It came about following an innocuous regional power outage and concluded as a technical feat that would serve as the baseline for nuclear reactor corrosion detection, repair and mitigation around the world. “This project was one of AECL’s most challenging feats, involving first-of-a-kind technologies and processes that were developed by our own people,” said Bill Pilkington, then-senior vice-president and chief nuclear officer for Atomic Energy of Canada Limited (AECL).
Chalk River Laboratories is located near Deep River, about two hours northwest of Ottawa, Ontario. Its versatile NRU research reactor supplies approximately one third of the world’s supply of medical isotopes. It is the only reactor in North America that produces molybdenum-99, which is used as the parent to produce technetium-99m – the most widely used isotope in nuclear medicine.
In May 2009, following a class 4 power outage that affected most of eastern Ontario, including the Chalk River Laboratories site, AECL announced to its various stakeholders that a small heavy-water leak in the NRU reactor had been detected during routine monitoring while the reactor was being readied for return to service. The leak did not present any evidence of immediate safety issues, but AECL staff decided that the prudent approach was to keep the reactor shut down while the source of the leak was determined.
While the shutdown posed a significant concern for the medical industry, Chalk River leaders recognized that rushing through the repair process and returning the NRU to service without a solid, long-term corrosion mitigation strategy could create more serious material condition problems in the future.
Hank Drumhiller, vice-president of operations and chief nuclear officer said, “Every day that we were shut down meant we weren’t supplying the isotopes needed for thousands of medical treatments. Even in light of that pressure, our goal was always to focus on doing the work safely and correctly and ensuring that it would be a lasting repair.”
That attitude of conservative decision-making, commitment to safety and open communication would extend throughout what would become a 15-month effort to repair and return the NRU to service.
From the onset, AECL worked closely with the country’s nuclear regulator, the Canadian Nuclear Safety Commission (CNSC), and established an AECL-CNSC protocol to provide the administrative framework, milestones and regulatory deliverables necessary to secure approval of the repair and restart of the reactor. Comprehensive submissions prepared for the CNSC also helped build regulatory confidence in the quality of AECL’s work.
And, during this critical period, the company was vigilant in its communications with its employees, stakeholders and the general public, creating a dedicated website to provide detailed video descriptions of the work being undertaken and issuing weekly status updates to help improve awareness of the project.
A search for the leak site by remotely-operated video cameras located it at a point on the lower outside wall of the NRU vessel. It was determined to be caused by corrosion due to the radiolytic formation of nitric acid in the annulus – a difficult-to-access area about 15 cm wide between the heavy-water filled reactor vessel and the light-water filled neutron reflector.
With the site identified, AECL began a progressive series of inspections to determine the condition of the vessel and the path forward to repair the leak.
From visual examinations of the outer surface of the vessel wall, it was evident that corrosion was not localized to the leak site. This discovery led to a decision to defuel and drain the vessel, thereby ensuring the safety of the fuel, allowing better access to the inner surface of the vessel wall for inspection and repair and to stop further heavy-water leakage.
While the visual results indicated areas with more advanced corrosion, non-destructive examination (NDE) techniques, applied to the inner surface of the vessel, were required to determine the actual depth of corrosion. One of the largest single NDE inspection campaigns ever carried out in the nuclear industry over a period of about six months was conducted. This involved using eddy-current and ultrasonic inspection techniques with over two million data points to determine the vessel’s condition and the areas requiring repair. In addition, material was removed from the vessel wall to investigate the condition of irradiated material and to determine the corrosion mechanism.
Building a ship in a bottle
Examination of the vessel was challenging as the work had to be performed from the reactor deck, about nine metres above the corrosion sites located at the base of the NRU vessel, and through small access ports, typically nine and 12 centimetres in diameter.
Workers were familiar with ultrasonic and eddy-current testing, but they had to develop new specialized tools to allow deployment of cameras and custom non-destructive evaluation heads that could fit through the ports, reach to the bottom of the vessel, and deploy out to reach the inside of the wall.
Technicians and engineers, with assistance from vendors, designed and fabricated more than 50 customized tools and heads, all hardened to withstand high radiation fields in or outside the reactor vessel and to function safely and reliably.
The next challenge was testing the tools to confirm reliability, and training staff on how to operate the tools remotely.
“The tools used an umbrella-type design, so it was like building a ship in a bottle,” Drumhiller said. “The collapsed tools were inserted in the small opening and expanded to reach the vessel wall. All of the work was done blindly, using only remote cameras and lights. The operator would watch a flat screen to manoeuvre and control the tools.”
