Nature is often an inspiration for scientists and especially so in fluid dynamics. We observe and admire the beauty of birds and fish as they move through the air and water, and wonder how these forms of locomotion evolved and whether they are optimised for efficiency. A common feature of this locomotion is that the thrust is generated by the flapping of a wing or fin, creating an unsteady flow. Other less readily observed animals such as salps or squid eject a pulsed jet, raising the question of whether there is an advantage of this unsteady forcing over a steady jet. This is the question addressed by Ruiz, Whittlesey & Dabiri (J. Fluid Mech., this issue, vol. 668, 2011, pp. 5–32) who have carried out detailed flow and power-consumption measurements with a self-propelled vehicle in water. The vehicle has a novel propulsion mechanism that allows a comparison of the efficiency of a pulsed and steady jet to be compared. They show that significant increases in efficiency are possible with the pulsed jet, even allowing for the additional power needed to create the pulsed flow. They also show that vortex rings are produced by the unsteady jet and that the additional entrainment of ambient fluid into the ring and the higher pressure at the front of the ring are the cause of this increased efficiency.

It has been previously suggested that the generation of coherent vortical structures in the near-wake of a self-propelled vehicle can improve its propulsive efficiency by manipulating the local pressure field and entrainment kinematics. This paper investigates these unsteady mechanisms analytically and in experiments. A self-propelled underwater vehicle is designed with the capability to operate using either steady-jet propulsion or a pulsed-jet mode that features the roll-up of large-scale vortex rings in the near-wake. The flow field is characterized by using a combination of planar laser-induced fluorescence, laser Doppler velocimetry and digital particle-image velocimetry. These tools enable measurement of vortex dynamics and entrainment during propulsion. The concept of vortex added-mass is used to deduce the local pressure field at the jet exit as a function of the shape and motion of the forming vortex rings. The propulsive efficiency of the vehicle is computed with the aid of towing experiments to quantify hydrodynamic drag. Finally, the overall vehicle efficiency is determined by monitoring the electrical power consumed by the vehicle in steady and unsteady propulsion modes. This measurement identifies conditions under which the power required to create flow unsteadiness is offset by the improved vehicle efficiency. The experiments demonstrate that substantial increases in propulsive efficiency, over 50 % greater than the performance of the steady-jet mode, can be achieved by using vortex formation to manipulate the near-wake properties. At higher vehicle speeds, the enhanced performance is sufficient to offset the energy cost of generating flow unsteadiness. An analytical model explains this enhanced performance in terms of the vortex added-mass and entrainment. The results suggest a potential mechanism to further enhance the performance of existing engineered propulsion systems. In addition, the analytical methods described here can be extended to examine more complex propulsion systems such as those of swimming and flying animals, for whom vortex formation is inevitable.

In recent microfluidic experiments with solutions of bacteria we observed the formation of biofilms in the form of thread-like structures, called ‘streamers’, which float in the middle plane of the channel and are connected to the side walls at the inner corners. Motivated by this observation, we discuss here the pressure-driven low-Reynolds-number flow around a corner bounded by the walls of a channel with rectangular cross-section. We numerically solve the flow field in a channel of constant cross-section, which exhibits 90° sharp corners, or turns with constant curvature, or portions with slowly changing curvature along the flow direction, for finite, but small, values of the Reynolds numbers and including the limit of vanishingly small Reynolds numbers. In addition, we develop a matched asymptotic expansion solution for the flow around two boundaries intersecting at an angle α and spanning the small gap h between two horizontal plates. We illustrate the basic features of the flow in these channel geometries by describing the three-dimensional velocity field and the distribution of streamwise vorticity and helicity, and comparing the numerical solutions with predictions based on the asymptotic approach. We demonstrate that near a corner or a change in the curvature of the side wall the flow is three-dimensional and pairs of counter-rotating vortical structures are present, as identified by Balsa (J. Fluid Mech., vol. 372, 1998, p. 25). Finally, we discuss how this secondary flow depends on the significant geometric parameters, the aspect ratio of the channel cross-section, the radius of curvature of the turn and, more generally, the variation of the curvature of the channel side boundary. We believe that these three-dimensional secondary flow structures are relevant to transport problems where accumulation of material at the boundary is possible.

