Basaltic volcanoes erupt in several different regimes which have not been explained. At Kilauea (Hawaii), eruption can take the form of either fire fountaining, where gas-rich jets propel lava clots to great heights in the atmosphere, or quiet effusive outflow of vesicular lava. Another regime is commonly observed at Stromboli, where large gas slugs burst intermittently at the vent. In an attempt to provide a unifying framework for these regimes, we investigate phenomena induced by degassing in a reservoir which empties into a small conduit. Laboratory experiments are done in a cylindrical tank topped by a thin vertical tube. Working liquids are silicone oils and glycerol solutions to investigate a range of viscosity and surface tension. Gas bubbles are generated at the tank bottom with known bubble diameter and total gas flux. The bubbles rise through the tank and accumulate in a foam layer at the roof. Depending on the behaviour of this foam layer, three different regimes can be distinguished: (i) steady horizontal flow of the foam leading to bubbly flow in the conduit; (ii) alternating regimes of foam build-up and collapse leading to the eruption of a single, large gas pocket; (iii) flow of the foam partially coalesced into larger gas pockets leading to intermittent slug flow in the conduit. These regimes have natural counterparts in basaltic volcanoes.

A simple theory is proposed to explain regimes (i) and (ii). The bubbles in contact with the roof deform under the action of buoyancy forces, developing flat contact areas whose size increases as a function of foam thickness. Maximum deformation corresponds to a critical thickness hc = 2σ/ερlgR, where σ is the coefficient of surface tension, ρl the liquid density, g the acceleration due to gravity, R the bubble radius and ε the gas volume fraction in the foam. The foam thickness is determined by a balance between the input of bubbles from below and the output into the conduit, and is proportional to (μlQ/ε2 ρlg)¼, where μl is the liquid viscosity and Q the gas flux. A necessary and sufficient condition for collapse is that it exceeds the critical value hc. In a liquid of given physical properties, this occurs when the gas flux exceeds a critical value which depends on viscosity, surface tension and bubble size. Experimental determinations of the critical gas flux and of the time between two events of foam collapse are in agreement with this simple theory.

In problems where a viscous fluid flows past a porous solid it has frequently been assumed that the tangential component of surface velocity is zero. When the porous solid has an open structure with large pores the external surface stress may produce a tangential flow below the surface. Recently, Beavers & Joseph (1967) have assumed that the surface velocity UB depends on the mean tangential stress $[\mu(d\overline{u}/dy)]_{y=0}$ in the fluid outside the porous solid through the relation \[ \left[\mu\frac{d\overline{u}}{dy}\right]_{y=0} = \frac{\mu\alpha}{k^{\frac{1}{2}}}(U_B-Q), \] where Q is the volume flow rate per unit cross-section within the porous material due to the pressure gradient, k is the Darcy constant and α is a constant which depends only on the nature of the porosity. An artificial mathematical model of a porous medium is proposed for which the flow can be calculated both inside and outside the surface. This conceptual model was materialized and the experimental results agree with the calculations. The calculated values of α so found are not quite independent of the external means of producing the external tangential stress.

Some general stability criteria for non-dissipative swirling flows are derived, and extended to the case of an electrically conducting fluid in the presence of axial magnetic field and current. In particular it is shown that the analogy between a rotating and a stratified fluid holds in this case, and that an important determinant of stability is a ‘Richardson number’ based on the analogue of the density gradient and the shear in the axial flow.

Direct numerical simulation databases have been used to study the effect of compressibility on mixing layers. The simulations cover convective Mach numbers from 0.2 to 1.2 and all contain a fully resolved turbulent energy cascade to small spatial scales. Statistical information is extracted from the databases to determine reasons for the reduced growth rate that is observed as the convective Mach number is increased. It is found that the dilatational contribution to dissipation is negligible even when eddy shocklets are observed in the flow. Also pressure-dilatation is not found to be significant. Using an accurate relation between the momentum thickness growth rate and the production of turbulence kinetic energy together with integrated equations for the Reynolds stress tensor it is shown that reduced pressure fluctuations are responsible for the changes in growth rate via the pressure–strain term. A deterministic model for the required pressure fluctuations is given based on the structure of variable-density vortices and the assumption that the limiting eddies are sonic. Simple anisotropy considerations are used to close the averaged equations. Good agreement with turbulence statistics obtained from the simulations is found.

