Remote control30 June 2000
The first phase of decommisioning of the Windscale piles, shut down by fire in 1957, has been successfully completed. Innovative procedures included using remotely-operated vehicles adapted fromthe North Sea oil fields
Phase 1 decommissioning of the Windscale piles was successfully completed in July 1999. The work lasted almost nine years and involved over 2000 hours of remotely operated vehicle and manipulator activity. It was managed by a small team from the UK Atomic Energy Authority (UKAEA), supported by a managing agency group from WS Atkins. Specialist contractors who had experience of working in hazardous environments within UKAEA safety management systems were employed to carry out various decommissioning tasks. Wherever possible, existing equipment and technologies were used, modified and adapted to suit the unique environment and requirements of the piles.
The two Windscale production piles were built in the years immediately following World War 2, largely to produce plutonium for weapons. The programme, set by the Ministry of Supply, included not only construction of the two air-cooled graphite moderated reactors, but also fuel handling, storage and reprocessing facilities on a scale not previously attempted outside the US.
The two piles are dimensionally similar but are placed to allow the use of a common fuel pond (B29), which is connected to the core discharge voids of each reactor by water ducts.
The piles operated from 1950 until 1957, when a fire caused severe damage to the core of Pile 1 and raised questions on the safety of their design. Pile 2 was shut down as soon as the fire in Pile 1 was discovered, although the decision to make this a permanent shutdown was not taken until some time later.
The fire in Windscale Pile 1 started during the ninth routine Wigner Energy release, an anneal of the reactor core carried out to release trapped energy in the core graphite caused by neutron bombardment at low temperature. Approximately 20% of the core was affected. The fire had become well established before the alarm was raised, and attempts were made to discharge fuel from around the heat-affected zone to form a fire break before deluging the core with water. It is thought that some 6700 fire damaged fuel elements and 1700 isotope cartridges still remain in the core.
Post-operational clean out
Between 1958 and 1961 the area around Pile 1 was decontaminated. The fuel was discharged from Pile 2 and from undamaged channels in Pile 1. Control and shutoff rods were fully inserted and the operating gear was dismantled and removed. An 80mm thick concrete screed was laid over the pile cap covering all of the mechanical equipment penetrations through the biological and thermal shields to prevent any further damage to the core.
The air blowers and filters were removed from the blower houses (Windscale buildings B3, B4, B13 and B14), and the air ducts connecting the blower houses to the pile cores were bricked up and sealed.
The work carried out at this time would now be referred to as post-operation clean out (POCO). An internationally acknowledged phased procedure for decommissioning redundant nuclear plant was not developed until the late 1980s.
The in-between years
Apart from routine surveillance and monitoring, no further work was carried out on either of the piles between 1961 and 1990. During this period the UKAEA was actively involved in the research and development programme for advanced gas cooled reactors (AGRs) and it used the pile buildings to house reactor test rigs and shielded facilities for post irradiation experiments (PIEs).
Dismantling the civil AGR and Windscale AGR rigs in B2 became part of the subsequent phase 1 decommissioning programme for Pile 1. The HERO reactor containment in the west blower house of Pile 1 (Windscale building B3) was demolished at the same time as B3 to provide space for an intermediate term ILW store for phase 2 decommissioning of the Pile 1 core.
Until the creation of BNFL in 1971, UKAEA had managed Windscale as two divisions: the Reactor Group and the Production Group. The Production Group, which included the Windscale AGR, Calder Hall, the fuel storage ponds and reprocessing plant, became BNFL Sellafield. UKAEA retained control of the piles complex and the plutonium laboratories in B33. B33 has subsequently been transferred to BNFL, where it has evolved into the Mixed Oxide Fuel Development Facility (MDF).
Areas of responsibility were defined for future decommissioning of obsolete and redundant nuclear facilities on the site. An anomaly in the demarcation of responsibilities is the pile chimneys (B16 and B6). BNFL has for years used the Pile 1 chimney lift shaft as a support structure for ventilation ductwork servicing the Magnox reprocessing area. It has retained responsibility for decommissioning of the chimneys even though these are part of the piles’ structures.
