John Wintle is a Consultant Engineer in Structural Integrity. He is responsible for developing reliability engineering at TWI and Inspection Qualification.
Richard Jones is Manager of the Arc and Surfacing Section of TWI where he has been involved in development and application of arc welding process technologies.
During a routine inspection of Reactor two at Chapelcross nuclear power station in 1997, a crack was found in the pressure retaining shell of heat exchanger six. Although difficult to detect in the dark confined space inside the warm heat exchanger, further examination revealed that the crack was large, extending about 230mm along the surface and up to 90% of the wall thickness. As John Wintle and Richard Jones report, there was no threat to nuclear safety with the reactor shutdown, but it was clear that a full investigation and a major repair were necessary before the reactor could safely re-start.
There are four Magnox type gas reactors at BNFL's works at Chapelcross near Annan in Dumfrieshire ( Fig.1). These reactors, together with those at Calder Hall, were the first power producing reactors to be constructed in the UK, and have been operating since 1957. Each reactor has four heat exchangers (boilers), positioned in a gantry outside the reactor building ( Fig.2), and gas is circulated to and from the boilers by ducting. Steam generated in the boilers is passed to drive turbines producing electricity for the grid.
The heat exchangers are massive cylindrical steel vessels with domed ends, some 23.5m tall by 5.3m in diameter, with a shell thickness of 35mm. Inside they are packed with banks of tubes in which water is heated to steam by hot gas from the reactor. The crack was located close to and running parallel with the main circumferential seam weld joining the barrel to the bottom dome, breaking the inside surface at the toe of a fillet weld attaching an internal bracket to the shell.
The metallurgical investigation required taking through-thickness core samples containing sections of the crack to establish the cracking mechanism. The texture of the crack surfaces indicated an arrested brittle fracture. A complete covering of black oxide confirmed that it had existed from before the stress relief heat treatment carried out during original construction in 1956, and it had not grown during over 40 years of service. The other heat exchangers were checked for cracking in equivalent positions, but none was found.
It was possible for BNFL to make a case to continue to operate Reactor two on three heat exchangers, while heat exchanger six was taken out of service and disconnected from the rest of the primary circuit. However, the loss of a heat exchanger reduced steam generating capacity and this had an inevitable effect on the plant economics. Pressure was on to make a repair.
Problems for the repair
Taking the core samples had left through wall holes in the boiler shell. This, together with the requirement to cut out the crack, committed BNFL to a major repair of the opening in the shell if the heat exchanger was ever to operate again (
Fig.3). Options reviewed included a fillet welded blanking plate (not allowed by the codes), and a narrow gap plug would have required unfavourable welding positions. In the end, a 'letter box repair', filling the opening of the shell with weld metal through thickness, was the only feasible option, even though the volume to be filled was large.
A number of difficulties in making and justifying a welded repair became immediately apparent:
- No post weld heat treatment could be undertaken for risk of damaging the internal tubing banks. In the as-welded condition, residual stresses in the repair weld could be high.
- Special welding procedures would be necessary to avoid coarse grained microstructures in the intercritical and transformed heat affected zones in the as-welded condition.
- The dome end material was a silicon killed boiler quality steel, equivalent to BS 1501:161C, and therefore susceptible in the as-welded condition to strain ageing embrittlement and cracking in the sub-critical heat affected zone(HAZ).
- The repair could not be practically pressure tested with the internal tube banks and the boiler in-situ next to the reactor. This would have been a normal code requirement.
In addition, access within the boiler was extremely restricted and required suits to be worn and stringent confined space safety precautions. External access was obtained from a scaffolding platform some 6m from the ground. The whole area was within a radiologically controlled zone.
As well as overcoming the technical and physical problems, there was a need to create confidence in the safety of repair for continued boiler operation. Ultimately, the nuclear regulator needed to be convinced. All these things made this a very challenging repair.
A repair strategy is planned
BNFL consulted TWI at an early stage. With so many difficulties and potential problems, the strategy was to take all possible steps to maximise the integrity of the repair. The key lay in achieving a defect free weld with maximum grain refinement to limit embrittlement in the HAZ regions. Once completed, the weld had to be demonstrated to be defect free by a programme of non destructive examinations in which there could be the utmost confidence.
The planned repair was to be made by completely cutting out the crack and core holes leaving an elongated opening through the shell. The inside surface of the opening was to be covered by a backing plate flush with the inside surface. The opening was then to be filled by controlled deposition of weld metal from outside the boiler in a sequence that would optimise residual stresses and minimise the risk of cracking.
