Peter E. Vickery, Ian E. Richardson and David Forsythe
BNFL

The paper concerns the recovery of radioactive sludge from a fuel storage pond which had been accumulated during the operational life of the pond. The presence of the sludge provided an operational challenge in managing the dose of those involved in the clean out and decommissioning of the pond and its recovery to modern storage facilities.

The nature of the sludge, the difficulty of gaining access to it and the age of the pond structure presented a major challenge to the pond operator and to those charged with designing and setting to work the recovery process. The paper tells how the sludge was characterised, methods for its recovery were identified and tested and how the eventual solution was justified. The installation, commissioning and the operational experience to date are then described.

The paper summarises the lessons learnt from the experience and how these are being applied to other clean-up tasks at Sellafield. The principal messages are that designs have to be fit for the purpose identified by the operational staff, and that continuing close liaison between designers and operators throughout such projects is essential to their success

The fuel storage pond (subsequently referred to as Pond 2) was commissioned in 1959/60 for the receipt, storage and de-canning of irradiated Magnox fuel. It includes a pond with a fuel storage capacity of 1,500 tonnes of Magnox fuel held in open steel storage skips. The layout and dimensions of the pond are shown at figure 1. It comprises 3 reinforced concrete storage ponds, an extension to the third pond, seven underwater de-canning bays and de-canning caves. Five of these bays, designated A-E, were originally used for underwater de-canning. Underwater de-canning activities were discontinued a few years after the commencement of operations and transferred to the de-canning caves. Bays A-E were used throughout the subsequent years of operation of Pond 2 for the storage of miscellaneous waste arisings.

Since 1986, current arisings of Magnox fuel have been stored and de-canned at the Fuel Handling Plant (FHP) on the Sellafield site, and over the last 10 years, work has been ongoing at Pond 2 to recover a backlog of stored fuel for reprocessing and to clear the accumulation of Magnox sludge and miscellaneous waste so as to prepare the structure for decommissioning.

Magnox alloy fuel cladding degraded under the prevailing storage conditions in the pond. This produced insoluble corrosion products comprised of magnesium hydroxide, Magnox metal, uranium and other fuel derived constituents. Over its operational life, the pond generated significant quantities of Magnox fuel corrosion sludge. During the mid 1970's, the radiation levels generated by the sludge were managed by the installation of a rotary skip washing facility to remove sludge accumulations from storage skips prior to de-canning operations.

At that time, retrieval, storage, and disposal solutions were being developed. As an interim solution, the sludge was hydraulically transferred to D bay, one of the redundant underwater de-canning facilities. A schematic layout of the bay is shown at figure 2. The original de-canning equipment was not removed prior to the start of sludge transfer operations. To minimise radiation levels within the building which the accumulation of sludge would cause, D bay was covered with concrete slabs with 4 apertures left for future access.

Sludge transfers were accomplished by using the bay as a settling tank. The supernatant water was removed to a set depth and the sludge transferred as a fixed volume batch of slurry comprising around 5% by weight of solids. This slurry then settled so that the sludge accumulated on the bay floor and the settling/transfer cycle could be repeated.

This method of transfer resulted in the accumulation of sludge under the inlet pipeline at the back of the bay. In order to maximise the storage volume available and to prevent the sludge from being exposed to air, the sludge was propelled towards the front of the bay by submersible pumps inserted through the floor penetrations. By the time transfer operations ceased in 1984, the bay had been filled to a sludge depth of around three metres, with a settled volume of around 230 cubic metres.

From this date onwards, sludge was transferred via a purpose built settling tank to new sludge storage tanks at the new Sellafield Site Ion Exchange Plant (SIXEP).

The accumulation of sludge in D Bay resulted in rising levels of radiation in the vicinity of the bay with an adverse effect on the occupational radiation dose to those involved in other work in the building. It was therefore agreed with the Regulator in the late 1980's that early recovery of the sludge from D bay to the tanks at SIXEP was necessary to minimise dose to those working in and around Pond 2.

The main constituent of the sludge was known to be Magnesium Hydroxide which, when generated by the corrosion of Magnox, has a lamella crystalline structure and rheological properties similar to those of a Bingham plastic. This had been established as a result of many years of study into the behaviour of Magnox sludge.

