A robotically-controlled actuation system has been developed and built to perform active open-loop flow control experiments on transitional and turbulent backward-facing step flows in water. Control of the reattaching shear layer used hydraulic suction-and-blowing actuation emanating from 128 individual ports along the separation edge of the step. Each ports perturbation was periodic in time, but individually controlled to produce either spanwise-invariant (2D) or spanwise-varying (3D) spatial actuation profiles. An image processing system and special aqueous tuft were developed to measure the length of the recirculation bubble. Multiple images of a tuft array were time-averaged to do so. In general, 3D forcing was no more effective in reducing bubble length than 2D forcing. However, greater local spanwise reductions in reattachment length were observed for some cases of spanwise-varying forcing. Backlit dye was used to track the evolution of vorticity in the flow in video and still images.

A interação entre troposfera e estratosfera tem grande influência e é de grande importância nos processos de ciclogênese em superfície. Entretanto, não se conhece exatamente a frequência destas interações e nem como podem influenciar na intensidade de ciclones em geral. Este trabalho tem como objetivo geral estudar como os altos níveis da atmosfera afetam o desenvolvimento de ciclones em superfície na América do Sul e Oceano Atlântico Sul, utilizando o conceito de vorticidade potencial (VP). Através de dados de ciclones rastreados em superfície e VP em 300 hPa desenvolveu-se um algoritmo que associa automaticamente os ciclones em superfície com anomalias de vorticidade potencial (AVP). Para o período 1998-2003, fez-se então a separação dos ciclones em associados (AAVP) e não-associados (NAVP) a AVP. De forma geral, observou-se que a maior parte dos ciclones AAVP concentra-se na região oceânica extratropical e os NAVP preferencialmente na região continental próximo de 30°S e em latitudes subtropicais. Para todo o período analisado, o número total de ciclones AAVP (55%) superou o número de NAVP (45%), sendo o ano de 2002 o único que apresentou número maior de eventos NAVP. Quanto à distribuição sazonal, os ciclones AAVP são mais frequentes nos meses de inverno e primavera, enquanto que os NAVP nos meses de verão. O tempo de vida dos NAVP é menor que dos AAVP, além de possuírem menor intensidade (de acordo com a pressão central média). Além destes fatores, a distância percorrida e a velocidade médias dos ciclones NAVP são menores do que dos AAVP. As composições dos campos sinóticos mostram que nos eventos NAVP, independente da estação do ano, a troposfera é mais quente que nos AAVP. Nos NAVP a forçante térmica é essencial para a formação do ciclone, enquanto que nos AAVP a AVP induz vorticidade ciclônica primeiro em altos níveis, que então se propaga para baixos níveis. Através da análise dos campos sinóticos, notou-se maior baroclinia nos casos NAVP, pois tanto os cavados em altos e médios níveis quanto a corrente de jato permanecem favorecendo o desenvolvimento do ciclone em superfície, enquanto que nos AAVP o centro do ciclone em superfície está verticalmente quase alinhado ao cavado. Centros de vorticidade relativa ciclônica em 500 hPa desprendem-se do escoamento de oeste em todas as estações para os casos AAVP, porém, no verão, isto também é visto nos NAVP. / The process of troposphere-stratosphere interaction has influence and is very important on surface cyclogenesis. However, the frequency of these interactions and how they influence the intensity of cyclones is not known exactly. The main objective of this work is to study how the upper levels affect the development of surface cyclones in South America and South Atlantic Ocean using the concept of potential vorticity (PV). Cyclone tracking data and 300 hPa PV were used to develop an algorithm that automatically associates the surface cyclones with potential vorticity anomaly (PVA). For the period 1998-2003, the cyclones were separated as associated (APVA) and non-associated (NPVA) with PVA. In general, it was observed that most of the APVA cyclone was concentrated in extratropical oceanic region, while NPVA cyclones form over the continent preferably around 30°S and subtropics. The total number of APVA cyclones (55%) exceeds the number of NPVA (45%), except for 2002. In regard to seasonal distribution, the APVA cyclones are more frequent in winter and spring months while NPVA in summer months. The lifetime of NPVA cyclones is shorter and they are less intense than APVA (according to the average central pressure). In addition to these factors, the mean traveled distance and mean velocity are smaller in the NPVA than in APVA. The composites of the synoptic fields show that in NPVA events, regardless of the season, the troposphere is warmer than in APVA. In NPVA cases the thermal forcing is essential to the cyclogenesis, while in the APVA the cyclonic vorticity induced by PVA at upper levels propagating to low levels is more important. The NPVA cases present more baroclinic characteristics which the upper and mid-level troughs accompanied by the jet stream favoring the surface cyclone development, whereas in the APVA the surface cyclone center remains almost vertically aligned with these troughs. For APVA cases, the centers of cyclonic vorticity at 500 hPa detach from westerly flow in all seasons however in summer this is also seen in NPVA.

