Thermo-hydraulic study of the flow of the moderator in the CANDU-6 reactor

Considering the size (6 mx7.6 m) and the quite complex hydrodynamic domain of CANDU reactors (i.e. 380 fuel channels embedded in the vessel), the motion of the the moderator flow in the calandria is still not perfectly understood until today. Data sampling process from a reactor in operation necessitates several modifications on the calandria vessel to get access to the core. Furthermore, the intense radiation area in the reactor core requires the use of quite particular measurement instruments. Consequently, the moderator flow must necessarily be studied through experimental and/or numerical models. Concerning experimental tests, fabrication and operation of such a huge facility is very expensive. Moreover, the scaling groups of CANDU reactors are in contradiction. Consequently, numerical simulations constitute an appealing. Because of the aforementioned constraints (i.e. complexity of the domain), the porous medium method has been used to study the moderator flow by the nuclear industry. This approach approximates the hydraulic domain as a continuous medium where, the tube bundle is replaced by hydraulic resistances. This model has the ability to predict macroscopic characteristics of the flow motion. However, it can not simulate the local effects such as temperature and velocity distributions near the calandria tubes. In the framework of nuclear safety, local effects around the calandria tubes are a point of high interest. Predictions of porous medium method previously showed that the moderator flow can develop several hydrodynamic configurations. For some, an asymmetric behaviour in calandria vessel was observed. This can result in boiling of the moderator flow at proximity of fuel channels. In such conditions, the reactivity coefficient can change significantly, resulting in variation of the reactor power. For this reason, a CFD model that takes into account the local effects is therefore necessary to study the motion of the moderator in the calandria. The purpose of this research is to model the complex behaviour of the flow moderator in CANDU-6 calandria, especially at proximity of calandria tubes. These simulations are used to identify possible flow configurations. This study is thus to formulate theoretical bases on behalf of macroscopic instabilities in the moderator, i.e. asymmetric motions which may cause boiling. The project's challenge was to determine the impact of these flow configurations on the reactivity of CANDU reactor. The reactor vessel's huge dimensions in conjunction with density variations make the simulations pretty tricky and costly. The previous CFD analyses do not coincide with the experimental data available in the literature, most likely due to not-converged results and/or use of inappropriate physical models which are not valid for this modelling. The presented numerical study identifies hydrodynamic instabilities that lead to oscillations in the system. This is the first time that this type of motion is predicted in the moderator flow. Indeed, such an asymmetry must be considered in the nuclear safety analysis. Simulations of this project were carried out using Code_Saturne finite volume solver, developed by electricite de France. At first, this code has been validated with experimental data. An important contribution of this thesis is the development of a moderator flow operating map. The predictions obtained from a transient model have allowed us to identify the oscillations frequencies related to each of the possible configurations of the moderator flow. In the next step, the predictions of the thermal hydraulic simulations were coupled with DONJON reactor physics solver. The coupling study revealed important features for reactor's reactivity by taking into account the moderator flow distributions. (author)