The asymptotic results (Kumaran 1998b) obtained for Λ∼1 for the flow in a flexible tube are extended to the limit Λ[Lt ]1 using a numerical scheme, where Λ is the dimensionless parameter Re1/3 (G/ρV2), Re=(ρVR/η) is the Reynolds number, ρ and η are the density and viscosity of the fluid, R is the tube radius and G is the shear modulus of the wall material. The results of this calculation indicate that the least-damped mode becomes unstable when Λ decreases below a transition value at a fixed Reynolds number, or when the Reynolds number increases beyond a transition value at a fixed Λ. The Reynolds number at which there is a transition from stable to unstable perturbations for this mode is determined as a function of the parameter Σ=(ρGR2/η2), the scaled wavenumber of the perturbations kR, the ratio of radii of the wall and fluid H and the ratio of viscosities of the wall material and the fluid ηr. For ηr=0, the Reynolds number at which there is a transition from stable to unstable perturbations decreases proportional to Σ1/2 in the limit Σ[Lt ]1, and the neutral stability curves have a rather complex behaviour in the intermediate regime with the possibility of turning points and isolated domains of instability. In the limit Σ[Gt ]1, the Reynolds number at which there is a transition from stable to unstable perturbations increases proportional to Σα, where α is between 0.7 and 0.75. An increase in the ratio of viscosities ηr has a complex effect on the Reynolds number for neutrally stable modes, and it is observed that there is a maximum ratio of viscosities at specified values of H at which neutrally stable modes exist; when the ratio of viscosities is greater than this maximum value, perturbations are always stable.

Axisymmetric gravity currents in a system rotating around a vertical axis, that result when a dense fluid intrudes horizontally under a less dense ambient fluid, are studied. Situations for which the density difference between the fluid is due either to compositional differences or to suspended particulate matter are considered. The fluid motion is described theoretically by the inviscid shallow-water equations. A ‘diffusion’ equation for the volume fraction in the suspension is derived for the particle-driven case, and two different models for this purpose are presented. We focus attention on situations in which the apparent importance of the Coriolis terms relative to the inertial terms, represented by the parameter [Cscr ] (the inverse of a Rossby number), is not large. Numerical and asymptotic solutions of the governing equations clarify the essential features of the flow field and particle distribution, and point out the striking differences from the non-rotating case (Bonnecaze, Huppert & Lister 1995). It is shown that the Coriolis effects eventually become dominant; even for small [Cscr ], Coriolis effects are negligible only during an initial period of about one tenth of a revolution. Thereafter the interface of the current acquires a shape which has a downward decreasing profile at the nose and its velocity of propagation begins to decrease to zero more rapidly than in the non-rotating situation. This relates the currents investigated here to the previously studied quasi-steady oceanographic structures called rings, eddies, vortices or lenses, and may throw additional light on the dynamics of their formation. The theoretical results were tested by some preliminary experiments performed in a rotating cylinder of diameter 90 cm filled with a layer of water of depth 10 cm in which a cylinder of heavier saline fluid of diameter 9.4 cm was released.

A numerical model based on the incompressible two-dimensional Navier–Stokes equations in the Boussinesq approximation is used to study mode-2 internal solitary waves propagating on a pycnocline between two deep layers of different densities. Numerical experiments on the collapse of an initially mixed region reveal a train of solitary waves with the largest leading wave enclosing an intrusional ‘bulge’. The waves gradually decay as they propagate along the horizontal direction, with a corresponding reduction in the size of the bulge. When the normalized wave amplitude, a, falls below the critical value ac=1.18, the wave is no longer able to transport mixed fluid as it propagates away from the mixed region, and a sharp-nosed intrusion is left behind. The wave structure is studied using a Lagrangian particle tracking scheme which shows that for small amplitudes the bulges have a well-defined elliptic shape. At larger amplitudes, the bulge entrains and mixes fluid from the outside while instabilities develop in the rear part of the bulge. Results are obtained for different wave amplitudes ranging from small-amplitude ‘regular’ waves with a=0.7 to highly nonlinear unstable waves with a=3.8. The dependence of the wave speed and wavelength on amplitude is measured and compared with available experimental data and theoretical predictions. Consistent with experiments, the wave speed increases almost linearly with amplitude at small values of a. As a becomes large, the wave speed increases with amplitude at a smaller rate, which gradually approaches the asymptotic limit for a two-fluid model. Results show that in the parameter range considered the wave amplitude decreases linearly with time at a rate inversely proportional to the Reynolds number. Numerical experiments are also conducted on the head-on collision of solitary waves. The simulations indicate that the waves experience a negative phase shift during the collision, in accordance with experimental observations. Computations are used to determine the dependence of the phase shift on the wave amplitude.

