Reference model parameter identification of space-time dependent reactivity in a CANDU-PHWR

The problem of estimating reactivity transients from an observed neutron flux transient is considered. This is relevant, for example, to analyzing a power rundown test or to estimating reactivity variations associated with some computer codes that do not specifically compute individual reactivity components. A method is presented which utilizes inverse space-time kinetics and optimal state estimators to extract the components of the reactivity transient from observed neutron flux measurements. The approach takes into account geometric characteristics and composition of the reactor core, as well as reactor operating conditions. Measurements from a limited number of in-core neutron flux detectors are the inputs used to extract reactivity components that fit a modal model of the reactor, referred to as the 'reference model'. An improved solution for the reactivity components is then generated using the modal approximation solution for the neutron transport equation in conjunction with optimal estimation techniques. The method has been applied to a reactivity initiated accident in which a transient is initiated by a non-uniform loss-of-coolant. This results in a spatially varying neutron overpower transient that is terminated by the asymmetric insertion of shutoff rods. In this paper the Joint Extended Kalman Filter and Rauch-Tung-Striebel smoother is employed to estimate the neutron flux distribution in the core and identify the reactivity components of the reference model. The reference model in the state space and the Kalman filter algorithm are shown. Results of numerical simulations of the reactor transient and the optimal estimation of the reactivity components are presented to demonstrate the capabilities of the method