Abstract
Instrumentation in a nuclear power plant is critical in monitoring the stability and
safety levels of a reactor. Temperature is a key measurement performed on the
core of a reactor to control the power output and sustain a safe thermal margin.
If there is a dramatic change in temperature, failure is likely to follow if action
is not taken to cool the system. Traditionally, to measure the temperature of a
reactor several resistance temperature detectors (RTD)s were placed in prede ned
locations on the system. RTDs are typically coiled platinum wire wrapped around
a ceramic cylinder and encased in a metal sheath. Due to the harsh environment of
nuclear reactors, RTDs degrade and their resistance measurements drift over time.
This drift in resistance can be misunderstood as a drift in reactor temperature.
In the past, RTDs would be serviced every few years either through calibration or
replacement. To service RTDs the reactor is shut down, and a person is sent into
a dangerous environment. Using Johnson noise thermometry (JNT) will reduce the
need for servicing and provide a high-accuracy temperature measurement. JNT is
a fundamental expression of temperature that is invulnerable to drift in the RTD's
physical condition. The signal processing behind JNT is presented in this document.
Spectral estimation methods are used to remove electromagnetic interference (EMI)
from the JNT measurement. These methods are unique to this dissertation. The EMI
estimation method is modeled and simulation results are presented. The modeling
of the EMI estimation involves locating EMI, analyzing EMI e ects, and removing
EMI without bias. Finally, results from numerical and experimental veri cation are presented. The research presented here is important to the future of the nuclear
industry for several reasons. With this technology applied to existing systems, reactor
shut down time can be decreased, technician exposure to dangerous radiation zones
is decreased, and the cost associated with lengthy shut downs can be reduced. The
instrumentation community will bene t through the innovation of signal processing
for very small signal versus noise interference.