It is well-known that fluidized beds are usually unstable to small perturbations and that this leads to the primary bifurcation of vertically travelling plane wavetrains. These one-dimensional periodic waves have been shown recently to be unstable to two-dimensional perturbations of large transverse wavelength in gas-fluidized beds. Here, this result is generalized to include liquid-fluidized beds and to compare typical beds fluidized with either air or water. It is shown that the instability mechanism remains the same but there are big differences in the ratio of the primary and secondary growth rates in the two cases. The tendency is that the secondary growth rates, scaled with the amplitude of a fully developed plane wave, are of similar magnitude for both gas- and liquid-fluidized beds, while the primary growth rate is much larger in the gas-fluidized bed. This means that the secondary instability is accordingly stronger than the primary instability in the liquid-fluidized bed, and consequently sets in at a much smaller amplitude of the primary wave. However, since the waves in the liquid-fluidized bed develop on a larger time and length scale, the primary perturbations need longer time and thereby travel farther until they reach the critical amplitude. Which patterns are more amenable to being visually recognized depends on the magnitude of the initially imposed disturbance and the dimensions of the apparatus. This difference in scale plays a key role in bringing about the differences between gas- and liquid-fluidized beds; it is produced mainly by the different values of the Froude number.

The proper orthogonal decomposition (POD) is applied to the minimal flow unit (MFU) of a turbulent channel flow. Our purpose is to establish a numerical validation of low-dimensional models based on the POD. The simplest (two-mode) model possible is built for the simplified flow in the minimal unit. The dynamical behaviour predicted by the model is compared with that actually occurring in the direct numerical simulation of the flow. The various modelling assumptions which underlie the construction of low-dimensional models are examined and confronted with numerical evidence. The relationship between intermittency in the MFU and intermittent low-dimensional parameters is investigated closely. The agreement observed is quite satisfactory, especially given the crudeness of the truncation considered. To further demonstrate the adequacy of the model, we develop a dynamical filtering procedure to recover information from realistic (partial) measurements. The success obtained illustrates the versatility of the low-dimensional paradigm.

Propagation of solitary waves in curved shallow water channels of constant depth and width is investigated by carrying out numerical simulations based on the generalized weakly nonlinear and weakly dispersive Boussinesq model. The objective is to investigate the effects of channel width and bending sharpness on the transmission and reflection of long waves propagating through significantly curved channels. Our numerical results show that, when travelling through narrow channel bends including both smooth and sharp-cornered 90°-bends, a solitary wave is transmitted almost completely with little reflection and scattering. For wide channel bends, we find that, if the bend is rounded and smooth, a solitary wave is still fully transmitted with little backward reflection, but the transmitted wave will no longer preserve the shape of the original solitary wave but will disintegrate into several smaller waves. For solitary waves travelling through wide sharp-cornered 90°-bends, wave reflection is seen to be very significant, and the wider the channel bend, the stronger the reflected wave amplitude. Our numerical results for waves in sharp-cornered 90°-bends revealed a similarity relationship which indicates that the ratios of the transmitted and reflected wave amplitude, excess mass and energy to the original wave amplitude, mass and energy all depend on one single dimensionless parameter, namely the ratio of the channel width b to the effective wavelength λe. Quantitative results for predicting wave transmission and reflection based on b/λe are presented.

