Dissertations / Theses on the topic 'Direct simulation'

A direct numerical simulation of a Batchelor vortex has been carried out in the presence of freely decaying turbulence, using both periodic and symmetric boundary conditions; the latter most closely approximates typical experimental conditions, while the former is often used in computational simulations for numerical convenience. A recently developed numerical method, based on compact schemes combined with three stage Runge-Kutta method for time integration, with projection method for enforcing continuity is used for numerical simulations. The Poisson solver used is a direct solver in spectral space. The higher-order velocity statistics were shown to be strongly dependent upon the boundary conditions, but the dependence could be mostly eliminated by correcting for the random, Gaussian modulation of the vortex trajectory, commonly referred to as wandering, using a technique often employed in the analysis of experimental data. Once this wandering had been corrected for, the strong peaks in the Reynolds stresses normally observed at the vortex centre were replaced by smaller local extrema located within the core region but away from the centre. Analysis of the budgets of turbulent kinetic energy and normal Reynolds stress suggest that the production budget during the growth phase of vortex development, resembles turbulent boundary layer type budgets. The analysis of the budgets of turbulent shear stresses shows that the formation and organization of `hairpin' (secondary) structures within the core is the main mechanism for turbulent production and the budget of TKE and radial tangential shear stress shows a turbulent boundary layer type budget.

The work described in the present thesis is related to a series of projects that I worked on toward the better understanding of fragmentation phenomena. In the past decades, the science of fragmentation has attracted many attentions within the researchers due to its wide range of applications. However, because of the complexity of the subject, even its basic concepts need more investigations. This thesis starts with an introduction to fragmentation of droplets using experimental or numerical approaches. It is discussed that the current mathematical and experimental tools are not able to describe all the details. Thus, high performance numerical simulations are the best alternatives to study the breakup of droplets. The introduction is followed by a discussion on the numerical method and the ranges of the non-dimensional groups. It is described that an adaptive, volume of fluid (VOF) method based on octree meshing is used, providing a notable reduction of computational cost. The rest of the thesis basically discusses the obtained results using direct numerical simulations. Two main geometries are investigated: falling droplets and droplets in a stream. For the case of falling droplets, three simulations with different Eötvös numbers are performed. For the case of droplets in a stream, two-dimensional and three-dimensional simulations are performed for a range of Weber number. The results are compared with the available mathematical theories and it is shown that the analysis presented here precisely demonstrates the mechanism of the bag breakup of falling droplets and instability growth over the droplets in an external high-speed flow. The outcomes can significantly assist the development of the secondary atomization and turbulent two-phase flows modelling.

International Telemetering Conference Proceedings / October 29-November 02, 1990 / Riviera Hotel and Convention Center, Las Vegas, Nevada
Direct digital synthesis technology has been employed in the development of a telemetry data simulator constructed for the Western Space and Missile Center (WSMC). The telemetry simulator, known as TDVS II, is briefly described to provide background; however, the principal subject is related to the development of programmable synthesizer modules employed in the TDVS II system. The programmable synthesizer modules (or PSMs) utilize direct digital synthesizer (DDS) technology to generate a variety of common telemetry signals for simulation output. The internal behavior of DDS devices has been thoroughly examined in the literature for nearly 20 years. The author is aware of significant work in this area by every major aerospace contractor, as well as a broad range of activity by semiconductor developers, and in the universities. The purpose here is to expand awareness of the subject and its basic concepts in support of applications for the telemetry industry. During the TDVS II application development period, new DDS devices have appeared and several advances in device technology (in terms of both speed and technique) have been effected. Many fundamental communications technologies will move into greater capacity and offer new capabilities over the next few years as a direct result of DDS technology. Among these are: cellular telephony, high-definition television and video delivery systems in general, data communications down to the general business facsimile and home modem level, and other communications systems of various types to include telemetry systems. A recent literature search of the topic, limited only to documents available in English, indicates that some 25 articles and dissertations of significance have appeared since 1985, with over 30% of these appearing in international forums (including Germany, Japan, Great Britain, Portugal, Finland...). Product advertisements can readily be found in various publications on test instruments, amateur radio, etc., which indicate that international knowledge and product application of the technology is becoming increasingly widespread.

Direct numerical simulation of Hz - O2 in the context of a temporally evolving mixing layer has been performed. Real molecular properties as well as the effects of the species differential diffusion were incorporated into an existing 3D parallel FORTRAN code. The geometry is a box with streamwise and spanwise directions being periodic whereas non-periodic boundaries were set up in transverse (vertical) directions which leads to inhomogeneity for the turbulent field in these directions. Initialisation were performed by error function distributions for streamwise velocity component, scalar mass fraction and temperature along the vertical axis of the domain, Initial pressure is set to be uniform and density Willi calculated based on ideal-gas law for the mixture. Disturbances were introduced by generating spanwise and streamwise vorticity in the middle of the mixing layer to enable transition from laminar to turbulent.

The purpose of this thesis was to develop a program that would accept, as input, a finite set of algebraic equations and simple if-then conditional expressions that model a natural system, and then produce a continuous computer simulation with graphics and tabular output. The equations and conditionals can be in any order and key elements can be missing. The program can be used to run existing models or as a development tool to produce immediate prototypic computer simulations through synergistic man-machine interactions. The theoretical aspects of automatic program generation were discussed, as well as the architectural design of the system. The simulation system was used to develop a computer simulation of an exploited Northern Utah pheasant population and the results were compared to the results from an earlier FORTRAN computer simulation of the same model. It was concluded that the simulation system developed for this thesis produces verified computer simulations from mathematical models that are at least as accurate as the corresponding simulation written in FORTRAN. The system was easy to use and should be useful for unsophisticated users. Some "tuning'' of the input was needed to produce a verified simulation and it was concluded that further work was needed here.

