The wakes of bluff objects and in particular of circular cylinders are known to undergo a ‘fast’ transition, from a laminar two-dimensional state at Reynolds number 200 to a turbulent state at Reynolds number 400. The process has been documented in several experimental investigations, but the underlying physical mechanisms have remained largely unknown so far. In this paper, the transition process is investigated numerically, through direct simulation of the Navier—Stokes equations at representative Reynolds numbers, up to 500. A high-order time-accurate, mixed spectral/spectral element technique is used. It is shown that the wake first becomes three-dimensional, as a result of a secondary instability of the two-dimensional vortex street. This secondary instability appears at a Reynolds number close to 200. For slightly supercritical Reynolds numbers, a harmonic state develops, in which the flow oscillates at its fundamental frequency (Strouhal number) around a spanwise modulated time-average flow. In the near wake the modulation wavelength of the time-average flow is half of the spanwise wavelength of the perturbation flow, consistently with linear instability theory. The vortex filaments have a spanwise wavy shape in the near wake, and form rib-like structures further downstream. At higher Reynolds numbers the three-dimensional flow oscillation undergoes a period-doubling bifurcation, in which the flow alternates between two different states. Phase-space analysis of the flow shows that the basic limit cycle has branched into two connected limit cycles. In physical space the period doubling appears as the shedding of two distinct types of vortex filaments.

Further increases of the Reynolds number result in a cascade of period-doubling bifurcations, which create a chaotic state in the flow at a Reynolds number of about 500. The flow is characterized by broadband power spectra, and the appearance of intermittent phenomena. It is concluded that the wake undergoes transition to turbulence following the period-doubling route.

A cylinder having mild variations in diameter along its span is subjected to controlled excitation at frequencies above and below the inherent shedding frequency from the corresponding two-dimensional cylinder. The response of the near wake is characterized in terms of timeline visualization and velocity traces, spectra, and phase plane representations. It is possible to generate several types of vortex formation, depending upon the excitation frequency. Globally locked-in, three-dimensional vortex formation can occur along the entire span of the flow. Regions of locally locked-in and period-doubled vortex formation can exist along different portions of the span provided the excitation frequency is properly tuned. Unlike the classical subharmonic instability in free shear flows, the occurrence of period-doubled vortex formation does not involve vortex coalescence; instead, the flow structure alternates between two different states.

Rapid-distortion theory is adapted to introduce a truly unsteady closure into a simple phenomenological turbulence model in order to describe the unsteady response of a turbulent wall layer exposed to a temporarily oscillating pressure gradient. The closure model is built by taking the ratio of turbulent shear stress to turbulent kinetic energy to be a function of the effective strain. The latter accounts for the history of the flow. The computed unsteady velocity fluctuations and modulated turbulent stresses compare favourably in the ‘non-quasi-steady’ frequency range, where quasi-steady assumptions would fail. This suggests that the concept of rapid distortion is especially appropriate for unsteady flows. This paper forms the basis for acoustical studies of the problem to be reported elsewhere.

The flow over a backward-facing step with laminar separation was investigated experimentally under controlled perturbation for a Reynolds number of 11000, based on a step height h and a free-stream velocity UO. The reattaching shear layer was found to have two distinct modes of instability: the ‘shear layer mode’ of instability at Stθ ≈ 0.012 (Stθ ≡ fθ/UO, θ being the momentum thickness at separation and f the natural roll-up frequency of the shear layer); and the ‘step mode’ of instability at Sth ≈ 0.185 (Sth ≡ fh/U0). The shear layer instability frequency reduced to the step mode one via one or more stages of a vortex merging process. The perturbation increased the shear layer growth rate and the turbulence intensity and decreased the reattachment length compared to the unperturbed flow. Cross-stream measurements of the amplitudes of the perturbed frequency and its harmonics suggested the splitting of the shear layer. Flow visualization confirmed the shear layer splitting and showed the existence of a low-frequency flapping of the shear layer.

In this paper we present experiments and an analysis of interfacial waves in core—annular flow; these waves are important for the flow to be stable. The observed wave velocity is about equal to the speed of the fluids near the interface, and the wavelength is 1–10 times the thickness of the annulus. These results are predicted by our analysis, which is valid provided the Reynolds number of the fluid in the annulus, and the ratio of the viscosities of the fluids in the core and the annulus, are large. The theory gives the growth rate of a wave as a function of this ratio, the Reynolds number, the surface tension and the wavenumber. For parameter values of interest, the growth rate is positive for a range wavenumbers which we compare with the experiments. Qualitative agreement between theory and experiment is excellent; quantitative comparison reveals discrepancies for which a possible explanation is the neglect of nonlinear terms.

