PSI - Issue 57

Gustav Hultgren et al. / Procedia Structural Integrity 57 (2024) 428–436 Hultgren & Barsoum / Structural Integrity Procedia 00 (2023) 000 – 000

432

5

associated with lower heat input in the same area. However, the presence of undercuts can potentially violate the 122 limitations imposed by the international weld class system (ISO 5817:2014, 2014), and is thus often avoided. 123 4. Analytical models of stress concentration factors 124 Multiple parametric investigations have, over the years, studied the stress-raising effect of the local weld 125 geometry using simulations of idealised weld geometry. Tab. 1 presents four models that all consider variation in the 126 weld toe radius for non-load-carrying tee joints. Fig. 4 illustrates the weld geometry idealisation derived from the 127 Winteria system alongside an idealised simulation geometry. These analytical models will be further compared with 128 the probabilistic sectional stress. A strong correlation with the analytical models would motivate considering the 129 current weld idealisations as reliable representations of the stress concentration factors that influence fatigue strength. 130 The correlation between the analytical SCFs and the ratio between the simulated sectional stress and nominal stress, 131 presented in Fig. 5, shows, however, no clear trend for any of the expressions. The size of the measured weld toe radii, 132 shown using the scatter colouring, highlights that the strong correlation for the analytical expressions with this quantity 133 is not as evident for the simulation results. One possible explanation for this disagreement is that the geometrical 134 complexity of real welds has more intricate interaction effects between the weld geometry parameters and the stress 135 concentrations than idealised geometrical simulations captures. Another possible explanation for the discrepancy can 136 be the higher uncertainty in weld toe radius evaluation for smaller radii. Preliminary results of an ongoing Round- 137 Robin study by Schubnell et al. (2022), which studies the accuracy of weld toe radius and weld toe angle estimations 138 for different measurement systems, show that the mean relative error is increasing for radii below 1 mm. Most 139 measurement systems overestimate the weld toe radius at this end of the spectrum for the studied idealised, machined 140 specimen. 141 Figure 3. Kernal distributions of the geometry measurements for the investigated specimens and their geometric impact seen in the ratio between the sectional stress and the nominal stress range.

Figure 4. Weld geometry that is (a) determined through the Winteria® system and the (b) idealised simulation geometry.