Search

Particle image velocimetry (PIV) is a ubiquitous optical based experimental technique for providing high resolution measurements of the velocity and derived turbulent quantities. Such data is relied upon for the validation of computer codes which are used to design and safety assure many engineering systems in use today. A desire for greater performance and safety within nuclear reactor designs increasingly demands the use of computational fluid dynamics (CFD). To fulfil this role CFD codes and models must be validated to the exceedingly high standards of the nuclear industry which in large part is expected to be performed using PIV. For nuclear reactors, water is the most commonly used coolant and thus the performance of CFD to simulate the related flow physics in water is of great interest. One particular area of interest is natural convection, which is relied upon in reactor designs as a passive heat removal capability in the event of emergency. Performing PIV measurements in water based natural convection experiments has traditionally been limited by the distorting effect of refraction which may result in substantial measurement error. In this thesis high resolution PIV measurements of natural convection from a uniformly heated vertical surface submerged in thermally stratified water are performed. The relatively high heat flux results in a transition to turbulence and substantial refractive distortion and blurring. A correction method is developed to correct for the refractive distortion and some amount of blurring. The method involves performing simultaneous background orientated schlieren (BOS) measurements to measure the distortion and provide a correction for the instantaneous PIV images as a pre-processing operation. The technical challenges of performing measurements in such circumstances and the methods employed to overcome them are discussed through this thesis. Novel methods for scaling the measured distortion and accounting for differing thermal conditions between measurements are presented. Finally, the uncorrected PIV result is compared against an instantaneous and time averaged corrected results. It is found that whilst the flow is laminar the effect of refraction is negligible, and as the flow transitions to turbulence the applicability of a time averaged correction method is limited.

This study was set out to perform a fluid-structure interaction benchmark simulation of the cylindrical cantilever beam subjected to the axial turbulent flow. The application has practical importance in the fuel rods in Pressurized Water Reactors (PWR). The modelling approach was chosen to be optimal for industrial applications. A recent experimental study from the authors' group on flow-induced vibrations generated by axial turbulent flow over a cantilever rod provides data on the rod displacement and local flow data, making this case highly attractive for CFD validation. The motivation for this research is thus provided by the availability of these data. This thesis investigates the computational modelling of flow-induced vibrations of cantilever rods subjected to turbulent axial flow at operating conditions relevant to those of fuel rods of PWR nuclear reactors. The aim is to assemble all the modelling elements needed for a cost-effective and thus URANS-based (Unsteady Reynolds Averaged Navier Stokes) modelling strategy, employing high-Reynolds-number turbulence models. This objective is pursued through three stages. It was first necessary to investigate the numerical FSI (Fluid-Structure Interaction) strategy adopted through the computation of flow-induced vibration of an elastic plate subjected to axial laminar flow. Then the suitability of URANS models was assessed through computations of turbulent flow over a forward-facing step, for which measurements of the fluctuating wall pressure are available. On the numerical side, these explorations led to adopting a two-way FSI strategy, using a single finite-volume solver, with the Arbitrary Lagrangian-Eulerian (ALE) approach, high-order convective discretization schemes and Laplacian smoothing for the displacement of the mesh in the fluid domain. On the physical modelling side, they resulted in the use of high-Reynolds-number Reynolds stress transport models. The resulting modelling strategy is subsequently validated against the experimentally investigated case of a steel cantilever rod caused to vibrate through exposure to turbulent axial flow. This is the first study that has successfully reproduced both the frequency and the amplitude of the flow-induced vibrations relevant to PWR applications, based on the use of URANS.

The present thesis focuses on the CFD analysis of natural circulation flows that are related to the passive safety loops of the forthcoming generation of water nuclear reactors. These flows involve inherent unsteadiness, laminar-to-turbulent transition and thermal stratification phenomena. As an initial validation study, a range of models that solve the Reynolds-Averaged Navier-Stokes equations (RANS), within the open-source CFD code Code_Saturne, are assessed in a forced convection 1-D channel flow case. A proposed numerical form of the Analytical Wall Function (AWF) for handling near-wall regions has been implemented within the code, and is further extended to account for buoyancy effects. A range of eddy-viscosity and second-moment closures with different near-wall treatments and different turbulent heat flux approximations are then tested using 2-D steady computations of the flow and heat transfer in a high aspect ratio differentially heated cavity. In this buoyancy driven flow, the k-epsilon model with AWF shows very promising results of the heat transfer and the turbulence mixing effects. Further comparisons of RANS models are performed in a differentially heated square cavity at Ra=1.58e9 and 10e11 with air using 2-D and 3-D transient simulations. It is shown that some models tend to laminarise the flow whilst others are deficient due to the log-law wall function. Significant improvements are shown when the effects of the 3-D structures are considered. The proposed strategy, AWF with k-epsilon model, outperforms the log-law based wall function, exhibiting robustness and improved Nusselt number and turbulence level predictions. The greatest resolution of this unsteady flow was achieved through a computationally demanding Large Eddy Simulation (LES), which was shown to agree closely with existing DNS data. The final series of 2-D and 3-D calculations concern unsteady RANS and conjugate heat transfer analysis for three vertical heater vertical cooler loop configurations. A range of flow regimes are studied, with modified Rayleigh numbers within 10e9-10e13, for water. The analysis mostly concerns the steady-state though some thermal transients and secondary motion in the cooler side are examined, and a way of appropriately non-dimensionalizing the predicted quantities is proposed. The developed strategy, k-epsilon with AWF, returned satisfactory predictions of the thermal and dynamic fields whereas the EBRSM and the Launder and Sharma form of the k-epsilon required a specific numerical treatment to maintain the flow turbulent. All computations were validated against either experimental or DNS data, or 1-D correlations where appropriate.

