PSI - Issue 19

Fumiyosi Yoshinaka et al. / Procedia Structural Integrity 19 (2019) 214–223 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

216

3

for higher stress amplitudes. The stress amplitudes ranged from 250 MPa, where the tests were run up to the N = 10 7 cycles without failure, to 350 MPa. The microstructure after the fatigue tests was investigated using EBSD microscopy for some specimens, including the run-out specimen in the HCF test. The observation was conducted on the section parallel to the loading direction (longitudinal section) finished with colloidal silica. A secondary crack was found on the section of the fatigue-fractured samples; therefore, this work investigated the microstructure around the secondary crack. 3.1. ε -N diagram Figure 1 shows the relationship between the total strain range Δ ε t and the number of cycles to failure N f ( ε - N diagram) obtained in the LCF tests. Here, N f is defined as the number of cycles where the maximum stress decreases by 25% compared to that at mid- life. Only at Δ ε t = 6% in Fe-15Mn-10Cr-8Ni-4Si, the tests were stopped because excessive elongation occurred and exceeded the measurable range of the extensometer; the number of cycles where the tests were stopped was used as N f . Probably for this reason, N f at Δ ε t = 6% was smaller than that expected from N f at the smaller Δ ε t . In the figure, the results obtained by us under similar test conditions for the following commercial materials are also shown for comparison: Fe-Mn-Si shape memory alloy Fe-28Mn-5Cr-6Si, stainless steel SUS304, and low-yield point steel LY225. Fe-28Mn-5Cr-6Si and SUS304 are also austenitic steels as with Fe 15Mn-10Cr-8Ni-4Si. LY225 is widely used in steel dampers due to its low yield ratio (yield stress / tensile stress). When comparing the N f at the same value of Δ ε t , N f of Fe-15Mn-10Cr-8Ni-4Si was significantly longer than that of the commercial materials, including Fe-28Mn-5Cr-6Si that showed a longer N f compared to that of SUS304 and LY225. Fig. 2 shows the probability plot of N f at Δ ε t = 2%. While the number of samples tested at Δ ε t = 2% was limited for the statistical discussion, the variability of N f seemed to follow a normal distribution. Assuming a normal distribution, mean N f at Δ ε t = 2%, i.e., the number of cycles to failure corresponding to the fracture probability of 50%, was 1.2 × 10 4 , which was 15 times longer than N f obtained for LY225. 3. Results and discussion

F ig. 1. Relationship between the total strain range Δ ε t and the number of cycles to failure N f .