Lagrangian properties of oceanic turbulent boundary layers were measured using neutrally buoyant floats. Vertical acceleration was computed from pressure (depth) measured on the floats. An average vertical vorticity was computed from the spin rate of the float. Forms for the Lagrangian frequency spectra of acceleration, ϕa(ω), and the Lagrangian frequency spectrum of average vorticity are found using dimension analysis. The flow is characterized by a kinetic energy dissipation rate, ε, and a large-eddy frequency, ω0. The float is characterized by its size. The proposed non-dimensionalization accurately collapses the observed spectra into a common form. The spectra differ from those expected for perfect Lagrangian measurements over a substantial part of the measured frequency range owing to the finite size of the float. Exact theoretical forms for the Lagrangian frequency spectra are derived from the corresponding Eulerian wavenumber spectra and a wavenumber–frequency distribution function used in previous numerical simulations of turbulence. The effect of finite float size is modelled as a spatial average. The observed non-dimensional acceleration and vorticity spectra agree with these theoretical predictions, except for the high-frequency part of the vorticity spectrum, where the details of the float behaviour are important, but inaccurately modelled. A correction to the exact Lagrangian acceleration spectra due to measurement by a finite-sized float is thus obtained. With this correction, a frequency range extending from approximately one decade below ω0 to approximately one decade into the inertial subrange can be resolved by the data. Overall, the data are consistent with the proposed transformation from the Eulerian wavenumber spectrum to the Lagrangian frequency spectrum. Two parameters, ε and ω0, are sufficient to describe Lagrangian spectra from several different oceanic turbulent flows. The Lagrangian Kolmogorov constant for acceleration, βa≡ϕa/ε, has a value between 1 and 2 in a convectively driven boundary layer. The analysis suggests a Lagrangian frequency spectrum for vorticity that is white at all frequencies in the inertial subrange and below, and a Lagrangian frequency spectrum for energy that is white below the large-eddy scale and has a slope of −2 in the inertial subrange.

The buoyancy-induced laminar flow and temperature fields associated with a line source of heat in an unbounded environment are described by numerically solving the non-dimensional Boussinesq equations with the appropriate boundary conditions. The solution is given for values of the Prandtl number, the single parameter, ranging from zero to infinity. The far-field form of the solution is well known, including a self-similar thermal plume above the source. The analytical description close to the source involves constants that must be evaluated with the numerical solution.

These constants are used when calculating the free convection heat transfer from wires (or cylinders of non-circular shape) at small Grashof numbers. We find two regions in the flow field: an inner region, scaled with the radius of the wire, where the effects of convection can be neglected in first approximation, and an outer region where, also in first approximation, the flow and temperature fields are those due to a line source of heat. The cases of large and small Prandtl numbers are considered separately. There is good agreement between the Nusselt numbers given by the asymptotic analysis and by the numerical analysis, which we carry out for a wide range of Grashof numbers, extending to very small values the range of existing numerical results; there is also agreement with the existing correlations of the experimental results. A correlation expression is proposed for the relation between the Nusselt and Grashof numbers, based on the asymptotic forms of the relation for small and large Grashof numbers.

Large-eddy simulation (LES) has been used to calculate the flow of a statistically two-dimensional turbulent boundary layer over a bump. Subgrid-scale stresses in the filtered Navier–Stokes equations were closed using the dynamic eddy viscosity model. LES predictions for a range of grid resolutions were compared to the experimental measurements of Webster et al. (1996). Predictions of the mean flow and turbulence intensities are in good agreement with measurements. The resolved turbulent shear stress is in reasonable agreement with data, though the peak is over-predicted near the trailing edge of the bump. Analysis of the flow confirms the existence of internal layers over the bump surface upstream of the summit and along the downstream trailing at plate, and demonstrates that the quasi-step increases in skin friction due to perturbations in pressure gradient and surface curvature selectively enhance near-wall shear production of turbulent stresses and are responsible for the formation of the internal layers. Though the flow experiences a strong adverse pressure gradient along the rear surface, the boundary layer is unique in that intermittent detachment occurring near the wall is not followed by mean-flow separation. Certain turbulence characteristics in this region are similar to those previously reported in instantaneously separating boundary layers. The present investigation also explains the driving mechanism for the surprisingly rapid return to equilibrium over the trailing flat plate found in the measurements of Webster et al. (1996), i.e. the simultaneous uninterrupted development of an inner energy-equilibrium region and the monotonic decay of elevated turbulence shear production away from the wall. LES results were also used to examine response of the dynamic eddy viscosity model. While subgrid-scale dissipation decreases/increases as the turbulence is attenuated/enhanced, the ratio of the (averaged) forward to reverse energy transfers predicted by the model is roughly constant over a significant part of the layer. Model predictions of backscatter, calculated as the percentage of points where the model coefficient is negative, show a rapid recovery downstream similar to the mean-flow and turbulence quantities.

