PSI - Issue 17

Moritz Hartweg et al. / Procedia Structural Integrity 17 (2019) 254–261 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

260

7

up to about 80,000 cycles would be covered by short crack propagation with only small increase in the crack size. To confirm these assumptions, further decidedly investigations must be carried out.

30

multiple cracks single crack

25

20

15

10

calculated angle [°]

5

0

N f - 10,000

N f - 9,000

N f - 8,000

N f - 7,000

N f - 6,000

N f - 5,000

cycle number prior to failure

Fig. 8. Angle of specimens with single and multiple cracks. The curves are shifted for better visibility. The scatter is higher in case of multiple cracks.

In case of multiple cracks, the scatter in the last cycles prior to failure of the specimen was found to be much higher compared to specimens with a single crack. Figure 8 shows the calculated angles for specimens with single and multiple cracks in the period from 10,000 to 5,000 cycles prior to final failure of the specimen. All curves are shifted for a better visibility, i.e. not the real calculated angle is shown. The specimen with multiple cracks shows a significant higher scatter of the calculated angle compared with the specimens with a single fatigue crack. The origin of this higher scatter can be attributed to the smaller length of the individual cracks, resulting in a lower potential drop and therefore a higher scatter of the potential drop measurement. 4. Discussion The experiments have shown that the detection of cracks in notched round bars with three potential drop probes is possible. The proposed analytical model allows an easy and fast determination of the crack initiation site on the specimen surface. However, for multiple cracks, the calculated angle only indicates the main emphasis of the sum of all cracks. The angle not only shows the location of the crack, it can also be used as a criterion for the initiation of a crack. In the case of circumferential cracks as found by Campagnolo et al (2019) in sharp notched specimen no defined angle can be calculated and only the increase of all three potential values indicates the formation of a crack. The scatter of the calculated angle was found to be decreasing with the cycle number. A quantification of the scatter can be performed by calculating the standard deviation over a defined interval. The calculated standard deviation (figure 5b) as well as the scatter of the standard deviation is decreasing with the cycle number. Obviously this parameter are coupled with the formation and propagation of a crack and can be used to develop a criterion for crack determination. This will be investigated in more detail in further experiments. For the determination of the moment of crack initiation this method has some advantages. Due to the good electrical conductivity of metallic materials only small voltages are measured with the PDM and the potential values are influenced by the specimen temperature (Halliday und Beevers (1980); Bauschke und Schwalbe (1985); Hartman und Johnson (1987)). An increase of the specimen temperature due to external influences can increase the measured potential and thus simulate the formation of a crack. This problem can be eliminated by using an additional potential probe away from the crack. Unfortunately, this additional potential probe can only be used for temperature