Abstract
The evaluation procedures for fatigue initiation of nuclear class 1 components are defined in ASME BPVC Section III NB-3200 (Design by Analysis) and NB-3600 (Piping Design). Design fatigue curves are provided to establish the suitability of a component for cyclic service and define the allowable number of cycles as a function of applied stress amplitude (S-N curves). The number of load cycles at a particular strain range is then divided by the cycles to failure to obtain a partial usage factor., and the cumulative usage factor (CUF) for the component site, calculated from the sum of the partial usage factors, must be less than one. The original fatigue evaluation procedures did not include the effects of the PWR or BWR coolant environments, but laboratory test data indicate that significant fatigue life reductions can occur under such conditions, depending on strain rates and temperatures. These observations led to the formulation of modified procedures, originally published in NUREG-CR/6909 which required the usage factors to be increased by an additional environmental factor, Fen, which accounts for the deleterious effects of high temperature water. An ASME Code Case N-792-1 has now been included in ASME Section III which is based on the NUREG-CR/6909 equations, with some minor modifications. The Fen factors are derived from testing of membrane-loaded solid round tensile or tubular specimens at different strain rates and temperatures. The data were obtained using simple triangular waveforms, i.e. at constant strain rate, and the temperature was also constant for each test. However, for components subject to plant loading, the situation is significantly more complicated, with most major transients being thermal in origin. For a thermal shock transient some key characteristics become apparent. These are (i) temperature is out-of-phase with strain (ii) strain rate and temperature vary through the cycle with a faster strain rate at the top of the cycle (iii) stress decays through the wall of the component. Several assumptions need to be made in order to simplify the assessment of these sorts of transients. Examples of such assumptions include the choice of temperature for the calculation (e.g. maximum or average through the transient) and the method of strain rate calculation (e.g. assumption of constant strain rate, or integration through the cycle, i.e. the modified strain rate approach). These assumptions can be overly conservative and hence very restrictive for plant operators when making safety justifications. Improved models have been developed which weight fatigue damage through the cycle, which is consistent with recent observations from testing under complex load cycles. Although these models can more accurately predict fatigue life for loading that is representative of PWR transients, they still assume membrane loading which is unrealistic for thermal shock transients in thin walled components. Details of a testing capability at Wood (formerly Amec Foster Wheeler) or thermal shock testing in a PWR environment were presented in a previous paper (ASME PVP2018-84923). The predictions of fatigue initiation indicated test durations of 2–3 months based on the latest fatigue models for austenitic stainless steel. The current paper presents the results of the first thermal shock tests carried out on a type 304L stainless steel. The predictions are compared with experimental observations and the accuracy of the models are assessed.