A wide range of engineering industrial applications requires both the thermal and optical efficiencies of the system to be maximized with a reasonable low penalty for the friction factor and subsequently low losses in pressure. Amongst the family of concentrated solar power systems, parabolic trough collectors (PTC) which have recently received significant attention, face similar challenges. To effectively enhance the thermal performance of the PTC system four enhancement techniques were numerically investigated and addressed in this thesis; changing heat transfer fluids, replacing the working fluids with nanofluids that have better thermal-physical properties than those of base fluids, inserting different turbulators with various design configurations and finally combining the advantages of nanofluids and swirl generators in the same application. All simulations were assumed to be in steady-state and three dimensions with a range of Reynold's number (Re=104-105). For the simulation the Monte Carlo Ray Tracing (MCRT) model was used to represent the non-uniform heat flux around the absorber tube of the PTC. Two low-Reynold's turbulence models were used; Launder and Sharma (LS) k-epsilon and Shear Stress Transport (SST) k-omega models. In order to assess the performance of each enhancement technique, a number of parameters were analyzed including average Nusselt (Nu) number, specific pressure drop distributions, thermal losses, overall collector efficiency and exergy efficiency of the PTC system. Three categorized-types of pure fluids were used firstly. Secondly, numerical simulations were performed for a solar collector to test the effectiveness of six non-metallic nanoparticles dispersed individually in three different base working fluids with three different volume fractions. The third step was to study the effect of the variation of geometrical properties of a single canonical insert to find the optimized shape then increase the number of strips to two, three and four around the central rod. The final step was to assess the effect of various straight strip shapes with and without nanofluids. Four different strip arrangements were considered; large conical-shape strips, small conical-shape strips, rectangular-shape strips and elliptical-shape strips. Results showed that, the largest enhancements in the overall collector efficiency and thermal exergy efficiency were achieved by the hybrid system of combining both large conical-shape strips and 6% of SiO2 dispersed in therminol VP-1 which are 15.41% and 15.32% respectively compared to a typical system.

The current study aims to validate and extend RANS approaches, using linear and non-linear eddy viscosity models, by considering the modelling of hypersonic flows, especially those with SWBLIs. The main objective is to assess the performance of three EVMs, widely used in industrial applications, in predicting flow separation and aerodynamic heating in hypersonic flows, and to explore adaptations that can improve their performance, by accounting for influences such as strong compressibility and discontinuity in flow variables. The turbulence models employed are the linear k-ε model of Launder and Sharma (1974) (LSY model), the linear k-ω shear stress transport model of Menter et al. (2003) (SST model) and the non-linear k-ε model of Craft et al. (2000) (CLS model). To give an objective, uniform and reliable evaluation of the performance of the turbulence models selected, the formulation of RANS equations for high-speed flows has first been investigated. A new density-based compressible solver, with various options of including, or excluding, several terms that were typically omitted in low speed (and sometimes high speed) flow simulations has developed within the framework of OpenFOAM. The influence of partially, or fully, ignoring these terms in the RANS equations has been analysed with comparisons of spatial quantities and wall quantities in supersonic and hypersonic cases. The numerical results show that while omitting these terms is typically fine for low-speed applications, they can have a significant effect on predictions of SWBLI cases in the hypersonic regime. There-fore, the form of RANS equations with all these terms fully included is selected to run all the simulations in this research. The non-linear CLS model has been implemented in the OpenFOAM framework in a similar way as the linear models tested in the current study and then examined, together with the other two linear models, by comparing results over a range of typical hypersonic SWBLIs benchmark cases, 2D, axisymmetric or 3D, over a wide range of Mach numbers and Reynolds numbers. The three models tested do generally return reasonable predictions of wall pressure in most cases, while the non-linear CLS model presented the best capability among the three in predicting the flow separations. Nevertheless, the wall heat flux in the interaction region is overpredicted in most cases by all three models (sometimes showing quite a dramatic overprediction). To improve the accuracy of predictions of wall heat flux, a new source term in the modelled dissipation rate equation, aimed at restricting the turbulent length scale in the shock wave boundary layer interaction region, is proposed. This term will cause no impact on incompressible flows and only has a limited influence on the predic-tions of wall pressure in hypersonic flows. The introduction of this new source term to the LSY and CLS models is shown to largely eliminate their over predictions of the peaks of wall heat flux in the interaction region (in some test cases, the peak value was reduced by around 80%), while the extra cost of the numerical simulation is less than 2%.

In this project, modelling of turbulent conjugate heat transfer within RANS frame is studied. The starting point is a four-equation transport model, developed by Craft et al. (2010). In this project, the model from Craft et al. (2010) has been tested on 1D fully-developed channel flow with heated wall at Pr=0.71 with Reτ=150, comparing predictions against DNS data from Flageul et al. (2017) covering a range of fluid and solid thermal properties. The results showed too rapid decay of temperature variance across the solid region. This investigation makes use of the wider range of available DNS data, to advance understanding of the processes involved and identify the influential parameters. Then the Craft et al. (2010) model is made sensitive to these physical phenomena. The three main elements of the model development are a) a model of the dissipation of thermal fluctuations in the solid region based entirely on solid-domain DNS data, b) a re-optimised model of the transport equations for the thermal fluctuations and their dissipation rate within the fluid domain and c) a revised set of interface conditions for the discontinuity of the dissipation rate of thermal fluctuations across the fluid-solid boundary. This was carried out using a 1D solver written by the author of this thesis using Fortran. Since the model is developed by considering fully-developed turbulent flow, its capability for handling more complex and challenging cases needs to be investigated. The revised new model has thus been incorporated into the open source CFD package Foam-extend 4.0, which can handle numerical simulations for complicated engineering problems. Then the solver developed within Foam-extend 4.0 has been used to test the revised model for 3D conjugate heat transfer in a channel flow with a hot jet injected from the bottom. The results have been compared with those from DNS data (Wu et al. (2017)) and those from the Craft et al. (2010) model, with the present model showing generally good agreement with the DNS.

The Large Hadron Collider is scheduled for upgrade during the middle of this decade, and with it will be upgraded most of the Silicon detectors used in the four experiments of the collider. The upgraded detectors will require a more advanced cooling system that will need to be larger and operate colder than ever before. This thesis describes activities related to the modelling and simulation of the two-phased pumped-loop cooling systems that will cool these Silicon detectors. The development of the tool is described, with details on modelling fluid properties, two-phase flow and some numerical techniques used to facilitate the solver's operation. The experimental setup that was refurbished and upgraded is also described. The setup was used to collect transient data to validate the performance of the tool. The application of the tool is discussed next. The tool was initially validated by simulating a residential heat pump unit. Simulation results were compared against experimental data as well as against another established simulation tool. Residential heat pumps are well-understood two-phase systems which exhibit transients of similar complexity to pumped loop systems. The comparison is, thus, informative. The tool was then used to validate the performance of a laboratory prototype of a pumped loop cooling system representative of the full-scale Silicon detector cooling systems. Startup and step change transients were compared and the tool captured the experimental trends well. The tool is now being used to design a prototype cooling system that will serve as a proof-of-concept for the future detector cooling systems to be deployed on the LHC. The tool is used to study plant behaviour in different dynamic scenarios such as startup, temperature set-point change and detector load change. The full control system to be used for the plant was implemented in the tool and control strategies were iterated. The simulation results are in line with expected behaviour, indicating a readiness of the tool to be used for tasks such as controller parameter tuning, virtual commissioning and operator training.

