PSI - Issue 19

Jan Papuga et al. / Procedia Structural Integrity 19 (2019) 405–414 Author name / Structural Integrity Procedia 00 (2019) 000–000

410

6

Fig. 2. S-N curves of unnotched specimens (left) and of 5 types of notched specimens (right).

Results of the tests on the various notched specimens can be found in Tab. 3 and also in Fig. 2, right. If the experiments on unnotched specimen show the reasons for using the bent Kohout-Věchet curve, Kohout and Věchet (2001), the common Basquin curve regression is sufficient for the notched configurations, see Fig. 2. 4. Analyses 4.1. FEA-models Contrary to the common engineering practice, the FE-models of notched specimens were prepared with extremely fine mesh, so that the stress gradient and also the effective stress in the critical distance could be correctly read. Tab. 4 documents the size of the mesh used at critical localities of individual notches. There is no explicit recommendation, which equivalent stress should be used to represent well the notch effect of axially loaded specimens. Therefore, Mises stress and first principal stress PS1 were evaluated typically.

Table 4. Description of FEA-models of notched specimens and of obtained parameters for Mises equivalent stress.

FEA-model and its output

Notch type

Mesh size [mm]

K t [-]

 ‘ [1/mm]

U-notch

0.020 0.014 0.014 0.052 0.074

2.11 2.51 2.19 2.42 2.89

1.807 7.185 1.760 1.449 1.349

Fillet

V-notch

Hole

Bi-notch

4.2. Type of evaluation Many variants of all three approaches were tested, and the most interesting outputs are presented in Tabs. 5, 7-8. The output values there concern the relative percentual error  FI in estimating the experimental stress values at given section cuts through S-N curves at 20, 50, 100, 200, 500 and 1000 thousands of cycles. Though different stress parameters are effective for different computational approaches, the parameter  FI is defined in all cases in the same