Once the Point Lepreau Candu unit was back on line, it began to claw its way back to the top of Nuclear Engineering International’s lifetime performance tables, where it had seemed a fixture for so long. Plant availability has risen from zero, to 17.7% for the first three months of 1997, to 80% in the second quarter and finally to 100% over the June – September period.

The series of problems at Point Lepreau arose at the end of an extended outage which included the checking of the positions of the garter springs that support the fuel channels inside the reactor, and which have been the subject of extensive checks and remedial measures on Candus since the issue first emerged over 15 years ago.

The first of the recent problems, which concerned steam generators, was discovered in September 1996. The control computers and the shutdown systems, which are each fed with boiler level readings, were found to be showing an anomaly when the boiler level was controlled to just below the steam-separator support deck at reactor powers less than 14% on steam generator 3. As a result it was decided to examine the inside of the steam generator.

The inspection revealed that the emergency water supply (EWS) header (shown in figure) was damaged by flow-accelerated corrosion inside. The header crosses the steam generator above the tube bundle and is subject to high internal fluid flow rates. In addition, the distribution header thermal expansion joint pipe connector was distorted and had been displaced, but this was found to be an incidental problem to that causing the level anomalies.

The other three steam generators were examined and the internal corrosion problem of the EWS header was found in all cases. The four steam generators were repaired with the work completed by late October. The incident was rated level 1 on the INES scale, but this was increased to level 2 when it had been found that, had they not been discovered, potentially more serious defects could have threatened the boiler internals.

HEAVY WATER LEAKAGE

The second incident began when radiation levels in the boiler room began to rise over several weeks, at the close of 1996 and beginning of 1997, indicating a leak of heavy water from the primary heat transport system. The initial leak rate in the first week of January was, at about 5 to 7 kg/h, too small to be able to easily pinpoint the source of the leak, but by 16 January it was clear that it had reached a level that would allow rapid detection and the station was shut down at 21:30 on that day.

Leak tracing procedures, set in motion after the unit had cooled sufficiently to allow personnel entry, located the problem in the inside bend of an outlet feeder pipe of the east feeder to one of the channels (S-O8). The crack was about 35 mm long on the outside and 55 mm on the inside. By the end of the month the failed feeder had been removed and was under examination at Atomic Energy of Canada Ltd’s (AECL’s) Chalk River nuclear laboratories. Initial measurements showed that the faulty pipe had not thinned beyond expectations. Further analysis showed that the crack had been initiated by stress corrosion cracking dating back for more than a year. This phenomenon is most unusual in carbon steel components unless several parameters are exceeded, including stress and temperature, and the presence of oxygen and other chemical conditions.

While this work was in hand at Chalk River, welding procedures to repair the damaged feeder were prepared. The repair was completed in February 1997 and extensive examination and review work undertaken to provide the required assurance levels for start up. To determine the implications of the wall-thinning, an expanded inspection of 65 outlet and three inlet pipes using specially prepared templates together with the latest high accuracy ultrasonic techniques was used to amass over 10 000 wall thickness measurements. The accuracy and repeatability of these tests were within ±0.003 inch. The results proved that all feeder pipes satisfied the fitness-for-service criteria for at least the next five years.

INSPECTION PROGRAMME

A continuing inspection programme will include on-line monitoring of thinning rates to validate the wall-thinning rate model and to assess planned remedial actions. It will include baseline thickness measurements of about 170 additional feeders over the next three years to improve end-of-life predictions. Other measurements are aimed at investigating the root causes of degradation and making mandatory thickness measurements to meet commitments to the Atomic Energy Control Board, Canada’s nuclear regulator.

As far as the stress corrosion cracking mechanism was concerned, a programme of inspection and assessment of other feeders, together with all components that could be a contributory factor in the cracking, was carried out. A total of 48 inlet and 109 outlet feeders were examined using ultrasonic techniques based on a guided wave principle. The pipes chosen were those that were most highly stressed; those with the most significant wall-thinning; those that had been drained since start-up; or those with supports found to be abnormal or with unusual layouts of any kind. No reportable conditions were found in any of these.

