Uncertainty and Sensitivity Analysis for Nuclear Reactor Noise Simulations
Examensarbete för masterexamen
Nuclear engineering, MSc
Neutron noise in nuclear reactors is related to the fluctuations of the neutron flux that can occur in stationary operational conditions. These fluctuations may be induced by different types of phenomena such as vibration of reactor components, perturbations of the coolant flow, stochastic aspects of nuclear fissions, etc. Neutron noise carries information about the actual properties of the reactor, and its analysis can help to diagnose anomalies that can evolve over time into more severe issues. The aim of the thesis is to evaluate the impact of modelling uncertainties on reactor noise simulations. For this purpose, the reactor noise simulator CORE SIM together with a statistical methodology for input uncertainty propagation are used. A simplified light water reactor with a thermal neutron absorber of variable strength placed in one point of the core, is analyzed. The uncertainties associated to the macroscopic neutron cross-sections of the core and to the neutron absorber, are considered. These uncertainties are assumed to behave as random variables with either a uniform or a normal probability density function. Then random samples can be generated for each type of cross-sections in each point of the core and for the noise source (i.e. the neutron absorber), and be propagated to the CORE SIM outputs. The calculations are performed for both the forward and the adjoint noise, and the results are used for the uncertainty and sensitivity analysis. The uncertainty analysis shows that the probability density functions of the outputs are close to normal distributions, whether the input uncertainties are normally or uniformly distributed. The Pearson correlation coefficient is used to quantify the sensitivity of the calculated forward and adjoint noise to the uncertain input parameters. The noise in the core is strongly correlated to the cross-sections taken at the position of the noise source, and to the characteristics of the noise source. In general terms, the most influential macroscopic cross-sections are found to be the thermal fission, thermal capture, and removal ones.
Fysik , Physical Sciences