Plant reliability enhancement using plant diagnosis

In order to enhance the reliability of nuclear power plants (NPPs), while also making use of the highly reliable equipment, diagnosis techniques to detect anomalies in the equipment at early stages are of great importance. The Tohoku Electric Power Company and the Toshiba Corporation started a research and development project on plant surveillance and diagnosis, using noise analysis techniques, before the startup tests of Onagawa-1 (524 MW(e), 4 x BWR, in commercial operation since June 1984) and accumulated a valuable characteristics database on the plant and also detected several anomalies. Apart from applying noise analysis based diagnosis techniques to the actual plant, a new technique to diagnose various transient phenomena was developed with the aim of decreasing the operator's load in inspecting the cause of observed transients. This technique makes use of a qualitative pattern consisting of three values: +1, 0, -1, corresponding to increase, no change/uncertain, decrease, extracted from the observed data. The candidates for the possible source of disturbance of the transient are then selected by evaluating the similarity between the pattern of the observed data and the typical transient patterns on the database. This technique has been verified against various transient test data during plant startup. After the joint research, the Tohoku Electric Power Company introduced a computerized plant diagnosis system that reflected the research results. This system has been under operation since July 1987 at Onagawa-1 and has experienced an actual plant transient, owing to a grid disturbance, and has functioned effectively. These plant diagnosis activities are considered to contribute a great deal to the high reliability of NPPs. (author). 1 ref., 11 figs