PSI - Issue 57

Sanjay Gothivarekar et al. / Procedia Structural Integrity 57 (2024) 487–493 S. Gothivarekar et al./ Structural Integrity Procedia 00 (2023) 000 – 000

492

6

In line with the observations of the stress-output, maximum values are found nearer to the grain boundaries. Subsequently, the fatigue life is calculated by finding the corresponding number of cycles in equation (1), using the fatigue properties stated in Table 1. The resulting field-output is overlaid on the grain structure in Figure 6 (b). Note that the scale is reversed and ranges from 210x10 3 to 250x10 3 cycles. The results confirm that the minimum fatigue life is found near the grain boundaries at the edge of the specimen, indicated by the red dotted circles. In addition, these grains exhibit a significant difference in stiffness. Nevertheless, the fatigue life plot suggests that a significant amount of damage is accumulated in the neighboring grains, as shown by the black dotted circles. In other words, the simulation demonstrates that crack initiation can occur around the edge of the specimen due to a combination of stress concentration and elastic heterogeneity of the peripheral grain structure. 3.3. Stress-life comparison To validate the numerical results, a comparison is made between the estimated (FEA) and experimental (EXP) stress-life data. The FEA data in Figure 7 includes the minimum fatigue life estimation obtained for every iteration of the finite element model. Here, a slight variety can be noticed in the estimated fatigue life, thereby representing a certain numerically-derived scatter. However, the average experimental standard deviation of 26x10 3 cycles remains approximately three times larger than the numerical standard deviation of 8.5x10 3 cycles. In other words, the current numerical framework does not fully capture the actual scatter found during fatigue testing. In future work, this factor might be reduced by introducing a larger range of elasticity, a varying distribution of section assignment, more variety in the average grain size, and so on.

Stress [MPa]

460

440

420

400

380

EXP FEA

360

1E+4

1E+5

1E+6

Fatigue life [cycles]

Fig. 7. Stress-life graph of experimental and finite element data at four maximum stress levels [2].

4. Conclusions From the current research several outcomes can be concluded. Given the fact that electrical steels are typically characterized by a non-oriented grain structure, the scatter on fatigue data can be significant. To this end, the implementation of Voronoi polycrystalline structures in FEA is useful for modelling the microstructural variation of different steel specimens for the same material. Especially when it comes to fatigue, where the lifetime estimations are usually quite sensitive to slight variations in local stress- and strain-output.