Ces travaux de doctorat portent sur la réduction du bruit d'origine aérodynamique émis par les ventilateurs et les doublets d'hélices contra-rotatifs. La méthodologie proposée consiste à intégrer des méthodes rapides et précises de prédiction des niveaux sonores dans le processus de conception. Cette thématique a vu son intérêt augmenter depuis que l'Union Européenne a restreint les limites d'exposition au bruit en milieu de travail et dans les zones habitées à proximité des aéroports. Parmi les méthodes numériques employées en aéroacoustique, les méthodes hybrides de prédiction du bruit sont considérées comme particulièrement appropriées pour la conception automatisée du fait de leur coût modéré en temps de calcul. Ces méthodes séparent la résolution de l'écoulement aérodynamique de celle de la génération du bruit et de sa propagation en champ lointain. L'écoulement aérodynamique est obtenu par simulation numérique, tandis que l'acoustique est traitée par méthodes analytiques. Ces méthodes analytiques développées et validées pour déterminer le bruit d'un profil aérodynamique placé dans un écoulement turbulent seront étendues pour traiter le réponse acoustique de pales en rotation. Ces travaux se concentrent sur deux configurations de ventilateurs basses vitesses. La première configuration traitée est le doublet d'hélices contra-rotatif de 4.2m de diamètre de la soufflerie L-1 de l'Institut von Karman (VKI). Ce système permet d'étudier le phénomène de bruit tonal et à large bande dû à l'impact des sillages turbulents, générés par l'hélice amont, sur l'hélice aval. La deuxième configuration traitée est un ventilateur à quatre pales du CETIAT (France) installé seul dans un large plenum. Ce système permet d'étudier le bruit propre ou bruit de bord de fuite causé par l'interaction des tourbillons générés par l'écoulement autour de la pale avec le bord de fuite de la pale. Pour cette configuration, des données expérimentales sont rendues disponibles dans le cadre d'un projet commun entre le VKI et le CETIAT. Les méthodes hybrides sont développées et mises en oeuvre pour ces deux mécanismes de bruit présents dans les deux configurations de ventilateur. L'objectif de ces travaux de thèse est d'employer les méthodes hybrides ainsi calibrées et validées pour réaliser l'optimisation du doublet d'hélices contra-rotatif de la soufflerie L-1. Le coeur de ces travaux portera sur l'extension des méthodes hybrides pour la prédiction du bruit d'un profil dans un écoulement turbulent uniforme au cas du bruit tonal et à large bande d'interaction de sillages et du bruit à large bande de bord de fuite dans des ventilateurs basses vitesses. Il sera montré qu'il est possible de déterminer le spectre de bruit de manière rapide et précise en s'appuyant sur la connaissance du champ aérodynamique dont les quantités seront extraites de simulations numériques stationnaires (RANS) pour alimenter la formulation analytique retenue. Cette dernière doit être adaptée au mécanisme de bruit étudié, à savoir l'interaction d'une pale de ventilateur avec un sillage ou celle du bord de fuite avec la turbulence qui s'est développé le long de la pale. Les deux mécanismes de bruit sont d'abord modélisés avec des fonctions analytiques qui sont calibrés avec les données des simulations numériques. Les modèles de sources de bruit ainsi que les estimations finales de spectre de bruit sont comparées aux données expérimentales disponibles et à des simulations directes. Enfin la méthodologie retenue est mise en oeuvre dans le cadre de l'optimisation du doublet d'hélices L-1 au moyen d'un algorithme génétique. L'étude détaillée de la sensibilité des paramètres et des contraintes de l'optimisation apporte un nouveau regard sur l'optimisation multi-objectif efficacité-bruit qui sera de plus en plus utilisée pour la conception de turbomachine dans le futur. / Abstract : The context of this thesis is the reduction of noise emitted by ventilation fans and aeronautical counter-rotating open rotors, which will be achieved by implementing fast and accurate noise prediction methods in the design process. The interest towards this subject has increased since the European Union enforced lower limits of exposure to noise in work environments and also to environmental noise in the proximity of airports. In the field of computational aeroacoustics, hybrid methods for noise prediction are considered particularly suitable for use in an automated design procedure due to their low computational cost. In fact they split the description of the flow field, which is made by computational fluid dynamics, from the quantification of the source of noise and of its propagation, obtained by using analytic formulations. Such analytic methods have already been used successfully for the prediction of the noise emitted by an airfoil placed in a turbulent flow; it is therefore natural