Conditions surrounding the transfer of the material and its subsequent management at D bay were also felt to be significant. The principal issues were perceived to be as follows :

• The process of rotary skip washing had led to the incorporation of both fragments of uranium fuel and occasional small items of fuel element furniture in the sludge.
• The process of moving material across D bay would have classified the material so that the smaller particle fraction would be deposited at the front, with the larger, more rapidly settling fraction, at the back under the inlet pipe.
• The inclusion of a significant quantity of uncorroded Magnox metal in the sludge was also going to give rise to ongoing hydrogen evolution, which would increase when the sludge was disturbed.

An extensive sampling and characterisation programme was therefore carried out. This had the two objectives of underpinning the safety case in the specific areas of uranium carry-over and hydrogen evolution and supporting the design of the retrieval process.

Samples were taken from the bay from all four access holes at varying depths. In this way, a full picture of the variation in sludge properties across the bay was generated. The following conclusions were drawn :

• The sludge varied in consistency across the bay, though samples at similar heights were comparable.
• The sludge near the top of the bay was similar to wet sand and was non-cohesive.
• The sludge was mainly Magnesium Hydroxide with up to 30% of the original Magnox material present.
• Considerable quantities of uranium were present mainly as uncorroded metal.
• Activity levels varied between samples by up to 2 orders of magnitude

The following properties were recorded from the characterisation exercise :

• Chemical and radiochemical data
• Particle size analysis (shown in fig. 3)
• Rheological data
• Settling characteristics

A summary of the characterisation data obtained on activity levels is given in table I. The sampling and characterisation work proceeded in parallel with the design process and was an invaluable source of data. In particular it confirmed the assumptions on property variability and uranium and Magnox content mentioned above.

The foregoing sections will have highlighted many of the constraints confronting the project team. These were :

• Radiation levels were very high, such that man access to the surrounding area for tasks such as surveying and installation were limited. Working times were very short (approx. 30 minutes). Access to the D bay area required cumbersome personal protective equipment due to the significant contamination hazard present.
• D bay is located in a busy area of the building, and shares services, such as steam, water and electricity with other facilities. This meant that limited space was available for new equipment and that existing work programmes would be affected. The area is serviced by a crane with limited working coverage and capacity.
• The building is 40 years old and was constructed to standards different to those which now prevail for nuclear plant and equipment.
• The ventilation system which served D bay had been designed to suit the bay's original function as a fuel de-canning facility and did not reflect its use as a sludge storage facility.
• The sludge was sealed in D bay and no means existed for conveying it to the storage tanks located at SIXEP.
• The sludge was going to be difficult to mobilise and convey on account of its inherent properties, its high radioactivity, the presence of miscellaneous solid items, the ongoing generation of hydrogen and the compaction resulting from its storage.
• The encast plant and equipment and any large items of waste which had not been recovered prior to filling the bay with sludge would prevent access for retrieval equipment to all corners of the bay.

The requirement to recover the sludge in D bay is one item in the overall programme of work identified across the Sellafield site to be undertaken as part of the recovery of historic waste accumulations. This programme is summarised in the Company's Waste Management Plans Review (WMPR), which is reviewed annually. This summarises the driving forces and constraints affecting the programme as follows :

• Health and Safety of the work force
• Meeting Regulatory Requirements
• Minimisation of Environmental Risk
• Adopt Cost Effective Methods
• Minimisation of Waste Volumes

and these headings are used as success criteria to direct the development of strategies and design schemes for the retrieval and treatment of waste streams such as D bay sludge. This ensures a consistency of approach across a large number of projects.

In order to generate viable options for sludge recovery in the context of this overall site strategy, BNFL has adopted the use of Value Engineering techniques. This process includes the use of workshops involving all those, from plant operators through to senior managers and Customers, with a stake in the final outcome. Extensive use was made of the experience of outside industries in generating and developing options.

This approach enabled the project to bring forward comparable schemes for evaluation without wasting money on excessively detailed "front end" design work . The design process was progressed to a point where alternative options could be defined in sufficient detail to allow them to be compared. A multi-attribute decision analysis approach was then employed, utilising success criteria based on the list above, to select a preferred scheme. This was then progressed to the point where capital sanction could be sought to bring the design to the point where it could be put out to tender for design and supply.

