Sequential Probability Ratio Test for Nuclear Plant Component Surveillance

It is common practice in nuclear power plants to install redundant sensors to monitor critical physical variables such as pressures, temperatures, and radiation levels. The design and testing of an extremely sensitive component-operability surveillance algorithm based on the sequential probability ratio test (SPRT) are reported. The SPRT technique processes the stochastic components of digitized signals from identical sensors on two or more components in an operating reactor for the detection and annunciation of off-normal operation. Information from the SPRT can provide a reactor operator with early identification of conditions that could lead to plant operational degradation, thus enabling him or her to terminate or avoid events that might challenge safety or radiological performance guidelines. The SPRT enhances plant availability and economics by minimizing unnecessary reactor trips caused in conventional systems by occasional spurious data that might exceed a simple high/low limit check. An example application of the SPRT for the surveillance of primary coolant pump operability in the Experimental Breeder Reactor II is presented.