No description available.

Flame-based synthesis is an attractive industrial process for the large scale generation of nanoparticles. In this aerosol process, a gasifi ed precursor is injected into a high-temperature turbulent flame, where oxidation followed by particle nucleation and other solid phase dynamics create nanoparticles. Precursor oxidation, which ultimately leads to nucleation, is strongly influenced by the turbulent flame dynamics. Here, direct numerical simulation (DNS) of a canonical homogeneous flow is used to understand the interaction between a methane/air flame and titanium tetrachloride oxidation to titania. Detailed chemical kinetics is used to describe the combustion and precursor oxidation processes. Results show that the initial precursor decomposition is heavily influenced by the gas phase temperature field. However, temperature insensitivity of subsequent reactions in the precursor oxidation pathway slow down conversion to the titania. Consequently, titania formation occurs at much longer time scales compared to that of hydrocarbon oxidation. Further, only a fraction of the precursor is converted to titania, and a signi cant amount of partially-oxidized precursor species are formed. Introducing the precursor in the oxidizer stream as opposed to the fuel stream has only a minimal impact on the oxidation dynamics. In order to understand modeling issues, the DNS results are compared with the laminar flamelet model. It is shown that the flamelet assumption qualitatively reproduces the oxidation structure. Further, reduced oxygen concentration in the near-flame location critically a ffects titania formation. The DNS results also show that titania forms on the lean and rich sides of the flame. A reaction path analysis (RPA) is conducted. The results illustrate the di ffering reaction pathways of the detailed chemical mechanism depending on the composition of the mixture. The RPA results corroborate with the DNS results that titania formation is maximized at two mixture fraction values, one on the lean side of the flame, and one on the rich side. / text

Spacer-filled channels are employed in membrane modules in many industrial applications where feed-flow spacers (employed to separate membrane sheets and create flow channels) tend to enhance mass transport characteristics, possibly mitigating fouling and concentration polarization phenomena. In this work direct numerical simulation was performed for the flow in the spacer-filled channels to obtain a better understanding of fluid flow and mass transfer phenomena in these channels. A solute with a Schmidt number of 1 at Reynolds numbers of 300, 500 and 800 (based on the bulk velocity and spacer diameter) was considered. The effect of spacer location was also studied for three different configurations, spacer at the centre of the channel, at off-centre location, and attached to the wall. Instantaneous velocity fields and flow structures such as separation of boundary layer on the walls and on the cylinder, eddies on the walls, recirculation regions and vortex shedding were investigated. A Fourier analysis was carried out on the time series velocity data. Using this analysis the Strouhal number was calculated and the development of the flow towards a broader turbulent state at higher Reynolds number was captured. Other statistical characteristics such as time-averaged velocities and wall shear rates are obtained and discussed. The average pressure loss which represents the operation cost of membrane modules was calculated for the channels and found to be highest for spacer at the centre of the channel and lowest for spacer attached to the wall. Scalar transport equation is directly solved along with Navier-Stokes equation to get the concentration field. Local Sherwood number is obtained on the walls and the relationship between shear stress, vortex shedding, and mass transfer enhancement was explored. The overall Sherwood number and Stanton number of the channels, which indicate the mass transfer performance of the channels, are obtained. It was observed that as spacer approaches the wall mass transfer rate is decreasing. / Thesis (Master, Mechanical and Materials Engineering) -- Queen's University, 2010-11-30 11:44:07.479

