Thermal stripping is a major safety challenge in nuclear power generation and propulsion systems. It arises as a consequence of the heat transfer from fluid to surrounding solid components varying in time and typically occurs in regions where the mixing of hot and cold fluids results in turbulent temperature fluctuations. It can occur in a range of components in reactors and thermal-hydraulics systems and may lead to structural failure by high-cycle thermal fatigue. Cases of cooling system pipes failing by this mechanism have been reported at the French Civaux and the Japanese Tsuruga-2 & Tomari-2 pressurised water reactor plants. CFD has great potential to provide predictions for flow fields in the pipe bends and junctions of nuclear plant thermal-hydraulics systems. The current project aims to use CFD to explore the physics of thermal mixing in plant components, and to develop \& validate CFD techniques for studying such problems in industry. Firstly, wall-resolved LES is used to demonstrate the importance of including nearby upstream pipe bends in CFD studies of thermal mixing in T-junctions. Swirl-switching of the Dean vortices generated at an upstream bend can give rise to an unsteady secondary flow about the pipe axis. This provides an additional mechanism for low-frequency near-wall temperature fluctuations downstream of the T-junction, over those that would be produced by mixing in the same T-junction with straight inlets. Wall-resolved LES is however currently computationally unaffordable for studying plant components in industry. Wall-functions offer a solution to this problem by imposing empirical results near walls, such that a coarser grid can be used. LES with blended wall-function predictions for flows in a 90 degree pipe bend and a simple T-junction with straight inlets are compared to experimental data. These studies highlight limitations in the predictive capabilities of the LES with wall-function approach. Predictions from a number of RANS models are also benchmarked. Finally, the consistent dual-mesh hybrid LES/RANS framework proposed by Xiao and Jenny (2012) is further developed as an alternative solution to the high computational cost of wall-resolved LES. Numerous modifications to the coupling between the two meshes are presented, which improve automation and accuracy. The approach is also extended to a passive temperature scalar field. Predictions for channel flows, a flow through periodic hills and thermal mixing in a T-junction between channel flows are all in excellent agreement with reference data.

The aim of this study is to compare the performance of a linear model to a nonlinear model focusing on flow separation based on a critical slope value. Specifically, the WindPRO WAsP model will be compared with the WindSIM CFD model over a simulated terrain to determine the point the two models differ in relation to the inclination of the terrain. The results of this study will verify if the proposed critical slope value of roughly 17 degrees is truly representative of the limitation of the WAsP model in producing accurate results as compared to a CFD model.  Multiple similar studies have been performed using existing sites with actual met mast data as a comparison to the model outputs. Many of these cases have come up with varying results due primarily to the large number of uncontrolled factors influencing the data. This study will be designed in a fully simulated environment where all variables can be controlled, allowing for the manipulation of a single variable to understand its’ specific influence over the model. The primary variable being tested in this study will be the slope of the terrain with all other factors held constant.   Based on the outcome of 7 alternative runs with ridge heights of 100, 120, 140, 160, 180, 200, and 300 meters and respective maximum slope values of 10.31, 12.32, 14.29, 16.23, 18.14, 20, and 28.63 degrees a defined separation point at a hub height of 94 meters could not be found. Each run demonstrated correlation between wind speeds and terrain slope variations but a considerable difference in estimated wind resources was present between the linear and non-linear CFD models where any slope in terrain is present. This, as expected, increases where terrain inclination increases, but a clearly defined difference between the two models is not evident at the previously established critical slope value of approximately 17 degrees (30%).

A fully coupled system in which a vehicle s aerodynamic and handling responses can be simulated has been designed and evaluated using a severe crosswind test. Simulations of this type provide vehicle manufacturers with a useful alternative to on road tests, which are usually performed at a late stage in the development process with a proto- type vehicle. The proposed simulations could be performed much earlier and help to identify and resolve any aerodynamic sensitivities and safety concerns before significant resources are place in the design. It was shown that for the simulation of an artificial, on-track crosswind event, the use of the fully coupled system was unnecessary. A simplified, one-way coupled system, in which there is no feedback from the vehicle s dynamics to the aerodynamic simulation was sufficient in order to capture the vehicle s path deviation. The realistic properties of the vehicle and accurately calibrated driver model prevented any large attitude changes whilst immersed in the gust, from which variations to the aerodynamics could arise. It was suggested that this system may be more suited to other vehicle geometries more sensitive to yaw motions or applications where a high positional accuracy of the vehicle is required.

