Performance of core thermocouples of FBTR – A technique for predicting accurate temperature for known transients in the presence of measurement delays

Abstract

Abstract FBTR is a 40 MWt research reactor, located at Kalpakkam, near Chennai, India. The reactor has been in operation since 1985 and has seen a maximum operating power of 30.2 MWt till July 2018. The fuel sub-assemblies (FSA) are located vertically in the grid plate along with other types of sub-assemblies namely reflectors and blanket sub-assemblies. Sodium enters the sub-assemblies from the bottom and exits through the top. There are 84 positions where thermocouples are provided to measure the temperature of the sodium jet passing out from the FSA’s. The mineral-insulated metal-sheathed ANSI Type K thermocouples are exposed to intense neutron fluence as well as gamma dosage. In this paper, the performance of the core thermocouples since the first criticality of FBTR has been comprehensively presented. A few problems, including failure of bare thermocouples (installed without thermo-wells), presence of normal-mode electrical noise in the signals etc., have been faced in FBTR. Suitable remedial actions have been taken for overcoming the same. Since the hot junctions of these thermocouples are inaccessible after installation, in-situ cross-calibration techniques have been employed for verifying their calibration. Using the time-domain data available during the safety actions such as simultaneous lowering-of-rods (LOR) and the dropping of all the six control rods (SCRAM), the in-situ process time-constants involved in core temperature measurement have been estimated. One of the problems faced in FBTR is that, subsequent to a reactor shut-down by LOR, the measured value of the sodium temperature difference across certain FSA’s rise above the expected value, leading to initiation of faster safety action (that is, reactor SCRAM), resulting in unwanted thermal shocks to the core components. This necessitated the need to analyze the transient behavior of the core during LOR, including the response of the core thermocouples. The whole measurement process was modeled, and with that, the actual temperature rise across the FSA’s could be predicted quite accurately from the observed temperatures which are influenced by both process and measurement delays. Using the predicted behavior, it could be established that the onset of unwanted reactor SCRAM can be considerably delayed, thereby minimizing the thermal shocks to the core components.