abstract: Many methods of passive flow control rely on changes to surface morphology. Roughening surfaces to induce boundary layer transition to turbulence and in turn delay separation is a powerful approach to lowering drag on bluff bodies. While the influence in broad terms of how roughness and other means of passive flow control to delay separation on bluff bodies is known, basic mechanisms are not well understood. Of particular interest for the current work is understanding the role of surface dimpling on boundary layers. A computational approach is employed and the study has two main goals. The first is to understand and advance the numerical methodology utilized for the computations. The second is to shed some light on the details of how surface dimples distort boundary layers and cause transition to turbulence. Simulations are performed of the flow over a simplified configuration: the flow of a boundary layer over a dimpled flat plate. The flow is modeled using an immersed boundary as a representation of the dimpled surface along with direct numerical simulation of the Navier-Stokes equations. The dimple geometry used is fixed and is that of a spherical depression in the flat plate with a depth-to-diameter ratio of 0.1. The dimples are arranged in staggered rows separated by spacing of the center of the bottom of the dimples by one diameter in both the spanwise and streamwise dimensions. The simulations are conducted for both two and three staggered rows of dimples. Flow variables are normalized at the inlet by the dimple depth and the Reynolds number is specified as 4000 (based on freestream velocity and inlet boundary layer thickness). First and second order statistics show the turbulent boundary layers correlate well to channel flow and flow of a zero pressure gradient flat plate boundary layers in the viscous sublayer and the buffer layer, but deviates further away from the wall. The forcing of transition to turbulence by the dimples is unlike the transition caused by a naturally transitioning flow, a small perturbation such as trip tape in experimental flows, or noise in the inlet condition for computational flows. / Dissertation/Thesis / M.S. Mechanical Engineering 2011

abstract: The flow around a golf ball is studied using direct numerical simulation (DNS). An immersed boundary approach is adopted in which the incompressible Navier-Stokes equations are solved using a fractional step method on a structured, staggered grid in cylindrical coordinates. The boundary conditions on the surface are imposed using momentum forcing in the vicinity of the boundary. The flow solver is parallelized using a domain decomposition strategy and message passing interface (MPI), and exhibits linear scaling on as many as 500 processors. A laminar flow case is presented to verify the formal accuracy of the method. The immersed boundary approach is validated by comparison with computations of the flow over a smooth sphere. Simulations are performed at Reynolds numbers of 2.5 × 104 and 1.1 × 105 based on the diameter of the ball and the freestream speed and using grids comprised of more than 1.14 × 109 points. Flow visualizations reveal the location of separation, as well as the delay of complete detachment. Predictions of the aerodynamic forces at both Reynolds numbers are in reasonable agreement with measurements. Energy spectra of the velocity quantify the dominant frequencies of the flow near separation and in the wake. Time-averaged statistics reveal characteristic physical patterns in the flow as well as local trends within dimples. A mechanism of drag reduction due to the dimples is confirmed, and metrics for dimple optimization are proposed. / Dissertation/Thesis / Ph.D. Mechanical Engineering 2011

Nous étudions par simulation numérique le transport et le dépôt turbulent d’inclusions d’oxydes métalliques (de l’ordre de 10 µm de diamètre) en suspension dans l’acier liquide. Deux surfaces de dépôt sont envisagées : l’interface acier liquide/paroi solide et l’interface métal liquide/laitier. Dans les deux cas, nous nous focalisons sur la couche limite adjacente à l’interface. Le comportement des inclusions en suspension est examiné à l’aide d’un suivi lagrangien où le poids, la poussée d’Archimède, la force d’accélération en volume, la force de masse ajoutée et la force de traînée sont prises en compte dans l’équation de la dynamique. Dans le cas de la paroi solide, nous nous appuyons sur une représentation schématique de l’écoulement du métal liquide dans la sous-couche visqueuse et dans la zone tampon, où les structures turbulentes qui apportent le liquide à la paroi (sweeps) ou l’éjectent (bursts) sont décrites analytiquement (modèle d’Ahmadi). Les simulations numériques montrent que les mécanismes principaux de dépôt des inclusions sont la sédimentation et dans une moindre mesure l’interception directe. Notons cependant que la contribution de l’interception directe croît avec l’intensité turbulente de l’écoulement et peut devenir prépondérante pour les vitesses de frottement les plus élevées (au-delà de 0,1 m.s-1). Les effets inertiels ont, quant à