An approximate set of equations is derived for a compressible liquid of infinite Prandtl number. These are referred to as the anelastic-liquid equations. The approximation requires the product of absolute temperature and volume coefficient of thermal expansion to be small compared to one. A single parameter defined as the ratio of the depth of the convecting layer, d, to the temperature scale height of the liquid, HT, governs the importance of the non-Boussinesq effects of compressibility, viscous dissipation, variable adiabatic temperature gradients and non-hydrostatic pressure gradients. When d/HT [Lt ] 1 the Boussinesq equations result, but when d/HT is O(1) the non-Boussinesq terms become important. Using a time-dependent numerical model, the anelastic-liquid equations are solved in two dimensions and a systematic investigation of compressible convection is presented in which d/HT is varied from 0·1 to 1·5. Both marginal stability and finite-amplitude convection are studied. For d/HT [les ] 1·0 the effect of density variations is primarily geometric; descending parcels of liquid contract and ascending parcels expand, resulting in an increase in vorticity with depth. When d/HT > 1·0 the density stratification significantly stabilizes the lower regions of the marginal state solutions. At all values of d/HT [ges ] 0·25, an adiabatic temperature gradient proportional to temperature has a noticeable stabilizing effect on the lower regions. For d/HT [ges ] 0·5, marginal solutions are completely stabilized at the bottom of the layer and penetrative convection occurs for a finite range of supercritical Rayleigh numbers. In the finite-amplitude solutions adiabatic heating and cooling produces an isentropic central region. Viscous dissipation acts to redistribute buoyancy sources and intense frictional heating influences flow solutions locally in a time-dependent manner. The ratio of the total viscous heating in the convecting system, ϕ, to the heat flux across the upper surface, Fu, has an upper limit equal to d/HT. This limit is achieved at high Rayleigh numbers, when heating is entirely from below, and, for sufficiently large values of d/HT, Φ/Fu is greater than 1·00.

The solution of the steady-state Navier–Stokes equations in three dimensions has been obtained by a numerical method for the problem of natural convection in a rectangular cavity as a result of differential side heating. In the past, this problem has generally been treated as though it were two-dimensional. The solutions explore the three-dimensional motion generated by the presence of no-slip adiabatic end walls. For Ra = 104, the three-dimensional motion is shown to be the result of the inertial interaction of the rotating flow with the stationary walls together with a contribution arising from buoyancy forces generated by longitudinal temperature gradients. The inertial effect is inversely dependent on the Prandtl number, whereas the thermal effect is nearly constant. For higher values of Ra, multiple longitudinal flows develop which are a delicate function of Ra, Pr and the cavity aspect ratios.

The structure of turbulent flow over large roughness consisting of regular arrays of cubical obstacles is investigated numerically under constant pressure gradient conditions. Results are analysed in terms of first- and second-order statistics, by visualization of instantaneous flow fields and by conditional averaging. The accuracy of the simulations is established by detailed comparisons of first- and second-order statistics with wind-tunnel measurements. Coherent structures in the log region are investigated. Structure angles are computed from two-point correlations, and quadrant analysis is performed to determine the relative importance of Q2 and Q4 events (ejections and sweeps) as a function of height above the roughness. Flow visualization shows the existence of low-momentum regions (LMRs) as well as vortical structures throughout the log layer. Filtering techniques are used to reveal instantaneous examples of the association of the vortices with the LMRs, and linear stochastic estimation and conditional averaging are employed to deduce their statistical properties. The conditional averaging results reveal the presence of LMRs and regions of Q2 and Q4 events that appear to be associated with hairpin-like vortices, but a quantitative correspondence between the sizes of the vortices and those of the LMRs is difficult to establish; a simple estimate of the ratio of the vortex width to the LMR width gives a value that is several times larger than the corresponding ratio over smooth walls. The shape and inclination of the vortices and their spatial organization are compared to recent findings over smooth walls. Characteristic length scales are shown to scale linearly with height in the log region. Whilst there are striking qualitative similarities with smooth walls, there are also important differences in detail regarding: (i) structure angles and sizes and their dependence on distance from the rough surface; (ii) the flow structure close to the roughness; (iii) the roles of inflows into and outflows from cavities within the roughness; (iv) larger vortices on the rough wall compared to the smooth wall; (v) the effect of the different generation mechanism at the wall in setting the scales of structures.

Experimental results are given for various statistical properties of the fluctuating wall-pressure field associated with a subsonic turbulent equilibrium boundary layer, developed on a smooth wind tunnel wall after natural transition from laminar to turbulent flow.

The statistical quantities of the wall-pressure field investigated were root-mean-square pressure, frequency power spectrum and space-time correlations. Space-time correlation measurements were made in both broad and narrow frequency bands. The experiments were made at flow Mach numbers of 0·3 and 0·5 and covered a Reynolds number range of about 5 to 1.