Using Stokes flow between eccentric, counter-rotating cylinders as a prototype for bounded, nearly parallel lubrication flow, we investigate the effect of a slender recirculation region within the flow field on cross-stream heat or mass transport in the important limit of high Péclet number Pe where the enhancement over pure conduction heat transfer without recirculation is most pronounced. The steady enhancement is estimated with a matched asymptotic expansion to resolve the diffusive boundary layers at the separatrices which bound the recirculation region. The enhancement over pure conduction is shown to vary as ε½ at infinite Pe, where ε½ is the characteristic width of the recirculation region. The enhancement decays from this asymptote as Pe−½. If one perturbs the steady flow by a time-periodic forcing, fast relative to the convective and diffusive times, the separatrices undergo a homoclinic entanglement which allows fluid elements to cross the separatrices. We establish the existence of this homoclinic entanglement and show that the resulting chaotic particle transport further enhances the cross-stream flux. We estimate the penetration of the fluid elements across the separatrices and their effective diffusivity due to this chaotic transport by a Melnikov analysis for small-amplitude forcing. These and the steady results then provide quantitative estimates of the timeaveraged transport enhancement and allow optimization with respect to system parameters. An optimum forcing frequency which induces maximum heat-transfer enhancement is predicted and numerically verified. The predicted optimum frequency remains valid at strong forcing and large Pe where chaotic transport is as important as the recirculation mechanism. Since most heat and mass transport devices operate at high Pe, our analysis suggests that chaotic enhancement can improve their performance and that a small amplitude theory can be used to optimize its application.

New subgrid-scale models for the large-eddy simulation of compressible turbulent flows are developed and tested based on the Favre-filtered equations of motion for an ideal gas. A compressible generalization of the linear combination of the Smagorinsky model and scale-similarity model, in terms of Favre-filtered fields, is obtained for the subgrid-scale stress tensor. An analogous thermal linear combination model is also developed for the subgrid-scale heat flux vector. The two dimensionless constants associated with these subgrid-scale models are obtained by correlating with the results of direct numerical simulations of compressible isotropic turbulence performed on a 963 grid using Fourier collocation methods. Extensive comparisons between the direct and modelled subgrid-scale fields are provided in order to validate the models. A large-eddy simulation of the decay of compressible isotropic turbulence – conducted on a coarse 323 grid – is shown to yield results that are in excellent agreement with the fine-grid direct simulation. Future applications of these compressible subgrid-scale models to the large-eddy simulation of more complex supersonic flows are discussed briefly.

The behaviour of shear-generated interfacial waves in a cocurrent gas-liquid flow in a small rectangular channel is studied experimentally at conditions close to neutral stability. It is found that the linearly most unstable mode, which typically has a frequency of 8–10 Hz and a wavelength 1–4 cm, grows initially — followed immediately by the first overtone. Measurements of the bicoherence spectrum indicate that the overtone and fundamental are coherent in phase, which suggests that energy is transferred from the fundamental to the linearly stable first overtone. This energy transfer mechanism can stabilize the system, as evidenced by data, which shows that the first mode can saturate with a wave slope small as as 0.005. Theory based on weakly nonlinear mode-interaction equations suggests that this steady state should be stable at conditions close to neutral stability where only overtone modes are present. However, under more severe conditions, where the amplitude of the fundamental mode becomes sufficiently large, a subharmonic mode may be excited. The generation of the subharmonic, when it is linearly stable with respect to the flat film base state, can be interpreted as a linear instability of the steady state containing the fundamental and overtones. Modes that are sidebands (with wavenumbers = k ± δk) to the main peak may also occur. These can participate in interactions with low-frequency modes (i.e. δk) and thereby transfer energy to frequencies much below the fundamental. It is expected that all of these interactions play important roles in determining the wave spectrum of conditions far away from neutral stability.

