PSI - Issue 38

Boris Spak et al. / Procedia Structural Integrity 38 (2022) 572–580 Author name / Structural Integrity Procedia 00 (2021) 000 – 000

578

7

4. Experimental results A series of fatigue tests has been conducted to assess the fatigue life if the clinched joint variants. The preparation of specimen and experimental setup has been done according to the guideline DVS/EFB 3480-1 (2007) using a Sincotec Power Swing resonance testing rig with stress ration R = 0.1. On average, the testing frequency leveled out at 35 Hz. Though, a variation in testing frequency indicates that, although the preparation of specimens had been done with great care, either the forming process or inhomogenity in the aluminum sheets might have led to deviations in joint properties, thus impacting the fatigue life and resulting in a relatively large scattering range. Fig. 3 presents the experimental results. The diagram on the left hand side shows the number of cycles to failure with the criterion fracture for force amplitudes F a ranging from 675 N to 1125 N. Comparing the two variants of clinched joints, it can be concluded that no significant difference in fatigue life exists with respect to the failure criterion fracture. The diagram on the right hand side presents the results of an investigation on the suitability of a drop in frequency of 0.05 Hz, 0.1 Hz and 0.2 Hz to predict crack initiation in the clinched joint variant 1.0. Two different force amplitudes were chosen to limit the test run time. For each drop in frequency at each chosen force amplitude, three tests were conducted. If the chosen drop in frequency was detected by the testing software, the test was stopped and a micrograph of the specimen was prepared to check for visible cracks. From the results of this investigation, it can be deduced that a drop of 0.1 Hz and 0.2 Hz is not sensitive enough to distinguish between crack initiation and fracture. A drop in frequency of 0.05 Hz appears to be more suitable to predict the number of cycles at crack initiation, although the specimens with detected cracks lie within the scattering range of the fractured specimen. A fatigue life estimation with the LSA was carried out. For the calculation of the damage Wöhler-Curve eq. (6), the set of cyclic material parameters with 0% pre-strain was chosen. The result of the fatigue life estimation for both variants is presented in the right diagram of fig. 3. Comparing the fatigue life estimation based on LSA with experimentally obtained number of cycles up to crack initiation for variant 1.0, the use of the damage parameter P SWT shows a very conservative estimation. This can be attributed to the inclusion of mean stress in the damage calculation. While it is reasonable to assume, that a higher mean stress will lead to a reduction in fatigue life, for a component such as the clinched joint manufactured from thin sheets with a strain history with high degrees of deformation and consequently change in microstructure as well as in material properties due to cold working, it is debatable how the mean stress evolves in the critical region during cyclic loading. Without any non-destructive methods to investigate the impact of the mean stress and any residual stresses on the fatigue life in the critical area of the clinched joint, a further attempt is made here by omitting the mean stress

in the damage parameter P SWT in eq. (4). The result is an increased fatigue life estimation and certainly closer to experimental results, as shown in fig. 3. For variant 1.4, the LSA using the damage parameter P SWT without and with estimates overall a shorter fatigue life compared to variant 1.0, except for very small force amplitudes. This appears to be in contradiction to the experimental results, as both variants show a similar fatigue life with respect to the criterion fracture. Without further investigation of crack initiation in the clinched joint variant 1.4, it is not possible to Fig. 3 Left: Experimental fatigue life with failure criteria fracture. Right: Comparison of fatigue life estimation of variants 1.0 and 1.4 using damage parameter P SWT (without and with mean stress). Three levels of drop in testing frequency to detect crack initiation for variant 1.0.