The collision of two equi-sized drops immersed in an immiscible liquid phase undergoing a shear flow in a parallel plate apparatus has been investigated over a range of capillary numbers. The drops were observed along the vorticity direction of shear flow by video enhanced contrast optical microscopy. Images of the colliding drops were processed by image analysis techniques. The distance Δy between the drop centres along the velocity gradient direction was measured as a function of time during approach, collision and separation of the two drops. It was found that Δy increases irreversibly after collision, thus providing a mechanism for drop dispersion in a concentrated system. Drop shape evolution during collision was characterized by measuring a deformation parameter and the angle made by the drop major axis with respect to the velocity gradient direction. The extent of the near-contact region when the drops are sliding on each other was also estimated. Coalescence was a rare event and was observed in the extensional quadrant of the shear flow. The experimental results show good agreement with numerical simulations recently reported in the literature.

Nikolayeva & Tsimring's (1986) collisionless Boltzmann model for surface-wave generation by a slowly fluctuating wind U(z, t) is transformed to an equivalent steady flow in which the wind speed in the reference frame of the wave (of speed c) is given by V(z)=〈(U−c)−2〉 −1/2, where 〈 〉 signifies a Gaussian average. This leads to a Sturm–Liouville equation for the Gaussian-averaged, complex amplitude of the wave-induced pressure. The wind-to-wave energy transfer for a logarithmic wind profile with the mean friction velocity κŪ1 (κ=Kármán's constant), the standard deviation δŪ1, and the roughness length z0=ΩŪ21/g is determined as a function of the parameters δ and Ω (Charnock's constant) through numerical integration of a Riccati equation (derived from the Sturm–Liouville equation). The energy transfer exceeds that predicted by the quasi-laminar model (Miles 1957; Conte & Miles 1959) by as much as 20–30% for δ≈1 and c (wave speed)[lsim ]6Ū1 but is decreased for c[gsim ]8Ū1 and may be negative for sufficiently large c/Ū1. These predictions contrast with the order-of-magnitude increase predicted by Nikolayeva & Tsimring.

Numerical studies are performed addressing the development of regions of high curvature and the spontaneous occurrence of cusped interfacial shapes in two-dimensional and axisymmetric Stokes flow. In the numerical simulations, the velocity field is computed using a boundary-integral method, and the evolution of the concentration of an insoluble surfactant over an evolving interface is computed using an implicit finite-volume method. Three configurations are considered in detail, and the results are used to elucidate three different aspects of cusp formation. In the first series, the deformation of a two-dimensional bubble immersed in a family of straining flows devised by Antanovskii, and of an axisymmetric bubble immersed in an analogous family of flows devised by Sherwood, are examined. The numerical results indicate that highly elongated and cusped two-dimensional shapes, and pointed or cusped axisymmetric shapes, are unstable and should not be expected to occur in practice. In the second series of studies, the role of an insoluble surfactant on the transient deformation of bubbles subject to the Antanovskii or Sherwood flow is investigated. Under certain conditions, the reduced surface tension at the tips raises the local curvature to high values and causes the ejection of a sheet or column of gas by means of tip streaming. In the third series of studies, the coalescence of a polygonal formation of five viscous columns of a fluid placed in an arrangement that differs only slightly from one proposed recently by Richardson is examined. The numerical results confirm Richardson's predictions that transient cusps may occur at a finite time in the presence of surface tension. The underlying physical mechanism is discussed on the basis of reversibility of surface-driven Stokes flow and with reference to the regularity of the motion driven by negative surface tension. Replacing the inviscid ambient gas with a slightly viscous fluid whose viscosity is as low as one hundredth the viscosity of the cylinders suppresses the cusp formation.

The linear stability of turbulent shear flow over a film-covered sea surface is studied theoretically. A compound matrix method (Wheless & Csanady 1993), is used to solve the eigenvalue problem numerically. The numerical method has been adjusted to a coupled air–sea system. In the stability problem the vertical component of the turbulent Reynolds stress has been taken into account. As pointed out by Wheless & Csanady, the second derivative of the traditional log–linear wind profile has a rather extreme behaviour near the matching point of the linear and logarithmic part. To improve the model, a new profile is calculated based on an eddy viscosity distribution for channel flow (Quarmby & Anand 1969), which has continuous derivatives all the way down to the surface. Calculations of the wave growth rates corresponds well with earlier theoretical results as well as laboratory measurements. The energy flux from the air to the sea caused by the pressure work at the surface has been calculated. An intriguing result obtained here is that this flux seems to be strongly dependent on the elastic property of the surface film. The flux attains a maximum for finite values of the film elasticity parameter.