The nature of the sludge, and the constraints surrounding its recovery made it important that a realistic scope was placed upon the design of the retrieval process. It was agreed by the "optioneering" process described above to follow a phased approach, and that the essential first step would be to recover as much sludge as possible prior to making any attempt to open up the bay, for example by removing the covers. Any attempt to recover or move large solid items during this first phase was therefore excluded. This would only be practicable when radiation levels had been reduced by the transfer of sufficient sludge away from the bay. This allowed the deployment of simple, low cost equipment during the first phase, when the operational constraints would be most severe.

The process is shown schematically at figure 4. The method of sludge retrieval is by hydraulic transfer using a fixed position jet pump positioned at the opposite end of the bay to that at which the sludge arrived. The slurry, at a concentration of not more than 10% by weight of solids, is transferred to a settling tank adjacent to Pond 2 which is also used for the transfer of other sludges and liquors to SIXEP. Transfers take place in batches of up to 70 cubic metres of slurry and flushing liquors. The supernate from the settling tank is transferred for treatment at SIXEP prior to sea discharge. Batch transfers continue until enough sludge has accumulated at the settling tank. It is then re-suspended and transferred to the bulk storage tanks at SIXEP. The sludge will eventually be recovered from the storage tanks and converted into a solid waste form for final disposal.

Submersible slurry pumps, inserted into the bay through the four access holes, are used to produce a jet of bay liquor which acts to liberate, mobilise and redistribute the sludge to bring it into the influence of the jet pump. The jet pump is designed on the basis of an assumed solids concentration generated by the pumps, which is then diluted by the motive water to deliver the reference transfer concentration of 10 % by weight.

Because the jet pump is fixed, the transfer concentration will diminish as the bay is emptied of sludge. To counteract this, a fifth slurry pump is located in the vicinity of the jet pump to direct a solids rich stream towards the jet pump when necessary. Provision is also included in the design for both the slurry pumps and the jet pump to be repositioned lower in the bay as the retrieval process proceeds.

The process described above was developed by means of the "optioneering" methods already outlined, supported by the development work which is described in the following section. As an example, a technically optimised solution for slurry transfer would have led to excessive mechanical complexity. The close liaison between designers and operators identified that, in this situation, radiation doses to those operating and maintaining the equipment, rather than process optimisation, was a dominant aspect of the scheme. As a result, the slurry transfer process is not optimum, but provides an equitable balance between process and operational requirements.

The recovery of the sludge from D bay represented a challenge from a difficult waste material in a unique location. The supporting development work, prior to the commitment of resources on the design phase onwards, had therefore to ensure that all aspects of the process were thoroughly understood. The project reviewed the process and identified all unknown aspects in a Development Requirements Document. Extensive literature searches were then carried out, both inside and outside the Company, to identify those issues for which some experimental work was required. Much data already existed regarding the performance of slurry pumps, jet pumps and the transfer of sludges through pipes, so that extensive studies into such aspects as erosion and critical velocity were not required.

The resulting programme, key elements of which are outlined below, therefore focused only on information that was not already available. All work was written up and approved to support the project's design and safety documentation.

Sludge sampling and characterisation - In addition to the quantitative data obtained, much useful information was derived by observing the sludge as it was extracted from the sampling tube, for example by observing how it initially maintained its coherence but subsequently collapsed in the absence of support.

Inactive simulant preparation - There was a clear need for demonstrations of aspects of the proposed process to establish design and operational parameters. This required that a simulant was prepared in sufficient quantities to enable inactive trials to be carried out. BNFL has extensive experience in the preparation of simulants representative of Magnox sludge, and the project therefore had access to a wealth of experience. However, the sludge properties measured by the characterisation work were too diverse to be economically represented in one simulant. Trials were therefore conducted using simulants appropriate to the property under review. For example, appropriate settling characteristics were increasingly seen as important to the process, and a simulant was therefore selected which was accurate in this respect.