In order to develop a safety case, BNFL had to show by an engineering critical assessment (ECA) that the limiting crack size was well in excess of the threshold size for the detection capability of the NDE. The ECA required data about the fracture toughness of the weld metal and HAZ regions and the operational and residual stresses in the repaired condition. This data was obtained from a programme of fracture toughness testing and stress analysis.
The parties
In order to undertake such a difficult repair, BNFL appointed Mitsui Babcock (MBEL) as the main contractor. The expertise and quality of its welding team based at Tipton, was already well known to TWI, and the team had recent relevant experience from working on a similar boiler at Sizewell A nuclear power station. In addition, the MBEL Technical Centre at Renfrew had the necessary project management capability and engineering workshops, as well as specialist technical skills in residual stress analysis and measurement and non-destructive testing for site inspection.
TWI had a number of roles. In the initial stages, TWI's job was to confirm that the welding procedure proposed by MBEL had been optimised by producing some trial welds. As the project developed, TWI undertook the tasks of testing the fracture toughness of the parent plate, weld, and heat affected zone materials, and of interpreting measurements of residual stress. In addition to these technical contributions, TWI was also asked by BNFL to peer review all technical work connected with the repair and to set up an inspection qualification body.
The TWI project was led by Richard Jones of the arc welding team with John Wintle of Structural Integrity responsible for leading the peer review and the inspection qualification body. Materials testing was co-ordinated by Mike Dawes. Peter Hart and Adrienne Barnes provided metallurgical support, and residual stress work was undertaken by Rick Leggatt.
Management of the technical work and development of the safety case was the responsibility of BNFL. A Technical Group was set up under the BNFL Project Manager to control and co-ordinate the technical work, and regular and lively meetings were held over a period of fifteen months as the technical difficulties were faced. Maintaining links with the station was crucial. They had to live with the result.
In order for the repair to be accepted and the boiler returned to service, BNFL had to satisfy a large number of parties. The approval authority for design, welding and site inspection was Eagle Star (now Zurich) Engineering. Lloyd's Register was appointed as the independent nuclear safety reviewer. The safety case had to be passed by BNFL's own Safety Committees and advisers. Ultimately, the assessors of the Nuclear Installations Inspectorate had to be satisfied.
Validating the repair procedure
The selected repair approach was to carry out a letter box type repair weld. This required the complete removal of a rectangular piece cut through the shell wall containing the holes left by the core sampling and the remaining existing defect. Because it was not possible to perform a stress relief heat treatment on completion of repair welding, MBEL proposed a repair weld procedure based on a controlled deposition MMA technique.
In this procedure, the surfaces of the weld preparation were to be buttered using low heat input weld beads ( Fig.4). Subsequently further weld buttering layers were to be deposited using controlled welding parameters so that the underlying coarse grained HAZ was refined and tempered. The intention was to generate a fine grained HAZ immediately adjacent to the fusion zone whose toughness and crack resistance was comparable with that of a heat-treated HAZ.
The remaining weld procedure was to follow a deposition sequence, which was to be modelled by finite element analysis to optimise, the distribution of residual stresses. Special batches of electrodes were also to be made to match the weld metal strength more closely to the base material than would otherwise have been possible using commercially available electrodes.
A weld repair procedure qualification test plate was manufactured and successfully tested in accordance with the requirements of ASME IX. In parallel, welding trials were conducted at TWI in order to confirm that the weld repair procedure had been optimised in terms of HAZ microstructural refinement. The limits to the procedure were explored such as the thickness of the buttering. Through this process, confidence was gained that likely variations in parameters during the actual repair would not lead to an unacceptable loss of HAZ microstructural refinement.
In view of the weld repair location and extremely restricted access to the internal side, it was important to demonstrate that the machining, welding and subsequent grinding operations were feasible under site conditions. In addition, it was also necessary to qualify the particular operators who would be involved for tasks required of them.
A full size mock-up of the heat exchanger shell in its site condition was constructed in the MBEL's engineering workshop in Renfrew. This represented access and working conditions that would apply on site including the attachments and internal constraints, although the mock-up was not set 6m above ground level. The machining operations required careful consideration of the tool design and fixtures to ensure that the preferred joint geometry could be produced in situ.
It was recognised at the outset that welding to the close tolerances required by the controlled deposition procedure would require highly skilled welders. For this reason, welders who had previous experience of similar welding on the Sizewell 'A' boiler repair were employed for the task. Successful completion of the repair weld on the mock up gave the necessary confidence that similar results could be obtained on the Chapelcross boiler shell itself. However, qualification of the inspection and the development of the safety case were required before the preparation was allowed to be put to the test on site.
(To be continued in next issue)