eux, une contribution négligeable sur le dépôt des inclusions (contrairement au cas des aérosols). Enfin, la prise en compte des interactions hydrodynamiques entre les inclusions et la paroi solide conduit à une diminution significative de la vitesse de dépôt des inclusions. Dans le cas de l’interface acier liquide/laitier, l’écoulement du métal liquide est calculé par simulation numérique directe (DNS) à l’échelle de la couche de surface. La turbulence, générée à distance de l’interface par un forçage aléatoire, diffuse vers l’interface métal liquide/laitier modélisée comme une surface libre indéformable. L’évolution des inclusions en suspension est obtenue par un suivi lagrangien à l’aide d’un couplage one-way. Le nombre de Reynolds de surface des simulations varie de 68 à 235. Le diamètre des inclusions varie de 10-5m à 5.10-5m et le rapport entre la densité des inclusions et la densité du métal varie de 0,5 (inclusions d’alumine) à 1 (inclusions fictives). Il apparaît que le dépôt des inclusions d’alumine est contrôlé par la sédimentation. En l’absence d’effet gravitaire, le dépôt d’inclusions est contrôlé par l’interception directe et dépend fortement du nombre de Reynolds de surface. Dans ce dernier cas, nous montrons que la vitesse de dépôt adimensionnée par la vitesse de Kolmogorov de surface est proportionnelle au diamètre des inclusions adimensionné par la longueur de Kolmogorov de surface. La prise en compte des interactions hydrodynamiques entre les inclusions et la surface libre conduit à une diminution de moitié de la contribution de l’interception directe mais affecte peu la contribution gravitationnelle. En outre, en l’absence d’effet gravitaire, la linéarité entre la vitesse de dépôt adimensionnée et le diamètre des inclusions adimensionné est conservée. / The deposition of metallic oxide inclusions (of about 10 µm in diameter) suspended in liquid steel is studied by numerical simulation. Two types of deposition surface are investigated, i.e., the liquid steel/solid wall interface and the liquid steel/liquid slag interface. In both cases, we focus on the boundary layer adjacent to the interface. The inclusion behavior is examined thanks to Lagrangian particle tracking: Newton’s second law governing inclusion motion includes the buoyancy force, the pressure gradient force, the added mass force and the steady drag force.For the liquid steel/solid wall interface, the inclusion behavior is analyzed in the buffer layer and in the viscous layer. These layers are described according to Ahmadi’s model, which provides a kinematic representation of the turbulent structures responsible for deposition, i.e., the sweeps and the bursts of liquid. The numerical simulations show that the deposition is mainly controlled by sedimentation. However, since the direct interception contribution increases with the turbulence intensity, direct interception becomes dominant for the highest values of the friction velocity (greater than 0.1 m.s-1). When the hydrodynamic interactions between the inclusions and the solid surface are taken into account, the deposition velocity is significantly reduced. Finally, it should be noted that the inertial forces have a negligible effect on the inclusion deposition velocity. For the liquid steel/liquid slag interface, the inclusion turbulent deposition is investigated using direct numerical simulation of the liquid flow combined with Lagrangian particle tracking under conditions of one-way coupling. The interface is modeled as a non-deformable free-slip surface. Unsheared turbulence is generated by random forcing in a finite-height region parallel to the free-slip surface. In between, the turbulence diffuses toward the free surface. The Reynolds number at the interface varies from 68 to 235. The inclusion diameter varies from 10-5m to 5.10-5m and the particle to liquid density ratio from 0.5 (alumina inclusions) to 1 (fictitious inclusions). It appears that the deposition of alumina inclusions is controlled by sedimentation whereas direct interception is the only deposition mechanism for non-buoyant inclusions. In the latter case, the deposition velocity strongly depends on the surface Reynolds number. It is shown that the deposition velocity made dimensionless by the free surface characteristic velocity scales as the inclusion diameter made dimensionless by the Kolmogorov length scale calculated at the free surface. When the hydrodynamic interactions between the inclusions and the free surface are taken into account, the direct interception contribution of the deposition velocity is significantly reduced (about half of the value without hydrodynamic retardation) but the scaling law is conserved.