Flow mechanisms of compound meandering channels are recognised to be far more complicated than compound straight channels. The compound meandering channels are mainly characterised by the continuous variation of mean and turbulent flow parameters along a meander wavelength; the existence of horizontal shear layer at the bankfull level and the presence of strong helical secondary flow circulations in the streamwise direction. The secondary flow circulations are very important as they govern the advection of flow momentum, distort isovels, and influence bed shear stress, thus producing a complicated and fully three-dimensional turbulent flow structures. A great deal of experiments has been conducted in the past, which explains flow mechanisms, mixing patterns and the behaviour of secondary flow circulations. However, a complete understanding of secondary flow structures still remains far from conclusive mainly because the secondary flow structures are influenced by the host of geometrical and flow parameters, which are yet to be investigated in detail. The three-dimensional Reynolds-averaged Navier-Stokes and continuity equations were solved using a standard Computational Fluid Dynamics solver to predict mean velocity, secondary flow and turbulent kinetic energy. Five different flow cases of various model scales and relative depths were considered. Detailed analyses of the measured and predicted flow variables were carried out to understand mean flow mechanisms and turbulent secondary flow structures in compound meandering channels. The streamwise vorticity equation was used to quantify the complex and three-dimensional behaviour of secondary flow circulations in terms of their generation, development and decay along the half-meander wavelength. The turbulent kinetic energy equation was used to understand energy expense mechanisms of secondary flow circulations. The strengths of secondary flow circulations were calculated and compared for different flow cases considered. The main findings from this research are as follows. The shearing of the main channel flow as the floodplain flow plunges into and over the main channel influences the mean and turbulent flow structures particularly in the crossover region. The horizontal shear layer at the inner bankfull level generates secondary flow circulations. As the depth of flow increases, the point of generation of secondary flow circulations moves downstream. The secondary shear stress significantly contributes towards the generation of streamwise vorticity and the production of turbulent kinetic energy. The rate of turbulence kinetic energy production was found to be higher than the rate of its dissipation in the crossover region, which demonstrates that the turbulence extracts more energy from the mean flu\\' than what is actually dissipated. This also implies that, in the crossover region, the turbulence is always advected downstream by the mean and secondary flows, The strength of geometry induced secondary flow circulation increases with the increase in the relative depth.

