Application of the Monte Carlo Method for Capacitive Pressure Transmitters Surveillance in Nuclear Power Plants

Abstract

In nuclear power plants (NPPs), according to current regulations, the response time of capacitive pressure transmitters is used as an index for surveillance. Such measurement can be carried out in situ applying the noise analysis techniques to the sensor output signal. The method is well established, and it is based on the autoregressive (AR) fitting optimized by the Akaike criterion (AIC). The sensor response is influenced by the sensing line, and its length is different in each plant. Recent empirical research has proved that the sensor inner structure can be modeled with a two real poles transfer function. In the present work, it has been proved that the noise analysis applied to the simulated response of a transmitter, modeled with two poles coupled with a sensing line, gives erroneous values for the ramp time delay when the sensing line is long. Specifically, the order of the AR model supplied by the Akaike criterion is not appropriate. Therefore, a Monte Carlo method is proposed to be applied in order to establish a new criterion, based on the statistical analysis of the repeatability of the ramp time delay obtained with the AR model.