The mixing of the round jet normal to a uniform crossflow is studied for a range of jet-to-crossflow velocity ratios, r, from 5 to 25. Planar laser-induced fluorescence (PLIF) of acetone vapour seeded into the jet is used to acquire quantitative two-dimensional images of the scalar concentration field. Emphasis is placed on r=10 and r=20 and a few select images are acquired up to r=200. The Reynolds number based on the jet exit diameter, d, and the exit velocity varies from 8400 to 41 500. Images are acquired for conditions in which the product rd is held constant, requiring decreasing d for increasing r.

Results from this experimental study concern structural events of the vortex interaction region, and mixing and mean centreline concentration decay in the near and far fields. The results cover all three regions of the transverse jet, and suggest that the jet scales with three length scales: d, rd and r2d.

Events within the vortex interaction region display d-scaling, including the crossflow boundary layer separation and roll-up. Over the range of velocity ratios studied, the vortex interaction region shows r-dependent variations in the flow field, including the emergence of jet fluid in the wake structures for r>10 and a slower development of the counter-rotating vortex pair (CVP) in higher-r jets.

The trajectory and physical dimension of the jet in both the near and far field display rd-scaling. The near field is characterized by a centreline concentration decay along the centreline coordinate s of s−1.3, different from the decay rate (s−1) of the free jet. When normalized by rd, the decay of each velocity-ratio jet branches away from the s−1.3 decay, approaching a decay of s−2/3, a rate predicted by modelling efforts. The branch points represent a transition in the flow field from enhanced mixing to reduced mixing compared to the free jet. When normalized by r2d, the branch points occur at a uniform jet position, s/r2d=0.3, which is viewed to be the division between the near and far fields. Self-similarity is not seen in the near field, but may be present in the far field.

The view of the branch points as a place of transition in the flow is supported by the probability density function (p.d.f.) of concentration along the upper edge of the jet. Before the branch points, the p.d.f.s are non-marching in character, and after the branch points, they are tilted in character.

Instantaneously, the CVP is asymmetric in shape and concentration. End views reveal extensive motion of the CVP and plan views show this motion can occur in both axisymmetric and sinusoidal motion. Ensemble-averaged images show the jet concentration is asymmetric about the centreline plane.

The stability of the flow of a fluid in a flexible tube is analysed over a range of Reynolds numbers 1<Re<104 using a linear stability analysis. The system consists of a Hagen–Poiseuille flow of a Newtonian fluid of density ρ, viscosity η and maximum velocity V through a tube of radius R which is surrounded by an incompressible viscoelastic solid of density ρ, shear modulus G and viscosity ηs in the region R<r<HR. In the intermediate Reynolds number regime, the stability depends on the Reynolds number Re=ρVR/η, a dimensionless parameter [sum ]=ρGR2/η2, the ratio of viscosities ηr= ηs/η, the ratio of radii H and the wavenumber of the perturbations k. The neutral stability curves are obtained by numerical continuation using the analytical solutions obtained in the zero Reynolds number limit as the starting guess. For ηr=0, the flow becomes unstable when the Reynolds number exceeds a critical value Rec, and the critical Reynolds number increases with an increase in [sum ]. In the limit of high Reynolds number, it is found that Rec∝[sum ]α, where α varies between 0.7 and 0.75 for H between 1.1 and 10.0. An analysis of the flow structure indicates that the viscous stresses are confined to a boundary layer of thickness Re−1/3 for Re[Gt ]1, and the shear stress, scaled by ηV/R, increases as Re1/3. However, no simple scaling law is observed for the normal stress even at 103<Re<105, and consequently the critical Reynolds number also does not follow a simple scaling relation. The effect of variation of ηr on the stability is analysed, and it is found that a variation in ηr could qualitatively alter the stability characteristics. At relatively low values of [sum ] (about 102), the system could become unstable at all values of ηr, but at relatively high values of [sum ] (greater than about 104), an instability is observed only when the viscosity ratio is below a maximum value η*rm.