Phase 1 decommissioning
Following licensing of the UKAEA sites in 1990, a programme of improvements was agreed with the Nuclear Installations Inspectorate (NII) to reduce the potential hazards associated with the Windscale Piles. This programme covered:
•Sealing the core containment with seismically qualified barriers.
•Installing a dedicated and filtered forced extraction ventilation system.
•Providing improved core condition monitoring systems.
•Removing fuel and debris from the charge and discharge voids.
•Draining the water ducts.
•Refurbishing the charge hoists.
•Long term option studies for the future management of the piles.
The programmes were to become known collectively as phase1 decommissioning.
It was clear from the outset that the knowledge available, on the core condition and the general state of the areas adjoining the core within the shielded structures, was inadequate. Extensive survey and investigative work would be required to support the limited objectives set for the phase 1 programme. This would prove useful in providing a foundation for a phase 2 programme which would include dismantling the Pile 1 core.
The differences in the condition of the two piles dictated decommissioning strategy. The core of Pile 1 was extensively damaged by the fire. It was estimated that up to 15t of fissile material could remain in the core and that the core structure itself could be unstable. The emergency discharge of fuel from around the fire-affected zone had been carried out under extreme emergency conditions. Unknown quantities of fuel and debris had been deposited in both the water and air ducts. Some of the fuel would almost certainly have suffered impact or heat damage to the cladding. Undamaged peripheral areas of the core graphite would contain Wigner Energy, as the anneal that was in progress when the fire started had not worked. Lighter particles and combustion products would be present in the charge and discharge air ducts.
By comparison Pile 2 was in relatively good condition. It had been shut down under controlled conditions and the core was undamaged. After the fire the complete fuel load had been discharged, although some fuel elements had become trapped on the burst slug scanning gear on the discharge face of the pile. Both piles therefore would be subject to the Phase 1 decommissioning programme but on completion of this Pile 1, which represented the greater hazard, would be subject to a phase 2 programme in which the core would be dismantled. Pile 2 would be placed onto an extended period of care and maintenance subject to achieving stringent safety criteria.
At the start of the phase 1 programme AEA Technology, formed as a division of UKAEA, was in the process of grouping core and non-core businesses into divisions in preparation for privatisation. UKAEA’s Government division was formed in 1992 to manage all of the nuclear assets and liabilities on UKAEA sites. UKAEA’s remit for piles phase 1 operations was to provide executive control of all aspects of the work to be carried out. WS Atkins Northern was appointed as managing agent for phase 1 operations as a result of a competitive tender in 1993. Its role was to take responsibility for the implementation of the programme.
Separating the core and containment areas from the general environment was considered a high programme priority. The water ducts had for some time been a cause of concern, being particularly vulnerable to impact or siesmic damage. The seals at the top of both exhaust stacks (pile chimneys) had deteriorated and the concrete overhang of the filter gallery was in poor condition.
The air ducts
Air dams were installed in the discharge air ducts of Pile 1 in 1990 to separate the chimney from the pile containment. As with all operations inside the biological shield walls of the core containment, high background radiation levels prevented any significant human access to the work site. The dams comprised steel box section fabrications inserted into the discharge air ducts through holes cut in the floor of the test house between the pile cap area and the pile chimney. The high radiation levels from the core and water duct prevented man access to the discharge air ducts for preparation of the mating concrete surfaces. With the box sections in place there was an irregular gap around 200mm wide between the concrete walls and the end of the dam sections. While the design intent of the dams had not been to provide a hermetic seal, the gap after installation was larger than anticipated and prevented the dedicated ventilation system from maintaining adequate depression within the containment. Resilient seals consisting of rubber tubes mounted on aluminium angle sections were installed to close the vertical gaps on each side of the dams. The rubber tubes were collapsed by vacuum during installation. When the vacuum was broken the tubes reverted to their normal shape, closing the gaps effectively. The base seal was created by injecting expanding foam sealant.
The inlet air ducts of both piles were sealed after the fire in Pile 1. In operation, the air stream from the blowers was directed into the core of each pile via the ducts. Two sets of cascade vanes were installed in each duct adjacent to the edge of the respective cores. The purpose of the vanes was to divert the flow of air across the whole face of the core to promote uniform air flow through the fuel and isotope channels. The vanes effectively divide the air ducts into three sections: east, west and the charge void directly in front of the core. The vanes converge at the bottom of the duct, forming a V which encloses the charge void.
