Environmental effects on fatigue. Possible reasons for the apparent mismatch between laboratory test results and operational experience

The current fatigue design curves from the ASME Boiler and Pressure Vessel Code, Section III, do not explicitly cover additional environmental effects from LWR environments. From laboratory test results, accelerating effects by these environments on both, crack initiation and crack growth due to fatigue are clearly evident since more than 30 years. However, operational experience did not reveal a systematic degradation of properly designed and manufactured pressure boundary components in the water-steam-circuit due to environmental fatigue in LWR plants worldwide. Therefore, the current fatigue design curves seem to cover even environmental effects in LWRs, although they were not originally intended to do so. Existing cases of fatigue failures with environmental assistance may be well explained by deviations from the specified conditions, e.g. unexpected, additional mechanical loadings, which were not considered in the initial design, or deficiencies regarding material and fabrication. After application of appropriate remedial measures no further degradation or even failure were observed. Consequently, there exists an apparent mismatch between observations from laboratory tests and service experience. This mismatch is assumed to be predominantly the consequence of significant differences in mechanical and environmental loadings applied in laboratory tests as compared to actual loadings during plant operation. These differences could be firstly different shapes of loading signals, i.e. triangular or sinusoidal wave shapes in laboratory tests versus complex loading signals during real plant operation, and secondly, differences in sequence of environmental and mechanical load. In laboratory tests, environmental and mechanical loads are applied simultaneously and continuously until crack initiation has occurred and during further crack advance until specimen failure as defined by individual criteria is reached. In real plant cycles, severe transients, which mostly are present during start-up and shut down of plants, are interrupted by long hold-times at constant load during steady-state operation. In order to study long-term hold times without causing unacceptable test durations, a special testing sequence was developed. This sequence consisted of saw-tooth loading cycles, which were interrupted by hold-times after a measurable crack advance has occurred. Hold-times of 3 days and 30 days were used. The tests were performed with both low-alloy steel and an austenitic stainless steel. The test environment was oxygenated high-temperature water (240 C, 400 ppb DO). Fatigue pre-cracked 1T-CT-specimens were used to measure crack advance. The test results revealed a trend of decreasing cyclic crack growth rates da/dN at hold times of 3 days for both materials. At hold times of 30 days, both materials revealed a significant and reproducible decrease of da/dN, which was more pronounced in stainless steel. (orig.)