Year
2015
Abstract
The Safeguards technical and scientific basis now has a history of over 50 years and is sufficiently mature to have acquired certain inertia, traditions, prevailing presumptions and some features that might indeed have emerged as “safeguards classics”. It is, therefore, important to perform periodical and critical checks on the validity of these prevailing concepts, to reveal where they might require revision or updating. This paper targets emerging concerns with some well-established paradigms (such as belief in the power of independent NDA measurements). As is well known from experimental physics, credible high-precision measurements usually require very thorough preparation, optimization of measurement conditions, removal of all possible negatively influencing factors, the ability to make iterative adjustments to measurement geometry, etc., which are rarely possible under the semi-industrial conditions normally associated with safeguards inspections. As a consequence, one might observe a questionable trend of essentially extrapolating ‘best achievable’ uncertainty parameters to all data obtained in the field. In practical terms, the problem of interpreting the results of field measurements becomes the challenge of judging whether an observed discrepancy between declaration and measurements is due to an actual diversion of material or is attributable to imperfect measurements. Imperfect measurements are generally much more likely to be encountered than deliberately falsified declarations. As a result, significant effort is usually dedicated towards analyzing ‘legitimate’ reasons for observed discrepancies. As a consequence of this analytical process, corrective factors, adjustments, and supplementary calibrations are often introduced, which may finally have an adverse impact on the detection of an actual diversion. There are no straightforward means of addressing this dilemma in full. However, one instrumental solution might be shifting the focus of NDA from striving to attain (arguable) precision in measuring a single quantity, towards expanding the set of directly measurable ‘observables’ and subsequently utilizing these to test “the hypothesis of a declaration being valid” through assessing its compatibility with such multifaceted observables. To ensure the success of this approach, the ability to derive and predict reliably physical observables from material declarations becomes a crucial factor, which depends critically on achieving progress in modelling and computer simulation techniques, plus making these techniques sufficiently user-friendly to utilize in the course of actual field activities.