Sludge mobilisation trials - Information was required on how far the sludge would be propelled by a jet of water so that the duty of the re-suspension pumps could be specified. These trials were carried out in a large tank. They confirmed the capability of the pumps to propel the sludge up to 5 metres and gave valuable design information on capacity, nozzle diameter and exit velocities.

Sludge transfer trials - The sludge had to be transferred over a distance of some 100-200 metres along a pipe route with a number of right angle bends. The design solution had to ensure that transfers could be accomplished without risk of pipeline blockage being caused by some of the larger particulate items. The transfer rig was set up at full scale and was used to support the selection of a jet pump, rather than a submersible slurry pump, as the transfer device and to confirm the pipeline geometry and transfer velocity parameters.

Scale model trials - Apart from the resolution of the specific design issues highlighted above, a principal area of uncertainty lay in the overall performance of the process as a whole. As already noted, the design did not attempt to optimise the transfer process, and it was important to establish how long it would take to empty the bay, as the sludge route to SIXEP is shared with other waste recovery projects, and an extended D bay programme would impact on the overall strategy. Rather than attempt costly and time consuming trials at full scale, the project commissioned a one fifth scale model of the process, and by the use of dimensional analysis techniques configured the key elements of the process in a perspex tank with associated pipe work. The rig used a simulant of the same particle size and nozzle velocities of the same magnitude as for full scale conditions. The validity of the approach was underpinned by external expert advice.

The strategy for design, supply and installation for D Bay involved an external contractor for the detailed design and supply of the main sludge transfer equipment. The initial conceptual design work was carried out by BNFL's own in house engineering design organisation. Installation of this equipment, service supplies, and construction of the pipe route to the settling tanks were carried out by BNFL. Before any of the on site work was carried out, a significant programme of preparation work was carried out:

The majority of preparation work was carried out on site. The whole project could be considered in two phases. The first stage involved the provision of a route for the sludge from D Bay to be pumped to the settling tank. This took the form of a shielded pipe route and the construction of a pipe bridge spur to join D bay to an existing pipe bridge which carried slurry from the original settling tank to the existing operational facility. This also included installation of a coaxial pipe from D bay to the settling tanks with localised lead shielding and a shielded instrument cabinet. The rest of the preparation was associated with ensuring that the best methods were being used to minimise the dose to the personnel involved:

ALARP Studies - This stands for "As Low As Reasonably Practicable" and refers to personnel dose. The whole installation procedure was examined as a number of work packs. Each work package was considered individually and in some cases work packages were split down further to individual tasks or operations. Each procedure was challenged against a structured series of guide words to prompt dose reduction ideas from a team made up from all levels of the safety and installation teams. The result was a set of installation procedures which would minimise dose uptake to personnel and minimise conventional safety risks.

Contractor's Works - The main sludge transfer equipment was assembled along with support steel work at the supplier's works. This allowed assembly techniques to be rehearsed.

Pre-installation dose reduction measures - It was also realised that it may be possible to improve the radiological conditions prior to commencement of installation in and around the D Bay area by replacing the liquor coverage of the sludge with clean de-mineralised water. The bay had not been significantly disturbed for quite some time and the covering liquor would now have dissolved some of the radioactive material. This was a contributory factor to the high dose rates measured in the D bay area. The liquor level in D Bay had not been routinely altered. The main part of the bay was in fact hydraulically sealed. There was no engineered system to add water to the bay but the level could be lowered by means of an installed pump which transferred the liquor to a part of an adjacent bay linked to the main pond. A new bay water top up system was installed with interlocks to the bay level system to prevent over filling or allowing sludge to become exposed. Purge operations were then carried out over several months and gave a five fold reduction in dose rates directly over the bay

Bay Hydraulic isolation - When the equipment was operated, it would be necessary to increase the level of liquor in the bay. This meant that there would be an increased chance of liquor overflowing from the bay via its engineered route. This route, however was no longer in an acceptable condition and a 100% seal needed to be made to prevent liquor inadvertently flowing down it.. A method of sealing was devised involving insertion of temporary plug using remote tools and then filling of the overflow weir box with a highly fluid grout. In order to test this, a mock up of the weir box and its access was constructed away from the plant. The team carrying out the job rehearsed the operation on the test rig. It was found that the bungs could be inserted quite quickly and that the grout provided a 100% seal when cured. The rehearsed procedure was repeated on the real plant. After the grout had cured the seal was confirmed by raising the bay level with clean water via the top up system described previously. Again the job proved to be very successful allowing installation of the equipment to proceed with confidence.

