A Socratic Approach to Independent Verification

In carrying out its inspection and verification responsibilities, the International Atomic Energy Agency (IAEA) performs a variety of chemical analyses and nondestructive assays (NDA) of various nuclear materials. Most of the IAEA's NDA measurements for materials accountability purposes are made in operating facilities using portable or transportable instruments. This approach has several advantages, including the prevention of tampering with the instrument by a potential diverter, since the instrument remains in the custody of the inspector. Recent interest in the possible use of NDA instruments that would not be in the direct custody of the inspector is due to two factors: 1) Greater use is being made of in-plant NDA instruments by facility operators for both process control and materials accountability. In many instances, these instruments provide data that would be useful to international inspectors. 2) There is a need for near-real-time accounting of material in certain types of facilities, obtainable only with in-plant measurement systems that cannot be easily moved. From these considerations, the question naturally arises as to how the IAEA might independently verify measurements made with an instrument that is in the custody of a facility operator. Although various people have examined the problem of independent verification for a specific NDA instrument monitoring a specific process, it appeared worthwhile to see if the problem could be approached in a more general way. In thinking about how actual tampering with an NDA instrument to conceal a diversion might be falsely interpreted by an inspector as being a routine measurement error, we posed the following (Socratic) questions: 1) Is one type of measurement error more likely than another? 2) Is one type of diversion attempt more likely than another? 3) Why do safeguards personnel normally test for measurement errors before exploring diversion possibilities? 4) What should be the difference in approach when testing for measurement errors vs testing for diversion? • This technical note is intended to help address these questions in a qualitative way. Although the present treatment is rather elementary, it may provide slightly different insight into some of the problems of independent verification, and perhaps stimulate further thought in an area of increasing importance to international safeguards.