En informatique graphique, les algorithmes de générations d'ombres évaluent la quantité de lumière directement perçue par une environnement virtuel. Calculer précisément des ombres est cependant coûteux en temps de calcul. Dans cette dissertation, nous présentons un nouveau système basé objet robuste, qui permet de calculer des ombres réalistes sur des scènes dynamiques et ce en temps interactif. Nos contributions incluent notamment le développement de nouveaux algorithmes de génération d'ombres douces ainsi que leur mise en oeuvre efficace sur processeur graphique. Nous commençons par formaliser la problématique du calcul d'ombres directes. Tout d'abord, nous définissons ce que sont les ombres directes dans le contexte général du transport de la lumière. Nous étudions ensuite les techniques interactives qui génèrent des ombres directes. Suite à cette étude nous montrons que mêmes les algorithmes dit physiquement réalistes se reposent sur des approximations. Nous mettons également en avant, que malgré leur contraintes géométriques, les algorithmes d'ombres basées objet sont un bon point de départ pour résoudre notre problématique de génération efficace et robuste d'ombres directes. Basé sur cette observation, nous étudions alors le système basé objet existant et mettons en avant ses problèmes de robustesse. Nous proposons une nouvelle technique qui améliore la qualité des ombres générées par ce système en lui ajoutant une étape de mélange de pénombres. Malgré des propriétés et des résultats convaincants, les limitations théoriques et de mise en oeuvre limite la qualité générale et les performances de cet algorithme. Nous présentons ensuite un nouvel algorithme d'ombres basées objet. Cet algorithme combine l'efficacité de l'approche basée objet temps réel avec la précision de sa généralisation au rendu hors ligne. Notre algorithme repose sur l'évaluation locale du nombre d'objets entre deux points : la complexité de profondeur. Nous décrivons comment nous utilisons cet algorithme pour échantillonner la complexité de profondeur entre les surfaces visibles d'une scène et une source lumineuse. Nous générons ensuite des ombres à partir de cette information soit en modulant l'éclairage direct soit en intégrant numériquement l'équation d'illumination directe. Nous proposons ensuite une extension de notre algorithme afin qu'il puisse prendre en compte les ombres projetées par des objets semi-opaque. Finalement, nous présentons une mise en oeuvre efficace de notre système qui démontre que des ombres basées objet peuvent être générées de façon efficace et ce même sur une scène dynamique. En rendu temps réel, il est commun de représenter des objets très détaillés encombinant peu de triangles avec des textures qui représentent l'opacité binaire de l'objet. Les techniques de génération d'ombres basées objet ne traitent pas de tels triangles dit "perforés". De par leur nature, elles manipulent uniquement les géométries explicitement représentées par des primitives géométriques. Nous présentons une nouvel algorithme basé objet qui lève cette limitation. Nous soulignons que notre méthode peut être efficacement combinée avec les systèmes existants afin de proposer un système unifié basé objet qui génère des ombres à la fois pour des maillages classiques et des géométries perforées. La mise en oeuvre proposée montre finalement qu'une telle combinaison fournit une solution élégante, efficace et robuste à la problématique générale de l'éclairage direct et ce aussi bien pour des applications temps réel que des applications sensibles à la la précision du résultat
Direct shadow algorithms generate shadows by simulating the direct lighting interaction in a virtual environment. The main challenge with the accurate direct shadow problematic is its computational cost. In this dissertation, we develop a new robust object-based shadow framework that provides realistic shadows at interactive frame rate on dynamic scenes. Our contributions include new robust object-based soft shadow algorithms and efficient interactive implementations. We start, by formalizing the direct shadow problematic. Following the light transport problematic, we first formalize what are robust direct shadows. We then study existing interactive direct shadow techniques and outline that the real time direct shadow simulation remains an open problem. We show that even the so called physically plausible soft shadow algorithms still rely on approximations. Nevertheless we exhibit that, despite their geometric constraints, object-based approaches seems well suited when targeting accurate solutions. Starting from the previous analyze, we investigate the existing object-based shadow framework and discuss about its robustness issues. We propose a new technique that drastically improve the resulting shadow quality by improving this framework with a penumbra blending stage. We present a practical implementation of this approach. From the obtained results, we outline that, despite desirable properties, the inherent theoretical and implementation limitations reduce the overall quality and performances of the proposed algorithm. We then present a new object-based soft shadow algorithm. It merges the efficiency of the real time object-based shadows with the accuracy of its offline generalization. The proposed algorithm lies onto a new local evaluation of the number of occluders between points (\ie{} the depth complexity). We describe how we use this algorithm to sample the depth complexity between any visible receiver and the light source. From this information, we compute shadows by either modulate the direct lighting or numerically solve the direct illumination with an accuracy depending on the light sampling strategy. We then propose an extension of our algorithm in order to handle shadows cast by semi opaque occluders. We finally present an efficient implementation of this framework that demonstrates that object-based shadows can be efficiently used on complex dynamic environments. In real time rendering, it is common to represent highly detailed objects with few triangles and transmittance textures that encode their binary opacity. Object-based techniques do not handle such perforated triangles. Due to their nature, they can only evaluate the shadows cast by models whose their shape is explicitly defined by geometric primitives. We describe a new robust object-based algorithm that addresses this main limitation. We outline that this method can be efficiently combine with object-based frameworks in order to evaluate approximative shadows or simulate the direct illumination for both common meshes and perforated triangles. The proposed implementation shows that such combination provides a very strong and efficient direct lighting framework, well suited to many domains ranging from quality sensitive to performance critical applications

A Lagrangian framework for computing subgrid-scale combustion physics in Large Eddy Simulations (LES) of fire is formulated and validated. The framework is based on coupling LES formulation, based on the Fire Dynamic Simulator (FDS) with the One-Dimensional Turbulence (ODT) model. The ODT model involves reaction-diffusion and turbulent transport along one-dimensional domains. The one-dimensional domains are attached to the flame brush positions, computed in LES, and are allowed to propagate along its surface. The Lagrangian LES-ODT framework involves various implementations including a) momentum, energy, and species solution along one-dimensional ODT domain, b) Tracking of ODT domains through their anchor points, c) Filtering of ODT solutions on the LES grid, d) Inverse filtering (interpolation) of LES velocity fields in ODT domains, and e) The management of ODT domains at the flow inlets and as they reach the flame tip. Comparison of LES-ODT solutions with FDS solutions shows that the LES-ODT implementation reproduces reasonably well the flame topology and structure.

Thesis (S.M.)--Massachusetts Institute of Technology, Computation for Design and Optimization Program, February 2012.
"February 2012." Cataloged from PDF version of thesis.
Includes bibliographical references (p. [73]-74).
This thesis develops and evaluates Gap-tooth DSMC (GT-DSMC), a direct Monte Carlo simulation procedure for dilute gases combined with the Gap-tooth method of Gear, Li, and Kevrekidis. The latter was proposed as a means of reducing the computational cost of microscopic (e.g. molecular) simulation methods using simulation particles only in small regions of space (teeth) surrounded by (ideally) large gaps. This scheme requires an algorithm for transporting particles between teeth. Such an algorithm can be readily developed and implemented within direct Monte Carlo simulations of dilute gases due to the non-interacting nature of the particle-simulators. The present work develops and evaluates particle treatment at the boundaries associated with diffuse-wall boundary conditions and investigates the drawbacks associated with GT-DSMC implementations which detract from the theoretically large computational benefit associated with this algorithm (the cost reduction is linear in the gap-to-tooth ratio). Particular attention is paid to the additional numerical error introduced by the gap-tooth algorithm as well as the additional statistical uncertainty introduced by the smaller number of particles. We find the numerical error introduced by transporting particles to adjacent teeth to be considerable. Moreover, we find that due to the reduced number of particles in the simulation domain, correlations persist longer, and thus statistical uncertainties are larger than DSMC for the same number of particles per cell. This considerably reduces the computational benefit associated with the GT-DSMC algorithm. We conclude that the GT-DSMC method requires more development, particularly in the area of error and uncertainty reduction, before it can be used as an effective simulation method.
by Jessica D. Armour.
S.M.

The aim of this work is to extend an existing CFD solver, named Shock/Boundary-Layer Interaction (SBLI) code, to include a fully 3D curvilinear capability in order to perform direct numerical simulation (DNS) of turbulent flows over complex geometries. The SBLI code solves the compressible Navier-Stokes equations by the finite difference method and uses the body-fitted curvilinear coordinate system approach to treat complex geometries. The extended version of the code has been used to perform a DNS of a channel flow with longitudinally ridged walls and a DNS of a turbulent flow over an axisymmetric hill geometry. Validation and comparison with previous experimental data and numerical results are also presented. In the first part of the work, the Navier-Stokes equations are presented in a strong conservation form and test validations of the code extension have been carried out such as free stream flow preservation on a wavy grid and a laminar plane channel flow on a skewed mesh. The free stream preservation test consists of a uniform flow computation on a cosinusoidal mesh and the objective is to evaluate the velocity components changes from their initial values due to the effect of a highly skewed mesh. The maximum discrepancy found is around 10-16. For the laminar plane channel flow simulation on a skewed mesh, the purpose is to verify the symmetrical propriety of numerical errors obtained in the velocity components while the main flow direction and the position of the walls are altered in rotation around the three physical coordinates. The symmetry of the numerical error is found to be well preserved as expected. The second part of the work contains DNS of laminar and turbulent flows in a channel with longitudinally ridged walls at different Reynolds numbers. The goal is to investigate the effect of ridged walls on the turbulent flow behavior and to provide quality DNS data-for assessing other numerical simulations, such as Large Eddy Simu-lation (LES) and Reynolds-Averaged Navier Stokes (RANS) modeling. Two Reynolds numbers have been simulated (ReT = 150 and ReT = 360, based on a reference velocity UT = vol Pb( -dPldx), the bulk density and the wall viscosity) on a domain of 1.257r0 x 20 X 0.3757r0 in the streamwise, wall normal and spanwise directions, respectively. This domain is similar to the minimal flow unit for a turbulent plane channel flow. Comparisons with previous experimental data and numerical prediction have show good agreement for the ReT = 150 case and a similar flow dynamics for the ReT = 360 case. In general, the effects of ridged walls on the turbulent flow, like the reduction of the normal Reynolds stress peak values, seems to be smaller when the Reynolds number increases. The third part of this work describes the main simulation of this thesis. DNS of a turbulent flow around an axisymmetric hill is carried out in order to investigate the three-dimensional boundary-layer flow separation which occurs behind the hill. Different domain sizes and grid resolutions have been tested up to a maximum of about 54 million points. A methodology for generating inflow conditions has been implemented and tested. Results are compared with previous experimental and numerical studies. Due to a low Reynolds number used (Reo* = 500, only 5% of an experimental simulation), the time averaged separation bubbles is much bigger and the flow seems to have a laminarisation process due to a strong adverse pressure gradient presented. A small recirculation bubble detected on the top of the hill seems to be the cause of the earlier separation of the turbulent boundary layer and, then, the bigger separation observed. However, similar to the full Reynolds number experiment, same flow dynamics, consisting in the formation of a counter rotating vortex pair merging in the streamwise current, have been captured well. The final part of the work presents an extension of the single-block SBLI code to a multiblock version. A pre-processor program has been developed in order to simplify the treatment of the interface between different blocks and a description of the algorithm is also given. As a demonstration study, DNS of a square jet in a turbulent cross flow has been performed at two Reynolds numbers (Reo* = 1000 and Res- = 2000) and different jet to cross flow velocity ratios. Compared with the available data, the results are in good agree, despite the lower Reynolds number used (half of value simulated in the available data). In conclusion, a fully 3D version of the SBLI code has been successfully derived and tested for various flow configurations. The 3D curvilinear capability has also been implemented and tested by simple, but not trivial, test cases. An option for simplified treatment of Cartesian mesh has been implemented and tests have shown a factor of 2 speedup in overall performance. Two main simulations have been carried out and for the turbulent flow in a ridged channel, the results are in good agreement with published data, while, for the flow over an axisymmetric hill case, simulation is compared qualitatively well and the noticeable discrepancies are primarily due to a reduced Reynolds number conditions. The code has also been successfully extended to a multiblock version and demonstrated on a two-block domain for a jet in cross flow case. Future works includes simulations of the hill problem at higher Reynolds number and LES extension of the SBLI code to fully 3D curvilinear capability.

