Bringing Beznau 1 back online21 June 2018
Beznau 1 recently returned to service after the Swiss regulator confirmed aluminium oxide inclusions in the reactor pressure vessel do not affect the unit’s safety case. Michael Dost looks back at over three years of testing and analyses that helped get the world’s oldest PWR back into operation.
IN EARLY SUMMER 2015, DURING an in-service inspection of the base material of the reactor pressure vessel (RPV) at Beznau 1, reportable ultrasonic testing (UT) indications were recorded. To determine the root cause of the indications and their effect on the vessel’s material properties more analysis was carried out. To co-ordinate these investigations and base them on state-of-the-art scientific knowledge, a project plan was established and approved by the regulator and its independent panel of experts.
Over the following almost three years, Axpo investigated following the plan in the roadmap. The project comprised three sub-projects (NDE and validation; root cause analysis and material characterisation; and structural integrity analysis), all of which interacted with one another. In December 2017, the final safety case was submitted to the Swiss nuclear regulator.
On 6 March 2018, the regulator issued a statement of acceptance of the safety case, with a permit to load fuel into the vessel. Beznau 1 was back on line by 19 March.
Design characteristics of the vessel
The primary systems and components at Beznau were designed and fabricated by or for Westinghouse Electric Company in the mid 1960s, based on the then-current versions of the ASME boiler and pressure vessel code, Westinghouse’s specifications and the internal guidelines of the steel manufacturer (Société des Forges et Ateliers du Creusot “SFAC”). The vessel head was replaced in 2015; the remaining major components of the vessel comprise the flange ring A, nozzle ring B (four nozzles), upper and lower core rings C and D, transition ring E (welded from three forged sections E1, E2 and E3) and bottom dome F.
The steel used for the forgings corresponds to specification ASTM SA508 Grade 3 Class 1, a low-alloy aluminium-killed ferritic steel.
A review of the manufacturing documents of the vessel was conducted with Westinghouse and SFAC to identify possible causes of the indications. The review, together with ultrasonic modelling of the fabrication examination, confirmed that the ultrasonic investigation performed during fabrication was adequate. A key conclusion was that if hydrogen flakes had been present in the Beznau vessel, they would have been detected by methods applied in the 1960s. The review of the shop internal control documents revealed ultrasonic echoes in rings C and E1, which were not reportable per code requirements at the time and therefore were not included in the end of manufacturing report.
The Westinghouse specifications, which exceeded ASME-Code, were met. The document review demonstrated proof of compliance of vessel fabrication with the ASME Code ed. 1965 Sections II/III, as well as with the current code edition. Further, the indications reported in Beznau in 2015 were acceptable according to the current edition of the ASME Code, Section III.
Root cause analysis (RCA)
The analysis of potential root causes included all metallurgical origins, combining all available knowledge from steel and forging manufacturers and experts in their respective fields. The most important task in this analysis was to establish whether the indications originated from the manufacturing process or whether they were introduced during the service life of the plant. The potential mechanisms are summarised in Figure 1.
The conclusion pinpointed aluminium oxide (Al2O3) inclusions as the root cause of the indications; the aluminium stems from steel casting operations. At the time aluminium was customarily added to kill the steel prior to solidification and to bind any oxygen in the molten steel. The investigation eliminated all other possible causes.
The investigation also covered the identification of the leading crack growth mechanism, which was determined to be fatigue crack growth. Structural integrity analysis confirmed fatigue crack growth in the present configuration would be negligible over the remaining service life of the Beznau plant.
The inspections of the base material were performed by Dekra, Intercontrôle and Areva. The procedure using primarily focused 0° longitudinal wave probe (L0°MER) was specifically qualified to detect hydrogen flaking. With the Dekra results it allowed a comparison of Beznau 1 with other recent vessel inspections that identified hydrogen flaking as the root cause of reportable ultrasonic indications (eg Doel and Tihange in Belgium).
Irregularities in the material that may lead to ultrasonic echoes, such as cracks, inclusions, pores, voids and artefacts, are categorised based on their shape and orientation as planar, laminar or volumetric.
Intercontrôle reported several regions with a higher density of indications in the lower part of ring C, located randomly in a small band of around 350mm width in axial direction above the circumferential weld RN5 and at a depth of up to 50mm from the wetted surface. The areas in these regions where the standard procedure was not able to resolve single indications were named “extended areas”. They were re-examined using a refined inspection procedure which included a smaller step size in the axial and circumferential directions, providing higher lateral resolution of separate closely-spaced indications. The extended areas were also inspected using a qualified weld inspection technique in search of planar flaws. No reportable planar indications were detected.
It is important to understand that the resolved indications inside the extended areas exhibit ultrasonic characteristics identical to the ones recorded outside of the extended areas, the only difference being the physical distance between single indications. A total of sixteen extended areas were resolved, yielding 2689 discernible indications. Out of these, two were important to the success of the safety case:
(i) EA600: 130mm (axial) by 560mm (circumferential), containing more than 50% of all resolved indications, thus exhibiting the highest density of imperfections. This set the upper density limit used to determine the properties of material with Al2O3 inclusions.
