The application of stress intensity factors derived from linear elastic fracture mechanics (LEFM) to fracture at creep temperatures has been considered. From tensile creep rupture tests on single edge notched and notched centre hole specimens of solution treated A.I.S.I. type 316 stainless steel, it is shown that a LEFM approach is inapplicable to predicting creep crack growth rates, whilst the net section stress is found to correlate well with the crack growth rates. These observations have been explained by considering the creep relaxation that takes place at the notch root, smoothing out the local stresses and thus making the LEFM stress distribution inapplicable. The resulting stress distribution supports the observation that the net section stress is a successful criterion on which to predict creep rupture in stainless steel.
The limitations as a fracture mechanics method are explored and it is found that a criterion based on the amount of creep rather than stress would have advantages in some respects. In this context the "crack opening displacement'" and the "fracture angle" criteria are considered and their use is found to hinge upon the development of suitable methods for relating the local displacement to the applied stress.