Rotational flow through narrow axial channels is considered in connection with a proposed technique to sort and separate particles according to sedimentation velocities. Nonlinear and linear axisymmetric flow through two channels connected by a slot in the vertical wall is studied numerically. A linearized formulation for the three-dimensional flow through a circumferentially blocked channel, with arbitrary positioning of the inlets and outlets, is examined analytically. Both approaches indicate that to have a sharp criteria for fractionation, the vertical shear layers on the channel walls must overlap. Otherwise, Coriolis effects, accompanying a strong azimuthal motion, make the sorting less precise. Results of an exploratory experiment with a simple two-stage machine demonstrate the feasibility of the basic process for simultaneous and continuous separation and fractionation.

A study of the electrophoretic motion of a chain of colloidal spheres along the line through their centres is presented. The spheres may differ in radius and in zeta potential and they are allowed to be unequally spaced. Also, the spheres can be either freely suspended in the fluid or linked by infinitesimally thin rods with arbitrary lengths. The fluid can contain an arbitrary combination of general electrolytes. Although the thin-double-layer assumption is employed, the polarization effect of the mobile ions in the diffuse layer is taken into account. A slip velocity of fluid and normal fluxes of ions at the outer edge of the double layer can be derived and used as the boundary conditions for the fluid domain outside the thin double layer. Using a collocation technique along with these boundary conditions, a set of electrokinetic equations governing this problem is solved in the quasi-steady state and the particle interaction effects are computed for various cases. The most important discovery is that a group of particles with the same zeta potential will interact with one another, unlike the no-interaction results obtained in previous investigations assuming that the double layer is infinitesimally thin. For most situations, the particle interaction among the spheres is a complicated function of the properties of the spheres and ions. Also, it no longer varies monotonically with the extent of separation for some cases. The phenomena cannot be predicted systematically by a simple general rule.

The orientations of fibres in a semi-dilute, index-of-refraction-matched suspension in a Newtonian fluid were observed in a cylindrical Couette device. Even at the highest concentration (nL3 = 45), the particles rotated around the vorticity axis, spending most of their time nearly aligned in the flow direction as they would do in a Jeffery orbit. The measured orbit-constant distributions were quite different from the dilute orbit-constant distributions measured by Anczurowski & Mason (1967b) and were described well by an anisotropic, weak rotary diffusion. The measured ϕ-distributions were found to be similar to Jeffery's solution. Here, ϕ is the meridian angle in the flow-gradient plane. The shear viscosities measured by Bibbo (1987) compared well with the values predicted by Shaqfeh & Fredrickson's theory (1990) using moments of the orientation distribution measured here.

A class of ‘exact’ steady and unsteady solutions of the Navier—Stokes equations in cylindrical polar coordinates is given. The flows correspond to the motion induced by an infinite disc rotating in the (x, y)-plane with constant angular velocity about the z-axis in a fluid occupying a semi-infinite region which, at large distances from the disc, has velocity field proportional to (x, — y,O) with respect to a Cartesian coordinate system. It is shown that when the rate of rotation is large Kármán's exact solution for a disc rotating in an otherwise motionless fluid is recovered. In the limit of zero rotation rate a particular form of Howarth's exact solution for three-dimensional stagnation-point flow is obtained. The unsteady form of the partial differential system describing this class of flow may be generalized to time-periodic flows. In addition the unsteady equations are shown to describe a strongly nonlinear instability of Kármán's rotating disc flow. It is shown that sufficiently large perturbations lead to a finite-time breakdown of that flow whilst smaller disturbances decay to zero. If the stagnation point flow at infinity is sufficiently strong the steady basic states become linearly unstable. In fact there is then a continuous spectrum of unstable eigenvalues of the stability equations but, if the initial-value problem is considered, it is found that, at large values of time, the continuous spectrum leads to a velocity field growing exponentially in time with an amplitude decaying algebraically in time.

Explicit or implicit filtered representations of chaotic fields like spectral cut-offs or numerical discretizations are commonly used in the study of turbulence and particularly in the so-called large-eddy simulations. Peculiar to these representations is that they are produced by different filtering operators at different levels of resolution, and they can be hierarchically organized in terms of a characteristic parameter like a grid length or a spectral truncation mode. Unfortunately, in the case of a general implicit or explicit filtering operator the Reynolds rules of the mean are no longer valid, and the classical analysis of the turbulence in terms of mean values and fluctuations is not so simple.

