PSI - Issue 37

A. Vshivkov et al. / Procedia Structural Integrity 37 (2022) 570–575 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

574

5

Fig. 5. Dependence between stored energy per cycle and crack length

In figure 5, there is linear part, until the cracks have reached a certain critical length, about 15-20 mm in this case. In this linear part, we can predict crack propagation by value of stored energy. Thus the stored energy can be the criteria of fatigue fracture.

5. Analysis of acoustic and thermal data Clustering of AE signals was carried out by the k-means method, and its quality was assessed using the silhouette coefficient. Thus, as a result of the application of cluster analysis, two most probable clusters were identified. These clusters can be correlated with structural mechanisms of damage. Figure 6 exemplify the typical results of the AE clustering analysis. The cumulative AE energy (blue and red lines) and the heat flux (green line) are plotted as functions of crack length. The decomposition of the cumulative AE energy into clusters associated with different deformation/fracture mechanisms allows establishing a correlation between the modes of the energy dissipation and the predominant type of acoustic signals accompanying the deformation and fracture processes. The first stage is characterized by the nearly linear growth of the cumulative AE energy. The second stage is featured by the avalanche-like growth of the cumulative AE energy for both clusters. This moment correlates with the significant changes in the heat flux. Thus, the two independent methods suggest that the transition between notably different stages of fatigue crack growth does exist.