We consider the stretching of a thin viscous thread, whose viscosity depends on temperature, that is heated by a radiative heat source. The thread is fed into an apparatus at a fixed speed and stretched by imposing a higher pulling speed at a fixed downstream location. We show that thermal effects lead to the surprising result that steady states exist for which the force required to stretch the thread can decrease when the pulling speed is increased. By considering the nature of the solutions, we show that a simple physical mechanism underlies this counterintuitive behaviour. We study the stability of steady-state solutions and show that a complicated sequence of bifurcations can arise. In particular, both oscillatory and non-oscillatory instabilities can occur in small isolated windows of the imposed pulling speed.

Nitric oxide formation in gas turbine combustion depends on four key factors: flame stabilization, heat transfer, fuel–air mixing and combustion instability. The design of modern gas turbine burners requires delicate compromises between fuel efficiency, emissions of oxides of nitrogen (NOx) and combustion stability. Burner designs allowing substantial NOx reduction are often prone to combustion oscillations. These oscillations also change the NOx fields. Being able to predict not only the main species field in a burner but also the pollutant and the oscillation levels is now a major challenge for combustion modelling. This must include a realistic treatment of unsteady acoustic phenomena (which create instabilities) and also of heat transfer mechanisms (convection and radiation) which control NOx generation.

In this work, large-eddy simulation (LES) is applied to a realistic gas turbine combustion chamber configuration where pure methane is injected through multiple holes in a cone-shaped burner. In addition to a non-reactive simulation, this article presents three reactive simulations and compares them to experimental results. The first reactive simulation neglects effects of cooling air on flame stabilization and heat losses by radiation and convection. The second reactive simulation shows how cooling air and heat transfer affect nitric oxide emissions. Finally, the third reactive simulation shows the effects of combustion instability on nitric oxide emissions. Additionally, the combustion instability is analysed in detail, including the evaluation of the terms in the acoustic energy equation and the identification of the mechanism driving the oscillation.

Results confirm that LES of gas turbine combustion requires not only an accurate chemical scheme and realistic heat transfer models but also a proper description of the acoustics in order to predict nitric oxide emissions and pressure oscillation levels simultaneously.

Three different approximations to the axisymmetric small-disturbance dynamics of a uniformly rotating thin spherical shell are studied for the equatorial region assuming time-harmonic motion. The first is the standard β-plane model. The second is Stern's (Tellus, vol. 15, 1963, p. 246) homogeneous, equatorial β-plane model of inertial waves (that includes all Coriolis terms). The third is a version of Stern's equation extended to include uniform stratification. It is recalled that the boundary value problem (BVP) that governs the streamfunction of zonally symmetric waves in the meridional plane becomes separable only for special geometries. These separable BVPs allow us to make a connection between the streamfunction field and the underlying geometry of characteristics of the governing equation. In these cases characteristics are each seen to trace a purely periodic path. For most geometries, however, the BVP is non-separable and characteristics and therefore wave energy converge towards a limit cycle, referred to as an equatorial wave attractor. For Stern's model we compute exact solutions for wave attractor regimes. These solutions show that wave attractors correspond to singularities in the velocity field, indicating an infinite magnification of kinetic energy density along the attractor. The instability that arises occurs without the necessity of any ambient shear flow and is referred to as geometric instability.

For application to ocean and atmosphere, Stern's model is extended to include uniform stratification. Owing to the stratification, characteristics are trapped near the equator by turning surfaces. Characteristics approach either equatorial wave attractors, or point attractors situated at the intersections of turning surfaces and the bottom. At these locations, trapped inertia–gravity waves are perceived as near-inertial oscillations. It is shown that trapping of inertia–gravity waves occurs for any monochromatic frequency within the allowed range, while equatorial wave attractors exist in a denumerable, infinite set of finite-sized continuous frequency intervals. It is also shown that the separable Stern equation, obtained as an approximate equation for waves in a homogeneous fluid confined to the equatorial part of a spherical shell, gives an exact description for buoyancy waves in uniformly but radially stratified fluids in such shells.

