HM Nuclear Installations Inspectorate

An Investigation into the Falsification of Pellet Diameter Data in the MOX Demonstration Facility at the BNFL Sellafield Site and the Effect of this on the Safety of MOX Fuel in Use

MOX FUEL MANUFACTURE AT SELLAFIELD

2.1 History of MOX Fuel
2.2 MOX Fuel Production Process

History of MOX Fuel

11. UKAEA and BNFL have been producing MOX fuel at Sellafield for more than 30 years. Extensive development work on the manufacture of MOX fuel in support of the UK Fast Reactor development programme has been completed. Associated fuel irradiation programmes carried out in the Dounreay Materials Testing Reactor, the Dounreay Fast Reactor and the Prototype Fast Reactor (PFR) were used to assess the influence of fuel properties such as density, fuel form and chemical composition on irradiation performance. In addition, the compatibility of fuels with cladding materials and the influence of manufacturing variables on the reprocessing behaviour of high burn-up fuels was assessed. In support of these programmes, more than two tonnes of experimental MOX fuel and 18 tonnes of driver charge MOX fuel was produced by BNFL in the Sellafield production plant facilities. Approximately 98,000 MOX fuel pins were irradiated in PFR between 1974 and 1994. The initial burn-up target set for PFR was approximately 75GWd/t HM, but large numbers of the pins reached burn-up levels well in excess of this target. Over 2,400 pins successfully reached burn-ups in excess of approximately 150GWd/t HM and 12 pins reached the level of approx 240GWd/t HM without failure. This programme has demonstrated the high quality of the fuel manufacturing process.

12. Following the Government's decision in 1988 to end the fast reactor programme, UKAEA and BNFL decided to collaborate in the development of a process to develop the use of MOX fuel to recycle separated plutonium in light water reactors. BNFL developed a two-stage strategy which involved:

the construction of a small-scale plant, MDF, to produce commercial quality fuel for irradiation in commercial reactors; and
the construction of the much larger scale plant, SMP, for bulk fuel supply.

13. BNFL's MOX Demonstration Facility is located in B33 at Sellafield. The building was originally owned and operated by UKAEA to support the Dounreay Fast Reactor Project, providing the experimental fuel for the Prototype Fast Reactor. 14. A decision was taken to build MDF in B33 where ventilation and support services already existed for plutonium operations. Construction of MDF commenced in 1991. The plant was handed over for initial commissioning in late 1992 and uranium commissioning started in May 1993. Plutonium active commissioning commenced in late 1993. In April 1994 ownership and operation of the Facility transferred entirely to BNFL. 15. The first fuel assemblies manufactured were for a utility in Switzerland. In late 1995 the plant produced fuel to a different design which was manufactured throughout 1996 for a German utility. This was followed by further Swiss fuel campaigns during 1997. At the end of 1997, the plant began to manufacture fuel to a Japanese design, which was manufactured in two separate tranches until September 1999 when the falsification came to light.

MOX Fuel Production Process

16. The Mixed Oxide Fuel Demonstration Facility is used to manufacture Mixed Oxide fuel containing natural or depleted uranium dioxide (UO₂) enriched with plutonium dioxide (PuO₂) for use in Pressurised Water (PWR) reactors. All the manufacturing processes from mixing the UO₂ and PuO₂ powders up to the stage where the completed fuel pellets are inserted into Zircaloy cladding tubes take place inside gloveboxes: stores are provided for completed fuel rods and for fuel assemblies. Details of the various checks made through the production process to provide quality assurance of the important fuel characteristics are given in Appendix 2.

Fuel Pellet Production

17. Accurately weighed quantities of PuO₂ and UO₂ powders are milled together with additives to produce homogenised MOX powder. This is then tumbled in a spheroidiser to produce free flowing granules. There is provision for sampling these granules for off-line analysis if required. The granules are pressed into pellets and a sample is checked for density. The pellets are then sintered under reducing conditions to produce ceramic MOX fuel pellets. The ceramic pellets are ground to meet closely defined customers' specifications, then checked using an automatic inspection system for dimension and visually inspected for surface defects. Samples are taken for physical/chemical analysis. Fuel pellets which pass the tests are then transferred to a buffer store.

18. The Automatic Inspection System in the Pellet Inspection Glovebox uses a precision calibrated laser micrometer to take three separate diameter measurements of each and every pellet. Any pellet containing one or more out of specification results is automatically rejected by a gate mechanism. This is failsafe by design in that failure of any part of the measuring or gate control mechanism results in the gate remaining in the closed position. Out of specification or unmeasured pellets are thereby guaranteed to be ejected from the process stream. The specification range for pellet diameters for the Kansai contract is ± 0.0125 mm. The claimed accuracy of the automatically measured diameters is ± 0.002mm.

19. The Lot identity of fuel pellets received from the pellet production area of MDF is recorded on receipt at the in-line pellet store glovebox. The pellet store has the capacity to store up to ten cassettes of fuel located on trays.

20. Each pellet Lot typically consists of up to 4000 pellets. The pellets are transported/ stored in a cassette containing 13 trays. Visual inspection is carried out on every single pellet in the Lot (this is referred to as screening). A different operator from that involved in "screening" then carries out the secondary sample checks for visual defects.

