SUMMARY OF THE SPECIAL SESSION ON EAF (2011) – Chaired by Anders Jenssen

Ten presentations were given in the special session on environmentally assisted fatigue (EAF), covering overviews of the existing knowledge, presentations on recent findings, and regulatory aspects.

Following a brief introduction by the session chairman, John Hickling gave a historical perspective on the subject. Early work on EAF in carbon and low-alloy steel (C&LAS) indicated that the ASME Section III design curve for ferritic materials was non-conservative for some lab data obtained in oxygenated high-temperature water. Other studies on cyclic crack growth rate testing of RPV LAS identified the important role of dissolved oxygen, while efforts by the ICG-EAC led to modification to the ASME Section XI code curves. The important role of water impurities and pitting during early cycling in reducing the fatigue life of C&LAS was demonstrated in a more recent study. In contrast, low oxygen environments (typical for PWRs) lead to a greater reduction in fatigue life than in oxygenated water (as in BWRs) for stainless steels (SS). Another difference between SS and LAS is the effect of flow rate, where high flow rates have a beneficial effect for C&LAS, but not for SS. Studies have also shown that the influence of the environment on fatigue of SS base metal also extends to weld metals and Ni-base alloys. Failure cases in Germany demonstrated the importance of slow, dynamic straining with localized, plastic deformation of the C&LAS, so called Strain Induced Corrosion Cracking (SICC). For SS, many failure cases caused by thermal fatigue have been reported, but the degree of environmental enhancement is not known. The presenter stressed that continued efforts to modify the ASME Section XI CGR curves for corrosion fatigue of C&LAS to better represent actual material behavior have so far been unsuccessful. The more recent work on S-N design curves involving an environmental correction factor was described. As a concluding remark it was stated that the situation concerning the adoption of new EAF design rules varies considerably around the world.

Claude Amzallag of Onet Technologies gave a presentation on EAF of SS in PWR environment. The presentation began with a short summary of failure cases related to EAF, where it was concluded that the service experience has so far been good in the low-cycle fatigue (LCF) domain, and failures have concerned the high-cycle fatigue (HCF) domain (thermal fatigue, vibrations, etc.). He then discussed the environmental correction factor, F_en, developed in Japan and the USA. Slower strain rates (down to ~10^-5 to 10^-6 s^-1 where saturation is observed) result in higher F_en, and consequently to shorter fatigue life, while F_en increases linearly with temperature. Regarding dissolved oxygen, it has been claimed that there is no effect on F_en. Recent laboratory data show that the fatigue life decreases with strain rate in the LCF domain. In strain-controlled tests, it was shown that secondary hardening at 300°C raises the 10⁷ limit, but much less so at 150°C, because the secondary hardening is not as pronounced at this temperature. The increased 10⁷ limit is believed to be caused by a continual decrease in the non-elastic strain as a result of the secondary hardening. Tests under load-control show that the fatigue life is shorter at 300°C than at 150°C. At a mean stress of 100 MPa, the fatigue life and the 10⁷ limit increases in air and the PWR primary environment, which is in contrast to 150°C, where the fatigue life and the 10⁷ limit decrease significantly. It was pointed out that comparisons between data from tests under strain or stress-control must be made with care, since the two types of tests results in different stress-strain histories. The presenter concluded with a summary of what he considered important areas for continued research. These involved a better understanding of secondary hardening, more data to assess heat-to-heat variability, additional testing to determine differences between the behavior in BWR and PWR environments, improved understanding of the role of surface finish and stress concentrators, and testing relevant to plant operation to assess the thresholds used in F_en analysis.

John Stairmand of Serco presented results from recent lab measurements of corrosion fatigue (CF) initiation and discussed the relevance of the results to PWRs. The tests were conducted at constant strain amplitude on 304L and 316L, at a temperature of 300°C in air or a PWR primary environment. Smooth specimen data were in good agreement with results presented in the literature. A ground surface finish reduces the fatigue life in air by a factor of 2.5 compared to polished specimens, whereas scratches and a ground finish reduce the fatigue life in PWR environment versus polished specimens by factors of 2.7 and 1.9, respectively. The presentation ended with a discussion on the transferability of lab data to actual plants, where it was concluded that the load forms applied in laboratory testing are simplified relative to the load cycles experienced by components in the plant. In addition, testing so far has been limited to isothermal conditions. A road map to address some of the issues related to the gaps in translation of lab data to the plant context was presented.