DESIRE P is a general purpose continuous time simulation language suitable for interactive simulation, dynamic system study, mathematical modeling, process control analysis. It includes an interactive editor, file manipulation facilities, and graphic packages, making it a completely self-contained system. The PDP-11 version of DESIRE P handles 20 state variables, while the VAX/VMS version runs 150 or more. An interpreted job-control language serves for interactive program entry, editing and file operations, and for programming multirun simulation studies. The dynamic segment, containing differential equations in first-order form, is entered just like the job-control statments and accesses the same variables. DESIRE P is largely written in PASCAL, and most of it can be transferred to different computers, with little change. The PASCAL implementation proves that the high-level language can be used to program direct executing languages, still keeping efficiency and speed comparable to assembly language. The runtime compiler of DESIRE P generates fast and efficient code. DESIRE P can incorporate existing and new precompiled FORTRAN numerical integration algorithms.

Existing numerical techniques for modeling saturated deformable porous media are based on homogenization techniques and thus are incapable of performing micro-mechanical investigations, such as the effect of micro-structure on the deformational characteristics of the media. In this research work, a numerical scheme is developed based on the parallelized hybrid lattice-Boltzmann finite-element method, that is capable of performing micro-mechanical investigations through direct numerical simulations. The method has been used to simulate compression of model saturated porous media made of spheres and cylinders in regular arrangements. Through these simulations it is found that in the limit of small Reynolds number, Capillary number and strain, the deformational behaviour of a real porous media can be recovered through model porous media when the parameters porosity, permeability and bulk compressive modulus are matched between the two media. This finding motivated research in using model porous geometries to represent more complex real porous geometries in order to perform investigations of deformation on the latter. An attempt has been made to apply this technique to the complex geometries of ªfeltº, (a fibrous mat used in paper industries). These investigations lead to new understanding on the effect of fiber diameter on the bulk properties of a fibrous media and subsequently on the deformational behaviour of the media. Further the method has been used to investigate the constitutive relationships in deformable porous media. Particularly the relationship between permeability and porosity during the deformation of the media is investigated. Results show the need of geometry specific investigations.

Numerical simulations of liquid flow in a micro-channel between two horizontal plates are performed. The channel is infinite in streamwise and spanwise directions and its height is taken as m, which falls within the dimension ranges of microchannels. The Navier-Stokes equations with the addition of Brinkman number (Br) to the energy equation are used as the governing equations and spectral methods based approach is applied to obtain the required accuracy to handle liquid flow in the microchannel. It is known for microchannels that Br combines the effects of conduction and viscous dissipation in liquids and is also a way of comparing the importance of latter relative to former. The present study aims to simulate the unusual behavior of decreasing of Nu with increasing Re in the laminar regime of microchannels and to show that Br can be introduced to explain this unexpected behavior. Consequently, it is seen at the end of the results that secondary effect of the Br is observed for the single-phase convective heat transfer. Therefore, a laminar flow of a liquid in a microchannel shows different characteristics compared to a similar flow in a macrochannel. To observe the differences, three different cases are run over each of a range of Reynolds numbers: one with no axial conduction assumption that corresponds to a case similar to macrochannel flow, another case with axial conduction included in the energy equation to simulate one of the main differences and lastly a case with the inclusion of Br number in the governing equations. A similar study is made for natural convection with the same numerical set-up for the same three cases. Formation of Rayleigh-Benard cells are observed for the critical numbers widely accepted in the literature. The results are compared with each other to see the effects of axial conduction and Br inclusion, in addition to Ra for natural convection.

Electron densities form field-aligned structured regions in the natural ionosphere and after a high altitude nuclear explosion (HANE). These electron densities, known as plumes, are made up of many smaller individual field-aligned regions called striations. Striation modeling for systems effects has traditionally been done use a statistical approach. This statistical approach evolves different moments of the electron density. Due to lack of test data it has never been validated. The purpose of this project was to use a direct numerical simulation to solve equations governing the differential motion of individual striations. It was done in five steps: 1) Transport a single striation, 2) solve potential equation, 3) combine transport and potential equations, 4) optimize combined solver, and 4) simulate a fully-striated plume for comparison with the statistical model.

It has been over 50 years since Toms [B.A. Toms, Proc. 1^st Int. Congr. Rheol., Sec. II, 135, North Holland (1949)] discovered that adding small quantities of long-chain polymer to turbulent pipe flow could drastically reduce the amount of turbulent drag. Since then a substantial amount of research has gone into examining dilute polymer solutions and their turbulent drag reducing properties, yet the precise mechanism by which the polymers interact with the turbulence to produce this affect is still unclear. From a theoretical standpoint, the difficulty lies in the analysis of accurate models of the fluid dynamics. The combination of the complex mathematical description required for turbulence and the intricate constitutive equations of polymer motion poses significant problems. However, recent developments in computing have resulted in machines powerful enough to simulate such flows using direct numerical simulation (DNS); a technique whereby all the important scales of turbulent fluid motion are fully resolved using an algorithm derived from the full momentum conservation equations. DNS has been successfully employed in previous studies of dilute polymer solutions in computational domains similar to experimental apparatus such as pipe flow [J. M. J. den Toonder, M. A. Hulsen, G. D. C. Kuiken and F. T. M. Nieuwstadt, J. Fluid Mech., 337, 193 (1997)] and channel flow, [R. Sureshkumar, A. N. Beris and R. A. Handler, Phys. Fluids, 9(3), 743 (1996)]. It was our aim to identify the changes within the turbulent dynamics produced by the presence of polymers in homogeneous isotropic turbulence - without an dependence on solid boundary conditions. We decided to simulate our solutions within a infinitely repeating cube subject to periodic boundary conditions, a well established technique for Newtonian fluids. In this thesis we examine a range of spectral measures and integral parameters for various non-Newtonian fluid models in statistically stationary homogeneous isotropic turbulence and compare them to an equivalent Newtonian flow using DNS. We begin by outlining the general theory of homogeneous isotropic non-Newtonian turbulence in the Fourier space domain. We then demonstrate how the general non-Newtonian momentum conservation equations are adapted for use in our DNS. This is followed by a literature review of turbulent drag reduction by long-chain polymer additives. The remainder of the thesis is concerned with the new work. We outline the four non-Newtonian fluid models we applied and present the results obtained from the DNS calculations. Each model embodies a particular characteristic of polymer solutions. The first is of our own construction and is based on the ability of polymers to increase the viscosity of the solution at small scales. Second, we simulate the nonlinear model of McComb [W. D. McComb, Int. J. Engng. Sci., 14, 239 (1976)] where the stress exhibits a nonlinear dependence on the rate of strain. For both of these we were able to obtain an analytical expression for the energy spectra and compared these to the DNS results. The third model is the viscous anisotropic model of den Toonder et al. (see above reference) which introduces a directional element based on the orientation of the polymers in the flow, by assuming they are of constant length and align with the instantaneous velocity. Finally we model a fully coupled FENE-P fluid [L. E. Wedgwood and R. B. Bird, Ind. Eng. Chem. Res., 27, 1313 (1988)] in which the polymers are finitely extensible, elastic and have their own equation of motion giving their orientation. In this way we have identified changes within the structure of turbulence itself which may be related to the drag reduction phenomenon.

