ESARDA NDA working group benchmark exercises

After the development of the co-incidence analysis and the time correlation analysis in the eighties, the methods have proven their reliability in the actual world of nuc]ear materials safeguards, The shift register analyser technique connected to active or passive equipment, as the HLNCC, AWCC and NCC, is widely used for in field inventory verification. Besides the equipment and the interpretation model, prediction methods are available based on a simple point reactor model, directly derived from the theory. This prediction method was satisfactory for the verification of homogeneous or symmetric samples, Once the sample becomes somewhat aberrant from the normal spectrum of samples treated with the existing prediction model, the approximations used become too heavy and the results of the prediction strongly diverge from the field measurement results, such that the need for a more sophisticated prediction model was felt. One of the very popular Monte-Carlo codes playing an important role in nuclear safeguards development of instrumentation and prediction models is MCNP [1], Inter comparison materialof this code with other well known codes is extensively available in literature for what concerns its use in reactor physics applications. Because the need for prediction codes in nuclear safeguards was increasingly felt over the last five years, and due to the fact that the price of CPU- time is decreasing with orders of magnitude every few years, Monte-Carlo simulation methods gained eriormously in importance. MCNP prediction is employed in safeguards R and D applications ta optimise instrument design and ta predict experimental results. A further application which is becoming attractive due to the aboveconsideration, is the field use of MCNP for predicting or improving sample verification results, e.g. reducing the need of physical standards in NDA applications. The ESARDA working group for Techniques and Standards for Non Destructive. Analysis decided to launch two international benchmark exercises. The first initiative aims to confront different approaches to "Reals" evaluation with each other and with measurement results obtained from a well documented experimental set-up. The present contribution will discuss briefly the " Reals prediction " benchmark exercise, the experimental set-up built in PERLA, and the conclusions of the inter-comparison will be drawn. In the second initiative the ESARDA-NDA working group decided to work out a benchmark exercise for MCNP on simple geometries typically related to the configurations frequently. encountered in safeguards applications. The exercise will confront the simulation results with experimental measurements in different fields as for instance, interaction of fast neutrons with 3He detectors, moderation, thermal and fast neutron induéed fission in Uranium, the effect of lead shielding on neutron transport, etc.. The exercise focuses mainly on the use and performance of MCNP, The present contribution will deal with the geometric aspects of the benchmark exercise.