PSI - Issue 82

Peter Haefele et al. / Procedia Structural Integrity 82 (2026) 174–181 Peter Haefele and Patrick Schwarz / Structural Integrity Procedia 00 (2026) 000–000

180

7

3.3. Comparison of damage patterns under constant and variable amplitude loading of electrical steel Unnotched specimens with polished edges made from NO30-15 show no deviations in the location of crack initiation between constant and variable amplitude loading. The damage mechanisms observed in the Wöhler tests can therefore be transferred to service-like loading conditions. Since the notch ground cannot be polished without damaging the specimen, the reference condition of the sharply notched specimen with K t = 6.4 specimens is milled. The milled edge of NO30-15 exhibits an edge hardening of approximately 6 % at the notch root (Schwarz and Haefele, 2025). Under constant amplitude loading, crack initiation occurs on the surface of the notch root, see Fig. 5 (a). In some cases, cracks initiate at the base of small burrs, but these initiation sites show no deviation in the number of cycles to failure. Only one intergranular crack initiation was detected. This low occurrence is attributed to the plastic deformation of the notch surface, which is comparable to that observed in shear-cut edges. Under variable amplitude loading, similar failure behaviour is observed, which again indicates a good transferability between the different types of loading concerning the failure location. For unnotched shear-cut specimens tested under variable amplitude loading, no load dependence of the crack initiation site is observed. All specimens exhibit crack initiation in the fracture zone of the edge, which deviates from the results obtained under constant amplitude loading. It is assumed that under spectrum loading, only a limited number of cycles exceed the yield strength. Since the smooth-cut region exhibits lower hardening and a smaller increase in yield strength, it begins to yield earlier. This leads to stronger cyclic hardening of the smooth-cut portion, thereby reducing the difference in fatigue strength between the two regions of the edge. The earlier yielding also promotes a higher reduction of residual stresses. Due to the high number of low-amplitude cycles and the lower cyclic hardening of the fracture zone, roughness-induced effects become more dominant again. This explains the increased influence of the edge condition observed under variable amplitude loading, as also discussed by Schwarz et al. (2023) and Thum et al. (2021). Notched specimens made from NO30-15 with shear cut edges show a random crack initiation site, independent of load level, occurring in both the smooth-cut and fracture area. This behaviour is similar under constant and variable amplitude loading. The difference compared to unnotched specimens can be explained by several factors. Hardness measurements at the notched edges show a homogeneous hardness distribution in both regions (Schwarz and Haefele, 2025). This lack of difference is attributed to the higher degree of plastic deformation at the punching caused by the notch radius. Moreover, the high notch sharpness leads to local yielding at low load levels, which reduces roughness induced stress concentrations. In addition to the different hardening behaviour, it is observed that the notch root of some K t = 6.4 specimens consists almost entirely of smooth-cut regions, see Fig. 5(b). The punching process causes material displacement, producing a smearing of the fracture zone. These specimens show no deviation in fatigue life, supporting the conclusion that roughness-related effects are mitigated by the relatively high plastic notch strain. Most rotors made from NO30-19 exhibit damage patterns under variable amplitude loading that are comparable to those observed in constant amplitude tests. Only two tests conducted under the highest load spectrum show, in addition to a crack initiation at the notch ground of the saturation bridge, a seconded crack at the rolled surface near the notch root, see Fig. 5(c). Since these tests do not show any anomalies in the number of cycles to failure, the crack at the rolled surface is interpreted as a secondary crack.

Fig. 5. Fracture images NO30-15 and NO30-19 R σ = 0.1; (a) K t = 6.4 specimen with milled edge NO30-15; (b) K t = 6.4 specimen with shear-edge NO30-15; (c) Rotor NO30-19 after variable amplitude loading.