La finalité de cette thèse est de fournir une meilleure compréhension de la turbulence homogène anisotrope soumise à un forçage barocline. À cette fin, nous utilisons une approche numérique pseudo-spectrale basée sur la transformation de Rogallo. L’utilisation d’un tel algorithme nous permet de considérer une asymétrie des fonctions de probabilité en faveur des évènements négatifs est observée. Le lien entre la distribution de vorticité potentielle et celle d’un scalaire passif est également étudié. Il est montré qu’à faible nombre de Richardson, c’est le mode vortex (à vorticité potentielle nulle) qui contient les plus importantes fuctuations de scalaire. Un écoulement homogène dans les trois directions de l’espace. Plusieurs simulations numériques directes (DNS) sont effectuées dans un contexte assez proche des écoulements géophysiques que l’on retrouve entre autre dans la stratosphère, où un gradient constant de vitesse zonale vient se coupler à un gradient constant de densité dans un repère tournant. Les résultats obtenus s’articulent autour de quatre axes principaux. Tout d’abord, une étude linéaire à temps fini est présentée en vue de compléter les résultats existants sur la dynamique linéaire asymptotique. La solution linéaire est décomposée en une partie ‘onde’ (qui se propage) et une partie dite ‘vortex’(stationnaire). L’étude analytique est complétée par un modèle synthétique de turbulence (Kinematic Simulation ou KS) basé sur la théorie de la distorsion rapide(RDT). Nous montrons qu’une distribution initiale non nulle de vorticité potentielle linéarisée peut conduire à d’importantes croissances transitoires. Ce résultat pourrait s’étendre à des modélisations du climat ou météorologique, où la distribution initiale de vorticité potentielle semble avoir autant d’importance que la distribution initiale de température ou de vitesse. Ensuite, nous consacrons une partie de notre étude à l’analyse paramétrique et à la stabilité de l’écoulement. Plusieurs DNS sont effectuées pour différents taux de rotation et stratification. Le diagramme de stabilité obtenu montre que pour de faibles taux de rotation, la limite de stabilité est identique à celle connue des écoulements sans rotation. À plus faible nombre de Rossby — lorsque la baroclinicité devient importante — la limite linéaire de stabilité Ri = 1 relative à l’instabilité symétrique est confirmée. La coexistance de l’instabilité barocline avec l’instabilité symétrique est également clarifiée. Une analyse énergétique détaillée mène à la conclusion suivante : la stratification doit être suffisamment importante (Ri ' 1) pour que l’instabilité barocline soit dominante i.e. que la conversion d’énergie potentielle soit la source principale d’énergie cinétique turbulente. Dans le cas contraire, l’instabilité symétrique — qui tire son énergie de l’énergie cinétique de l’écoulement moyen et non de son énergie potentielle — domine la dynamique de l’écoulement. Le troisième axe d’étude concerne la turbulence à proprement parler. En conséquence de l’ajustement géostrophique, le vent thermique force la turbulence d’une manière naturelle, en opposition à d’autres méthodes de forçage stochastique. L’émergence de structures dans le contexte barocline est approfondie. Des statistiques Euleriennes sont présentées afin de fournir une caractérisation fine de l’anisotropie de l’écoulement. Enfin, nous étendons notre étude à la caractérisation de la vorticité potentielle turbulente. Les fonctions de probabilité de la vorticité potentielle d’Ertel montrent que des anomalies sont présentes dans les configurations instables. En particulier, une asymétrie des fonctions de probabilité en faveur des évènements négatifs est observée. Le lien entre la distribution de vorticité potentielle et celle d’un scalaire passif est également étudié. Il est montré qu’à faible nombre de Richardson, c’est le mode vortex (à vorticité potentielle nulle) qui contient les plus importantes fuctuations de scalaire. / The main objective of this thesis is to provide a better understandingof homogeneous turbulence dynamics under an external baroclinic forcing.To achieve this goal, we use a pseudo-spectral code based on the Rogallo transformation.The use of such an algorithm allows to assume homegeneity in the threespatial directions. Direct Numerical Simulations (DNS) are done in a context representativeof geophysical baroclinic flows in the middle atmosphere: superpositionof a uniform mean zonal flow with stable vertical stratification and frame rotation.The results we obtained are then presented along four axes.First, a finite-time linear analysis is done to complete previous asymptotic results.The linearized flow is decomposed into a propagating (wave) and stationary (vortex)part. The analytical work is completed by a Kinematic Simulation (KS) modelbased on Rapid Distortion Theory (RDT). It is shown that the linearized potentialvorticity mode can produce dramatic transient growth of the kinetic energy if nonzeroinitially. The consequence of such a result is then of capital interest in climatemodelling, where the initial distribution