to try to extend their applicability to the case of rotating blades. Two application cases have been chosen for this thesis. The first one is the 4.2 m diameter counter-rotating fan of the von Karman Institute (VKI) L1 low-speed wind tunnel, which is used to study the phenomenon of wake-interaction tonal and broadband noise. The second application case is a four-bladed low-speed ventilation fan in which the dominant source of noise is the trailing-edge or self-noise caused by the turbulent eddies passing over the trailing-edge of the blade. In this case, an experimental database has been made available by CETIAT, France, in the framework of a collaborative project with VKI. The final step of the project will be to use the prediction codes developed for both the noise phenomena in the geometric optimization of the L1 counter-rotating fan. The fundamental question that will be addressed in the thesis is how to extend the hybrid CFD-analytic methods to predict noise from an airfoil in a uniform turbulent flow to the case of tonal and broadband wake-interaction noise and trailing-edge broadband noise in low-speed fans. It will be shown that it is possible to provide a fast and reasonably accurate prediction of the spectrum of noise emitted by low-speed fans by extracting flow data from Reynolds Averaged Navier-Stokes (RANS) simulations and using them as input to Amiet's analytic formulation, provided that this has been carefully adapted to the studied noise generation phenomenon, i.e. the interaction of the leading-edge of a fan blade with an incoming wake or of the trailing-edge with the turbulent boundary layer over the blade surface. Concerning the methodology, both noise generation mechanisms will first be modeled with analytic functions, then the necessary flow field input will be extracted from RANS simulations and the models will be validated with respect to experimental data, whenever possible, or to higher fidelity simulations. The last step of the project is the application of these noise prediction methods to the shape optimization of the L-1 fan blades by means of a genetic algorithm. The sensitivity analysis of the design parameters and of the constraints used in the optimization process provides a new perspective on the multi-objective efficiency-noise optimization approach which will be increasingly used in turbomachinery design in the future.

Diel vertical migrations (DVM) comprise the largest animal migration on the planet and are a phenomenon present in all bodies of water on Earth. A strong sound scattering layer undergoing DVM was observed in the Straits of Florida via a bottom-mounted Acoustic Doppler current profiler (ADCP) Workhorse Longranger 75 kHz (Teledyne RD Instruments) located at the 244 m isobath. ADCP average backscatter showed a clear periodicity corresponding with sunrise and sunset times indicating the presence of a nocturnal DVM. Analysis of the ADCP backscatter data indicated zooplankton swimming velocities were faster during sunrise than sunset times. In several cases the zooplankton swimming velocity appeared to be faster at the beginning of the descent, after which the swimming velocity decreased. Analysis of ADCP velocity data indicated a measureable decrease in the northward component of the current velocity field during migrations (sunrise and sunset) compared to three hours prior. This was presumably associated with an increase in drag due to turbulent friction associated with DVM. A non-hydrostatic computational fluid dynamics (CFD) model with injection of Lagrangian particles was utilized to simulate the effects of DVM on the velocity field and turbulence signature of the Florida Current. A domain simulating a section of the Florida Current was created and zooplankton were represented by particle injection with a discrete phase model. The model was run with and without particles, holding all other parameters the same, for comparison. Idealized temperature stratification and velocity profiles were set for both summer and winter conditions to observe seasonal differences. For each case, velocity and turbulence with particles were compared to results without particles to confirm the changes in profiles were due to the zooplankton (Lagrangian particles). In several cases there was an observable change in average x-velocity profiles due to the injection of particles into the domain. In all cases there was an observable increase in subgrid turbulent viscosity in the wake of the injected particles. This effect was much stronger in the winter case, most likely due to stratification of the water column which gave a near critical Richardson number. These results indicated that DVM does in fact have an effect on the velocity profile and turbulence signature in a strong current under certain conditions and that there was a seasonal difference due to stratification profiles.

