Application of Autoregressive Models to In-Service Estimation of Transient Response for LMFBR Process Instrumentation

The response of a sensor depends on its operating conditions, and thus it is desirable to develop an in-service method for response time estimation. The applicability of the autoregressive (AR) model for this purpose was examined in the case of the fuel subassembly outlet coolant thermocouples and the primary circuit electromagnetic flowmeter (EMF) of Monju, the prototype fast breeder reactor in Japan.

The use of an AR model with exogenous input (ARX model) is possible when the physical variable to be sensed can be observed by an alternative means with a faster response time than that of the sensor in question. In the case of the subassembly outlet thermocouple, the temperature output from an eddy-current sensor, during pseudorandom reactor power variation, served as the exogenous input.

In respect to the thermocouple response, AR and ARX modeling were shown to be applicable, and the transient responses thus derived agreed well with each other and with the results measured by means of a step change in sodium temperature.

However, the primary circuit EMF response time, estimated using the AR model, decreased with increasing flow rate even when approaching the rated flow, demonstrating that the method was not completely applicable. Nevertheless, it can be concluded that the response is faster than that estimated in the rated condition.