Performance improvement in temperature thermometry using data analytics - A nuclear power plant perspective

The power density of the nuclear reactor core is high. In the event of inadequate cooling of Fuel subassembly
(FSA), the fuel-clad temperature can increase to a high value leading to clad rupture. Hence, it is very important
to detect core anomalies with adequate and suitable instrumentation. It is also required to monitor continuously
the adequacy of reactor core cooling and initiate suitable safety actions in case of any abnormality.
K-Type Thermocouples are normally used for measuring the core temperature. Computer based Core
temperature-monitoring system is provided for acquiring the temperature signals, detection of core anomalies,
error in core loading, fuel enrichment error and fuel orifice error. The control system monitor the temperature
and provides signals for safety system to initiate protective action in case of incidents. These safety actions
prevent the clad hot spot and fuel temperature from reaching the design safety limits.
Hence, it is necessary to factor in the performance uncertainties of the safety instrumentation so that the
process is operated efficiently within the limiting conditions of operation, limiting safety settings and the design
safety limits. The errors and uncertainties of the thermocouples, signal processing electronics and the display

systems shall be quantitatively estimated to implement an efficient operation strategy and optimize the re-
quirements of calibration & ageing management.

This paper deals with application of data analysis techniques to analyze the performance characteristics of

core temperature monitoring thermocouples, estimate errors and suggest correction factors so that the re-quirements of availability and reliability are met within the actual safety limits for the system in a nuclear powerreactor. The challenges in electronic calibration of measurement channels are discussed in view of distributedlocations. The uncertainties and errors in thermocouple sensors are analysed using data analytics models and anintegrated methodology of compensating for overall uncertainties in the system is proposed based on the