Detection of Sensor Abnormalities in a Pressurizer by Means of Analytical Redundancy

To improve the reliability of pressurizer level measurement channels against sensor common cause failures (CCFs), a sensor fault detection scheme has been developed based on analytical redundancy in this paper. The fault detection mechanism has been transformed into statistical tests on a Kalman filter generated innovation sequence. An exponentially weighted moving average (EWMA) scheme has been developed to work with other sensor channels to ensure that the abnormality can be detected and isolated as quickly as possible. Because analytical redundancy does not share the same failure modes as existing sensor channels, the overall reliability of the measurement system has been improved. The paper provides a detailed procedure for formulating analytical redundancy, generating the corresponding innovation sequence, and detailing the design and implementation of the proposed detection scheme. Finally, the performance of the proposed scheme has been evaluated in a CANDU (Canada deuterium uranium) pressurizer to demonstrate its potential benefits. The speed of the fault detection is also evaluated against that of the average run length of the Shewhart control chart. It can be concluded that the proposed scheme is significantly more sensitive to sensor abnormalities at an earlier stage in their development. Even though level measurement channels in a pressurizer are considered, the methodology and design techniques can easily be extended to other systems/applications.