Using the Hurst's exponent as a monitor and predictor of BWR reactor instabilities

Since the decade of the 1950s, when the development of boiling water reactor technology began, unstable situations have existed, which involve a high amplitude self-oscillatory process in the reactor's thermal power. As the development progressed and the reactors increased power density, the possibility of instability under certain circumstances increased. Thus, in 1985, Caorso nuclear plant (Italy) reported the first event of this type, and in 1988 (NRC. 1988), such an event was reported at La Salle as well. Since then, multiple instability events have been reported. The danger of these unstable power situations resides in the possibility of exceeding a thermal limit, as expressed in Appendix A of 10FR50. Thus, the need arises to monitor and correct these situations in the industry. The most common way to monitor and control these instability situations involves the use of Decay Ratio (DR) and Resonance Frequency (RF) (Verdu et al., 2001; Montesinos et al., 2003). The use of these parameters is polemical, because their use involves certain simplifications and operations prior to the calculation which question how well they represent the reality. The most important simplifications are those which lead to the interpretation of the power time series as the result of a second order system. With regard to the previous operations, the time series needs to be standardized and filtered. The result is loss of information during prediction, due to the operations, and the results, therefore, lack accuracy. In this paper, the system is considered without simplifications, that is to say that it is treated as dynamic and, as we shall see, chaotic, in the mathematical sense of the term (Mandelbrot, 1982). The series will be used in pure form without manipulations. The parameter used for monitoring and prediction of the core's behaviour will be the Hurst's exponent (H) Hurst, 1957. The concept used for this proposal is that the response of a complex dynamic system depends not only on the last excitation, but on the prior history. The processes and systems are catalogued in three forms: persistent (H > 0.5), neutral (H = 0.5) and anti-persistent or unstable (H < 0.5). Thus, the instability of the core will occur when H is under 0.5 and the prediction shall be made in relation to the evolution of H over time, that is to say, evaluating its tendency. (C) 2009 Elsevier Ltd. All rights reserved.