The stability of a two-dimensional surfactant-free (gas–liquid) foam in a gravitational field is considered. The foam is assumed to have low liquid fraction, so the gas phase can be divided into approximately polygonal bubbles separated by thin liquid films. These free films drain toward accumulations of liquid at the bubble vertices, the Plateau borders, and eventually rupture due to van der Waals intermolecular attractions; this drives foam coarsening through the coalescence of neighbouring bubbles. In particular, we demonstrate how gravitational effects strongly modify the shape of the Plateau border interfaces and enhance the drainage flow in the liquid films, driving non-uniform thinning with exponential decay of the minimum film thickness, significantly faster than the power-law thinning predicted when gravitational effects are negligible.

Interference heating effects generated by a blunt fin-type protuberance on a flat plate exposed to a hypersonic flow have been investigated experimentally and numerically. Experiments and simulations were carried out at a free-stream Mach number of 6.7 under laminar flow conditions. The surface heating on the plate was measured experimentally using liquid-crystal thermography, which provided quantitative data with high spatial resolution. Complementary surface oil flow and schlieren experiments were also carried out to gain a better understanding of the interference flow field. The effects of fin leading-edge diameter on the heating distribution on the flat plate surface were explored. The results of the experiments and simulations agree well and reveal a highly complex interaction region which extends over seven diameters upstream of the fin. Within the interaction region surrounding the fin, heating enhancements up to ten times the undisturbed flat plate value were estimated from the experimental data. However, the liquid crystals have a limited range, and the numerical simulations indicated localized peak heating many times this value both on the plate and the fin itself.

In commonly used formulations of the Boussinesq approximation centrifugal buoyancy effects related to differential rotation, as well as strong vortices in the flow, are neglected. However, these may play an important role in rapidly rotating flows, such as in astrophysical and geophysical applications, and also in turbulent convection. Here we provide a straightforward approach resulting in a Boussinesq-type approximation that consistently accounts for centrifugal effects. Its application to the accretion-disc problem is discussed. We numerically compare the new approach to the typical one in fluid flows confined between two differentially heated and rotating cylinders. The results justify the need of using the proposed approximation in rapidly rotating flows.

Previous studies have shown that the formation of coherent vortex rings in the near-wake of a self-propelled vehicle can increase propulsive efficiency compared with a steady jet wake. The present study utilizes a self-propelled vehicle to explore the dependence of propulsive efficiency on the vortex ring characteristics. The maximum propulsive efficiency was observed to occur when vortex rings were formed of the largest physical size, just before the leading vortex ring would pinch off from its trailing jet. These experiments demonstrate the importance of vortex ring pinch off in self-propelled vehicles, where coflow modifies the vortex dynamics.

Laminar–turbulent transition in the boundary layer at supersonic speeds can be initiated by small solid particles present in the free stream. Particulates interacting with the boundary-layer flow generate unstable wavepackets related to Tollmien–Schlichting (TS) waves. The latter grow downstream and ultimately break down to turbulent spots. This scenario of TS-dominated transition is modelled using the Mack amplitude method. A theoretical model describing the receptivity mechanism is developed to predict the initial spectrum of TS waves. With these initial conditions the downstream growth of TS instability is calculated using the linear stability theory. The transition onset is associated with the point where the disturbance amplitude reaches a threshold value. As an example, calculations are carried out for a 14° half-angle sharp wedge flying in the standard atmosphere at altitude 20 km, Mach number 4 and zero angle of attack. It is shown that spherical particles of radius from $10$ to $20~\unicode[.5,0][STIXGeneral,Times]{x03BC} \mathrm{m} $ and density ${\geqslant }1~\mathrm{g} ~{\mathrm{cm} }^{- 3} $ can cause transition onset corresponding to the amplification factor $N= 9{\unicode{x2013}} 10$, which is in the empirical range of flight data. This indicates that atmospheric particulates may be a major source of TS-dominated transition on aerodynamically smooth surfaces at supersonic speeds. The receptivity model provides a foundation for further treatments of different cases associated with transition in dusty environments. It can also be used for predictions of particle-induced transition at subsonic and hypersonic speeds.

