PSI - Issue 68

Lars A. Lingnau et al. / Procedia Structural Integrity 68 (2025) 303–309 L. A. Lingnau et al. / Structural Integrity Procedia 00 (2025) 000–000

307

5

was reduced by a factor of 0.89 to N f = 91 for a shoulder opening angle of 2α = 90°. A similar reduction by a factor of 0.87 was observed at a total strain amplitude of ε a,t = 0.0025, where the specimen with a shoulder opening angle of 2α = 30° exhibited a higher number of cycles to failure (N f = 361) compared to the specimen with a shoulder opening angle of 2α = 90° (N f = 314). Overall, the load curves for both damage states are comparable, showing continuous cyclic softening, a phenomenon also observed in other tests without phase shift. The resulting material responses, in terms of force and torsional moment, were smaller compared to tests conducted without phase shift. 3.2. Microstructure analyses In order to investigate the damage mechanisms and the interaction between fatigue loading and forming-induced ductile damage, the fracture surfaces of the specimens subjected to axial-torsional loading without a phase shift (Fig. 5a)) and with a phase shift of d = 90° (Fig. 5b)) were analyzed microstructurally. Fig. 5a) shows the fracture surface of an axial-torsional loaded specimen without a phase shift in the LCF regime. This test was conducted with a strain ratio of R = -1, a total strain amplitude of ε a,t = 0.01, and a test frequency of f = 0.01 Hz, along with a superimposed angular amplitude of θ = 10° at the same frequency. The specimen failed after N f = 102 cycles. The fracture surface reveals multiple crack initiation points typical of the LCF regime and exhibits a combination of characteristic fatigue and ductile fracture areas. The cyclic damage progressed until the specimen fractured at an angle of about 45° to the maximum normal stress in the volume. The material degradation was further exacerbated by the development of longitudinal cracks. Compared to the fracture surface of a similarly loaded specimen with a phase shift of d = 90°, and with otherwise comparable test parameters and a number of cycles to failure of N f = 91, no significant differences in fracture morphology were observed (Fig. 4b)). The typical damage mechanisms of the axial torsional loaded specimens without phase shift were comparable to those seen on the fracture surface described before (Fig. 4a)). Regarding the influence of forming-induced ductile damage, it was observed that cracks predominantly initiated at MnS (manganese sulfides) inclusions or in voids located in their immediate vicinity. The significant relevance of MnS inclusions and voids on crack propagation was identified for both damage states independent of the phase shift (Fig. 4c)).

Fig. 4. (a) Macroscopic fracture surface of a full forward rod extruded specimen with a shoulder opening angle of 2α = 30° after axial–torsional fatigue stress without a phase shift, (b) a phase shift of d = 90° and (c) MnS inclusions on the fracture area.

3.3. 3D-model of the defects The Zeiss Zen Analyzer software was used together with the Zeiss Zen Intellesis module for image segmentation. Using a deep learning algorithm, models were trained to automatically and reliably segment voids in SEM SE images and MnS inclusions in SEM BSE images. This enabled automatic identification and segmentation of voids and MnS inclusions. Fig. 5a) shows the first steps of processing FIB-SEM SE images. From a series of 510 images, a region