The main conclusion to which the measurements lead is that the wall-pressure field has a structure produced by contributions from pressure sources in the boundary layer with a wide range of convection velocities, and comprises two families of convected wave-number components. One family is of high wave-number components and is associated with turbulent motion in the constant stress layer; the components are longitudinally coherent for times proportional to the times taken for them to be convected distances equal to their wavelengths and laterally coherent over distances proportional to their wavelengths. The other family comprises components of wavelength greater than about twice the boundary-layer thickness, which lose coherence as a group more or less independently of wavelength and are associated with large-scale eddy motion in the boundary layer, outside the constant stress layer. The evolution of the pressure field is discussed in terms of these two wave-number families.

Flow visualization and laser-anemometry measurements are reported in the flow downstream of a plane 3: 1 symmetric expansion in a duct with an aspect ratio of 9·2: 1 downstream of the expansion. The flow was found to be markedly dependent on Reynolds number, and strongly three-dimensional even well away from the channel corners except at the lowest measurable velocities. The measurements at a Reynolds number of 56 indicated that the separation regions behind each step were of equal length. Symmetric velocity profiles existed from the expansion to a fully developed, parabolic profile far downstream, although there were substantial three-dimensional effects in the vicinity of the separation regions. The velocity profiles were in good agreement with those obtained by solving the two-dimensional momentum equation. At a Reynolds number of 114, the two separation regions were of different lengths, leading to asymmetric velocity profiles; three dimensional effects were much more pronounced. At a Reynolds number of 252, a third separation zone was found on one wall, downstream of the smaller of the two separation zones adjacent to the steps. As at the lower Reynolds numbers, the flow was very stable. At higher Reynolds numbers the flow became less stable and periodicity became increasingly important in the main stream; this was accompanied by a highly disturbed fluid motion in the separation zones, as the flow tended towards turbulence.

We present results from a direct numerical simulation of the passive transport of solid particles by a fully developed turbulent channel flow with a Reynolds number of 180 based on the friction velocity and the channel half-width. Three particle sets are studied, ranging in diameter from 0.5 to 1.4 viscous wall units and in aerodynamic time constant from 0.6 to 56 centreline Kolmogorov time scales. We use particle number density histograms and fractal dimensions to show that the level of order in the particle spatial distribution peaks near a Stokes number of unity based on the Kolmogorov time scale. We then quantify the relationship between this spatial distribution and the instantaneous flow topology. The results indicate that the previously reported preferential concentration of particles in low-speed streaks leads to a suppression of particle velocities in the viscous sublayer and buffer region even in the presence of streamwise gravitational acceleration. In other regions of the flow, the particles' non-random spatial distribution is shown to be uncorrelated with the local flow topology. We compare our results with the experimental data of Kulick et al. (1994) and Fessler et al. (1994) and confirm that the latter authors' results were not influenced by turbulence modification.

The Kelvin–Helmholtz rollup of three-dimensional temporally evolving plane mixing layers with an initial Reynolds number of 500 based on vorticity thickness and half the velocity difference have 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 profile. A standard set of ‘clean’ structures develops in the majority of the simulations. The spanwise vorticity rolls up into a corrugated spanwise roller with vortex stretching creating strong spanwise vorticity in a cup-shaped region at the bends of the roller. Predominantly streamwise rib vortices develop in the braid region between the rollers. For sufficiently strong initial three-dimensional disturbances these ribs ‘collapse’ into compact axi-symmetric vortices. The rib vortex lines connect to neighbouring ribs and are kinked in the direction opposite to that of the roller vortex lines. Because of this, these two sets of vortex lines remain distinct. For certain initial conditions, persistent ribs do not develop. In such cases, the development of significant three-dimensionality is delayed.

In addition, simulations of infinitesimal three-dimensional disturbances evolving in a two-dimensional mixing layer were performed. Many features of the fully nonlinear flows are remarkably well predicted by the linear computations. Such computations can thus be used to predict the degree of three-dimensionality in the mixing layer even after the onset of nonlinearity. Several nonlinear effects can also be identified by comparing linear and nonlinear computations. These include the collapse of rib vortices, the formation of cups of spanwise vorticity, and the appearance of spanwise vorticity with sign opposite that of the mean vorticity. These nonlinear effects have been identified as precursors of the transition to turbulence (Moser & Rogers 1991).