The study of magnetohydrodynamics unifies the fields of fluid mechanics and electrodynamics to describe the interactions between magnetic fields and electrically conducting fluids. Flows described by magnetohydrodynamics form a significant aspect in a wide range of engineering applications, from the liquid metal blankets designed to surround and remove heat from nuclear fusion reactors, to the delivery and guidance of nanoparticles in magnetic targeted drug delivery. The ability to optimize these, and other, processes is increasingly reliant on the accuracy and stability of the numerical models used to predict such flows. This thesis addresses this by providing a detailed assessment on the performance of two electromagnetically extended Reynolds-averaged Navier-Stokes models through computations of a number of electromagnetically influenced simple channel and Rayleigh-Bènard convective flows. The models tested were the low-Re k-ε linear eddy-viscosity model of Launder and Sharma (1974), with electromagnetic modifications as proposed by Kenjereš and Hanjalić (2000), and the low-Re stress-transport model of Hanjalić and Jakirlić (1993), with electromagnetic modifications as proposed by Kenjereš and Hanjalić (2004). First, a one-dimensional fully-developed turbulent channel flow was considered over a range of Reynolds and Hartmann numbers with a magnetic field applied in both wall-normal and streamwise directions. Results showed that contributions from the electromagnetic modifications were modest and, whilst both models inherently captured some of the reduction in mean strain that a wall-normal field imposed, results from the stress-transport model were consistently superior for both magnetic field directions. Then, three-dimensional time-dependent Rayleigh-Bènard convection was considered for two different Prandtl numbers, two different magnetic field directions and over a range of Hartmann numbers. Results revealed that, at sufficiently high magnetic field strengths, a dramatic reorganization of the flow structure is predicted to occur. The vertical magnetic field led to a larger number of thinner, more cylindrical plumes whilst the horizontal magnetic field caused a striking realignment of the roll cells' axes with the magnetic field lines. This was in agreement with both existing numerical simulations and physical intuition. The superior performance of the modified stress-transport model in both flows was attributed to both its ability to provide better representation of stress generation and other processes, and its ability to accommodate the electromagnetic modifications in a more natural, and exact, fashion. The results demonstrate the capabilities of the stress-transport approach in modelling MHD flows that are relevant to industry and offer potential for those wishing to control flow structure or levels of turbulence without recourse to mechanical means.

In computational fluid dynamics (CFD), there are two widely-used methods for computing the near-wall regions of turbulent flows: high Reynolds number (HRN) models and low Reynolds number (LRN) models. HRN models do not resolve the near-wall region, but instead use wall functions to compute the required parameters over the near-wall region. In contrast, LRN models resolve the flow right down to the wall. Simulations with HRN models can take an order of magnitude less time than with LRN models, however the accuracy of the solution is reduced and certain requirements on the mesh must be met if the wall function is to be valid. It is often difficult or impossible to satisfy these requirements in industrial computations. In this thesis the near-wall domain decomposition (NDD) method of Utyuzhnikov (2006) is developed and implemented into the industrial code, Code_Saturne, for the first time. With the NDD approach, the near-wall regions of a fluid flow are removed from the main computational mesh. Instead, the mesh extends down to an interface boundary, which is located a short distance from the wall, denoted y*. A simplified boundary layer equation is used to calculate boundary conditions at the interface. When implemented with a turbulence model which can resolve down to the wall, there is no lower limit on the value of y*. There is a Reynolds number-dependent upper limit on y*, as there is with HRN models. Thus for large y*, the model functions as a HRN model and as y*→ 0 the LRN solution is recovered. NDD is implemented for the k−ε and Spalart-Allmaras turbulence models and is tested on five test cases: a channel flow at two different Reynolds numbers, an annular flow, an impinging jet flow and the flow in an asymmetric diffuser. The method is tested as a HRN and LRN model and it is found that the method behaves competitively with the scalable wall function (SWF) on simpler flows, and performs better on the asymmetric diffuser flow, where the NDD solution correctly captures the recirculation region whereas the SWF does not. The method is then tested on a ribbed channel flow. Particular focus is given to investigating how much of the rib can be excluded from the main computational mesh. It is found that it is possible to remove 90% of the rib from the mesh with less than 2% error in the friction factor compared to the LRN solution. The thesis then focuses on the industrial case of the flow in an annulus where the inner wall, referred to as the pin, has a rib on its surface that protrudes into the annulus. Comparison is made between CFD calculations, experimental data and empirical correlations. It is found that the experimental friction factors are significantly larger than those found with CFD, and that the trend in the friction factor with Reynolds number found in the experiments is different. Simulations are performed to quantify the effect that a non-smooth surface finish on the pin and rib surface has on the flow. This models the situation that occurs in an advanced gas-cooled nuclear reactor, when a carbon deposit forms on the fuel pins. The relationship between the friction factor and surface finish is plotted. It is demonstrated that surface roughness left over by the manufacturing process in the experiments is not the source of the discrepancy between the experimental and CFD results.