In this paper, we formulate a self-consistent model of thermal superfluid dynamics. By solving it, we analyse the problem of superfluid vorticity cloud propagation in normal-fluid turbulence. We show that superfluid cloud expansion is driven by pattern-forming superfluid vortex instabilities taking place in the interface layer between the cloud's bulk and the outer undisturbed normal-fluid turbulence. The radius of the cloud increases linearly with time. Mutual friction transfers energy from the normal-fluid turbulence to the superfluid cloud, whilst damping the smallest normal-fluid turbulence motions. This damping action is much weaker than viscous dissipation effects in a corresponding pure normal-fluid turbulence. The energy spectrum of superfluid turbulence presents the k−3 scaling that characterizes the spiral superfluid vorticity patterns of normal vortex tube–superfluid vortex interactions. The corresponding k−2 pressure spectrum signifies the singular nature of superfluid vorticity. These two scalings coincide in wavenumber space with the Kolmogorov regime in the normal-fluid turbulence. We compute a fractal dimension df ≈ 1.652 for superfluid vorticity. Due to simpler underlying superfluid vortex dynamics in relation to the strongly nonlinear classical vortex dynamics, this fractal dimension is smaller than the corresponding dimension of vortex tube centrelines in classical turbulence.

An analytical theory is developed which illustrates dynamics of the spontaneous generation of large-scale structures in the unforced two-dimensional eddying flows. The eddy field is represented by the closely packed array of standing coherent vortices whose intensity is weakly modulated by the long-wavelength perturbations introduced into the system. The asymptotic multiscale analysis makes it possible to identify instabilities resulting from the positive feedback of the background eddies on large-scale perturbations. Initially, these instabilities amplify at a rate proportional to the square root of their wavenumber. Linear growth is arrested when the amplitude of the long-wavelength perturbations reaches the level of background eddies. The subsequent evolutionary pattern is characterized by the emergence of relatively sharp features in the large-scale streamfunction field – features suggestive of the coherent jets commonly observed in eddying geophysical flows. The proposed solutions differ substantially from their counterparts in forced-dissipative systems, exemplified by the canonical model of Kolmogorov flow. The asymptotic model is successfully tested against numerical simulations.

The flow of a viscous compressible fluid in a circular tube generated by a sudden impulse at a point on the axis is studied on the basis of the linearized Navier–Stokes equations. A no-slip boundary condition is assumed to hold on the wall of the tube. An efficient numerical scheme has been developed for the calculation of flow velocity and pressure disturbance as a function of position and time.

The transition process in a supersonic flat-plate boundary layer at Mach 2 is investigated numerically using linear stability theory (LST) and direct numerical simulations (DNS). The experimental investigations by Kosinov and his co-workers serve as a reference and provide the physical conditions for the numerical set-up. In these experiments, the weakly nonlinear regime of transition was studied. This led to the discovery of asymmetric subharmonic resonance triads, which appear to be relevant for transition in a Mach 2 boundary layer. These triads were composed of one primary oblique wave of frequency 20kHz and two oblique subharmonic waves of frequency 10kHz. While the experimentalists have focused on this new breakdown mechanism, we have found that the experimental data also indicate the presence of another mechanism related to oblique breakdown. This might be the first experimental evidence of the oblique breakdown mechanism in a supersonic boundary layer. With the simulations presented here, the possible presence of oblique breakdown mechanisms in the experiments is explored by deliberately suppressing subharmonic resonances in the DNS and by comparing the numerical results with the experimental data. The DNS results show excellent agreement with the experimental measurements for both linear and nonlinear transition stages. Most importantly, the results clearly show the characteristic features of oblique breakdown. In addition, we also investigated the subharmonic transition route using LST and DNS. When forcing both the subharmonic and the fundamental frequencies in the DNS, a subharmonic resonance mechanism similar to that in the experiments can be observed.