This present study aims to predict the flow past Flettner Rotor. The effective predicted approaches are also explored to provide the most accurate result of flow past both a stationary and rotating cases. The Flettner Rotor comprises of cylinder (curved) and disc (flat) walls; flow always interacts with these surfaces; the curved cylinder surface causes to produce re-circulating flow; the disc associates the increasing of flow speed and producing more propulsion force exerting on the cylinder. The research starts from laminar flow past a rotating cylinder at Re of 200 (Coutanceau and Menard, 1985). The study applies the pure laminar approach (Navier-Stokes Equations) examining 3-D flow. The result shows that the process of wake development is consistent with previous experimental study. The predicted aerodynamic coefficients are a little different with previous 2-D computational study (Mittal and Kumar, 2003) due to 3-D effect. The next test case is to predict turbulent flow past a rotating cylinder at Re of 130,000 (Aoki and Ito, 2001). A number of turbulence modelling are employed which are low-Re k-epsilon and high-Re k-epsilon with a series of eddy viscosity model including linear, quadratic and cubic models. The Reynolds Stresses Equation model (RSM) with different versions (GL: Gibson and Launder (1978) and TCL : Two Component Limit (Craft and Launder, 1996)) are also applied. The standard and analytical wall treatments are in conjunction with those of high-Re k-epsilon approaches to examine the effect of turbulence across fully turbulent region; they require low computational resource. All turbulence models predict the increase and decrease of lift and drag coefficient when spin ratios (alfa : the proportional tangential cylinder to free flow velocity) are larger. However, the predicted results are not effective in the transition flow from laminar to turbulence at spin ratios less than 0.5, the pure laminar approach provides the accurate result in this range. The installation of multiple and end discs on cylinder are predicted at Re of 5,600, 5,800, 12,200 and 48,900 (Thom, 1934; Clayton, 1985). The flow past multiple discs along cylinder is simplified by the flow past disc spacing. Whilst, the flow past a Flettner Rotor is investigated on fully domain size corresponding to real experimental setup at Re of 48,900 (Clayton, 1985). The accurate predicted results are obtained by low-Re k-epsilon and RSM-GL approaches with near wall functions on either disc or cylinder.

This study presents an experimental investigation of rotating flows related to gas turbine blade cooling. It focuses on measuring the development of turbulent flow and heat transfer inside a cooling passage which consists of two straight square-sectioned ducts connected with a square-ended bend, using porous metal foam blocks as turbulent promoters. The cooling passages are either stationary, or in orthogonal rotation. This study aims to improve the understanding of the effects of using porous metal foam on flow and thermal development in rotating cooling passages and exploring the possibility of using porous media for gas turbine blade cooling. Particle image velocimetry (PIV) and Thermochromic liquid crystals (TLC) experiments have been performed to measure the flow and heat transfer characteristics, respectively. A set of 12 static pressure taps around the passage provide detailed picture of the pressure variation along both sides of the rotating passage. Aluminium porous foam blocks of an aspect ratio of 1.5 with 0.93 porosity and pore density per cm of 2, were attached to two opposite walls of the straight sections normal to the bend in a staggered manner. The ratio of blocks spacing to the duct's hydraulic diameter (D) is 1; whereas the block's height ratio (h/D) is 0.6. Water is used as the working fluid at Reynolds numbers of 16,000, 26,000 and 36,000 and rotation numbers of 0.32 and 0.64. The resulting tests have generated original knowledge and information which advances our understanding of the effects of porous metallic foams on the flow and thermal development in rotating cooling passages and also provides detailed data for CFD validation. The results generated by the PIV method show the serpentine manner of the flow both upstream and downstream of the bend region due to the presence of the porous blocks. Within the bend, in contrast to the case with a smooth upstream section, a single vortex dominates the flow. While rotation does not change the overall flow character, it does force more fluid through the blocks on the trailing (pressure) side of the duct. For both stationary and rotating conditions, the upstream section is long enough for the flow to become periodic over successive rib intervals, which produces very attractive data for CFD validation. The Nusselt number distribution is significantly affected by the presence of the porous blocks and the heat transfer levels are improved by about 17%. The local Nusselt number distribution shows that the influence of rotation is negligible compared to the effect of the porous blocks.

This doctoral thesis concerns itself with the simulations of turbulent fluid flows with heat transfer about serpentine tube boiler geometries, with the aim of developing the tools necessary to tackle research questions put forward by the EDF Energy R&D UK Centre with regard to their continued use beyond their original decommissioning date in a safe manner. This is in support of the Plant Lifetime Extension, or PLEX, scheme currently being implemented by EDF to safely lengthen the operational lifespan of their Advanced Gas-Cooled Reactor (AGR) fleet, which provide a significant portion of the UK's energy demands, without large-scale reactor deployments to replace them. Among the key components that ageing affects most significantly are the serpentine boiler systems used by some of the newer AGR designs, where fatigue and corrosion damage have forced operators to plug individual tube platens to prevent feedwater escaping and coming into contact with the graphite core. The primary coolant is CO 2 , which is heated by the reactor core before coming into thermal contact with the boiler tubes, producing steam. The flow domain is an in-line tube bank with alternating longitudinal pitch, and so this report utilises Code Saturne, a general-purpose Computational Fluid Dynamics package developed by the wider EDF R&D teams, on an idealised square in-line tube bank validation case assuming flow periodicity, along with supporting software packages. Chapter 1 contains a discussion on AGR technology, introduces the problem definition, and the main research outcomes expected by EDF Energy by the doctorate's end. Chapter 2 documents a review of pertinent literature concerning flows about cylinders and tube banks, and the heat transfer processes which occur across them. Chapter 3 outlines the software packages and the methods used to simulate fluid problems via the Finite Volume Method, along with the reasons for their selection. Chapter 4 is entirely dedicated to turbulence modelling and explores in detail many of the strategies used to account for it, though it is not exhaustive. Chapter 5 introduces the concept and underlying reasoning behind the Analytical Wall Function (AWF) of Suga et al. (2006), which enables the possibility of improved near-wall behaviour for high-Reynolds number turbulence modelling approaches and an extension to account for augmented surface roughness. Chapter 5 also documents precisely how this AWF modelling approach has been implemented in Code Saturne for the first time as part of this research work. Chapter 6 presents the methodology behind simulating a square in-line tube bank using flow periodicity, and the results obtained from 2D and 3D appraisals using Unsteady Reynolds-Averaged Navier-Stokes (URANS) and Large Eddy Simulation (LES) approaches, with temperature operating as a passive scalar while thermal fluid properties are held constant, both with and without the AWF implementation. Chapter 7 addresses a similar methodology, except on a problem of greater interest to the application, a spanwise-periodic platen geometry consisting of four tube platens with ten passes through the flow domain, with confining walls, where tube blanking is represented and the properties of carbon dioxide can now vary freely with temperature. Attempts are made also to definitively quantify the effect of blanking and augmented roughness on impairments of heat transfer through the system, to give some indication as to its safe operational lifespan. Finally, the conclusions from each chapter are presented, along with a discussion on potential avenues for further exploration. Appendices A and B list some of the more lengthy formulations and derivations used within the AWF implementation respectively, while Appendix C contains some supplementary results from Chapter 7 not critical to the discussion taking place there. The findings here indicate that in a worst-case scenario, at current reactor capacity, approximately 30% of heat transfer is negatively impacted by blanking of tubes and near-wall roughness effects.