Ventilation Modification - As part of the installation the ventilation system in D bay was up graded to include hydrogen monitoring and critical flow instrumentation.

Building Access Control - The sludge transfer route from D bay to the settling tanks could not be fully shielded due to structural limitations of the existing building and the predicted radiation dose rates were therefore unacceptably high. It was therefore necessary for personnel to be excluded from the area during sludge transfers including active commissioning. It was decided that a new building access control system was needed to give assurances that the area could be adequately segregated from surrounding plants and access roads. This was designed, installed and commissioned as part of the project.

D Bay equipment support steel work - It was first necessary to provide a sound supporting structure for the main equipment. In the main bay this meant installation of support steel work to spread direct the load of the main pumps and mast handling units to the east and west walls of the bay. The load to the south wall needed to be minimised. It was also important that the steel work would be level and correctly supported. A raised "platform" covered D bay and it was essential that this was free from loading. This was especially difficult since the gap needed to be minimal to enable a seal to be effected for efficiency of the ventilation system. To achieve this, areas were prepared for the steel work feet, levelled and surveyed. The steel work was then manufactured to suit. Pre assembly was rehearsed off site.

An extremely difficult import route to D bay meant that the steel beams needed to be fabricated in 2 sections for assembly within the building. A lay down area was prepared close to D bay to be used as an assembly position. The assembly itself was critical and a detailed procedure was produced to ensure that the fastening of the connection plates left no residual stresses in the beams after assembly.

Withdrawal Well Area Installation - This area had been a flasking area for the export of fuel for reprocessing when D bay had been an underwater fuel de-canning bay. There had been a balustrade allowing access to the open liquor surface for flasks and long handled tools, lights, cameras etc. This had been previously covered over with steel slabs as a biological shield. The south walkway and balustrade wall are cantilevered from the south wall of D bay. Although approx. 1m thick, this wall was not of suitable structural integrity to support further construction. The existing steel slabs were supported from the balustrade and angles fixed to the dividing wall. The slabs would support the weight of one of the main pumps. The balustrade wall would also support the weight of the fixed jet pump.

The dose studies had shown that it would be impractical to remove the existing covers and modify them to support the pump. New covers were therefore designed along with a new angle to allow a greater proportion of the load to be carried by the dividing wall. In order to position the fifth pump in the correct location along with access to the jet pump etc. it was necessary to replace the five existing covers with four different size ones. This meant that the installation procedure was complicated and temporary covers were required for the gaps created during installation. The new angle was fixed to the dividing wall above the level of the slabs with a considerably larger web to reduce the load transfer to the balustrade. In addition the channel section of the balustrade which would take some proportion of the load needed to be stiffened by the addition of two additional parallel webs.

The second main job in the withdrawal well was the installation of the support for the jet pump. A large support bracket was fabricated called the encast item. This supports the jet pump assembly and locates it in a position to allow easy connection of the pipe work from the walkway without the need for the installers to lean over the balustrade with an associated radiation dose. The encast item obviously needed to be installed before the jet pump and was to be grouted into the balustrade wall. This required a "letter box" hole to be trepanned through the wall. With the covers removed the bracket part which would support the pump was craned into position and offered to the hole. An external fixing plate was then positioned on the outside face of the wall. Sealing compound was used on all faces and the plates were bolted together before the bracket was unslung. The external plate contained holes to allow the back filling of grout.

Main Pump Installation - These were fitted by the plant maintenance teams as practised on the contractor's test rig. Each pump was tested before installation and installed with an initial 1.5m mast length, i.e. the pumps remained clear of the active liquor

From the very earliest concept designs, BNFL had maintained a close relationship with the regulators. A thorough safety case had been produced to cover the installation and operation of the sludge transfer equipment. The commissioning phases were used to prove this safety case and the regulator scrutiny during this time was significantly higher than anticipated. Though satisfied with BNFL's thoroughness in preparing the safety case and preparing for installation and commissioning, the project raised issues which the Regulator had to address thoroughly. This had a big impact on the commissioning work and inevitably delayed progress.

