PSI - Issue 19

Helen Bartsch et al. / Procedia Structural Integrity 19 (2019) 395–404 Helen Bartsch, Benno Hoffmeister, Markus Feldmann / Structural Integrity Procedia 00 (2019) 000 – 000

399

5

Fig. 3. Fractures of a large-scale specimen C, small-scale specimen TB and a cracked bolt of series A

The tests results of series B and TB as well as C and TC can be found in Fig. 4. For the statistical evaluation of the fatigue tests data the prediction interval has been used, which is mentioned as approach in the background document to EC 3-1-9 (Sedlacek, G. et al., unpublished) and also incorporated into Annex D of EC 0 (Drebenstedt, et al., 2018).

1000

1000

100 Stress range ∆σ [ N/mm²]

100 Stress range ∆σ [ N/mm²]

Small-scale tests TB Large-scale tests B Regression line m = 3 Design S-N-curve m = 3

Small-scale tests TC Large-scale tests C Regression line m = 3 Design S-N-curve m = 3

10

10

1.E+04 1 ∙ 10 4

1.E+05 1 ∙ 10 5

1.E+06 ∙ 1 6

1.E+07 ∙ 1 7

1.E+07 ∙ 1 7

1.E+05 1 ∙ 10 5

1 ∙ 10 4

1.E+06 ∙ 1 6

1.E+04

Number of cycles to failure N [-]

Number of cycles to failure N [-]

Fig. 4. Test results and design S-N-curves of series B and TB as well as C and TC (from left to right)

Series B and TB where failure occurred in the weld root show a large scatter, which is characteristic for this detail. Imperfections and flaws in the weld root can oftentimes not be detected, which is why test results of this detail show a great variance. The statistical evaluation of all failures in the weld root result in a reference value of fatigue strength of Δσ c = 47.6 MPa at N C = 2 · 10 6 . This results leads to a detail classification of the weld root into FAT class 45, which exceeds the current class 36* or 40 incorporated into (Eurocode 3-1-9, 2005). The S-N diagram of series C and TC, where failure occurred at the weld toe, shows a smaller scatter. A reference value of fatigue strength of Δσ c = 78.2 MPa at N C = 2 · 10 6 results, shortly below FAT class 80 of (Eurocode 3-1-9, 2005). When modified nominal stresses are taken into account, a fatigue strength of Δσ c = 100 MPa at N C = 2 · 10 6 is achieved. Fig. 4 illustrates, that the fatigue strength of the large-scale specimens shortly lies below the results of the small-scale tests. Regarding series A and LA, where failure occurred in the bolts, additional numerical simulation is required in order to take into account bending stresses of the bolts not measured by strain gauges. During these simulations further described in section 4, also prestressing losses have been taken into account. Under these circumstances, a rather high reference value of fatigue strength of Δσ c = 75.1 MPa at N C = 2 · 10 6 results for the bolts, as seen in Fig. 5. The right hand side of Fig. 5 illustrates the tension as well as the bending stress of one bolt in an LA fatigue tests,