Methodology for Identification of the Coolant Thermalhydraulic Regimes in the Core of Nuclear Reactors

The paper deals with one of the acute for the nuclear energy problem of accident regimes of NPPs recognition diagnostics using noise signal diagnostics methodology. The methodology intends transformation of the random noise signals of the main technological parameters at the exit of a nuclear facility (neutron flow, dynamic pressure etc.) which contain the important information about the technical status of the equipment. The effective algorithms for identification of random processes wore developed. After proper transformation its were considered as multidimensional random vectors. Automatic classification of these vectors in the developed algorithms is realized on the basis of the probability function in particular Bayes classifier and decision functions. Till now there no mathematical models for thermalhydraulic regimes of fuel assemblies recognition on the acoustic and neutron noises parameters in the core of nuclear facilities. The two mathematical models for analysis of the random processes submitted to the automatic classification is proposed, i.e. statistical (using Bayes classifier of acoustic spectral density diagnosis signals) and geometrical (on the basis of formation in the featured space of dividing hyperplane). The theoretical basis of the bubble boiling regimes in the fuel assemblies is formulated as identification of these regimes on the basis of random parameters of auto spectral density of acoustic noise (ASD) measured in the fuel assemblies (dynamic pressure in the upper plenum in the paper). The elaborated algorithms allow recognize realistic status of the fuel assemblies. For verification of the proposed mathematical models the analysis of experimental measurements was carried out. The research of the boiling onset and definition of the local values of the flow parameters in the seven-beam fuel assembly (length of 1.3 m, diameter of 6 mm) have shown the correct identification of the bubble boiling regimes. The experimental measurements on real WWER core assemblies were analysed as well. On the basis of model of Bayes classifier for bubble structure of two-phase flow in fuel assemblies of WWBR-440 (intends usage of 28 dimensional accidental realizations of ASD of neutron noise) the reliable identification of the pointed regimes of fuel assemblies in WWERs up to 98% was obtained. On the basis of geometrical mathematical model of identification at essentially more limited volume of teaching sampling the recognition of ASD realizations of the neutron noise of the same both dimensions and quantity of the reliability of correct identification of these parameters was up to 92%. The recognition of the pointed thermalhydraulic parameters was carried out on the basis of experimental research of ASD of acoustic noise parameters of the experimental fuel assembly with electrically heated imitators using the two recognition models — statistical and geometrical. It confirmed high efficiency of the algorithms developed. The average reliability of identification of the first vapor bubbles activation regime at the heat transfer surface was not low then 90%.