We present 10 new equilibrium solutions to plane Couette flow in small periodic cells at low Reynolds number Re and two new travelling-wave solutions. The solutions are continued under changes of Re and spanwise period. We provide a partial classification of the isotropy groups of plane Couette flow and show which kinds of solutions are allowed by each isotropy group. We find two complementary visualizations particularly revealing. Suitably chosen sections of their three-dimensional physical space velocity fields are helpful in developing physical intuition about coherent structures observed in low-Re turbulence. Projections of these solutions and their unstable manifolds from their ∞-dimensional state space on to suitably chosen two- or three-dimensional subspaces reveal their interrelations and the role they play in organizing turbulence in wall-bounded shear flows.

Measurements were made of two components of the average and fluctuating velocities, and of the local self-diffusion coefficients in a flow of granular material. The experiments were performed in a 1 m-high vertical channel with roughened sidewalls and with polished glass plates at the front and the back to create a two-dimensional flow. The particles used were glass spheres with a nominal diameter of 3 mm. The flows were high density and were characterized by the presence of long-duration frictional contacts between particles. The velocity measurements indicated that the flows consisted of a central uniform regime and a shear regime close to the walls. The fluctuating velocities in the transverse direction increased in magnitude from the centre towards the walls. A similar variation was not observed for the streamwise fluctuations. The self-diffusion coefficients showed a significant dependence on the fluctuating velocities and the shear rate. The velocity fluctuations were highly anistropic with the streamwise components being 2 to 2.5 times the transverse components. The self-diffusion coefficients for the streamwise direction were an order-of-magnitude higher than those for the transverse direction. The surface roughness of the particles led to a decrease in the self-diffusion coefficients.

This paper presents the first detailed comparisons between experiments and direct numerical simulations (DNS) of inertial particle clustering in nearly isotropic ‘box turbulence’. The experimental system consists of a box 38cm in each dimension with fans in the eight corners that sustain nearly isotropic turbulence in the centre of the box. We inject hollow glass spheres with a mean diameter of 6 μm and measure the locations of several hundred particles in a 1 cm3 volume in the centre of the box using three-dimensional digital holographic particle imaging. We observe particle concentration fluctuations that result from inertial clustering (sometimes called ‘preferential concentration’). The radial distribution function (RDF), a statistical measure of clustering, has been calculated from the particle position field. We select this measure because of its relevance to the collision kernel for particles. DNS of the equivalent system, with nearly perfect parameter overlap, have also been performed. We observe good agreement between the RDF predictions of the DNS and the experimental observations, despite some challenges in the interpretation of the experiments. The results provide important guidance on ways to improve the measurement.

The forced reattachment of flow to an inclined flat surface, simulating a simple flap, was investigated experimentally. The transition from a separated to an attached state of the flow was initiated by an abrupt change in the frequency and the amplitude of periodic perturbations emanating from a slot at the flap shoulder. The excitation parameters determined the total duration of the reattachment process. Minimum reattachment time occurred at an optimal excitation frequency of $F_{\hbox{\scriptsize{\it opt}}}^{+} \approx 1.5$, which was independent of amplitude and flap inclination. The control over the process was achieved by enhancing large spanwise vortices in the flow. Spatial amplification of consecutive vortices induces mean transport of fluid away from the flap surface which causes the main stream to reattach. The time scales of the excitation are at least an order of magnitude smaller than the typical reattachment times.

The effects of resistive forces on unsteady shallow flows over rigid horizontal boundaries are investigated theoretically. The dynamics of this type of motion are driven by the streamwise gradient of the hydrostatic pressure, which balances the inertia of the fluid and the basal resistance. Drag forces are often negligible provided the fluid is sufficiently deep. However, close to the front of some flows where the depth of the moving layer becomes small, it is possible for drag to substantially influence the motion. Here we consider three aspects of unsteady shallow flows. First we consider a regime in which the drag, inertia and buoyancy (pressure gradient) are formally of the same magnitude throughout the entire current and we construct a new class of similarity solutions for the motion. This reveals the range of solution types possible, which includes those with continuous profiles, those with discontinuous profiles and weak shocks and those which are continuous but have critical points of transition at which the gradients may be discontinuous. Next we analyse one-dimensional dam-break flow and calculate how drag slows the motion. There is always a region close to the front in which drag forces are not negligible. We employ matched asymptotic expansions to combine the flow at the front with the flow in the bulk of the domain and derive theoretical predictions that are compared to laboratory measurements of dam-break flows. Finally we investigate a modified form of dam-break flow in which the vertical profile of the horizontal velocity field is no longer assumed to be uniform. It is found that in the absence of drag it is no longer possible to find a kinematically consistent front of the fluid motion. However the inclusion of drag forces within the region close to the front resolves this difficulty. We calculate velocity and depth profiles within the drag-affected region, and obtain the leading-order expression for the rate at which the fluid propagates when the magnitude of the drag force is modelled using Chézy, Newtonian and power-law fluid closures; this compares well with experimental data and provide new insights into dam-break flows.