In numerous applications, two-phase liquid-gas transport at sub-millimeter length scales plays a substantial role in the determination of the behavior of the system at hand. As its main application, the present work focuses on the polymer electrolyte membrane (PEM) fuel cells. Desirable performance and operational life-time of this class of high-throughput energy conversion devices requires an effective water management, which per se relies on proper prediction of the water-air transport mechanisms. Such two-phase flow involves interfacial forces and phenomena, like hysteresis, that are associated with the physicochemical properties the liquid, gas, and if present, the solid substrate. In this context, numerical modeling is a viable means to obtain valuable predictive understanding of the transport mechanisms, specially for cases that experimental analyses are complicated and/or prohibitively expensive. In this work, an efficient finite element/level-set framework is developed for three-dimensional simulation of two-phase flow. In order to achieve a robust solver for practical applications, the physical complexities are consistently included and the involved numerical issues are properly tackled; the pressure discontinuity at the liquid-gas interface is consistently captured by utilizing an enriched finite element space. The method is stabilized within the framework of variational multiscale stabilization technique. A novel treatment is further proposed for the small-cut instability problem. It is shown that the proposed model can provide accurate results minimizing the spurious currents. A robust technique is also developed in order to filter out the possible noises in the level-set field. It is shown that it is a key to prevent irregularities caused by the persistent remnant of the spurious currents. It is shown how the well-established contact-line models can be incorporated into the variational formulation. The importance of the inclusion of the sub-elemental hydrodynamics is also elaborated. The results presented in the present work rely on the combination of the linearized molecular kinetic and the hydrodynamic theories. Recalling the realistic behavior of liquids in contact with solid substrates, the contact--angle hysteresis phenomenon is taken into account by imposing a consistent pinning/unpinning mechanism developed within the framework of the level-set method. Aside from the main developments, a novel technique is also proposed to significantly improve the accuracy and minimize the the loss in the geometrical features of the interface during the level-set convection based on the back and forth error compensation correction (BFECC) algorithm. Within the context of this thesis, the numerical model is validated for various cases of gas bubble in a liquid and liquid droplets in a gas. For the latter scenario, besides free droplets, the accuracy of the proposed numerical method is assessed for capturing the dynamics droplets spreading on solid substrates. The performance of the model is then analyzed for the capturing the configuration of a water droplet on an inclined substrate in the presence the contact--angle hysteresis. The proposed method is finally employed to simulate the dynamics of a water droplet confined in a gas channel and exposed to air-flow.
Existen numerosas aplicaciones industriales en las que transporte bifásico (líquido-gas) a escalas submilimétricas resulta crucial para la determinación del comportamiento del sistema en cuestión. Entre todas ellas, el presente trabajo se centra en las pilas de combustible con membrana de electrolito polimérico (PEMFC). El rendimiento deseable y la vida útil operativa de esta clase de dispositivos de conversión de energía de alto rendimiento requieren una gestión eficaz del agua (conocida como “water management”), que per se depende de la predicción adecuada de los mecanismos de transporte de agua y aire. Así pues, el análisis del flujo microfluídico de dos fases obliga considerar fuerzas y fenómenos interfaciales, tales como la histéresis, que están asociados con las propiedades fisicoquímicas del líquido, el gas y, si está presente, el sustrato sólido. En este contexto, la modelización numérica es una alternativa viable para obtener una predicción precisa de los mecanismos de transporte, especialmente en aquellos casos en los que los análisis experimentales son prohibitivos, ya sea por su complejidad o coste económico. En este trabajo, se desarrolla un marco eficiente, basado en la combinación del método de elementos finitos y el método de “level-set”, para la simulación tridimensional de flujos bifásicos. Con el fin de lograr una herramienta numérica robusta para aplicaciones prácticas, las complejidades físicas se incluyen consistentemente y los problemas numéricos involucrados se abordan adecuadamente. Concretamente, la discontinuidad de la presión en la interfaz líquido-gas se captura consistentemente utilizando un espacio de elementos finitos enriquecido. La estabilización del método se consigue mediante la introducción de la técnica de multiescalas variacionales. Asimismo, se propone también un tratamiento novedoso para el problema de la inestabilidad de tipo “small-cut”. Se muestra que el modelo propuesto puede proporcionar resultados precisos minimizando las corrientes espurias en la interfaz liquido-gas. Complementariamente, se presenta una nueva metodología para filtrar el ruido en el campo de “level-set”. Esta metodología resulta ser crucial para prevenir las irregularidades provocadas por el remanente persistente de las corrientes espurias. El comportamiento de la línea de contacto es considerado a través de la inclusión los modelos correspondientes en la formulación variacional. A este respecto, el presente trabajo aborda la importancia de la inclusión de la hidrodinámica subelemental. Los resultados presentados se basan en la combinación de la cinética molecular linealizada y las teorías hidrodinámicas. Para representación del comportamiento realista de los líquidos en contacto con sustratos sólidos, el fenómeno de histéresis del ángulo de contacto se tiene en cuenta imponiendo un mecanismo de anclado / desanclado consistente desarrollado en el marco del método de level-set. Aparte de los desarrollos principales, también se propone una técnica novedosa para la convección de la función ”level-set”. Ésta permite mejorar significativamente la precisión, minimizando a su vez la pérdida en las características geométricas de la interfaz asociadas al transporte. Esta nueva metodología está basada en el algoritmo de corrección de compensación de errores (BFECC). La herramienta numérica desarrollada en esta tesis es validada para varios casos que involucran burbujas de gas en un líquido y pequeñas gotas de líquido en un gas. Para el último escenario, además de las gotas libres, se evalúa la precisión de la herramienta propuesta para capturar la dinámica de las gotas sobre sustratos sólidos. A continuación, se analiza el rendimiento del modelo para capturar la configuración de una gota de agua sobre un sustrato inclinado en presencia de la histéresis del ángulo de contacto. El método propuesto finalmente se aplica
Enginyeria civil