This research focuses on the development of wall functions suitable for the prediction of high-speed compressible flows. Wall-functions avoid the need for prohibitively expensive fine near-wall meshes and low-Re models of turbulence which still involve a certain amount of approximation. The conventional log-law-based wall functions, however, have limitations in even incompressible cases, which are further compounded when applied to high-speed compressible flows. The objective of this study is to examine the performance of an advanced analytical wall-function treatment which has been successfully used in a range of incompressible flows and explore how compressibility effects could be accounted for in such approaches. The starting point was the implementation of the analytical wall function proposed by Craft et al (2002) in OpenFoam and its subsequent use for the prediction of the impinging shock interaction and compression corner cases up to a Mach number of 3. The wall pressure and skin friction results obtained by the original version result in improvements over those of the standard wall function (log-law based) and are close to those obtained by the low-Re number modelling for supersonic flows. However, an unphysical behaviour is encountered when applying it to higher Mach number cases. A compressible flow version of the analytical wall function is proposed which includes the following modifications: a)inclusion of thermal dissipation terms in the analytical equation for the energy variation over the near-wall cells, b) Variable molecular viscosity (due to temperature variations) over the viscous sub-layer, c) improved variation of the convection terms in the near-wall cell analytical equations. The resultant model has been applied to the above flows up to Mach numbers of 9 and comparisons drawn with experimental data and with predictions from the log-law based wall functions and from the Low-Re Launder and Sharma model. The present results are consistently closer to the data than those of other wall functions in some instances even better than those of the low-Re number. Improvements are especially noticeable in the prediction of the wall heat flux rates, where the log-law wall function generally predicts too low values in the shock interaction region, while the low-Re model, predicts too high heat transfer rates in the highest Mach number cases, as a result of overpredicting turbulence levels where extremely rapid near-wall temperature variations are found.

The phenomenon of fluid-structure interaction is present in many industrial applications, such as bridge cables, transmission wires, drilling risers in petroleum production, biomedical engineering, etc. In the nuclear industry, fluid-structure interactions have also been identified, one form of which is Flow-Induced Vibrations (FIVs) of the fuel rods in a typical pressurized nuclear water reactor core. As a result, the vibrating fuel rods may have contact with the neighbouring structures, such as spacer grids which are technically a structure for preventing the fuel rods from excessive movement, and concurrently initializing fretting on the fuel rod surfaces, called grid-to-rod fretting. Thus, for safety concerns, the characteristics of the flow-induced rod vibrations in such a system are required to be understood, and accordingly monitoring the wear-through failure using a prediction tool. In the present study, an experimental approach is proposed to advance our understanding of these phenomena and extend the range of available correlations. In the present experiment, a flow-induced structural vibration system has been designed, in which the geometry is prototypical of pressurized water nuclear reactor core and the flow parameters replicate the flow conditions during its full power operation. Following a validation of the methodology, a series of tests on a cantilever rod in pipe flows directed from the rod free end towards the fixed end has been carried out, in which the rod features either a blunt or a tapered free-end shape, and has been filled internally with either air or lead (for mimicking the fuel pellets in a real nuclear fuel rod). Through analysis of the resulting data, it has been found that the vibrating amplitude of a cantilever rod is more sensitive to the free-end shape of the rod, while the vibrating frequency is mostly influenced by the internal loading material. These findings give an insight into the future design of relevant structures in nuclear reactor cores, the fluid-structure interaction community will also benefit from the availability of such data to fine-tune relevant numerical codes.

This thesis focuses on the study of flow and heat transfer in porous media in both laminar and turbulent flow regimes, by using Volume Averaged Reynolds Navier Stokes (VARNS) approach. The main concern is to investigate the possibility of using porous media for the gas turbine blade cooling. Very recently, using this technique in blade cooling, particularly with internal cooling, has motivated many researchers due to an effective enhancement in the blade cooling. In this study turbulence is represented by using the Launder-Sharma low-Reynolds-number k-Îμ turbulence model, which is modified via proposals by Nakayama and Kuwahara (2008) and Pedras and de Lemos (2001) for extra source terms in the turbulent transport equations to account for the porous structure, which is treated as rigid and isotropic. Due to the changing of the effective porosity as the clear fluid region is approached, the porosity and additional source term in the macroscopic Reynolds averaged Navier-Stokes equations are relaxed across a thin transitional layer at the edges of the porous media. This is achieved by utilizing exponential damping relations to consider these changes. The Local Thermal Equilibrium (LTE) (one-energy equation) model is used for the thermal analysis in porous media. In order to investigate the validity of the extended model, laminar and turbulent flow in different cases, fully developed and developing flows, have been considered. For laminar flows, fully developed plane channel flows with one and two porous layers, a channel with a single porous block and partially filled porous channel flows have been examined for the purpose of validating the extra drag terms in the momentum equations. For the validation purpose for turbulent flows in porous media, the extended model has been tested in homogeneous porous media, turbulent porous channel flows, turbulent solid/porous rib channel flows, and repeated turbulent porous baffled channel flows. Results of all laminar cases show excellent qualitative agreements with the available numerical calculations and experimental data. Results of all turbulent cases show that the extended model returns generally satisfactory accuracy through the comparisons with the available data, except for some predictive weaknesses in regions of either impingement or adverse pressure gradients, both of which are largely due the underlying eddy-viscosity model formulation employed. Thus, from all results, it can be confirmed that the extended model is promising for engineering applications.