of potential vorticity seems to have moreimportance than other eulerian quantities (temperature or velocity).The second axis is dedicated to a parametric analysis of the flow stability. SeveralDNS are done for different rotation and stratification rates. The derived stabilitydiagram shows that at low rotation rates, the stability bound for purely shearedstratifiedflows is recovered. At higher rotation rates — when baroclinicity is dominant— the linear bound for the so-called symmetric instability is confirmed. Thecoexistence of baroclinic and symmetric instabilies is also clarified. A complete energeticanalysis leads to the conclusion that stratification must be sufficiently highto enhance potential energy release through baroclinic instability. If not, symmetricinstability — driving its energy from the kinetic energy of the mean flow and notfrom the potential energy of the mean flow — is found to dominate the dynamics.The third axis is devoted to a characterisation of homogeneous turbulence submittedto an external baroclinic forcing. As coming from the geostrophic adjustment,the thermal shear allows an organic forcing of turbulence, in opposition to ad-hocarticificial forcing. The structures associated with the simultaneous presence of rotation,stratification and shear are investigated. Eulerian statistics are gathered togive a sharp characterisation of the spatial anisotropy of the flow.Finally, we open our work to the study of turbulent potential vorticity. Probabilitydensity functions of Ertel’s potential vorticity show that potential vorticityanomalies are present in unstable configurations. In particular, an asymmetry ofthe probability density functions toward negative events is observed. An attemptto link potential vorticity dynamics with scalar mixing in baroclinic flows is donethrough joint probability functions analysis.

La première partie concerne les interactions acoustique-vorticité dans les écoulements cisaillés linéaires incompressibles, qui peuvent être décomposés en la somme d’une partie hyperbolique et d’une partie rotation solide. L’écoulement de Couette en est un exemple. En utilisant la démarche non-modale , les équations d’évolution de perturbations compressibles se réduisent à une EDO de dimension trois en temps, qui dépend d’un paramètre adimensionné ε représentant le rapport entre le taux de cisaillement de l’écoulement et la fréquence des perturbations. Pour ε faible, la méthode WKB permet d’exhiber naturellement trois modes (deux modes acoustiques et un mode de vorticité) et permet de mettre en évidence des couplages entre ces modes. Ces couplages sont exponentiellement faible en 1/ε, et ne peuvent être pris en compte par une méthode asymptotique. Ils semblent être liés à la partie hyperbolique de l’écoulement. La seconde partie traite de la réflexion d'une onde par une surface de géométrie complexe. Une transformation conforme permet de transformer une frontière complexe en une frontière plane, mais fait apparaître des coefficients non constants dans les équations en volume. Celles-ci sont résolues au moyen de la méthode de la matrice d’impédance multimodale qui ramène le problème à une équation de Riccati pour la matrice d’impédance. Une méthode pour trouver des géométries admettant des modes piégés est proposée. Puis la méthode de résolution est appliquée à la modélisation de la couche limite visqueuse d’un fluide oscillant au contact d’une surface complexe périodique. Une solution perturbative est proposée. La présence de zones de recirculation est étudiée. / The first part is a study of the interactions between acoustic and vorticity perturbations in linear incompressible shear flows, which can decomposed as a sum of a hyperbolic part and of a rigid rotation part. The plane Couette flow is an example of such flows. By using the non-modal approach, the equations governing the evolution of compressible perturbations reduce to an ODE of dimension three in time, which depends on a dimensionless parameter ε representing the ratio between the shear rate of the flow and the frequency of the perturbations. For small ε values, the WKB method allows us to exhibit naturally three modes (two acoustic modes and one vorticity mode) and to highlight couplings between these modes. These couplings are exponentially small in 1/ε, and cannot be taken into account by an asymptotic method. They seem to be linked to the hyperbolic part of the flow.The second part deals with the reflection of a wave by a geometrically complex surface. A conformal mapping allows us to transform a complex boundary into a plane boundary, but makes appear varying coefficients in the bulk equations. These equations are then solved with the multimodal impedance matrix method, which reduce the problem to a Riccati equation for the impedance matrix. A method to find geometries allowing for the existence of trapped modes is proposed. Then the solving method is applied to the modeling of the viscous boundary layer of a fluid oscillating near a periodical rough surface. A perturbative solution is proposed. The presence of recirculation areas is studied.

