Transitional and turbulent flows over a backward-facing step are physically highly complex. Apart from vastly different mean flow regimes and the rapid generation of turbulence, additional complexities arise from the presence of large coherent structures. For the present study, the mean flow, turbulence statistics and the origin of large coherent structures were investigated using Direct Numerical Simulations and turbulence modeling approaches. The latter included Large Eddy Simulations (LES) and state-of-the-art Reynolds-Averaged Navier-Stokes (RANS) computations. Wall-distance independent forms of the RANS models were developed, validated and calibrated. The ability of computing the step flows investigated and the associated computational costs were evaluated, for both LES and RANS. By employing harmonic forcing of the shear layer and a Fourier analysis in time and in the lateral direction the generation of coherent structures was linked to specific hydrodynamic instabilities. Comparison with references in the literature, resolution and domain size studies, and variations of inflow conditions established an accurate description of the mean flow and turbulence quantities and the level of sensitivity of the flow field to boundary conditions. From the controlled environment of the simulations, a simplified scenario was proposed for the creation of large coherent structures in transitional and turbulent step flows. The scenario suggests that Kelvin-Helmholtz, elliptical and centrifugal instabilities may be the relevant physical mechanisms for the observed primary, secondary and tertiary instabilities of the shear layer, respectively. The onset of the elliptical instability can also be described as a fundamental resonance of two waves. A cascade of subharmonic resonances is regarded to be responsible for vortex mergings and the generation of low frequency waves in the flow field. Furthermore, the simulations indicate that a three-dimensional global instability of the time and spanwise averaged separation bubble may be present. It was observed that the range of all unstable lateral wavelengths has a short-wave cutoff depending on Reynolds number and an upper bound on the order of the reattachment length.

Transitional and turbulent supersonic axisymmetric wakes are investigated by conducting various numerical experiments. The main objective is to identify hydrodynamic instability mechanisms in the flow at M = 2.46 for several Reynolds numbers, and relating these to coherent structures that are found from various visualization techniques. The premise for this approach is the assumption that flow instabilities lead to the formation of coherent structures. The effect of these structures on the mean flow is of particular interest, as they strongly affect the base drag. Three high-order accurate compressible codes were developed in cylindrical coordinates for this research: A spatial Navier-Stokes (N-S) code to conduct Direct Numerical Simulations (DNS), a linearized N-S code for linear stability investigations using two-dimensional basic states, and a temporal N-S code for performing local stability analyses. The ability of numerical simulations to deliberately exclude physical effects is exploited. This includes intentionally eliminating certain azimuthal/helical modes by employing DNS for various circumferential domain-sizes. With this approach, the impact of structures associated with certain modes on the global wake-behavior can be scrutinized. It is concluded that azimuthal modes with low wavenumbers are responsible for a flat mean base-pressure distribution and that k = 2 and k = 4 are the dominant modes in the trailing wake, producing a four-lobe wake pattern. Complementary spatial and temporal calculations are carried out to investigate whether instabilities are of local or global nature. Circumstantial evidence is presented that absolutely unstable global modes within the recirculation region coexist with convectively unstable shear-layer modes. The flow is found to be absolutely unstable with respect to modes k > 0 for ReD > 5,000 and with respect to the axisymmetric mode for ReD > 100,000. Furthermore, it is investigated whether flow control measures designed to weaken the naturally most significant modes can decrease the base drag. Finally, the novel Flow Simulation Methodology (FSM), using state-of-the-art turbulence closures, is shown to reproduce DNS results at a fraction of the computational cost.

A layered heat flux gauge, which can withstand a high temperature environment for applications such as for use on thermal protection shields on aeroentry vehicles, is analyzed, designed, fabricated, and tested. The heat flux gauge consists of two resistance temperature detectors on the top and bottom faces of a thin ceramic substrate. The heat flux is calculated from temperature measurements of the two temperature detectors. An analytical model is used to simulate the gauge response. Several numerical methods to calculate the heat flux are investigated to improve the time response of the gauge. The error due to gauge intrusiveness and the validity of one-dimensional heat transfer within the gauge is studied by solving a steady state two-dimensional composite problem using a semi-analytical approach. Gauge fabrication techniques and measurement devices are discussed. Testing apparatus, including a "close-to-entry" condition apparatus using an arcjet at low pressure and a conduction calibration furnace, are explained. Experimental data showing qualitative gauge response is presented.

