The flow rate distribution at the entrance of the core plays a key role for reactor design since it has important implications for the performance, and efficient safety of a nuclear reactor. When the coolant passes from the downcomer to the core, it changes direction due to the inertia force and the curvature of the bottom vessel head. The internal components inside lower plenum work to homogenize the flow distribution. Their purpose is to prevent the formation of instabilities and the creation of vortices due to the flow reversal. In the frame of EDF’s new reactor design there is a desire to identify an optimal flow diffuser. The future intention is to study five different types of flow diffuser including EPR, VVER, Konvoï, APR+ and Westinghouse to look at the pros and cons of each design. The authors underline that the geometries of each Reactor Pressure Vessel (RPV) and associated diffuser device are quite different therefore a generic form needs to be used to make an equivalent comparison. The goal of the present work is to find the optimal mesh refinement and associated numerical parameters for the simulation of the lower plenum flow. This work is a preliminary step for a future study to compare existing diffuser concepts. Thus in the future work only the section containing the flow diffuser structure will be changed.
The PIRT methodology is applied to better define the physical phenomena and key parameters that will influence the flow distribution at the entrance of the core. In order to better understand the fluid distribution and the function of the diffuser component, 3D computational fluid dynamics (CFD) simulations are launched to improve our knowledge on the flow pattern inside the lower plenum.
Both the geometry and mesh are generated by Salomé1. Simulations are carried out using Code_Saturne2, an EDF in-house open-source CFD code. The generic test case is a 1/5 scale EDF “BORA” 4 loop mock-up with a flow rate of 0.1 m3/s injected into each cold leg. The unsteady flow algorithm with standard k-epsilon turbulence model has been used with a full explicit meshing except for the reactor core where a porous approach is adopted. The physical time for each calculation case is 5s for a converged simulation. Mesh sensitivity tests have been carried out ranging from 8 million cells to 28 million cells. A mesh of 22 million cells is found to provide the most appropriate balance between simulation quality and feasibility. Due to the size of the simulations, high performance computers are necessary to provide timely results. The results indicate that CFD can provide extra capacity to engineers for reactor design to evaluate the pros and cons of different existing diffuser concepts.