An analytical framework is developed to understand and predict the thermoacoustic instability in solid rocket motors, taking into account the non-orthogonality of the eigenmodes of the unsteady coupled system. The coupled system comprises the dynamics of the acoustic field and the propellant burn rate. In general, thermoacoustic systems are non-normal leading to non-orthogonality of the eigenmodes. For such systems, the classical linear stability predicted from the eigenvalue analysis is valid in the asymptotic (large time) limit. However, the short-term dynamics can be completely different and a generalized stability theory is needed to predict the linear stability for all times. Non-normal systems show an initial transient growth for suitable initial perturbations even when the system is stable according to the classical linear stability theory. The terms contributing to the non-normality in the acoustic field and unsteady burn rate equations are identified. These terms, which were neglected in the earlier analyses, are incorporated in this analysis. Furthermore, the short-term dynamics are analysed using a system of differential equations that couples the acoustic field and the burn rate, rather than using ad hoc response functions which were used in earlier analyses. In this paper, a solid rocket motor with homogeneous propellant grain has been analysed. Modelling the evolution of the unsteady burn rate using a differential equation increases the degrees of freedom of the thermoacoustic system. Hence, a new generalized disturbance energy is defined which measures the growth and decay of the oscillations. This disturbance energy includes both acoustic energy and unsteady energy in the propellant and is used to quantify the transient growth in the system. Nonlinearities in the system are incorporated by including second-order acoustics and a physics-based nonlinear unsteady burn rate model. Nonlinear instabilities are analysed with special attention given to ‘pulsed instability’. Pulsed instability is shown to occur with pressure coupling for burn rate response. Transient growth is shown to play an important role in pulsed instability.

The results of point-particle Eulerian–Lagrangian direct numerical simulation (DNS) calculations of dilute particle-laden turbulent channel flow are used to study the effect of the particles on the local flow topology. It is found that in the viscous sublayer, the flow becomes increasingly more two-dimensional as the two-way coupling effect (due to interaction between particles and fluid flow) increases with increasing particle load. Beyond the viscous sublayer the modifications in flow topology are not strongly related to the preferential concentration of particles in the flow field, which is in contrast to previous channel flow simulations. The effect of particles on the turbulent flow beyond the viscous sublayer is mostly a result of the overall changing near-wall dynamics of the fluid flow.

This paper presents an experimental study on turbulent flow characteristics in submerged plane wall jets subjected to injection (upward seepage) and suction (downward seepage) from the wall. The vertical distributions of time-averaged velocity components, turbulence intensity components and Reynolds shear stress at different horizontal distances are presented. The horizontal distributions of wall shear stress determined from the Reynolds shear stress profiles are also furnished. The flow field exhibits a decay of the jet velocity over a horizontal distance. The wall shear stress and the rate of decay of the jet velocity increase in the presence of injection and decrease with suction. Based on the two-dimensional Reynolds-averaged Navier–Stokes equations of a steady turbulent flow, the velocity and Reynolds shear stress distributions in the fully developed zone subjected to no seepage, injection and suction are theoretically computed. The response of the turbulent flow characteristics to injection and suction is analysed from the point of view of similarity characteristics, growth of the length scale and decay of the velocity and turbulence characteristics scales. The significant observation is that the velocity, Reynolds shear stress and turbulence intensities in the fully developed zone are reasonably similar under both injection and suction on applying the appropriate scaling laws. An analysis of the third-order moments of velocity fluctuations reveals that the inner layer of the jet is associated with the arrival of low-speed fluid streaks causing an effect of retardation. On the other hand, the upper layer of the jet is associated with the arrival of high-speed fluid streaks causing an effect of acceleration. Injection influences the near-wall distributions of the third-order moments by increasing the upward turbulent advection of the streamwise Reynolds normal stress. In contrast, suction influences the near-wall distributions of the third-order moments by increasing the downward turbulent advection of the streamwise Reynolds normal stress. Also, injection and suction change the vertical turbulent flux of the vertical Reynolds normal stress in a similar way. The streamwise turbulent energy flux travels towards the jet origin within the jet layer, while it travels away from the origin within the inner layer of the circulatory flow. The turbulent energy budget suggests that the turbulent and pressure energy diffusions oppose each other, and the turbulent dissipation lags the turbulent production. The quadrant analysis of velocity fluctuations reveals that the inward and outward interactions are the primary contributions to the Reynolds shear stress production in the inner and outer layers of the jet, respectively. However, injection induces feeble ejections in the vicinity of the wall.