The main objective of the thesis is to provide a detailed assessment of the performance of four types of Low Reynolds Number (LRN) Eddy Viscosity Models (EVM), widely used for industrial purposes, on flows featuring SWBLI, using experimental and direct numerical simulation data. Within this framework the two-equation linear k-ε of Launder and Sharma (1974) (LS), the two-equation linear k-ω SST, the four-equation linear φ-f of Laurence et al. (2004) (PHIF) and the non-linear k-ε scheme of Craft et al. (1996b,1999) (CLSa,b) have been selected for testing. As initial test cases supersonic 2D compression ramps and impinging shocks of different angles and Reynolds numbers of the incoming boundary layer have been selected. Additional test cases are then considered, including normal shock/isotropic turbulence interaction and an axisymmetric transonic bump, in order to examine the predictions of the selected models on a range of Mach numbers and shock structures. For the purposes of this study the PHIF and CLSa,b models have been implemented in the open source CFD package OpenFOAM. Some results from validation studies of these models are presented, and some explorations are reported of certain modelled source terms in the ε-equation of the PHIF and CLSb models in compressible flows. Finally, before considering the main applications of the study, an examination is made of the performance of different solvers and numerical methods available in OpenFOAM for handling compressible flows with shocks. The performance of the above models, is analysed with comparisons of wall-quantities (skin-friction and wall-pressure), velocity profiles and profiles of turbulent quantities (turbulent kinetic energy and Reynolds stresses) in locations throughout the SWBLI zones. All the selected models demonstrate a broadly consistent performance over the considered flow configurations, with the CLSb scheme generally giving some improvements in predictions over the other models. The role of Reynolds stress anisotropy in giving a better representation of the evolution of the boundary layer in these flows is discussed through the performance of the CLSb model. It is concluded that some of the main deficiencies of the selected models is the overestimation of the dissipation rate levels in the non-equilibrium regions of the flow and the underestimation of the amplification of Reynolds stress anisotropy, especially within the recirculation bubble of the flows. Additionally, the analysis of the performance of the considered EVM's in a normal shock/isotropic turbulence interaction illustrates some drawbacks of the EVM formulation similar to the ones observed in normally-strained incompressible flows. Finally, a hybrid Detached Eddy Simulation (DES) approach is incorporated for the prediction of the transonic buffet around a wing.

