An investigation of the influence of various wall perturbations on the development of instability waves has been carried out through a series of experiments to study different aspects of the stability/transition problem. The first experiment was devoted to an investigation of the effects of an isolated step on the transition of the boundary layer over a flat plate. Steps of various heights, both positive and negative, produced an upstream movement of the transition point, which was interpreted as arising from an increase in the amplification of the instability waves. The transition Reynolds number and the resulting change in the so-called 'N-factor', used for transition prediction, has been correlated with the step height to provide a useful empirical engineering design tool. The second experiment concerned the effect of a two-dimensional (2-D) sharpedged short bump on instability waves (Tollmien-Schlichting (TS) waves). The experiment was designed to study the mechanisms by which a sharp-edged surface roughness element, in the form of a 2-D bump, modified the evolution of TS wave and hence affected the process of transition. This investigation demonstrated that the interaction of a 2-D TS wave with a 2-D bump was a many-sided complex problem. The bump played multiple roles as a disturbance generator, an amplifier and a promoter of three-dimensionality. The crucial parameter was the height. Nonlinearity was the main factor that caused the enhancement of boundary layer transition and manifested itself in the appearance of higher harmonics in the spectra. In the last experiment the bump was three-dimensional (3-D) and in the form of a cylinder. The experimental results revealed that the effect of a 3-D bump was confined to a wedge-shaped region downstream of the bump. The mean velocity distortion had a distinct spanwise distribution from which a pair of streamwise vortex structures could be inferred. The interaction of a 2-D TS wave with a 3-D bump generated oblique waves with a dominant band of spanwise wavenumbers as predicted by theoretical analysis. Three thresholds were found that defined bump height ranges where there was no-effect, a linear behaviour and a nonlinear one.

No description available.

No description available.

This thesis presents developments in the use of probability density function (PDF) kinetic equations to model the dispersion of inertial particles in turbulent boundary layers. The PDF kinetic equation is used as a master equation from which to construct continuum equations for the particle-phase, and these continuum equations form an in nite set of coupled equations which require closure in the particle velocity statistics. Furthermore, the continuum equations contain dispersion tensors which describe the e ect of the underlying uid turbulence on the dispersion of the particles throughout the ow eld. These dispersion tensors themselves require closure and in this thesis new closure models are developed which are non-local and attempt to take into account the e ects of turbulence inhomogeneity, anisotropy and particle-wall collisions on the dispersion tensors. The rst closure model developed is for particles dispersing under Stokes drag forcing only; appropriate for particles whose material density is much greater than that of the uid in which they are dispersed. This closure model is tested against equivalent particle tracking simulation data over a range of particle sizes and the closure model predictions are found to be in excellent agreement. In contrast to the new closure model predictions, the traditional `local' approximations to the dispersion tensors are found to be in signi cant error when compared to the particle tracking data. The closure model is then developed to account for particles dispersing under Stokes drag, added mass and gravitational forcing; added mass forcing being important for particles whose material density is comparable to or less than that of the uid in which they are dispersed. The modelling is presented and a discussion is given regarding the various complex terms that require approximation in this closure model. The closure model predictions are then compared against the alternative local approximations. It is seen that with added mass forcing the local approximations can be qualitatively and quantitatively di erent to the non-local predictions, whereas under only a drag force, errors in the local approximations are mainly quantitative. Finally, consideration is given to the forms of the dispersion tensors appearing in the PDF and continuum equations. It is shown theoretically that the dispersion tensors (and therefore the PDF and continuum equations themselves) are free from the so called `spurious drift' phenomena associated with certain types of models for predicting the dispersion of uid particles in incompressible, inhomogeneous turbulent ows. However, it is also shown that closure approximations applied to the dispersion tensors may result in the introduction of a spurious drift. Nevertheless, it is demonstrated that the arti cial drift introduced by closure approximations does not have any appreciable a ect on the dispersion tensors when they are describing the dispersion of inertial particles.

