Streamwise oscillations of a circular cylinder in a steady uniform flow are investigated experimentally using a technique of high-image-density particle image velocimetry, in conjunction with instantaneous force measurements. This approach allows insight into the relationship between the loading and the patterns of vorticity and streamline topology in the near wake.

In analogy with the classical locked-on state arising from transverse oscillations of a cylinder in uniform flow, it is possible to attain either locked-on or quasi-locked-on states due to streamwise oscillations. In these cases, however, the repetitive signature of the transverse force is not sinusoidal; rather, it is strongly modulated and the corresponding spectra can exhibit several sharply defined peaks. The predominant peak can vary over a remarkably wide range, extending from the subharmonic to the third harmonic of the cylinder oscillation frequency; for certain locked-on states of the transverse force signature, the spectral peak at the cylinder oscillation frequency is actually suppressed. Corresponding instantaneous traces and spectra of the in-line force simply show dominance of the spectral peak at the cylinder oscillation frequency. Further interpretation of the loading is provided in terms of Lissajous patterns of the transverse and in-line force coefficients.

All of these features are related to the instantaneous patterns of vortex formation in the near wake. During a typical cycle of the cylinder oscillation, these patterns can be divided into two broad categories: Kármán-like shedding; and a nearly ‘frozen’ array of shed vortices. The order of occurrence of these basic patterns during an oscillation cycle dictates the instantaneous signatures and time-averaged spectra of the transverse force.

A cylinder in a steady current beneath a free surface is subjected to oscillations in the streamwise direction. Techniques of high-image-density particle image velocimetry and instantaneous force measurement provide the relationship between the instantaneous, global flow patterns and the unsteady loading on the cylinder.

The existence of locked-on states for the fully submerged cylinder is addressed in the companion study of Cetiner & Rockwell (2001). The present investigation shows that it is possible to generate distinctly different locked-on states of vortex formation, provided the cylinder is located immediately beneath the free surface. As a consequence, the time-dependent transverse force is phase-locked to the cylinder motion. In the event that a finite gap exists between the cylinder and the free surface, however, instantaneous, jet-like flow through the gap acts to destabilize such locked-on states. Lissajous representations of the forces demonstrate the degree of phase-locking or, in some cases, a loss of lock-on and associated phase drift. Moreover, the degree of submergence of the cylinder beneath the free surface has remarkable consequences for the magnitudes of positive and negative spikes of the time-dependent force signatures, as well as the averaged spectra of the transverse force. In turn, these alterations of the unsteady transverse force are accompanied by substantial changes of the averaged in-line and transverse forces.

Vortex systems can exist at locations both upstream and downstream of the cylinder. They are due to vorticity from the cylinder surface and/or the free surface. The space–time development of the entire system of vorticity concentrations is interpreted in terms of the time histories of the relative velocity of the cylinder and the instantaneous forces on the cylinder. In turn, these features of the vorticity field are related to critical points near the free surface, deduced from topologies of the corresponding velocity and streamline patterns. Despite the fact that changes in the patterns of vorticity and the corresponding topologies occur in conjunction with large fluctuations of the transverse force coefficient, the dimensionless strength of the vortices is below the threshold for which distinguishable, localized deformations of the free surface occur.

The formation of an axisymmetric vortex ring by forcing uid impulsively through a pipe is examined. An idealized model of the circulation, impulse and energy provided by the injected plug is developed, and these quantities are equated to the corresponding properties of the class of rings with finite cores described by Norbury (1973). It is shown that, as the length-to-diameter aspect ratio L/D of the plug increases, the size of the core increases in comparison with all the fluid carried along with the ring, until the limiting case of Hill's spherical vortex is reached. For aspect ratios larger than a certain value it is not possible to produce a single ring while conserving circulation, impulse, volume and energy. This implies that the limiting vortex is ‘optimal’ in the sense that it has maximum impulse, circulation and volume for a given energy input. While this matching calculation makes the physical mechanism clear, the L/D ratio that can be achieved in practice is more appropriately taken from the direct experimental measurements of Gharib et al. (1998) who concluded that the limiting value is L/D = 4. This is close to the value found in our calculation.

