Assessment of the effectiveness of safeguards

Assessments of the effectiveness of nuclear material safeguards have so far been mainly verbal and qualitative. Such assessments cannot give any clear indication of the probability that undetected diversion of material occurs, or indicate what limit is placed (or should be placed) upon this risk. Attempts to develop quantitative assessment methods have concentrated attention upon the probability of detection of diversion, should diversion occur. This approach could lead to some false conclusions, because detection probability is not a complete measure of the effectiveness of safeguards. Deterrence of diversion attempts is also important. Attention is drawn in the paper to a range of factors which could influence the risk of diversion, and a mathematical model is proposed to illustrate their possible effects. This model indicates an approach of the type which would have to be taken in order to evaluate and assess the effectiveness of safeguards in explicit, quantitative terms. The general approach is referred to as probabilistic assessment of safeguards effectiveness (PASE). The value which probabilistic assessment has already demonstrated in the field of safety suggests that this may be a fruitful line of development in safeguards work. However, some problems of application are clear. In particular: (i) the effectiveness of safeguards is strongly influenced by human decision processes; (ii) PASE contains factors which can only be subjectively evaluated. PASE appears to be most readily applicable to optimisation studies. For compliance assessments, it should be used initially to supplement rather than replace existing methods. (author)