In this paper a new operatorial approach to the study of turbulence based on the general algebraic properties of the filtered representations of a turbulence field at different levels is presented. The main results of this analysis are the averaging invariance of the filtered Navier—Stokes equations in terms of the generalized central moments, and an algebraic identity that relates the turbulent stresses at different levels. The statistical approach uses the idea of a decomposition in mean values and fluctuations, and the original turbulent field is seen as the sum of different contributions. On the other hand this operatorial approach is based on the comparison of different representations of the turbulent field at different levels, and, in the opinion of the author, it is particularly fitted to study the similarity between the turbulence at different filtering levels. The best field of application of this approach is the numerical large-eddy simulation of turbulent flows where the large scale of the turbulent field is captured and the residual small scale is modelled. It is natural to define and to extract from the resolved field the resolved turbulence and to use the information that it contains to adapt the subgrid model to the real turbulent field. Following these ideas the application of this approach to the large-eddy simulation of the turbulent flow has been produced (Germano et al. 1991). It consists in a dynamic subgrid-scale eddy viscosity model that samples the resolved scale and uses this information to adjust locally the Smagorinsky constant to the local turbulence.

A hydroelastic model describing the mechanics of the human semicircular canal is presented that extends previous work to address the influence of the shape of the labyrinth and the interaction between the endolymph and the cupula. The analysis is based extensively on the three-dimensional geometry and structure of the system and exploits the slender toroidal geometry to obtain an asymptotic solution describing the velocity distribution of the endolymph, the pressure distribution and the deflection of the cupula. All parameters appearing in the model are explicitly defined in terms of physical properties and the geometry. Results for the structure of an infant human endolymphatic canal agree well with experimental measurements of the end-organ velocity gain and phase over the entire physiological range of angular head frequencies. From 0.09 to 1.5 Hz the mechanical response relative to head velocity is essentially constant and the end-organ acts as an angular velocity transducer. Below 0.09 Hz the velocity gain is diminished and above 1.5 Hz the velocity gain is enhanced. For a 1 rad sinusoidal rotation of the head, the analysis predicts an average cupula displacement for the infant canal of approximately 8 × 10−5 cm at 0.09 Hz and 2 × 10−3 cm at 2.0 Hz.

We consider the behaviour of an internal gravity wave encountering a critical level in a stratified fluid, assuming the critical-level flow to be dominated by nonlinear effects. The background flow is a shear layer, and the stratification is sufficiently strong to support wave propagation everywhere. Incident and reflected waves are permitted below the critical level, and a radiation condition is imposed far above it. For this geometry we construct, by a combination of asymptotic and numerical means, steady, nonlinear solutions, and discuss the associated transmission coefficients, reflection coefficients, phase shifts, and resonance positions when the system is forced from below.

The inviscid solutions we exhibit have continuous density and velocity everywhere, and so do not require the introduction of internal viscous boundary layers. Further, the streamlines bounding the recirculating cat's-eye regions have corners, just as in the unstratified case. For weak stratification, the transmitted wave is nearly as strong as the incident wave, and there is accompanying strong over-reflection. As the stratification increases, the critical level becomes a nearly perfect reflector. The amount of transmission depends on wave amplitude, and the sensitivity increases with increasing stratification.

There are regions of parameter space for which steady solutions could not be found. The critical-layer structure appears to break down by unbounded thickening when the stratification becomes too strong, suggesting that in these cases some neglected physical process must intervene to limit growth of the recirculating region.

Models for the transport of passive scalar in turbulent flow were investigated using databases derived from numerical solutions of the Navier—Stokes equations for fully developed plane channel flow, these databases being generated using large-eddy and direct numerical simulation techniques. Their reliability has been established by comparison with the experimental measurements of Hishida. Nagano & Tagawa (1986). The present paper compares these simulations and calculations using the Nagano & Kim (1988) ‘two-equation’ model for the scalar variance (kθ) and scalar variance dissipation (εθ). This model accounts for the dependence of flow quantities on the Prandtl number by expressing eddy diffusivity in terms of the ratio of the timescales of velocity and scalar fluctuations. However, the statistical analysis by Yoshizawa (1988) showed that there was an inconsistency in the definition of the isotropic eddy diffusivity in the Nagano—Kim model, the implications of which are clearly demonstrated by the results of this paper where large-eddy simulation and direct numerical simulation (LES/DNS) databases are used to compute the quantities contained in both models. An extension of the Nagano-Kim model is proposed which resolves these inconsistencies, and a further development of this model is given in which the anisotropic scalar fluxes are calculated. Near a rigid surface, a third-order ‘anisotropic representation’ of scalar fluxes may be used as an alternative model for reducing the eddy diffusivity, instead of the conventional ‘damping functions’. This model is similar but distinct from the algebraic scalar flux model of Rogers, Mansour & Reynolds (1989). A third aspect of this paper is the use of the LES/DNS databases to evaluate certain coefficients (those for modelling the pressure-scalar gradient terms) of another model of a similar type, namely the algebraic scalar flux model of Launder (1975).