Wavelike spiral asymmetries are an intriguing aspect of Tropical Cyclone dynamics. Previous work hypothesized that some of them are Vortex Rossby Waves propagating on the radial gradient of mean–flow relative vorticity. In the Intermittently Forced Vortex Rossby Wave theory, intermittent convection near the eyewall wind maximum excites them so that they propagate wave energy outward and converge angular momentum inward. The waves’ energy is absorbed as the perturbation vorticity becomes filamented near the outer critical radii where their Doppler–shifted frequencies and radial group velocities approaches zero. This process may initiate outer wind maxima by weakening the mean–flow just inward from the critical radius. The waves are confined to a relatively narrow annular waveguide because of their slow tangential phase velocity and the narrow interval between the Rossby wave cut–off frequency, where the radial wavenumber is locally zero, and the zero frequency, where it is locally infinite.

Lean flame blowout is investigated experimentally within a high-speed combustor to analyze the temporal extinction dynamics of turbulent premixed bluff body stabilized flames. The lean blowout process is induced through fuel flow reduction and captured temporally using simultaneous high-speed particle imaging velocimetry (PIV) and CH* chemiluminescence. The evolution of the flame structure, flow field, and the resulting strain rate along the flame are analyzed throughout extinction to distinguish the physical mechanisms of blowout. Flame-vortex dynamics are found to be the main driving mechanism of flame extinction; namely, a reduction of flame-generated vorticity coupled with an increase of downstream shear layer vorticity. The vorticity dynamics are linked to hydrodynamic instabilities that vary as a function of the decreasing equivalence ratio. Frequency analysis is performed to characterize the dynamical changes of the hydrodynamic instability modes during flame extinction. Additionally, various bluff body inflow velocity regimes are investigated to further characterize the extinction instability modes. Both equivalence ratio and flow-driven instabilities are captured through a universal definition of the Strouhal number for the reacting bluff body flow. Finally, a Karlovitz number-based criterion is developed to consistently predict the onset of global extinction for different inflow velocity regimes.

This document details original numerical studies performed by the author pertaining to the propellant oxidizer, ammonium perchlorate (AP). Detailed kinetic mechanisms have been utilized to model the combustion of the monopropellants AP and ADN, and a two-dimensional diffusion flame model has been developed to examine the flame structure above an AP/HTPB composite propellant. This work was part of an ongoing effort to develop theoretically based, a priori combustion models. The improved numerical model for AP combustion utilizes a “universal” gas-phase kinetic mechanism previously applied to combustion models of HMX, RDX, GAP, GAP/RDX, GAP/HMX, NG, BTTN, TMETN, GAP/BTTN, and GAP/RDX/BTTN. The universal kinetic mechanism has been expanded to include chlorine reactions, thus allowing the numerical modeling of AP. This is seen as a further step in developing a gas-phase kinetic mechanism capable of modeling various practical propellants. The new universal kinetic mechanism consists of 106 species and 611 reactions. Numerical results using this new mechanism provide excellent agreement with AP's burning rate, temperature sensitivity, and final species data. An extensive literature review has been conducted to extract experimental data and qualitative theories concerning ADN combustion. Based on the literature review, the first numerical model has also been developed for ADN that links the condensed and gas phases. The ADN model accurately predicts burning rates, temperature and species profiles, and other combustion characteristics of ADN at pressures below 20 atm. Proposed future work and modifications to the present model are suggested to account for ADN's unstable combustion at pressures between 20 and 100 atm. A two-dimensional model has been developed to study diffusion in composite propellant flames utilizing a vorticity formulation of the transport equations. This formulation allows for a more stable, robust, accurate, and faster solution method compared to the Navier-Stokes formulations of the equations. The model uses a detailed gas-phase kinetic mechanism consisting of 37 species and 127 reactions. Numerical studies have been performed to examine particle size, pressure, and formulation effects on the flame structure above an AP/HTPB propellant. The modeled flame structure was found to be qualitatively similar to the BDP model. Results were consistent with experimental observations. Three different combustion zones, based on particle size and pressure, were predicted: the AP monopropellant limit, the diffusion flame, and a premixed limit. Mechanistic insights are given into AP's unique combustion properties.