The streamwise growth rate of the mixing layer downstream of a splitter plate with a 60° swept lambda notch and velocity ratio of 0.5 was spanwise independent and self-preserving once the velocity deficit from the splitter plate had disappeared. It was hypothesized the baseline mixing layer had a rapidly divergent growth component normal, and a non-divergent component parallel to the trailing edge of the notch. The streamwise growth rate of flow was not dependent on the spanwise dependence of the boundary layer. For forced cases the interaction of waves propagating from the oscillating flaperons influenced the streamwise growth rate of mixing layer. The forced case had a divergent growth component parallel to the trailing edge associated with streamwise rib vortices. The initial growth stages of the forced cases of this turbulent 3D mixing layer were dependent on the local Strouhal Number; similar to certain 2D cases.

Direct numerical simulations (DNS) are employed to investigate laminar boundary layer separation and its control by vortex generator jets (VGJs), i.e. by injecting fluid into the flow through an array of small holes. Particular focus is directed towards identifying some of the relevant physical mechanisms associated with VGJ control of low Reynolds number separation, as encountered in low-pressure turbine applications. In a comparison of selected controlled cases, pulsed VGJs are shown to be much more effective than steady VGJs, when the same momentum coefficient is used for the actuation. The formation and the dynamics of steady as well as unsteady flow structures are subsequently investigated in more detail. For steady VGJs, up to a certain "threshold" amplitude, angled injection is shown to be more effective than vertical injection, which is attributed to the fact that the generated longitudinal vortices remain closer to the wall while penetrating deeper into the boundary layer (in the spanwise direction). Beyond this "threshold" amplitude, however, vertical VGJ injection "suddenly" yields fully attached flow along the entire surface. This change in the global flow dynamics is explained by the formation of symmetric horseshoe-type vortices which are shown to augment the entrainment of high-momentum fluid from the free stream. For pulsed VGJs, the increased control effectiveness is attributed to the fact that hydrodynamic instabilities of the underlying flow can be exploited. When pulsing with frequencies to which the separated shear layer is naturally unstable, instability modes are shown to develop into large scale, spanwise coherent structures. These structures provide the necessary entrainment of high-momentum fluid to reattach the flow. In a series of additional simulations, the effects of varying the frequency as well as the duty cycle are investigated. While deviations from the "optimal" pulsing frequency are shown to result in increased separation losses, changes in the duty cycle have only a minor influence on the effectiveness of the control.

An experimental investigation into the characteristics of the unsteady laminar shear layer developed over a wall-mounted cavity was conducted, and the effects of the addition of a flush-mounted rotating cylinder to the leading edge or trailing edge of the cavity were quantified through extensive velocity and frequency measurements. It is shown that the injection of circulation resulting from the rotation of the leading edge cylinder caused the shear layer deflect into the cavity and re-attach to the cavity floor at greater cavity depths; also, the additional momentum caused a delay of the formation of shear-layer vortices, interrupting the cavity vortex formation-impingement feedback phenomenon. Higher leading-edge cylinder rotation speeds induced rapid laminar-to-turbulent transition, precluding any of the periodicity associated with laminar shear-layer flow. Furthermore, though the trailing-edge is the source of the feedback information for the self-sustained vortex formation, rotation of a cylinder at the trailing edge had little effect on the flow fields. Experiments were carried out with a ratio of separating boundary layer momentum thickness to impingement length of 6.07 x 10-3 .

An experimental study of the process of shock initiation of detonation in liquid nitromethane sensitized with diethylenetriamine (DETA) was conducted. In particular, the effect of the lateral boundary conditions on the sensitivity of the explosive to shock stimuli was investigated. The explosive was tested in a "gap test" arrangement. Various charge diameters and different materials for the capsule that contained the test explosive were used. The arrival time of the shock front was recorded at different locations along the test charge and the light emitted in the cases where detonation was initiated in the test explosive was monitored. / It is found that the minimum shock strength necessary to initiate detonation in the explosive test mixture is greater if the test charge diameter is small. It is also found that for a given shock strength, the minimum charge radius that permits initiation corresponds to the distance necessary to establish detonation. / In charges with small diameters, interactions between the input shock and the charge capsule walls permit initiation for input shock strengths that would normally not be sufficient to cause initiation. The nature of these interactions is complex and the outcome of a shock initiation experiment cannot be predicted from mere knowledge of the impedance of the capsule material or the shock speed within it.

In this thesis, an analytical investigation of the dynamics of three-body tethered satellite systems is presented. The analysis is performed for two cases: two-dimensional and three-dimensional motion. At first, the equations governing the motion of such systems are obtained assuming that the system center of mass moves in a circular orbit around the Earth. Then, the various equilibrium configurations are obtained. The equilibrium configurations are classified in eleven groups. Three of these groups are collinear configurations, and the others are triangular configurations. A stability analysis is performed for each equilibrium configuration. It is found that two of the collinear configurations are marginally stable. These are configurations with the three masses on the local vertical. The other collinear configurations are unstable. The triangular configurations are unstable except for some systems with a small middle mass. The results for two-dimensional motion are consistent with those for three-dimensional motion.