The modification of a flat-plate turbulent boundary layer resulting from the injection of drag-reducing polymer solutions through a narrow inclined slot into the near-wall region of the flow has been studied. Two-component coincident laser-Doppler velocity profile measurements were taken with a free-stream velocity of 4.5 m/s with polymer injection, water injection, and no injection. Polyethylene oxide solutions at concentrations of 500 and 1025 w.p.p.m. were injected. These data are complemented by polymer concentration profile measurements that were taken using a laser-induced-fluorescence technique. Also, integrated skin friction measurements were made with a drag balance for a range of polymer injection conditions and free-stream velocities. The immediate effects of polymer injection are a deceleration of the flow near the wall, a dramatic decrease of the vertical r.m.s. velocit}’ fluctuation levels and the Reynolds shear stress levels, and a mean velocity profile approaching Virk's asymptotic condition. These effects relax substantially with increasing stream wise distance from the injection slot and become similar to the effects observed for dilute homogeneous polymer flows.

Various difficulties can be expected in trying to extract from experimental data the distribution of singularities, the f(α) function, of the energy dissipation. One reason is that the multifractal model of turbulence implies a dependence of the viscous cutoff on the singularity exponent. It is an open question if current hot-wire probes can resolve the scales implied by exponents a significantly less than 1.

Two exactly soluble models are used to show how spurious scaling can occur, due to finite Reynolds number effects. In the Gaussian model the true velocity signal is replaced by independent Gaussian random variables. The dissipation, defined as the square of the difference of successive variables, has trivial scaling in so far as all the moments of spatial averages of the dissipation behave asymptotically as a uniform dissipation. Still, contamination by subdominant terms requires that scaling exponents for high-order moments be identified over an increasingly large range of scales. If the available range is limited by the Reynolds number, scaling exponents for high orders will be systematically underestimated and spurious intermittency will be inferred. Burgers’ model is used to highlight further problems. At finite Reynolds numbers, regions with no small-scale activity (away from shocks) have a residual dissipation which contributes a spurious point (α = 1,f(α) = 1). In addition, when the f(α) function is obtained by Legendre transform techniques, convex hull effects generate an entire spurious segment.

Finally, Burgers’ model also indicates that the relation between exponents of structure functions and exponents of local dissipation moments, which goes back to Kolmogorov's (1962) work, leads to an inconsistency for structure functions of low positive order.

Results are presented for a class of self-similar solutions of the steady, axisymmetric Navier–Stokes equations, representing the flows in slender (quasi-cylindrical) vortices. Effects of vortex strength, axial gradients and compressibility are studied. The presence of viscosity is shown to couple the parameters describing the core growth rate and the external flow field, and numerical solutions show that the presence of an axial pressure gradient has a strong effect on the axial flow in the core. For the viscous compressible vortex, near-zero densities and pressures and low temperatures are seen on the vortex axis as the strength of the vortex increases. Compressibility is also shown to have a significant influence upon the distribution of vorticity in the vortex core.

A systematic perturbation procedure, based on a small mean flow Mach number and large duct Reynolds number, is employed to formulate and solve an initial-boundary-value problem for acoustic processes in a shear flow contained within a rigid-walled parallel duct. The results describe the general transient evolution of acoustic waves driven by a plane source located at a given duct cross-section. Forced bulk oscillations near the source and oblique wave generation are shown to result from refraction of the basic planar axial disturbance by the shear flow. Refraction also causes the axial waves to exhibit higher-order amplitude variations in the transverse direction. As the source frequency approaches certain critical values, specific refraction-induced oblique waves evolve into amplifying purely transverse waves. As a result, the magnitude of the refraction effect increases with time, and quasi-steady solutions do not exist. The analysis is extended to the thin acoustic boundary layer adjacent to the solid walls to examine the shear-layer structure induced by the variety of acoustic waves in the core flow. Nonlinear effects and acoustic streaming are shown to be negligibly small on the scale of a few axial wavelengths.