[ES] La comprensión de los fenómenos físicos que acontecen en la región densa (también conocida como campo cercano) durante la atomización de los sprays ha sido una de las mayores incógnitas a la hora de estudiar sus aplicaciones. En el sector industrial, el rango de interés abarca desde toberas en aplicaciones propulsivas a sprays en aplicaciones médicas, agrícolas o culinarias. Esta evidente falta de conocimiento obliga a realizar simplificaciones en la modelización, provocando resultados poco precisos y la necesidad de grandes caracterizaciones experimentales en la fase de diseño. De esta manera, los procesos de rotura del spray y atomización primaria se consideran problemas físicos fundamentales, cuya complejidad viene dada como resultado de un flujo multifásico en un régimen altamente turbulento, originando escenarios caóticos. El análisis de este problema es extremadamente complejo debido a la ausencia sustancial de teorías validadas referentes a los fenómenos físicos involucrados como son la turbulencia y la atomización. Además, la combinación de la naturaleza multifásica del flujo y su comportamiento turbulento resultan en una gran dificultad para afrontar el problema. Durante los últimos 10 años, las técnicas experimentales han sido finalmente capaces de visualizar la región densa, pero la confianza, análisis y efectividad de dichos experimentos en esta región del spray todavía requiere de mejoras sustanciales. En este contexto, esta tesis trata de contribuir al entendimiento de estos procesos físicos y de proporcionar herramientas de análisis para estos flujos tan complejos. Para ello, mediante Direct Numerical Simulations se ha afrontado el problema resolviendo las escalas de movimiento más pequeñas, y capturando todas las escalas de turbulencia y eventos de rotura. Uno de los objetivos de la tesis ha sido evaluar la influencia de las condiciones de contorno del flujo entrante en la atomización primaria y en el comportamiento turbulento del spray. Para ello, se han empleado dos condiciones de contorno diferentes. En primer lugar se ha empleado una condición de contorno sintética para producir turbulencia homogenea a la entrada, simulando el comporamiento de la tobera. Una de las características más interesantes de este método es la posibilidad de retocar los parámetros dentro del algoritmo. En particular, la escala de longitud integral se ha variado para evaluar la influencia de las estructuras mas grandes de la tobera en la atomización primaria. El análisis de la condición de contorno sintética también ha permitido el diseño óptimo de simulaciones de las cuales se han derivado estadísticas turbulentas significativas. En este escenario, se han llevado a cabo estudios más profundos sobre la influencia de propiedades de las estructuras turbulentas como la homogeneidad y la anisotropía tanto en el espectro de los flujos como en las estadísticas de las gotas. Para tal fin, se han desarrollado metodologías novedosas para computar el análisis espectral y la estadística de las gotas Entre los resultados de este análisis destaca la independencia de la condición de contorno de entrada en las estadísticas de las gotas, mientras que por otra parte, recalca que las características turbulentas desarrolladas en el interior de la tobera afectan a la cantidad total de masa atomizada. Estas consideraciones se encuentran respaldadas por el análisis espectral realizado, mediante el cuál se concluye que la turbulencia multifásica comparte el comportamiento universal descrito por las teorías de Kolmogorov.
[CAT] La comprensió dels fenòmens físics que succeïxen en la regió densa (també coneguda com a camp pròxim) durant l'atomització dels sprays ha sigut una de les majors incògnites a l'hora d'estudiar les seues aplicacions. En el sector industrial, el rang d'interés comprén des de toveres en aplicacions propulsives a sprays en aplicacions mèdiques, agrícoles o culinàries. Esta evident falta de coneixement obliga a realitzar simplificacions en la modelització, provocant resultats poc precisos i la necessitat de grans caracteritzacions experimentals en la fase de disseny. D'esta manera, els processos de ruptura del spray i atomització primària es consideren problemes físics fonamentals, la complexitat dels quals ve donada com resultat d'un flux multifàsic en un règim altament turbulent, originant escenaris caòtics. L'anàlisi d'este problema és extremadament complex a causa de l'absència substancial de teories validades dels fenòmens físics involucrats com són la turbulència i l'atomització. A més, la combinació de la naturalesa multifàsica del flux i el seu comportament turbulent resulten en una gran dificultat per a afrontar el problema. Durant els últims 10 anys les tècniques experimentals han sigut finalment capaces de visualitzar la regió densa, però la confiança, anàlisi i efectivitat dels experiments en esta regió del spray encara requerix de millores substancials. En este context, esta tesi tracta de contribuir en l'enteniment d'estos processos físics i de proporcionar ferramentes d'anàlisi per a estos fluxos tan complexos. Per a això, per mitjà de Direct Numerical Simulations s'ha afrontat el problema resolent les escales de moviment més menudes, al mateix temps que es capturen totes les escales de turbulència i esdeveniments de ruptura. Un dels objectius de la tesi ha sigut avaluar la influència que les condicions de contorn del flux entrant tenen en l'atomització primària i en el comportament turbulent del spray. Per a això, s'han empleat dos condicions de contorn diferents. En primer lloc s'ha empleat una condició de contorn sintètica per a produir turbulència homogènia a l'entrada, simulant el comportament de la tovera. Una de les característiques més interessants d'este mètod és la possibilitat de retocar els paràmetres dins de l'algoritme. En particular, l'escala de longitud integral s'ha variat per a avaluar la influència de les estructures mes grans de la tovera en l'atomització primària. L'anàlisi de la condició de contorn sintètica també ha permés el disseny òptim de simulacions de les quals s'han derivat estadístiques turbulentes significatives. En este escenari, s'han dut a terme estudis més profunds sobre la influència de propietats de les estructures turbulentes com l'homogeneïtat i l'anisotropia tant en l'espectre dels fluxos com en les estadístiques de les gotes. Per a tal fi, s'han desenrotllat metodologies noves per a computar l'anàlisi espectral i l'estadística de les gotes. Entre els resultats d'esta anàlisi destaca la independència de la condició de contorn d'entrada en les estadístiques de les gotes, mentres que d'altra banda, es recalca que les característiques turbulentes desenrotllades en l'interior de la tovera afecten a la quantitat total de massa atomitzada. Estes consideracions es troben recolzades per l'anàlisi espectral realitzat, per mitjà del qual es conclou que la turbulència multifásica compartix el comportament universal descrit per les teories de Kolmogorov.
[EN] The understanding of the physical phenomena occurring in the dense region (also known as near field) of atomizing sprays has been long seen as one of the biggest unknown when studying sprays applications. The industrial range of interest goes from nozzles in combustion and propulsion applications to medical sprays, agricultural and food process applications. This substantial lack of knowledge is responsible for some important simplification in modeling, that often result to be inaccurate or simply partial, leading to the evident need of large experimental characterization during the design phase. In fact, the spray breakup and primary atomization processes are indeed fundamental problems of physics, which complexity results from the combination of a multiphase flow in a highly turbulent regime that leads to chaotic scenarios. The analysis of this problem is extremely problematic, due to a substantial lack of definitive theories about the physical phenomena involved, namely turbulence and atomization. Furthermore, the combination of the multiphase nature of the flow and its turbulent behavior makes substantially difficult to address the problem. Only within the last 10 years, experimental techniques have been capable of visualizing the dense region, but the experiments reliability, analysis and effectiveness in this region still requires vast improvements. In this scenario, this thesis aims to contribute in the understanding of these physical process and to provide analysis tools for these complex flows. In order to do so, Direct Numerical Simulations have been used for addressing the problem at its smallest scale of motion, while reliably capturing all turbulence scales and breakup events. The multiphase nature of the flow is accounted for by using the Volume of Fluid method. One of the goal of the thesis was to assess the influence of the inflow boundary conditions on the primary atomization and on the spray's turbulence behavior. In order to do so, two different boundary conditions were used. In a first place, a synthetic inflow boundary condition was used in order to produce a homogeneous turbulence inflow, simulating the nozzle behavior. One of the interesting features of this method was the possibility of tweaking the parameters within the algorithm. In particular, the integral length scale was varied in order to assess the influence of nozzle larger turbulent structures on the primary atomization. The analysis on the synthetic boundary condition also allowed to optimally design simulations from which derive meaningful turbulence statistics. On this framework, further studies were carried over on the influence of turbulent structures properties, namely homogeneity and anisotropy, on both the flows spectra and droplets statistics. In order to achieve this goal, novel procedures for both computing the flow spectra and analyzing droplets were developed and are carefully addressed in the thesis. The results of the analysis highlight the independence of droplets statistics from the inflow boundary condition, while, on the other hand, remarking how the total quantity of atomized mass is significantly affected by the turbulence features developed within the nozzle. This considerations are supported by the spectrum analysis performed, which also highlighted how multiphase turbulence shares the universal features described in Kolmogorov theories.
Crialesi Esposito, M. (2019). Analysis of primary atomization in sprays using Direct Numerical Simulation [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/133975
TESIS

Thesis (Ph. D.)--University of Sydney, 2004.
Title from title screen (viewed 16 April 2008). Submitted in fulfilment of the requirements for the degree of Doctor of Philosophy to the School of Aerospace, Mechanical and Mechatronic Engineering. Degree awarded 2004; thesis submitted 2003. Contains published article co-authored by Joung. Includes bibliographical references. Also available in print form.