The impact of a spherical water drop onto a water surface has been studied experimentally with the aid of a 35 mm drum camera giving high-resolution images that provided qualitative and quantitative data on the phenomena. Scaling laws for the time to reach maximum cavity sizes have been derived and provide a good fit to the experimental results. Transitions between the regimes for coalescence-only, the formation of a high-speed jet and bubble entrapment have been delineated. The high-speed jet was found to occur without bubble entrapment. This was caused by the rapid retraction of the trough formed by a capillary wave converging to the centre of the cavity base. The converging capillary wave has a profile similar to a Crapper wave. A plot showing the different regimes of cavity and impact drop behaviour in the Weber–Froude number-plane has been constructed for Fr and We less than 1000.

Evolution of a nonlinear wave field along a laboratory tank is studied experimentally and numerically. The numerical study is based on the Zakharov nonlinear equation, which is modified to describe slow spatial evolution of unidirectional waves as they move along the tank. Groups with various initial shapes, amplitudes and spectral contents are studied. It is demonstrated that the applied theoretical model, which does not impose any constraints on the spectral width, is capable of describing accurately, both qualitatively and quantitatively, the slow spatial variation of the group envelopes. The theoretical model also describes accurately the variation along the tank of the spectral shapes, including free wave components and the bound waves.

The velocity and pressure fields produced in a homogeneous rapidly rotating fluid driven by an isolated buoyant parcel are investigated. Gravity and rotation are allowed to have arbitrary orientations and the parcel shape is assumed Gaussian. Inertial forces and time-dependent effects are ignored. The linear problem is easily solved by three-dimensional Fourier transform, and the inversion is facilitated by assuming the Ekman number, E, to be very small. In this limit the fields form a Taylor column extended in the direction of the rotation axis. In the absence of rigid boundaries no boundary layers occur. The velocity and pressure in the vicinity of the parcel are found in closed form while elsewhere (within the Taylor column) they are expressed in terms of relatively simple scalar integrals which are easily evaluated.

Within the buoyant parcel, the momentum balance is baroclinic, involving Coriolis, pressure and buoyancy forces. Outside the parcel, the balance is geostrophic at unit order. The viscous force is important at order E and determines the axial structure of the Taylor column. In contrast to the case of flow driven by a rigid body, no ‘Taylor slug’ of recirculating flow occurs. The velocity and pressure decay algebraically with distance from the parcel, with the scale of variation being a/E in the axial direction and a in the radial direction, where a is the parcel radius. In the vicinity of the parcel, the return flow occurs in a broad region surrounding the parcel. The structure of flow in the vicinity of the parcel is independent of the Ekman number. This return flow sweeps the fringes of the parcel backward, making the net rise speed significantly slower than that of a rigid sphere of identical buoyancy. The return flow also acts to deform the parcel; this deformation is quantified.

Quasi-one-dimensional cavitating nozzle flows are considered by employing a homogeneous bubbly liquid flow model. The nonlinear dynamics of cavitating bubbles is described by a modified Rayleigh–Plesset equation that takes into account bubble/bubble interactions by a local homogeneous mean-field theory and the various damping mechanisms by a damping coefficient, lumping them together in the form of viscous dissipation. The resulting system of quasi-one-dimensional cavitating nozzle flow equations is then uncoupled leading to a nonlinear third-order ordinary differential equation for the flow speed. This equation is then cast into a nonlinear dynamical system of scaled variables which describe deviations of the flow field from its corresponding incompressible single-phase value. The solution of the initial-value problem of this dynamical system can be carried out very accurately, leading to an exact description of the hydrodynamic field for the model considered.

A bubbly liquid composed of water vapour–air bubbles in water at 20 °C for two different area variations is considered, and the initial cavitation number is chosen in such a way that cavitation can occur in the nozzle. Results obtained, when bubble/bubble interactions are neglected, show solutions with flow instabilities, similar to the flashing flow solutions found recently by Wang and Brennen. Stable steady-state cavitating nozzle flow solutions, either with continuous growth of bubbles or with growth followed by collapse of bubbles, were obtained when bubble/bubble interactions were considered together with various damping mechanisms.

