The nonlinear development of the most unstable Görtler vortex mode in boundary-layer flows over curved walls is investigated. The most unstable Görtler mode is confined to a viscous wall layer of thickness O(G−1/5) and has spanwise wavelength O(G−1/5); it is, of course, most relevant to flow situations where the Görtler number G [Gt ] 1. The nonlinear equations governing the evolution of this mode over an O(G−3/5) streamwise lengthscale are derived and are found to be of a fully non-parallel nature. The solution of these equations is achieved by making use of the numerical scheme used by Hall (1988) for the numerical solution of the nonlinear Görtler equations valid for O(1) Görtler numbers. Thus, the spanwise dependence of the flow is described by a Fourier expansion whereas the streamwise and normal variations of the flow are dealt with by employing a suitable finite-difference discretization of the governing equations. Our calculations demonstrate that, given a suitable initial disturbance, after a brief interval of decay, the energy in all the higher harmonics grows until a singularity is encountered at some downstream position. The structure of the flow field as this singularity is approached suggests that the singularity is responsible for the vortices, which are initially confined to the thin viscous wall layer, moving away from the wall and into the core of the boundary layer.

Many processes in the Earth, such as magma migration, can be described by the flow of a low-viscosity fluid in a viscously deformable, permeable matrix. The purpose of this and a companion paper is to develop a better physical understanding of the equations governing these two-phase flows. This paper presents a series of analytic approximate solutions to the governing equations to show that the equations describe two different modes of matrix deformation. Shear deformation of the matrix is governed by Stokes equation and can lead to porosity-driven convection. Volume changes of the matrix are governed by a nonlinear dispersive wave equation for porosity. Porosity waves exist because the fluid flux is an increasing function of porosity and the matrix can expand or compact in response to variations in the fluid flux. The speed and behaviour of the waves depend on the functional relationship between permeability and porosity. If the partial derivative of the permeability with respect to porosity, ∂kϕ/∂ϕ, is also an increasing function of porosity, then the waves travel faster than the fluid in the pores and can steepen into porosity shocks. The propagation of porosity waves, however, is resisted by the viscous resistance of the matrix to volume changes. Linear analysis shows that viscous stresses cause plane waves to disperse and provide additional pressure gradients that deflect the flow of fluid around obstacles. When viscous resistance is neglected in the nonlinear equations, porosity shock waves form from obstructions in the fluid flux. Using the method of characteristics, we quantify the specific criteria for shocks to develop in one and two dimensions. A companion paper uses numerical schemes to show that in the full equations, viscous resistance to volume changes causes simple shocks to disperse into trains of nonlinear solitary waves.

Using numerical schemes, this paper demonstrates how viscous resistance to volume changes modifies the simplest shock wave solutions presented in Part 1. For an initial condition chosen to form a step-function shock, viscous resistance causes the shock to disperse into a rank-ordered wavetrain of solitary waves. Large obstructions in flux produce large-amplitude, slow-moving wavetrains while smaller shocks shed small-amplitude waves. While the viscous resistance term is initially important over a narrow boundary layer, information about obstructions in the flux can propagate over many compaction lengths through the formation of non-zero wavelength porosity waves. For large-amplitude shocks, information can actually propagate backwards relative to the matrix. The physics of dispersion is discussed and a physical argument is presented to parameterize the amplitude of the wavetrain as a function of the amplitude of the predicted shock. This quantitative relationship between the prediction of shocks and the development of solitary waves also holds when mass transfer between solid and liquid is included. Melting causes solitary waves to decrease in amplitude but the process is reversible and freezing can cause small perturbations in the fluid flux to amplify into large-amplitude waves. These model problems show that the equations governing volume changes of the matrix are inherently time dependent. Perturbations to steady-state solutions propagate as nonlinear waves and these problems demonstrate several initial conditions that do not relax to steady state. If these equations describe processes such as magma migration in the Earth, then these processes should be inherently episodic in space and time.

A comparison is made between a number of properties of a quasi-homogeneous isotropic turbulent field obtained from a direct numerical simulation of the Navier–Stokes equation and its random counterpart with the same energy spectrum. It is demonstrated that some effects in a real flow have a considerable contribution of a kinematic nature (e.g. reduction of nonlinearity), while others are mostly dynamical (e.g. alignment between vorticity and eigenvectors of the rate of strain).