In this paper, we present results of three-dimensional direct numerical simulations of the spiral turbulence phenomenon in a range of moderate Reynolds numbers, in which alternating intertwined helical bands of turbulent and laminar fluids co-exist and propagate between two counter-rotating concentric cylinders. We show that the turbulent spiral is comprised of numerous small-scale azimuthally elongated vortices, which align into and collectively form the barber-pole-like pattern. The domain occupied by such vortices in a plane normal to the cylinder axis resembles a ‘crescent moon’, a shape made well known by Van Atta with his experiments in the 1960s. The time-averaged mean velocity of spiral turbulence is characterized in the radial–axial plane by two layers of axial flows of opposite directions. We also observe that, as the Reynolds number increases, the transition from spiral turbulence to featureless turbulence does not occur simultaneously in the whole domain, but progresses in succession from the inner cylinder towards the outer cylinder. Certain aspects pertaining to the dynamics and statistics of spiral turbulence and issues pertaining to the simulation are discussed.

The streamwise velocity component is studied in fully developed turbulent channel flow for two very rough surfaces and a smooth surface at comparable Reynolds numbers. One rough surface comprises sparse and isotropic grit with a highly non-Gaussian distribution. The other is a uniform mesh consisting of twisted rectangular elements which form a diamond pattern. The mean roughness heights (±) the standard deviation) are, respectively, about 76(±42) and 145(±150) wall units. The flow is shown to be two-dimensional and fully developed up to the fourth-order moment of velocity. The mean velocity profile over the grit surface exhibits self-similarity (in the form of a logarithmic law) within the limited range of 0.04≤y/h≤0.06, but the profile over the mesh surface does not, even though the mean velocity deficit and higher moments (up to the fourth order) all exhibit outer scaling over both surfaces. The distinction between self-similarity and outer similarity is clarified and the importance of the former is explained. The wake strength is shown to increase slightly over the grit surface but decrease over the mesh surface. The latter result is contrary to recent measurements in rough-wall boundary layers. Single- and two-point velocity correlations reveal the presence of large-scale streamwise structures with circulation in the plane orthogonal to the mean velocity. Spanwise correlation length scales are significantly larger than corresponding ones for both internal and external smooth-wall flows.

The Kraichnan–Leith–Batchelor (KLB) theory of statistically stationary forced homogeneous isotropic two-dimensional turbulence predicts the existence of two inertial ranges: an energy inertial range with an energy spectrum scaling of k−5/3, and an enstrophy inertial range with an energy spectrum scaling of k−3. However, unlike the analogous Kolmogorov theory for three-dimensional turbulence, the scaling of the enstrophy range in the two-dimensional turbulence seems to be Reynolds-number-dependent: numerical simulations have shown that as Reynolds number tends to infinity, the enstrophy range of the energy spectrum converges to the KLB prediction, i.e. E ~ k−3. The present paper uses a novel optimal control approach to find a forcing that does produce the KLB scaling of the energy spectrum in a moderate Reynolds number flow. We show that the time–space structure of the forcing can significantly alter the scaling of the energy spectrum over inertial ranges. A careful analysis of the optimal forcing suggests that it is unlikely to be realized in nature, or by a simple numerical model.

We study the dynamics of neutrally buoyant particles with diameters varying in the range [1, 45] in Kolmogorov scale units (η) and Reynolds numbers based on Taylor scale (Reλ) between 590 and 1050. One component of the particle velocity is measured using an extended laser Doppler velocimetry at the centre of a von Kármán flow, and acceleration is derived by differentiation. We find that the particle acceleration variance decreases with increasing diameter with scaling close to (D/η)−2/3, in agreement with previous observations, and with a hint for an intermittent correction as suggested by arguments based on scaling of pressure spatial increments. The characteristic time of acceleration autocorrelation increases more strongly than previously reported in other experiments, and possibly varying linearly with D/η. Further analysis shows that the probability density functions of the acceleration have smaller wings for larger particles; their flatness decreases as well, as expected from the behaviour of pressure increments in turbulence when intermittency corrections are taken into account. We contrast our measurements with previous observations in wind-tunnel turbulent flows and numerical simulations.