The buoyancy range, which represents a transition from large-scale wave-dominated motions to small-scale turbulence in the oceans and the atmosphere, is investigated through large-eddy simulations. The model presented here uses a continual forcing based on large-scale standing internal waves and has a spectral truncation in the isotropic inertial range. Evidence is presented for a break in the energy spectra from the anisotropic k−3 buoyancy range to the small-scale k−5/3 isotropic inertial range. Density structures that form during wave breaking and periods of high strain rate are analysed. Elongated vertical structures produced during periods of strong straining motion are found to collapse in the subsequent vertically compressional phase of the strain resulting in a zone or patch of mixed fluid.

A study of the Lagrangian statistical properties of velocity and passive scalar fields using direct numerical simulations is presented, for the case of stationary isotropic turbulence with uniform mean scalar gradients. Data at higher grid resolutions (up to 5123 and Taylor-scale Reynolds number 234) allow an update of previous velocity results at lower Reynolds number, including intermittency and dimensionality effects on vorticity time scales. The emphasis is on Lagrangian scalar time series which are new to the literature and important for stochastic mixing models. The variance of the ‘total’ Lagrangian scalar value (ϕ˜+, combining contributions from both mean and fluctuations) grows with time, with the velocity–scalar cross-correlation function and fluid particle displacements playing major roles. The Lagrangian increment of ϕ˜+ conditioned upon velocity and scalar fluctuations is well represented by a linear regression model whose parameters depend on both Reynolds number and Schmidt number. The Lagrangian scalar fluctuation is non-Markovian and has a longer time scale than the velocity, which is due to the strong role of advective transport, and is in contrast to results in an Eulerian frame where the scalars have shorter time scales. The scalar dissipation is highly intermittent and becomes de-correlated in time more rapidly than the energy dissipation. Differential diffusion for scalars with Schmidt numbers between 1/8 and 1 is characterized by asymmetry in the two-scalar cross-correlation function, a shorter time scale for the difference between two scalars, as well as a systematic decrease in the Lagrangian coherency spectrum up to at least the Kolmogorov frequency. These observations are consistent with recent work suggesting that differential diffusion remains important in the small scales at high Reynolds number.

The transient adjustment process of a compressible fluid in a rapidly rotating pipe is studied. The system Ekman number E is small, and the assumptions of small Mach number and the heavy-gas limit (γ = 1.0) are invoked. Fluid motion is generated by imposing a step-change perturbation in the temperature at the pipe wall Tw. Comprehensive analytical solutions are obtained by deploying the matched asymptotic technique with proper timescales O(E−1/2) and O(E−1). These analytical solutions are shown to be consistent with corresponding full numerical solutions. The detailed profiles of major variables are delineated, and evolution of velocity and temperature fields is portrayed. At moderate times, the entire flow field can be divided into two regions. In the inner inviscid region, thermo-acoustic compression takes place, and the process is isothermal–isentropic with the angular momentum being conserved. In the outer viscous region, diffusion of angular momentum occurs. The principal dynamic mechanisms are discussed, and physical rationalizations are offered. The essential differences between the responses of a compressible and an incompressible fluid are highlighted.

The issue of stability of the analytically obtained flow is addressed by undertaking a formal stability analysis. It is illustrated that, within the range of parameters of present concern, the flow is stable when ε ∼ O(E).