Probability density function (p.d.f.) methods are extended to include modelling of wall-bounded turbulent flows. A p.d.f. near-wall model is developed in which the generalized Langevin model is combined with a model for viscous transport. This provides exact treatment of viscous inhomogeneous effects, and enables consistent imposition of the no-slip condition in a particle framework. The method of elliptic relaxation is combined with additional boundary conditions and with the generalized Langevin model to provide an analogy for the near-wall fluctuating continuity equation. This provides adequate representation of the near-wall anisotropy of the Reynolds stresses. The model is implemented with a p.d.f./Monte Carlo simulation for the joint p.d.f. of velocity and turbulent frequency. Results are compared with DNS and experimental profiles for fully developed turbulent channel flow.

The relative velocity of two fluid particles in homogeneous and stationary turbulence is considered. Looking for reduced dynamics of turbulent dispersion, we apply the nonlinear Mori–Zwanzig projector method to the Navier–Stokes equations. The projector method decomposes the Lagrangian acceleration into a conditionally averaged part and a random force. The result is an exact generalized Langevin equation for the Lagrangian velocity differences accounting for the exact equation of the Eulerian probability density. From the generalized Langevin equation, we obtain a stochastic model of relative dispersion by stochastic estimation of conditional averages and by assuming the random force to be Gaussian white noise. This new approach to dispersion modelling generalizes and unifies stochastic models based on the well-mixed condition and the moments approximation. Furthermore, we incorporate viscous effects in a systematic way. At a moderate Reynolds number, the model agrees qualitatively with direct numerical simulations showing highly non-Gaussian separation and velocity statistics for particle pairs initially close together. At very large Reynolds numbers, the mean-square separation obeys a Richardson law with coefficient of the order of 0.1.

The stability of viscous shear is studied for flows that consist of predominantly linear shear, but contain localized regions over which the vorticity varies rapidly. Matched asymptotic expansion simplifies the governing equations for the dynamics of such ‘vorticity defects’. The normal modes satisfy explicit dispersion relations. Nyquist methods are used to find and classify the possible instabilities. The defect equations are analysed in the inviscid limit to establish the connection with inviscid theory. Finally, the defect approximation is used to study nonlinear stability using weakly nonlinear techniques, and the initial value problem using Laplace transforms.

This work extends the study of the structure of wall-bounded flows using the topological properties of eddying motions as developed by Chong et al. (1990), Soria et al. (1992, 1994), and as recently extended by Blackburn et al. (1996) and Chacin et al. (1996). In these works, regions of flow which are focal in nature are identified by being enclosed by an isosurface of a positive small value of the discriminant of the velocity gradient tensor. These regions resemble the attached vortex loops suggested first by Theodorsen (1955). Such loops are incorporated in the attached-eddy model versions of Perry & Chong (1982), Perry et al. (1986), and Perry & Marusic (1995), which are extensions of a model first formulated by Townsend (1976). The direct numerical simulation (DNS) data of wall-bounded flows studied here are from the zero-pressure-gradient flow of Spalart (1988) and the boundary layer with separation and reattachment of Na & Moin (1996). The flow structures are examined from the viewpoint of the attached eddy hypothesis.

An investigation is made into the dynamics involved in the movement of the contact line when a single liquid with an interface moves into a vacuum over a smooth solid surface. In order to remove the stress singularity at the contact line, it is postulated that slip between the liquid and the solid or some other mechanism occurs very close to the contact line. It is assumed that the flow produced is inertia dominated with the Reynolds number based on the slip length being very large. Following a procedure similar to that used by Cox (1986) for the viscous-dominated situation (in which the Reynolds number based on the macroscopic length scale was assumed very small) using matched asymptotic expansions, we obtain the dependence of the macroscopic dynamic contact angle on the contact line velocity over the solid surface for small capillary number and small slip length to macroscopic lengthscale ratio. These results for the inertia-dominated situation are then extended (at the lowest order in capillary number) to an intermediate Reynolds number situation with the Reynolds number based on the slip length being very small and that based on the macroscopic lengthscale being very large.