This work extends a previously developed finite-volume overset-grid fluid flow solver to enable the characterisation of rigid-body-fluid interaction problems. To this end, several essential components have been developed and blended together. The inherent time-dependent nature of fluid-solid interaction problems is captured through the laminar transient incompressible Navier-Stokes equations for the fluid, and the Euler-Newton equations for rigid-body motion. First and second order accurate time discretisation schemes have been implemented for the former, whereas second and third order accurate time discretisation schemes have been made available for the latter. Without doubt the main advantage the overset-grid method offers regarding moving entities is the avoidance of the time consuming grid regeneration step, and the resulting grid distortion that can often cause numerical stability problems in the solution of the flow equations. Instead, body movement is achieved by the relative motion of a body fitted grid over a suitable background mesh. In this case, the governing equations of fluid flow are formulated using a Lagrangian, Eulerian, or hybrid flow description via the Arbitrary Lagrangian-Eulerian method. This entails the need to guarantee that mesh motion shall not disturb the flow field. With this in mind, the space conservation law has been hard-coded. The compliance of the space conservation law has the added benefit of preventing spurious mass sources from appearing due to mesh deformation. In this work, two-way fluid-solid interaction problems are solved via a partitioned approach. Coupling is achieved by implementing a Picard iteration algorithm. This allows for flexible degree of coupling specificationby the user. Furthermore, if strong coupling is desired, three variants of interface under-relaxation can be chosen to mitigate stability issues and to accelerate convergence. These include fixed, or two variants of Aitken’s adaptive under-relaxation factors. The software also allows to solve for one-way fluid-solid interaction problems in which the motion of the solid is prescribed. Verification of the core individual components of the software is carried out through the powerful method of manufactured solutions (MMS). This purely mathematically based exercise provides a picture of the order of accuracy of the implementation, and serves as a filter for coding errors which can be virtually impossible to detect by other means. Three instances of one-way fluid-solid interaction cases are compared with simulation results either from the literature, or from the OpenFOAM package. These include: flow within a piston cylinder assembly, flow induced by two oscillating cylinders, and flow induced by two rectangular plates exhibiting general planar motion. Three cases pertaining to the class of two-way fluid-interaction problems are presented. The flow generated by the free fall of a cylinder under the action of gravity is computed with the aid of an intermediate ‘motion tracking’ grid. The solution is compared with the one obtained using a vorticity based particle solver for validation purposes. Transverse vortex induced vibrations (VIV) of a circular cylinder immersed in a fluid, and subject to a stream are compared with experimental data. Finally, the fluttering motion of a rectangular plate under different scenarios is analysed.

In this thesis, the flow and heat transfer in dimpled channels have been investigated using the Reynolds-averaged Navier-Stokes (RANS) approach. The primary objective of this investigation is to identify the capabilities of RANS models to reproduce the characteristics of the flow and heat transfer in dimples. The flow in dimpled channels has been chosen as the test case due to their relevance to gas turbine cooling applications, as well as the fairly complex flow features over dimples, which poses a challenge to turbulence modelling. Five turbulence models have been tested in the present work. These include: the Launder and Sharma k-epsilon model, both the Craft et al. (1996) and (2000) cubic k-epsilon models, the Hanjalic and Jakirlic Reynolds stress model (RSM), as well as the Craft (1998) two-component limit (TCL) RSM. The models have been chosen such that all three classes of RANS closure were tested. The tested models have been applied to two dimpled channel configurations with increasing complexity. In the first, the flow over a single dimple in a channel has been considered, while in the second, the case of a staggered array of dimples has been examined. Moreover, across these two configurations, the effect of the dimple depth, the channel height and the Reynolds number have also been investigated. The results show that all models produce a physically viable solution for the problem of the flow in dimpled channels. Nevertheless, the Craft et al. (1996) and (2000) cubic k-ε models, as well as the Craft (1998) TCL RSM, predicted dimple flow structures that deviate from the expected state. In general, the main flow characteristics are reproduced by the RANS models, and the predicted mean velocity profiles are in good agreement with the data. All models report an overall enhancement in heat transfer levels when using dimples in comparison to those of a plane channel.

This Thesis aims to assess the reliability of turbulence models in predicting the flow fields around the horizontal axis wind turbine (HAWT) rotor blades and also to improve our understanding of the aerodynamics of the flow field around the blades. The simulations are validated against data from the NREL/NASA Phase VI wind turbine experiments. The simulations encompass the use of fourteen turbulence models including low-and high-Reynolds-number, linear and non-linear eddy-viscosity models and Reynolds stress models. The numerical procedure is based on the finite-volume discretization of the 3D unsteady Reynolds-Averaged Navier-Stokes equations in an inertial reference frame with the sliding mesh technique to follow the motion of the rotor blades. Comparisons of power coefficient, normalised thrust, local surface pressure coefficients (CP) and the radial variation of the section average of normal force coefficients with published experimental data over a range of tip-speed ratios, lead to the identification of the turbulence models that can reliably reproduce the values of the key performance indicators. The main contributions of this study are in establishing which RANS models can produce quantitatively reliable simulations of wind turbine flows and in presenting the flow evolution over a range of operating conditions. At low (relative to the blade tip speed) wind speeds the flow over the blade surfaces remains attached and all RANS models return the correct values of key performance coefficients. At higher wind speeds there is circumferential flow separation over the downwind surface of the blade, which eventually spreads over the entire surface, Moreover, within the separation bubble the centrifugal force pumps the flow outwards, which at the higher wind speeds suppresses the formation of the classical tip vortices. More refined RANS models which do not rely on the linear effective viscosity approximation generally lead to more reliable predictions over this range of higher wind speeds. In particular the Gibson-Launder version of the Reynolds stress transport model and the high-Re versions of the Lien et al non-linear k-ε produce consistently reliable simulations over the entire range of wind speeds. By contrast some popular linear effective viscosity models, like the SST (k-ω) and the v^2-f, perform the poorest over this complex flow range. Finally all RANS models are also able to predict the dominant (lowest) frequency of the pressure fluctuations and the non-linear effective viscosity models, the Launder and Shima version of RSM and the SST are also able to return some of the higher frequencies measured.

