PSI - Issue 7

Ivo Černý et al. / Procedia Structural Integrity 7 (2017) 431 – 437 Ivo Černý / St ructural Integrity Procedia 00 ( 2017) 000–000

435

5

24

X60_p1 X60_p2 X60_p3 X60_s2 X60_s3 X60_z1 X60_z2 X65_1 X65_2 X65_7 X65_9 X70_2 X70_6 X70_8 X70_9

23

22

21

20

a (mm)

19

18

17

0

50000

100000

150000

200000

250000

300000

Number of cycles

Fig. 6. FCG as a dependence on crack length on number of cycles in different specimens of the pipeline steels

a

b

c

d

Fig. 7. FCG from Fig. 5 evaluated as Paris diagram with regression lines

Fig. 8. Probability density of parameter C evaluated from Fig. 6, a) X70 steel, normal distribution, b) X65 steel, normal distribution, c) X60 , Weibull distribution and d) X60, normal distribution

3. Probabilistic assessment FCG data for the X60, X65 and X70 steel were used as a basis for probabilistic simulations and safety assessment of DN700 pipes of 10 mm thickness containing typical surface axial defect of 2 mm depth and surface length 20 mm. Note cracks of such dimensions are the smallest defects still detectable by NDT methods used for pipeline inspections. Crack of the depth 8 mm, i.e. 80% of total pipe thickness, was considered as a safety limit. The simulations were performed using ALIAS HIDA software recently developed during European project HIDA Applicability within the European Framework Programme – Deschanels et. al. ( 2006), Jovanovič et. al. (2001). The software uses methods of Monte Carlo simulations. Though the software enables to randomise many of parameters, including e.g. pipe dimensions, just parameter C of the Paris dependence was considered as randomly distributed. Otherwise, if parameter m also would be randomised, the approach and calculations would be much more complicated, as these two parameters are coupled – partially dependent on each other. On the other hand, the probability assessment would not be significantly improved – Černý (2004). Scatter of the parameter C was evaluated in terms of a distance of individual FCG points