PSI - Issue 7

E. Vacchieri et al. / Procedia Structural Integrity 7 (2017) 182–189

188

E. Vacchieri et al. / Structural Integrity Procedia 00 (2017) 000–000

7

instruction, in order to obtain its associate Weibull plot, can be used to predict as a function of cycles the percentage of not repairable blades in the set. The prediction based on Weibull statistics are in good agreement for the two considered critical locations with respect to the creep-fatigue lifing procedure evaluation. The agreement between the two predictions validates the developed lifing procedure and verifies the applicability of Weibull statistics to field feedback data.

3.4. Maximum allowable defect size for safe operation through Weibull statistics

To be sure that the crack length corresponds to a ”safe” crack depth, some further analyses have been performed. To obtain a Weibull curve describing the probability to have a crack so deep to be critical for the mechanical integrity of the blade, two information are necessary: • the critical depth of the crack; • the correlation between surface crack length and crack depth. FE simulation and crack growth calculation allow the evaluation of the critical crack depth, d c , considered danger ous for the blade mechanical integrity, on the basis of the stress, local geometry and wall thickness. Moreover, d c can be evaluated on the basis of the repairability of the di ff erent critical locations of the blade. For the correlation between length and depth, destructive analyses of some operated blades have to be done to experimentally determine this relationship. In fact, all the data that has been statistically treated up to now come from non-destructive examinations of operated blades, having information only about surface crack size. The probability functions for the di ff erent crack sizes can be used for the evaluation of the crack growth rate on the blade surface. This evaluation is completely based on a statistical approach and is not based on fracture mechanic calculations. However, considering that temperature and stress value are locally constant, a good approximation of mean crack propagation rate on blade surface can be achieved. The relationship between crack surface length and crack depth in the substrate can be used to evaluate the growth rate in the base material. The experimental data come from the results of destructive examination on seven operated first stage blades. For each blade a cracks mapping on the surface has been done through a comprehensive visual inspection after FPI. The length of each identified crack has been measured. Then, the maximum crack depth in the substrate has been measured through a complete stratigraphy of each studied crack. Finally, the length and the maximum penetration of each crack have been plotted as reported in Fig.6-a.

Fig. 6. (a) Linear relationship between maximum length and maximum depth in the substrate for all the cracks observed in a critical location, (b) correlation between maximum crack depth, considering the blade repairability criteria, and NSU for all the destructively analysed blades.

The relation between surface length and crack penetration in the substrate is close to be linear and it shows that the minimum observable surface length corresponds to the beginning of the propagation in the substrate. Since the metallic coating of the analysed blades does not represent any resistance to the crack growth, as reported in Vacchieri (2017), crack propagation in its whole thickness can be considered to be instantaneous. So, the propagation time