Reactor Physics - Modeling and Evaluation of Cross Section

In the 1920s-1940s, the theoretical and experimental proliferation of nuclear physics, a discipline in full expansion at the time, also had to respond to the need to control atomic energy. Indeed, quantifying and controlling the chain reaction in a nuclear system requires the resolution of equations describing the behavior of neutron flux and the evolution of the concentrations of the various nuclei; equations whose coefficients are constants known as "nuclear data". This theme has therefore been of prime importance for a very long time. This objective corresponds to what Emilio Segrè called having "good numbers": "In an enterprise such as the building of the atomic bomb the difference between ideas, hopes, suggestions and theoretical calculations, andsolid numbers based on measurement, is paramount." At the time, this concept was already associated with the need to develop theoretical approaches (models of nuclear reactions, fission), to initiate the measurement of fundamental physical observables through microscopic experiments, and, very quickly, to set up so-called "integral" experiments (e.g. critical mass measurement, etc.). These three pillars still characterize the nuclear data evaluation activity today.

Nuclear data continues to play an essential role, along with numerical methods and associated algorithms, in design and analysis calculations for all nuclear energy applications, from radiation protection to criticality. As a result of the reduction in computational biases due to advances in the digital sciences (applied mathematics, software engineering, computer science, etc.), the dependence of results on the quality of nuclear data is becoming paramount.

This article is devoted to the process of evaluating these data, with particular emphasis on the basic physical phenomena, for which certain modelling elements can still be perfected. The aim of this process is to obtain quantities that can be used by neutronics calculation codes, which enable the design and analysis of nuclear systems, particularly reactors. The evaluation combines physics models, mathematical and computer techniques, experiments designed to obtain quantities and validation of the whole at various levels. Effective cross-sections with a small number of groups were used very early on, for military and exploratory applications, with summary calculation tools. Since the first libraries of effective sections for computational codes in the 1960s-1970s, the process has evolved considerably towards the use of ever finer, ever more predictive models, but the overall complexity is such that it is still illusory to replace the ensemble with calculations from first principles. The aim of what follows is to explain the approach and provide food for thought for improving the whole towards ever greater...