The potential for active control of low Reynolds number boundary layers using synthetic jet generators (SJG) has been established. The results from a four stage experimental study are presented in which the operational and geometric parameters of a rectangular slot choice SJG are optimised. A time-dependent, analysis of the SJG velocity profile is carried out in quiescent conditions prior to application of the SJG in (i) a nominally zero pressure gradient at plate boundary layer (l.54 > < l05 < ReX < 2.86 > <l05), and (ii) a simulated longitudinal pressure gradient, in order to asses the impact of the device on boundary layer development downstream A piezoelectric driven synthetic jet is used in this study in which the resonance characteristics of an intimal cavity are used to establish a fluid jet. Two characteristic resonance frequencies are identified; the mechanical resonance frequency of the diaphragm (FD), and the acoustic resonance frequency of the cavity (FC). The latter is shown to be the most energy efficient excitation frequency in terms of creating the highest synthetic jet velocity. Since the optimum operational and geometrical parameters are difficult to predict theoretically, a parametric study is presented in which, for a constant slot width (H=0.28mm) and resonance cavity geometry, an optimum slot length (L/H=l7.86~2l.43), which creates the highest synthetic jet velocity, can be established. Similarly, for a constant resonance cavity diameter, a narrower resonance cavity results in a higher synthetic jet velocity. Although not a experimental parameter, diaphragm clamping force is also seem to be significant in terms of synthetic jet velocity. When studied in conjunction with a zero-pressure gradient boundary layer - such that the jet exits normal to the surface -the orientation of the rectangular slot relative to the freestream (ß) is seen to result in a flow downstream of the slot which changes from a counter rotating vortex pair (when ß=0°) to a single streamwise vortex (when ß=20°). Increasing the slot angle beyond 20° diminishes the induced vortex structure such that, at ß=90°, the effect is similar to a surface obstruction. A time dependent analysis of the synthetic jet velocity profile shows that the jet has a structured fluctuation which, in the longitudinal sense, is strongest in the middle of slot (X/H=O, Y/I-I=0) and in a plane close to the slot (Z/H=3). The suction stroke of the synthetic jet is seen to induce a flow in the surrounding air which is identified by measurements at various lateral locations with a single component HWA probe. When a dual diaphragm is used for a SJG, synchronisation of the diaphragm displacement is seen to be important. If the velocity peaks corresponding to each diaphragm occur at a different phase in the excitation cycle, the performance increase attributed to the dual diaphragm operation is reduced. Studies of SJG effectiveness within a 2D boundary layer subjected to a controlled longitudinal pressure gradient show that activation of the SJG can both trigger laminar-turbulent transition and eliminate a laminar separation bubble. The potential for low Reynolds number flow control by virtual aeroshaping is therefore demonstrated.

No description available.

No description available.

The present thesis is concerned with mechanisms of sound generation by unsteady hydrodynamic motions in shear flows. Such problems are of great importance for the aviation industry, where noise reduction has long been a serious concern. The current state of the art in this field primarily rests on the acoustic analogy theory initiated by Lighthill (1952) and its numerous variants. An alternative method, based on matched asymptotic expansions, was developed by Crow (1970) among others. It was recently further advanced further within the framework of triple-deck theory to predict sound emission from certain viscous flows in the presence of a boundary (Wu 2002). We apply this approach to three aeroacoustic problems, which are deemed suitable for a triple-deck approach, with a view to analyze some fundamental sound generation processes on a self-consistent first-principle basis. The first problem addresses sound generation in a transonic boundary layer subjected to unsteady suction and injection. The analogous problem was analysed by Wu (2002) for subsonic flows but the theory breaks down at transonic speeds. The transonic effect first manifests itself when 1 −M2 = O(R−1/9), where M is the free-stream Mach number and R the global Reynolds number, which is assumed large. The unsteady flow due to the suction/injection is accommodated by a somewhat different triple-deck structure, in which the unsteady effect appears at the leading order in the upper deck, as was found by Bowles & Smith (1993). It is found that the resulting sound field is fundamentally different from its subsonic counterpart. The subsonic flow field involves an outer region, in which the pressure assumes an acoustic character. Such an outer region is not present in the transonic case, and it is not possible to express the solution in terms of multipoles since the source is not compact. Most importantly, it is found that the radiated sound produces a leading-order ‘back action’ effect on the source. The second problem is concerned with the acoustic radiation emitted by instability waves as they undergo rapid distortion, which has been recognized as one of prime mechanisms by which instability waves generate sound. We consider the situation where the rapid distortion in the Tollmien-Schlichting (T-S) waves is caused by a localized surface roughness in a compressible subsonic boundary layer. We find that in order to predict the leading-order acoustic pressure fluctuation, the first four terms in the expansion for the hydrodynamic field have to be determined. They contribute equally to the radiated sound because they act as octupole, quadrupole, dipole and monopole sources respectively. The analysis reveals two types of cancellations, which may explain the difficulties in accurately predicting aerodynamic sound. The first occurs in spectral space in the small-wavenumber limit; this cancellation renders the leading-order source to act as an octupole rather than a quadrupole source. The second type occurs in physical space, among the sources in different regions of the flow. The analysis also shows that a localized roughness influences the energetics of a T-S wave and this effect can be characterized by a transmission coefficient. The third problem that we analyze is that of sound emission due to a stationary source embedded in a boundary layer. Our aim is to obtain some generic results, which generalise the studies of individual cases to encompass a broad class of flows. For that purpose, we seek Green’s functions corresponding to an arbitrarily specified source located in the main deck. The temporal and spatial scales of the source are assumed to be compatible with those of triple-deck theory so that the near-field hydrodynamics can be described by a triple-deck structure and a fourth, outer region accommodates the acoustic pressure. The solution is sought in terms of an asymptotic series for the pressure. The required order of approximation depends on the radiating nature of the effective source q, which can be characterized by its behaviour in the small wavenumber k limit. If q = O(1) for k = 0, the first two terms are needed. They act as dipole and monopole sources respectively and so contribute equally to the sound in an asymptotic sense. If q = O(k) for k 1, the first three terms have to be obtained, which act as quadrupole, dipole and monopole sources respectively.

