This thesis studies the cooling of rod bundles within the Advanced Gas-cooled Reactor (AGR) fuel route at non-design conditions using a variety of methods. The aims of the thesis are (i) to contribute to the general understanding of the detailed flow, heat transfer, and
turbulence phenomena in AGR rod bundles. (ii) to develop a 3-D porous model software package for the thermal-hydraulics analysis of the fuel route. Of primary concern to this project are scenarios where the fuel bundle is distorted as a result of being dropped or damaged during refuelling operations. The model developed herein will complement the current 1-D thermal codes in use at EDF Energy. This is particularly for the cases where the latter would be excessively pessimistic or inaccurate due to their inability to capture the 3-D characteristics of the flow.
The open-source, co-located, and segregated Computational Fluid Dynamics (CFD) solver Code Saturne developed by EDF has been used. For the
first aim, three studies have been carried out, that is, (a) Large Eddy Simulation (LES) study of natural circulation in a short 0.25 m enclosed bundle (b) Large Eddy
Simulation study of natural circulation in a 1 m tall enclosed bundle and (c) Reynolds Averaged Navier Stokes (RANS) study of forced convection in a damaged bundle. In a short bundle, (a) the flow is largely laminar and constrained to the thin boundary layers around the fuel rods and containment wall. Away from the walls, in the core, the flow is stagnant. The vertical temperature distribution is heavily stratified. The natural circulation flow in the 1 m domain is heavily influenced by a vertically developing boundary layer on the containment surface, which is initially laminar but transitions to turbulence at about a quarter of the height from the top. The Nusselt number on the containment wall can be correlated using a well established expression over a vertical plate in a free space. Laminar boundary layers observed in both the long and short domains compare very well with similarity solutions, though for those over the fuel rods, the curvature needs to be considered. Forced convection in a damaged WheatSheaf bundle shows the flow to swirl around the rods as it is diverted to regions of less resistance through the rod gaps. Hot spots on the fuel at any axial location are found on the leeward side of the cross-flow. To fulfil the requirements of the second aim a thermal-hydraulics code for the fuel route, named FREEDOM has been developed. FREEDOM aims to predict AGR fuel component temperatures under potential fault conditions while the fuel is being handled or stored within the AGR fuel route. FREEDOM has two modes, one for intact and another for damaged fuel. The main focus of this thesis is on damaged fuel. The model comprises of two domains, the fluid domain computed using Code Saturne and the solid domain computed using Syrthes. In the fluid domain, the porous media representation is used to simplify the mesh generation and lessen computation cost. Thermal conduction and radiation are associated with the solid domain. The two domains are coupled together through the exchange of temperatures and heat transfer coefficients. In addition, oxidation due to the fuel and carbon deposit have been modelled considering both diffusion and reaction dynamics controlled conditions. The code is validated by performing experimental and code-to-code comparisons for a variety of flow conditions and idealised geometries. Forced convection comparisons were in good agreement and natural convection comparisons ranged from good to acceptable. The validated FREEDOM has then been successfully used to support a safety case argument by taking into account the three-dimensional effects for the flow, radiation, and solid conduction.