The present thesis considers the flow mechanics of a natural convection boundary layer (NCBL) along an isothermally heated vertical wall. Large scale direct numerical simulations are carried out to investigate the laminar stability and the turbulent mechanics of the flow. In this study, a computationally efficient temporal framework, where periodic boundary conditions are imposed in the homogeneous directions, has been used to develop a temporally evolving (instead of a spatially evolving) flow. The stability properties of the laminar temporally developing NCBL, with Prandtl number 0.71, are numerically investigated in the configuration of a temporally evolving parallel flow. By assuming the timescales of the laminar base flow and the perturbations are separate, the instantaneous linear stability of the flow is investigated by an eigenvalue approach with a quasi-steady assumption, whereby the unsteady base flow is frozen in time. Temporal responses of the discrete perturbation modes are numerically obtained by solving the two-dimensional linearised disturbance equations using a `frozen' base flow as an initial-value problem at various 〖Gr〗_δ. The resultant amplification rates of the discrete modes are compared with the quasi-steady eigenvalue analysis, and both two-dimensional and three-dimensional direct numerical simulations (DNS) of the temporally evolving flow. The amplification rate predicted by the linear theory compares well with the direct numerical simulation solutions up to a transition point. The extent of the linear regime where the perturbations linearly interact with the base flow is thus identified. The value of the transition 〖Gr〗_δ, according to the three-dimensional DNS results, is dependent on the initial perturbation amplitude. Beyond the transition point, the DNS results diverge from the linear stability predictions as nonlinear mechanisms become important. For the turbulent NCBL flows, three-dimensional direct numerical simulations (DNS) with different initial conditions were carried out to investigate the turbulent mechanics up to 〖Gr〗_δ=1.2×10^8. The turbulent NCBL is examined in two distinct regions separately: a near-wall boundary-layer-like region and an outer bulk plume-like region. In the near-wall region, a constant heat flux layer (see also in George & Capp, 1979; Ho ̈lling & Herwig, 2005) and a constant forcing layer are identified for the turbulent NCBL. In the close vicinity of the wall (y^+<5) a laminar-like sublayer has developed, and the temperature profile follows the linear relation, consistent with the studies of spatially developing flows (Tsuji & Nagano. 1988a); whereas such a linear relation cannot be observed for the velocity profile due to the extra buoyancy. Similar to earlier studies (Ng et al., 2017), this buoyancy effect is shown to asymptotically approach zero with increasing 〖Gr〗_δ. Further away from the wall (y^+>50), there is a log-law region for the mean temperature profile as reported by Tsuji & Nagano (1988a). In this region, the turbulent length scale which characterises mixing scales linearly with distance from the wall once 〖Gr〗_δ is sufficiently large. By taking the varying buoyancy into consideration with the robust mixing length model, a modified log-law for the mean velocity profile for y^+>50 is proposed. The effect of the initialization is shown to persist until relatively high 〖Gr〗_δ as a result of slow adjustment of the buoyancy (temperature) profile. Once these differences are accounted for, our two DNS cases and the spatially developing data of Tsuji & Nagano (1988a) show excellent agreement with the modified log-law. Beyond a wall-normal distance δ_i, the NCBL can be characterised as an outer bulk plume-like region. This region is found to be well described by an self-similar integral model with profile coefficients (cf. van Reeuwijk & Craske, 2015) which are 〖Gr〗_δ-independent after 〖Gr〗_δ=10^7. The entrainment coefficient for the plume-like region is investigated by decomposing contributions from shear production, buoyancy, viscosity and boundary conditions. For the turbulent NCBL, the entrainment coefficient is found to be mainly affected by the buoyancy in the flow and appears constant beyond 〖Gr〗_δ=10^7. Solution to the self-similar integral model are analytically obtained by solving ordinary differential equations (ODE) with profile coefficients empirically obtained from the DNS results. The DNS results also suggest that the wall heat transfer of the NCBL is directly related to the top-hat scales which characterise the plume-like region. The Nusselt number of the NCBL is found to follow 〖Nu〗_δ∝〖Gr〗_δ^0.373, similar to the observations by Ng et al. (2017) for a vertical NCBL in differentially heated slot and He et al. (2012) for a Rayleigh--Be ̀nard convection. This power-law correlation is higher than the empirical 1/3-power-law correlation reported for spatially developing NCBLs at lower 〖Gr〗_δ, but appears consistent with the ultimate heat transfer with a logarithmic correction suggested by Grossmann & Lohse (2011). Using empirical correlations for the wall shear stress, it is shown that the buoyancy effect in the near-wall region would become negligible, and the near-wall mechanics of the NCBL would become similar to that of a neutrally buoyant turbulent boundary layer above 〖Gr〗_δ>2×10^9 for the present study.

A new Direct Simulation fibre model was developed which allowed flexibility in the fibre during the simulation of fibre suspension flow.This new model was called the 'Chain-of-Spheres' model.It was hypothesised that particle shape and deformation could significantly affect particle dynamics,and also suspension bulk properties such as viscosity.Data collected from the simulation showed that flexible fibres in shear flow resulted in an order of 7 −10% bulk relative viscosity increase over the 'rigid' fibre result.Results also established the existence of a relationship between bulk viscosity and particle stiffness. In comparison with experimental results,other more conventional rigid fibre based methods appeared to underpredict relative viscosity.The flexible fibre method thus markedly improved the ability to estimate relative viscosity.The curved rigid fibre suspension also exhibited increased viscosity of the order twice that of the equivalent straight rigid fibre suspension.With such sensitivity to fibre shape,this result has some important implications for the quality of fibre inclusions used.For consistent viscosity,the shape quality of the fibres was shown to be important. The 'Chain of Spheres' simulation was substantially extended to create a new simulation method with the ability to model the dynamics of arbitrarily shaped particles in the Newtonian flow field.This new '3D Particle' simulation method accounted for the inertial force on the particles,and also allowed particles to be embedded in complex flow fields.This method was used to reproduce known dynamics for common particle shapes,and then to predict the unknown dynamics of various other particle shapes in shear flow. In later sections, the simulation demonstrated inertia-induced particle migration in the non-linear shear gradient Couette cylinder flow,and was used to predict the fibre orientation within a diverging channel flow.The performance of the method was verified against known experimental measurements,observations and theoretical and numerical results where available.The comparisons revealed that the current method reproduced single particle dynamics with great fidelity. The broad aim of this research was to better understand the microstructural dynamics within the fibre-filled suspension and from it,derive useful engineering information on the bulk flow of these fluids.This thesis represents a move forward to meet this broad aim.It is hoped that future researchers may benefit from the new approaches and algorithms developed here.

A new Direct Simulation fibre model was developed which allowed flexibility in the fibre during the simulation of fibre suspension flow.This new model was called the �Chain-of-Spheres �model.It was hypothesised that particle shape and deformation could signi ficantly a ffect partic e dynamics,and also suspension bulk properties such as viscosity.Data collected from the simulation showed that flexible fibres in shear flow resulted in an order of 7 −10% bulk relative viscosity increase over the �rigid �fibre result.Results also es- tablished the existence of a relationship between bulk viscosity and particle sti ffness.In comparison with experimental results,other more conventional rigid fibre based methods appeared to underpredict relative viscosity.The flexible fibre method thus markedly improved the ability to estimate relative viscosity.The curved rigid fibre suspension also exhibited increased viscosity of the order twice that of the equivalent straight rigid fibre suspension.With such sensitivity to fibre shape,this result has some important implications for the quality of fibre inclusions used.For consistent viscosity,the shape quality of the fibres was shown to be important. The �Chain of Spheres �simulation was substantially extended to create a new simulation method with the ability to model the dynamics of arbitrarily shaped particles in the Newtonian flow field.This new �3D Particle �simulation method accounted for the inertial force on the particles,and also allowed particles to be embedded in complex flow fields.This method was used to reproduce known dynamics for common particle shapes,and then to predict the unknown dynamics of various other particle shapes in shear flow. In later sections, the simulation demonstrated inertia-induced particle migration inthe non-linear shear gradient Couette cylinder flow,and was used to predict the fibre orientation within a diverging channel flow.The performance of the method was verified against known experimental measurements,observations and theoretical and numerical results where available.The comparisons revealed that the current method reproduced single particle dynamics with great fidelity. The broad aim of this research was to better understand the microstruc- tural dynamics within the fibre-filled suspension and from it,derive useful engineering information on the bulk flow of these fluids.This thesis represents a move forward to meet this broad aim.It is hoped that future researchers may bene fit from the new approaches and algorithms developed here.