This paper is a continuation of the work begun in Goldstein et al. (J. Fluid Mech., vol. 644, 2010, p. 123), who constructed an asymptotic high-Reynolds-number solution for the flow over a spanwise periodic array of relatively small roughness elements with (spanwise) separation and plan form dimensions of the order of the local boundary-layer thickness. While that paper concentrated on the linear problem, here the focus is on the case where the flow is nonlinear in the immediate vicinity of the roughness with emphasis on the intermediate wake region corresponding to streamwise distances that are large in comparison with the roughness dimension, but small in comparison with the distance between the roughness array and the leading edge. An analytical O(h2) asymptotic solution is obtained for the limiting case of a small roughness height parameter h. These weakly nonlinear results show that the spanwise variable component of the wall-pressure perturbation decays as x−5/3 ln x when x → ∞ (where x denotes the streamwise distance scaled on the roughness dimension), but the corresponding component of the streamwise velocity perturbation (i.e. the wake velocity) exhibits an O(x1/3 ln x) algebraic/transcendental growth in the main boundary layer. Numerical solutions for h = O(1) demonstrate that the wake velocity perturbation for the fully nonlinear case grows in the same manner as the weakly nonlinear prediction – which is considerably different from the strictly linear result obtained in Goldstein et al. (2010).

A spectral direct numerical simulation (DNS) code was used to study the growth and spreading of turbulent spots in a nominally laminar, zero-pressure-gradient boundary layer. In addition to the flat-plate case, the interaction of these spots with riblets was investigated. The flat plate, riblets and initial spot perturbation were simulated via an immersed boundary method, and a ‘suction wall’ allowed the available channel code to model a boundary layer. In both flat-wall and riblet cases, self-similar arrowhead-shaped spots formed. The λ2 variable of Jeong & Hussain (1995) was used to visualize the vortical structures within a spot, and a spot was seen to consist primarily of a multitude of entwined hairpin vortices. The range of scales of the hairpin vortices was found to increase as the spot matures. Ensemble averaging was used to obtain more accurate results for the spot spreading angle, both for the flat-wall case and the riblet case. The spreading angle for the flat-wall spot was 6.3°, in reasonably good agreement with prior DNS work. The spreading angle for the spot over riblets was 5.4°, a decrease of 14% compared with the flat-wall.

We describe experiments of the entrainment and mixing of water (sub-Kolmogorov scale) droplets across a turbulent–non-turbulent interface (TNI) as well a turbulent–turbulent interface (TTI) in shearless grid turbulence, over a time scale in which evaporation is insignificant. The flow is produced by means of a splitter plate with an active grid and water sprays on one side and screens or an active grid on the other side. The Taylor microscale Reλ on the turbulent side is 275 and the average dissipation scale Stokes number, Stη ≈ 0.2, and based on the integral scale, Stl ≈ 0.003. By changing the orientation of the grid system, gravitational effects may be excluded or included. We show that in the absence of gravity, for the Stokes number range studied (0.06 ≤ Stη ≤ 1.33), the droplet distribution does not change across the interface. With gravity, the larger drops are selectively mixed and this is more pronounced for the TNI than for the TTI. The particle concentration distribution is an error function for the TTI but departs significantly for the TNI due to the intermittency in the flow. In terms of particle concentration, the entrainment is most efficient for the TTI with gravity. The results are related to droplet entrainment in clouds.

The effect of blockage on the onset of instability in the two-dimensional uniform flow past a cascade of cylinders is investigated. The same techniques as those described in Gajjar & Azzam (J. Fluid Mech., vol. 520, 2004, p. 51) are used to tackle the generalized eigenvalue problem arising from a global stability analysis of the linearized disturbance equations. Results have been obtained for the various mode classes, and our results show that for the odd–even modes, which correspond to anti-phase oscillatory motion about the midplane between the cylinders and are the modes most extensively studied in the literature, the effect of blockage has a marginal influence on the critical Reynolds numbers for instability. This is in sharp contrast to results cited in many studies with a fully developed inlet flow past a cylinder placed between confining walls. We are also able to find other unstable modes and in particular for low blockage ratios, the odd–odd modes which correspond to the in-phase oscillatory motion about the midplane between the cylinders are the first to become unstable as compared with the odd–even modes, and with much lower frequencies.

An approximate decomposition of fluid-flow variables satisfying unbounded compressible Navier–Stokes equations into acoustically radiating and non-radiating components leads to well-defined source terms that can be identified as the physical sources of aerodynamic noise. We show that, by filtering the flow field by means of a linear convolution filter, it is possible to decompose the flow into non-radiating and radiating components. This is demonstrated on two different flows: one satisfying the linearised Euler equations and the other the Navier–Stokes equations. In the latter case, the corresponding sound sources are computed. They are found to be more physical than those computed through classical acoustic analogies in which the flow field is decomposed into a steady mean and fluctuating component.

