Doctoral thesis, 2001

The study of reactor noise theory concerns questions about cause and effect relationships, and the utilisation of random noise in nuclear reactor systems. The diversity of reactor noise theory arises from the variety of noise sources, the various mathematical treatments applied and the various practical purposes. The neutron noise in zero-energy systems arises from the fluctuations in the number of neutrons per fission, the time between nuclear events, and the type of reactions. It can be used to evaluate system parameters. The mathematical treatment is based on the master equation of stochastic branching processes. The noise in power reactor systems is given rise by random processes of technological origin such as vibration of mechanical parts, boiling of the coolant, fluctuations of temperature and pressure. It can be used to monitor the reactor behaviour with the possibility of detecting malfunctions at an early stage. The mathematical treatment is based on the Langevin equation. The unity of reactor noise theory arises from the fact that the useful information from noise is embedded in the second moments of random variables, which lends the possibility of building up a unified mathematical description and analysis of the various reactor noise sources. Exploring such possibilities is the main subject among the three major topics reported in this thesis.
The first subject is within the zero power noise in steady media, and we reported on the extension of the existing theory to more general cases. In Paper I, by use of the master equation approach, we have derived the most general Feynman- and Rossi-alpha formulae so far by taking the full joint statistics of the prompt and all the six groups of delayed neutron precursors, and a multiple emission source into account. The involved problems are solved with a combination of effective analytical techniques and symbolic algebra codes (Mathematica). Paper II gives a numerical evaluation of these formulae. An assessment of the contribution of the terms that are novel as compared to the traditional formulae has been made.
The second subject treats a problem in power reactor noise with the Langevin formalism. With a very few exceptions, in all previous work the diffusion approximation was used. In order to extend the treatment to transport theory, in Paper III, we introduced a novel method, i.e. Pade approximation via Lanczos algorithm to calculate the transfer function of a finite slab reactor described by one-group transport equation. It was found that the local-global decomposition of the neutron noise, formerly only reproduced in at least 2-group theory, can be reconstructed. We have also showed the existence of a boundary layer of the neutron noise close to the boundary.
Finally, we have explored the possibility of building up a unified theory to account for the coexistence of zero power and power reactor noise in a system. In Paper IV, a unified description of the neutron noise is given by the use of backward master equations in a model where the cross section fluctuations are given as a simple binary pseudorandom process. The general solution contains both the zero power and power reactor noise concurrently, and they can be extracted individually as limiting cases of the general solution. It justified the separate treatments of zero power and power reactor noise. The result was extended to the case including one group of delayed neutron precursors in Paper V.

neutron fluctuations

master equations

correlations

stochastic processes

the Langevin technique

power spectra

reactor noise

covariances

Chalmers, Department of Reactor Physics

Physical Sciences

91-7291-017-8

CTH-RF -Chalmers University of Technology, Department of Reactor Physics: 159

Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie: 1701