Moving contact line problems appear in many natural and industrial processes. A contact line is formed where the interface between two immiscible fluids meets a solid wall. Examples from everyday life include raindrops falling on a window and water bugs resting on water surfaces. In many cases the dynamics of the contact line affects the overall behavior of the system. Industrial applications where the contact line behavior is important include gas and oil recovery in porous media, lubrication, inkjet printing and microfluidics. Computer simulations are fundamental tools to understand and predict the behavior.   In this thesis we look at numerical simulations of dynamic contact line problems. Despite their importance, the physics of moving contact lines is poorly understood. The standard Navier-Stokes equations together with the conventional no-slip boundary condition predicts a singularity in the shear stresses at the contact line. Atomistic processes at the contact line come into play, and it is necessary to include these processes in the model to resolve the singularity. In the case of capillary driven flows for example, it has been observed that the microscopic contact line dynamics has a large impact on the overall macroscopic flow. In Paper I we present a new multiscale model for numerical simulation of flow of two immiscible and incompressible fluids in the presence of moving contact points (i.e. two-dimensional problems). The paper presents a new boundary methodology based on combining a relation between the apparent contact angle and the contact point velocity, and a similarity solution for Stokes flow at a planar interface (the analytic Huh and Scriven velocity). The relation between the angle and the velocity is determined by performing separate microscopic simulations. The classical Huh and Scriven solution is only valid for flow over flat walls. In Paper II we use perturbation analysis to extend the solution to flow over curved walls. Paper III presents the parallel finite element solver that is used to perform the numerical experiments presented in this thesis. Finally, the new multiscale model (presented in Paper I) is applied to a relevant microfluidic research problem in Paper IV. For this problem it is very important to have a model that accurately takes the atomistic effects at contact lines into account.

Improving the hydraulic performance of waste stabilization ponds (WSPs) is an important management strategy to not only ensure protection of public health and the environment, but also to maximize the potential reuse of valuable resources found in the treated effluent. To reuse effluent from WSPs, a better understanding of the factors that impact the hydraulic performance of the system is needed. One major factor determining the hydraulic performance of a WSP is sludge accumulation, which alters the volume of the pond. In this study, computational fluid dynamics (CFD) analysis was applied to investigate the impact of sludge layer geometry on hydraulic performance of a facultative pond, typically used in many small communities throughout the developing world. Four waste stabilization pond cases with different sludge volumes and distributions were investigated. Results indicate that sludge distribution and volume have a significant impact on wastewater treatment efficiency and capacity. Although treatment capacity is reduced with accumulation of sludge, the latter may induce a baffling effect which causes the flow to behave closer to that of plug flow reactor and thus increase treatment efficiency. In addition to sludge accumulation and distribution, the impact of water surface level is also investigated through two additional cases. Findings show that an increase in water level while keeping a constant flow rate can result in a significant decrease in the hydraulic performance by reducing the sludge baffling effect, suggesting a careful monitoring of sludge accumulation and water surface level in WSP systems.

