On the estimation of the unknown reactivity coefficients in a CANDU reactor

A space-time kinetics based inverse architecture method is suggested to analyze the reactivity variations associated with power excursions in a generic CANDU reactor. It is intended to provide diagnosis tools to gain enhanced control thereby ensuring safe operation of the plant. A methodology for analyzing the data available from the in core flux detectors and extracting the unknown reactivity coefficients is presented. The proposed system uses a reference model in conjunction with an optimal estimator. The reference model is composed of a state space representation of the space-time dynamics of neutron flux in the core, based on modal expansion approximation, and a time domain optimal estimator filter. We investigated three different estimation techniques based on recursive prediction error method (RPEM), dual extended Kalman filter (DEKF), and joint extended Kalman filter (JEKF). We compared their applicability to the estimation of coolant-void dynamic reactivity in loss-of-coolant accident in a CANDU reactor. The state equations also include the characteristics of the detector responses. The thermal hydraulic models were not included in the calculations. Two different types of detectors are considered in this analysis, the over prompt responsive Platinum detector of the reactor shutdown systems, and the under delayed responsive Vanadium detector of the flux mapping system.

► An inverse architecture method is presented for studying the space-time dynamics of neutron flux in power excursions in CANDU reactors.
► The results from three algorithms based on recursive prediction error method, dual extended Kalman filter, and joint extended Kalman filter are compared.
► A methodology for analyzing the data available from the in core flux detectors and extracting the unknown reactivity coefficients is presented.
► The model is useful as a diagnosis tool to gain enhanced control thereby ensuring safe operation of the plant.