A bas nombre de Reynolds, le concept de voilure battante apparaît comme une alternative auxconcepts conventionnels de voilure fixe et voilure tournante. Dans le cadre d’une applicationpratique (micro-drones ou MAVs), l’évaluation de l’adaptabilité d’un tel mode de sustentationrequiert la compréhension fine des principaux mécanismes aérodynamiques mis en jeu et de leurimpact sur les efforts résultants. Ces derniers se caractérisent par une instationnarité forte et descomportements complexes.Les travaux de cette thèse se concentrent sur l'étude du mouvement de vol battu à bas Reynolds (del'ordre de 1000), dans une configuration de vol stationnaire. Le modèle est constitué d’une ailerectangulaire à profil symétrique animé d’un mouvement couplé de battement et de tangage. Cemode de sustentation se caractérise par la génération à proximité de l'aile de structurestourbillonnaires plus ou moins persistantes influant fortement les efforts appliqués à l’aile.L'objectif consiste à analyser l'évolution des mécanismes instationnaires et des efforts en résultant.L'étude porte en particulier sur une analyse approfondie d'un cas de référence, comparé ensuite àd'autre résultats lors d'une étude paramétrique portant sur l’influences de l'allongement d'une part, etde la cinématique du mouvement d'autre part.Les moyens d’investigations adoptés pour mener cette étude sont à la fois numériques etexpérimentaux, L’analyse repose d’une part sur une approche numérique DNS utilisant unetechnique de maillage « chimère », et d’autre part sur des mesures en bassin de type PIV 3D-3Crésolues en temps. La mesure directe des efforts instationnaires de faible niveau étant difficilementenvisageable, une part importante du travail a consisté à adapter une méthode d’évaluation desefforts par bilan de quantité de mouvement à partir des champs de vitesse PIV résolus en temps. Lespoints durs de cette approche, en particulier l’évaluation de la pression à partir des champs devitesse, font l’objet d’une attention particulière. / A low Reynolds number, flapping wings appears as an alternative to conventional concepts of fixed wings and rotary wings aircrafts. In the context of a practical application (micro air vehicles, MAVs), assessing the suitability of such mode of lift generation requires a detailed understanding of the key aerodynamic mechanisms involved and their impact on resulting forces. These are characterized by a strong unsteadiness and complex behaviors.This work focuses on the study of flapping flight at low Reynolds (around 1 000), in a hover configuration. The model consists of a rectangular wing with a symmetrical profile in a flapping motion. This mode is characterized by the generation of vortex structures more or less persistent that strongly influence the forces applied to the wing.The objective is to analyze the evolution of unsteady mechanisms and resulting forces. The study focuses in particular on a thorough analysis of e reference case, then compared to other results in a parametric study on the influence of aspect-ratio on the one hand, and on the kinematic of movement on the other.The means of investigation adopted for this study are both numerical and experimental. The analysis is based in part on a numerical approach using a DNS meshing technique “chimera”, and on experimental approach with 3C-3D TR-PIV measures. Direct measurement of unsteady low forces being difficult to consider, an important part of the work was to adapt a method for evaluating loads by applying momentum equation using PIV velocity fields. The bottleneck of this approach which is the evaluation of the pressure from the velocity fields is subject to special attention.