David Tice of Serco described recent developments in corrosion fatigue CGR testing of stainless steels in a simulated PWR environment. Substantial enhancement of the CF CGR can occur in PWR primary coolant, where the degree of enhancement increases with decreasing frequency and increasing rise time of the loading waveform. Enhanced CGRs have been observed for different heats of 304 and 304L at 250°C down to rise times of several hours. The behavior can be described by a time domain model. Under some circumstances, CGRs may be retarded relative to the fully enhanced rates. Temperature is one factor that promotes retardation of the CGR at long rise times, where the effect is more likely to be observed at 300°C than 250°C. Other factors promoting retardation are high R value, longer hold times, higher flow rate, and high sulfur content in the steel. Testing with sulfide injection to the crack tip demonstrated that retardation occurred shortly after injection of Na₂S. These results suggest that dissolution of MnS at the crack tip may sometimes be important, possibly by leading to crack blunting. There was also some evidence for differences in local deformation between steels with high and low sulfur contents, where the deformation tended to be more homogeneous in the high-sulfur steel.

Hans-Peter Seifert of PSI gave a presentation on the effect of static load hold periods on CF of austenitic stainless steel. A short background and a summary of previous observations at PSI are available in the presentation accessibly via the minutes, but – for reasons of time – they were not presented at the meeting. The motives for performing the tests were described. In the plant, there may be very long periods of moderate static stress, as opposed to most laboratory tests where the stress is cycled continuously. Regarding CF initiation, it was concluded that fatigue life increased with increasing static load periods (compared to continuous saw tooth loading) at P_max or P_mean. For CF CGRs of short cracks, there is little effect of hold time on the technical fatigue initiation life. If the material is sensitized, SCC usually dominates over CF. Regarding CF CGRs of solution annealed SS, there was no significant effect of long static hold times on either the short or long crack growth behavior in BWR NWC or HWC. Thus, there is no real effect of hold times on the technical CF life, and the current NUREG approach was considered adequate. Retardation of the CF CGR may occur under cyclic saw tooth loading, as well as under loading conditions involving long static hold times, but this response was not regarded as representative of the general behavior. It was concluded that saw-tooth loading usually gives an upper bound for environmental effects on CF (initiation and growth) at low as well as high corrosion potentials, if the contribution from SCC in sensitized SS is considered separately.

Armin Roth of Areva NP presented an overview of the proposed changes to the German KTA rules, and summarized recent results on the effects of hold time on EAF. The proposed revision of KTA is currently being discussed, and it has caused some controversy. If the changes become effective, environmental effects must be considered in more detail in the future. More specifically, the changes will require increased attention for components where the cumulative usage factor (CUF) exceeds 0.2 for SS and 0.4 for C&LAS. For cases where the CUF exceeds these limits, continued operation may be justified by inspection programs, relevant laboratory testing, or detailed EAF analysis. In laboratory CGR tests on LAS and SS in oxygenated water, recovery effects have been observed for long hold-times. Similar effects have been observed for SS in SN fatigue tests.

Takashi Hirano of IHI presented recent activities on EAF in Japan. Following a brief introduction and a historical overview, the basic concept for evaluating EAF was described. The concept applied in Japan involves an EAF correction factor, F_en. In 2009 the JSME code was revised to take into account knowledge gained from a Japanese national project. The revision involved the redefinition of F_en, where the strain rate thresholds for C&LAS, SS and Ni-based alloys were changed. In addition, the effect of strain holding has been added, a requirement that is applicable only in specific cases. One remaining issue in the code development is the influence of SCC on environmental fatigue of Type 316NG in a BWR environment. In 2007, JSME established a fatigue evaluation task group. The group consists of three teams; the design fatigue curve team, the environmental fatigue effect team, and the flaw tolerance team. The first team has proposed new best fit curves and design factors for SS and C&LAS. The second team focuses on mechanistically based fatigue evaluation and open issues, including the effects of oxygen and hydrogen (PWR environment) on SS, the effect of the oxide film for C&LAS (flow rate effects, strain holding, type of loading, etc.). Issues being dealt with by the flaw tolerance team are the assumed initial crack size (particularly the aspect ratio) and the NDE detection limit.

