PSI - Issue 79

Andrea Avanzini et al. / Procedia Structural Integrity 79 (2026) 88–96

94

The  T-N curves for each test are presented Fig. 5a, where the highlighted underlying area was used to compute  , following the method suggested (Santonocito et al., 2021) to account for the stepwise nature of the procedure when using a single specimen. The resulting predicted S-N curves are shown in Fig. 5b.

Fig. 5. Determination of  parameter and predicted life curves

Considering fatigue strength, the application of IRT methods suggests that the best performances can be achieved after S+A, followed by AN and SR, while DA is less effective. These findings were compared with preliminary evidence from conventional tests conducted using the same test rig, frequency and load ratio, from an ongoing campaign. These tests were run up to 2E7 cycles and revealed that HT has a profound influence on fatigue. A summary of the main results is reported in Tab.3, where  IRT refers to the range of values predicted with the different methods and  1E6 ,  2E6 ,  2E7 to the stress limit corresponding to 1E6, 2E6 and 2E7 cycles in the conventional test.

Table 3. Comparison between predicted and actual fatigue limit (from conventional tests)

Heat Treatment

 IRT [MPa]

 IRT /  1E6

 IRT /  2E6

 IRT /   E7

AN SA SR DA

355-359 417-449 333-370 277-286

1.014 - 1.026 1.043 - 1.123 0.900 - 1.000 0.865 - 0.893

1.044 – 1.056 1.191 – 1.283 0.979 - 1.089 0.923 - 0.953

1.183 – 1.196 1.603 - 1.727 1.332 - 1.480 1.385 - 1.430

In general, the order of magnitude of the IRT fatigue limit is similar to conventional tests, confirming that the method is sensitive to variations of material microstructure induced by HT and can provide useful information for screening purposes. Quite interestingly, the predictions obtained with the IRT method are in good agreement with conventional tests especially when considering the life range 1E6 ÷ 2E6. On the other hand, for tests at 2E7 the correlation is weaker and actually, failures were observed in the range 2E6 ÷ 2E7 even for stress below the  IRT predicted range. While the tested material exhibited a porosity lower than 0.2% for all conditions, this difference is likely the result of the competition between defects and microstructure on triggering crack nucleation and propagation. For the low stress levels associated with longer lives, the micro-plastic deformation occurring on the surface are indeed limited, and the IRT method (or the setup used in the present work) may not be sufficiently sensitive to capture very small temperature variations. While this should correspond to what is expected when testing the material below the fatigue limit, the presence of sub-surface or inner defects may act as an alternative source, capable of activating fatigue crack nucleation and propagation, ultimately resulting in failure below the “expected” stress level. Considering the S-