Cette étude est consacrée à l’analyse des écoulements réactifs supersoniques cisailléset, plus particulièrement, des couches de mélange compressibles pouvant se développerdans les moteurs ramjet et scramjet. Des méthodes numériques appropriées ont été implémentéeset vérifiées pour aboutir au développement d’un code de calcul numériquemassivement parallèle, appelé CREAMS (compressible reactive multi-species solver). Cedernier a été spécialement conçu pour conduire des simulations numériques haute précision(simulations numériques directes ou DNS) de ce type d’écoulements. Une attentionparticulière a été portée à la description des termes de transport moléculaire et des termessources chimiques de façon à considérer la description physique la plus fidèle possible desmélanges des gaz réactifs à haute vitesse, au sein desquelles les temps caractéristiqueschimiques et de mélange aux petites échelles sont susceptibles d’être du même ordre degrandeur. Les simulations des couches de mélange bidimensionnelles et tridimensionnelles,inertes et réactives, confirment l’importance des effets associés à la compressibilité et autaux de dégagement de chaleur. Les résultats ainsi obtenus diffèrent en certains points deceux issus d’autres simulations qui introduisaient certaines hypothèses simplificatrices :développement temporel, emploi d’une chimie globale ou encore lois de transport simplifiées.En revanche, ils reproduisent certains tendances déjà observées dans un certainnombre d’études expérimentales conduites dans des conditions similaires. / This study is devoted to the analysis of supersonic reactive shear flows and, in particular,compressible mixing layers that can develop inside the ramjet and scramjet engines.Appropriate numerical methods have been implemented and tested to achieve the developmentof a massively parallel numerical solver, called CREAMS (compressible reactivemulti-species solver). This tool was designed to conduct high-precision numerical simulations(direct numerical simulations or DNS) of such flows. Particular attention waspaid to the description of the molecular transport terms and chemical source terms toconsider the most accurate physical description of reactive gas mixtures at high velocity,in which the chemical and mixing time scales, corresponding to the smallest scalesof the flow, are susceptible to be of the same order of magnitude. Simulations of twoandthree-dimensional, inert and reactive, mixing layers confirm the importance of theeffects associated with compressibility and rate of heat release. The results obtained differin some points from other simulations which introduced simplifying assumptions such astemporal development, use of a global chemistry or a simplified description of the moleculartransport terms. Nevertheless, they reproduce some trends already observed in severalexperimental studies conducted under similar conditions.

Après plus de 50 ans de recherche, l'intérêt de la simulation des grandes échelles pour la simulation des écoulements instationnaires a été largement démontré et cette méthode est aujourd'hui utilisée pour une grande variété d'applications industrielles. Plusieurs classes de modèles sous-maille ont été proposées dont celle très connue des modèles de viscosité sous-maille souvent préférée pour sa simplicité et sa robustesse. Leur formulation comporte un coefficient qui doit être ajusté pour chaque type d'écoulement et qui a été analysé pour des géométries simples. L'objectif de ce travail est de réaliser des analyses a-priori de ces modèles dans un canal plan et un canal convergent-divergent à relativement grand nombre de Reynolds. Les influences du type de filtre et de la largeur du filtre sont systématiquement abordées pour chacune des statistiques. Le transfert d'énergie sous-maille et la dissipation sous-maille sont tout d'abord étudiés. Ensuite, les coefficients des modèles Smagorinsky, Smagorinsky dynamique, WALE et du modèle Sigma nouvellement proposé sont estimés a-priori. Il est démontré que les coefficients des quatre modèles sont non-homogènes dans le domaine de simulation et sont largement affectés par le gradient de pression adverse, principalement dans la zone de recirculation. Enfin, les corrélations entre les quantités exactes et leurs équivalents modélisés sont examinées. Les résultats montrent un faible niveau de prédiction des modèles sous-maille et une grande variabilité des quantités modélisées dans les régions de fort gradient de pression adverse. Ceci peut expliquer les difficultés pour obtenir de bons résultats LES dans une telle configuration / After more than 50 years of investigation, Large Eddy Simulation has demonstrated its benefit for unsteady flow simulation and is currently applied in a wide variety of engineering applications. Several classes of subgrid scale models were proposed, including the well known eddy viscosity models, usually preferred because of their simplicities and robustness. The formulation of