Magister Scientiae - MSc / Nuclear reactors release small amounts of radioactivity during their normal operations. The most common method of calculating the dose to the public that results from such releases uses Gaussian Plume models. We are investigating these methods using CAP88-PC, a computer code developed for the Environmental Protection Agency (EPA) in the USA that calculates the concentration of radionuclides released from a stack using Pasquill stability classification. A buoyant or momentum driven part is also included. The uptake of the released radionuclide by plants, animals and humans, directly and indirectly, is then calculated to obtain the doses to the public. This method is well established but is known to suffer from many approximations and does not give answers that are accurate to be better than 50% in many cases. More accurate, though much more computer-intensive methods have been developed to calculate the movement of gases using fluid dynamic models. Such a model, using the code FLUENT can model complex terrains and will also be investigated in this work. This work is a preliminary study to compare the results of the traditional Gaussian plume model and a fluid dynamic model for a simplified case. The results indicate that Computational Fluid Dynamics calculations give qualitatively similar results with the possibility of including much more effects than the simple Gaussian plume model. / South Africa

The main objective of this Master Thesis project is the development of a new methodology to perform Computational Fluid Dynamics (CFD) conjugate heat transfer simulations for internal combustion engines, at the Fluid and Combustion Simulations Department (NMGD) at Scania CV AB, Södertalje, Sweden. This new method allows to overcome the drawbacks identified in the former methodology, providing the ability to use the more advanced polyhedral mesh type to generate good quality grids in complex geometries like water cooling jackets, and integrating all the different components of the engine cylinder in one unique multi-material mesh. In the method developed, these advantages can be used while optimizing the process to perform the simulations, and obtaining improved accuracy in the temperature field of engine components surrounding the water cooling jacket when compared to the experimental data from Scania CV AB tests rigs. The present work exposes the limitations encountered within the former methodology and presents a theoretical background to explain the physics involved, describing the computational tools and procedures to solve these complex fluid and thermal problems in a practical and cost-effective way, by the use of CFD.A mesh sensitivity analysis performed during this study reveals that a mesh with low y+ values, close to 1 in the water cooling jacket, is needed to obtain an accurate temperature distribution along the cylinder head, as well as to accurately identify boiling regions in the coolant domain. Another advantage of the proposed methodology is that it provides new capabilities like the implementation of thermal contact resistance in periodical contact regions of the engine components, improving the accuracy of the results in terms of temperature profiles of parts like valves, seats and guides. The results from this project are satisfactory, providing a reliable new methodology for multi-material thermal simulations, improving the efficiency of the work to be performed in the NMGD department, with a better use of the available engineering and computational resources, simplifying all the stages of multi-material projects, from the geometry preparation and meshing, to the post-processing tasks.

Industrial tray air-dryers are increasingly used for the drying of agricultural products. The main drawback of these dryers is the non-uniform velocity distribution in the drying zone resulting in a non-uniform drying of the product. Computational Fluid Dynamics (CFD) software was implemented to predict and decrease the non-uniform velocity distribution of various dryer configurations. Tunnel dryers in commercial use were used to obtain experimental data. The CFD results were correlated with the test data. Trolley and tray tunnel dryers provide a relatively simple, low capital intensive and versatile method for drying a wide range of products. Artificial drying has the advantage of controlled drying conditions compared to traditional sun drying. The main focus of every tunnel design should be the improvement of the quality of the product in terms of colour, texture and aroma. Increasing the evaporation rate without increasing the energy required to do so, should always be done in-line with this main objective. Many studies focus on the mango structure and food dehydration principles that influence the uniform drying product with the assumption that the airflow over the produce is uniform. Few have been conducted on the air movement inside industrial dryers. CFD analysis predicts the airflow without influencing the airflow pattern compared to the measuring equipment inside test dryers. The experimental data were obtained from an empty dryer without a flow diverter. This was compared to dryer with the flow diverter included and compared to a dryer with the trolleys, trays and mango slices included. The test results showed that turbulence created by this configuration, still played a major role in the nonuniform velocity distribution along the drying zone of the tunnel. The inclusion of a flow diverter did however dampen the swirl effect of the main fan. Measuring the velocity distribution was practically difficult with the handheld devices used, which influenced the accuracy of the measurements taken. This justified the CFD analysis in order to better visualise and predict the airflow