Eddy-current testing proved to be particularly valuable in providing detailed analyses of specific vessel areas. Ultimately, high special-resolution eddy-current testing was used on all inspection points.
Representatives from a variety of organizations with relevant expertise provided input on possible repair strategies. Two methods were deemed worth pursuing – the mechanical attachment of a patch to the vessel wall, and the use of welding, either as a build up or as a method of attaching plates.
After preliminary development of both technologies, welding was identified as the primary mode of repair, with mechanical repair continuing to be developed as a back-up. Welding could be used to build up thinned areas and restore structural integrity – directly or over backing strips for thinned areas – or to attach structural plates to the vessel wall.
It was also determined that, with appropriately qualified procedures and staff, the weld-repaired vessel could be shown to meet applicable codes and standards to achieve a permanent repair.
Precision welds
The vessel leak repair project team, under the leadership of David Cox, appointed by AECL as director of the repair project, faced enormous technical and logistical issues that they ultimately overcame.
The critical-path scope of the outage focused on 10 sites around the base of the vessel requiring non-destructive evaluations and extensive welding to build up the portions of the wall weakened by corrosion.
The leak in the vessel and regions of the wall that had thinned were repaired through a number of unique and complex welding processes. Specialized techniques and tooling were required to conduct and verify a qualified repair once again from the reactor deck, through the very narrow access points and in the highly radioactive environment.
“When you think about it, it was like trying to change the oil in your car from your living room, but more difficult,” said Cox.
To conduct the weld repairs, two types of remote welding tools were produced: one with a welding head that operates in a vertical direction and another with a welding head that operates horizontally. These tools provided flexibility in addressing the specific geometry of weld sites and in-vessel restrictions. A tool capable of deploying aluminium plates and tack welding them to the vessel wall was also developed.
To gain a better understanding of the corrosion and determine the most effective repair method before welding started, the team cut out a through-wall 1.5-inch diameter coupon of the vessel near the leak site. The radioactive coupon was transferred to a shielded hot cell facility near the reactor, equipped with manipulators and specialized equipment that technicians used to analyze the vessel material. In addition to the corrosion coupon, shallow scoop samples from the vessel wall were also obtained. These scoops were used to analyze the metallurgy of the irradiated vessel wall and to assess the weldability of this material.
Welding proved to be the most demanding phase of the outage, taking seven months to complete. New aluminium is difficult to weld, and at the time, no organization in the world had ever worked on highly irradiated aluminium. To further complicate the task, work had to be performed remotely with welding tools guided by arc-resistant viewing systems that provided a close-up view to the welders, who operated the weld torch and wire feed system from on top of the reactor.
Prior to conducting each weld, the area underwent a thorough engineering assessment and testing and qualification in mock-ups – typically reproductions of a segment of the bottom of the NRU vessel, and in most cases modeled at full-height – built at Chalk River and suppliers’ locations to ensure successful deployment of tooling and execution of the repair.
The success of a weld process is dependent upon good surface preparation. Two cleaning tools were developed; one for bulk cleaning and a second for final cleaning of the weld area just prior to welding. Weld trials were performed to ensure the surface preparation was acceptable and that the weld on the vessel wall would behave as expected, including the use of the coupon and scoops of irradiated aluminium removed from the vessel wall.
While weld tools were being finalized and performance-tested, work was also underway on qualifying a weld procedure and weld operators. The procedure was developed to meet applicable codes and standards, and qualification of the procedure was witnessed by CNSC staff and Canada’s Technical Standards and Safety Authority (TSSA). All weld operators were specially trained on the weld procedure and qualified by the TSSA.
“We significantly challenged our welders,” Cox said. “They had to perform complex and precise welds in three dimensions without the ability to feel, hear or directly view the weld area, working with a two-dimensional vision system. Also, we initially contemplated applying a uniform weld build-up around the corroded segment of the vessel, but after detailed computer modeling of the welding stresses we instead implemented 10 unique weld build-up repairs as the optimum technique to rebuild vessel-wall thickness.”
With successful qualification and demonstration of the weld-repair process, a repair plan was submitted to the CNSC describing the high-level requirements and approvals needed to execute the 10 repairs. The initial plan was based on a combination of weld build-up and weld build-up over backing plates in locations where the vessel wall was too thin to weld directly. In one case, a backing strip was used to cover the hole where the coupon was removed from the vessel wall. Subsequently, the use of structural plates attached to the vessel wall, but not welded over, was added to the repair plan.
The work on individual weld repairs started in December 2009. To keep welding work on track, Chalk River Laboratories assigned third-party reviewers to provide expert oversight and scrutiny of welding plans, decisions and activities.