A new facility has been developed to investigate the directional solidification of transparent aqueous solutions forming mushy layers in a quasi-two-dimensional system. Experiments have been conducted on NaCl–H2O solutions by translating a Hele-Shaw cell at prescribed rates between fixed heat exchangers providing a temperature gradient of approximately 1°C mm−1. The mush–liquid interface remained planar at all freezing velocities larger than 8 μm s−1, while steepling occurred at lower velocities. No significant undercooling of the mush–liquid interface was detected at freezing velocities up to 12 μm s−1. Mathematical predictions of the steady-state temperature profile and mushy-layer thickness as functions of freezing rate are in excellent agreement with experimental measurements.

The Ffowcs Williams–Hawkings equation represents a standard approach in the prediction of noise from rotating blades. It is widely used for linear aeroacoustic problems concerning helicopter rotors and aircraft propellers and over the last few years, through the use of the so called porous (or permeable) surface formulation, has replaced the Kirchhoff approach in the numerical solution of nonlinear problems. Nevertheless, because of numerical difficulties in evaluating the contribution from supersonic sources, most of the computing tools are still unable to treat the critical velocities at which the shock delocalization occurs. At those conditions, the attention is usually limited to the comparison between the noise prediction and the experimental data in the narrow time region where the pressure peak value is located, but there has been little attention paid to the singular behaviour of the governing equation at supersonic speeds. The aim of this paper is to couple the advantages of the porous formulation to an emission surface integration scheme in order to show if and how the singularities affect the noise prediction and to demonstrate a practical way to remove them. Such an analysis enables an investigation of some interesting and somewhat hidden features of the numerical solution of the governing equation and suggests a new solution approach to predicting the noise of a rotor at any rotational velocity.

This paper deals with perturbed unsteady laminar flows in a pipe. Three types of flows are considered: a flow accelerated from rest; a flow in a pipe generated by the controlled motion of a piston; and a water hammer flow where the transient is generated by the instantaneous closure of a valve. Methods of linear stability theory are used to analyse the behaviour of small perturbations in the flow. Since the base flow is unsteady, the linearized problem is formulated as an initial-value problem. This allows us to consider arbitrary initial conditions and describe both short-time and long-time evolution of the flow. The role of initial conditions on short-time transients is investigated. It is shown that the phenomenon of transient growth is not associated with a certain type of initial conditions. Perturbation dynamics is also studied for long times. In addition, optimal perturbations, i.e. initial perturbations that maximize the energy growth, are determined for all three types of flow discussed. Despite the fact that these optimal perturbations, most probably, will not occur in practice, they do provide an upper bound for energy growth and can be used as a point of reference. Results of numerical simulation are compared with previous experimental data. The comparison with data for accelerated flows shows that the instability cannot be explained by long-time asymptotics. In particular, the method of normal modes applied with the quasi-steady assumption will fail to predict the flow instability. In contrast, the transient growth mechanism may be used to explain transition since experimental transition time is found to be in the interval where the energy of perturbation experiences substantial growth. Instability of rapidly decelerated flows is found to be associated with asymptotic growth mechanism. Energy growth of perturbations is used in an attempt to explain previous experimental results. Numerical results show satisfactory agreement with the experimental features such as the wavelength of the most unstable mode and the structure of the most unstable disturbance. The validity of the quasi-steady assumption for stability studies of unsteady non-periodic laminar flows is discussed.

Drag and lift forces acting on a spherical water droplet in a homogeneous linear shear air flow were studied by means of a three-dimensional direct numerical simulation based on a marker and cell (MAC) method. The effects of the fluid shear rate and the particle (droplet) Reynolds number on drag and lift forces acting on a spherical droplet were compared with those on a rigid sphere. The results show that the drag coefficient on a spherical droplet in a linear shear flow increases with increasing the fluid shear rate. The difference in the drag coefficient between a spherical droplet and a rigid sphere in a linear shear flow never exceeds 4%. The lift force acting on a spherical droplet changes its sign from a positive to a negative value at a particle Reynolds number of Rep ≃ 50 in a linear shear flow and it acts from the high-speed side to the low-speed side for Rep ≥ 50. The behaviour of the lift coefficient on a spherical droplet is similar to that on a stationary rigid sphere and the change of sign is caused by the decrease of the pressure lift. The viscous lift on a spherical droplet is smaller than that on a rigid sphere at the same Rep, whereas the pressure lift becomes larger. These quantitative differences are caused by the flow inside a spherical droplet.

