The high-enthalpy, hypersonic flow over a compression corner has been examined experimentally and theoretically. Surface static pressure and heat transfer distributions, along with some flow visualization data, were obtained in a free-piston shock tunnel operating at enthalpies ranging from 3 MJ kg−1 to 19 MJ kg−1, with the Mach number varying from 7.5 to 9.0 and the Reynolds number based on upstream fetch from 2.7×104 to 2.7×105. The flow was laminar throughout. The experimental data compared well with theories valid for perfect gas flow and with other relevant low-to-moderate enthalpy data, suggesting that for the current experimental conditions, the real gas effects on shock wave/boundary layer interaction are negligible. The flat-plate similarity theory has been extended to include equilibrium real gas effects. While this theory is not applicable to the current experimental conditions, it has been employed here to determine the potential maximum effect of real gas behaviour. For the flat plate, only small differences between perfect gas and equilibrium gas flows are predicted, consistent with experimental observations. For the compression corner, a more rapid rise to the maximum pressure and heat transfer on the ramp face is predicted in the real gas flows, with the pressure lying slightly below, and the heat transfer slightly above, the perfect gas prediction. The increase in peak heat transfer is attributed to the reduction in boundary layer displacement thickness due to real gas effects.

The velocity field of a binary mixture of incompressible miscible liquids is non-solenoidal when the densities of the two liquids are different. If the mixture density is linear in the volume fraction, as in the case of simple (ideal) mixtures or very nearly for glycerin and water, then the velocity can be decomposed into a solenoidal and an expansion part. In the context of this theory, we derive a new solution which describes the smoothing of an initial plane discontinuity in concentration across a channel bounded by sidewalls. The requirement that the velocity vanishes on the sidewall introduces a different initial discontinuity not present in the solenoidal theory. The problem may be reduced to a partial differential equation in two similarity variables, one for the smoothing of a concentration discontinuity without sidewalls and the other for the smoothing the velocity discontinuity at the sidewall. The similarity equations are solved explicitly in a special case.

A two-dimensional temporal mixing layer is generated in a stratified tilting tank similar to that used by Thorpe (1968). Extensive flow dynamics visualization is carried out using, for the top and bottom layers, fluids of different densities but of the same index of refraction. The two-dimensional density field is measured with the laser-induced fluorescence technique (LIF). The study examines further the classical problem of the two-dimensional mixing layer and explores the effects of cross-shear on a nominally two-dimensional mixing layer, a situation widespread in complex industrial and natural flows. Cross-shear is another component of shear, in plane with but perpendicular to the main shear of the base flow, generated by tilting the tank around a second axis.

A good approximation to modelling the shape of a liquid–air meniscus advancing or receding in a capillary tube of radius a can be constructed by balancing the curvature of the interface with the sum of a viscous stress valid near the contact line and a constant static pressure. This model has unique solutions for each value of the boundary condition, i.e. the dynamic contact angle. When the meniscus recedes at very small capillary numbers, the model predicts a critical receding velocity beyond which a liquid layer of the receding fluid (a liquid tail) develops along the solid (see figure 4). The length of the layer increases as the receding speed and the contact angle decrease. This layer regime is characterized by menisci whose macroscopic curvature is >1/a.

We consider Stokes flow in a cylindrical container of circular section induced by the uniform translatory motion of one of the endwalls. This flow field is of interest because it is possible to get reliable analytical descriptions of important three-dimensional structures such as the primary and corner eddies. It is shown, using a result of Tran-Cong & Blake, that separable solutions exist which can be combined to yield vector eigenfunctions that satisfy the sidewall boundary conditions provided the eigenvalues satisfy the transcendental equation

formula here

A turbulent separation bubble is forced by single- and double-frequency sinusoidal disturbances, with the emphasis placed on the reattachment length as a function of the forcing amplitude and frequency. The separation bubble is that formed along the side of a blunt circular cylinder with a square leading edge. In single-frequency forcing, the reattachment length attains a minimum at a particular forcing frequency, F, which scales with the frequency of shedding of vortices from the reattachment region of the separated shear layer. A flow model is presented to interpret the frequency F. Forcing of sufficiently high amplitude eliminates the recirculating region in a range of the forcing frequency. Flow visualization and a survey of the mean flow and turbulence properties demonstrate how the flow in the separated shear layer is modified by the forcing. In double-frequency forcing, the superposition of the F-component on its higher or subharmonic components is considered. A non-resonant combination of the two frequencies is also considered.