these models includes a coefficient which needs to be analyzed for each flow configuration and which has been investigated in simple geometries.The aim of the present work is to perform a-priori analysis of subgrid scale models in plane channel flow and in a converging-diverging channel flow at fairly large Reynolds number.The influences of the filter type and filter width are systematically addressed in analyses of all statistics. The SGS energy transfer and energy dissipation are firstly analyzed.Then, the a priori estimate of the coefficients of subgrid scale models, including the standard Smagorinsky, Dynamic Smagorinsky, the WALE and the new updated sigma models, are investigated in detail. It is shown that, the coefficients of the four models are non-homogeneous in the simulation domain and are largely affected by the adverse pressure gradient, especially in the recirculation region. Finally, the correlations between the exact quantities and their counterparts modeled by the subgrid scale models with respect to three criteria are explored. The results show a low predictability of subgrid scale models and a strong variability of the modeled quantities in the region of strong adverse pressure gradient. This may explain the difficulty to obtain accurate LES results in such flow configuration

This thesis investigates turbulent wall-bounded flows using the Smoothed Particle Hydrodynamics (SPH) method. The first part focuses on the SPH method itself in the context of the Navier-Stokes equations with a special emphasis on wall boundary conditions. After discussing classical wall boundary conditions a detailed introduction to unified semi-analytical wall boundary conditions is given where the key parameter is a renormalization factor that accounts for the truncated kernel support in wall-bounded flows. In the following chapter it is shown that these boundary conditions fulfill energy conservation only approximately. This leads to numerical noise which, interpreted as form of Brownian motion, is treated using an additional volume diffusion term in the continuity equation where it is shown to be equivalent to an approximate Riemann solver. Two extensions to the boundary conditions are presented dealing with variable driving forces and a generalization to Robin type and arbitrary-order interpolation. Two modifications for freesurface flows are then presented, one for the volume diffusion term and the other for the algorithm that imposes Robin boundary conditions. The variable driving force is validated using a Poiseuille flow and the results indicate an error which is five orders of magnitude smaller than with the previous formulation. Discretising the wave equation with Robin boundary conditions proves that these are correctly imposed and that increasing the order of the interpolation decreases the error. The two modifications for flows under the influence of external forces significantly reduce the error at the free-surface. Finally, a dam break over a wedge demonstrates the capabilities of all the proposed modifications. With the aim of simulating turbulent flows in channels, the thesis moves on to extending the unified semi-analytical wall-boundary conditions to three dimensions. The thesis first presents the consistent computation of the vertex particle mass. Then, the computation of the kernel renormalization factor is considered, which in 3-D consists of solving an integral over a two dimensional manifold where the smoothing kernel intersects the boundary. Using a domain decomposition algorithm special integration areas are obtained for which this integral can be solved for the 5 th -order Wendland kernel. This algorithm is successfully applied to several validation cases including a dam break with an obstacle which show a significant improvement compared to other approximative methods and boundary conditions. The second part of this thesis investigates turbulent flows, in particular turbulent channel flow. This test case is introduced in detail showing both the physical properties as well as established numerical methods such as direct numerical simulation (DNS) and large eddy simulation (LES). In the penultimate chapter several SPH simulations of the turbulent channel flow are shown. The first section deals with a quasi DNS of the minimal-flow unit, a channel flow with a minimal domain size to sustain turbulent flow structures. The Eulerian statistics are compared to literature and show good agreement except for some wall-normal quantities. Furthermore, preliminary Lagrangian statistics are shown and compared to results obtained from a mesh-based DNS. The final simulation shows a LES of a full-sized channel at Reynolds number Re τ = 1000. The Eulerian statistics are compared to literature and the discrepancies found are explained using simulations of the Taylor-Green vortex, indicating that the momentum is not transferred appropriately due to an unresolved velocity-pressure-gradient tensor.