The heat transfer from the surface of a circular cylinder into a crossflow has been computed using a two-dimensional model, for a range of Reynolds numbers from Re=200 to 15550. The boundary-layer separation, the local and overall heat-transfer rates, the eddy- and flare-detachment frequencies and the width of the flares were determined from the numerical simulations. In this range of Reynolds numbers, the heat-transfer process is unsteady and is characterized by a viscous length scale that is inversely proportional to the square root of the Reynolds number. To ensure uniform numerical accuracy for all Reynolds numbers, the dimensions of the computational mesh were selected in proportion to this viscous length scale. The small scales were resolved by at least three nodes within the boundary layers. The frequency of the heat flares increases, and the width of each flare decreases, with the Reynolds number, in proportion to the viscous time and length scales. Despite the presence of three-dimensional structures for the range of Reynolds numbers considered, the two-dimensional model captures the unsteady processes and produced results that were consistent with the available experimental data. It correctly simulated the overall, the front-stagnation and the back-to-total heat-transfer rates.

It has been estimated that the number of spatial modes (or nodal values) required to uniquely determine a two-dimensional turbulent flow at a specific time is finite, and is bounded by Re4/3 for forced turbulence and Re for decaying turbulence. The usual computational estimate of the number of space–time modes required to calculated decaying two-dimensional turbulence is . These bounds neglect intermittency, and it is not known how sharp they are. In this paper we use an adaptive multi-scale wavelet collocation method to estimate for the first time the number of space–time computational modes necessary to represent two-dimensional decaying turbulence as a function of Reynolds number. We find that for 1260 ≤ Re ≤ 40400 over many eddy turn-over times, and that temporal intermittency is stronger than spatial intermittency. The spatial modes alone scale like Re0.7. The β-model then implies that the spatial fractal dimension of the active regions is 1.2, and the temporal fractal dimension is 0.3. These results suggest that the usual estimates are not sharp for adaptive numerical simulations. The relatively high compression confirms the importance of intermittency and encourages the search for reduced mathematical models of two-dimensional turbulence (e.g. in terms of coherent vortices).

We study an interfacial instability in the coupled flow of ice and subglacial sediment, both modelled as viscous media. Unlike other interfacial instabilities in coupled viscous flows at zero Reynolds number, the mechanism considered here does not rely on buoyancy or the effect of an upper free surface, but on the pressure-dependence of the sediment viscosity. Specifically, the instability relies on sediment rheology being such that, when sediment flows in simple shear, sediment flux increases with compressive normal stress at the ice–sediment interface when the velocity of the interface is kept constant. When ice moves over a shallow bump in the interface, it generates a higher compressive stress on the bump's upstream side than in its lee. If in addition the effective sediment viscosity is low compared with that of ice, interfacial velocity remains approximately constant, and this then implies that more sediment flows into the bump than out of it, causing it to grow. Modelling ice as a Newtonian material, we show that this mechanism works for a wide range of sediment rheologies, including the highly nonlinear shear-thinning ones typically thought most appropriate for the description of ‘nearly plastic’ sediment. The instabilities predicted are essentially two-dimensional, with infinite transverse wavelength, and a nonlinear model shows that growth is unbounded until cavitation occurs in the lee of evolving bumps on the interface. The instability mechanism does not seem to predict the formation of common glacial landforms, but may explain the formation of water-filled cavities on deformable glacier beds.

The Stokes boundary layer in the turbulent regime is investigated by using large-eddy simulations (LES). The Reynolds number, based on the thickness of the Stokes boundary layer, is set equal to Reδ = 1790, which corresponds to test 8 of the experimental study of Jensen et al. (J. Fluid Mech. vol. 206, 1989, p. 265).

Our results corroborate and extend the findings of relevant experimental studies: the alternating phases of acceleration and deceleration are correctly reproduced, as is the sharp transition to turbulence, observable at a phase angle between 30° and 45°, and its maximum between 90° and 105°. Overall, a very good agreement was found between our LES first- and second-order turbulent statistics and those of Jensen et al. (1989). Some discrepancies were observed when comparing turbulent intensities in the phases of the cycle characterized by a low level of turbulent activity.

In the central part of the cycle, namely from the mid acceleration to the late deceleration phases, fully developed equilibrium turbulence is present in the flow field, and the boundary layer resembles that of a canonical, steady, wall-bounded flow. In those phases characterized by low turbulent activity, two separate regions can be detected in the flow field: a near-wall one, where the vertical turbulent kinetic energy varies much more rapidly than the other two components, thus giving rise to the formation of horizontal, pancake-like turbulence; and an outer region where both vertical and spanwise velocity fluctuations vary much faster than the streamwise ones, hence producing cigar-like turbulence.

