PSI - Issue 79

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

427

0.02 mm 0.012 mm

0.01 mm

0.006 mm

Fig. 4. Representative finite element mesh at the weld toe region, illustrating the refinement used for notch stress evaluation.

3. Results

As a result of the fatigue testing outcomes, several specimens were excluded from the evaluations presented in this article. The first exclusions involved specimens that failed at the weld root, specifically one from the CHS fillet, two from the CHS lap, and three from the CHS-Plate fillet S configuration. Subsequently, two Plate fillet T Ten specimens were omitted based on the observation reported by Gurney (1997) that excessive defects likely compromised their fatigue performance. Finally, all run-outs and specimens with fatigue lives exceeding 10 7 cycles were excluded, as this threshold defines the transition to the shallower slope region of the S–N curve, according to Hobbacher and Baumgartner (2024). This limit was exceeded by two specimens from the CHS fillet configuration. All excluded specimens are shown in the S–N curves as white-filled markers outlined in their configuration color, annotated with the symbol ⊥ or, for run-outs, with an arrow. The final number of accepted specimens for each configuration is summarized in Table 1. In addition to individual specimen exclusions, entire configurations were omitted from further evaluation when they produced misleading results for any of the prediction approaches. This was the case for the Plate fillet T Ben and Plate fillet T Ten configurations, where uncertainty in the notch stress evaluation arose from incomplete documentation of weld dimensions. As only nominal weld sizes were reported by Gurney (1997), they were applied consistently along the weld seam, including the anticipated crack initiation locations at the weld ends, where the actual geometry clearly di ff ered from the regular nominal weld profile. Both RHS-fillet configurations were also excluded, as the definition of nominal stress for these joint types was inherently ambiguous. In all figures, specimens from omitted configurations are depicted as white-filled markers outlined in the color assigned to their respective configuration. Each specimen configuration was assigned a characteristic FAT class for all fatigue life estimation approaches considered in this study. For the nominal stress-based evaluation, FAT classes were initially selected according to Hobbacher and Baumgartner (2024) and then increased by 10 % to account for the thickness e ff ect, as recommended by Rennert et al. (2020). Regarding the hot-spot stress-based evaluation, the same procedure of initial FAT class selection followed by a 10 % increase was applied to plate-based configurations. Hollow section-based configurations, however, were directly assigned FAT193, according to Zhao and Packer (2000). The FAT classes assigned to each configuration under the nominal and hot-spot stress-based approaches are summarized in Table 1. For the notch stress based evaluation, a uniform FAT class of FAT500 was applied to all configurations, as recommended by Hobbacher and Baumgartner (2024) for weldments failing at the weld toe. The S–N data obtained from the individual fatigue life estimation approaches are shown in Figures 5, 6, and 7. All calculated stress ranges were normalized to their respective FAT classes to enable meaningful comparison across the di ff erent fatigue life estimation approaches as well as across configurations assigned to various FAT classes. The normalization was performed by dividing each stress range by the FAT class associated with the given configuration and prediction approach, as follows:

∆ σ FAT

(3)

∆ λ =