The collision of a solitary wave, travelling over a horizontal bed, with a vertical wall is investigated using a boundary-integral method to compute the potential fluid flow described by the Euler equations. We concentrate on reporting new results for that part of the motion when the wave is near the wall. The wall residence time, i.e. the time the wave crest remains attached to the wall, is introduced. It is shown that the wall residence time provides an unambiguous characterization of the phase shift incurred during reflection for waves of both small and large amplitude. Numerically computed attachment and detachment times and amplitudes are compared with asymptotic formulae developed using the perturbation results of Su & Mirie (1980). Other features of the flow, including the maximum run-up and the instantaneous wall force, are also presented. The numerically determined residence times are in good agreement with measurements taken from a cine film of solitary wave reflection experiments conducted by Maxworthy (1976).

A numerical bifurcation study in plane Couette flow is performed by computing successive finite-amplitude equilibrium states, solutions of the Navier–Stokes equations. Plane Couette flow being linearly stable for all Reynolds numbers, first two-dimensional equilibrium states are computed by extending nonlinear travelling waves in plane Poiseuille flow through the Poiseuille–Couette flow family to the plane Couette flow limit. The resulting nonlinear states are stationary with a spatially localized structure; they are subject to two-dimensional and three-dimensional secondary disturbances. Three-dimensional disturbances dominate the dynamics and three-dimensional stationary equilibrium states bifurcating at criticality on the two-dimensional equilibrium surface are computed. These nonlinear states, periodic in the spanwise direction and spatially localized in the streamwise direction, are computed for Reynolds numbers (based on half the velocity difference between the walls and channel half-width) close to 1000. While a possible relationship between the computed solutions and experimentally observed coherent structures in turbulent plane Couette flow has to be assessed, the present findings reinforce the idea that subcritical transition may be related to the existence of finite-amplitude states which are (unstable) fixed points in a dynamical systems formulation of the Navier–Stokes system.

The equations of motion for ‘rapid’ flow of a granular material have fully developed solutions representing flow driven by a body force, such as gravity, along a channel bounded by plane parallel walls. The stability of these solutions to small perturbations is investigated. For given properties of the particles and the channel walls it is found that the condition of critical stability is a relation between the mean concentration of the particles and the width of the channel. When the base state is unstable the fastest growing modes are travelling waves propagating in the axial direction and these induce characteristic patterns of variations in particle concentration, as well as velocity. These instabilities are contrasted with those found for Couette shearing in an earlier publication.

The model equation

formula here

arises as the equation for solitary-wave solutions to a fifth-order long-wave equation for gravity–capillary water waves. Being Hamiltonian, reversible and depending upon two parameters, it shares the structure of the full steady water-wave problem. Moreover, all known analytical results for local bifurcations of solitary-wave solutions to the full water-wave problem have precise counterparts for the model equation.

At the time of writing two major open problems for steady water waves are attracting particular attention. The first concerns the possible existence of solitary waves of elevation as local bifurcation phenomena in a particular parameter regime; the second, larger, issue is the determination of the global bifurcation picture for solitary waves. Given that the above equation is a good model for solitary waves of depression, it seems natural to study the above issues for this equation; they are comprehensively treated in this article.

The equation is found to have branches of solitary waves of elevation bifurcating from the trivial solution in the appropriate parameter regime, one of which is described by an explicit solution. Numerical and analytical investigations reveal a rich global bifurcation picture including multi-modal solitary waves of elevation and depression together with interactions between the two types of wave. There are also new orbit-flip bifurcations and associated multi-crested solitary waves with non-oscillatory tails.