The results presented here are part of a long-term study in which we analyse the segregation of inertial particles in turbulent flows using the so called full Lagrangian method (FLM) to evaluate the ‘compressibility’ of the particle phase along a particle trajectory. In the present work, particles are advected by Stokes drag in a random flow field consisting of counter-rotating vortices and in a flow field composed of 200 random Fourier modes. Both flows are incompressible and, like turbulence, have structure and a distribution of scales with finite lifetime. The compressibility is obtained by first calculating the deformation tensor Jij associated with an infinitesimally small volume of particles following the trajectory of an individual particle. The fraction of the initial volume occupied by the particles centred around a position x at time t is denoted by |J|, where J ≡ det(Jij) and Jij ≡ ∂xi(x0, t)/∂x0,j, x0 denoting the initial position of the particle. The quantity d〈ln|J|〉/dt is shown to be equal to the particle averaged compressibility of the particle velocity field 〈∇ · v〉, which gives a measure of the rate-of-change of the total volume occupied by the particle phase as a continuum. In both flow fields the compressibility of the particle velocity field is shown to decrease continuously if the Stokes number St (the dimensionless particle relaxation time) is below a threshold value Stcr, indicating that the segregation of particles continues indefinitely. We show analytically and numerically that the long-time limit of 〈∇ · v〉 for sufficiently small values of St is proportional to St2 in the flow field composed of random Fourier modes, and to St in the flow field consisting of counter-rotating vortices. If St > Stcr, however, the particles are ‘mixed’. The level of mixing can be quantified by the degree of random uncorrelated motion (RUM) of particles which is a measure of the decorrelation of the velocities of two nearby particles. RUM is zero for fluid particles and increases rapidly with the Stokes number if St > Stcr, approaching unity for St ≫ 1. The spatial averages of the higher-order moments of the particle number density are shown to diverge with time indicating that the spatial distribution of particles may be very intermittent, being associated with non-zero values of RUM and the occurrence of singularities in the particle velocity field. Our results are consistent with previous observations of the radial distribution function in Chun et al. (J. Fluid Mech., vol. 536, 2005, p. 219).

We study non-Newtonian fluid displacements in horizontal narrow eccentric annuli in the situation where the inner cylinder is moving. This represents a practically important extension of the model analysed by Carrasco-Teja et al. (J. Fluid Mech., vol. 605, 2008, pp. 293–327). When motion of the inner cylinder is included, the Hele-Shaw model closure becomes significantly more complex and extremely costly to compute, except for Newtonian fluids. In the first part of the paper we address the model derivation and closure relations. The second part of the paper considers the limit of large buoyancy number, in which the interface elongates along the annulus. We derive a lubrication-style model for this situation, showing that the leading-order interface is symmetric. Rotation of the inner cylinder only affects the length of the leading-order interface, and this occurs only for non-Newtonian fluids via shear-thinning effects. At first order, casing rotation manifests in an asymmetrical ‘shift’ of the interface in the direction of the rotation. We also derive conditions on the eccentricity, fluid rheology and inner cylinder velocity, under which we are able to find steady travelling wave displacement solutions.

