Sensitivity in Creep Crack Growth Predictions of Components Due to Variability in Deriving the Fracture Mechanics ParameterC*

Abstract

AbstractIn this paper the uncertainties in estimating C* creep fracture mechanics parameter, on component defect assessment, is considered. Comparisons of reference stress σref and the stress intensity factor K for a surface cracked plate under tension and bending and a pipe geometry under internal pressure and external bending using formulae from available codes of practices such as the R5/R6 [1-2], BS7901 [3] and A16 [4] are made with three-dimensional Finite Element (3D FE) analysis calculations. It is shown, using the same material properties and specimen dimensions, that the uncertainty associated with calculating σref could be as much as ±40% whereas for the stress intensity K it is about ±10%. The σref values are used to calculate C* using the mean properties of a base 316LN type stainless steel plate tested at 650 °C and a cross-weld P22 circumferentially welded pipe tested at 565 °C. Both the pipe and the plate were tested within a European collaborative programme ‘HIDA’ between (1996–2000) [5]. The tendency is that ‘global’ limit load analysis will give a lower and more conservative C* than the ‘local’ analysis. From the comparisons of the pipe and plate initial cracking rates, using different C* estimations, with standard creep crack growth rate data on compact tensions (CT) specimens, using the ASTME1457-2002 to evaluate C*, it is shown that predictions of cracking rates could vary by as much as two decades in magnitude. This inconsistency yields itself to an analysis based on an empirical model for predicting crack growth in components.