A complete nonlinear self-similar solution that characterizes the impact of two liquid wedges symmetric about the velocity direction is obtained assuming the liquid to be ideal and incompressible, with negligible surface tension and gravity effects. Employing the integral hodograph method, analytical expressions for the complex potential and for its derivatives are derived. The boundary value problem is reduced to two integro-differential equations in terms of the velocity modulus and angle to the free surface. Numerical results are presented in a wide range of wedge angles for the free surface shapes, streamline patterns, and pressure distributions. It is found that the splash jet may cause secondary impacts. The regions with and without secondary impacts in the plane of the wedge angles are determined.

Exact compressible one-dimensional nozzle flow solutions at steady state are determined in various limit situations of two-phase liquid–gas mixtures. First, the exact solution for a pure liquid nozzle flow is determined in the context of fluids governed by the compressible Euler equations and the ‘stiffened gas’ equation of state. It is an extension of the well-known ideal-gas steady nozzle flow solution. Various two-phase flow models are then addressed, all corresponding to limit situations of partial equilibrium among the phases. The first limit situation corresponds to the two-phase flow model of Kapila et al. (Phys. Fluids, vol. 13, 2001, pp. 3002–3024), where both phases evolve in mechanical equilibrium only. This model contains two entropies, two temperatures and non-conventional shock relations. The second one corresponds to a two-phase model where the phases evolve in both mechanical and thermal equilibrium. The last one corresponds to a model describing a liquid–vapour mixture in thermodynamic equilibrium. They all correspond to two-phase mixtures where the various relaxation effects are either stiff or absent. In all instances, the various flow regimes (subsonic, subsonic–supersonic, and supersonic with shock) are unambiguously determined, as well as various nozzle solution profiles.

A depiction of the mean and fluctuating vorticity structure in turbulent wall flows is presented and described within the context of the self-similar properties admitted by the mean dynamical equation. Data from a relatively wide range of numerical and physical experiments are used to explore and clarify the structure postulated. The mean vorticity indicator for the onset of the four-layer regime of the mean dynamics is revealed. With increasing Reynolds number, the mean vorticity is shown to segregate into two increasingly well-defined domains. Half of the mean vorticity concentrates into a near-wall region of width (relative to the overall flow width) that diminishes proportionally to the inverse square root of Reynolds number. The remainder of the mean vorticity is spread, with diminishing amplitude, over an outer domain that approaches the overall flow width at high Reynolds number. Vorticity stretching and reorientation are surmised to be the characteristic mechanisms accounting for the inner domain behaviour of both the mean and fluctuating vorticity. Vorticity dispersion via advective transport is surmised to be the characteristic mechanism in the outer domain. In this domain, the fluctuating enstrophy approaches that of the instantaneous enstrophy with increasing Reynolds number. This underpins an emerging self-similarity between the mean and r.m.s. vorticity in the domain where the mean velocity profile is logarithmic. The Reynolds number dependence of a number of properties associated with the vorticity field is explored and quantified. The study closes with brief account of the combined vortical and mean dynamical structure of turbulent wall flows.

Convection in a porous medium at high Rayleigh number $\mathit{Ra}$ exhibits a striking quasisteady columnar structure with a well-defined and $\mathit{Ra}$-dependent horizontal scale. The mechanism that controls this scale is not currently understood. Motivated by this problem, the stability of a density-driven ‘heat-exchanger’ flow in a porous medium is investigated. The dimensionless flow comprises interleaving columns of horizontal wavenumber $k$ and amplitude $\widehat{A}$ that are driven by a steady balance between vertical advection of a background linear density stratification and horizontal diffusion between the columns. Stability is governed by the parameter $A= \widehat{A}\mathit{Ra}/ k$. A Floquet analysis of the linear-stability problem in an unbounded two-dimensional domain shows that the flow is always unstable, and that the marginal-stability curve is independent of $A$. The growth rate of the most unstable mode scales with ${A}^{4/ 9} $ for $A\gg 1$, and the corresponding perturbation takes the form of vertically propagating pulses on the background columns. The physical mechanism behind the instability is investigated by an asymptotic analysis of the linear-stability problem. Direct numerical simulations show that nonlinear evolution of the instability ultimately results in a reduction of the horizontal wavenumber of the background flow. The results of the stability analysis are applied to the columnar flow in a porous Rayleigh–Bénard (Rayleigh–Darcy) cell at high $\mathit{Ra}$, and a balance of the time scales for growth and propagation suggests that the flow is unstable for horizontal wavenumbers $k$ greater than $k\sim {\mathit{Ra}}^{5/ 14} $ as $\mathit{Ra}\rightarrow \infty $. This stability criterion is consistent with hitherto unexplained numerical measurements of $k$ in a Rayleigh–Darcy cell.