(ii) EA740: 34mm (axial) by 480mm (circumferential), this EA is considerably smaller than EA600, but due to its location relative to the safety injection nozzle, it is leading in terms of the ability of the vessel to withstand a PTS event.
The Intercontrôle evaluation provided the size and location of all reportable ultrasonic indications. The sizing of the indications was based on the 6dB drop method, which typically oversizes reflectors smaller than the beam size of 4mm. The great majority of the indications were sized at 3-5mm (Figure 2). Numerical simulations suggested that the reflectors present in Beznau 1 were significantly smaller than the reported sizes. The reporting threshold of the indications was conservatively selected at -24dB (0dB being the reference value of a 2mm diameter side-drilled hole in an unclad block) based on the Intercontrôle qualification. The measured amplitude of most of the indications in the EAs was -14dB to -24dB.
Figure 3 shows the location of the indications, top view in the x, y plane, seen from the inside of the vessel. These indications can be represented in a cross-sectional view across the thickness of the vessel (Figure 4).
From the onset of the project it was obvious that to demonstrate the integrity of the Beznau 1 reactor vessel, Axpo would have to locate a considerable quantity of suitable test material:
- to conduct extensive material tests, particularly focused on determining any deviations of the reference temperature in the brittle to ductile transition region, given the presence of aluminium inclusions;
- to validate that the ultrasonic test equipment will accurately record flaws in the base material;
- to perform chemical and hardness tests on material with inclusions.
Unfortunately, a global search extending over many months remained largely unsuccessful; consequently, a decision was taken to fabricate a new ring C, a replica, duplicating as much as possible the exact metallurgical features and properties of the original forging. Ring C was chosen over other rings because it exhibited the largest number of indications.
In summer 2016, Sheffield Forgemasters were commissioned with manufacturing the replica, based on the original specifications and shop documents, manufacturing methods and the outcome of interviews with the shop specialists of Creusot Forge, who provided invaluable expertise on processes that were in use some fifty years earlier when the original RPV was manufactured.
The complete process from casting to final product was meticulously monitored, with the prescribed heat treatment sequences. Both the original ring C and the replica were subject to two heat treatments: primary (precautionary) and quality heat treatments.
Comparison ring C vs. replica C
Analyses found that Creusot and Sheffield exhibited identical compositions of the heats for about twenty chemical elements. In terms of those elements identified as being the major contributors to neutron embrittlement, namely phosphorus, copper and nickel, their presence in the replica as a whole, the segments used for specimen extraction and the specimens themselves was meticulously analysed in order to demonstrate the representativeness of the replica. The results are presented in Table 1.
The areas in the vicinity of the Al2O3 inclusions were investigated for local accumulations of P, Cu and Ni; no such regions with chemical imbalances exist, neither in general, nor due to the presence of aluminium inclusions, as shown using EDX spectrometry.
The ultrasonic fingerprints, were virtually identical (Figures 5&6).
The ferritic–bainitic structure was found to be a close match between replica and ring C (Figure 7).
A large number of metallurgical analyses were conducted by slicing suitable test material every few millimetres, to validate the accuracy of the ultrasonic test equipment, and provide a qualitative comparison of the true nature of the ultrasonic indications. The material shown to the right in Figure 8 was sourced from a partial specimen of the acceptance test ring C; shown to the left is replica material originating from segment MQ1965, micro section S1.
The single most important task was to provide evidence that the replica is representative of ring C; any specimen test results derived from replica material are only of value if it represents ring C and specifically the areas of interest. A list of criteria permitted the unambiguous conclusion that it was representative. The UT scans of the replica matched those obtained from ring C strikingly well (Figure 9).
In a further refinement, more quantitative methods of comparison such as the location of the inclusions and their density, their spatial distribution in the base material, their ultrasonic echo amplitude and amplitude distribution were also applied.
These characteristics provided the basis to compare zones with UT indications in the vessel and in the replica. Therefore, UT density and amplitude in extended areas in ring C were compared with zones in the replica selected for material test specimen extraction. The representativeness of specimens was demonstrated by comparison of C-scan patterns from the specimens of the replica and the governing extended areas of ring C.
Using replica material, the flaw detection, location and sizing capability of the IC L0°MER technique at the chosen reporting threshold was successfully validated. The selected reporting threshold of -24dB detects and reports all Al2O3 inclusions larger than 1 mm2, corresponding to an inclusion 2mm (circumferential) and 0.6mm (axial). Some higher amplitude indications were analysed with destructive testing during the validation, confirming their Al2O3 origin. The validation confirmed that the indications are caused by Al2O3 inclusions and their conglomerates. Neither the metallurgical nor the fractografical analysis showed any other sources.
Zones were selected based on UT characteristics and used to extract test specimens to determine material properties. The RCA, the fractography of the fracture surfaces and the ultrasonic validation all demonstrated that the indications are caused by Al2O3 inclusions and their conglomerates, which are not connected to the steel matrix. They are inert and consist of micrometer sized particles. There was no evidence for any other manufacturing effects.
