Effects of Prior Stress Relaxation on the Prediction of Creep Life Using Time and Strain Based Methods

Abstract

A critical requirement for both next generation conventional and nuclear plants is the development of simplified inelastic design and fitness for purposes procedures that give a reasonably accurate prediction of the complex multiaxial time dependent stress strain behavior. The accumulation of this inelastic strain in the form of coupled creepfatigue damage over time is one of the principal damage mechanisms which will eventually lead to crack initiation in critical, high temperature equipment. Two main procedures that address creepfatigue loading are generally used, either a time fraction or a ductility exhaustion approach. It is generally accepted that these methods enable conservative predictions within a factor of 2 to 3 and hence are reliable methods for code based design and fitness for purpose type assessments. However, for complex cycles, this may not be the case; for example, prior relaxation cycles are found to accelerate the creep rupture of the material with the result that a significant reduction in creep life can be observed. An investigation was undertaken into the influence of prior relaxation on resultant failure using a typical low alloy ferritic power station steel. Both timebased and strain based methods were used to predict the damage caused by the stress relaxation cycles followed by operation at steady state. The predictions found that while ductility exhaustion methodologies based on mean properties were adequate in predicting the failure life, time fraction methods were found to be extremely nonconservative for mean properties and only lower bound solutions provided an estimate of remaining creep life. The ASME time fraction approach, using isochronous curves was found to be extremely conservative for K = 0.67, but was able to predict similar damages to ductility exhaustion when K = 1 was used. The MonkmanGrant approach resulted in predictions that erred on the conservative side. The results have implications for both current and future conventional and nuclear power stations as it may be difficult for time based approaches to account accurately for complex cycling, shakedown conditions or stress relaxation at welds.