Al Ahluwalia of EPRI gave a presentation entitled “EPRI Environmental Assisted Fatigue projects”. Additional regulatory requirements regarding EAF have resulted in difficulties in demonstrating acceptable fatigue usage for new plants, as well as for license renewal approval. At present, there is a lack of clear design rules and this creates uncertainties and affects the overall costs to plant owners. EPRI is therefore involved in efforts to develop a technical basis for code modifications, to develop code cases that provide evaluation procedures for assessing fatigue environmental factors, and to support ASME Sections III and XI code revisions that include EAF procedures. Initial work (where a simple fatigue problem was solved) showed that there is no technical basis for the current requirement of CUF<0.1. A probabilistic approach showed that a higher CUF value can be justified. Work is also in progress to develop ASME code cases and to develop specifications for the FatiguePro 4.0 software. A gap analysis will describe a long-term plan to address EAF issues and to identify areas where further work is necessary. There are plans underway for Argonne National Laboratory to review code cases that have been adopted into the ASME code, Section III regarding EAF, as well as review available laboratory data produced subsequent to the publication of regulatory guide 1.207. EPRI has formed an international expert panel to review and advise the industry on technical issues and priorities.

Yrjö Hytönen of STUK gave a presentation entitled “Environmental Effects in Fatigue of Olkiluoto 3 Primary Circuit and Stainless Steel Components – a short overview of Finnish Nuclear Safety Authority”. STUK will adopt RG 1.207 in their new guide YVL E.4. Fatigue analyses have been performed for OL3 components using currently existing design curves. For selected locations, the factors characterizing environmental effects (F_en and allowable F_en) will then be determined. For cases where the environmental effects result in F_en lower than the allowable F_en, no additional analysis (with regard to fatigue damage) is necessary. If the environmental effects lead to F_en being greater than the allowable F_en, then additional analyses will be required. The locations for which environmental effects should be considered include components such as the RPV, steam generators and the pressurizer. For each component, a representative location must be studied, where the CUF is > 0.1 based on preliminary analyses excluding environmental effects, and the surface is wetted. The worst case should be selected for analysis if several locations in the component are subjected to fatigue. F_en factors are based on NUREG/CR-6909.

Darrell Dunn of NRC presented the NRC staff perspective and activities on EAF. For EAF evaluations, there are various options for carbon steel, stainless steel and Alloy 600. New or refined evaluations should use all six stress components and the analyst judgments need to be well-documented. The NRC staff positions on the four code cases (relevant to EAF) in development for ASME Section III were then presented. Regarding fatigue code case N‑761, it was not considered to have an adequate technical basis. The F_en code case N-792 was accepted by NRC, as it adopts the NUREG/CR-6909 methodology and the strain amplitude modification was considered conservative. NRC has not yet investigated the strain-rate code case, while the flaw tolerance code case was not accepted because allowance for the possibility of fatigue flaws in vessel design was considered contrary to the philosophy behind ASME Section III. Concerning EAF evaluation for new plants, NRC expects applicants to evaluate pressure boundary components exposed to reactor coolant using RG 1.207 methodology. The need to include RPV core support structures in EAF evaluations of new plants will be considered. NRC is currently involved in research to support the technical basis for high-energy line postulated break locations in operating reactors requesting license renewal and for new plant applications.

Summary of the panel discussion – prepared by Karen Gott and edited by John Hickling:

As an introduction to the panel discussion, Ian Armson provided a brief update on the progress of the ASME XI Committee to incorporate environmental effects on the stainless steel fatigue crack growth FCG of stainless steel in deaerated water. Revised curves were currently being drafted, but he highlighted that these curves were based on data and a model developed in an excellent assessment by Mills (Bettis) and presented at an EPRI MRP Conference in Colorado Springs in 2010. In view of the significance of this presentation, some key aspects were highlighted and the Mills presentation is now included in the minutes for this session. In this paper, Mills highlights that the corrosion fatigue databases developed at Bettis and reported elsewhere in the literature reveal FCG rates for 304 and 316 stainless steels in deaerated water that are significantly accelerated, with da/dN values typically one or two orders of magnitude greater than their air counterparts. The databases were critically reviewed by Mills to establish a FCG rate model that correlates rise time, stress ratio, ΔK, temperature and material. The steady state, non-retarded data were used to establish a Paris power law relationship for predicting nominal crack growth rates for wrought stainless steel in deaerated water. Comparison of the normalised crack growth rates for the entire database shows that the fatigue behaviour obtained by the various laboratories is in reasonably good agreement. Despite this good fit of the test data to the model, Mills makes particular reference in his abstract to Japanese data at an R of 0.95 which exhibit an exceptionally large effect. Peter Andresen had earlier commented on difficulties in testing at high R levels.

Mike Wright pointed out that there is a disconnect between laboratory data and field experience which may be because of the thresholds at which various effects start. Raj Pathania said that it is important to resolve this, since one must be able to define clearly when to apply Section III or Section XI. The definition of initiation leads to difficulties in deciding when to use Section III or Section XI; there is also an option to exclude Section III.

