CFD-analysis of buoyancy-driven flow inside a cooling pipe system attached to a reactor pressure vessel

Sammanfattning: In this work a cooling system connected to a reactor pressure vessel has been studied using the CFD method for the purpose of investigating the strengths and shortcomings of using CFD as a tool in similar fluid flow problems within nuclear power plants. The cooling system is used to transport water of 288K (15°C) into a nuclear reactor vessel filled with water of about 555K (282°C) during certain operating scenarios. After the system has been used, the warm water inside the vessel will be carried into the cooling system by buoyancy forces. It was of interest to investigate how quickly the warm water moves into the cooling system and how the temperature field of the water changes over time. Using the open source CFD code OpenFOAM 2.3.x and the LES turbulence modelling method, a certain operating scenario of the cooling system was simulated. A simplified computational domain was created to represent the geometries of the downcomer region within the reactor pressure vessel and the pipe structure of the cooling system. Boundary conditions and other domain properties were chosen and motivated to represent the real scenario as good as possible. For the geometry, four computational grids of different sizes and design were generated. Three of these were generated using the ANSA pre-processing tool, and they all have the same general structure only with different cell sizes. The fourth grid was made by the OpenFOAM application snappyHexMesh, which automatically creates the volume mesh with little user input. It was found that for the case at hand, the different computational grids produced roughly the same results despite the number of cells ranging from 0,14M to 3,2M. A major difference between the simulations was the maximum size of the time steps which ranged from 0,3ms for the finest ANSA mesh to 2ms for the snappy mesh, a difference which has a large impact on the total time consumption of the simulations. Furthermore, a comparison of the CFD results was made with those of a simpler 1D thermal hydraulic code, Relap5. The difference in time consumption between the two analyses were of course large and it was found that although the CFD analysis provided more detailed information about the flow field, the cheaper 1D analysis managed to capture the important phenomena for this particular case. However, it cannot be guaranteed that the 1D analysis is sufficient for all similar flow scenarios as it may not always be able to sufficiently capture phenomena such as thermal shocks and sharp temperature gradients in the fluid. Regardless of whether the CFD method or a simpler analysis is used, conservativeness in the flow simulation results needs to be ensured. If the simplifications introduced in the computational models cannot be proved to always give conservative results, the final simulation results need to be modified to ensure conservativeness although no such modifications were made in this work.