Rod bundle is a typical constitutive element of a very wide range of nuclear reactor designs. This thesis describes the investigation of such geometry with wall-resolved Large Eddy Simulation (LES). In order to alleviate the mesh constraint, imposed by the near wall resolution, the usage of embedded refinements and polyhedral meshes is analysed firstly with a inviscid laminar case (Taylor Green vortices) and secondly with a fully turbulent case (channel flow only with embedded refinement). The inviscid test case shows that the addition of embedded refinements decreases the conservation properties of the code. Indeed the accuracy decreases from second order in a structured conformal mesh, to something in between first and second order depending on the quality of the unstructured mesh. Better results are obtained when the interface between refined and coarse areas presents a more regular and structured pattern, reducing the generation of skewed and stretched cells. The channel flow simulation shows that the Reynolds stresses, of some embedded refined meshes, are affected by spurious oscillations. Surprisingly this effect is present in the unstructured meshes with the best orthogonal properties. Indeed analysis of Reynolds stress budgets shows that terms, where the gradient in the wall normal direction is dominant, have a largely oscillatory behaviour. The cause of the problem is attributed to the convective term and in particular in the method used for the gradient reconstruction. As a consequence of these contradictory signs between the inviscid and the fully turbulent cases, the rod bundle test case is analysed using a conventional body fitted multiblock mesh. Two different Reynolds numbers are investigated reporting Reynolds stresses and budgets. The flow is characterised by an energetic and almost periodic azimuthal flow pulsation in the gap region between adjacent sub-channels, which makes turbulent quantities largely different from those in plane channel and pipes and enhances mixing. Experiments found that a constant Strouhal number, with the variation of the Reynolds number, characterises the phenomenon. The frequency analysis finds that present simulations are distinguished by three dominant frequencies, the first in agreement with the experimental value and two higher ones, which might be due to the correlation of the azimuthal velocity in the streamwise direction. Several passive temperature fields are added at the simulations in order to study the effects of the variation of the Prandtl number and the change in boundary conditions (Neumann and Dirichlet). A simplified case where an imbalance of the scalar between adjacent sub-channels is also investigated in order to evaluate the variation of the heat fluxes with respect to the homogeneous case. An alternative solution, to reduce the mesh constraint imposed by the wall, is to hybridize LES with RANS. The main achievement of this work is to integrate the heat transfer modelling to the already existing model for the dynamic part. Further investigations of the blending function, used to merge the two velocity fields, are carried out in conjunction with a study of the model dependency on the mesh resolution. The validation is performed on a fully developed channel flow at different Reynolds numbers and with constant wall heat flux. On coarse meshes the model shows an improvement of the results for both thermal and hydraulic parts with respect to a standard LES. On refined meshes, suitable for wall-resolved LES, the model suffers from a problem of double counting of modelled Reynolds stresses and heat fluxes because the RANS contribution does not naturally disappear as the mesh resolution increases.

Cette thèse porte sur la prédiction haute-fidélité de phénomènes visqueux turbulents modélisés par les équations Reynolds-Averaged Navier-Stokes (RANS). Si l’adaptation de maillage a été appliquée avec succès aux simulations non-visqueuses comme la prédiction du bang sonique ou la propagation d’explosion, prouver que ces méthodes s’étendent et s’appliquent également aux simulations RANS avec le même succès reste un problème ouvert. Dans ce contexte, cette thèse traite des problématiques relatives aux méthodes numériques (solveur de mécanique des fluides) et aux stratégies d’adaptation de maillage. Pour les méthodes numériques, nous avons implémenté un modèle de turbulence dans notre solveur et nous avons conduit une étude de vérification et validation en deux et trois dimensions avec comparaisons à l’expérience. Des bons résultats ont été obtenus sur un ensemble de cas tests, notamment sur le calcul de la traînée pour des géométries complexes. Nous avons également amélioré la robustesse et la rapidité de convergence du solveur, grâce à une intégration en temps implicite, et grâce à une procédure d’accélération multigrille. En ce qui concerne les stratégies d’adaptation de maillage, nous avons couplé les méthodes multigrilles à la boucle d’adaptation dans le but de bénéficier des propriétés de convergence du multigrille, et ainsi, améliorer la robustesse du processus et le temps CPU des simulations. Nous avons également développé un algorithme de génération de maillage en parallèle. Celui-ci permet de générer des maillages anisotropes adaptés d’un milliard d’éléments en moins de 20 minutes sur 120 coeurs de calcul. Enfin, nous avons proposé une procédure pour générer automatiquement des maillages anisotropes adaptés quasi-structurés pour les couches limites. / This thesis deals with the high-fidelity prediction of viscous turbulent flows modelized by the Reynolds-Averaged Navier-Stokes (RANS) equations. If mesh adaptation has been successfully applied to inviscid simulations like the sonic boom prediction or the blast propagation, demonstrating that these methods are also well-suited for 3D RANS simulations remains a challenge. This thesis addresses research issues that arise in this context, which are related to both numerical methods (flow solver) and mesh adaptation strategies. For the numerical methods, we have implemented a turbulence model in our in-house flow solver and carried out its verification & validation study. Accurate results were obtained for a representative set of test cases, including the drag prediction workshop. Additional developments have been done to improve the robustness and the convergence speed of the flow solver. They include the implementation of an implicit time integration and of a multigrid acceleration procedure. As regards mesh adaptation, we have coupled the adaptive process to multigrid in order to benefit from its convergence properties and thus improve the robustness while preventing losses of computational effort. We also have devised a parallel mesh generation algorithm. We are able to generate anisotropic adapted meshes containing around one billion elements in less than 20min on 120 cores. Finally, we introduced a procedure to automatically generate anisotropic adapted quasi-structured meshes in boundary layer regions.