Supporting components and hangers connected to the feeders were also inspected and adjusted if necessary. In addition, the locking mechanisms of all the fuel channels were checked together with the chemistry monitoring equipment on the primary heat transport system in both normal and “off-normal” conditions.

The future inspection programme will include specialised ultrasonic inspection techniques to detect flaws in feeder pipes due to stress corrosion cracking. Several feeders were inspected in 1997, and others from the previously inspected sample will be examined again in 1998. From these operations a continuous inspection programme will be devised. A similar procedure will be adopted as far as hangers and supports are concerned.

Assistance in finding the solution to the problem was also provided by the Research and Productivity Council of New Brunswick, Thielsch Engineering (specialists in pipe analysis and repair), Babcock & Wilcox and Ontario Hydro.

RESOLVING THE PROBLEM

After a thorough examination of the operating records for the station, the crack was found to have resulted from a fuel channel being left unlocked during operations following the 1995 outage (for the relocation of fuel channel spacers that was carried out on all Candu units over a certain age).

“We measure the position of each fuel channel each time we visit it with the fuelling machine, which is about once in six months,” explained Stuart Broom, Point Lepreau’s Technical Manager. “The fact that the channel had moved could be seen in the data from the fuelling machine, although it was not alarmed. So when the leak occurred we had cause to refer to the data and saw quite clearly that this fuel channel was floating” (ie not locked at either end and moving axially in the reactor).

Candu fuel channels have locking mechanisms at both ends of each channel which are used to control the elongation of the channel due to the effects of radiation-induced creep. This is taken into account in the design life of the reactor by allowing the channels to extend first on one side of the reactor and then the other by locking the opposite end in position. Design allowances for growth in the Point Lepreau type of reactor will accommodate 150 mm of axial growth in 30 years of operation.

At about 12 to 15 years of operation, with about 75 mm of extension, the unlocked end has come close to the bearing allowance, so this end is then locked and the opposite end allowed to extend for the remainder of the design allowance. This mid-life maintenance process is called reconfiguration and was carried out in the 1995 outage. Up to 1995, all the west ends of the channels had been locked, and the channel extension had been eastwards. During the reconfiguration, it was found, S-O8 had not been properly locked at the east end, and so was left floating.

The history of fuelling machine visits to the 380 fuel channels included data covering all visits since the Spring of 1996. During on-load refuelling of a Candu reactor, when the fuelling machine locates on to a locked end of a channel (the east end of the channels since the reconfiguration), it records the position automatically. This is compared with the previous visit to make sure it is exactly the same. When it locks onto the west end of a channel, now free to extend, the position measurement monitors the extension, and allows a check that each channel is extending at the expected rate.

When the fuelling machine visits a channel, it puts an axial thrust on it. On the unlocked channel, this in turn puts a bending moment on the feeder pipe that enters the end of the channel at right angles to its length. The strain and the rate of strain were both important in deciding whether the pipe could stand this treatment, and in this case the relatively rapid movement of the fuelling machine produced a strain rate sufficient to rupture the oxide of the carbon steel and render it prone to continuing the cracking mechanism through stress corrosion cracking.

A ‘WAKE UP CALL’

So the fault poses no generic problems, but, according to Stuart Broom, is a “wake up call” to show that the physical supports of carbon steel piping need to be functioning in accordance with their design. A broken hanger or support represents a potential threat by allowing the pipe to be in a position where it could be strained or be subject to a high strain rate. AECL has notified all its customers about the fault at Point Lepreau, explaining what occurred and recommending details for an inspection programme for the primary heat transport system pipework hangers and supports.

The costs of the repair, inspection and related work were estimated at Can$7 million, with replacement generation costs amounting to Can$400 000 per day. In view of the maintenance and inspection work that was carried out during the first quarter of this year, the annual planned outage was postponed from the Spring to September.