PSI - Issue 58

Mikkel L. Larsen et al. / Procedia Structural Integrity 58 (2024) 73–79

77

M.L. Larsen et al. / Structural Integrity Procedia 00 (2024) 000–000

5

Pin observed to deform sligthly

Equivalent springs

Fixed support

Fig. 6. Picture of pin that is observed to deform during testing and updated FE model with equivalent springs.

gauges G2, G5, G6 and G11 in accordance with Fig. 3 are used as response parameters and the spring sti ff ness is used as the optimization parameter. The spring sti ff ness is then optimized such that the response from the chosen gauges is as similar as possible to the experimentally obtained strains. This optimization is based on a fixed deformation measured from the experiments in the vertical direction ( d V in Fig. 4) that is applied to the FE model as well. The initial spring sti ff ness is set to 100,000 N / mm and the allowed range in the optimization is from 50,000 to 200,000 N / mm. After performing the optimization, the updated spring sti ff ness is 142,130 N / mm. The updated strain comparisons using the updated FE model are shown in Fig. 7, where all strains have been normalized.

a) Global gauges.

b) Global rosette gauges.

c) Local gauges.

Fig. 7. Comparison of FE predicted strains and experimental strains using the updated FE model with equivalent springs.

As seen in Fig. 7 (a) and (b), the updated FE model predicts the strains at the global gauges and global rosette gauges quite accurately, with only a few gauges showing relatively high discrepancies. The local gauges, Fig. 7 (c) are seen to show higher discrepancies. This is expected, as the gauges are placed closer to stress raising e ff ects and discontinuities. Thus, a minor misplacement between the location of the gauges in the experiment as compared to the FE model will result in large discrepancies in the strain results. However, in general, the FE model is seen to be quite accurate. In Tab. 1, the root-mean-square error and averaged prediction ratio P µ is given. As seen from Tab. 1, a high improvement is obtained as compared to the original model and the FE model is accurately predicting the strains.