This work investigates fluid mechanical processes involving movement of shock waves inside channels. In particular, the prediction and control of shock motion in intakes is of great importance to high-speed flight and future spacecraft engines--it is a fundamental problem whose solution currently enjoys a high degree of interest. In the first part of the thesis, an analytical, numerical, and experimental study is performed in two-dimensions to assess the quasi-steady flow starting characteristics in an external-compression ramp-type intake. Using the popular model based on common understanding in the field and consistent with well-cited literature, it is shown that the traditional model based on one-dimensional flow across a planar shock is inaccurate. To remedy this, a novel analytical model is presented to predict off-design self-starting behaviour using a curved shock wave. The analytical results are supported by the numerical evidence from CFD simulations as well as by the experimental evidence; taken together, these results constitute overwhelming evidence supporting the arguments presented. In the second part of the thesis, an analytical and numerical study is performed in the unsteady, quasi-one-dimensional setting to assess the feasibility of predicting shock dynamics inside an intake using a simple, approximate, yet representative model. There is very little widely-available knowledge on impulsive flow starting in the literature, and there is no existing theory for analytically predicting impulsive flow starting phenomena discussed here. The well-known Chester-Chisnell-Whitham formulation is extended using a novel model (labelled lifting isentrope); it is effectively a provision for non-uniform, unsteady flow upstream of the shock wave. It is found that a singularity exists in the CCW model, which prevents its application to problems in impulsive flow starting. There is little or no literature on this singularity. To resolve these issues, a novel for / Ce travail examine des processus de la mécanique des fluides impliquant le mouvement des ondes de choc a travers des chaînes. En particulier, la prédiction et le contrôle de mouvement du choc dans les entrées d'air ont une grande importance pour le vol ultra-rapide et pour les moteurs des vaisseau spatiaux futurs--c'est un problème fondamental dont la solution attire présentement un haut degré d'intérêt. Même si les processus de démarrage et non-démarrage d’entrées d’air sont considérés bien compris au niveau conceptuel, d'importants détails restent à régler.Dans la première partie de la thèse une étude analytique, numérique et expérimentale est réalisée. L'étude est bidimensionnelle dans le but d’évaluer les caractéristiques de l'écoulement quasi-stationnaire dans une entrée d’air de type rampe avec compression externe. À l'aide d’un modèle bien-connu basé sur la compréhension commune dans le monde scientifique et selon la littérature de spécialité on montre que le model traditionnel basé sur l'écoulement unidimensionnel sur un plan de choc est inexact. Pour corriger cette situation un nouveau modèle analytique est présenté pour montrer le démarrage d’entrées d’air en utilisant une onde de choc courbe. Les résultats analytiques sont bases sur des preuves numériques obtenus des simulations CFD ainsi que par éléments expérimentales. Mises ensemble ces résultats constituent des preuves écrasantes à l'appui les arguments présentés.Dans la deuxième partie de la thèse une étude analytique et numérique est réalisée sur le sujet de l'écoulement instable, quasi-unidimensionnel. Le but est d’évaluer la faisabilité de la prédiction de la dynamique des chocs dans une entrées d’air en utilisant un modèle e simple, approximatif mais représentative.Il n’y a pas beaucoup des études et connaissances sur le démarrage impulsive d’entrées d’air disponibles et il n’existe pas une théori

The unsteady flows generated by the harmonically variable inflow velocities are studied in this thesis by using a numerical method for the time-accurate solution of the Navier-Stokes equations. This method, which has been developed by Mateescu and Venditti, uses an implicit three-point-backward scheme for the real time discretization, and a pseudo-time discretization based on a relaxation procedure using artificial compressibility. A special decoupling procedure is used to eliminate the pressure from the momentum equations with the aid of the continuity equation in pseudo-time, in order to reduce the problem to the efficient solution of scalar tridiagonal systems of equations. The method uses a finite difference formulation on a stretched staggered grid, which was validated for the steady incompressible flows in a duct with a downstream-facing step, by comparison with previous computational and experimental results. / This thesis first presents the solutions obtained for the unsteady flows with multiple separation regions in a duct with fixed geometry, which are generated by the variation in time of the inflow velocities. Then, the solutions for the unsteady flows generated by both an oscillating wall and the variation in time of the inflow velocity are also presented. The influence of the Reynolds number, of the inflow velocity amplitudes, and of the phase difference is also thoroughly studied.