PSI - Issue 19

Fabian Becker et al. / Procedia Structural Integrity 19 (2019) 645–654 F. Becker et al. / Structural Integrity Procedia 00 (2019) 000–000

649

5

2.3. Experimental set-up of the fatigue testing with clamping

As initial step in this second testing series, the bending modulus was determined with open clamping group like in the reference tests. The bending strain in a clamped specimen is di ff erent due to the additional sti ff ness of the clamping. However, the algorithm for calculation of the bending strain with open clamping group was used and the result was defined as equivalent bending strain. The bolt loads were adjusted to the desired value of 5 kN prior to the beginning of the fatigue test with alternating tightening of the bolts. During the tests, the bolt loads were recorded with a frequency of 32 Hz. To determine the damaging behaviour during the test, the fatigue tests were interrupted every 5000 cycles to take one photo of the damaged tension side of the specimen via a mirror. All tests were conducted with a testing frequency of f = 2 Hz to avoid heating of the clamping group due to friction of damaged parts of the specimen and sinusoidal waveform with R = 0 . 1.

3. Results and discussion

In a postprocessing routine programmed with the software Matlab, The MathWorks, Natick USA, the bolt load data and the corresponding sti ff ness curves were correlated. The results of all testing series are summarized in table 1. All strain values are normalized with respect to the maximum applied equivalent bending strain and the number of cycles reached are normalized with respect to the highest measured cycles. The nominal equivalent strain value of specimen ”7” is denoted as ”low”, the value of specimen ”10” as ”intermediate” and the values of specimens ”8” and ”9” is denoted as ”high” in the following.

Table 1. Results of reference testing series and testing series with a target bolt load of 5 kN Specimen ID Testing series (-) Bending modulus (MPa)

Norm. max. equivalent strain (-)

Norm. cycles to failure (-)

1 2 3 4 5 6 7 8 9

Reference Reference Reference Reference Reference Reference

46458 44556 42994 42145 45003 45127 45263 39711 42985 45460

0.83 0.92 0.88 1.00 0.77 0.96 0.58 0.87 0.73 0.65

0.042 0.022 0.053 0.007 1.000 0.008 0.158 0.006 0.023 0.058

5 kN 5 kN 5 kN 5 kN

10

As an example, the experimental results for the intermediately loaded specimen ”10” is plotted in figure 4. The mean values of the bolt loads were extracted from the measurement data for each cycle during the fatigue test. Prior to the beginning of the experiment the force values show a decrease in mean value of 0 . 5 kN and a constant loss during the experiment, which is caused by the creeping in the clamping group. Long term oscillations are visible, that may have their origin in a variable oil temperature during the fatigue test ing, leading to heat transmission to the clampings’ steel plates by conduction and a subsequent thermal expansion of the polyamide material. The hypothesis of increasing force values due to thermal expansion were confirmed by static creep testing where a thick flat metal plate was clamped within the developed device and the clamping force increase due to thermal expansion under ambient temperature was observed and measured. The normalized dynamic sti ff ness shows a minor increase in its value during the first 40 % of the specimen’s life and one steep decrease between 70 % − 80 % followed by a constant, but flatter loss in sti ff ness until 90 % of the specimen’s life. After 90 % of the specimen’s life, a second steep loss in dynamic sti ff ness is visible. While the increase in sti ff ness during the early stages is unlikely to be ascribable to a certain material behavior, the steeper loss in sti ff ness can be attributed to the beginning of delaminations. Three images are shown that were taken before and after the first loss in sti ff ness (figure 5). With the comparison of figures 6 and 7 the beginning of a delami-