Before the ducts could be cleared of debris, door apertures had to be cut through the walls. Rumic deployed two tracked ROVs, called Cyclops, to retrieve fuel and debris. The Cyclops vehicles were electrically powered and each had a small five-function electric manipulator mounted on the front deck of the chassis to recover fuel elements and place them in shielded containers. A vacuum cleaner collected small pieces of debris and dust. The V configuration of the vanes allowed the ROVs access to the entire floor area of the ducts.
The water ducts
Of all the phase 1 tasks, the most difficult and protracted was draining and sealing the water ducts.
The ducts are concrete tunnels 75m long which connect the discharge voids of both piles to the central storage pond (B29). The walls and roof of the ducts are 1.4m thick concrete and include a waterproof membrane. The ducts were subject to the full hydrostatic head of the pond water, nominally 2.8m higher than the top of the duct roof. The steel sluice gates, originally designed to isolate the pond from the ducts, were beyond any possible repair. Any major breach of the water ducts would lead to the release of large quantities of contaminated water and a potential reduction in the storage pond water level.
Concrete barriers were cast and installed at the ends of both ducts in 1990. The barriers did not seal the entrance to the ducts completely: initially the access shaft containing the sluice gates was left open at the top to facilitate sludge pumping and the removal of the sluice gates at a later date. The barriers provide effective isolation of the ducts from the ponds. Bund walls were constructed along the sides of both ducts to contain any minor leakage that could occur from any deterioration of the duct membrane. With the barriers and bunds in place, cleaning and draining of the duct could be considered. The work required was split into seven packages:
•A survey of the duct using underwater vehicles.
*Design, manufacture and installation in the water duct of a seismically qualified and watertight barrier at the east side of the Pile 1 bioshield and the west side of the Pile 2 bioshield.
*Pumping of sludge from both ducts to B29 pond.
*Removal of fuel and debris from the water ducts to other BNFL and UKAEA facilities for conditioning prior to reprocessing or disposal.
*Decontamination of the duct internal surfaces.
*Design, manufacture and installation of airtight covers over each of the duct access shafts (two in B2, two in B12 and two on B29 pond).
*Phased draining of both ducts.
Information available in 1990 on conditions inside the water ducts was limited. Over a two year period AEA Technology did some preliminary survey work using a small remotely-operated submersible. It became apparent at an early stage that conditions in the duct were very bad and that a much more detailed survey would be needed before a programme of work could be commissioned to clear the duct.
Specialist contractors were selected on a competitive basis with weighting for previous experience in offshore and remote operations. Rumic, a local company with extensive offshore experience, was contracted to carry out a detailed duct survey. Rumic’s choice of vehicle to carry out the task was a Hydrovision Hyball submersible, modified to include: sonar for position identification; profilers to establish sludge levels on the duct floor; and a gamma probe to record the levels of radiation and locate fuel elements.
That Rumic’s choice of vehicle was appropriate is testified by the fact that the original vehicle continued to operate usefully up until the end of the duct programme in June 1999, when it was transferred to BNFL for pond operations. All submersible excursions were recorded on video and these were made available to organisations tendering for duct work packages.
Steel seismically-qualified barriers were installed in both water ducts, dividing each into two sections. For easy reference and location identification, each end of the ducts is referenced by the building number through which it extends. In the case of Pile 1 the duct section behind the core became the B2 water duct and the duct section between the barrier and the pond became the B4 water duct.
The Pile 1 barrier was supplied and installed by Demtech and the Pile 2 barrier by Nukem Nuclear. Each barrier has two 4in ball valves installed to facilitate sludge pumping from behind the core (B2 duct) to the outer duct section (B4 duct). In a similar manner to the air dams the barriers had to fit between the two walls and be sealed to the floor and roof lintel; this time a completely watertight seal was required and installation took place under water.
Space limitations in the access shaft effectively ruled out the use of a remotely operated vehicle to make the fixings holes and end connections between the barriers and concrete structure. A tool post arrangement comprising a frame and cross slide carrying a vertical slide was built. The frame was bolted to the floor around the access shaft opening and quick release tooling deployed on the vertical slideway to prepare and drill the holes for the barrier fixings.