Electrohydrodynamic (EHD) flow around a charged spherical colloid near an electrode was studied theoretically and experimentally to understand the nature of long-range particle–particle attraction near electrodes. Numerical computations for finite double-layer thicknesses confirmed the validity of an asymptotic methodology for thin layers. Then the electric potential around the particle was computed analytically in the limit of zero Péclet number and thin double layers for oscillatory electric fields at frequencies where Faradaic reactions are negligible. Streamfunctions for the steady component of the EHD flow were determined with an electro-osmotic slip boundary condition on the electrode surface. Accordingly, it was established how the axisymmetric flow along the electrode is related to the dipole coefficient of the colloidal particle. Under certain conditions, the flow is directed toward the particle and decays as r−4, in accord with observations of long-range particle aggregation. To test the theory, particle-tracking experiments were performed with fluorescent 300 nm particles around 50μm particles over a wide range of electric field strengths and frequencies. Treating the particle surface conductivity as a fitting parameter yields velocities in excellent agreement with the theoretical predictions. The observed frequency dependence, however, differs from the model predictions, suggesting that the effect of convection on the charge distribution is not negligible as assumed in the zero Péclet number limit.

The frequencies of vortex shedding from circular cylinders forced to oscillate transversely in low-turbulence uniform and shear flows were investigated. The stream velocity in the shear flow varied linearly with spanwise distance.

In both flows the vortex shedding frequency locked on to the cylinder frequency and to submultiples of the cylinder frequency. In uniform flow the range of cylinder frequencies for locking-on was dependent on the amplitude of oscillation and Reynolds number. At the boundaries of locking-on at the cylinder frequency locked-on shedding was intermittent with unforced shedding and locking-on was accompanied by a change in wake width. At a particular cylinder frequency near mid-range it is conjectured that the wake width jumped from being greater to being less than that for the stationary cylinder. In shear flow the spanwise extent of locking-on at the cylinder frequency was explained by considering the uniform flow results and the inclination of shed vortices in shear flow. At the spanwise boundaries of this locking-on, locked-on cells were shed intermittently with unforced cells which were more stable in frequency than the corresponding cells for the stationary cylinder.

The sound generated by vortex pairing in axisymmetric jets is determined by direct solution of the compressible Navier–Stokes equations on a computational grid that includes both the near field and a portion of the acoustic far field. At low Mach number, the far-field sound has distinct angles of extinction in the range of 60°–70° from the jet's downstream axis which can be understood by analogy to axisymmetric, compact quadrupoles. As the Mach number is increased, the far-field sound takes on a superdirective character with the dominant sound directed at shallow angles to the jet's downstream axis. The directly computed sound is compared to predictions obtained from Lighthill's equation and the Kirchhoff surface method. These predictions are in good agreement with the directly computed data. The Lighthill source terms have a large spatial distribution in the axial direction necessitating the introduction of a model to describe the source terms in the region downstream of the last vortex pairing. The axial non-compactness of the quadrupole sources must be adequately treated in the prediction method.

The Orr-Sommerfeld stability problem has been studied for velocity profiles appropriate to turbulent channel flow. The intent was to provide an evaluation of Malkus's theory that the flow assumes a state of maximum dissipation, subject to certain constraints, one of which is that the mean velocity profile is marginally stable. Dissipation rates and neutral stability curves were obtained for a representative two-parameter family of velocity profiles. Those in agreement with experimental profiles were found to be stable; the marginally stable profile of greatest dissipation was not in good agreement with experiments. An explanation for the apparent success of Malkus's theory is offered.

A potential flow model of two-dimensional vortex shedding behind an inclined plate is developed. The free shear layers which emanate from the sides of the plate are represented by discrete vortices through the use of the appropriate complex-velocity potential, the Kutta condition and the Joukowsky transformation between a circle and the plate cross-section. The analysis is then applied to predict the kinematic and dynamic characteristics of the flow for various angles of attack. The results compare favourably with the available experimental data as far as the form of vortex shedding and the Strouhal number are concerned. The calculated normal-force coefficients are 20−25% yo larger than those measured by Fage & Johansen (1927).