A stochastic eddy-damped quasi-normal Markovian (EDQNM) approach is used to investigate self-similar decaying isotropic turbulence at a high Reynolds number (400 ≤ Reλ ≤ 104). The realistic energy spectrum functional form recently proposed by Meyers & Menevau (Phys. Fluids, vol. 20, 2008, p. 065109) is generalized by considering some of the model constants as random parameters, since they escape measure in most experimental set-ups. The induced uncertainty on the solution is investigated, building response surfaces for decay power-law exponents of usual physical quantities. Large-scale uncertainties are considered, the emphasis being put on Saffman and Batchelor turbulences. The sensitivity of the solution to initial spectrum uncertainties is quantified through probability density functions of the decay exponents. It is observed that the initial spectrum shape at very large scales governs the long-time evolution, even at a high Reynolds number, a parameter which is not explicitly taken into account in many theoretical works. Therefore, a universal asymptotic behaviour in which kinetic energy decays as t−1 is not detected. However, this decay law is observed at finite Reynolds numbers with low probability for some initial conditions.

This paper presents a finite-element simulation of the interfacial flow during propulsion of water walkers such as fishing spiders and water striders. The unsteady stroke of the driving leg is represented by a two-dimensional cylinder moving on a specified trajectory. The interface and the moving contact lines are handled by a diffuse-interface model. We explore the mechanism of thrust generation in terms of the interfacial morphology and flow structures. Results show that the most important component of the thrust is the curvature force related to the deformation of the menisci and the asymmetry of the dimple. For water walkers with thick legs, the pressure force due to the inertia of the water being displaced by the leg is also important. The viscous force is negligible. An extensive parametric study is performed on the effect of leg velocity, stroke depth, leg diameter and surface wettability. The propulsive force is insensitive to the contact angle on the leg. However, the hydrophobicity of the leg helps it detach from the surface during the recovery stroke and thus decreases the resistance. It is also important for averting or delaying penetration of the interface at large rowing velocity and depth. In two dimensions, surface waves are more efficient than vortices in transferring the momentum imparted by the leg to the water.

This study uses spatio-temporal stability analysis to investigate the convective and absolute instability properties of a steady unconfined planar liquid jet. The approach uses a piecewise linear velocity profile with a finite-thickness shear layer at the edge of the jet. This study investigates how properties such as the thickness of the shear layer and the value of the fluid velocity at the interface within the shear layer affect the stability properties of the jet. It is found that the presence of a finite-thickness shear layer can lead to an absolute instability for a range of density ratios, not seen when a simpler plug flow velocity profile is considered. It is also found that the inclusion of surface tension has a stabilizing effect on the convective instability but a destabilizing effect on the absolute instability. The stability results are used to obtain estimates for the breakup length of a planar liquid jet as the jet velocity varies. It is found that reducing the shear layer thickness within the jet causes the breakup length to decrease, while increasing the fluid velocity at the fluid interface within the shear layer causes the breakup length to increase. Combining these two effects into a profile, which evolves realistically with velocity, gives results in which the breakup length increases for small velocities and decreases for larger velocities. This behaviour agrees qualitatively with existing experiments on the breakup length of axisymmetric jets.

We address the sequence of events accompanying the transition from an initially compact volume of liquid – a drop – into dispersed fragments when it impacts a solid surface. We describe the change of topology of the drop to a radially expanding sheet and discuss the reasons of its rim destabilization, responsible for the emergence of radial ligaments which ultimately fragment into smaller drops. The dynamics ruling the radius of the sheet, its stability and the resulting fragment drop size distribution are documented experimentally. The radius dynamics results from a simple balance between inertia of the initial drop and capillary restoring forces at the rim, with damping due to the continuous transfer of momentum from the sheet to the rim. The ligaments expelled from the rim originate from a Rayleigh–Taylor mechanism localized at the rim. The final drop size distribution in the spray is shown to be a linear superposition of gamma distributions characteristic of ligament breakup, leading generically to Bessel functions.