PSI - Issue 75

Arthur THIBAULT et al. / Procedia Structural Integrity 75 (2025) 509–518 Arthur THIBAULT / Structural Integrity Procedia (2025)

517

9

The main limitation of these results stems from one of the initial assumptions: indeed, the tests presented here were conducted on samples from scooter decks, assuming that these results are transposable to other types of products. Ideally, the same kind of tests should be conducted on different products from different suppliers and using different welding technologies, in order to account for production dispersion. There are several possible avenues for improvement for this work: firstly, it would be interesting to increase the complexity of the constitutive laws determined by the AI, for example, by seeking to identify multi-linear elasto plastic laws or by considering, for instance, transverse isotropy behaviour models. Furthermore, a modified version of the AI used here could be used to determine the HAZ size in addition to the constitutive laws. This implies adding an additional variable and therefore expanding the database. One could also consider other ways of sectioning the specimens based on strain criteria rather than on the microstructure. 4. Outlooks This work represents the first step towards improving the model used internally at Decathlon; the next step will be to perform multiaxial fatigue tests under representative loading.The purpose of these multiaxial fatigue tests is to quantify the influence of the loading type on the fatigue performance of welded joints. A secondary objective is to validate, on a structure similar to bicycle frames, the results obtained from the micro-tensile tests presented earlier. To this end, the actual loads applied to bicycle frames were studied and reproduced as faithfully as possible, considering the available testing equipment. Two series of multiaxial tests were conducted on butt-welded tubes using a biaxial testing machine combining tension and torsion (Fig. 9). The two series were carried out with different biaxiality ratios in order to approximate the loading cases observed on bicycle frames. In the first series, the loading was predominantly tensile with a relatively small torsional component, while in the second series, torsion was dominant over tension. These tests are still ongoing, but the initial results show a significant influence of the loading type on both fatigue life and crack initiation location. Specifically, the first series predominantly exhibited cracking at the weld toe, whereas the second series showed cracks initiating directly within the weld metal.

Figure 9: Samples used for the multiaxial tests (a). Experimental set up (b). Illustration of various crack initiation locations (c) and (d)