PSI - Issue 68

Lea Aydin et al. / Procedia Structural Integrity 68 (2025) 1280–1286 L. Aydin and S. Marzi / Structural Integrity Procedia 00 (2024) 000–000

1285

6

Fig. 6. Comparison of TSLs from external measurement and OFS. Left: crack initiation. Right: crack propagation.

4.4. Determination of fracture properties

To obtain further results and to compare the already evaluated test, two additional tests were carried out with di ff erent initial crack lengths. The results of the three key parameters that are extracted from the TSL are shown in Fig. 7. The mean values with error bars of the cohesive sti ff ness ¯ k s , the maximum COD ¯ w max COD and the ERR ¯ J are shown for each of them. In addition, the CTOD and the ERR can also be determined from the external measurement. Here, the ERR is averaged from the plateau in Fig. 5, and therefore also has an error. For the COD, however, we only obtain one value. It can be seen that, regardless of the initial crack length, the results correlate very well with each other concerning the scattering, but also provide good results in direct comparison with the external measurement. This indicates a reproducible method with promisingly reliable data.

Fig. 7. Comparison of results of each key parameter with the external measurement.

5. Summary

In this study, a new method for evaluating DCB tests using optical fibers with high spatial resolution is presented. Moreover, the limits of the optical fiber method were discussed, as well as the potential for performing tests based solely on strain measurements to derive the fracture energy, traction separation laws, and, in perspective cohesive zone models. The widely used J-integral typically only allows the evaluation of the traction-separation law at the ini tial crack tip position. In contrast, the new approach o ff ers greater flexibility and additionally allows for the detection of defects within the adhesive layer and the identification of material behavior inhomogeneities. Even after excluding some initially unusable data, we were able to extract approximately 700 TSLs from a single experiment, demonstrat ing the method’s capability to generate an extensive data set. This large amount of data enables a more detailed and