A stochastic model of a nuclear reactor with directed percolation. Overjump and maximum power

Abstract

A stochastic risk model is applied to simulating the behavior of a nuclear reactor in a situation where the neutron chain length is described by a distribution with heavy "tails," such as the Pareto distribution. Probabilities of a fluctuation exceeding a critical threshold are obtained, and risk bounds for power-law distributions of jumps are estimated. Functionals of the reactor power maximum, the instant of first reaching the maximum, and the distribution of the overjump magnitude are considered. A relationship between the shape parameter and the physical constants of reactors is obtained, as well as the relationship with the noise spectrum and physical constants. The finite dimensions of a real reactor are taken into account. The autocorrelation function of the truncated Lévy process and its relationship with the frequency filters of the neutron flux monitoring equipment are considered.