PSI - Issue 79

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

423

ness correction may be applied according to Hobbacher and Baumgartner (2024), although no explicit guidance was provided. By contrast, Rennert et al. (2020) proposed a more detailed thickness correction for wall thicknesses be tween 25 mm and 10 mm, expressed as: FAT = FAT ref · t ref t 0 . 1 (1) where the subscript ref denotes the reference values. For structures with wall thicknesses below 10 mm, Rennert et al. (2020) recommended adopting the FAT class corresponding to a wall thickness of 10 mm. As noted by Radaj et al. (2006), the hot-spot stress-based approach was initially formulated for fatigue assessment using strain-gauge measurements and was later adapted for finite element analysis, which now constitutes its principal area of application. According to Hobbacher and Baumgartner (2024), hot-spot stress can be determined by linear or quadratic extrapolation of surface stresses evaluated at absolute or thickness-relative distances from the weld toe. Experimental comparisons on 10 mm thick non-load-carrying fillet-welded specimens, reported by Lee et al. (2010), revealed negligible di ff erences among these evaluation techniques. As described by Radaj et al. (2006), a notable advantage of the hot-spot stress-based approach is its ability to account for the actual geometric conditions of the weld detail, while the finite element models required for its evalu ation remain relatively simple. Furthermore, the results are less sensitive to local imperfections, such as undercuts or overlaps, than those obtained using other local approaches. However, the standard hot-spot stress-based approach is unsuitable for situations where fatigue cracks initiate at the weld root, and alternative techniques described by Fricke (2013) should then be employed. Similar to the previous approach, the influence of plate thickness is reflected by adjusting the FAT class. For joints not involving hollow sections, Hobbacher and Baumgartner (2024) specified FAT classes only for a reference wall thickness of 25 mm and provided no explicit guidance on how to incorporate the beneficial thickness e ff ect associated with thinner walls. Consequently, following the procedure used in the nominal stress-based approach, the FAT classes specified by Hobbacher and Baumgartner (2024) were increased by 10 %, in accordance with the recommendation of Rennert et al. (2020). For hollow-section joints, Hobbacher and Baumgartner (2024) referred to Zhao and Packer (2000), who defined FAT classes for components with wall thicknesses between 4 mm and 50 mm using the following expression: where t is the wall thickness. For components thinner than 4 mm, Zhao and Packer (2000) recommended adopting the S–N curve corresponding to 4 mm, since the beneficial trend of increasing fatigue strength with decreasing wall thickness may no longer apply. Among the fatigue life estimation methods considered in this study, the notch stress-based approach, as reported by Radaj et al. (2006), is the most general, being suitable for the fatigue assessment of virtually all welded geometries and potential failure locations. This flexibility, however, entails greater computational demands and increased sensitivity to local imperfections, such as undercuts or overlaps. Thee ff ective notch stress is defined by introducing an artificially enlarged notch radius, as illustrated schematically in Figure 1. As noted by Radaj et al. (2013), this artificial enlargement serves as a practical equivalent to averaging the elastic stress over a small, material-dependent distance ahead of the notch root, which is a principle that under lies Neuber’s microstructural support hypothesis. However, since the actual notch radius of a real weld toe is often unknown, Radaj (1990) noted that it is conservatively assumed to be zero, and the notch root is then rounded with a fictitious radius of 1 mm, determined from the material-dependent distance, ρ ∗ , and the multiaxiality coe ffi cient, s . Considering thin-walled structures, Marulo et al. (2017) reported that applying a fictitious notch radius of 1 mm represents a significant change to the weld geometry and tends to excessively reduce the stress concentration, resulting in non-conservative fatigue life predictions. To mitigate this limitation, Hobbacher and Baumgartner (2024) proposed smaller reference radii of 0 . 05 mm for very thin components and 0 . 3 mm for structures of intermediate wall thickness. With respect to their practical use, Baumgartner et al. (2020) recommended applying these radii at the weld root and weld toe in structural steel and aluminium alloy components when they represent no more than 10 % and 20 % of the wall thickness, respectively. Although these smaller reference radii improve prediction accuracy for thin-walled FAT = 10 ( 1 3 ( 12 . 476 − log ( 2 · 10 6 )) + 0 . 06log ( 2 · 10 6 ) log ( 16 t )) (2)