Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of GraphSearch.
Nuclear reactors are inherently stochastic systems, in which neutronic and thermal-hydraulic parameters fluctuate continuously even during steady-state conditions. In addition, structural components vibrate due to the coolant hydraulic forces. This stochasticity is the cause of the neutron population fluctuating behavior, a phenomenon referred as neutron noise. The neutron noise is monitored over the reactor lifetime, providing valuable knowledge of the core behavior. More importantly, the neutron noise monitoring is used for the early detection of reactor anomalies. The neutron noise phenomenon is an intensively-studied research topic, leading the development of noise surveillance methods, signal processing techniques and analytical solvers. Nevertheless, an unexpected neutron noise level increase trend, observed in the last decade in several reactors causing undesirable costly operational consequences, triggered an increasing interest. The new observations revealed the need for an improved neutron noise behavior understanding. In this context, the current doctoral research intents to systematically analyze the neutron noise phenomenon by developing innovative noise modelling methodologies. First, main attention is given to the neutron noise characteristics identification of the Swiss Gösgen nuclear reactor (KKG). KKG plant data are analyzed in the time and frequency domains using traditional signal processing techniques, and key aspects of the KKG noise phenomenology are revealed, allowing the better characterization of the noise sources affecting its dynamic behavior. Then, a neutron noise modelling methodology is developed, utilizing advanced neutronic solvers. These codes are used to systematically model key noise sources (i.e. fuel assembly vibration, and inlet coolant temperature and flow fluctuations), and to study their impact on the neutron noise. For the first time, the fuel assembly vibration model in the utilized codes is systematically studied and is qualified at a 3D full core level. This detailed work, demonstrate the capabilities and the robustness of the newly developed PSI neutron noise methodology to successfully model key noise sources and to reproduce neutron noise phenomena. In addition, a comparative study between the simulated results and the KKG measured data revealed that, the stochastic fluctuation of the inlet coolant temperature in combination with the fuel assembly vibration have a primary role on the KKG neutron noise behavior. Most importantly, it is observed that, the neutron noise increase trend observed in KKG could be explained, at some extent, by the introduction in the core of a newer fuel design which is more susceptible in lateral vibrations. Last, a new in-house methodology is established serving as a supportive diagnostic tool for the detailed identification of signals connectivity patterns. To this aim, the PSI connectivity analysis methodology is based on the causality analysis principles to indicate the cause-and-effect interactions between reactor signals. The nuclear core is analyzed utilizing the most prominent causality analysis techniques in the frequency domain (i.e. rPDC and DTF). The current research is the first application of causality analysis techniques on nuclear reactor data. The application of the developed methodology in measured and simulated datasets showed that, it can successfully indicate the causal interconnections and the perturbation root-cause.
Andreas Pautz, Hakim Ferroukhi, Gaëtan Girardin, Abdelhamid Dokhane, Dionysios Chionis