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

INTRODUCTION

1.1 Circumstances Leading up to this Investigation
1.2 Objectives
1.3 Methodology
1.4 Statistical Analysis

Circumstances Leading up to this Investigation

1. In September 1999 BNFL reported to NII that some of the secondary checks on mixed oxide (MOX) fuel pellet diameter in the MOX Demonstration Facility (MDF) at Sellafield had been falsified.

2. Non-compliance with Nuclear Site Licence requirements is a serious matter, especially the deliberate falsification of records. NII therefore mounted an investigation into the events at MDF. This report sets out how NII conducted the investigation, its findings and recommendations to ensure that such events will not be repeated in the future.

Objectives

3. The objectives of the NII investigation were to: determine if any falsification of QA records had occurred, and if so; determine the extent of the falsification; determine who was involved; determine why falsification happened; examine the adequacy of BNFL's own investigation; determine if the data falsification had prejudiced the ability of the fuel to be safe in use; and make recommendations to HM Chief Inspector on the requirements for BNFL to restart operations in MDF.

Methodology

4. The investigation was carried out following the normal procedures for an event on a nuclear licensed site. The Site Inspector responsible for the area of Sellafield which included MDF carried out the investigation, supported as necessary by specialist colleagues.

5. The first phase of the investigation looked at the MOX fuel production process, particularly the quality control and quality assurance procedures, to determine the ease with which data relating to the parameters governing the safety of the fuel in use could be falsified. The second phase was to carry out a statistical analysis of the fuel pellet diameter measurements used for the Acceptable Quality Level (AQL) checks for the fuel manufactured for supply to the Japanese electricity utility, Kansai Electric Power Co. This statistical analysis was carried out by HSE's Epidemiology and Medical Statistics Unit. The third phase involved Inspectors who visited the plant and interviewed the staff to determine the extent of their knowledge of the falsification of the QC records, why this had happened, and to gain a view on the adequacy of the Management of the plant and their control and supervision of operations.

6. In addition to these activities, the Inspectors looked at the adequacy of BNFL's own investigation. This was done to check that NII had a complete picture of events. It was also done to check the thoroughness of the BNFL investigation process to see if there were implications for other parts of the Sellafield site. Finally, NII specialists examined the BNFL case justifying the safety of fuel in use against the criteria normally used in the UK and the USA, particularly the effect of fuel pellet diameter variations.

7. The Site Inspector carried out most of the on-site interviews and co-ordinated the activities and input from NII's specialists.

Statistical Analysis

8. To help determine the extent of the fuel pellet diameter data falsification, NII commissioned an independent statistical analysis of the AQL data. BNFL supplied NII with both the AQL and the 100% automatic laser measurements of pellet diameter. (The process is described later in this report.) NII consulted HSE's Epidemiology and Medical Statistics Unit (EMSU) who did the analysis. The timescale for the analysis limited the types of possible data falsification for which it was possible to test: the intention was to determine whether falsification had taken place and to establish its extent.

9. EMSU identified three types of data copying which were worthy of study: copying a whole spreadsheet and replacing some data entries - falsification being detected by assessing the number of cell for cell matches between spreadsheets; copying whole rows of three data entries, ie. the three diameter readings for one pellet, within a spreadsheet - falsification being detected by counting the number of duplications of rows within each spreadsheet; other manipulation and invention of data - falsification being detected by internal checks within the sample spreadsheet and by external comparison with data on the population from which it was supposedly drawn.

10. Although the above approach was judged at the time to be the most fruitful based upon the information supplied by BNFL and from a priori reasoning, it was recognised that the analysis would not be exhaustive, and other methods of falsifying the data could have been used. Item (c) above was intended to cover such things as pure invention of pellet diameter data and some other forms of copying and manipulation of data from one spreadsheet to another. However HSE's statistical experts advised that for many of the more sophisticated forms of manipulation, it would be difficult to devise appropriate tests and virtually impossible to interpret the results.

Added to the HSE website on 18th February 2000