Each repair site was unique and required development of specific weld techniques and parameters. Once the individual components of a repair had been developed, the entire sequence was verified with integration tests using all required tooling in a full-scale mock-up. Welding trials were conducted with replicas machined to match the profile of the vessel wall to ensure the weld process was appropriate for the local thickness-variations resulting from the pattern of corrosion.
Complex repair sites required many replicas. For example, more than 100 replicas were used in the weld trials for the largest repair site. The welds performed on replicas of each repair site were subjected to non-destructive and destructive examinations to confirm that acceptable welds were being achieved. When acceptable welds were reliably reproduced, the welders were deemed to be qualified to in-vessel weld operations at a specific site.
The primary complicating factor in carrying out the weld repairs was the potential impact on the lower vessel seal. The NRU vessel is a cylinder bolted to upper and lower headers, with a mechanical seal between the vessel and headers. The welded repairs were conducted close to the seal with the lower header. For larger repair areas, contraction of the weld as it cooled created tensile stresses that could counter the force being exerted by the seal bolts. A series of finite-element analyses was completed prior to each repair to ensure that the lower seal would not be compromised by the weld. To avoid large areas of weld build-up, structural plates were welded to the vessel, requiring only fillet or groove welds on their perimeter. Laser profilometry before and after welding was used to confirm that actual strain was consistent with the analysis, and the integrity of the lower vessel seal was monitored continuously during welding.
Convergence on a repair process for a given site required iteration between the engineering team to ensure the repair met fitness for service requirements, the finite-element analysts to ensure the forces on the lower seal were acceptable, and the welding teams to ensure the proposed repair sequence could be reliably executed.
A series of inspections were required to demonstrate that the repairs were acceptable. Following each repair, a remote visual inspection was performed. Eddy-current testing was conducted in a post-weld surface examination of the weld crown for the detection of surface cracks. The thickness of the weld build-up was measured using ultrasonic inspection. The ultrasonic weld thickness examination was also used to detect any lack of fusion between the weld and the vessel wall. Lastly, a phased array ultrasonic angle beam examination was used to scan for both near-wall and far-wall cracking in the heat-affected zone of the parent material.
An in-service leak test was successfully performed after completion of the inspection of the final repair and the introduction of heavy water into the vessel. All repairs were successful in meeting acceptance criteria.
Early in the forced outage, AECL recognized there was an opportunity to conduct additional work not directly related to the leak repair that could not be completed within the normal operating cycle. This extensive maintenance and improvement programme, conducted concurrently with the repair, would help improve the safety and reliability of the NRU for the long term. Improvements were made to various systems, with a focus on those requiring the reactor to be in a defuelled or drained state. One particular improvement was the replacement of the heavy-water moderator with low-tritium heavy water. This has resulted in nearly a tenfold reduction in tritium concentrations in the moderator, which will improve worker safety and reduce releases to the environment.
Other areas of improvement included maintenance activities on the main heavy water system, the process water system, the on-power fuel handling system, control rod systems and safety systems.
Returning the NRU to service
The resumption of NRU operations required a series of measures to be undertaken to ensure the safety and integrity of the NRU.
A comprehensive readiness-for-service assessment was conducted to verify that all systems were fully operational. This included a series of detailed test procedures and comprehensive check lists, factoring in relevant operating experience, and complete activities, such as loading fuel or refilling the heavy-water system. Thorough readiness reviews were conducted before proceeding to key milestones, including refill of the NRU vessel, loading of fuel, the controlled sequence of sub-critical instrumentation tests and start-up to low neutron level.
A systematic review of the training requirements for operational personnel was also conducted prior to start-up, including routine tasks associated with normal operations that had not been performed due to the extended shutdown. This included refresher training on such areas as control and safety systems and electrical systems. In addition, several unique one-time activities and tasks required to return the NRU reactor to operation were identified, and just-in-time and infrequently-performed training was conducted.
As a result, more than 600 training activities were identified and delivered in preparation for and during the return to service.
Upon completing satisfactory inspection results for the last repair and the vessel leak-test, the NRU vessel was qualified for operation according to the fitness for service assessment. Successful execution of the return-to-service plan confirmed that all systems identified in the NRU safety analysis report for safe operation were functioning and met its requirements. The CNSC granted approval to complete refueling of the NRU reactor on 15 July 2010. On 16 August, the NRU was successfully returned to high-power operation and the first medical isotopes were shipped two days later.
Today, AECL is continuing to work with the CNSC to renew, for another five years, the Chalk River Laboratories site licence, which expires at the end of October 2011.
