Time Series Analysis for BWR Stability Studies

Analysis of the power time series evolution is used to investigate a stable or unstable process after the disturbance in a light water reactor of the boiling water reactor (BWR) type. Several different methodologies are currently used and the uncertainties of the various approaches are in some cases very different. In this work, the time series model known as the Autoregressive Moving Average model was used to calculate the decay ratio (DR), and the natural frequency (NF) due to power oscillations in a BWR. The method consists of locating the appropriate dominant pole of the transfer function. The autoregressive methods are quite often used to study the stability of BWR reactors. In this work the Box-Cox transformation is implemented to stabilize the variances of the power signals in order to maintain the linear assumptions that the calculation of DR and NF needs; that is, to correct biases in the distribution of errors to stabilize the variance and mainly so that the signal approaches a linear behavior. The MATLAB code was used for this purpose. This work also presents a nonlinear analysis of the power series, determining the values of the largest Lyapunov exponents with Rosenstein’s algorithm in order to analyze the stability of the system. The results of the DR and NF calculated by the used methodology are very close to the values obtained in the benchmark.