A Worthy Note Issue in Fatigue Reliability Analysis and PSA: The Random Cyclic Stress-Strain Responses of a NPP Material

摘要

Attention is paid to the random cyclic stress-strain (CSS) responses of a nuclear power plant (NPP) material, 1Crl8Ni9Ti stainless steel pipe-weld metal. Intrinsic cause of the responses and non-conservative evaluation without considering the responses are indicated, and a probability safety assessment (PSA) with the responses is made. The observations of short fatigue crack behavior indicate that the intrinsic cause is the difference of microstructural growth conditions in the initial zone and the zones ahead of the tips of dominant crack, which results in specimen failure. Existence of the responses implies that a random strain applied history will be introduced under any of loading modes even a deterministic loading history. Taking into account the responses, a reliability analysis indicates that conventionally deterministic CSS model is non-conservative. In addition, the responses yield irregularly distributed virtual stress amplitude data. The virtual stress amplitude-fatigue crack initiation life (S-N) data have thus a character of double random variables. The PSA should be different from that using the test data of commonly stress-controlled group and maximum likelihood fatigue tests. By the probability-based S-N curves suggested to model the S-N data, a PSA indicates that the ASME Code-based design curves are non-conservative in the lower virtual stress amplitude regime and most conservative in the higher virtual stress amplitude regime. Therefore, the responses should be considered in the fatigue reliability analysis and the PSA of engineering practice.