Experimental methods for the investigation of trailing vortex strengths, total lift, and lift distributions for three-dimensional wings in close proximity flight were developed. With these experiments we model compound aircraft flight either docked tip-to-tip, or flying in formation. There is a distinct lack of experimental formation flight data using three-dimensional wing models for tests. The absence of fixed walls on either end of the wing permits the development of the asymmetric shedding of vortices, and the determination of the asymmetric circulation distribution induced by the proximity of the leading wing. The pair consisted of a swept NACA-0012 non-cambered wing simulating one half of a leading aircraft and a rectangular cambered NACA 63-420 wing simulating the trailing aircraft. Important aspects of the work included theoretical development, experimental setup, data acquisition and processing, and results validation. Experimentally determining the lift for formation flight, in addition to the local flow behavior for a pair of wings, can provide valuable insight for the proposition of flying actual aircraft in formation to increase mission efficiency. To eliminate the need for bulky mounting stings and direct load measurement devices that can potentially interfere with the local flowfield, a minimally invasive velocity probe method is developed. A series of experiments were performed to assist with the development of the method. Velocity and vorticity distributions obtained along a near-field plane were processed to calculate wingtip vortex strengths. Additionally, vortex position instabilities and the shedding of vorticity inboard of the wingtips were observed. To determine the circulation distributions for the trailing wing, the initial method is modified. By processing velocity information acquired in a near-field plane, both the lift and induced drag were calculated for the trailing airfoil. Comparisons are made to directly measured loads and to results reported earlier. Directly measured lift and drag coefficients were found to agree with existing literature. / Master of Science

This thesis describes research on acoustically excited bluff body flow-fields, motivated by the problem of combustion instabilities in devices utilizing these types of flame-holders. Vortices/convective-structures play a dominant role in perturbing the flame during these combustion instabilities. This thesis addresses a number of issues related to the origin, evolution and the interaction of these structures with the flame. The first part of this thesis reviews the fluid mechanics of non-reacting and reacting bluff body flows. The second part describes the spatio/temporal characteristics of bluff-body flames responding to excitation. The key processes controlling the flame response have been identified as 1) the anchoring of the flame at the bluff body, 2) the excitation of flame-front wrinkles by the oscillating velocity field and 3) flame propagation normal to itself at the local flame speed. The first two processes control the growth of the flame response and the last process controls the decay. The third part of this thesis describes the effect of acoustic excitation on the velocity field of reacting bluff body flows. Acoustic disturbances excite the Kelvin-Helmholtz (KH) instability of the reacting shear layer. This leads to a spatially decaying vorticity field downstream of the bluff body in the shear layers. The length over which the decay occurs was shown to scale with the length of the recirculation zone of the bluff body, i.e. the length over which the velocity profile transitions from shear layer to wake. The flame influences this decay process in two ways. Gas expansion across the flame reduces the extent of shear by reducing the magnitude of negative velocities within the recirculation zone. This combined with the higher product diffusivity reduces the length of the recirculation zone, thereby further augmenting the decay of the vorticity fluctuations. Lastly, these results also revealed phase jitter - a cycle-to-cycle variation in the position of the rolled-up vortices. Close to the bluff-body, phase jitter is very low but increases monotonically in the downstream direction. This leads to significant differences between instantaneous and ensemble averaged flow fields and, in particular, the decay rate of the vorticity in the downstream direction.