Sensitivity Analysis for Activation Problems

¹ UK Atomic Energy Authority, Culham Science Centre, Abingdon OX14 3DB, United Kingdom
² Culham Electromagnetics Ltd, Culham Science Centre, Abingdon OX14 3DB, United Kingdom

Abstract

A study has been made as to how to develop further the techniques for sensitivity analysis used by Fispact-II. Fispact-II is a software suite for the analysis of nuclear activation and transmutation problems, developed for all nuclear applications. The software already permits sensitivity analysis to be performed by Monte Carlo sampling, and a faster uncertainty analysis is made possible by a powerful graph-based approach which generates a reduced set of nuclides on pathways leading to significant contributions to radiological quantities. The peculiar aspects of the sensitivity analysis problem for activation are the large number, typically thousands, of rate equation parameters(decay rates and reaction cross-sections) which all have some degree of associated error, and the fact that activity as a function of time varies as a sum of exponentials, so appears discontinuous as rate parameters are varied unless the sampling frequency is impractically fast. Nevertheless, Monte Carlo sampling is a generic approach and it is therefore conceivable that techniques more targeted to the activation problem might be beneficial. Moreover, recent theoretical developments have highlighted the importance of a two-stage approach to mathematically similar problems, where in the first stage, information is collected about the global behaviour of the problem, such as the identification of the rate parameters which cause the greatest variation in dose or nuclear activity, before a second stage examines a problem with its scope restricted by the information from the first. In the second stage, for example, Quasi-Monte Carlo sampling may be used in a restricted parameter space. The current work concentrates on the first stage and consists of a review of possible techniques with a detailed examination of the most promising pathways reduction approach, examined directly using Fispact-II. All the evidence obtained demonstrates the strong potential of this approach.