumptions, the Wöhler (S-N) curve is described by the parameters documented in Fig. 12. While the stress amplitude increases from ideal flat (  a,vM,k = 44.7 MPa) over smoothed (  a,vM,k = 60.9 MPa) to “real” scanned (  a,vM,k = 89.0 MPa) specimens due to the surface-related stress concentration, the scatter decreases from ideal flat T  = 1 : 1.9) to smoothed or “real” scanned T  = 1 : 1.6), which indicated that the results merge by consideration of the main effect on fatigue. The slope also changes slightly from k = 3.1 for ideal flat to k = 3.8 the others. The same slope but different stress amplitudes between specimens with smoothed and “real” scanned surface are comprehensible, since the failure location of the specimens tested is still at the same spot, but the notch shape is different. Based on the knee points of S - N curves, surface-related stress concentration factors of K t,sur = 1.5 (smoothed) and K t,sur = 2.0 (“real” scanned). This factor can be used to calculate the local equivalent stresses based on the design and simulation using the ideal flat geometry.