Probability distribution functions of fluctuation velocities (P(ux) and P(uy), where ux and uy are the fluctuation velocities in the x- and y-directions, respectively; the gravity is acting along the periodic x-direction and the flow is bounded by two walls parallel to the y-direction) and the density and the spatial velocity correlations are studied using event-driven simulations for an inelastic smooth hard disk system undergoing gravity-driven granular Poiseuille flow (GPF). It is shown that for GPF with smooth and/or perfectly rough walls the Maxwellian/Gaussian is the leading-order distribution over a wide range of densities in the quasi-elastic limit, which is a surprising result, especially for a dilute granular gas for which the Knudsen number belongs to the transitional flow regime. The signature of wall-roughness-induced dissipation mainly shows up in the P(ux) distribution in the form of a sharp peak for negative velocities in the near-wall region. Both P(ux) and P(uy) distributions become asymmetric with increasing dissipation at any density, and the emergence of density waves, which appear in the form of sinuous wave/slug at low-to-moderate values of mean density, makes these asymmetries stronger, especially in the presence of a slug. At high densities, the flow degenerates into a dense plug (where the density approaches its maximum limit and the shear rate is negligibly small) around the channel centreline and two shear layers (where the shear rate is high and the density is low) near the walls. The distribution functions within the shear layer follow the characteristics of those at moderate mean densities. Within the dense plug, the high-velocity tails of both P(ux) and P(uy) appear to undergo a transition from Gaussian in the quasi-elastic limit to power-law distributions at large inelasticity of particle collisions. For dense flows, it is shown that although the density correlations play a significant role in enhancing the velocity correlations when the collisions are sufficiently inelastic, they do not induce velocity correlations when the collisions are quasi-elastic for which the distribution functions are close to Gaussian. The combined effect of enhanced density and velocity correlations around the channel centreline with increasing inelastic dissipation seems to be responsible for the emergence of non-Gaussian high-velocity tails of distribution functions.

We present a study of Eulerian and Lagrangian statistics from a high-resolution numerical simulation of isotropic and homogeneous turbulence using the FLASH code, with an estimated Taylor microscale Reynolds number of around 600. Statistics are evaluated over a data set with 18563 spatial grid points and with 2563 = 16.8 million particles, followed for about one large-scale eddy turnover time. We present data for the Eulerian and Lagrangian structure functions up to the tenth order. We analyze the local scaling properties in the inertial range. The Eulerian velocity field results show good agreement with previous data and confirm the puzzling differences previously found between the scaling of the transverse and the longitudinal structure functions. On the other hand, accurate measurements of sixth-and-higher-order Lagrangian structure functions allow us to highlight some discrepancies from earlier experimental and numerical results. We interpret this result in terms of a possible contamination from the viscous scale, which may have affected estimates of the scaling properties in previous studies. We show that a simple bridge relation based on a multifractal theory is able to connect scaling properties of both Eulerian and Lagrangian observables, provided that the small differences between intermittency of transverse and longitudinal Eulerian structure functions are properly considered.

Receptivity of the two-dimensional boundary layer on a flat plate with elliptic leading edge is studied by numerical simulation. Vortical perturbations in the oncoming free stream are considered, impinging on two leading edges with different aspect ratio to identify the effect of bluntness. The relevance of the three vorticity components of natural free-stream turbulence is illuminated by considering axial, vertical and spanwise vorticity separately at different angular frequencies. The boundary layer is most receptive to zero-frequency axial vorticity, triggering a streaky pattern of alternating positive and negative streamwise disturbance velocity. This is in line with earlier numerical studies on non-modal growth of elongated structures in the Blasius boundary layer. We find that the effect of leading-edge bluntness is insignificant for axial free-stream vortices alone. On the other hand, vertical free-stream vorticity is also able to excite non-modal instability in particular at zero and low frequencies. This mechanism relies on the generation of streamwise vorticity through stretching and tilting of the vertical vortex columns at the leading edge and is significantly stronger when the leading edge is blunt. It can thus be concluded that the non-modal boundary-layer response to a free-stream turbulence field with three-dimensional vorticity is enhanced in the presence of a blunt leading edge. At high frequencies of the disturbances the boundary layer becomes receptive to spanwise free-stream vorticity, triggering Tollmien–Schlichting (T-S) modes and receptivity increases with leading-edge bluntness. The receptivity coefficients to free-stream vortices are found to be about 15% of those to sound waves reported in the literature. For the boundary layers and free-stream perturbations considered, the amplitude of the T-S waves remains small compared with the low-frequency streak amplitudes.

Transient linear growth in laminar magnetohydrodynamic duct flow is analysed. The duct is straight with rectangular cross-section and electrically insulating walls. The applied uniform magnetic field is oriented perpendicular to the mean flow direction and parallel to one of the walls. Optimal perturbations and their maximum amplifications over finite time intervals are computed. The optimal perturbations are increasingly damped by the magnetic field, localized in the boundary layers parallel to the magnetic field irrespective of the duct aspect ratio. Typically, the optimal perturbations have non-vanishing streamwise wavenumber as found in magnetohydrodynamic channel flow with spanwise magnetic field. The Hartmann boundary layers perpendicular to the magnetic field do not contribute to the transient growth.

