Boiling water reactor stability analysis by stochastic transfer function identification

The univariate and the bivariate ARMA models are proposed as the stochastic transfer function models for the identification of BWR systems. This technique has been developed as a new method for on-line system identification, optimum control, and malfunction monitoring of nuclear power plants. The relationships between the stochastic transfer function model and the differential equation model are derived. The estimation algorithms are developed through the related covariance functions and Green's function by the least squares method. It has been shown that the stochastic models can also be used for fitting the stochastic data which are contaminated with sinusoidal waves. Both the univariate and the bivariate modeling are applied in the BWR system identification and stability analysis. The univariate modeling is applied to decompose the pressure dynamics from the neutron data. From both of the normal operation data and the perturbation experiment data, the reactor dynamics are consistently estimated. The dynamics of the reactor core are estimated as a second order mode with a natural frequency of 0.4 Hz and a damping ratio of 0.1. The univariate modeling is also applied to monitor the local performance of the coolant channel in the reactor. The transfer functions between system's variables are obtained by use of bivariate modeling. The obtained transfer functions are closely related to the stability analysis of thermal-hydraulics in the reactor. The transition of the system dynamics from normal operation to the perturbation experiment are observed