La présence de contraintes de plus en plus strictes sur les émissions de polluants on poussé les contruteurs vers l'injection directe essence (IDE), afin d'améliorer les performances et réduire la consommation de carburant et les émissions des moteurs à combustion interne. Par conséquent, de nouveaux défis sont introduits en termes d'optimisation de la combustion, en raison d'une plus complexe phénomenologie tandis que les modéles système demande des paramètres de calibration supplémentaires.Cette thèse présente le développement et la validation d'un modèle zéro-dimensionnel (0D) de combustion en IDE pour application en simulation système. Le modèle proposé détaille la physique de l'atomisation, et évaporation des gouttes, de la préparation du mélange air/carburant, de la propagation de flamme dans un mélange non-homogène ainsi que l'intéraction entre ces phénomènes.La phase liquide est discretisés en paquets groupant des gouttes de la même taille.Un modèle d'atomisation empirique basé sur la vitesse d'injection, les propriétés du carburant et les conditions thermodynamiques fournit les diamètres initiaux. Un modèle Lagrangien détaillant une dynamique de trainée/inértie, échange thermique et convection forcée décrit la pénétration liquide et l'evaporation des paquets. La formation du mélange air/carburant est décrite avec une PDF qui discretise la charge en un mécanisme de classes intéragissant les unes avec les autres et avec les paquets de gouttes. La propagation de flamme prend en compte les effets de l'hétérogéneité du mélange sur la vitesse de flamme et la formation des polluants.Le modèle proposé a été implémenté dans la plateforme Simcenter Amesim, dédiée á la modélisation de systémes multi-physiques, et intégrée dans le modèle de combustion essence CFM1D, de la librairie IFP-Engine.Des approche de modélisation de l'evaporation de carburant, de la dynamique de spray et de la formation du mélange, inspirés de la literature sur les moteurs Diesel, ont été adaptés aux conditions IDE.Le modèle a initialement été validé sur des mesures et des simulations RANS 3D réalisées avec le code IFP-C3D, d'une bombe d'injection à volume constant.Un vortex de tumble, dans un premier temps, et des variations rapides du voulume de la chambre ensuite, ont été ajoutés aux expériments numériques afin d'évaluer la réponse du modèle à l'aérodynamique dans la chambre de combustion et à des conditions thermodynamiques variables, en termes d'évaporation, développement du spray et distribution de la richesse. Des simulations d'injections dans un moteur entraîné,dont les résultats ont été comparés avec des mesures et des calculs CDF,complètent la validation du modèle avec à la fois des conditions thermodynamiques variable et de l'aérodynamique
Future constraints on pollutant emissions pushed car manufacturers towards gasoline direct injection (GDI) technologies to improve engine performances and reduce fuel consumption and emissions. New challenges are then introduced in terms of combustion optimization due to a more complex phenomenology while system models require additional calibration parameters.This PhD work presents the development and validation of a Zero-Dimensional (0D) model of GDI combustion for system simulation. The proposed model focuses on physics of atomization and drop evaporation, fuel/air mixing, flame propagation in heterogeneous charge and mutual interaction between these phenomena.The liquid phase is discretized in parcels grouping drops of the same size. An empirical atomization model based on injection velocity, fuel characteristics and thermodynamic conditions provides initial diameters. A Lagrangian model including drag-inertia dynamics, heat-up and forced convection describes drop parcel penetration and evaporation. Fuel / air mixing is described using a discrete Probability Density Function (PDF) approach, based on constant-mixture-fraction classes interacting with each other and with the drop parcels. Flame propagation takes into account mixture heterogeneity effects on flame speed and pollutant production is modelled.The model was implemented in the Simcenter Amesim platform for multi-physical modelling and integrated in a generic Spark Ignition (SI) combustion chamber submodel, CFM1D, from the IFP-Engine library.Fuel evaporation, spray dynamics and mixture formation modelling approaches, inspired by literature on Diesel engines, were adapted to GDI operating conditions. The model was first validated on a constant-volume vessel with quiescent gas in different thermodynamic conditions by means of experiments and 3D RANS CFD simulations performed with IFP-C3D. A tumble vortex in a constant volume vessel, in a first time, and rapid variations of the vessel volume, in a second time, were then added to the numerical experiment in order to test the model response to in-cylinder flow aerodynamics and variable thermodynamic conditions, respectively, in terms of fuel evaporation, spray development and fuel/air mixing and equivalence ratio distribution. Computations of fuel injections in a motored engine complete the model validation campaign in variable thermodynamic conditions and with realistic aerodynamics and the results were compared to both experiments and CFD computations

Le présent travail se concentre sur les propriétés statistiques des scalaires réactifs subissant des réactions chimiques réversibles en turbulence incompressible. Une analyse théorique des propriétés statistiques des scalaires à différents ordres de moments a été réalisée sur la base d'approximations et de modèles convenablement proposés. Les résultats théoriquement dérivés ont ensuite été comparés aux résultats numériques obtenus par simulation numérique directe (DNS). Dans la simulation numérique directe, les dérivés spatiales ont été principalement approximées en utilisant une méthode pseudi-spectrale, car la vitesse turbulente et les champs scalaires sont généralement des conditions aux limites périodiques. Pour les configurations spéciales dans lesquelles la condition aux limites n'est pas périodique, une méthode aux différences finies avec des schémas fins a été utilisée pour approximer les dérivées spatiales. L'intégration temporelle numérique a été mise en oeuvre par un schéma Runge-Kutta du troisième ordre. Tous les travaux menés dans cette thèse sont consacrés aux explorations numériques et théoriques des scalaires réactifs en turbulence incompressible de différentes configurations. Nos résultats suggèrent de nouvelles idées pour de futures études, qui sont discutées dans les conclusions
The present work focuses on the statistical properties of reactive scalars undergoing reversible chemical reactions in incompressible turbulence. Theoretical analysis about the statistical properties of scalars at different order of moments were carried out based on appropriately proposed approximations and models. The theoretically derived results were then compared with numerical results obtained by direct numerical simulation (DNS). In the direct numerical simulation, the spatial derivatives were mainly approximated by using a pseudo-spectral method, since the turbulent velocity and scalar fields are generally of periodic boundary conditions. For the special configurations in which the boundary condition is not periodic, a finite difference method with fine schemes was used to approximate the spatial derivatives. The numerical time integration was implemented by a third order Runge-Kutta scheme. All the works carried out in this thesis are devoted to the numerical and theoretical explorations about reactive scalars is incompressible turbulence of different configurations. Our finding suggest new ideas for future studies, which are discussed in the conclusions

In this study, which is numerical in nature, direct numerical simulation (DNS) of the pipe flow is performed. For the DNS a solenoidal spectral method is employed, this involves the expansion of the velocity using divergence free functions which also satisfy the prescribed boundary conditions, and a subsequent projection of the N-S equations onto the corresponding dual space. The solenoidal functions are formulated in Legendre polynomial space, which results in more favorable forms for the inner product integrals arising from the Petrov-Galerkin scheme employed. The developed numerical scheme is also used to investigate the effects of spanwise oscillations and phase randomization on turbulence statistics, and drag, in turbulent incompressible pipe flow for low to moderate Reynolds numbers (i.e. $mathrm{Re} sim 5000$) ).

Offshore marine applications often include configurations of cylindrical structures that produce complex three-dimensional flow features. Catenary risers, for instance, can create complex flow patterns when subjected to hydrodynamic loads. In recent published studies, the shape of a catenary riser has been approximated by a quarter segment of a ring followed by a horizontal extension, obtaining a curved circular cylinder. In the present Master thesis, Direct Numerical Simulations at Re = 100 and 500 have been conducted in order to study the flow past such geometry. The main flow direction was parallel to the plane of curvature of the cylinder and directed towards the convex face of the quarter-of-ring. Additionally, a sheared incoming flow has been considered in the analysis by imposing a linearly varying velocity profile at the inlet. The shedding mechanism observed in uniform flow was similar to that reported in previous published studies. One single shedding frequency prevailed along the entire span of the cylinder at Re = 100 and 500. Moreover, the vortex cores at Re = 100 were normal to the flow direction and exhibited slight distortions as they were convected downstream, whereas at Re = 500 the wake topology was characterized by three-dimensional structures of smaller scale. A sheared inflow, on the other hand, gave rise to an oblique and cellular vortex shedding pattern with two cells of different shedding frequencies. The strong slanting of the vortices, as well as the cellular pattern, was clearly induced by the variation of the local Reynolds number along the front stagnation point. The basic knowledge gained from this thesis appear as very promising in the context of marine structures, it is therefore expected that this work will constitute a basis for further investigations considering this type of geometry.