A three-dimensional computational model is developed for simulating the flag motion in a uniform flow. The nonlinear dynamics of the coupled fluid–flag system after setting up of flapping is investigated by a series of numerical tests. At low Reynolds numbers, the flag flaps symmetrically about its centreline when gravity is excluded, and the bending in the spanwise direction is observed near the corners on the trailing edge. As the Reynolds number increases, the spanwise bending is flattened due to the decrease of the positive pressure near the side edges as well as the viscous force of the fluid. At a certain critical Reynolds number, the flag loses its symmetry about the centreline, which is shown to be related to the coupled fluid–flag instability. The three-dimensional vortical structures shed from the flag show a significant difference from the results of two-dimensional simulations. Hairpin or O-shaped vortical structures are formed behind the flag by connecting those generated at the flag side edges and the trailing edge. Such vortical structures have a stabilization effect on the flag by reducing the pressure difference across the flag. Moreover, the positive pressure near the side edges is significantly reduced as compared with that in the center region, causing the spanwise bending. The Strouhal number defined based on the flag length is slightly dependent on the Reynolds number and the flag width, but scales with the density ratio as St ~ ρ−1/2). On the other hand, the flapping-amplitude-based Strouhal number remains close to 0.2, consistent with the values reported for flying or swimming animals. A flag flapping under gravity is then simulated, which is directed along the negative spanwise direction. The sagging down of the flag and the rolling motion of the upper corner are observed. The dual effects of gravity are demonstrated, i.e. the destabilization effect like the flag inertia and the stabilization effect by increasing the longitudinal tension force.

The aerodynamic noise of a thin rigid annulus (referred to as the ring here) placed in the mixing layer of a subsonic circular jet is investigated in the paper, both theoretically and experimentally. From the experimental point of view, the jet–ring configuration is understood as an axisymmetric alternative to more usual ones involving a rectangular aerofoil held between parallel side plates, dedicated to the study of the noise due to the impingement of upstream turbulence. The main advantages of the circular geometry are a minimum background noise, the absence of tip effects and more specifically the account for all radiation angles from the surface in the far-field acoustic signature. The circular set-up is well suited for the study of pure broadband interaction noise only if the flow remains free of self-sustained oscillations. This is ensured by keeping a sufficient interaction distance between the nozzle and the ring, and by shaping serrations on the nozzle lip. From the theoretical point of view, an analytical model is derived as a straightforward extension of existing formulations. The induced unsteady lift forces on the ring are first inferred from a linearized unsteady aerodynamic theory and the far field is calculated in a second step by a radiation integral. This relates the far-field acoustic pressure power spectral density (PSD) to the two-wavenumber spectrum of the radial turbulent velocity at the ring location, by means of an aeroacoustic transfer function. The latter is shown asymptotically identical to the one detailed in the Appendix for a rectangular aerofoil, in the limit of relatively high frequencies. The analytical acoustic predictions are found to agree well with the measurements over an extended frequency range, provided that the model is fed with turbulent velocity input data measured by a hot-wire probe. Indirectly, this agreement validates the transfer function for a rectangular aerofoil at oblique radiation angles, which is not achievable in a set-up involving side plates and a rectangular nozzle.

We use direct numerical simulation to study the mixing behaviour of pulsed jets in crossflow. The pulse is a square wave and the simulations consider several jet velocity ratios and pulse conditions. Our objective is to study the effects of pulsing and to explain the wide range of optimal pulsing conditions found in experimental studies of the problem. The central theme is that pulsing generates vortex rings; the effect of pulsing on transverse jets can therefore be explained by the behaviour of vortex rings in crossflow. Sau & Mahesh (J. Fluid Mech., vol. 604, 2008, pp. 389–409) show that vortex rings in crossflow exhibit three distinct flow regimes depending on stroke and ring velocity ratios. The simulations of pulsed transverse jets in this paper show that at high velocity ratios, optimal pulse conditions correspond to the transition of the vortex rings produced by pulsing between the different regimes. At low velocity ratios, optimal pulsing conditions are related to the natural time scale on which hairpin vortices form. An optimal curve in the space of stroke and velocity ratios is presented. Data from various experiments are interpreted in terms of the properties of the equivalent vortex rings and shown to collapse on the optimal curve. The proposed regime map allows the effects of experimental parameters such as pulse frequency, duty cycle, modulation and pulse energy all to be predicted by determining their effect on the equivalent stroke and velocity ratios.