The stability of small two-dimensional travelling-wave disturbances in an incompressible laminar boundary layer over a flexible surface is considered. By first determining the wall admittance required to maintain a wave of given wave-number and phase speed, a characteristic equation is deduced which, in the limit of zero wall flexibility, reduces to that occurring in the ordinary stability theory of Tollmien and Schlichting. The equation obtained represents a slight and probably insignificant improvement upon that given recently by Benjamin (1960). Graphical methods are developed to determine the curve of neutral stability, as well as to identify the various modes of instability classified by Benjamin as ‘Class A’, ‘Class B’, and ‘Kelvin-Helmholtz’ instability, respectively. Also, a method is devised whereby the optimum combination of surface effective mass, wave speed, and damping required to stabilize any given unstable Tollmien-Schlichting wave can be determined by a simple geometrical construction in the complex wall-admittance plane.

What is believed to be a complete physical explanation for the influence of an infinite flexible wall on boundary-layer stability is presented. In particular, the effect of damping in the wall is discussed at some length. The seemingly paradoxical result that damping destabilizes class A waves (i.e. waves of the Tollmien-Schlichting type) is explained by considering the related problem of flutter of an infinite panel in incompressible potential flow, for which damping has the same qualitative effect. It is shown that the class A waves are associated with a decrease of the total kinetic and elastic energy of the fluid and the wall, so that any dissipation of energy in the wall will only make the wave amplitude increase to compensate for the lowered energy level. The Kelvin-Helmholtz type of instability will occur when the effective stiffness of the panel is too low to withstand, for all values of the phase speed, the pressure forces induced on the wavy wall.

The numerical examples presented show that the increase in the critical Reynolds number that can be achieved with a wall of moderate flexibility is modest, and that some other explanation for the experimentally observed effects of a flexible wall on the friction drag must be considered.

The analysis reported in Part 1 is extended here to the case in which the conductivity κ is large compared with the viscosity ν, the conduction ‘cut-off’ to the θ-spectrum then being at wave-number (ε/κ3)¼. It is shown, with a plausible and consistent hypothesis, that the convective supply of $\overline {\theta^2}$-stuff to Fourier components of θ with wave-numbers n in the range (ε/κ3)¼ [Lt ] n [Gt ] (ε/ν3)¼ is due primarily to motion on a length scale of order n-1 acting on a uniform gradient of θ of magnitude $[(\overline {\nabla \theta)^2}]^{\frac {1}{2}}$. The consequent form of the theta;-spectrum within this same wave-number range is $\Gamma (n) = \frac {1}{3}C \chi \epsilon ^{\frac {2}{3}} k^{-3}n^ {-\frac {17} {3}}.$

The way in which conduction influences (and restricts) the effect of convection on the distribution of θ at these wave-numbers beyond the conduction cut-off is discussed.

A parabolic approximation to the reduced wave equation is investigated for the propagation of periodic surface waves in shoaling water. The approximation is derived from splitting the wave field into transmitted and reflected components.

In the case of an area with straight and parallel bottom contour lines, the asymptotic form of the solution for high frequencies is compared with the geometrical optics approximation.

Two numerical solution techniques are applied to the propagation of an incident plane wave over a circular shoal.

Several models are developed for the high-wavenumber portion of the spectral transfer function of scalar quantities advected by high-Reynolds-number, locally isotropic turbulent flow. These models are applicable for arbitrary Prandtl or Schmidt number, v/D, and the resultant scalar spectra are compared with several experiments having different v/D. The ‘bump’ in the temperature spectrum of air observed over land is shown to be due to a tendency toward a viscous-convective range and the presence of this bump is consistent with experiments for large v/D. The wavenumbers defining the transition between the inertial-convective range and viscous-convective range for asymptotically large v/D (denoted k* and k1* for the three- and one-dimensional spectra) are determined by comparison of the models with experiments. A measurement of the transitional wavenumber k1* [denoted (k1*)s] is found to depend on v/D and on any filter cut-off. On the basis of the k* values it is shown that measurements of β1 from temperature spectra in moderate Reynolds number turbulence in air (v/D = 0·72) maybe over-estimates and that the inertial-diffusive range of temperature fluctuations in mercury (v/D ≃ 0·02) is of very limited extent.