Kalliadasis & Chang (J. Fluid Mech., vol. 261, 1994, pp. 135–168) showed within lubrication theory that if a liquid film is thicker than a critical value then drops will accelerate and grow, whereas with thinner films the drops fall at a constant velocity. As the thickness of the film increases to the critical value, the drops move faster and are larger. We revisit their asymptotic analysis of these large drops. While we recover their results for the leading-order and first correction, we do not agree on further corrections. In particular we find it necessary to evaluate the third correction, which they do not consider, before we obtain a first approximation to the dependence of the speed on the non-dimensional control parameter. We proceed to two further corrections in order to improve this first approximation to the speed.

We study the modal and non-modal linear instability of inertia-dominated channel flow of viscoelastic fluids modelled by the Oldroyd-B and FENE-P closures. The effects of polymer viscosity and relaxation time are considered for both fluids, with the additional parameter of the maximum possible extension for the FENE-P. We find that the parameter explaining the effect of the polymer on the instability is the ratio between the polymer relaxation time and the characteristic instability time scale (the frequency of a modal wave and the time over which the disturbance grows in the non-modal case). Destabilization of both modal and non-modal instability is observed when the polymer relaxation time is shorter than the instability time scale, whereas the flow is more stable in the opposite case. Analysis of the kinetic energy budget reveals that in both regimes the production of perturbation kinetic energy due to the work of the Reynolds stress against the mean shear is responsible for the observed effects where polymers act to alter the correlation between the streamwise and wall-normal velocity fluctuations. In the subcritical regime, the non-modal amplification of streamwise elongated structures is still the most dangerous disturbance-growth mechanism in the flow and this is slightly enhanced by the presence of polymers. However, viscoelastic effects are found to have a stabilizing effect on the amplification of oblique modes.

The physical properties of a recently proposed feedback-stabilization method of a vortex flow in a finite-length straight pipe are studied for the case of a solid-body rotation flow. In the natural case, when the swirl ratio is beyond a certain critical level, linearly unstable modes appear in sequence as the swirl level is increased. Based on an asymptotic long-wave (long-pipe) approach, the global feedback control method is shown to enforce the decay in time of the perturbation’s kinetic energy and thereby quench all of the instability modes for a swirl range above the critical swirl level. The effectiveness of an extended version of this feedback flow control approach is further analysed through a detailed mode analysis of the full linear control problem for a solid-body rotation flow in a finite-length pipe that is not necessarily long. We first rigourously prove the asymptotic decay in time of all modes with real growth rates. We then compute the growth rate and shape of all modes according to the full linearized control problem for swirl levels up to 50 % above the critical level. We demonstrate that the flow is stabilized in the whole swirl range and can be even further stabilized for higher swirl levels. However, the control effectiveness is sensitive to the choice of the feedback control gain. A potentially best range of the gain is identified. An inadequate level of gain, either insufficient or excessive, could lead to a marginal control or failure of the control method at high swirl levels. The robustness of the proposed control law to stabilize both initial waves and continuous inlet flow perturbations and the elimination of the vortex breakdown process are demonstrated through numerical computations.

We report on turbulent Rayleigh–Bénard convection of a nematic liquid crystal while it undergoes a transition from the nematic to the isotropic phase in a cylindrical convection cell with a height equal to twice the diameter (aspect ratio $\Gamma = 0. 50$). The difference between the top and bottom plate temperature $ \mathrm{\Delta} T= {T}_{b} - {T}_{t} $ was held constant, while the average temperature ${T}_{m} = ({T}_{b} + {T}_{t} )/ 2$ was varied. There was a significant increase of the transported heat when the phase transition temperature ${T}_{NI} $ was between ${T}_{b} $ and ${T}_{t} $. Measurements of the temperatures along the sidewall of the sample as a function of ${T}_{m} $ showed several ranges with qualitatively different behaviour of quantities such as the time-averaged sidewall temperature, temperature gradient, or temperature fluctuations. We interpret these different ranges in terms of properties of the thermal boundary layers close to the top and bottom plates whose stability and nature depends on the location within the sample of ${T}_{NI} $.