This analysis makes use of asymptotic analyses and numerical methods to address, in the limit of small Reynolds and ionic Péclet numbers and small clearances, the canonical problem of the forces exerted on a small solid spherical particle undergoing slow translation and rotation in an incompressible fluid moving parallel to an elastic substrate, subject to electric double-layer and van der Waals intermolecular forces, as a representative example of particle gliding and the idealized swimming dynamics of more complex bodies near soft and sticky surfaces in a physiological solvent. The competition of the hydrodynamic, intermolecular and surface-deformation effects induces a lift force, and drag-force and drift-force perturbations, which do not scale linearly with the velocities, and produce a non-additivity of the intermolecular effects by reducing the intensity of the repulsive forces and by increasing the intensity of the attractive forces. The lift force enhances a reversible elastohydrodynamic adhesion regime in both ionized and deionized solvents, in which lateral motion and lift-off from the substrate can occur. An irreversible elastohydrodynamic adhesion regime, produced by elastic instabilities in the form of surface bifurcations in the substrate, is found to exist for both positive and negative lift forces and is enhanced by small gliding velocities and large substrate compliances, for which critical thresholds are calculated for both ionized and deionized solvents. Elastohydrodynamic corrections are derived for the critical coagulation concentration of electrolyte predicted by the Derjaguin–Landau–Verwey–Overbeek (DLVO) standard theory of colloid stabilization. The corrected DLVO critical concentration is unable to describe the adhesion process when the substrate is sufficiently compliant or when the solvent is deionized. These effects may have consequences on the lateral motility and adhesion of small particles and swimming micro-organisms to soft and sticky substrates, in which the reversible or irreversible character of the adhesion process may be influenced not only by the solvent ionic strength, as described by the DLVO theory, but also by the motion kinematics and the substrate mechanical properties.

A model is presented to predict the orientation and concentration state of a semi-dilute, rigid fibre suspension in a plane channel flow. A probability distribution function is used to describe the local orientation and concentration states of the suspension and evolves according to a Fokker–Planck equation. The fibres are free to interact with each other hydrodynamically and are modelled using the approach outlined by Folgar & Tucker (J. Reinf. Plast. Comp. vol. 3, 1984, p. 98). Near the channel walls, the no-flux boundary conditions as proposed by Schiek & Shaqfeh (J. Fluid Mech. vol. 296, 1995, p. 271) are applied on the orientation distribution function. With this approach, geometric constraints are used to couple the fibres' rotary motion with their translational motion. This eliminates physically unrealistic orientation states in the near-wall region. The concentration distribution is modelled in a similar manner to that proposed by Ma & Graham (Phys. Fluids vol. 17, 2005, art. 083103). A two-way coupling between the fibre orientation state and the momentum equations of the suspending fluid is considered. Experiments are performed to validate the numerical model by visualizing the motion of tracer fibres in an index-of-refraction matched suspension. The orientation distribution function is determined experimentally based on these observations of fibre motion and a comparison is made with the model predictions. Good agreement is shown particularly near the channel walls. The results indicate that at distances less than one-half of a fibre length from the channel walls, the model accurately predicts the available fibre orientation states and the distribution of fibres amongst these states. The model further predicts a large concentration gradient in this region that is also observed experimentally. The magnitude of the concentration gradient in the near-wall region is shown to increase with increasing fibre concentration.

