A COMPARISON OF CREEP-FATIGUE ASSESSMENT AND MODELLING METHODS

Design codes, such as RCC-MRx and ASME III NH, for generation IV nuclear reactors use interaction diagram based method for creep-fatigue assessment. In the interaction diagram the fatigue damage is expressed as the ratio of design cycles over the allowable amount of cycles in service and the creep damage as the ratio of time in service over the design life. With this approach it is assumed that these quantities can be added linearly to represent the combined creep-fatigue damage accumulation. Failure is assumed to occur when the sum of the damage reaches a specified value, usually unity or less. The fatigue damage fraction should naturally be unity when no creep damage is present and creep damage should be unity when no fatigue damage is present. However, strict fatigue limits and safety factors used for creep rupture strengths as well as different approaches to relaxation calculation can cause a situation where creep-fatigue test data plotted according to the design rules are three orders of magnitude away from the interaction diagram unity line. Thus, utilizing the interaction diagram methods for predicting the number of creep-fatigue cycles may be inaccurate and from design point of view these methods may be overly conservative. In this paper the results of creep-fatigue tests carried out for austenitic stainless steel 316 and heat resistant ferritic-martensitic steel P91, which are included in the design codes, such as RCC-MRx, are assessed using the interaction diagram method with different levels of criteria for the creep and fatigue fractions. The test results are also compared against the predictions of a recently developed simplified creep-fatigue model which predicts the creep-fatigue damage as a function of strain range, temperature and hold period duration with little amount of fitting parameters. The Phi-model utilizes the creep rupture strength and ultimate tensile strength (UTS) of the material in question as base for the creep-fatigue prediction. Furthermore, challenge of acquiring representative creep damage fractions from the dynamic material response, i.e. cyclic softening with P91 steel, for the interaction diagram based assessment is discussed.