Redundant Sensor Calibration and Estimation for Monitoring and Control of Nuclear Power Plants

Performance, reliability and safety of nuclear power plants depend upon validity and accuracy of sensor signals that measure plant conditions for information display, health monitoring and control [1]. Validity of measurements is important because a sensor failure can have serious consequences. Thus, it is essential to regularly ensure correct operation of sensors, in particular for those having great importance for operating safety, to locate and identify any possible degradations and faults. However, periodic maintenance strategies cause the unnecessary calibration of instruments that are operating correctly which can result in premature aging, damaged equipment, plant downtime, and improper calibration under non-service conditions. Recent studies have shown that less than 5% of process instrumentation being manually calibrated requires any correction at all. Therefore, plants are interested in monitoring sensor performance during operation and only manually calibrating the sensors that require correction [2]. Redundant sensors are often installed to generate spatially averaged time-dependent estimates of critical variables so that reliable monitoring and control of the plant are assured. For example, temperature, pressure, and flow sensors are installed with redundancy in nuclear power plants. Redundancy can be classified into two groups: direct redundant data and analytical redundant data (using mathematical models of the physical system). In practice, analytical measurements may be the only source of supplemental redundancy for detection of plant component and sensor failures. This paper presents a calibration and estimation filter for redundancy management of sensor data and analytical measurements. The filter is validated based on redundant sensor data of primary coolant temperature collected from simulator of IRIS (International Reactor Innovative and Secure) Nuclear Power Plant.