Controlled, amplitude-modulated excitation of a cylinder at low Reynolds number (Re equals; 136) in the cross-stream direction generates several states of response of the near wake including: a locked-in wake structure, which is periodic at the modulation frequency; a period-doubled wake structure, which is periodic at a frequency half the modulation frequency; and a destabilized structure of the wake, which is periodic at the modulation frequency, but involves substantial phase modulations of the vortex formation relative to the cylinder displacement. The occurrence of each of these states depends upon the dimensionless modulation frequency, as well as the nominal frequency and amplitude of excitation. Transition through states of increasing disorder can be attained by either decreasing the modulation frequency or increasing the amplitude of excitation at a constant value of nominal frequency. These states of response in the near wake are crucial in determining whether the far wake is highly organized or incoherent. Both of these extremes are attainable by proper selection of the parameters of excitation.

The behaviour of a granular flow at a boundary cannot be specified independently of what is happening in the rest of the flow field. This paper describes a study of two fictitious, but instructive, flat boundary types using a computer simulation of a two-dimensional granular flow with the goal of trying to understand the possible effects of the boundary on the flow. The two boundary conditions, Type A and Type B, differ largely in the way that they apply torques to the flow particles. During a particle–wall collision, the Type A boundary applies the force at the particle surface, thus applying the largest mechanistically possible torque to the particle, while the Type B boundary applies the force directly to the particle centre, resulting in the application of zero torque. Though a small change on continuum scales (i.e. the point at which the force is applied has only been moved by a particle radius) it makes a huge difference to the macroscopic behaviour of the system. Generally, it was found that, near boundaries, large variations in continuum properties occur over distances of a particle diameter, a non-continuum scale, throwing into doubt whether boundaries may be accurately modelled via continuum mechanics. Finally, the large torques applied to the particles by the Type A boundary induce asymmetries in the stress tensor, which, in these steady flows, are balanced by gradients in a couple stress tensor. Thus, near boundaries, a frictional granular material must be modelled as a polar fluid.

The global behaviour of a granular flow is critically dependent on its interaction with whatever solid boundaries with which it comes into contact, whether they be used to drive, retard or simply bound the flow field. This paper describes the results of a computer simulation study of the effects of roughening boundaries by ‘gluing’ particles to the surfaces. Roughness is commonly used in experimental devices as a way of approximating a no-slip condition between a granular material and the driving surfaces. On a microscopic level, this produces a boundary that extends out into the flow field to the limit of the roughness elements. This has a strong effect on the way that forces, and, in particular, torque, is transmitted to the particles in the neighbourhood of the boundary.

The stability and nonlinear evolution of a ridge of fluid on an inclined plane is investigated. This model was introduced by Hocking (1990). Here we present numerical solutions of the model showing the evolution of the ridge and in some cases the formation of droplets. Also, we investigate the linear stability of the fluid ridge allowing for contact-line motion. We find a preferred wavelength for the linear stability of spanwise disturbances.

Two-dimensional, unsteady flow of a viscous, incompressible fluid in a stepped channel has been studied by the numerical solution of the Navier–Stokes equation using an accurate finite-difference method.

With a sinusoidal mass flow rate, the problem has three governing parameters: the Reynolds number, the Strouhal number, and the step height. The effects on the flow of varying all three parameters has been investigated systematically. In appropriate parameter regimes, a strong ‘vortex wave’ is generated during the forward phase when the flow is over the step into the expansion. Secondary effects on the wave can result in a complex flow pattern with each major structure of the flow consisting of an eddy with more than one core. No such wave is found during the reverse phase of the flow. The generation and development of the wave is examined in some detail, and our results are compared and contrasted with those of previous studies, both experimental and theoretical, of flow in non-uniform vessels.

The effect of a geostrophic boundary current on internal gravity waves is studied with a reduced-gravity model. We found that the boundary current not only modifies the coastal Kelvin wave, but also forms wave guides for short internal gravity waves. The combined effects of current shear, the boundary, and the slope of the interface create the trapping mechanism. These trapped internal gravity waves appear as groups of discrete zonal modes. They have wavelengths comparable to or shorter than the internal Rossby radius of deformation. Their phase speeds are close to that of the internal Kelvin wave. However, they can propagate both in, or opposite to, the direction of the Kelvin wave. The results of the present work suggest the possibility of finding an energetic internal gravity wave phenomenon with near-inertial frequency in a broad geostrophic boundary current.

