The nuclear industry has a long history of usage of reliable and well-validated software for design and safety analysis of nuclear power generation systems. However, the majority of codes employed are based on the assumption that the flow is essentially one-dimensional, and they often provide excessively conservative predictions when three-dimensional phenomena have a significant influence.
The aims of this study are twofold:
- Furthering the understanding of three-dimensional flow patterns that may occur within the AGR fuel stringer during refuelling operations.
- Developing a numerical tool for refuelling cooling analysis capable of supporting the existing methodology in scenarios where neglecting three-dimensional phenomena would affect significantly the accuracy of the calculations.
The software used in the calculations comprised of Code_Saturne, a finite volume CFD solver, and Syrthes, a finite element solver for thermal conduction and radiation. Both solvers are open source and are developed by EDF.
To pursue the first objective, a LES of natural convection in an enclosed fuel pin bundle was carried out using the WALE model. The computational domain represented a 60 degrees sector of a 250mm long pin bundle. The roughness of the pins was neglected. The results of the calculation showed a pattern in which fluid motion takes place in thin boundary layers adjoining the active walls, separated by a stagnant core region. It has been found that the flow between the various pins resembles natural convection in a rectangular enclosure and that the local Nusselt number correlation takes the same form as that of the cavity flow.
The second line of work resulted in the development of POSTR, a numerical three-dimensional tool designed for routine usage in refuelling cooling calculations. The tool is based on the approximation of the fuel pin bundle as a porous medium to reduce the computational cost. A two-scale approach has been followed, with a coarse grid used for the calculation of the modelled quantities and a finer mesh used for the solution of the governing equations. A solver for conduction and thermal radiation in the solid components of the stringer is coupled with the CFD model for the gas, and a simple turbulence model is adopted.
A demonstration of the capabilities of the code has been performed. The predictions of the code have also been validated against experimental data experiment for scenarios encompassing forced, natural, and mixed convection conditions. The solid temperatures predicted by POSTR are found to be in good agreement with those calculated by detailed CFD calculations and by legacy software formerly used in the industry, as well as with the measurements taken in a large scale experiment. Recommendation for the usage of the tool are provided and directions for future work are suggested.
Two main original contributions of this study can be highlighted:
- To the knowledge of the author, the LES presented in this thesis is the first numerical investigation of turbulent natural convection in an enclosed bundle of concentrically arranged pins. The observations described in the thesis could provide useful guidance for further investigation on the topic.
- The hybrid approach between CFD and sub-channel analysis introduced in this thesis, based on the use in the calculation of two meshes at two different scales has been adopted in subsequent research projects on coarse-grid CFD, carried out at the Heat, Flow and Turbulence Research Group of the University of Sheffield.