Direct numerical simulations (DNS) are performed to investigate the evolution of turbulence in a uniformly sheared and stably stratified flow. The spatial discretization is accomplished by a spectral collocation method, and the solution is advanced in time with a third-order Runge–Kutta scheme. The turbulence evolution is found to depend strongly on at least three parameters: the gradient Richardson number Ri, the initial value of the Taylor microscale Reynolds number Reλ, and the initial value of the shear number SK/<ε. The effect of each parameter is individually studied while the remaining parameters are kept constant. The evolution of the turbulent kinetic energy K is found to follow approximately an exponential law. The shear number SK/<ε, whose effect has not been investigated in previous studies, was found to have a strong non-monotone influence on the turbulence evolution. Larger values of the shear number do not necessarily lead to a larger value of the eventual growth rate of the turbulent kinetic energy. Variation of the Reynolds number Reλ indicated that the turbulence growth rate tends to become insensitive to Reλ at the higher end of the Reλ range studied here. The dependence of the critical Richardson number Ricr, which separates asymptotic growth of the turbulent kinetic energy K from asymptotic decay, on the initial values of the Reynolds number Reλ and the shear number SK/<ε was also obtained. It was found that the critical Richardson number varied over the range 0.04<Ricr<0.17 in our DNS due to its strong dependence on Reynolds and shear numbers.

Measurements of the spanwise and wall-normal components of vorticity and their constituent velocity derivative fluctuations have been made in a turbulent boundary layer over a mesh-screen rough wall using a four-hot-wire vorticity probe. The measured spectra and variances of vorticity and velocity derivatives have been corrected for the effect of spatial resolution. The high-wavenumber behaviour of the spectra conforms closely with isotropy. Over most of the outer layer, the normalized magnitudes of the velocity derivative variances differ significantly from those over a smooth wall layer. The differences are such that the variances are much more nearly isotropic over the rough wall than on the smooth wall. This behaviour is consistent with earlier observations that the large-scale structure in this rough wall layer is more isotropic than that in a smooth wall layer. Isotropy-based approximations for the mean energy dissipation rate and mean enstrophy are consequently more reliable in this rough wall layer than in a smooth wall layer. In the outer layer, the vorticity variances are slightly larger than those over a smooth wall; reflecting structural differences between the two flows.

Free convection in liquid metals is of significant practical interest to the crystal-growing community since the adverse effects of convective instabilities in the melt phase can be frozen into the solid product. Here, we present the results of a combined numerical and experimental study of steady convective flows in a sample of liquid gallium which is heated at one end and cooled at the other. Experimental measurements of temperature distributions in the flow are compared with the standard Hadley-cell solution and with the numerical results obtained from a two-dimensional model. Excellent quantitative agreement is found between all three for low Grashof numbers but a systematic divergence between the results is seen as this parameter is increased.

Extinction limits and the lean flammability limit of non-adiabatic stretched premixed methane–air flames are investigated numerically with detailed chemistry and two different Planck mean absorption coefficient models. Attention is paid to the combined effect of radiative heat loss and stretch at low stretch rate. It is found that for a mixture at an equivalence ratio lower than the standard lean flammability limit, a moderate stretch can strengthen the combustion and allow burning. The flame is extinguished at a high stretch rate due to stretch and is quenched at a low stretch rate due to radiation loss. A O-shaped curve of flame temperature versus stretch rate with two distinct extinction limits, a radiation extinction limit and a stretch extinction limit respectively on the left- and right-hand sides, is obtained. A C-shaped curve showing the flammability limit of the stretched methane–air flame is obtained by plotting these two extinction limits in the mixture strength coordinate. A good agreement is shown on comparing the predicted results with the experimental data. For equivalence ratio larger than a critical value, it is found that the O-shaped temperature curve opens up in the middle of the stable branch, so that the stable branch divides into two stable flame branches; a weak flame branch and a normal flame branch. The weak flame can survive between the radiation extinction limit and the opening point (jump limit) while the normal flame branch can survive from its stretch extinction limit to zero stretch rate. Finally, a G-shaped curve showing both extinction limits and jump limits of stretched methane–air flames is presented. It is found that the critical equivalence ratio for opening up corresponds to the standard flammability limit measured in microgravity. Furthermore, the results show that the flammability limit (inferior limit) of the stretched methane–air flame is lower than the standard flammability limit because flames are strengthened by a moderate stretch at Lewis number less than unity.