Scaling of the streamwise velocity spectrum ${\phi }_{11} ({k}_{1} )$ in the so-called sink-flow turbulent boundary layer is investigated in this work. The present experiments show strong evidence for the ${ k}_{1}^{- 1} $ scaling i.e. ${\phi }_{11} ({k}_{1} )= {A}_{1} { U}_{\tau }^{2} { k}_{1}^{- 1} $, where ${k}_{1} $ is the streamwise wavenumber and ${U}_{\tau } $ is the friction velocity. Interestingly, this ${ k}_{1}^{- 1} $ scaling is observed much farther from the wall and at much lower flow Reynolds number (both differing by almost an order of magnitude) than what the expectations from experiments on a zero-pressure-gradient turbulent boundary layer flow would suggest. Furthermore, the coefficient ${A}_{1} $ in the present sink-flow data is seen to be non-universal, i.e. ${A}_{1} $ varies with height from the wall; the scaling exponent −1 remains universal. Logarithmic variation of the so-called longitudinal structure function, which is the physical-space counterpart of spectral ${ k}_{1}^{- 1} $ scaling, is also seen to be non-universal, consistent with the non-universality of ${A}_{1} $. These observations are to be contrasted with the universal value of ${A}_{1} $ (along with the universal scaling exponent of −1) reported in the literature on zero-pressure-gradient turbulent boundary layers. Theoretical arguments based on dimensional analysis indicate that the presence of a streamwise pressure gradient in sink-flow turbulent boundary layers makes the coefficient ${A}_{1} $ non-universal while leaving the scaling exponent −1 unaffected. This effect of the pressure gradient on the streamwise spectra, as discussed in the present study (experiments as well as theory), is consistent with other recent studies in the literature that are focused on the structural aspects of turbulent boundary layer flows in pressure gradients (Harun et al., J. Fluid Mech., vol. 715, 2013, pp. 477–498); the present paper establishes the link between these two. The variability of ${A}_{1} $ accommodated in the present framework serves to clarify the ideas of universality of the ${ k}_{1}^{- 1} $ scaling.

Linearized boundary conditions are a commonplace numerical tool in any flow problems where the solid wall is nominally flat but the effects of small waviness or roughness are being investigated. Typical examples are stability problems in the presence of undulated walls or interfaces, and receptivity problems in aerodynamic transition prediction or turbulent flow control. However, to pose such problems properly, solutions in two mathematical distinguished limits have to be considered: a shallow-roughness limit, where not only roughness height but also its aspect ratio becomes smaller and smaller, and a small-roughness limit, where the size of the roughness tends to zero but its aspect ratio need not. Here a connection between the two solutions is established through an analysis of their far-field behaviour. As a result, the effect of the surface in the small-roughness limit, obtained from a numerical solution of the Stokes problem, can be recast as an equivalent shallow-roughness linearized boundary condition corrected by a suitable protrusion coefficient (related to the protrusion height used years ago in the study of riblets) and a proximity coefficient, accounting for the interference between multiple protrusions in a periodic array. Numerically computed plots and interpolation formulas of such correction coefficients are provided.

A model of hydrodynamic loads acting on a rigid floating body during the lifting of the body from the liquid surface is presented. The liquid is of infinite depth, inviscid and incompressible. Initially the liquid is at rest. The body suddenly starts to move upwards from the liquid at a constant acceleration. Boundary conditions on the liquid surface are linearized and imposed on the equilibrium position of the liquid surface. The resulting boundary problem is solved by the methods of analytical functions. Negative pressures are allowed and the pressure is assumed continuous at the periphery of the wetted area. The unknown size of the wetted area is determined by the condition that the speed of the contact points is proportional to the local velocity of the flow. This condition provides a nonlinear Abel-type integral equation which is solved explicitly. Both two-dimensional and axisymmetric configurations are considered. Predicted hydrodynamic forces are compared with the computational fluid dynamics results by Piro & Maki (11th International Conference on Fast Sea Transport. Honolulu, Hawaii, USA, 2011) for both a rigid wedge and circular cylinder, which initially enter the water and then exit from it.