As a side result, the range of application of the plane-averaged dynamic mixed model was assessed based on the qualitative behaviour over the cycle of a significant parameter representing the ratio between a turbulent time scale and a free-stream time scale associated with the oscillatory motion.

This paper investigates the three-dimensional stability of a horizontal flow sheared horizontally, the hyperbolic tangent velocity profile, in a stably stratified fluid. In an homogeneous fluid, the Squire theorem states that the most unstable perturbation is two-dimensional. When the flow is stably stratified, this theorem does not apply and we have performed a numerical study to investigate the three-dimensional stability characteristics of the flow. When the Froude number, Fh, is varied from ∞ to 0.05, the most unstable mode remains two-dimensional. However, the range of unstable vertical wavenumbers widens proportionally to the inverse of the Froude number for Fh ≪ 1. This means that the stronger the stratification, the smaller the vertical scales that can be destabilized. This loss of selectivity of the two-dimensional mode in horizontal shear flows stratified vertically may explain the layering observed numerically and experimentally.

The mean wall-normal gradients of the Reynolds shear stress and the turbulent kinetic energy have direct connections to the transport mechanisms of turbulent-boundary-layer flow. According to the Stokes–Helmholtz decomposition, these gradients can be expressed in terms of velocity–vorticity products. Physical experiments were conducted to explore the statistical properties of some of the relevant velocity–vorticity products. The high-Reynolds-number data (Rθ ≃ O(106), where θ is the momentum thickness) were acquired in the near neutrally stable atmospheric-surface-layer flow over a salt playa under both smooth- and rough-wall conditions. The low-Rθ data were from a database acquired in a large-scale laboratory facility at 1000 > Rθ > 5000. Corresponding to a companion study of the Reynolds stresses (Priyadarshana & Klewicki, Phys. Fluids, vol. 16, 2004, p. 4586), comparisons of low- and high-Rθ as well as smooth- and rough-wall boundary-layer results were made at the approximate wall-normal locations yp/2 and 2yp, where yp is the wall-normal location of the peak of the Reynolds shear stress, at each Reynolds number. In this paper, the properties of the vωz, wωy and uωz products are analysed through their statistics and cospectra over a three-decade variation in Reynolds number. Here u, v and w are the fluctuating streamwise, wall-normal and spanwise velocity components and ωy and ωz are the fluctuating wall-normal and spanwise vorticity components. It is observed that v–ωz statistics and spectral behaviours exhibit considerable sensitivity to Reynolds number as well as to wall roughness. More broadly, the correlations between the v and ω fields are seen to arise from a ‘scale selection’ near the peak in the associated vorticity spectra and, in some cases, near the peak in the associated velocity spectra as well.

A systematic procedure for initializing supersonic and hypersonic turbulent boundary layers at controlled Mach number and Reynolds number conditions is described. The initialization is done by locally transforming a true direct numerical simulation flow field, and results in a nearly realistic initial magnitude of turbulent fluctuations, turbulence structure and energy distribution. The time scales necessary to forget the initial condition are studied. The experimental conditions of previous studies are simulated. The magnitude of velocity and temperature fluctuations, as well as the turbulent shear stresses given by the direct numerical simulations are in agreement with the experimental data.

Rotating stably stratified geophysical flows can exhibit a near ‘balanced’ evolution controlled by the conservative advection of a single scalar quantity, the potential vorticity (PV). This occurs frequently in the Earth's atmosphere and oceans where motions tend to be weak compared with the background planetary rotation and where stratification greatly inhibits vertical motion. Under these circumstances, both high-frequency acoustic waves and lower-frequency inertia–gravity waves (IGWs) contribute little to the flow evolution compared with the even-lower-frequency advection of PV. Moreover, this ‘slow’ PV-controlled balanced evolution appears unable to excite these higher-frequency waves in any significant way – i.e. balance persists.

The present work pushes the limits of balance by systematically exploring the evolution of a range of highly nonlinear flows in which motions are comparable with the background rotation. These flows do not possess a frequency separation between PV advection and IGWs. Nonetheless, the flows exhibit a remarkable persistence of balance. Even when flows are not initialized to minimize the amount of IGWs initially present, and indeed even when flows are deliberately seeded with significant IGW amplitudes, the flow evolution – over many inertial periods (days) – remains strongly controlled by PV advection.

