PSI - Issue 19

Sandro Citarelli et al. / Procedia Structural Integrity 19 (2019) 336–345 Sandro Citarelli, Markus Feldmann / Structural Integrity Procedia 00 (2019) 000 – 000

341

6

e = 15 mm could be derived, which are the basis for the subsequent fatigue assessment of all investigated crane runway girders. Further influencing parameters have been investigated which cannot be dealt with in detail in this paper. The main influencing parameters that have been additionally investigated and taken into account to derive the fatigue strengths include multi-axial stress states, different damage hypotheses, different stress concepts and weld execution. These effects were considered very precisely and as individually as possible.

-100

-90 -80 -70 -60 -50 -40 -30 -20 -10 0

-80

-60

-40

e zz [µm/m]

SG FEM

-20 e zz FEM [µm/m]

0

0

-20

-40

-60

-80

-100

0 200 400 600 800 1000

distance to upper weld [mm]

e zz SG [µm/m]

Fig. 9. Validation FEM and measurements for one crossing (left) and for all crossings (rigth)

3. Statistical evaluation based on Maximum-likelihood

The data collected within the project (Citarelli, S. and Feldmann, M., 2018) needed to be evaluated using suitable statistical evaluation methods. As presented in section 2, a total of 66 crane runway girders were available for the evaluation. The special characteristic of the database, however, show only 3 of the 66 girders with real fatigue cracks on the web to flange connections, so that the remaining 63 girders would be regarded as runouts. If then a conventional statistical evaluation approach was used, consequently, all 63 runouts had to be excluded from the evaluation. Nevertheless, also the runouts without fatigue cracks contain inherent information on fatigue resistance. For this reason, the probabilistic model described below was used on the basis of the maximum likelihood method (D’Angelo, L., 2015; Pollak, R.D. and Palazotto, A.N., 2009; Pascual, F.G. and Meeker, W.Q., 1996; D'Angelo et al., 2014). The model can use the S-N-curve definition according to (EN 1993-1-9, 2010) for variable stress ranges, see Fig. 10 and Eq. (2). log ∆ = { 1 1 (log − log ) + log ∆ für < 1 2 (log − log ) + log ∆ für ≤ < log ∆ für ≤ (2) where log ∆ : logarithm of stress range ∆ log : logarithm of cycles : number of cycles corresponding to the fatigue limit according to EC3-1-9 ( = 5 ∙ 10 6 ) : number of cycles corresponding to the cut-off limit according to EC3-1-9 ( = 10 8 )