Diagnosis and decision aided for the monitoring of nuclear reactor systems

Abstract

One of the major challenges in instrumentation is to identify wrong data (signal) measurements and perform their validation. This can be done by regularly ensuring a correct operation of the different process components, particularly those having great importance for safety, in order to detect, isolate and identify any possible degradation or fault. The fault monitoring, considered as part of fault supervision, is composed mainly of two principal functions: fault detection and diagnosis (diagnostic). On the other hand, diagnosis is composed of several functions principally: isolation, identification and localization. The operation of on line monitoring should be done as early as possible, before any fault causes failure in equipment which can lead to the downtime of the plant and even to severe catastrophes and disasters. Thus, the early indication of faults in these systems becomes highly crucial due to the negative consequences since it provides early warning to operators and gives enough information and time to take corrective or decisive actions. Consequently, if process faults are not well monitored, they cause a serious impact on process operation as the increase of the down time and the incorrect control actions. Therefore, these consequences influence negatively on productivity, availability and environment. Due to the complexity and size of current industrial systems, the operators (decision-makers) are brought to treat (manipulate) volumes of more and more considerable information, what leads to monitor an increasing number of variable and make so difficult the work of the operators. Therefore, the conception of a system of supervision coupled with a tool of help (assistant) in the decision seems important. At Triga-Mark II (Training Research and Isotope Production General Atomic) nuclear research reactor, the heat exchangers are provided for removing generated heat from the reactor pool water throw cooling circuits. Therefore, the monitoring of the evolution of its thermal hydraulic parameters is necessary to ensure the safety of the reactor. Among several developed techniques, analytical redundancy has been recognized as an effective method for fault monitoring. It is the process of identifying a faulty instrument in a system through a comparison of its output to an estimate data. This estimation is based on the model and the measurements provided by the data acquisition channels of the existing sensors during all the operating modes of the installation. The aim of this thesis is to monitor and accommodate some parameters of the core and the heat exchanger of Triga-Mark II nuclear research reactor at LENA (Laboratory of Nuclear Applications), since these systems are the most commonly monitored. We underline the theory on which the monitoring approach, analytical redundancy proposed in this thesis are based. So, the main motivation for this research is to exploit the potential of artificial intelligence and physics relationships to design faulty free model behaviors and to generate residuals for systems to be monitored. In this thesis we review the theory of the supervision of fault (i.e., fault detection, diagnosis, and accommodation) including the different methods used in this domain. In addition, we present a comparative result by using different mathematical models, and Kalman filter and artificial neural networks approaches for the monitoring and accommodation of some parameter of the core and heat exchanger in Triga-Mark II research reactor.