PSI - Issue 7

Marton Groza et al. / Procedia Structural Integrity 7 (2017) 438–445 M. Groza et al. / Structural Integrity Procedia 00 (2017) 000–000

443

6

Fig. 3. (a) parameter identification for the Crossland criterion, (b) parameter identification for the DSG criterion, (c) experimental data and the Kitagawa curve simulated with the DSG criterion for tension and torsion R-1 (d). Experimental data from Nadot et al. (2004).

Fig. 4. (a) analysis setup with the designations of the different analysed surface defects, (b) von Mises equivalent stresses with the deformed shape with enlarged scale.

of different extent, elastic-shakedown is reached in the second loading cycle. The amplitude and mean stress is evaluated after the second cycle, since this behavior is considered to be representative for the majority of fatigue life. Computation with the EIM has been conducted for D2 and D3, because the analytical formulas have not yet been coded for the hemispherical shape of D1, which is only a technical problem. Some of the difference between the analytical ( ,max EIM Cr σ ) and linear elastic FE-results ( . ,max FEA elas Cr σ − ) can be explained with the fact, that EIM considers an inhomogeneity embedded in the matrix, not one on the surface. This has been detected and quantified by Vincent et al. (2014). However, the major contributor in the inaccuracy are the numerical errors originating from the CAD and FEA geometry representation, even with extreme high mesh density. The same location in the FEA model space and EIM are not clearly equivalent, which makes the comparison more difficult. The FE and analytical ellipsoid surfaces are not exactly at the same location, which leads to significant error when dealing with stresses with high gradients. After some calibration, accuracy of the EIM solution could be greatly improved in general cases. In Fig. 5b. the stress distributions in surface normal direction from the hot-spot are compared for the three different computations methods. The Crossland equivalent stress fields and DSG utilization factor plots in Fig. 5 c-h. (computed with elastic-plastic FEA) help to visually interpret the results. The DSG utilization factor ( DSG k ) plot is a practical way