Inactive Testing - The inactive tests were carried out at the premises of the contractor supplying the main sludge transfer equipment. The majority of the testing was centred around the complex PLC control system (over 600 lines of ladder logic). However it was essential that the inputs to the system be as realistic as possible to avoid re testing in the radioactive environment. This meant that real signals needed to be used from the proximity switches, encoders, MCC's etc. wherever possible. To this end the rig was constructed from scaffolding to represent the full height of the real D Bay (5.5m). The support steel work and mast handling units were fixed at the top of the scaffolding allowing the full mast lengths to be installed with the pumps close to ground level. Large tanks were used as a means of providing water to submerge the pumps for the tests.

This rig was also used as a means of introducing both operators and maintenance personnel to the new sludge transfer equipment. When used on site this would be operated remotely to avoid radiation dose. i.e. the operators would have only a SCADA system to indicate what was happening to the mechanical equipment.

The test rig therefore provided an excellent facility for the operators to see the relationship between the SCADA feedback and control with the actual operations of pumps, valves etc.. Operators were given "classroom" instruction followed by a walk around of the rig and a demonstration of the full operation. Plant maintenance personnel were also used to practice installation methods on the test rig along with all mechanical maintenance activities.

Pre Active Testing - Once again, the control system was extensively tested without any transfer of active material. Mimic signals were used instead of starting the main pumps and jet pump motive water was diverted to a drain so that no pumps actually started. Full pumping operations were simulated under these conditions.

As part of pre active commissioning it was found that the required flow rates for ventilation (especially hydrogen removal) were not achieved. A design review was carried out to solve the problem. The result was a further significant modification to the ventilation system which was subsequently successfully commissioned.

Active testing - Active commissioning comprised testing of the whole system with the main pumps able to run and motive water re-supplied to the jet pump. BNFL had developed a detailed safety case which included the potential for releases of hydrogen when the sludge was disturbed. A model had been produced which predicted the worst case volumes of hydrogen for a given sludge disturbance. The active trials were used to substantiate this model. Initially only agitation nozzles around the jet pump were used for a short time. The time was increased over a number of test runs and the process was repeated, increasing run times for the main pumps. Each time the volume of hydrogen released was assessed before proceeding with a longer run time.

The overall commissioning strategy proved to be extremely successful in terms of reducing dose to the commissioning team and ensuring that all major faults were rectified before the active phase of commissioning. This is borne out by the number of faults raised during each commissioning phase. There were 98 faults raised during inactive commissioning. These were generally associated with the overall control system and were resolved with no dose penalty. This number reduced to 44 for pre active commissioning. Most were related to the plant inputs that were only simulated in the previous phase and were largely solved remotely from the plant. Only 9 faults were raised during active commissioning. All related to aspects of the control system that could not have been identified easily until real plant conditions were used. Most of these were rectified without the need to work in a high radiation environment.

The benefits of the dose reduction measures described earlier have been realised over the whole project. The anticipated dose for the installation, commissioning and operation of the D bay equipment was expected to be between 487 - 994 mSv. The actual dose received was only 295 mSv, which includes 33 mSv from additional work. Only two pumps have been used in the main bay thus far. with another two to be installed once debris in the bay is removed. A fifth pump will be installed once the sludge level in the withdrawal well area is sufficiently low. Overall operations to date have removed approximately 1/3 of the estimated 230m³ of sludge. This figure is derived from a recent sonar mapping exercise which also depicted the sludge profile within the bay.

Operational experience has shown that the radiation dose rates along the sludge transfer route are less than anticipated and the requirement to prevent overall building access has been removed. Instead, access to the area immediately surrounding D bay is prevented during sludge transfer operations. This allows other building operations to continue unhindered. As sludge has been removed, the radiation dose rates around D bay have continued to fall.

The development of this project and its successful implementation has been the work of a number of enthusiastic and capable teams to which the authors wish to extend their thanks.

BACK