Fossil fuel reserves are projected to be decreasing, and emission regulations are becoming more stringent due to increasing atmospheric pollution. Alternative fuels for power generation in industrial gas turbines are thus required able to meet the above demands. Examples of such fuels are synthetic gas, blast furnace gas and coke oven gas. A common characteristic of these fuels is that they are multi-component fuels, whose composition varies greatly depending on their production process. This implies that their combustion characteristics will also vary significantly. Thus, accurate and yet flexible enough combustion sub-models are required for such fuels, which are used during the design stage, to ensure optimum performance during practical operating conditions. Most combustion sub-model development and validation is based on Direct Numerical Simulation (DNS) studies. DNS however is computationally expensive. This, has so far limited DNS to single-component fuels such as methane and hydrogen. Furthermore, the majority of DNS conducted to date used one-step chemistry in 3D, and skeletal chemistry in 2D only. The need for 3D DNS using skeletal chemistry is thus apparent. In this study, an accurate reduced chemical mechanism suitable for multi-component fuel-air combustion is developed from a skeletal mechanism. Three-dimensional DNS of a freely propagating turbulent premixed flame is then conducted using both mechanisms to shed some light into the flame structure and turbulence-scalar interaction of such multi-component fuel flames. It is found that for the multi-component fuel flame heat is released over a wider temperature range contrary to a methane flame. This, results from the presence of individual species reactions zones which do not all overlap. The performance of the reduced mechanism is also validated using the DNS data. Results suggest it to be a good substitute of the skeletal mechanism, resulting in significant time and memory savings. The flame markers commonly used to visualize heat release rate in laser diagnostics are found to be inadequate for the multi-component fuel flame, and alternative markers are proposed. Finally, some popular mean reaction rate closures are tested for the multi-component fuel flame. Significant differences are observed between the models’ performance at the highest turbulence level considered in this study. These arise from the chemical complexity of the fuel, and further parametric studies using skeletal chemistry DNS would be useful for the refinement of the models.

In this thesis models for the simulation of transient phenomena in high voltage direct-current systems are developed. The new converter model is versatile and the solution algorithm is free from numerical oscillations. A new generic inverter control described in this thesis is based on a predictive approach. Steady-state and transient simulations of two-terminal and multi-terminal (i.e., a parallel converter system) high voltage direct-current systems are carried out using the new converter system model. Comparison between the two-terminal transient simulation results and the high voltage direct-current simulator outputs shows good agreement. An alternating-current/direct-current initialization procedure for the Electromagnetic Transients Program (EMTP) has been investigated and a novel initialization algorithm has been suggested in this thesis.
Applied Science, Faculty of
Electrical and Computer Engineering, Department of
Graduate

La thèse est axée sur l'examen de l'effet de la contrainte sur la conductivité thermique des matériaux piézoélectriques. Les matériaux piézoélectriques sont des cristaux qui présentent une déformation mécanique lors de l'application d'un champ électrique. Des exemples de tels systèmes sont ZnO, AlN, et SiO2. En utilisant des simulations de dynamique moléculaire, nous avons calculé la conductivité thermique de cristaux de ZnO et AlN sous contrainte. Nous avons aussi calculé la résistance thermique des interfaces SiO/C et ZnO/C soumis à un champ électrique.Nous commençons par le calcul des propriétés piézoélectriques et élastiques de ZnO. Celles-ci serviront à valider les potentiels interatomiques utilisés, et à montrer l'ampleur de la contrainte qu’il est possible d'appliquer. En utilisant la dynamique moléculaire d'équilibre, nous avons estimé le coefficient élastique c33 de ZnO, qui se trouve être en accord avec les valeurs expérimentales. Il a aussi été déterminé que la limite élastique d'un cristal de ZnO est de 6 GPa, ce qui correspond à une déformation de 6%. Nous avons ensuite établi les coefficients piézoélectriques de ZnO en utilisant la dynamique moléculaire de non-équilibre, et il a été constaté que les coefficients piézoélectriques dij sont en accord avec les valeurs de la littérature.Deuxièmement, nous avons examiné l'effet de la pression sur la conductivité thermique intrinsèque de ZnO et d’AlN. La dynamique moléculaire de non-équilibre inverse a été mise en œuvre pour calculer la conductivité parce que les coûts de calcul sont nettement inférieurs à ceux de la méthode d'équilibre, d’autant plus pour ZnO dont le potentiel inter-atomique contient les interactions Coulombiennes. L'effet de taille sur la conductivité thermique de ZnO et AlN a ensuite été étudié. Nous avons montré que la formule de Schelling peut en effet être mise en œuvre pour les deux cristaux pour différentes valeurs de la contrainte. La conductivité thermique pour un cristal de ZnO de taille infinie est extraite de la formule de Schelling, et elle se révèle être de 410 W/mK. La conductivité thermique de cristaux de ZnO sous contrainte a ensuite été analysée. Nous avons montré que, après correction de l'effet de taille, la conductivité thermique suit une dépendance en loi de puissance à la contrainte uniaxiale. De plus, la conductivité thermique de ZnO est affectée par un champ statique externe en raison de la contrainte induite. La conductivité thermique d'AlN est estimée à 3000 W/mK, l'effet de la contrainte ne modifie pas cette valeur du fait du potentiel inter-atomique utlisé. Par conséquent, AlN n’est pas un matériau pertinent pour faire office de switch thermique.Troisièmement, nous avons exploré l'effet d’un déplacement piézoélectrique sur la conductance thermique d’interface de Si2O/graphène et ZnO/graphène. Utilisant la dynamique moléculaire d’équilibre, la conductivité thermique d'un super-réseau dont la période est composée de silice et de graphène polyfeuillet. Le super-réseau a été évalué pour différentes valeurs du champ électrique externe. Nous avons constaté que l'application d'un champ électrique de 20 MV/m positif parallèle à la direction hors-plan du super-réseau conduit à la réduction de la conductivité thermique d'un facteur deux. D'autre part, aucun changement dans la conductance thermique n’est noté pour le super-réseau ZnO/graphène. Cette différence est due aux différences de déformations induites au niveau des interfaces dans le super-réseau. L'effet est recréé dans un super-réseau Si/Ge en appliquant une déformation pour former les interfaces. Cette approche crée une déformation non uniforme qui est susceptible de diffuser les phonons
The thesis is focused on investigating the effect of strain on the thermal conductivity of piezoelectric materials. Piezoelectric materials are crystals which display a mechanical deformation upon application of an electric field. Examples of such material are ZnO, AlN, and SiO2. Using Molecular Dynamics simulations, we calculate the thermal conductivity of unstrained and strained ZnO and AlN crystals. We also calculate the thermal resistance of SiO/graphene interfaces under strain.We calculate the piezoelectric and elastic properties of ZnO. These will serve as confirmation of the correctness of the inter-atomic potential used, and will serve to show the magnitude of strain that is possible to apply. Using non-equilibrium molecular dynamics, we determine the elastic coefficient of ZnO c33, and we see that it agrees with experimental values. We also determine that the elastic limit of a perfect ZnO crystal is 6 GPa which corresponds to a 6% strain. We also determine the piezoelectric coefficient of ZnO using NEMD, and we find that the piezoelectric coefficient d33 also agrees with literature values.Second, we look at the effect of strain on the intrinsic thermal conductivity of ZnO and AlN. We use reverse non-equilibrium molecular dynamics to calculate the conductivity because the computational costs are significantly lower than those for the equilibrium method; especially for ZnO whose inter-atomic potential contains Coulomb interaction. We also study the size-effect on the thermal conductivity of ZnO and AlN. We show that the Schelling formula can indeed be implemented to both crystals for different values of strain. The infinite length thermal conductivity for ZnO is extracted from the formula, and it is found to be 410 W/mK. We then calculate the thermal conductivity of strained ZnO crystals. We show that after correcting for the size effect the thermal conductivity follows power-law dependence to uniaxial strain. Also, we demonstrate that the thermal conductivity of ZnO can be affected by a static external field due to the induced strain. The infinite length thermal conductivity of AlN is found to be 3000 W/mK. We show that for the case of AlN the effect of strain does not affect the thermal conductivity due to the different inter-atomic bonding. Hence, AlN might not be a useful material for piezothermal application.Third, we explore the effect of piezoelectric strain on the thermal conductance of SiO2/graphene and ZnO/graphene superlattices. Using EMD we calculate the thermal conductivity of a superlattice composed of silica and graphene monolayers. The thermal conductance of the superlattice was evaluated under different values of external electric field. We find that applying a positive electric field parallel to the Z-direction leads to reduction of the thermal conductance by a factor of 2 for an electric field of 20 MV/m. On the other hand, no change in the thermal conductance is noted for ZnO/graphene superlattice. The effect is due to the non-uniform strain induced at the superlattice junctions. The effect is recreated in Si/Ge superlattice by mechanically applying a non-uniform strain at the interface. This approach might be responsible for the scattering of phonons

Thesis (Ph.D. in Mechanical Engineering)--S.M.U.
Title from PDF title page (viewed July 20, 2007). Source: Dissertation Abstracts International, Volume: 67-02, Section: B, page: 1125. Adviser: Radovan Kovasevic. Includes bibliographical references.