The ASME Boiler & Pressure Vessel Code Section III method for the evaluation of fatigue in nuclear plant components uses a fatigue design curve derived from the testing of standard cylindrical specimens to describe the fatigue endurance of austenitic stainless steel components. The test results describe the number of cycles to achieve a 25% total load drop within a standard specimen (approximately equal to a 3 mm crack) under membrane loading conditions and the design curve is commonly associated with fatigue initiation. However, for non-standard loading conditions, such as the case of a thermal gradient within a component where a crack may be growing into a decreasing stress field, this description of initiation may be overly conservative.
Alternative approaches, such as the total life approach, may provide better representations of fatigue life in real plant components. By separating out quantifiable portions of long crack growth (Stage II) from the current design curve, alternative definitions of initiation can be derived and subsequently used in conjunction with standard fracture mechanics in order to model fatigue more accurately.
In this paper numerical methods are used to model the fatigue crack growth between starting crack depths of 0.25 mm, and the depth associated with a 25% load drop in a standard cylindrical specimen. The numerical predictions are compared with striation spacings measured at a range of crack depths on the fracture surfaces of austenitic stainless steel specimens tested in both air and water environments, under strain control. A good correlation between numerical predictions and the measured striation spacings was obtained and the results are used to characterise different stages of fatigue cracking.
Based upon the methods developed in this paper, modified fatigue design curves, using alternative definitions of crack initiation, are proposed and their applications in total life approaches to fatigue assessment are discussed, based on a worked example of a thermally fatigued stepped pipe experiment.