Experiments were conducted to study the effects of heat release in a planar, gas-phase, reacting mixing layer formed between two free streams, one containing hydrogen in an inert diluent, the other, fluorine in an inert diluent. Sufficiently high concentrations of reactants were utilized to produce adiabatic flame temperature rises of up to 940 K (corresponding to 1240 K absolute). The temperature field was measured at eight fixed points across the layer. Flow visualization was accomplished by schlieren spark and motion picture photography. Mean velocity information was extracted from Pitot-probe dynamic pressure measurements. The results showed that the growth rate of the layer, for conditions of zero streamwise pressure gradient, decreased slightly with increasing heat release. The overall entrainment into the layer was substantially reduced as a consequence of heat release. A posteriori calculations suggest that the decrease in layer growth rate is consistent with a corresponding reduction in turbulent shear stress. Large-scale coherent structures were observed at all levels of heat release in this investigation. The mean structure spacing decreased with increasing temperature. This decrease was more than the corresponding decrease in shear-layer growth rate, and suggests that the mechanisms of vortex amalgamation are, in some manner, inhibited by heat release. The mean temperature rise profiles, normalized by the adiabatic flame temperature rise, were not greatly changed in shape over the range of heat release of this investigation. A small decrease in normalized mean temperature rise with heat release was however observed. Imposition of a favourable pressure gradient in a mixing layer with heat release resulted in an additional decrease in layer growth rate, and caused only a very slight increase in the mixing and amount of chemical product formation. The additional decrease in layer growth rate is shown to be accounted for in terms of the change in free-stream velocity ratio induced by the pressure gradient.

This investigation focuses on the characteristics of near-bed turbulence in fully rough gravel-bed open-channel flows. The analysis combines results obtained with the double-averaging methodology and local flow characterization, using velocity measurements provided by a high-resolution three-axis Acoustic Doppler Velocity Profiler (ADVP). As a result of the flow heterogeneity induced by the bed topography, the flow is not locally uniform in the near-bed region, and a double-averaging methodology is applied over a length scale much greater than the gravel size. In smooth- and rough-bed flow conditions, without macro-roughness bed elements, maximum turbulent kinetic energy (TKE) production occurs very close to z = 0, while in our case with fully rough flows with macro-roughness elements, maximum turbulence activity is found to occur at gravel crest levels zc (zc/h = 0.1). Turbulent diffusion also reaches a maximum at this elevation. The characteristics of the spatially averaged TKE budget are in good agreement with those obtained in flows over canopies. The hydrodynamic double-averaged properties have strong similarities with mixing layers and reattached mixing layers in flows over backward facing steps. Local time-averaged velocity profiles can be split into three typical classes, namely log, S-shaped and accelerated. It appears that the S-shaped class profiles, located in the wakes of the macro-roughness elements, exhibit an inflectional profile typical of mixing layers. They are of major importance in the double-averaged TKE budget, as they provide a local high contribution to the double-averaged TKE flux, TKE production and dissipation compared to the log class profiles. Consequently, double-averaged TKE production is roughly 75% greater than the dissipation rate at the point of maximal TKE production. Moreover the macro-roughness bed elements imply mixing-layer-type hydrodynamics that play a dominant role in the overall structure of mean near-bed turbulence of gravel-bed channel flows.

A generalized Reynolds analogy (GRA) is proposed for compressible wall-bounded turbulent flows (CWTFs) and validated by direct numerical simulations. By introducing a general recovery factor, a similarity between the Reynolds-averaged momentum and energy equations is established for the canonical CWTFs (i.e. pipes, channels, and flat-plate boundary layers that meet the quasi-one-dimensional flow approximation), independent of Prandtl number, wall temperature, Mach number, Reynolds number, and pressure gradient. This similarity and the relationships between temperature and velocity fields constitute the GRA. The GRA relationship between the mean temperature and the mean velocity takes the same quadratic form as Walz’s equation, with the adiabatic recovery factor replaced by the general recovery factor, and extends the validity of the latter to diabatic compressible turbulent boundary layers and channel/pipe flows. It also derives Duan & Martín’s (J. Fluid Mech., vol. 684, 2011, pp. 25–59) empirical relation for flows at different physical conditions (wall temperature, Mach number, enthalpy condition, surface catalysis, etc.). Several key parameters besides the general recovery factor emerge in the GRA. An effective turbulent Prandtl number is shown to be the reason for the parabolic profile of mean temperature versus mean velocity, and it approximates unity in the fully turbulent region. A dimensionless wall temperature, that we call the diabatic parameter, characterizes the wall-temperature effects in diabatic flows. The GRA also extends the analysis to the fluctuation fields. It recovers the modified strong Reynolds analogy proposed by Huang, Coleman & Bradshaw (J. Fluid Mech., vol. 305, 1995, pp. 185–218) and explains the variation of the temperature–velocity correlation coefficient with wall temperature. Thus, the GRA unveils a generalized similarity principle behind the complex nonlinear coupling between the thermal and velocity fields of CWTFs.

