PSI - Issue 75

Xiru Wang et al. / Procedia Structural Integrity 75 (2025) 85–93 Wang / Structural Integrity Procedia 00 (2025) 000 – 000

90 6

a minor stress concentration. This indicates that the stress concentration is influenced not only by the weld toe radius and angle but also by 3D-effects that determine the flow of stress through the component.

Fig. 3: Derived stresses and cycles to failure from 2D-models calculated in plain strain condition and one 3D-model

Fig. 4: Stress distribution and coordinates at the 3D-model. Dark grey color indicate a stress 1 ≤2× In addition to the local 3D-effect, there is a global at play: Although the weld geometry remains approximately constant between coordinates 0 and 40 as well as between 80 and 120, the stress concentration in the 3D-model is not uniform; it continuously decreases towards the edges of the specimen. This variation is due to a non-uniform stress condition in the width direction: at the edges of the specimen, a plane stress condition exists, while in the middle, particularly at the weld, a plane strain condition can be observed. This heterogeneous condition results in an uneven stress concentration at the weld. The maximum stress concentration in the 2D-model is approximately 20 % lower than that observed in the 3D model. This difference results in a slightly higher calculated fatigue life of 25%, determined by fracture mechanics ( 10 2.36−2.26 =1.25 ). Therefore, both modeling approaches yield similar fatigue lives in principle. When considering the overall effect of the weld end and start position by comparing the number of cycles at the weld start and stop position (coordinate 55) with that at the continuous weld (position 30), the difference is significant, resulting in a factor of 2 in cycles ( 10 2.6−2.3 =2.0) . Despite some scatter in the experimental data, ( 1: =1.1 see Fig. 3), a factor of 2 in cycles is not evident. This raises the question why the experimental fatigue life does not reflect a difference in life. The likely explanation is the differing crack propagation: at a weld without distinct start and stop positions, the entire weld is subjected to high stress. Cracks can initiate at multiple locations, resulting in a small ratio of crack depth to crack width. This leads to faster crack propagation compared to welds where only one area is highly stressed, resulting in a larger / -ratio. Another interpretation of the influence of weld start-stop positions can be derived by examining the effective stress. In the 3D-model, analyzing the coordinates 30 and 55, a factor in stress of 1.52/1.06 = 1.43 is derived those leads,