We study the longitudinal linear optimal perturbations (which maximize the energy gain up to a prescribed time $T$) to inviscid parallel shear flow, which present unbounded energy growth due to the lift-up mechanism. Using the phase invariance with respect to time, we show that for an arbitrary base flow profile and optimization time, the computation of the optimal longitudinal perturbation reduces to the resolution of a single one-dimensional eigenvalue problem valid for all times. The optimal perturbation and its amplification are then derived from the lowest eigenvalue and its associated eigenfunction, while the remainder of the infinite set of eigenfunctions provides an orthogonal base for decomposing the evolution of arbitrary perturbations. With this new formulation we obtain, asymptotically for large spanwise wavenumber ${k}_{z} , $ a prediction of the optimal gain and the localization of inviscid optimal perturbations for the two main classes of parallel flows: free shear flow with an inflectional velocity profile, and wall-bounded flow with maximum shear at the wall. We show that the inviscid optimal perturbations are localized around the point of maximum shear in a region with a width scaling like ${ k}_{z}^{- 1/ 2} $ for free shear flow, and like ${ k}_{z}^{- 2/ 3} $ for wall-bounded shear flows. This new derivation uses the stationarity of the base flow to transform the optimization of initial conditions in phase space into the optimization of a temporal phase along each trajectory, and an optimization among all trajectories labelled by their intersection with a codimension-1 subspace. The optimization of the time phase directly imposes that the initial and final energy growth rates of the optimal perturbation should be equal. This result requires only time invariance of the base flow, and is therefore valid for any linear optimal perturbation problem with stationary base flow.

We study the flow forced by precession in rigid non-axisymmetric ellipsoidal containers. To do so, we revisit the inviscid and viscous analytical models that have been previously developed for the spheroidal geometry by, respectively, Poincaré (Bull. Astronomique, vol. XXVIII, 1910, pp. 1–36) and Busse (J. Fluid Mech., vol. 33, 1968, pp. 739–751), and we report the first numerical simulations of flows in such a geometry. In strong contrast with axisymmetric spheroids, where the forced flow is systematically stationary in the precessing frame, we show that the forced flow is unsteady and periodic. Comparisons of the numerical simulations with the proposed theoretical model show excellent agreement for both axisymmetric and non-axisymmetric containers. Finally, since the studied configuration corresponds to a tidally locked celestial body such as the Earth’s Moon, we use our model to investigate the challenging but planetary-relevant limit of very small Ekman numbers and the particular case of our Moon.

The present work provides an optimal control strategy, based on the nonlinear Navier–Stokes equations, aimed at hampering the rapid growth of unsteady finite-amplitude perturbations in a Blasius boundary-layer flow. A variational procedure is used to find the blowing and suction control law at the wall providing the maximum damping of the energy of a given perturbation at a given target time, with the final aim of leading the flow back to the laminar state. Two optimally growing finite-amplitude initial perturbations capable of leading very rapidly to transition have been used to initialize the flow. The nonlinear control procedure has been found able to drive such perturbations back to the laminar state, provided that the target time of the minimization and the region in which the blowing and suction is applied have been suitably chosen. On the other hand, an equivalent control procedure based on the linearized Navier–Stokes equations has been found much less effective, being not able to lead the flow to the laminar state when finite-amplitude disturbances are considered. Regions of strong sensitivity to blowing and suction have been also identified for the given initial perturbations: when the control is actuated in such regions, laminarization is also observed for a shorter extent of the actuation region. The nonlinear optimal blowing and suction law consists of alternating wall-normal velocity perturbations, which appear to modify the core flow structures by means of two distinct mechanisms: (i) a wall-normal velocity compensation at small times; (ii) a rotation-counterbalancing effect al larger times. Similar control laws have been observed for different target times, values of the cost parameter, and streamwise extents of the blowing and suction zone, meaning that these two mechanisms are robust features of the optimal control strategy, provided that the nonlinear effects are taken into account.

New flow-visualization experiments on mixing layers of various velocity and density ratios are reported. It is shown that, in mixing layers developing from laminar initial conditions, the familiar mechanism of growth by vortex amalgamation is replaced at the mixing transition by a previously unrecognized mechanism in which the spanwise-coherent large structures individually undergo continuous linear growth. In the organized post-transition flow it is this continuous linear growth of the individual structures that produces the self-similar growth of the mixing-layer thickness, with the occasional interactions between neighbouring structures occurring as a consequence of their growth, not its cause. It is also observed that periods during which the post-transition mixing layer comprises orderly processions of large structures alternate with periods during which no large-scale organization is apparent downstream of the transition location. These two fully turbulent flow states are characterized by different growth rates, entrainment ratios and orientations of the mixing layer relative to the free streams. The implications of these findings are discussed.