This paper presents the results of a detailed experimental investigation into the effectiveness of sinusoidal leading edge serrations on aerofoils for the reduction of the noise generated by the interaction with turbulent flow. A detailed parametric study is performed to investigate the sensitivity of the noise reductions to the serration amplitude and wavelength. The study is primarily performed on flat plates in an idealized turbulent flow, which we demonstrate captures the same behaviour as when identical serrations are introduced onto three-dimensional aerofoils. The influence on the noise reduction of the turbulence integral length scale is also studied. An optimum serration wavelength is identified whereby maximum noise reductions are obtained, corresponding to when the transverse integral length scale is approximately one-fourth the serration wavelength. This paper proves that, at the optimum serration wavelength, adjacent valley sources are excited incoherently. One of the most important findings of this paper is that, at the optimum serration wavelength, the sound power radiation from the serrated aerofoil varies inversely proportional to the Strouhal number $St_{h}=fh/U$, where $f$, $h$ and $U$ are frequency, serration amplitude and flow speed, respectively. A simple model is proposed to explain this behaviour. Noise reductions are observed to generally increase with increasing frequency until the frequency at which aerofoil self-noise dominates the interaction noise. Leading edge serrations are also shown to reduce aerofoil self-noise. The mechanism for this phenomenon is explored through particle image velocimetry measurements. Finally, the lift and drag of the serrated aerofoil are obtained through direct measurement and compared against the straight edge baseline aerofoil. It is shown that aerodynamic performance is not substantially degraded by the introduction of the leading edge serrations on the aerofoil.

A linear solution modelling the shock-cell structure of an axisymmetric supersonic jet operated at slightly off-design conditions is developed by the method of multiple scales. The model solution takes into account the gradual spatial change of the mean flow in the downstream direction. Turbulence in the mixing layer of the jet has the tendency to smooth out the sharp velocity and density gradients induced by the shocks. To simulate this effect, eddy-viscosity terms are incorporated in the model. Extensive comparisons between the numerical results of the present model and experimental measurements gathered at the NASA Langley Research Center over the Mach number range of |Mj2 − Md2| [les ] 1.0 for underexpanded and overexpanded supersonic jets are carried out. Here Mj is the fully expanded jet Mach number and Md is the design Mach number of the convergent–divergent nozzle. Very favourable agreement is found. This is especially true for the gross features of the shock cells, including the shock-cell spacings and the pressure amplitudes associated with the shocks. The measured data show that the pressure distributions over the first three or four shock cells usually are rich in fine structures. These fine structures are reproduced by the calculated results. Beyond the first few shock cells the model predicts that the shock-cell structure can be represented by a single Fourier mode of the mean flow. This is confirmed by a careful examination of the experimental data. The appropriate turbulent Reynolds number for shock-cell structure calculation is investigated. It is shown that the best choice is the same as the value found to give the best results for jet mean-flow calculation. The present model is used to explain some of the observed characteristics of broadband shock-associated noise.

Attached cavitation on a wall with continuous curvature is investigated on the basis of experiments carried out on various bodies (circular and elliptic cylinders, NACA 16 012 foil). Visualization of the boundary layer by dye injection at the leading edge shows that a strong interaction exists between attached cavitation and the boundary layer. In particular, it is shown that the cavity does not detach from the body at the minimum pressure point, but behind a laminar separation, even in largely developed cavitating flow. A detachment criterion which takes into account this link between attached cavitation and boundary layer is proposed. It consists of connecting a cavitating potential-flow calculation and a boundary-layer calculation. Among all the theoretically possible detachment points, the actual detachment point is chosen to be the one for which the complete calculation predicts a laminar separation just upstream. This criterion, applied to the NACA foil, leads to a prediction which is in good agreement with experimental results.