The interaction of a laser-induced cavitation bubble with an elastic boundary is investigated experimentally by high-speed photography and acoustic measurements. The elastic material consists of a polyacrylamide (PAA) gel whose elastic properties can be controlled by modifying the water content of the sample. The elastic modulus, E, is varied between 0.017 MPa and 2.03 MPa, and the dimensionless bubble–boundary distance, γ, is for each value of E varied between γ = 0 and γ = 2.2. In this parameter space, jetting behaviour, jet velocity, bubble migration and bubble oscillation time are determined. The jetting behaviour varies between liquid jet formation towards or away from the elastic boundary, and formation of an annular jet which results in bubble splitting and the subsequent formation of two very fast axial liquid jets flowing in opposite directions. The liquid jet directed away from the boundary reaches a maximum velocity between 300 ms−1 and 600 ms−1 (depending on the elastic modulus of the sample) while the peak velocity of the jet directed towards the boundary ranges between 400 ms−1 and 800 ms−1 (velocity values averaged over 1 μs). Penetration of the elastic boundary by the liquid jet is observed for PAA samples with an intermediate elastic modulus between 0.12 and 0.4 MPa. In this same range of elastic moduli and for small γ-values, PAA material is ejected into the surrounding liquid due to the elastic rebound of the sample surface that was deformed during bubble expansion and forms a PAA jet upon rebound. For stiffer boundaries, the bubble behaviour is mainly characterized by the formation of an axial liquid jet and bubble migration directed towards the boundary, as if the bubble were adjacent to a rigid wall. For softer samples, the bubble behaviour becomes similar to that in a liquid with infinite extent. During bubble collapse, however, material is torn off the PAA sample when bubbles are produced close to the boundary. We conclude that liquid jet penetration into the boundary, jet-like ejection of boundary material, and tensile-stress-induced deformations of the boundary during bubble collapse are the major mechanisms responsible for cavitation erosion and for cavitation-enhanced ablation of elastic materials as, for example, biological tissues.

The attached eddy hypothesis developed for zero pressure gradient boundary layers and for pipe flow is extended here to boundary layers with arbitrary streamwise pressure gradients, both favourable and adverse. It is found that in order to obtain the correct quantitative results for all components of the Reynolds stresses, two basic types of eddy structure geometries are required. The first type, called type-A, is interpreted to give a ‘wall structure’ and the second, referred to as type-B, gives a ‘wake structure’. This is in analogy with the conventional mean velocity formulation of Coles where the velocity is decomposed into a law of the wall and a law of the wake.

If the above mean velocity formulation is accepted, then in principle, once the eddy geometries are fixed for the two eddy types, all Reynolds stresses and associated spectra contributed from the attached eddies can be computed without any further empirical constants. This is done by using the momentum equation and certain convolution integrals developed here based on the attached eddy hypothesis. The theory is developed using data from equilibrium and quasi-equilibrium flows. In Part 2 the authors’ non-equilibrium data are used.

This paper is concerned with the dispersion of a material quantity in the steady flow of a viscous fluid through a random network of capillaries (which is a useful model of a porous medium), for the case in which molecular diffusion and macroscopic mixing, due to the randomness of the streamlines, are both important. A Lagrangian correlation function is introduced and the longitudinal and lateral effective diffusivities are thereby calculated for all values of Ul/κ less than some large value. Here, l denotes the length of a capillary, U the mean velocity of the fluid, and κ the molecular diffusivity of the material quantity. The theory is compared with experimental observations of dispersion in flow through granular beds.

A drop of low viscosity hitting a solid may bounce, provided that the material is highly hydrophobic. As a model of such a situation, we consider here the case of a very hot solid. Then, as discovered by Leidenfrost, a thin layer of vapour sustains the drop, preventing any contact with the substrate. On hitting such a solid, a drop rebounds, and we discuss here the elasticity of the shock. Two very different cases are described: at a large velocity, the weaker the impact velocity, the weaker the elasticity; at a small velocity, a quasi-elastic regime is found. The boundary between the two domains is set by a Weber number, which compares the kinetic and surface energies of the drop, of order unity.

The dynamic transition from combustion noise to combustion instability was investigated experimentally in two laboratory-scale turbulent combustors (namely, swirl-stabilized and bluff-body-stabilized backward-facing-step combustors) by systematically varying the flow Reynolds number. We observe that the onset of combustion-driven oscillations is always presaged by intermittent bursts of high-amplitude periodic oscillations that appear in a near-random fashion amidst regions of aperiodic low-amplitude fluctuations. These excursions to periodic oscillations last longer in time as operating conditions approach instability and finally the system transitions completely into periodic oscillations. A continuous measure to quantify this bifurcation in dynamics can be obtained by defining an order parameter as the probability of the signal amplitude exceeding a predefined threshold. A hysteresis zone was observed in the bluff-body-stabilized configuration that was absent in the swirl-stabilized configuration. The recurrence properties of the dynamics of intermittent burst oscillations were quantified using recurrence plots and the distribution of the aperiodic phases was examined. From the statistics of these aperiodic phases, robust early-warning signals of an impending combustion instability may be obtained.