As part of the isotope supply reliability programme, AECL has been renewing facilities, equipment and staff capabilities that are required for reliable, long-term isotope production. An integrated safety review of the NRU, a key component of that programme, and the repairs and enhancements made to the NRU during the repair and return-to-service project, will support the renewal of the Chalk River Laboratories operating licence.
Root causes
No successful repair is of long-term value without understanding the root cause of the issue and ensuring it does not recur.
Early inspections of the NRU reactor found that corrosion was occurring low in the reactor vessel. Annulus drain lines were becoming blocked with corrosion products. When drains became plugged, water laced with dilute nitric acid – a by-product of air in a moist high-radiation field during reactor operations – pooled at the bottom of the vessel and annulus wall.
The finding led to a review of a 2005 condition assessment of the vessel. It had reported no problems, even though the new inspections showed corrosion was in progress at that time. Drumhiller said the report ultimately revealed the root cause of the oversight: the lack of a sufficient questioning attitude.
“We knew nitric acid was produced in the annulus,” Drumhiller said. “The belief was that if nitric acid was causing corrosion, the corrosion would be located high on the vessel wall. So the assessment only inspected there and identified no problems. But corrosion was taking place low in the reactor vessel where the nitric acid pooled. During the 2005 assessment, we did not find videos that showed the corrosion, and the organizational communications to help people understand the corrosion were not adequate.”
As a result, the Chalk River Laboratories came out of the extended outage in early August 2010 with a new corrosion-mitigation strategy in place: to keep annulus drain lines free-flowing to prevent water from pooling, and flood the area with carbon dioxide to drive out nitrogen and reduce the production of nitric acid.
To ensure the vessel continues to be fit for service, an in-service inspection programme has been put in place. Inspections will monitor the vessel for service-induced degradation, including vessel wall corrosion.
Chalk River also revised its outage schedule to optimize testing, maintenance and inspection frequencies. Before this extended outage, the plant, which is refueled online, was shut down every three weeks for a five-day outage.
The NRU now shuts down every four weeks for a five-day outage. In the spring of 2011, the unit will stay down for 28 days to perform more complex maintenance, detailed testing and in-service inspections. These key activities will verify the efficacy of corrosion-mitigation strategies, which can then be adjusted based on results. Additionally, AECL is planning annual four-week maintenance shutdowns to ensure continued and ongoing fitness for service.
Safety culture improvements have also been incorporated into AECL’s Voyageur programme, which drives broad organizational improvements. The next phase of this programme has been developed to include corrective actions that will address equipment reliability, reinforce desired behaviours among staff, ensure broad organizational involvement and improve interpersonal communication. A multi-year action plan has been put in place and improvements will be closely monitored and measured.
Going forward, further work will be performed to reduce annulus air and reflector water leakage, the conditions in the annulus will be monitored, the vessel will be inspected on a regular basis, and technology for sealing small vessel leaks is being developed. The technology would be used in the unlikely event that a leak occurs due to a highly-localized corrosion mechanism.
Additional mitigation measures are also being considered for possible future implementation, including the application of an aluminium cold spray to the external vessel or reflector wall to serve as a sacrificial coating, and changes to the reflector water chemistry to reduce the corrosive potential of leaking water.
A successful conclusion
Drumhiller said that while the repair outage took longer to complete than initially estimated, “the extra time and effort have helped to avert protracted shutdowns and material-condition problems in the future and ensure the NRU reactor is safe for continued long-term operation.”
“At no time before or during the extended shutdown was there any immediate threat to nuclear safety, workers, the public or the environment, and we want to keep it that way for some time to come,” he said.
In the end, AECL managed to do what no other organization has done before, he added.
“We have essentially revolutionized corrosion detection and mitigation in reactors. Very few nuclear facilities have the capability to design, manufacture and qualify ultrasonic and eddy-current testing probes on their own, and we have that capability at the Chalk River Laboratories, all in-house. No other facility in the world could have done what we did.”
The 15-month odyssey also tested the innovation, resourcefulness, persistence and perseverance of some 300 AECL staff – and according to Drumhiller, they passed with flying colours.
“We could not have achieved this feat without the commitment of our employees,” he said. “Despite challenges and schedule extensions, our staff stayed focussed and ensured we were doing the right things to make the NRU safer and more reliable in the long run. Likewise, our suppliers and external partners brought significant knowledge and expertise to the project and made it possible to meet this challenge head on.”
“Their combined contributions helped restore the NRU to isotope production and research advancement, which will continue to benefit Canadians and millions of people around the world for some time to come.”
Related ArticlesThe NRU and the leak site
| NRU return-to-service in numbers |
|
-4,000 hours of in-vessel NDE |