In order to gain a better understanding of the fundamental structure associated with turbulent flows over very rough, urban-type surfaces, laboratory experiments were undertaken in a subsonic wind tunnel facility at the University of Southampton. It was anticipated that undertaking this work would provide better insight into fundamental differences in the flow structure compared to smooth-wall surfaces. After modification of an existing facility to accommodate a longer working section length, testing proceeded over a regular array of cube roughness elements with a 25% area density. Measurements were conducted with hot-wire anemometry, Laser Doppler anemometry, and particle image velocimetry. Development of the particle image velocimetry technique to obtain accurate turbulence statistics over and among the roughness elements was successfully undertaken providing significant new analysis opportunities and results. Initial testing characterised a large-scale spanwise variation discovered within the boundary layer developing over the cube surface. It was found that mean velocity variation in the span at a height of 50% of the boundary layer thickness could exceed ±5%. Further testing was conducted at locations far enough downstream to minimise the amplitude of the variation. Time-averaged mean velocity and turbulence statistics were collected revealing the averaged flow features. The peak Reynolds shear stress near the cube surface was found to be a strong function of the relative boundary layer thickness compared to the roughness size. Quadrant analysis showed the instantaneous sweep motions found near the rough surface intermittently producing large percentages of the local shear stress. Spatial correlation analysis of the instantaneous field data collected with particle image velocimetry revealed long, streamwise-stretched regions of streamwise mean velocity cross correlation. Correlation analysis also allowed calculation of the structure angle of the streamwise velocity cross correlation and the associated integral length scales of the turbulence structure. Two integral length scales were found in certain locations near the cube surface highlighting the complex nature of the flow and inherent difference compared to smooth wall flows. Comparisons were made with existing direct numerical simulation studies over identical geometries showing many general similarities but also indicating differences associated with the assumptions governing each approach. Together, the experiments and analysis establish a broad picture of the distinct flow structure found in urban-type flows.

Surface flow visualisation and Particle Image Velocimetry (PIV) were used to investigate the flow-field of sub-boundary layer vane vortex generators (VVGs) and steady jet vortex generators (SJVGs) in a separated flow at 20m/s. The vortex generators were mounted on a 2D bump which was situated on the floor of a 350mm by 250mm wind tunnel and had a separation zone on its trailing edge. Surface flow visualisation was used to select VVG spacings and SJVG velocity ratios for effective separation control and investigate surface flow. Instantaneous and mean parameters downstream of the VVG and SJVG were measured non-intrusively using PIV. The instantaneous behaviour of the vortices was assessed using the coherence of a vortex and fluctuations in the location of instantaneous peak vorticity. Coherence is a parameter that was developed in the course of this research and had not been encountered in the literature at the time experiments ceased. The mean behaviour of the vortices was assessed using traditional variables, such as, mean, peak vorticity and its location, diameter and circulation of the vortex. A flat plate VVG study was performed to allow direct comparisons with previous work to be made. It was found that vortices generated by both VVGs and SJVGs were unsteady irrespective of geometry, became incoherent with downstream distance and had an exponential decay in mean peak vorticity. Surface flow visualisation indicated that for multiple VVG configurations the co-rotating configuration with a spacing of 3h effectively minimised separation on the 2D bump. For SJVGs, a velocity ratio of 1.0 was sufficient for separation control