PSI - Issue 79

Martin Sladký et al. / Procedia Structural Integrity 79 (2026) 421–432

430

tially lower, reaching only about 60 % of that observed for the nominal and notch stress-based approaches. In con trast, the notch stress-based approach, which exhibited a scatter range index comparable to the nominal stress-based approach, produced two approximately parallel S–N curves corresponding to plate-based and hollow section-based configurations, showing a clear separation between the two trends.

4. Discussion

Since the FAT class increase associated with wall thickness reduction in both nominal and hot-spot stress-based approaches is constrained by the threshold values corresponding to limit wall thicknesses of 4 mm and 10 mm, as specified by Zhao and Packer (2000) and Rennert et al. (2020), respectively, the potential benefits of extending these limits were investigated. For the nominal stress-based approach, the currently applied 10 % FAT class increase, corresponding to a limit thickness of 10 mm, already led to the first non-conservative predictions for several Plate fillet L specimens, as shown in Figure 5. A further FAT class increase, determined according to Equation 1, resulted in additional non-conservative fatigue life estimates for more Plate fillet L specimens and, subsequently, for the first Plate-RHS and CHS fillet specimens, while the scatter remained essentially unchanged. As the dominant factor influencing scatter is likely the degree of similarity between the evaluated and reference structural details, addressing this correspondence would probably enhance predictive accuracy more e ff ectively than further extending the thickness correction, which in this case appears to play only a secondary role. Conversely, the S–N data presented in Figure 6 indicate that increasing the FAT classes beyond the current thresh olds associated with the limit wall thicknesses could further improve the predictive accuracy of the hot-spot stress based approach. To investigate this, alternative FAT classes were established for each configuration based on its actual wall thickness, using Equation 2 for hollow section-based and Equation 1 for plate-based configurations, as recom mended by Zhao and Packer (2000) and Rennert et al. (2020), respectively. The resulting S–N data, normalized to these alternative FAT classes, are presented in Figure 8, with the slope of the fitted S–N curve fixed at m = 5 to maintain consistency with the preceding analyses.

Hot-spot stress-based approach with FAT classes derived from actual thickness

9

3 ( 12 . 476 − log ( 2 · 10

6 )) +0

. 06log ( 2 · 10

6 ) log ( 16

t ))

Hollow section-based configurations: FAT =10 ( 1 Plate-based configurations: FAT =25 0 . 1 FAT ref ( t ) −

0 . 1

6

CHS lap CHS fillet

1 Normalized stress range ∆ λ [-] ∆ λ P 50% = 1.613 T σ = 1:1.400 m= 5 2 4

CHS-Plate fillet S CHS-Plate fillet G RHS fillet OPB RHS fillet IPB Plate fillet T Ben Plate fillet T Ten Plate-RHS fillet Plate fillet L Best-fit S-N curve Design S-N curve

0.7

10 4

10 5

10 6

10 7

Fatigue life in cycles N [-]

Fig. 8. S–N data for the hot-spot stress-based approach, normalized to the FAT classes determined for each configuration from its actual wall thickness using the equations shown in the top annotation box. White-filled markers denote specimens excluded from the evaluation of parameters listed in the bottom annotation box.