PSI - Issue 38

Jinchao Zhu et al. / Procedia Structural Integrity 38 (2022) 621–630 Author name / Structural Integrity Procedia 00 (2021) 000 – 000

624

4

(

) 2

    

lognormal

ln ,  

:

FAT

ln FAT FAT

5.73 0.157



= =

(5)

ln ln

FAT FAT



It is noted that the FAT value is defined at 97.7% referring to mean value minus two standard deviations (Sonsino et al. (2012)). Referring to survival probability of 95% at two sides confidence level of the mean of 75% and considering the usual scatter of fatigue tests, the difference in terms of stress is less than 2% and thus may be neglected (Hobbacher (2016)). 2.3. Notch stress method with r ref = r actual + 1 mm and FAT 200 Fricke et. al (2007) proposed taking into account the actual toe radius in the definition of the idealized geometry and reducing the FAT value. The method is applicable when the actual toe radius is in the range 1-3 mm. The fictitious radius is defined as the actual radius enlarged by 1 mm, that is r ref = r actual + 1 mm (Fricke (2012)). The notch stress range is computed from the stress concentration factor at the fictitious notch using FEA as ( ) notch t a e ctua r f l 1mm S K r S r = = + (6) The FAT class is reduced to 200 instead of 225 and the same standard deviation as in Eq. (5) is assumed (Fricke (2012)). The FAT distribution is therefore given by ( ) 2 ln ln lognormal , FAT FAT FAT   :

    

5.61 0.157



= =

(7)

ln ln

FAT FAT



The FAT distributions for both NS methods (Eq. (5) and Eq. (7)) are shown in Fig. 2.

Fig. 2. FAT distributions for both notch stress methods according to Eq. (5) and Eq. (7).

2.4. Finite-Element analysis

The FEA is performed using ANSYS (2020) software based on a 2D plane strain model with quadratic displacement function. The geometry of the cruciform specimen, loading condition and the local geometry parameters are presented Section 2.1. A quarter of the specimens is modelled with double-symmetry boundary conditions. The