The influence of aspect ratio (χ = diameter/thickness) on the vortex shedding behaviour of fixed, and freely moving, circular disk has been investigated numerically. The aspect ratio significantly changes the structure of the vortices shed from the disk, thus altering the fluid induced forces. Disks of χ = 2 and 4 were selected, and by choosing Re = 240 periodic behaviour was observed for both the ‘fixed’ and ‘freely’ moving disks. First, the vortex structures shed from a ‘fixed’ circular disk of χ = 2 and 4 were computed for Re = 240. This was followed by a computation of their trajectories falling ‘freely’ under the action of gravity at Re = 240. For the ‘fixed’ disk of χ = 2, periodic shedding of one-sided hairpin-shaped vortex loops was observed. The flow field had a spatial planar symmetry and the vortices were shed from the same location, resulting in an asymmetric lateral force on the disk. The Strouhal number (St), calculated using the fluctuation in the axial velocity in the far-wake, was 0.122. This vortex shedding behaviour is referred to as the ‘single-sided’ vortex shedding mode. For the ‘fixed’ disk of χ = 4, periodic shedding of hairpin-shaped vortex loops was observed from the diametrically opposite location of the disk. The flow field had a spatial planar symmetry, and also a spatio-temporal one, with respect to a plane orthogonal to the spatial symmetry plane. The shed vortices induced a symmetric lateral force on the disk with a zero mean. The computed Strouhal number, was equal 0.122, same as that for χ = 2. This vortex shedding behaviour is referred as the ‘double-sided’ vortex shedding mode. For the ‘freely falling’ disk of χ = 2, an oscillatory motion was observed in a plane with a 83° phase lag between the lateral and angular velocity. The Strouhal number (Stb), calculated using the oscillation frequency of the ‘freely’ falling disk was equal to 0.116, which is comparable to the St of the fixed disk. For a ‘freely falling’ disk of χ = 4, oscillatory motion was observed in a plane with a 21° phase lag between the lateral and angular velocity. The Strouhal number (Stb) was equal to 0.171, which differs from the St observed in the wake of the fixed disk.

The rheology of a dilute two-dimensional suspension of vesicles (closed bags of a lipid bilayer membrane) is studied by numerical simulations. The numerical methods used are based on the boundary integral formulation (Green's function technique) and the phase field approach, which has become a quite popular and powerful tool for the numerical study of free-boundary problems. The imposed flow is an unbounded linear shear. The goal of the present study is to elucidate the link between the rheology of vesicle suspensions and the microscopic dynamics of the constituent particles (tank-treading and tumbling motions). A comparison with emulsion rheology reveals the central role played by the membrane. In particular, at low viscosity ratio λ (defined as the viscosity of the internal fluid over that of the ambient one), the effective viscosity decreases with λ, while the opposite trend is exhibited by emulsions, according to the classical Taylor result. This fact is explained by considering the velocity field of the ambient fluid. The area-incompressibility of the vesicle membrane modifies the surrounding velocity field in a quite different manner than what a drop does. The overall numerical results in two dimensions are in reasonable agreement with the three-dimensional analytical theory derived recently in the small deformation limit (quasi-spherical shapes). The finding that the simulations in two dimensions capture the essential features of the three-dimensional rheology opens the way for extensive and large-scale simulations for semi-dilute and concentrated vesicle suspensions. We discuss some peculiar effects exhibited by the instantaneous viscosity in the tumbling regime of vesicles. Finally, the rheology is found to be relatively insensitive to shear rate.

We investigate the laminar shedding of hairpin vortices in the wake of a truncated square cylinder placed in a duct, for Reynolds numbers around the critical threshold of the onset of vortex shedding. We single out the formation mechanism of the hairpin vortices by means of a detailed analysis of the flow patterns in the steady regime. We show that unlike in previous studies of similar structures, the dynamics of the hairpin vortices are entwined with that of the counter-rotating pair of streamwise vortices, which we found to be generated in the bottom part of the near wake (these are usually referred to as ‘base vortices’). In particular, once the hairpin structure is released, the base vortices attach to it, forming its legs, so these are streamwise, and not spanwise as previously observed in unconfined wakes or behind cylinders of lower aspect ratios. We also single out a trail of Ω-shaped vortices, generated between successive hairpin vortices through a mechanism that is analogous to that active in near-wall turbulence. Finally, we show how the dynamics of the structures we identified determine the evolution of the drag coefficients and Strouhal numbers when the Reynolds number varies.

In Pinelli et al. (2010) the correct figure 7 with the corresponding caption should appear as follows.

We apologise for omitting the Reviewer's name at the end of this book review (Page 2010).