Autoregressive Multivariate Analysis of BWR Bistable Flow

Abstract

Bi-stable flow patterns can induce flow oscillations possibly leading to power fluctuations. In other cases, bi-stable flow can generate loads on components, so that structural stresses may become a potential cause of failure. This type of flow occurs in BWRs at different operation conditions, so there is no an absolute methodology for detection and prediction of such phenomenon. In this work, a multivariate autoregressive (MAR) analysis is performed to different signals related to a bi-stable flow event that occurred in one of the BWR Units at the Laguna Verde Nuclear Power Plant. The signal analysis was performed with the home-developed NOISE computer program, which, among several other applications, computes the autoregressive coefficients which contain the information of the dynamics of the signal, and that later are used to determine the relative power contribution (RPC) ratio, which in turn allows establishing the influence of the different signals on each other. From the signal analysis, among the important results obtained, it was found that no new frequencies appeared during the event. Also, it was determined through the Relative Power Contribution ratios that the most probable cause of reactor power change was the flow variation in the recirculation flow of loop B. Maximum variations (both above and below) from the initial average reactor power were 0.5%, so the bi-stable flow impact was of no safety concern.