PSI - Issue 75

6

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

514

The tests were conducted using the CRAPPY Python module in order to achieve simultaneous control of the cameras and the micro tensile testing machine. CRAPPY is a home Python module developed at the LaMcube, as described by Couty et al. (2021), which enables the simultaneous control and acquisition from multiple sensors or actuators. In our case, the micro-machine was operated under load control, using successive loading/unloading cycles with an increment of 100 N on the maximum force for each cycle until failure. During the tests, images were acquired simultaneously by both cameras at 1hz. Additionally, the tensile force applied by the machine was also recorded. All data were synchronised using a common timestamp, allowing easy retrieval of the tensile force applied for each pair of images. Thanks to the indentations on both side of the speckle pattern, the displacement and strain fields obtained by image correlation can be plotted onto the microstructure of the specimens, thus allowing visualisation of the most highly strained zones. The mechanical fields resulting from image correlation show a localisation of displacement and strain within the weld bead (Fig. 6).

Fig. 6. Example of experimental results, equivalent Von Mises Strain (standardised)

Early yielding of the weld bead compared to the rest of the structure is also observed, indicating a lower yield strength. The transverse strains (ϵyy ) show a contraction of the weld bead, which is consistent with the raw test images; indeed, significant necking of the specimen is observed in this zone before failure. It is also observed (Fig. 6.) that the strain fields, although primarily localised in the weld bead, are neither homogeneous nor perfectly symmetrical within it; this calls into question the isotropy of the weld bead. 1.3. Numerical and AI models A numerical model was developed in parallel with the experimental tests, with the aim of quantitatively determining the differences in mechanical behaviour in the welded zones. First, a simulation representative of the micro-tensile tests was defined. To make the model geometry as close as possible to the experiments, it was derived directly from the micrographs taken of the specimens after chemical etching. Thus, the numerical model comprises three distinct zones: the weld bead, based on the actual geometry of each specimen; a HAZ (Heat-Affected Zone) with an arbitrarily set width; and the base metal (BM) (Fig. 7a). For each of the three zones, a different bilinear constitutive law is implemented. This simulation uses an isotropic behaviour model.