Quantification of integral data effectiveness using the concept of active subspace in evaluated nuclear data validation

Abstract

ABSTRACT In order to know the performance of evaluated nuclear data in reactor physics or radiation shielding calculations, its benchmark testing with integral data is mandatory. Nowadays we have a huge amount of integral data, but some of them are quite similar to each other. We need to know the independency of available integral data, and also to choose a proper set of a limited number of integral data for benchmark calculations. Furthermore, it is beneficial to know how effective a set of integral data is for independent validation of each nuclear data in performing the validation test of nuclear data. In order to quantify the effectiveness of integral data in nuclear data validation, we propose several methods based on a concept of the active subspace. With the proposed methods, we can quantify the independency of a set of integral data, choose a minimum set of proper integral data, and quantify the possibility of independent validation of nuclear data from a set of integral data. These methods are adopted to a set of fictitious integral data and a set of actual integral data including experimental data about and reaction rate ratio. Through these applications, effectiveness of these integral data has been successfully quantified. Furthermore, the proposed concept is utilized to interpret the nuclear data compensation effect, which has been recently discussed in the community of nuclear data.