Fig. 4 Fatigue growth behavior of a short surface crack whose results are presented in Fig. 3 should its length be measured at intervals of (A) 10000 cycles, (B) 20000 cycles and (C), and 30000 cycles
The above example considers one crack out of fifteen cracks at a surface site. Should other individual cracks be in focus, difference in results is expected. Further, cracks existing at other surface sites are expected to have different activities. All the above aspects give rise to the experimentally observed anomalous fatigue growth behavior of short cracks.
Figure 5 shows examples of the computed \(A_{C}\) plotted against the number of cycles \(N\) as a result of running four virtual tests relevant to the experimental parameters listed in Table 1(a). The four runs started with the same specimen configuration. The two lines O-A-D\ and OBC\ present the results relatively out of the two constant stress amplitudes (CSA) 390 MPa and 432.5 MPa. In both two-step loading (TSL) tests, the application of\(\sigma_{o_{1}}\)for \(N_{1}\) cycles results in a behavior similar to that given due to the application of the CSA of 432.5 MPa and 390 MPa , i.e. the line O-A in Fig. 5 for the L-H sequence and the line O-B for the H-L sequence. The remaining life\(\text{\ N}_{2}\ \)at\(\text{\ σ}_{o_{2}}\) results in\(\ \sum{\ \frac{N}{N_{f}\ }}\) equal to 1.25 for the virtual L-H test and 0.79 for the H-L virtual test. Here, \(N_{f}\) refers to the number of cycles to failure at a CSA test. The corresponding experimental \(\ \sum{\ \frac{N}{N_{f}\ }}\) are 1.35 and 0.823. The lines B-C and A-D give the duration\(\text{\ N}_{2}\ \)for the sequences L-H and H-L respectively. Another specimen differently configured was utilized to repeat the above four virtual runs with the same\(\ N_{1}\). The resulting\(\ \sum{\ \frac{N}{N_{f}\ }}\) is and 1.18 for the virtual L-H test and 0.83 for the H-L virtual test.