PSI - Issue 61

Aliyye Kara et al. / Procedia Structural Integrity 61 (2024) 98–107 Author name / Structural Integrity Procedia 00 (2019) 000 – 000 3(a). Two different variance values ( 0 =3000 MPa 2 and 0 = 5000 MPa 2 ) were considered in order to observe the changes in the results for a moderate and larger amount of plasticity at the notch. The remote stress-time histories were generated by means of the WAFO toolbox by using 500000 points. They had a zero mean and skewness and a variance of 3007.2 MPa 2 and a kurtosis of 2.95 for the case of 0 = 3000 2 PSD, and a variance and kurtosis of 5011.8 MPa 2 and 3.05, respectively, for the case 0 = 5000 2 . After generating the stress-time histories, turning points were extracted and then used as input in the finite element model for the subsequent time-domain simulation. The use of turning points in place of stress-time values allowed the computational time to be shortened by four to five times. Fig. 2(b) compares the simulated stress-time histories with the sequence of turning points used in the finite element simulation. The output of the finite element analysis, next analyzed in Section 6, is the random stress and strain time histories at the notch tip. 103 6

b.

Fig. 3. (a) Power spectral densities (b) A sequence of stress-time history for 0 = 5000 2 .

4.3. Material model The mechanical properties of 30NiCrMo8 steel, taken from Rognon et al. (2014) and reported in Table 1, were used. The S-N curve is made to correspond to the elastic part of the Manson-Coffin curve; accordingly, the fatigue strength constant = 1/2( ′ ) − 1 ⁄ and S-N slope = −1/ can be calculated from Manson-Coffin- parameters. Table 1. Static and fatigue properties of 30NiCrMo8 steel Mechanical properties Coffin-Manson parameters E (MPa) 195000 b -0.085 ν 0.33 c -0.752 (MPa) 945 ′ 1.741 (MPa) 1135 ′ (MPa) 1750 In the finite element model, the material elasto-plastic behavior was simulated by a kinematic hardening model in Eq. (8), which correlates the stress amplitudes to plastic strain amplitudes for stabilized cycles (Benasciutti et al., 2018). = +∑ ℎ ( , ) =1 (8) Eq. (8) was calibrated on the cyclic stress-strain curve represented by a Ramberg-Osgood model. In the calibration procedure, the cyclic yield stress σ yc =814 MPa, corresponding to a 0.2% plastic strain offset on the established Ramberg-Osgood curve, was estimated first. By exploiting Eq. (8), the kinematic variables and were determined with the help of least squares fitting. Two sets of kinematic model parameters, and , =1,2 , were chosen, leading to the estimated kinematic parameters 1 =10558 , 1 =35 and 2 =7865 , 2 =124 ; they were input in finite element simulations.