21. The secondary sample check for diameter (known as "overinspection" for diameter) is also carried out. An approximately equal number of pellets is selected at random from each tray to give a total of 200 pellets. Each of these 200 pellets is then manually measured for diameter by placing them lengthwise on a "V" shaped block and measuring top, middle and bottom diameters using a calibrated laser micrometer (the measurement takes place inside a glovebox, with the pellets being manually positioned by one of the operators using tweezers). The micrometer readings are presented on an LED display outside the glovebox and the measurement readings manually input to a spreadsheet specific to the Lot being measured, by a second operator using a computer adjacent to the glovebox. The overinspection for pellet diameter requires two operators and takes between 1.5 and 2 hours to complete.

22. 600 diameter readings in total are therefore recorded covering the 200 pellets which form the 5% sample out of each Lot of approximately 4000 pellets. The spreadsheet automatically highlights individual readings which are outside specification (i.e. too high or too low), and calculates the mean diameter for each pellet based on the top, middle and bottom readings. If the mean diameter is outside specification, then the pellet is automatically identified on the spreadsheet as a reject. The Lot will pass or fail depending on the number of rejects found. Under the normal criterion, the Lot will pass on five or less reject pellets but fail on six or more.

23. If a Lot fails diameter "overinspection" it is returned for 100% automatic diameter inspection followed by re-overinspection for diameter measurement. A tightened inspection regime is applied to any Lots which undergo such inspection for a second time. Such Lots are required to pass on three or less rejects, and fail on four or more rejects.

24. There is a third independent check of diameter as part of the density measurement, but this involves a smaller sample of 20 pellets.

25. Following additional visual and sample diameter checks on the pellets, and also when the Pu enrichment has been confirmed as being within specification, each cassette is 'released' from its store position and the pellets are fed forward from the cassette trays to form pellet sub-stacks in the stack make-up area of the pellet load glovebox. Several sub-stacks are required to fill each cladding tube.

2.2.2 Fuel Rod Production

26. The next stage, rod fabrication involves the loading of sub-stacks of fuel pellets into the cladding tubes and sealing and decontaminating the rods prior to inspection. Facilities for rod recovery and the reclamation of fuel from reject rods are also located in this area. All operations in rod fabrication are carried out in a series of interconnected and free-standing gloveboxes.

27. Pellets are loaded into each tube through a closely toleranced insert. This insert would itself prevent the loading of any significantly oversized pellets which had somehow bypassed the 100% automatic diameter check stage.

28. The sub-stacks of pellets are length checked and weighed prior to being manually loaded through the toleranced insert into the neck of the fuel rod. The space between the end of the pellet stack and the top of the rod, is measured (plenum length check) and pellets are added, as required, to make up the correct stack length.

29. A plenum spring is manually loaded into the fuel rod and the top end cap is fitted and welded into position.

30. The full rod is then filled with helium through a small hole in the end cap to a required pressure and the hole is welded closed.

31. After further visual checks, checks on sample rods to ensure the correct internal helium pressure, and checks for surface contamination, rods are transported six at a time to the Fuel Assembly Area for further inspection.

32. Trays of finished rods received from the Rod Fabrication Area are subjected to a helium mass spectrometry leak test. Each tray of six rods is placed in a vacuum chamber where the pressure is reduced to a pre-set level. The atmosphere is then analysed for the presence of helium. This sequence of operations is carried out automatically. In the event of a positive helium result, indicating a loss of integrity in a rod tube, the six rods are tested individually. Rejected rods are sent for reclamation or recovery. After leak testing, the rods are examined by X-radiography. The rod is laid on a strip of photographic film on a table and passed through an X-ray machine. The film is processed and used to examine girth and seal welds, to detect defective pellets, and to confirm the presence of the correct components. After this, a special detection system checks the enrichment of every pellet in each rod, and the results are recorded on a scan output.

33. Manual checks for rod straightness, length and diameter are then carried out. The only stage at which the operator has significant hands-on contact with the fuel rods is during the rod straightness check. This operation involved manually rolling the rod on a flat table and inserting feeler gauges under any revealed gaps, and a visual inspection for weld form, scratches and surface blemishes is also carried out.

34. Reject rods are returned to the rod reclamation glovebox for reworking, or pellet recovery as appropriate. Accepted rods are transferred on a shielded trolley to the rod store where they are loaded into magazines.

2.3 Quality Control / Quality Assurance Procedures

35. The fuel manufactured in MDF for commercial purposes is covered by Quality Control Plans which must be approved by the customer before fuel manufacture can begin. In this way, BNFL provides assurance that the fuel supplied will perform in reactor in accordance with the design specification. The quality control plan (amongst other things) defines for each characteristic the specific requirement, the method of analysis or measurement, the frequency of measurement and/or sampling, and how the information is recorded. The manufacturer must adhere to the quality control plan and the customer indicates which of the checks he wishes to witness.

36. The customer approves the quality control plan for manufacture of MOX fuel pellets, MOX fuel rods and MOX fuel assemblies in MDF before the production campaign begins and only those modifications approved by the customer can be incorporated once the quality control plan has been approved.

37. Customers carry out their own checks on BNFL's adherence to the QC plans as the fuel is manufactured.

Added to HSE website on 18th February 2000