PSI - Issue 76

Afshin Khatammanesh et al. / Procedia Structural Integrity 76 (2026) 115–122

118

3.2. Fractographic investigations The fracture surfaces of all specimens were analysed using scanning electron microscopy (SEM). Figure 3 shows examples of surface and interior failure for the three steel sheets investigated (overview images were taken with light optical microscopy). The inclusions observed at the origin of crack initiation were identified as TiN using energy dispersive X-ray diffraction. Surface failure occurred solely at the machined surface, as shown in Figs. 3(a), (c), and (e). The absence of surface failure from the original surface suggests the presence of compressive residual stresses, which impedes fatigue crack initiation. Interior failure originated mainly from the centre of the specimens with similar distances from the original surfaces, as shown in Figs. 3(b), (d), and (f). This indicates that tensile residual stresses are present that promote fatigue crack initiation.

Fig. 3. Fracture surfaces with high magnification of crack-initiation sites (SEM) of specimens tested at (a) σ a = 750 MPa, N f = 2.67×10 5 cycles; (b) σ a = 550 MPa, N f =1.58×10 8 cycles; (c) σ a = 880 MPa, N f = 4.90×10 4 cycles; (d) σ a = 580 MPa, N f = 2.60×10 7 cycles; (e) σ a = 830 MPa, N f = 1.20×10 5 cycles; (f) σ a = 550 MPa, N f = 6.72×10 7 cycles. Furthermore, the size of each inclusion was measured in terms of √ area , which – according to Murakami (2019) – is the square root of the projection area of the inclusion perpendicular to the loading direction. The inclusion sizes were analysed based on statistics of extremes and plotted in Fig. 4. This evaluation assumes that the largest inclusion within each specimen’s gauge section is the origin of fatigue failure; for more details see Murakami (2019). Since surface inclusions possess a higher crack-driving force, i.e., a higher stress intensity factor, only the sizes of interior crack-initiating inclusions are plotted in Fig. 4. It is obvious that the detrimental inclusions in material B are, in mean, significantly larger than those in materials A and C. The latter materials exhibit a very similar distribution of maximum inclusion sizes.