Probabilistic techniques for predicting reactor coolant pump motor failure rates

The Calvert Cliffs Nuclear Plant, a two-unit pressurized water reactor nuclear station located on the western shore of the Chesapeake Bay in Maryland, employs 13-kV motors to drive each primary system reactor coolant pump. When a motor start is initiated, protection of the stator windings during breaker closure is achieved by the use of surge capacitors. These capacitors are located at the motor in the harsh environment of the containment building. During normal plant operation, there have been several capacitor failures, each of which led to a reactor plant trip. In addition to these capacitors proving unreliable in service, the current supplier has indicated that in the future, procurement of replacements will not be possible. This led to the necessity for defining a strategy to resolve these problems. The following options for remedial action were to be included: (1) replacement of the surge capacitor protection system with inductors located at the switchgear (this is a mild environment and easily accessible), and (2) complete removal of the surge protection system. The latter option appeared to be the more attractive, but clearly could not be proposed until its impact on plant availability and the resultant economic risk could be assessed. The remainder of this paper describes how probabilistic techniques proved useful in performing this assessment