Technical Note: Development of a generalized source model for flux estimation in nuclear reactors

Abstract

Although very precise, the Monte Carlo method is difficult to exploit in an industrial context. It requires a relatively long calculation time to obtain results with a reasonable statistical accuracy. In the context of so-called fleet reactors, and especially, to facilitate the transposition from one reactor to another, it is proposed to develop a method allowing the modeling of the reactor source in a synthetic and adjustable way. One approach is to perform a complete Monte Carlo simulation and to generate during this simulation a phase-space file (PSF) by judiciously placing a storage surface. This PSF provides a precise distribution of the particle parameters and can be used as a source in a second simulation. This method is interesting but can lead to the introduction of a latent variance. An approach that overcomes this limitation is to create a so-called generalized source model based on histograms. By sampling the initial parameters of the particles, no particle can be generated twice with exactly the same parameters, which leads to a reduction of the latent variance present in PSF. In addition, histogram-based sampling is faster than PSF reading and also allows for correlations between particle parameters.