Numerical models of insect flapping flight have previously been developed and used to simulate the performance of insect flight. These models were commonly developed via Blade Element Theory, offering efficient computation, thus allowing them to be coupled with optimisation procedures for predicting optimal flight. However, the models have only been used for simulating hover flight, and often neglect the presence of the induced flow effect. Although some models account for the induced flow effect, the rapid changes of this effect on each local wing element have not been modelled. Crucially, this effect appears in both axial and radial directions, which influences the direction and magnitude of the incoming air, and hence the resulting aerodynamic forces. This thesis describes the development of flapping wing models aimed at advancing theoretical tools for simulating the optimum performance of insect flight. Two models are presented: single and tandem wing configurations for hawk moth and dragonfly, respectively. These models are designed by integrating a numerical design procedure to account for the induced flow effects. This approach facilitates the determination of the instantaneous relative velocity at any given spanwise location on the wing, following the changes of the axial and radial induced flow effects on the wing. For the dragonfly, both wings are coupled to account for the interaction of the flow, particularly the fact that the hindwing operates in the slipstream of the forewing. A heuristic optimisation procedure (particle swarming) is used to optimise the stroke or the wing kinematics at all flight conditions (hover, level, and accelerating flight). The cost function is the propulsive efficiency coupled with constraints for flight stability. The vector of the kinematic variables consists of up to 28 independent parameters (14 per wing for a dragonfly), each with a constrained range derived from the maximum available power, the flight muscle ratio, and the kinematics of real insects; this will prevent physically-unrealistic solutions of the wing motion. The model developed in this thesis accounts for the induced flow, and eliminates the dependency on the empirical translation lift coefficient. Validations are shown with numerical simulations for the hover case, and with experimental results for the forward flight case. From the results obtained, the effect of the induced velocity is found to be greatest in the middle of the stroke. The use of an optimisation process is shown to greatly improve the flapping kinematics, resulting in low power consumption in all flight conditions. In addition, a study on dragonfly flight has shown that the maximum acceleration is dependent on the size of the flight muscle.

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 the present thesis turbulent natural convection of air within different challenging test cases are investigated numerically by means of an unstructured finite volume code, Code_Saturne. First, flow within both two-dimensional vertical and inclined differentially heated rectangular cavities at 60° and 15° to the horizontal for an aspect ratio of H/L=28.6 and Rayleigh number of 0.86×10e6 is computed using several high and low-Re models. Here the effectiveness of the RANS models in Code_Saturne is assessed through comparisons with a range of available experimental data. After some tests of thermal field inside vertical cavity, the “two-velocity-scale wall function” is chosen to be used with high-Re models. In both vertical and inclined cases the overall flow pattern appears similar, with a single circulation cell, and a boundary layer at the wall. The levels of turbulence energy are generally slightly lower in the inclined case. Most models give a reasonable prediction of measured Nusselt number, with the two low-Re approaches generally being closer to the data than the schemes employing wall functions. For the 15° inclined cavity, a multi cellular motion is shown by the high-Re models. Nevertheless, all the model predictions disagree with experimental data due to the presence in real flow of 3-D unsteady structures as found in Benard convection problems. These cannot, definitely, be reproduced using a 2-D geometry. Both highly resolved LES and unsteady RANS computations are then conducted, for turbulent natural convection of air inside 15° unstably and stably stratified cavities. In accordance with recent experimental data, the LES computations for both enclosures returned three-dimensional time-averaged flow fields. In the case of the unstably stratified enclosure, the flow is highly unsteady with coherent turbulent structures in the core of the enclosure. Results of LES computations show close agreement with the measured data. Subsequent comparisons of different URANS schemes with the present LES are used in order to explore to what extent these models are able to reproduce the large-scale unsteady flow structures. All URANS schemes have been found to be able to reproduce the 3-D unsteady flow features present in the 15° unstable cavity. However, the low-Re model tested as well as requiring a high resolution near-wall grid, also needed a finer grid in the core region than the high-Re models, thus making it computationally very expensive. Flow within the 15° stable cavity also shows some 3-D features, although it is significantly less unsteady, and the URANS models tested here have been less successful in reproducing this flow pattern. Finally, natural convection of CO2 inside a horizontal annular penetration enclosure, which can be found in AGR's, has been performed using a highly resolved LES and a set of RANS models. The Rayleigh number is 1.5×10e9. RANS models agree with the present LES on the fact that the flow is unsteady and there are large-scale oscillations present which decrease in amplitude as one moves from the open towards the closed end of the annular enclosure. Overall heat transfer and thermal quantitative and dynamic results show that RANS schemes are in close agreement with the current LES data except some discrepancies shown by the high-Re model which can be returned to the limitation of the simple wall function used to predict such complex flow.

After the end of the NASA space shuttle programme, there has been resurgence of interest in developing a single stage-to-orbit spacecraft. The key technology to realize this dream is the airbreathing scramjet engine. The scramjet concept has been around for decades, but much work is still needed in order to eliminate the remaining obstacles to develop a practical working prototype of the engine. Many such obstacles are related to the inlet which functions as the main compression unit for the engine. Typically, a high speed inlet is designed to function properly in a single flight condition. Such an inlet would experience adverse flow conditions related to various shock-shock interactions, viscous effects, shock-boundary layer interactions, and many other flow phenomena at off-design conditions. The traditional mechanism to mitigate the adverse flow conditions is by varying the inlet geometry at off-design conditions. There are still gaps in understanding the behaviour of inlets at off-design conditions and the effectiveness of variable geometry as inlet flow control. This is partly due to complex flow diagnostics setup, which limits the type, quantity and quality of information that can be extracted from the inlet flow. The first objective of this thesis was to develop a global inlet measurement system that can provide an abundance of information on inlet flow. The pressure sensitive paint method was employed together with other methods to provide comprehensive understanding on inlet flow characteristics. Calculation of Mach number at the isolator exit using the isolator sidewall pressure map was successfully demonstrated. The measurement of Mach number at the isolator exit has allowed for performance of the inlet to be calculated without the need for intrusive flow diagnostics tools used by previous researchers. The global measurement system was then employed to investigate the characteristics of the scramjet inlet operating at various off-design conditions. Complex shock structures were observed at the inlet cowl entrance as the angle-of-attack was increased. The relationship of flow quality and inlet performance was examined and discussed. General improvements on the inlet performance were obtained if the size of separation on the compression ramp was reduced. The inlet was also observed to perform poorly when compression shocks impinged on the inner cowl surface. Cowl deflections were demonstrated to be effective in controlling the internal flow of the inlet and improving its performance. An exploratory study on the role of micro-vortex generators to control boundary layer separation on scramjet inlets has been included as well. Strategies for optimizing an inlet at off-design conditions were analysed, and it was found that any variable geometry combination must maintain high throat-to-freestream Mach number ratio in order to preserve high inlet performance.