pattern inside the dryer. The total average speed CFD results of the sections in the drying zone (without mangoes and trolleys) of the dryer without a flow diverter was 11.2% higher compared to the test results. It was 14% higher for the dryer with the flow diverter included. The dryer with the mangoes, trays, trolleys and flow diverter showed a large difference where the total average speed of the CFD analysis was 49% higher compared to the test results. The main reason for the difference of the CFD analysis compared to the measured results are the factors that influenced the uncertainty of the experimental set up. The CFD analysis showed that the coefficient of variance (CV) of the dryer with the flow diverter (mangoes and trolleys included) was 3% lower compared to the dryer without one. Various dryer configurations were analysed using the CFD software to investigate what the best combination of flow diverter, vanes and blanking-off plates would be. A dryer configuration where flow diverters (Up-and-downstream of the main fan) above the false ceiling and inside the drying zone was analysed. A 16% decrease in terms of the CV value was obtained compared to the dryer with just the flow diverter downstream of main fan above the false ceiling. There was however a large region of swirl upstream of the main above the false ceiling resulting in a larger loss of heated air through the outlet fan before it reached the drying zone. The cost of manufacturing a simple vane and flow diverter for an existing dryer is 4% of the initial building costs (excluding the initial cost of the trolleys). The overall drying uniformity of this dryer is improved according to the CFD analysis by 7%. A cost analysis (taking into account the 15 year life cycle of a dryer) in terms of the energy requirement to evaporate water from the drying zone, showed that the dryer with the flow diverter was 6% less expensive to run on a yearly basis. Labour costs will be lower due to man-hours saved in terms of sorting out the wet slices from the dried product. Resources (dryers and trolleys) that would have been used for re-drying the wet produce, could now be implemented to increase the production rate of the plant. Copyright / Dissertation (MEng)--University of Pretoria, 2010. / Mechanical and Aeronautical Engineering / unrestricted

Asset protection in the form of fluid filtration makes up an ever-increasing part of the civilized and industrialized world. Fluid filtration applications in the conditioned environment and life sciences are affording the world’s population a chance to better realize their potential, while industrial applications help ensure that high demand processes can be carried out safely, reliably, and effectively. In the present work, a tool has been developed, using the computational fluid dynamics package FLUENT, to allow the designer to better predict the magnitude of geometric imperfections within a given pleat configuration. Pleated rectangular filters, intended to improve the quality of air for human occupants, with a U-shaped pleat form have been chosen as the focus of this study. A simulation study is developed to investigate the maximum local velocity normal to the filtration surface and to characterize the magnitude of the pleatwise velocity distribution across a range of pleated geometries and flow conditions. The geometry of the U-shaped pleat form can be characterized by, amongst other parameters, the width of the pleat channel, the overall height of the individual pleat, as well as the thickness of the filtration medium. The various geometries of the current study were developed by changing the width of the pleat channel, as well as the channel height, while keeping the medium thickness constant throughout. Changing the width of the pleat channel allows the designer to achieve varying pleat densities, expressed as a number of pleats along a one inch section of the overall pleated pack. Pleat densities of 6.5, 7, 7.5, 8, and 8.5 pleats per inch are considered in the current study. Pleat heights of 1.0, 0.75, and 0.50 inches are also investigated in the current study. Furthermore, the filter velocity can be characterized by the free stream velocity at the face of the filter pack, termed the face velocity, and by the velocity of the fluid at the interface with the filtration medium, referred to as media velocity. In the present work, the face velocity was adjusted in each case to achieve the desired media velocities across the study, which are 10.5, 9.0, 7.5, and 6.0 feet per minute. In an effort to more clearly communicate the results of the study, the results are presented in the form of a non-dimensionalized plots which present the designer with a way to quickly gauge the effect of pleat geometry on maximum velocity. Additionally, two tools are presented to aid the designer in more accurately predicting the maximum filtration velocity. These tools are then evaluated for effectiveness using the method of absolute relative percent error. The assumption of uniform flow through the filtration media leads to an average absolute relative percent error of 27%. The first tool the reader is presented with is a simple correction factor which predicts the maximum filtration velocity with an average absolute relative percent error of 10% over the study domain. The second tool, which takes a slightly more complicated y-intercept form, characterizes the maximum filtration velocity as a function of average velocity and aspect ratio. This approach further reduces the average absolute relative percent error to 4%. The results of the simulation herein are successfully employed to develop a set of simple yet effective tools that allow the filter designer to more accurately predict maximum velocities through a pleated air filter.

