Abstract :
The failure of a circumferentially cracked stainless steel piping system can be predicted by assuming that failure conforms to a net-section stress criterion, using as input an appropriate value for the critical net-section stress together with a knowledge of the anticipated loadings. The usual procedure is to calculate the stress acting on the cracked section via a purely elastic analysis based on the piping system being uncracked. However (a) the piping is built-in at the system ends into a larger component, and (b) the onset of crack extension requires some plastic deformation. Consequently, use of the net-section stress approach to predict the onset of crack extension, can give overly conservative failure predictions. Earlier work by the author has quantified the extent of this conservatism, and has shown how it depends on the material ductility and the geometry of the cracked section. It has also been demonstrated that the extent of conservatism depends on the elastic flexibility of the system, and in particular on the location of the cracked section within the system. The present paper extends these earlier studies, and shows that the location for which crack extension is favoured is not necessarily that location where the stress (calculated on the basis of an uncracked system analysis) is a maximum, as is implicitly assumed when the net-section stress approach is used.