A model inviscid and incompressible flow problem is studied in which an infinite array of equi-spaced identical rectilinear line vortices moves in a uniform stream over a wall in which is embedded an equi-spaced array of discrete line sources of variable strength. It is shown that for a suitable choice of source spacing and strength, a flow that is periodic both in time and in the streamwise direction is possible. The flow is shown to be stable to small two-dimensional disturbances for a range of values of vortex height above the wall and source strength. The implications for the corresponding viscous problem and active flow control are discussed.

The hydrodynamic stability of steady axisymmetric thermocapillary flow in a cylindrical liquid bridge is investigated by linear stability theory. The basic state and the three-dimensional disturbance equations are solved by various spectral methods for aspect ratios close to unity. The critical modes have azimuthal wavenumber one and the most dangerous disturbance is either a pure hydrodynamic steady mode or an oscillatory hydrothermal wave, depending on the Prandtl number. The influence of heat transfer through the free surface, additional buoyancy forces, and variations of the aspect ratio on the stability boundaries and the neutral mode are discussed.

The evolution of three-dimensional temporally evolving plane mixing layers through as many as three pairings has been simulated numerically. All simulations were begun from a few low-wavenumber disturbances, usually derived from linear stability theory, in addition to the mean velocity. Three-dimensional perturbations were used with amplitudes ranging from infinitesimal to large enough to trigger a rapid transition to turbulence. Pairing is found to inhibit the growth of infinitesimal three-dimensional disturbances, and to trigger the transition to turbulence in highly three-dimensional flows. The mechanisms responsible for the growth of three-dimensionality and onset of transition to turbulence are described. The transition to turbulence is accompanied by the formation of thin sheets of spanwise vorticity, which undergo secondary rollups. The post-transitional simulated flow fields exhibit many properties characteristic of turbulent flows.

Direct numerical simulations of temporally evolving plane mixing layers undergoing as many as three pairings have been examined for evidence of spanwise scale change. All simulations were begun from a few low-wavenumber disturbances, usually derived from linear stability theory, in addition to the mean velocity. The amplitude of the initial three-dimensional disturbances varied from infinitesimal to large enough to trigger a rapid transition to turbulence. The time required for a change of characteristic spanwise scale with infinitesimal three-dimensional disturbances was found to be very long, requiring three or more pairings to complete a doubling of the spanwise scale. Stronger three-dimensionality can produce more rapid scale changes, but it is also likely to trigger transition to turbulence. No evidence was found for a change from an organized array of rib vortices at one spanwise scale to a similar array at a larger spanwise scale.

The evolution of vortex structure in the vicinity of a pattern defect or dislocation, generated experimentally by forcing a high Reynolds number mixing layer, is studied using a new two-dimensional wavelet transform called Arc. This transform localizes spectral information in physical space – as all wavelets do – but is not direction-specific in wavenumber space. Various types of forcing, including forcing at the fundamental and subharmonic wavenumbers, produce a range of mixing-layer responses. The most significant finding is that a dislocation site acts as a nucleus and initiates a rapid, localized evolution to larger scales. The area of the localized ‘patch’ grows approximately as the square of downstream distance. Defect-initiated patches bear generic similarities to the disturbed regions in cylinder wakes – commented upon recently by many researchers – and in particular to the Λ-structures described by Williamson (1992).

The instability of hypersonic boundary-layer flow over a flat plate is considered. The viscosity of the fluid is taken to be governed by Sutherland's formula, which gives a more accurate representation of the temperature dependence of fluid viscosity at hypersonic speeds than Chapman's approximate linear law. A Prandtl number of unity is assumed. Attention is focused on inviscid instability modes of viscous hypersonic boundary layers. One such mode, the ‘vorticity’ mode, is thought to be the fastest growing disturbance at high Mach numbers, M [Gt ] 1; in particular it is believed to have an asymptotically larger growth rate than any viscous instability. As a starting point we investigate the instability of the hypersonic boundary layer which exists far downstream from the leading edge of the plate. In this regime the shock that is attached to the leading edge of the plate plays no role, so that the basic boundary layer is non-interactive. It is shown that the vorticity mode of instability operates on a different lengthscale from that obtained if a Chapman viscosity law is assumed. In particular, we find that the growth rate predicted by a linear viscosity law overestimates the size of the growth rate by O((log M)½). Next, the development of the vorticity mode as the wavenumber decreases is described. It is shown, inter alia, that when the wavenumber is reduced to O(M-3/2) from the O(1) initial, ‘vorticity-mode’ scaling, ‘acoustic’ modes emerge.