A new theoretical approach for turbulent flows based on Lie-group analysis is presented. It unifies a large set of ‘solutions’ for the mean velocity of stationary parallel turbulent shear flows. These results are not solutions in the classical sense but instead are defined by the maximum number of possible symmetries, only restricted by the flow geometry and other external constraints. The approach is derived from the Reynolds-averaged Navier–Stokes equations, the fluctuation equations, and the velocity product equations, which are the dyad product of the velocity fluctuations with the equations for the velocity fluctuations. The results include the logarithmic law of the wall, an algebraic law, the viscous sublayer, the linear region in the centre of a Couette flow and in the centre of a rotating channel flow, and a new exponential mean velocity profile not previously reported that is found in the mid-wake region of high Reynolds number flat-plate boundary layers. The algebraic scaling law is confirmed in both the centre and the near-wall regions in both experimental and DNS data of turbulent channel flows. In the case of the logarithmic law of the wall, the scaling with the distance from the wall arises as a result of the analysis and has not been assumed in the derivation. All solutions are consistent with the similarity of the velocity product equations to arbitrary order. A method to derive the mean velocity profiles directly from the two-point correlation equations is shown.

This paper presents a combined experimental/computational study of a surface cavity in a low Reynolds number Mach 9 flow. The geometry is based on a body of revolution, which produces highly two-dimensional time-averaged flow for all experimental test cases. A range of cavity length-to-depth ratios, up to a maximum of 8, is investigated. These correspond to ‘closed’ cavity flows, with the free shear layer bridging the entire cavity. For most cases the free shear layer is laminar. However, there is evidence of three-dimensional unsteadiness which is believed to be the consequence of Taylor–Görtler-type vortex formation. The effect of this is first detected on the cavity floor but progressively spreads as the cavity length is increased. For the longest cavities the flow is also influenced by the early stages of laminar–turbulent transition in the free shear layer.

In experiments, Plesniak, Mehta & Johnson (1994) have noted that curved two-stream mixing layers are susceptible to centrifugal instabilities under the condition that the slower of the streams curves towards the faster one; this condition is analogous to the concave curvature condition for the stability of the flow over a plate. The modes which arise manifest themselves as vortices aligned with the dominant flow direction. Previous numerical and analytical work has elucidated the structure of these vortices within incompressible mixing layers; Otto, Jackson & Hu (1996). In this paper we go on to investigate the rôles of compressibility and heating in determining the streamwise fate of Görtler vortices within these situations.

The development of the disturbances is monitored downstream and curves of neutral stability are plotted. The effect of changing the Mach number and free-stream temperatures is studied in detail. It is found that for certain parameter régimes modes can occur within convexly curved, or ‘stable’ mixing layers; these ‘thermal modes’ have no counterpart within incompressible mixing layers. By making use of a large Görtler number analysis we are able to verify our numerical results, and derive a very simple condition which yields information about the parameter ranges for which certain modes are likely to occur. As an aside this method can be used to show that no degree of wall cooling will allow sustained growth of Görtler vortices within boundary layers over convex plates.

An analysis is presented for the steady, two-dimensional, free convection around line sources of heat and heated cylinders in unbounded saturated porous media. It is extended to account also for the effects of forced convection. The study is based on the Boussinesq equations, with the velocities calculated using Darcy's law.

The analysis begins with the non-dimensional formulation and numerical solution of the problem of pure free convection around a line source of heat. When this analysis is extended to include the effects of forced convection, two parameters appear in the non-dimensional formulation: the non-dimensional value, V∞, of the free-stream velocity and its angle γ of inclination with respect to the vertical. We first describe the asymptotic form of the solution for large and small values of the distance to the source. The far-field description, which is also applicable to the flow around heated cylinders, is needed to facilitate the numerical solution of the problem. It includes a thermal wake, aligned with the free stream, and an outer irrotational flow with a sink and a vortex at the line source. The temperature distribution near the source involves a constant A0(V∞, γ), to be calculated with the numerical solution of the complete problem, which is used in the evaluation of the heat transfer from heated cylinders when the Rayleigh and Péclet numbers are small compared with unity. In this case we find an inner region where heat conduction is dominant, and an outer region where the cylinder appears as a line source of heat. The asymptotic analysis is complemented with the numerical solution of the general problem for circular cylinders with a wide range of Rayleigh numbers and some representative values of V∞ and γ. We give correlations for the Nusselt number in the limiting cases of pure free convection and pure forced convection.