This thesis provides an assessment of various computational strategies for modelling the turbulent flow and heat transfer around in-line tube banks. The research has direct application to the heat exchanger of an Advanced Gas-cooled Reactor (AGR). The suitability and accuracy of different Computational Fluid Dynamic (CFD) techniques were investigated first on generic square in-line tube banks where experimental data are available. The assumption of flow periodicity in all three Cartesian directions is initially investigated whereby the domain size was varied. Wall-resolved Large Eddy Simulations (LES) predict an increasing flow asymmetry with decreasing tube spacing. Two dimensional (2D) and three dimensional (3D) Unsteady Reynolds Averaged Navier-Stokes (URANS) models were simulated at the tube spacing known to be close to the flow pattern transition from symmetric to asymmetric. Marked differences were observed between the flow pattern predicted by turbulence models resolving the boundary layer and those that rely on wall functions. Ultimately, an improved understanding of the flow physics and heat transfer mechanisms encountered within in-line tube banks was gained. The assumption of flow periodicity was then removed and the effects of confining walls were investigated by reproducing experimental conditions. The correct pressure forces and heat transfer around the central tubes could only be accurately predicted when the walls in the crossflow direction were modelled. The inclusion of walls in the spanwise direction gave rise to small flow asymmetries which have been reported on similarly-spaced in-line tube banks. The latter half of the thesis focuses on the reasons for the enhanced thermal mixing and 3D secondary flow patterns observed in the in-line section of the AGR heat exchanger. A wall-resolved periodic LES was conducted at the lower Reynolds number of 11,000 along with URANS calculations of the full experimental conditions at both Reynolds numbers 11,000 and 66,000. These calculations required the use of High Performance Computing (HPC) facilities. Large 3D secondary flow structures were predicted that produced the same level of crossflow temperature drifting as that reported experimentally. Multiple upward and downward flow paths were observed which qualitatively explained why the experimental temperature profiles reported at different spanwise locations indicated multiple spirals (or secondary vortices). Quantification of the levels of thermal diffusion were investigated using both decaying temperature spikes and blanked tube platens. Thus the CFD provided recommendations about the thermal diffusivity assumptions used by the AGR heat exchanger code.

The generation of structured grids around complex geometries is generally a difficult task. Thistask is typically a major bottleneck in the overall solution procedure; however, the overset gridmethod can be used to relieve much of this burden. An overset grid consists of a set of simplecomponent grids, which can overlap arbitrarily (provided there is sufficient overlap to interpolatefrom). The union of all simple grids should then delineate the global domain. This allows complex domains to meshed using a series of simple meshes. Interpolation boundary conditions are enforced at internal boundaries to ensure a continuous solution. Standard tri-linear interpolation is typically used for this purpose, although there are alternative methods that attempt to enforce global conservation. A new CFD code has been developed that incorporates the overset grid method in three spatialdimensions. This code uses the steady state, finite volume discretisation method. SIMPLE pressure velocity coupling has been used on a colocated grid with Rhie-Chow interpolation for face velocities. Different interpolation methods have been compared for the information transfer at internal boundaries from one grid to the next. It has been shown that for a variety of test cases, continuous and accurate solutions are obtained from one grid to another, which are comparable to those of the single-block or block-structured solutions, or to experimental data (where available). A new hole cutting algorithm and bulk correction outlet condition are presented. Improvements to existing digital tree data structures are also described. Lid driven cavity flow, the flow around rotating cylinders, and flow impingement onto a concavesurface are considered in order to demonstrate the method. The flow over a backward facing step, over a multi-element airfoil, through a bifurcating artery and over a wing-body junction are then considered (with experimental comparison). This demonstrates the range of applicability of the method. In all cases, the overset method offers significant advantages over block-structured solutions that are available in the literature. It is shown that greater numerical efficiency is generally achievable via the use of an overset simulation: Since the gridding is flexible, high aspect ratio cells need not propagate into the domain (as is often the case for a block-structured arrangement). Also, much of the domain away from localised regions of geometrical complexity can be meshed with efficientCartesian grids.

Conjugate heat transfer is a crucial issue in a number of turbulent engineering fluidflow applications, particularly in nuclear engineering and heat exchanger equipment. Temperature fluctuations in the near-wall turbulent fluid lead to similar fluctuationsin the temperature of the solid wall, and these fluctuations in the solid cause thermalstress in the material which may lead to fatigue and finally damage. In the present study, the Reynolds Average Navier-Stokes (RANS) modelling approachhas been adopted, with four equation k−ε−θ2−εθ eddy viscosity based modelsemployed to account for the turbulence in the fluid region. Transport equations forthe mean temperature, temperature variance, θ2, and its dissipation rate, εθ, have beensimultaneously solved across the solid region, with suitable matching conditions forthe thermal fields at the fluid/solid interface. The study has started by examining the case of fully developed channel flow withheat transfer through a thick wall, for which Tiselj et al. [2001b] provide DNS dataat a range of thermal activity ratios (essentially a ratio of the fluid and solid thermalmaterial properties). Initial simulations were performed with the existing Hanjali´cet al. [1996] four-equation model, extended across the solid region as described above. However, this model was found not to produce the correct sensitivity to thermal activityratio of the near wall θ2 values in the fluid, or the decay rate of θ2 across the solid wall. Therefore, a number of model refinements are proposed in order to improve predictionsin both fluid and solid regions over a range of thermal activity ratios. These refinementsare based on elements from a three-equation non-linear EVM designed to bring aboutbetter profiles of the variables k, ε, θ2 and εθ near the wall , and their inclusion is shownto produce a good matching with the DNS data of Tiselj et al. [2001b].Thereafter, a further, more complex test case has been investigated, namely an opposedwall jet flow, in which a hot wall jet flows vertically downward into an ascendingcold flow. As in the channel flow case, the thermal field is also solved across the solidwalls. The modified model results are compared with results from the Hanjali´c modeland LES and experimental data of Addad et al. [2004] and He et al. [2002] respectively. In this test case, the modified model presents generally good agreement with the LESand experimental data in the dynamic flow field, particularly the penetration point ofthe jet flow. In the thermal field, the modified model also shows improvements in the θ2predictions, particularly in the decay of the θ2 across the wall, which is consistent withthe behaviour found in the simple channel flow case. Although the modified model hasshown significant improvements in the conjugate heat transfer predictions, in some instancesit was difficult to obtain fully-converged steady state numerical results. Thusthe particular investigation with the inlet jet location shows non-convergence numericalresults in this steady state assumption. Thus, unsteady flow calculations have beenperformed for this case. These show large scale unsteadiness in the jet penetration area. In the dynamic field, the total rms values of the modelled and mean fluctuations showgood agreement with the LES data. In the thermal field calculation, a range of the flowconditions and solid material properties have been considered, and the predicted conjugateheat transfer predicted performance is broadly in line with the behaviour shownin the channel flow.

