PSI - Issue 57

Lewis Milne et al. / Procedia Structural Integrity 57 (2024) 365–374 Lewis Milne et al. / Structural Integrity Procedia 00 (2019) 000 – 000

369

5

The ultrasonic frequency fatigue tests were carried out at 20kHz using a Shimadzu USF-2000A ultrasonic fatigue testing machine, and conducted according to the procedures specified in WES 1112 (2017). The test set-ups for both 20Hz and 20kHz fatigue testing are shown in Figure 3. The ultrasonic specimens were screwed into an acoustic horn on each end of the specimen, and the height of the top horn was adjusted to ensure there was no mean stress acting on the specimen. The failure point was taken using the USF software’s automatic fracture detection, which stops the test when the resonant frequency drops to below 19.5kHz. The sample can then be pulled apart by applying a tensile load of 1kN to view the fracture surface. In order to counteract the heat generation from internal plasticity dissipation in UFT, air cooling and intermittent pulse loading was applied. The length of the load pulse was kept at 110ms throughout the test, and the cooling pause was adjusted from 0.5s-5s as necessary to give specimens sufficient time to cool back down to room temperature before the next load pulse. The specimens were painted with a high-emissivity matte black paint, and the surface temperature at the centre of the gauge section was monitored using a PyroCube infrared spot sensor with a sampling interval of 20ms. The temperature measurements were then fed into a LabView program which would pause the test if the specimen temperature exceeded 30°C, to mitigate any heating effects. 5. Fatigue Results The fatigue results, along with their corresponding Basquin model fit, are presented for both steels at 20Hz and at 20kHz test frequencies in Figure 4. As expected, a large discrepancy between the SN curves at conventional and ultrasonic frequencies was observed for both of the steels. The scatter was relatively low for the conventional frequency tests for both materials and for the high frequency S355JR tests, with R 2 values of 0.76, 0.81 and 0.96 for the S355JR 20Hz, 20kHz and Q355B 20Hz tests respectively. For the Q355B 20kHz tests, however, a much larger amount of scatter was observed, with an R 2 value of just 0.42. Some amount of scatter is to be expected due to the inherent probabilistic nature of fatigue, however the reason for the larger amount of scatter in the Q355B UFT data is unclear. Taking the run out values as the fatigue limits for each frequency, a clear increase in fatigue limit with test frequency was observed for the S355JR specimens, with the ultrasonic frequency fatigue limit appearing to be 54% higher than the conventional frequency fatigue limit. Due to the lack of material available, and scatter observed around the fatigue limit, no clear fatigue limit could be observed for Q355B at either frequency. As such, this necessitated the use of finite-life region comparison methods to allow comparison of the frequency sensitivity of the two steels. 6. Influence of Specimen Geometry At 20kHz, the two steels behaved similarly. The Q355B exhibited slightly higher fatigue resistance than the S355JR, although it is within the same region of scatter. This is as expected, with both materials being so similar, apart from the yield strength of S355JR being 8% lower than that of Q355B. At 20Hz, however, the fatigue results deviate significantly between the two materials. At an equivalent number of cycles to failure, Q355B specimens were able to withstand upwards of 25% higher stress amplitudes than the S355JR specimens. As this significant deviation was not seen at ultrasonic frequencies, it was likely caused by size effects due to difference in specimen geometries used at 20Hz. This highlights how significantly the size effect can influence results when comparing across different test frequencies, and reiterates the assertions made in previous literature that consistent geometries must be used (Fitzka et al. 2021; Tridello et al. 2021). As such, the difference between the Q355B results at the two frequencies should be considered more representative of the true influence of the test frequency. A limitation of using the hourglass specimen geometry at low frequencies is highlighted in the Q355B results, however. Because of the small risk volume in the samples, the probability of there being a critical defect within the highly stressed region is much lower than for the larger risk volume in the cylindrical fatigue specimens. As such, this leads to added difficulty in accurately identifying the fatigue limit, as it is possible for no critical defects to be present within the risk volume for a given specimen, leading to specimens running out above the fatigue limit. This can be observed for Q355B in the current investigation, where a run out test is observed at 300MPa, whereas another specimen is seen to fail at the much lower stress amplitude of 270MPa. It is clear that more robust fatigue limit determination methods would therefore be necessary when carrying out conventional frequency testing with this sample geometry in future.