The decay of high-Reynolds-number inhomogeneous turbulence in an unbounded domain is considered. The turbulence may be initially localized in one to three spatial directions and the fluid is assumed to be at rest at infinity in those directions. Previous arguments used to determine the decay laws of homogeneous turbulence are extended to the decay of inhomogeneous turbulence by integrating the turbulence statistics over the inhomogeneous directions. Dimensional arguments based on the invariance or near-invariance of low-wavenumber spectral coefficients associated with the integrated mean-square velocity are used to determine asymptotic decay laws for inhomogeneous turbulence. These decay laws depend on the number of inhomogeneous directions of the flow field and reduce to the well-known decay laws of homogeneous turbulence when this number is zero. Different decay laws are determined depending on the spectral behaviour at low wavenumbers. Asymptotic similarity states of the spectrum during the decay and of the distribution of the mean-square velocity along the inhomogeneous directions are also determined. An analytical result for the decay of the mean-square velocity at the centre of the initial disturbance is found, and the decay proceeds more rapidly with increasing number of inhomogeneous directions due to the transport of energy along those directions.

Large-eddy simulations of decaying turbulence homogeneous in a plane and localized in a single direction are performed to test the theoretical scaling laws. The numerically determined asymptotic decay laws of the integrated mean-square velocity agree well with the theoretical predictions. A self-similar decay of the spectra and mean-square velocity distributions is also observed. The simulation results suggest that when the low-wavenumber spectral coefficient is an exact invariant, a unique similarity state depending only on the initial value of this invariant and independent of all other aspects of the initial conditions is attained asymptotically.

The coupling of the temperature and velocity fields by buoyancy in a laminar two-dimensional wall jet over a finite-length horizontal plate is studied numerically and analytically in the asymptotic limit of infinite Reynolds number. Two configurations are considered leading to a cold layer of fluid over the plate, namely an ambient-temperature jet over a cooled plate and a cold jet over an insulated plate. In both cases buoyancy generates an adverse pressure gradient that may separate the flow if the Froude number is sufficiently small and always makes the solution everywhere over the plate dependent on the conditions at the downstream boundary. In the limit of very small Froude number separation occurs in a viscous–inviscid interaction region near the origin of the jet, leading to a separation bubble that covers a fraction of the plate dependent on the Prandtl number. The scalings of the solution in this asymptotic limit are obtained by order of magnitude estimations in the different regions of the bubble and in the buoyancy-dominated flow beyond the bubble, and the results are checked against the numerical solutions of the boundary layer equations. A separate analysis is carried out for very large Prandtl numbers showing that the recirculation bubble is then much shorter than the plate, also in agreement with the numerical results.

When scaled properly, the high-wavenumber and high-frequency parts of wind-wave spectra collapse onto universal curves. This collapse has been attributed to a dynamical balance and so these parts of the spectra have been called the equilibrium range. We develop a model for this equilibrium range based on kinematical and dynamical properties of breaking waves. Data suggest that breaking waves have high curvature at their crests, and they are modelled here as waves with discontinuous slope at their crests. Spectra are then dominated by these singularities in slope. The equilibrium range is assumed to be scale invariant, meaning that there is no privileged lengthscale. This assumption implies that: (i) the sharp-crested breaking waves have self-similar shapes, so that large breaking waves are magnified copies of the smaller breaking waves; and (ii) statistical properties of breaking waves, such as the average total length of breaking-wave fronts of a given scale, vary with the scale of the breaking waves as a power law, parameterized here with exponent D.

An axisymmetric flow due to a submerged sink in water of infinite depth is considered, with a stagnation point on the free surface above the sink. Forbes & Hocking (1990) calculated numerically a flow for each value of the Froude number F smaller than a critical value Fc. For F close to Fc there is a ring-shaped bump on the free surface. At F=Fc, the crest of the bump becomes a ring of stagnation points. We use the numerical procedure of Hocking & Forbes to show that the bump is the first crest of a train of axisymmetric waves. The wave amplitude decreases with increasing distance from the source. Then we give a local analysis of axisymmetric free-surface flows with a circular ring of stagnation points. We find flows in which the surface has a discontinuity in slope with an enclosed angle of 120° all along the ring. This behaviour is consistent with the numerical solution for F=Fc near the crest of the bump.