Finally, the inviscid instability of the boundary layer near the leading-edge interaction zone is discussed. Particular attention is focused on the strong-interaction zone which occurs sufficiently close to the leading edge. We find that the vorticity mode in this regime is again unstable. The fastest growing mode is centred in the adjustment layer at the edge of the boundary layer where the temperature changes from its large, O(M2). value in the viscous boundary layer, to its O(1) free-stream value. The existence of the shock indirectly, but significantly, influences the instability problem by modifying the basic flow structure in this layer.

This paper considers the dynamics of a gas bubble in response to either a pressure pulse or a pressure step at t = 0, both in the presence and absence of a mean flow. Our work utilizes small-deformation, domain perturbation analysis carried to second and higher order in the amplitude of deformation, ε. In the absence of a mean flow, our analysis of the small deformation problem for an initial impulsive perturbation of the bubble volume and shape is closely related to recently published work by Longuet-Higgins on the time-dependent oscillations of an initially deformed bubble in a quiescent fluid. However, in the presence of a mean flow which deforms the bubble, the bubble response to pressure changes is more complex. Specifically, the present analysis identifies a number of different mechanisms for resonant interaction between shape deformation modes and the volume or radial breathing mode of oscillation. This includes not only a fundamental change in the resonant interactions at 0(ε2) - where resonant interaction is also found in the absence of mean flow – but resonant interactions also at the level of 0(ε3/2;) which are not present without the mean flow. On the other hand, the bubble dynamics in response to a step change in the pressure distribution in a quiescent fluid exhibits similar resonant interactions at 0(ε2) to those obtained for a pressure pulse in the presence of mean flow because the bubble oscillates around a non-spherical steady-state shape owing to the non-uniform pressure distribution on the bubble surface in both the cases.

Intermittent bursting events, similar to those characterizing the dynamics of near-wall turbulence, have been observed in a low-dimensional dynamical model (Aubry et al. 1988) built from eigenfunctions of the proper orthogonal decomposition (Lumley 1967). In the present work, we investigate the persistency of the intermittent behaviour in higher - but still of relatively low-dimensional dynamical systems. In particular, streamwise variations which were not accounted for in an explicit way in Aubry et al.'s model are now considered. Intermittent behaviour persists but can be of a different nature. Specifically, the non-zero streamwise modes become excited during the eruptive events so that rolls burst downstream into smaller scales. When structures have a finite length, they travel at a convection speed approximately equal to the mean velocity at the top of the layer (y+ ≈ 40). In all cases, intermittency seems to be due to homoclinic cycles connecting hyperbolic fixed points or more complex (apparently chaotic) limit sets. While these sets lie in the zero streamwise modes invariant subspace, the connecting orbits consist of nonzero streamwise modes travelling downstream. Chaotic limit sets connected by quasi-travelling waves have also been observed in a spatio-temporal chaotic regime of the Kuramoto–Sivashinsky equation (Aubry & Lian 1992a). When the limit sets lose their steadiness, the elongated rolls become randomly active, as they probably are in the real flow. A coherent structure study in our resulting flow fields is performed in order to relate our findings to experimental observations. It is shown that streaks, streamwise rolls, horseshoe vortical structures and shear layers, present in our models, are all connected to each other. Finally, criteria to determine a realistic value of the eddy viscosity parameter are developed.

The linear stability of co-extrusion flow of two upper convected Maxwell fluids through a pipe at low Reynolds numbers is studied for arbitrary wavelength disturbances. The fluids interface can become unstable due to various mechanisms. It is shown that elasticity of the fluids can stabilize the capillary instability and a linearly stable interface can be achieved by appropriate choice of controlling parameters.

In a hypersonic boundary layer over a wall of variable curvature, the region most susceptible to Görtler vortices is the temperature adjustment layer sitting at the edge of the boundary layer. This temperature adjustment layer is also the most dangerous site for Rayleigh instability. In this paper, we investigate how the existence of large-amplitude Görtler vortices affects the growth rate of Rayleigh instability. The effects of wall cooling and gas dissociation on this instability are also studied. We find that all these mechanisms increase the growth rate of Rayleigh instability and are therefore destabilizing.