Installing the barriers was not entirely without problems: the as-built drawings of the duct cross section were inaccurate and additional grout formulation had to be forced into the joint space to seal it. The barriers were subsequently tested during duct draining and proved to be capable of withstanding the pressure exerted by the full static head of water contained in the water duct behind the core.
In the tenders for sludge removal and fuel/debris collection contracts UKAEA offered the use of a remotely operated tracked vehicle (Norman) which had been developed for underwater operations. In the event, the successful bidders for both packages of work elected to design and build somewhat simpler vehicles, which ran on the rails originally used for the fuel skip trains.
Rumic was contracted to supply the sludge removal vehicle (S.ROV) while Comex Nuclear Services supplied the fuel/debris collecting vehicle (D.ROV). Both vehicles made extensive use of equipment developed for offshore work and, after some teething problems, carried out their tasks successfully.
Both vehicles were converted for dry use after draining the ducts and were used to pressure wash the duct walls to remove loose contamination. In total 8m3 of sludge with a particle size of less than 5mm was transferred from the ducts to B29 pond. Some 210 fuel elements and 261 isotope cartridges were recovered before the ducts were clear.
Testing showed that approximately 10mm of concrete would have to be removed from the internal surfaces of the ducts to completely remove all traces of contamination. A vehicle was designed to carry out this task but failed in operation and became stuck 10m from the access shaft launch site. The subsequent recovery operation was difficult and involved placing, by ROV, of an air bag and hydraulic ram to right and return the vehicle to the rails. Placing this equipment was hindered by poor visibility and the presence of the ROV umbilical and debris collection hose. Considerable operator skill was required to free and retrieve the vehicle. No further attempts were made to decontaminate using this method.
The two original sluice gates were removed from the duct ends in 1999, immediately before draindown. The gates were closed steel box section fabrications that were seized in their vertical operating guides. Each gate comprised two sections and had an overall weight of 7t.
Structural considerations ruled out direct force as a means of recovery. As part of its contract to drain the Pile 1 duct, Nukem Nuclear had to remove the gate. After option studies which considered various tooling packages and delivery systems, a waterjet cutting system was selected as the most suitable for deployment in the confined space of the pond end access shaft. An additional advantage was that jetting operations, when carried out underwater, would have little impact on the working area above and most of the equipment could be sited outside of the controlled area, reducing the potential for contamination.
The small access area around the gate apertures was tented over to contain any contamination released during cutting and lifting. Cuts were made vertically down both sides of each gate to free them from the guides and horizontally to separate the upper and lower sections. On completion of the cutting operations the gate sections were raised above water level and washed down to remove surface contamination. Holes were cut in the closed box sections to allow trapped water to drain out.
Nuclear Decommissioning carried out similar work on the Pile 2 gates and adopted the Nukem system.
Before final disposal as low level waste the gates were reduced to obtain optimum packing density in a Drigg container. The potential for radioactive sludge to remain inside the gate sections effectively ruled out the use of any hot cutting techniques: waterjet cutting in a normal environment was thought to be high-risk and would generate secondary waste from the process. An industrial bandsaw modified to run on a machine slide was used to reduce the gates to 35cm wide sections.
The work was carried out in an existing facility in B12 by Nuclear Decommissioning. Each pass of the saw took on average about three hours to cut a 3m-long strip. While slow, the process generated little heat and consequentially no airborne contamination.
During the last nine years, many other difficult remote operations tasks that stretched the ingenuity of all concerned have been completed. Fuel elements were recovered from the BSSGs in both piles using manipulators on 30m-long masts; the skip train pulleys were recovered from the water ducts of both piles, all of the 3444 fuel channels in Pile 2 were surveyed and graphite samples were taken from both cores to establish the rate of release of Wigner energy.
While work within the containments was technically challenging, a significant number of other engineering tasks which did not involve remote operations were carried out to fulfill the agreement made with NII. Both charge hoists were modified and returned to service. The CAGR rig and Windscale building B3 were demolished. B2 was reclad and new dedicated ventilation systems installed for both piles. Finally, engineering option studies were carried out to identify potential methods to be used for phase 2 dismantling of the Pile 1 core.