The reliable prediction of turbulent non-equilibrium flows is of high academic and industrial interest in several engineering fields. Most turbulent flows are often predicted using single-time-scale Reynolds-Averaged-Navier-Stokes (RANS) turbulence models which assume the flows can be modelled through a single time or length scale which is an admittedly incorrect assumption. Therefore they are not expected to capture the lag in the response of the turbulence in non-equilibrium flows. In attempts to improve prediction of these flows, by taking into consideration some features of the turbulent kinetic energy spectrum, the multiple-time-scale models arose. A number of two-scale models have been proposed, but so far their use has been rather limited.This work thus focusses on the development of two-time-scale approaches. Two two-time-scale linear-eddy-viscosity models, referred to as NT1 and NT2 models, have been developed and the initial stages of the development of two-time-scale non-linear-eddy-viscosity models are also reported. The models' coefficients have been determined through asymptotic analysis of decaying grid turbulence, homogeneous shear flows and the flow in a boundary layer in local equilibrium. Three other important features of these models are that there is consistent partition of the large and the small scales for all above limiting cases, model sensitivity to the partition and production rate ratios and sensitivity of the eddy viscosity sensitive to the mean strain rates.The models developed have been tested through computations of a wide range of flows such as homogeneous shear and normally strained flows, fully developed channel flows, zero-pressure-gradient, adverse-pressure-gradient, favourable-pressure-gradient and oscillatory boundary layer flows, fully developed oscillatory and ramp up pipe flows and steady and pulsated backward-facing-step flows.The proposed NT1 and NT2 two-scale models have been shown to perform well in all test cases, being, among the benchmarked models tested, the models which best performed in the wide range of dimensionless shear values of homogeneous shear flows, the only linear-eddy-viscosity models which predicted well the turbulent kinetic energy in the normally strained cases and the only models which showed satisfactory sensitivity in predicting correctly the reattachment point in the unsteady backward facing step cases with different forcing frequencies. Although the development of the two-time-scale non-linear-eddy-viscosity models is still in progress, the interim versions proposed here have resulted in predictions of the Reynolds normal stresses similar to those of much more complex models in all test cases studied and in predictions of the turbulent kinetic energy in normally strained flows which are better than those of the other models tested in this study.

Buoyant flows are present in nature and also in many engineering applications,from domestic heating to the cooling of nuclear power plants. This experimental study focuses on the effects of angle of inclination on buoyancy-driven flows inside tall, rectangular, differentially-heated cavities. The objective is to produce detailed local flow and thermal data, which will advance our understanding of the flow physics and also provide CFD validation data. It considers a 2.18m × 0.52m × 0.0762m cavity, resulting in an aspect ratio of 28.6, with its two opposing long walls maintained at constant but different temperatures, while all the remaining walls are thermally insulated. The Rayleigh number, based on the temperature difference and spacing of the long sides, is 0.86 x 106 for most cases and the working fluid is air (Prandtl number0.71). Experimental data for the flow and the thermal fields, using laser Doppler anemometry and Chromel-Alumel thermocouple traverses respectively, are presented for the cavity inclined at 60° and 15° to the horizontal, for both stable (the hot surface being the upper surface) and unstable (the hot surface the lower one) orientations. The 15° stable case is investigated at a higher Rayleigh number of 1.54 x106 and some additional data for the 15° unstable case are also presented at this high value of Rayleigh number. Comparisons with the measurements of Betts and Bokhari [1], for the same cavity at the vertical position, are also included. For moderate angles of inclination, the flow is two-dimensional and the effects of inclination are primarily confined to the fluctuating fields. For large angles of inclination, the flow becomes three-dimensional. In the unstable 15° angle of inclination case, a set of four longitudinal vortices are formed over the entire length of the cavity, with four counter-rotating re-circulation cells within the cross-section parallel to the thermally active walls. The enhanced mixing at 15° unstable inclination leads to uniform temperature in the cavity core and thus only minor deviations from two dimensionality in the thermal field. A modest rise in Rayleigh number, in the 15° unstable case, does not affect the mean motion, but causes an increase in the normalised turbulence intensities, which in turn leads to a more uniform temperature within the cavity core and a practically two-dimensional thermal field. The stable 15° angle of inclination case, surprisingly, leads to the formation of two longitudinal vortices and two re-circulation cells. The lack of mixing, in the 15° stable case, leads to more noticeable three-dimensional thermal field. The thesis includes a full set of flow and thermal predictions and also spectral analysis of thermal fluctuations, which show a significant effect of the angle of inclination on both the power density level and the ranges of frequencies involved.

This Thesis describes the numerical simulation of fluid-structure interaction (FSI) problems. A finite-volume based stress analysis code was developed and coupled to an existing in-house CFD code to form a general purpose FSI solver capable of being used with the advanced turbulence and near-wall models developed within the research group. The code has been used to study a number of physiological flows in the present work, although the general nature of the solver allows it to be used for other applications also. By using the same numerical method, implemented in a consistent manner, for both fluid and solid domains, the inefficiencies associated with using separate packages for the fluid and solid were avoided. Separate packages typically store information in different data structures; some form of software interface is required to transfer information between the two packages. This additional software layer, which is called during each FSI iteration, causes a considerable overhead. By using a single numerical mesh across both domains, the inaccuracies associated with boundary interpolation were also avoided. Typically, separate packages use meshes which do not conform at their common boundary. In order to find nodal values of the fluid pressure, say, at the solid nodes, some form of interpolation is necessary. The interpolation leads to the introduction of truncation errors. These improvements allow for more accurate and efficient FSI simulations, particularly transient cases, to be performed. The solid solver was verified against analytical solutions for a number of test cases, including: planar stress distribution in a square plate with a circular hole in the centre; axisymmetric stress in a thick walled cylinder under internal pressure, and unsteady displacement of a cantilevered beam under free vibration. Before coupled analyses were performed, the flow solver was also validated through a number of rigid walled test cases, including steady flow through a stenosed tube and unsteady flow through an aneurysm. Many physiological flows are difficult to capture due to flow separation and early transition to turbulence. The use of a low-Reynolds number k-ε turbulence model was successful at capturing the flow field over a range of physiologically relevant flow rates. Once the solid body and flow solvers had been validated in isolation, they were coupled together and applied to a number of physiological flows, namely: steady flow through an initially straight tube with a compliant wall; steady flow through a compliant stenosis, and unsteady flow through a compliant aneurysm. The results from all three test cases showed good agreement with the available experimental and numerical data in terms of wall deformation. The solid body solver also proved itself to be capable of producing high quality numerical meshes for use in other simulations. The fluid mesh was considered to be a solid body with arbitrary material properties; the required deformation was specified as prescribed displacement boundary conditions. The main benefit of this method, compared to simple elliptical grid generation methods, is that near-wall grid spacing was preserved throughout the coupled simulation.