This work details modelling buoyancy-driven viscous flow and heat transfer through heterogeneous saturated packed pebble beds via a set of volume-averaged conservation equations in which local thermal disequilibrium is accounted for. The latter refers to the two phases considered viz. solid and fluid, differing in temperature. This is effected by describing each phase with its own governing equation. Further to the aforementioned, the governing equation set is written in terms of intrinsic volume-averaged material properties that are fully variant with respect to temperature. The heterogeneous solid phase is described with a porosity field varying from 0.39 to 0.99. The intent of the stated upper bound is to explicitly model typical packed bed near-wall phenomena such as wall-channelling and pebble-wall heat transfer as true to reality as possible, while maintaining scientific rigour. The set of coupled non-linear partial differential equations is solved via a locally preconditioned artificial compressibility method, where spatial discretisation is effected with a compact finite volume edge-based discretisation method. The latter is done in the interest of accuracy. Stabilisation is effected via JST scalar-valued artificial dissipation. This is the first instance in which an artificial compressibility algorithm is applied to modelling heat and fluid flow through heterogeneous porous materials. As a result of the aforementioned, calculation of the acoustic velocities, stabilisation scaling factors and allowable time-step sizes were revised. The developed technology is demonstrated by application to the modelling of SANA test cases, i.e. natural convective flow inside a heated porous axisymmetric cavity. Predicted results are shown to be within 12% of experimental measurements in all cases, while having an average deviation of only 3%. / Dissertation (MEng)--University of Pretoria, 2008. / Mechanical and Aeronautical Engineering / unrestricted

Large Eddy Simulation (LES) has potential to address unsteady phenomena in turbulent premixed flames and to capture turbulence scales and their influence on combustion. Thus, this approach is gaining interest in industry to analyse turbulent reacting flows. In LES, the dynamics of large-scale turbulent eddies down to a cut-off scale are solved, with models to mimic the influences of sub-grid scales. Since the flame front is thinner than the smallest scale resolved in a typical LES, the premixed combustion is a sub-grid scale (SGS) phenomenon and involves strong interplay among small-scale turbulence, chemical reactions and molecular diffusion. Sub-grid scale combustion models must accurately represent these processes. When the flame front is thinner than the smallest turbulent scale, the flame is corrugated by the turbulence and can be seen as an ensemble of thin, one-dimensional laminar flames (flamelets). This allows one to decouple turbulence from chemistry, with a significant reduction in computational effort. However, potentials and limitations of flamelets are not fully explored and understood. This work contributes to this understanding. Two models are identified, one based on an algebraic expression for the reaction rate of a progress variable and the assumption of fast chemistry, the other based on a database of unstrained flamelets in which reaction rates are stored and parametrised using a progress variable and its SGS variance, and their potentials are shown for a wide range of premixed combustion conditions of practical interest. The sensitivity to a number of model parameters and boundary conditions is explored to assess the robustness of these models. This work shows that the SGS variance of progress variable plays a crucial role in the SGS reaction rate modelling and cannot be obtained using a simple algebraic closure like that commonly used for a passive scalar. The use of strained flamelets to include the flame stretching effects is not required when the variance is obtained from its transport equation and the resolved turbulence contains predominant part of the turbulent kinetic energy. Thus, it seems that SGS closure using unstrained flamelets model is robust and adequate for wide range of turbulent premixed combustion conditions.