PSI - Issue 79

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

92

3. Results and discussion As explained previously, the first and preliminary aim of this work was to implement and compare different approaches, concerning both the nature of the thermal parameter, temperature or energy based, both the strategies used to define the transition stress level corresponding to the fatigue limit. A summary of the results obtained for the examined conditions using dT/dt,  T steady and Q parameters is provided in Fig. 3, in which the results are presented after elaboration using the OCM method. As can be noticed, all the specimens exhibited a characteristic transition for a stress level, which according to the principles of IRT methods for fatigue estimations, corresponds to fatigue limit. Scatter can be noticed for low stress levels, with the notable exception on AN in which case transition is more abrupt. It is worth noting that the trends closely resemble those commonly reported in literature for conventional materials. When applying TCM methods for the identification of the transition stress level to the set of data under investigation, the differences among the algorithms implemented were in general limited, although not negligible. Considering the different parameters, the values of the fatigue limit obtained after averaging on different analysis techniques (i.e. OCM, TCM), are shown in Tab.1. The mean values for each test and method applied to a specific specimen are relatively consistent; however, the standard deviation of the results shows significant variability, and in this respect the steady temperature parameter gives better results compared to slope parameters. Nevertheless, the differences are not so dramatic and operatively it is probably easier to find the initial slope compared to the steady temperature. Actually, the concept of stabilization temperature is not so clearly defined and for the studied materials, such a stabilized trend was not always evident. Moreover, for the tests at low load levels, the stabilization temperature can be near, or even lower, compared to the thermal camera measurement uncertainties. In line with these considerations, for AN a very low dispersion of the data is associated with the relatively higher temperature rise, which is reflected in the quality of the data acquired. In this respect, our findings underscore the importance of studying fatigue across multiple specimens as well as the necessity of ensuring a noticeable temperature rise.

Fig. 3. Thermal indexes dT/dt, DT steady, Q for different HT condition (elaboration with OCM method).