Thermal convection in a horizontal fluid layer heated uniformly from below usually produces an array of convection cells of roughly equal amplitudes. In the presence of a vertical magnetic field, convection may instead occur in vigorous isolated cells separated by regions of strong magnetic field. An approximate model for two-dimensional solutions of this kind is constructed, using the limits of small magnetic diffusivity, large magnetic field strength and large thermal forcing.

The approximate model captures the essential physics of these localized states, enables the determination of unstable localized solutions and indicates the approximate region of parameter space where such solutions exist. Comparisons with fully nonlinear numerical simulations are made and reveal a power-law scaling describing the location of the saddle-node bifurcation in which the localized states disappear.

The formation of a toroidal bubble towards the end of the bubble collapse stage in the neighbourhood of a solid boundary has been successfully studied using the boundary integral method. The further evolution (rebound) of the toroidal bubble is considered with the loss of system energy taken into account. The energy loss is incorporated into a mathematical model by a discontinuous jump in the potential energy at the minimum volume during the short collapse–rebound period accompanying wave emission. This implementation is first tested with the spherically oscillating bubble system using the theoretical Rayleigh–Plesset equation. Excellent agreement with experimental data for the bubble radius evolution up to three oscillation periods is obtained. Secondly, the incorporation of energy loss is tested with the motion of an oscillating bubble system in the neighbourhood of a rigid boundary, in an axisymmetric geometry, using a boundary integral method. Example calculations are presented to demonstrate the possibility of capturing the peculiar entity of a counterjet, which has been reported only in recent experimental studies.

The motion of a spherical particle along a rough bed in a simple shear viscous flow is studied experimentally for a wide range of parameters, varying the particle size and density, the fluid viscosity and the shear rate γ. The instantaneous particle velocity is calculated in the stream, transverse and vertical directions, using a high-speed video imaging system. It is found that the normalized streamwise mean particle velocity U/US, where US is the Stokes settling velocity, depends only on the dimensionless shear rate μ γ/(Δ ρ g d), this relationship being independent of the particle Reynolds number Rep. This result holds for small Rep, which was the case in our experiments (Rep < 10). The characteristic amplitude and frequency of the velocity fluctuations are also given and discussed. A model is then proposed for the mean streamwise velocity, based on ideas of Bagnold (Proc. R. Soc. Lond. A, vol. 332, 1973, p. 473) and calculations of Goldman et al. (Chem. Engng Sci., vol. 22, 1967b, p. 653) for the velocity of a particle close to a smooth plane. From this model an equivalent bed roughness and an effective friction coefficient are deduced.

We study the motion of large bubbles in curved channels both semi-analytically using the lubrication approximation and computationally using a finite-volume/front-tracking method. The steady film thickness is governed by the classical Landau–Levich–Derjaguin–Bretherton (LLDB) equation in the low-capillary-number limit but with the boundary conditions modified to account for the channel curvature. The lubrication results show that the film is thinner on the inside of a bend than on the outside of a bend. They also indicate that the bubble velocity relative to the average liquid velocity is always larger in a curved channel than that in a corresponding straight channel and increases monotonically with increasing channel curvature. Numerical computations are performed for two-dimensional cases and the computational results are found to be in a good agreement with the lubrication theory for small capillary numbers and small or moderate channel curvatures. For moderate capillary numbers the numerical results for the film thickness, when rescaled to account for channel curvature as suggested in the lubrication calculation, essentially collapse onto the corresponding results for a bubble in a straight tube. The lubrication theory is also extended to the motion of large bubbles in a curved channel of circular cross-section.

A formula is derived that shows exactly how much the discrepancy between the volume flux in laminar and in turbulent flow at the same pressure gradient increases as the pressure gradient is increased. We compare laminar and turbulent flows in channels with and without flow control. For the related problem of a fixed bulk-Reynolds-number flow, we seek the theoretical lowest bound for skin-friction drag for control schemes that use surface blowing and suction with zero-net volume-flux addition. For one such case, using a crossflow approach, we show that sustained drag below that of the laminar-Poiseuille-flow case is not possible. For more general control strategies we derive a criterion for achieving sublaminar drag and use this to consider the implications for control strategy design and the limitations at high Reynolds numbers.