We study the shear-induced gradient diffusion of particles in an inhomogeneous dilute suspension of neutrally buoyant spherical particles undergoing a simple shearing motion, with all inertia and Brownian motion effects assumed negligible. An expansion is derived for the flux of particles due to a concentration gradient along the directions perpendicular to the ambient flow. This expression involves the average velocity of the particles, which in turn is expressed as an integral over contributions from all possible configurations. The integral is divergent when expressed in terms of three-particle interactions and must be renormalized. For the monolayer case, such a renormalization is achieved by imposing the condition of zero total macroscopic flux in the transverse direction whereas, for the three-dimensional case, the additional constraint of zero total macroscopic pressure gradient is required. Following the scheme of Wang, Mauri & Acrivos (1996), the renormalized integral is evaluated numerically for the case of a monolayer of particles, giving for the gradient diffusion coefficient 0.077γa2c¯2, where is the applied shear rate, a the radius of the spheres and c¯ their areal fraction.

The coupled radiation-diffraction problem due to a floating body with slow (time-dependent) rotation about the vertical axis in incoming waves is studied by means of potential theory. The water depth may be finite. First, the radiation problem is described. It is shown how the various components of the velocity potential may be obtained by means of integral equations. The first-order forces in the coupled radiation-diffraction problem are then considered. Generalized Haskind relations for the exciting forces and generalized Timman–Newman relations for the added mass and damping forces are deduced for bodies of arbitrary shape with vertical walls at the water line. The equation of motion is obtained, and the frequencies of the linear body responses superposed on the slow rotation are identified. Formulae for the wave-drift damping coefficients in the yaw mode of motion are derived in explicit form, and the energy equation is discussed. Computations illustrating the various aspects of the method are performed for two ships. The wave-drift damping moment is found to become positive in the present examples. When the rotation axis is moved far away from the body, the slow motion becomes effectively unidirectional, and results of the translational case are recovered.

We investigate the evolution of nearby like-sign vortices whose centres are at different vertical levels in a stably stratified rotating fluid. We employ two differently singularized representations of the potential vorticity distribution in the quasi-geostrophic equations (QG), in order to elucidate the pair-interaction behaviour previously seen in non-singular QG numerical solutions. The first is an analytically tractable conservative (Hamiltonian) elliptical-moment model (EM) for thin-core vortices, which exhibits a regime of very strong horizontal elongation of a vortex in response to the strain induced by its partner. We interpret this as an early evolutionary stage towards the irreversible dissipative merger and alignment interactions. This interpretation is strengthened by weakly dissipative numerical solutions of a thin-core contour-dynamics model (CD), which exhibit even further progress towards the completion of these vortex interactions in the same regime.

In the EM model we classify the co-rotating stationary states which exist always for vertically offset thin-core vortices. However, the mutual strain field among the vortices cannot be balanced by co-rotation in a weakly elongated stationary state for a certain class of neighbouring, but substantially non-aligned, vortex configurations, and our interpretive assumption is that such configurations will rapidly evolve in non-singular QG solutions towards a more aligned configuration through significantly non-conservative reorganizations of the potential vorticity field. Both the EM and CD models show qualitatively similar regime boundaries between evolutions with weakly and strongly deformed vortices. In particular, there is a fairly close correspondence between the occurrence of strong vortex elongation in the EM solutions and significant filamentation and splitting in the CD solutions.

The near- and far-field break-up and atomization of a water jet by a high-speed annular air jet are examined by means of high-speed flow visualizations and phase Doppler particle sizing techniques. Visualization of the jet's near field and measurements of the frequencies associated with the gas–liquid interfacial instabilities are used to study the underlying physical mechanisms involved in the primary break-up of the water jet. This process is shown to consist of the stripping of water sheets, or ligaments, which subsequently break into smaller lumps or drops. An entrainment model of the near-field stripping of the liquid is proposed, and shown to describe the measured liquid shedding frequencies. This simplified model explains qualitatively the dependence of the shedding frequency on the air/water momentum ratio in both initially laminar and turbulent water jets. The role of the secondary liquid break-up in the far-field atomization of the water jet is also investigated, and an attempt is made to apply the classical concepts of local isotropy to explain qualitatively the measurement of the far-field droplet size distribution and its dependence on the water to air mass and momentum ratios. Models accounting for the effect of the local turbulent dissipation rate in the gas on both the break-up and coalescence of the droplets are developed and compared with the measurements of the variation of the droplet size along the jet's centreline. The total flux of kinetic energy supplied by the gas per unit total mass of the spray jet was found to be the primary parameter determining the secondary break-up and coalescence of the droplets in the far field.