Cases of dam failure occur around the world almost each year. Dam failures can result in the formation and propagation of fast-moving unsteady flows that can cause loss of life as well as significant environmental and economic consequences in downstream flooded areas. The initial stages of a dam break are important due to wave-breaking front and the associated turbulence. Furthermore, characteristics of the river bed downstream of the dam (topography and bathymetry) as well as the presence of obstacles in the dam break wave path such as man-made or natural obstacles like bridges, trees, and local sills affect flow dynamics, which can lead to the formation of hydraulic jumps and the reflection of the flood wave. Accordingly, the precise prediction of flood parameters such as arrival times, free surface profiles, and flow velocity profiles is essential in order to mitigate flood hazards. This study aimed to assess the performance of various turbulence models in predicting and estimating dam-break flows and related positive and negative flood wave characteristics over different downstream bed conditions. Three-dimensional (3-D) Computational Fluid Dynamics (CFD) models were created to solve the unsteady Reynolds equations in order to determine the initial stages of the free surface profiles over dry and wet beds and to investigate the generation and propagation of dam-break flows and reflected flood waves in the presence of a bed obstacle. The performance of different Reynolds-averaged Navier-Stokes (RANS) turbulence models has been investigated, and the standard k-ε, RNG k-ε, realizable k-ε, k-ω SST, and v^2-f turbulence models have been studied using OpenFOAM software. Dam-breaks were modelled using the Volume of Fluid (VOF) method employing the Finite Volume Method (FVM). Both qualitative and quantitative comparisons of numerical simulations with laboratory experiments were completed in order to assess the suitability of different turbulence models. The results of the first study showed that the RNG k-ε model exhibited better performance in capturing the flood wave free surface profiles over both dry- and wet-bed downstream conditions, while from the second study, it was concluded that the k-ω SST model was able to accurately predict the formation and propagation of reflected waves against a bottom obstacle in terms of free surface profiles and negative bore propagation speeds.

Wake has a high impact on wind turbine performance and durability. Its impact, without a doubt, should be considered in wind resource assessment. However, wakes in the forest is a niche that is only examined by a couple of countries which area has a significant per cent land covered in forests. In this thesis, the wake over a homogenous forest is analysed, using the realisable k-e model. Simulations were performed using OpenFOAM software. Firstly, simulations were performed for one wind turbine to examine if RANS is a reliable tool for wake calculation. Secondly, the power production of two wind turbines in tandem are examined ,where one is in the wake of another one. Wind speed velocity analysis concluded that RANS credible predicts velocity deficit in the wake region and could be used in forested areas. For instance, the correlation coefficient for velocity mostly lies above 0.9. Also, the distance between wind turbine swhich would be erected in the forest could be reduced. The reduction could exceed up to 3D with only 35% power loss, which means that wind farms in forests could be more compact. Further studies could examine multiple wake interactions and how a combination of forest clearing, vegetation growth and snow cover affects wind farm performance over a lifespan.