Analysis of Log Rate Noise in Ontario's CANDU Reactors

Abstract

In the fall of 2003, the operators noticed that in the recently-refurbished Bruce A Shutdown System #1 (SDS1) the noise level in Log Rate signals were much larger than before. At the request of the Canadian Nuclear Safety Commission (CNSC), all Canadian CANDU reactors took action to characterize their Log Rate noise. Staff of the Inspection and Maintenance Services division of Ontario Power Generation (OPG) has collected high-speed high-accuracy noise data from nearly all 16 Ontario reactors, either as part of routine measurements before planned outages or as a dedicated noise recording. This paper gives the results of examining a suitable subset of this data, with respect to the characteristics and possible causes of Log Rate noise. The reactor and instrumentation design is different at each station: the locations of the moderator injection nozzles, the location of the ion chambers for each system, and the design of the Log Rate amplifiers. It was found that the Log noise (source of Log Rate noise) was much larger for those ion chambers in the path of the moderator injection nozzles, compared to those which were not in the path. This "extra" Log noise would then be either attenuated or amplified depending on the transfer function (time constants) of the Log Rate amplifier. It was also observed that most of the Log and Log Rate noise is independent of any other signal measured. Although all CANDU reactors in Ontario have Log and Log Rate noise, the Bruce A SDS1 system has the largest amount of Log Rate noise, because (a) its SDS1 (and RRS) ion chambers are at the top of the reactor in the path of the moderator injection nozzles, and (b) its SDS1 Log Rate amplifiers have the smallest time constants.