PSI - Issue 5

M. Dabiri et al. / Procedia Structural Integrity 5 (2017) 385–392 M. Dabiri et al. / Structural Integrity Procedia 00 (2017) 000 – 000

390

6

The loads were applied corresponding to the experimental values, and the model was run for ten virtual cycles to achieve stabilization. Although elastic-plastic analysis of notched components under cyclic loading can be performed via the same method of applying monotonic loads but using the cyclic stress-strain curve as the material response instead of the monotonic true stress-strain one (Zeng & Fatemi, 2001), it has been recommended to run the analysis for a few virtual cycles to allow the stress-strain values to achieve stabilization (Susmel & Taylor, 2015). It was, also, shown (Dabiri, et al., 2016) that using the maximum values on the first cycle (corresponding to the application of load in a monotonic way by using the cyclic stress-strain curve) could give slightly higher strain values, especially at notches with higher stress concentration factors. To obtain the value of the material characteristic length to be used with the point method by Taylor (Susmel & Taylor, 2010), the experimental results of notched components should be used for the calibration. The sharper the notch radius is, the more accurate the calibration process is. Although using one specimen for calibration is acceptable, more than one specimen can be used in the case of a high number of experimental tests. In this study, the specimen N3 was selected. The required steps, including experiments and modelling to determine the material characteristic length by TCD, are shown in Fig. 2. The total life to failure of this specimen (8715 cycles) was used to obtain the corresponding strain value ( ) using the Coffin-Manson equation. This value was then searched on the strain gradient from the notch root towards the centre (demonstrated in the cut view in Fig. 2) to determine its distance from the notch root, which correlates to /2 . Following the steps shown in Fig. 2, the value of = 0.506 correlated to /2 = 0.072 mm. Therefore, the material characteristic length of = 0.144 mm would be obtained for this material and used for the strain analysis of other specimens. It should be highlighted that the condition of having a fully reversed local strain ratio at the notch root must be satisfied for the method to yield accurate results. This was checked for all models in this study.

Fig. 2. Steps required in strain-life prediction for a notched member under constant amplitude loading using TCD.

6. Results and discussion A comparison of the results of both configurations is shown in Fig. 3. As expected, the predictions of Neuber’s rule led to highly unnecessarily conservative life estimations by overestimation of strain values at the notch root. In addition, the high strength level of the material under investigation resulted in a slight difference between the predictions using K t or K f . This difference was even smaller in specimens with blunter notches, which made the values of K t and K f fairly close. However, using the modified version of Neuber’s rule significantly increased its accuracy under plane strain conditions and led to better predictions than those of the linear rule, which is a simple approximation method under plane strain conditions. The SED method similarly failed to make accurate life estimations in its original form. However, it surprisingly yielded accurate predictions when it was used with the “biaxial plane strain” stress -