We investigate the stability of interfacial waves in conducting fluids under the influence of a vertical current density, paying particular attention to aluminium reduction cells in which such instabilities are commonly observed. We develop a wave equation for the interface in which the Lorentz force is expressed explicitly in terms of the fluid motion. Our wave equation differs from previous models, most notably that developed by Urata (1985), in that earlier formulations rested on a more complex, implicit coupling between the fluid motion and the Lorentz force. Our formulation furnishes a number of quite general stability results without the need to resort to Fourier analysis. (The need for Fourier analysis typifies previous studies.) Moreover, our equation supports both travelling and standing waves. We investigate each in turn.

We obtain three new results. First, we show that travelling waves may become unstable in the presence of a uniform, vertical magnetic field. (Our travelling waves are quite different to those discovered by previous investigators (Sneyd 1985 and Moreau & Ziegler 1986) which require more complex magnetic fields to become unstable.) Second, in line with previous studies we confirm that standing waves may also become unstable. In this context we derive a simple energy criterion which shows which types of motion may extract energy from the background magnetic field. This indicates that a rotating, tilted interface is particularly prone to instability, and indeed such a motion is often seen in practice. Finally, we use Gershgorin's theorem to produce a sufficient condition for the stability of standing waves in a finite domain. This allows us to place a lower bound on the critical value of the background magnetic field at which an instability first appears, without solving the governing equations of motion.

Cross-stream migration and stable orientations of elliptic particles falling in an Oldroyd-B fluid in a channel are studied. We show that the normal component of the extra stress on a rigid body vanishes; lateral forces and torques are determined by the pressure. Inertia turns the long side of the ellipse across the stream and elasticity turns it along the stream; tilted off-centre falling is unstable. There are two critical numbers: the elasticity and Mach numbers. When the elasticity number is smaller than critical the fluid is essentially Newtonian with broadside-on falling at the centreline of the channel. For larger elasticity numbers the settling turns the long side of the particle along the stream in the channel centre for all velocities below a critical one, identified with a critical Mach number of order one. For larger Mach numbers the ellipse flips into broadside-on falling again. The critical numbers are functions of the channel blockage ratio, the particle aspect ratio and the retardation/relaxation time ratio of the fluid. Two ellipses falling near to each other, attract, line-up vertically and straighten-out with long sides vertical. Stable, off-centre tilting is found for ellipses falling in shear-thinning fluids and for cylinders with flat ends in which particles tend to align their longest diameter with gravity.

Collapsing shock-bounded cavities in fast/slow (F/S) spherical and near-spherical configurations give rise to expelled jets and vortex rings. In this paper, we simulate with the Euler equations planar shocks interacting with an R12 axisymmetric spherical bubble. We visualize and quantify results that show evolving upstream and downstream complex wave patterns and emphasize the appearance of vortex rings. We examine how the magnitude of these structures scales with Mach number. The collapsing shock cavity within the bubble causes secondary shock refractions on the interface and an expelled weak jet at low Mach number. At higher Mach numbers (e.g. M=2.5) ‘vortical projectiles’ (VP) appear on the downstream side of the bubble. The primary VP arises from the delayed conical vortex layer generated at the Mach disk which forms as a result of the interaction of the curved incoming shock waves that collide on the downstream side of the bubble. These rings grow in a self-similar manner and their circulation is a function of the incoming shock Mach number. At M=5.0, it is of the same order of magnitude as the primary negative circulation deposited on the bubble interface. Also at M=2.5 and 5.0 a double vortex layer arises near the apex of the bubble and moves off the interface. It evolves into a VP, an asymmetric diffuse double ring, and moves radially beyond the apex of the bubble. Our simulations of the Euler equations were done with a second-order-accurate Harten–Yee-type upwind TVD scheme with an approximate Riemann Solver on mesh resolution of 803×123 with a bubble of radius 55 zones.