John Hickling noted that reaching a CUF of 1 was never meant to imply components would then contain cracks: it is only a design concept. In many cases, conservatism within the design procedure would have been sufficient for us never to see cracks in the field. The question is whether or we are going to retain this conservatism in the future, both with life extension and the use of more modern design approaches (e.g. using finite element analysis), In the case of LAS,  laboratory data and field experience are in good agreement. Hans-Peter Seifert agreed and considers that there is no real discrepancy between the laboratory and the field overall if the correct boundary conditions are taken into account, i.e. real laboratory conditions and real operational loads. Armin Roth wondered if we are only looking for the missing link – compare the situation with SCC 25 years ago. He suggested that we should make a coordinated effort to analyze differences in testing methods and details between laboratories.

Ulf Ilg said that, in Germany, plants are today very aware of loading transients, since they are required to perform extensive analysis as part of the AMP. Transients which previously were expected to lead to failure would be monitored today. An evaluation of the actual transients over a 10 year period is made, and can thus show that only 20 – 25 relevant transients have actually occurred, compared to the 100 that were considered in the design procedure. This means that the CUF can be re-analyzed.

Claude Amzallag noted that there is considerable conservatism in design curves, but there is insufficient conservatism on real stresses, plant cycles and strain rates.

Raj Pathania pointed out that plant cycles are often 1 -2 years, but few laboratory tests are run for so long. Are the laboratory tests in fact run for long enough to build up the right oxides on the specimens? John Hickling questioned if the strain rates in some of the tests are relevant – he noted that some cracked LAS piping in BWRs had been monitored and extremely low strain rates detected during startup. Martin Widera also pointed out that more extensive load following will increase the importance of fatigue analysis.

The panel discussion was concluded with the following contribution from Armin Roth as a comment to John Hickling’s remarks on the measurement of strain rates acting upon plant components during start-up:

The strain-induced “thermal” stress on components caused by thermal gradients, which per se cause thermal strain, can occur in three different ways:

1)      Stress from axial thermal expansion of pipes, e.g. on bends/elbows between straight pipes. This case causes primarily axial strain due to bending and related thermal stress. These strains can basically be measured at surfaces, provided they are large enough, i.e. greater than the resolution limit.

2)      Stress from inhomogeneous circumferential expansion due to thermal stratification of hot and cold water. This case causes primarily circumferential strain and related circumferential thermal stress. These can also be measured when large enough.

3)      Local thermal shock at surfaces due to injection of cold water into high-temperature-water systems, e.g. at mixing T-joints. In this case, no macroscopic change of total strain occurs, provided that the wall thickness of the components provides significant constraint, i.e. when it is large compared to the penetration of the colder temperatures into the wall thickness due to thermal shocks. The thermally expanded material would like to contract in thermally-shocked areas due to the lower thermal expansion at the correlated lower temperatures. This is, however, prevented by the geometric constraints and therefore causes a thermal stress without the occurrence of macroscopic strain (and related strain rate).
This latter case can easily be visualized by a rod, which is fixed at one end and not at the other. Increasing the temperature would cause free thermal expansion of the rod. If the rod is then also fixed at the other end, it cannot change its total elongation due to the fixation at both ends. If it is then cooled down, the thermal expansion tries to decrease accordingly, but is prevented from doing so by the existing constraint. This causes the establishment of a mechanical stress, because the prior thermal expansion at higher temperatures is maintained by the constraint and must be converted into corresponding mechanical strain. Since no macroscopic change of strain occurs upon cooling, no (macroscopic) strain rate is generated either: it is zero. Therefore, it cannot be measured by strain gauges. This is the reason why fatigue monitoring systems are based on temperature measurements. The microscopic (localized) situation is different, because cyclic thermal shock certainly leads to fatigue damage, e.g. as seen in Civaux. Regarding the environmental effect, it is, however, questionable how the protective oxide layer is affected by the macroscopic zero strain rate. This situation cannot be simulated easily in simple laboratory tests, nor is it covered by conventional mechanical testing techniques.
Splash tests used by some authors, where cold water is splashed onto the hot surface of specimens or parts which are heated in air, e.g. by induction, are therefore not representative of components exposed to HT-water during service. In this comparatively easy experimental set-up, both heat transfer and environmental conditions are significantly different as compared to actual component loadings. A realistic simulation of thermal shocks in HT-water environment and respective parametric studies of environmental effects has seldom (if ever) been performed anywhere in the world. Tests like this could be considered as innovative. To perform them is certainly not easy and requires considerable efforts regarding design, duration and cost.