The influence of the inclusions on material properties was assessed by grouping the representative specimens after testing. The number and size of inclusions in the process zone of the cleavage fracture of compact specimens (C(T)) and the fracture surface (tensile specimens) were investigated by fractographic methods. Specimens were grouped into specimens with no inclusions, small inclusions and large inclusions. Then the mechanical properties of the groups were compared. The value of the grouping criterion was based on the results of the UT validation, demonstrating reliable detection and sizing of flaws equal to or larger than 2mm linear dimension in axial orientation (T-L) and radial orientation (S-L) direction and larger than 0.6mm in tangential (L-T) direction.
The mechanical properties investigated were fracture toughness in the brittle to ductile transition range in all three orientations (T-L, L-T and S-L); hardness; yield and tensile strengths; ductility; fracture toughness in the ductile range (upper shelf). As evidenced by a very large number of specimens with inclusions in the process zone of the cleavage fracture, crack initiation never occurred in the vicinity of an Al2O3 inclusion, but rather followed the expected behaviour of the initiation site being associated with the presence of a carbide (Figure 10).
The master curve (MC) concept was applied to investigate the possible effects of Al2O3 inclusions and their conglomerates on fracture toughness based on specimen material from the replica.
The transition temperature (T0) was not negatively influenced by the presence of Al2O3 inclusions in the process zone of the cleavage fracture. Also, the presence of Al2O3 inclusions in the cleavage fracture process zones of the C(T)12.5, C(T)25 and tensile specimens had no significant influence on mechanical properties such as fracture toughness in the brittle to ductile transition and ductile range, tensile strength and upper shelf energy.
The chemical mapping showed that the matrix is not affected by the Al2O3 nonmetallic inclusions and their conglomerates regarding any chemical elements relevant to embrittlement. Neutron embrittlement is not affected by the inclusions.
The chemical mapping did not show any impurities in the matrix. Manganese sulfides, as detected on the fracture surfaces of specimens, are common in pressure vessel steels. Comprehensive investigations on replica material containing inclusions concluded that the fracture toughness is not affected by the MnS inclusions or the Al2O3 inclusion conglomerates.
Specimens extracted from replica material demonstrated that none of the relevant material properties were negatively affected by the presence of either large or small inclusions. Neither the reference temperatures (RTref) in unirradiated nor irradiated condition were affected by the presence of the Al2O3 inclusions.
A large number of micro hardness and chemical composition measurements were recorded on RPV and replica materials. When the structure of the steel matrix with inclusions was investigated it found:
- Al2O3 inclusions have no influence on the embrittlement behaviour of the RPV steel;
- More than 600 analyses show that Al2O3 inclusions have no influence on the chemical composition of the steel after solidification;
- More than 800 analyses show that micro hardness of the material is not affected by Al2O3 inclusions, even those nearby.
Structural integrity assessment (SIA)
The purpose of the structural integrity assessment (SIA) is to verify that the structural integrity of the pressure vessel is maintained under all operating and accident conditions. The ASME Boiler & Pressure Vessel Code applies because the design and construction of Beznau were based on it and the legal requirements of the Swiss regulator refer to it.
The acceptability of flaws is assessed by ASME Section XI, rules for in-service inspections of nuclear power plant components (Figure 11).
A crack is the governing model to conservatively determine the effect on structural integrity of a material imperfection in vessel steel. A crack oriented normal to the maximum principal stresses governs assessment of the structural integrity. This principle is applied to the 3D boxes projecting the enveloped indications for the purpose of evaluation into axial and circumferential planes. Proximity rules are used to group flaws according to ASME Section XI, IWA-3300. The flaw assessment considers stresses and temperatures for all the governing operating and accident conditions. For the governing accident condition (PTS) the methodology is the same the standard analysis for Beznau.
The vast majority of flaws met the acceptance criteria per ASME XI, tables IWB-3500; some areas required analytical methods and a few were qualified using the numerical approach. All were found to be acceptable.
Proof of integrity of the vessel is demonstrated by ruling out crack initiation. In the schematic Figure 12 shown below this goal is achieved if the loading curves of the material (red loading curve) remain to the right side of the green resistance curve with an adequate margin. Increasing neutron embrittlement of the vessel material causes a shift of the resistance curve towards higher temperatures, reducing the available margins.
The influence of neutron embrittlement on material properties is being monitored in a surveillance programme at Beznau from the onset of operations. The last capsule in unit 1 was removed from the vessel in 2011, with an equivalent service life of more than 65 years.
Given the specific location of some of the areas with inclusions in the vicinity of the cold plume during a PTS event, numerical methods had to be deployed.
The origin of the UT indications is fully understood and supported by state-of-the-art knowledge about the specifics of manufacturing steel in ingots.
The aluminium oxides found in the RPV base material have no negative effect on the mechanical properties of the steel in the unirradiated state; further, they have no effect on the mechanisms defining neutron embrittlement; so evidence about embrittlement behaviour of material without nonmetallic inclusions remains valid.
The 2015 in-service UT examinations of the pressure vessel were adequate.
The integrity of the pressure vessel is ascertained for all operating and accident conditions and all technical requirements are met, even for the very conservative approach of modelling the Al2O3 inclusions as cracks.
Michael Dost is Head of Beznau nuclear power plant