Entrainment of ambient fluid into both two-dimensional and axisymmetric gravity currents is investigated experimentally using a novel neutralization technique. The technique involves the titrative neutralization of an alkaline gravity current which intrudes into and entrains an acidic ambient, and is visualized using a pH indicator solution. Using this technique, we can determine quantitative results for the amount of dilution in the head of the current. The head of the current is able to entrain ambient fluid both by shear instabilities on the current/ambient interface and by over-riding (relatively light) ambient fluid. Guided by our experimental observations, we present two slightly different theoretical models to determine the entrainment into the head of the current as a function of distance from the source for the instantaneous release of a constant volume of fluid in a two-dimensional geometry. By dimensional analysis, we determine from both models that the dimensionless entrainment or dilution ratio, E, defined as the ratio of the volumes of ambient and original fluid in the head, is independent of the initial reduced gravity of the current; and this result is confirmed by our experiments in Boussinesq situations. Our theoretical evaluation of E in terms of the initial cross-sectional area of the current agrees very well with our experimental measurements on the incorporation of an entrainment coefficient α, evaluated experimentally to be 0.063 ± 0.003. We also obtain experimental results for constant-volume gravity currents moving over horizontal surfaces of varying roughness. A particularly surprising result from all the experiments, which is reflected in the theoretical models, is that the head remains essentially unmixed – the entrainment is negligible – in the slumping phase. Thus the heads of gravity currents with identical initial cross-sectional areas but different initial aspect ratios (lock lengths) will begin to be diluted by ambient fluid at different positions and hence propagate at different rates. A range of similar results is determined, both theoretically and experimentally, for the instantaneous release of a fixed volume of (heavy) fluid in an axisymmetric geometry. By contrast, the results of our experiments with gravity currents fed by a constant flux exhibit markedly different entrainment dynamics due to the continual replenishment of the fluid in the head by the constant input of undiluted fluid from the tail.

Schwinger's variational formulation is applied to the radiation of surface waves due to small oscillation of bodies. By means of Haskind's theorem the wave forces on a stationary body due to a plane incident wave are found using only far-field properties. Results for horizontal rectangular and vertical circular cylinders are presented.

In this paper, we study the transverse vortex-induced vibrations of a cylinder with no structural restoring force (k = 0). In terms of the conventionally used normalized flow velocity, U*, the present experiments correspond to an infinite value (where U* = U/fND, fN = natural frequency, D = diameter). A reduction of mass ratios m* (mass/displaced mass) from the classically studied values of order m* = 100, down to m* = 1, yields negligible oscillations. However, a further reduction in mass exhibits a surprising result: large-amplitude vigorous vibrations suddenly appear for values of mass less than a critical mass ratio, m*crit = 0.54. The classical assumption, since the work of den Hartog (1934), has been that resonant large-amplitude oscillations exist only over a narrow range of velocities, around U*∼5, where the vortex shedding frequency is comparable with the natural frequency. However, in the present study, we demonstrate that, so long as the body’s mass is below this critical value, the regime of normalized velocities (U*) for resonant oscillations is infinitely wide, beginning at around U*∼5 and extending to U*→∞. This result is in precise accordance with the predictions put forward by Govardhan & Williamson (2000), based on elastically mounted vibration studies (where k > 0). We deduce a condition under which this unusual concept of an infinitely wide regime of resonance will occur in any generic vortex-induced vibration system.

This paper describes part of a detailed study of the initial region of coaxial jets of three different mean-velocity ratios. Similarity of the mean-velocity and turbulent-intensity profiles within the two mixing regions inside the initial merging zone, and within the mixing region inside the fully-merged zone, has been observed. Similarity with single-jet results has been obtained. Overall pressure measurements and hot-wire and microphone spectra inside coaxial jets yield two pronounced peaks, suggesting the existence of two types of vortices at different frequencies. The higher-frequency or primary vortices are found to be generated in the primary or inner mixing region, and the lower-frequency or secondary ones in the secondary or outer mixing region. It seems that the former are generated further upstream than the latter. Both types of vortices are found in the initial merging zone. However, their growth or decay and their dominance depend on the mean-velocity ratio. Low mean-velocity ratios indicate the dominance of the high-frequency vortices. At high mean-velocity ratios, the low-frequency vortices are dominant. The Strouhal number of the vortices plays an important part in their growth and decay. The complicated flow structure of coaxial jets can be very simply related to, and described by, the much simpler structure of single jets.