We present theoretical models describing the quasi-steady downward transport of buoyant fluid across a gravitationally stable density interface separating two unbounded quiescent fluid masses. The primary transport mechanism is turbulent entrainment resulting from the localised impingement of a vertically forced high-Reynolds-number axisymmetric jet with steady source conditions. The entrainment across the interface is examined in the large-time asymptotic state, wherein the interfacial gravity current, formed by the fluid entrained from the upper layer and the jet, becomes infinitesimally thin and a two-layer stratification persists. Characterising flows with small interfacial Froude numbers $\def \xmlpi #1{}\def \mathsfbi #1{\boldsymbol {\mathsf {#1}}}\let \le =\leqslant \let \leq =\leqslant \let \ge =\geqslant \let \geq =\geqslant \def \Pr {\mathit {Pr}}\def \Fr {\mathit {Fr}}\def \Rey {\mathit {Re}}({{\mathrm{Fr}}}_i)$ as an axisymmetric semi-ellipsoidal impingement dome, we combine conservation equations with a mechanistic model of entrainment and reveal that, in this regime, the dimensionless entrainment flux $E_i$ across the interface follows the power law $E_i = 0.24{{\mathrm{Fr}}}_i^2$. For large-${{\mathrm{Fr}}}_i$ impingements, modelled as a fully penetrating turbulent fountain, we show that $E_i$ no longer scales with ${{\mathrm{Fr}}}_i^2$, but linearly on ${{\mathrm{Fr}}}_i$, following $E_i = 0.42{{\mathrm{Fr}}}_i$. We establish the intermediate range of ${{\mathrm{Fr}}}_i$ over which there is a transition between these quadratic and linear power laws, thus enabling us to classify the dynamics of entrainment across the interface into three distinct regimes. Finally, the close agreement of our solutions with existing experimental results is illustrated.

Wing or fin flexibility can dramatically affect the performance of flying and swimming animals. Both laboratory experiments and numerical simulations have been used to study these effects, but analytical results are notably lacking. Here, we develop small-amplitude theory to model a flapping wing that pitches passively due to a combination of wing compliance, inertia and fluid forces. Remarkably, we obtain a class of exact solutions describing the wing’s emergent pitching motions, along with expressions for how thrust and efficiency are modified by compliance. The solutions recover a range of realistic behaviours and shed new light on how flexibility can aid performance, the importance of resonance, and the separate roles played by wing and fluid inertia. The simple robust estimates afforded by our theory may prove valuable even in situations where details of the flapping motion and wing geometry differ.

The travelling wave instability in a channel with small-amplitude longitudinal grooves of arbitrary shape has been studied. The disturbance velocity field is always three-dimensional with disturbances which connect to the two-dimensional waves in the limit of zero groove amplitude playing the critical role. The presence of grooves destabilizes the flow if the groove wavenumber $\def \xmlpi #1{}\def \mathsfbi #1{\boldsymbol {\mathsf {#1}}}\let \le =\leqslant \let \leq =\leqslant \let \ge =\geqslant \let \geq =\geqslant \def \Pr {\mathit {Pr}}\def \Fr {\mathit {Fr}}\def \Rey {\mathit {Re}}\beta $ is larger than $\beta _{tran}\approx 4.22$, but stabilizes the flow for smaller $\beta $. It has been found that $\beta _{tran}$ does not depend on the groove amplitude. The dependence of the critical Reynolds number on the groove amplitude and wavenumber has been determined. Special attention has been paid to the drag-reducing long-wavelength grooves, including the optimal grooves. It has been demonstrated that such grooves slightly increase the critical Reynolds number, i.e. such grooves do not cause an early breakdown into turbulence.

The diffraction of an initially one-dimensional plane expansion wave over a 90° corner was explored using experiment and numerical simulation. Unlike studies of shock diffraction, expansion wave diffraction has hardly been documented previously. The planar expansion wave was produced in a shock tube by bursting a diaphragm. Two independent parameters were identified for study: (i) the initial diaphragm shock tube pressure ratio, which determines the strength (pressure ratio) of the expansion, and (ii) the position of the diaphragm from the apex of the 90° corner, which determines the width of the wave. The experimentation only considered variation in the shock tube pressure ratio whereas the simulation varied both parameters. A Navier–Stokes solver with Menter’s shear stress transport $\def \xmlpi #1{}\def \mathsfbi #1{\boldsymbol {\mathsf {#1}}}\let \le =\leqslant \let \leq =\leqslant \let \ge =\geqslant \let \geq =\geqslant \def \Pr {\mathit {Pr}}\def \Fr {\mathit {Fr}}\def \Rey {\mathit {Re}}k\mbox{--}\omega $ turbulence model was found to adequately model the overall flow field. A number of major flow features were identified occurring in the vicinity of the corner. These were: a shear layer that originated by flow separation at the apex of the corner; a vortex within a separation bubble that remained attached to the wall, in sharp contrast to what happens in the shock wave diffraction case, where the vortex convects downstream; and a reflected compression wave arising from perturbation signals generated by the diffraction. For a narrow-width expansion wave existing prior to diffraction, the reflected compression wave steepens into an outwardly propagating, weak cylindrical shock wave. Regions of supersonic flow are identified surrounding the bubble and can extend downstream depending on the pressure ratio. Another major flow feature identified in some cases was an oblique shock located near the separation bubble. A large wake region is evident immediately downstream of the bubble and appears to consist of two distinct layers. The experimental results showed large-scale turbulent structures within the separation bubble, and shear layer instability and vortex shedding from the separation bubble were also evident.

A laminar water jet issuing at high speed from a short circular nozzle into air exhibits various instability features at different distances from the nozzle exit. Near the exit, the effects of gaseous friction and pressure are relatively weak. Deformation of the jet surface in this region is mainly due to the instability of a thin liquid shear layer flow, which relaxes from the velocity profile produced by the nozzle wall. In this paper, a model for this type of instability based on linear stability analysis is investigated to describe the process initiating the formation of liquid ligaments disintegrating into fine droplets near the nozzle exit. The modelling comprises identifying unstable waves excitable in the liquid shear layer and exploring a self-destabilizing mechanism by which unstable waves responsible for the formation of liquid ligaments are naturally reproduced from the upstream-propagating capillary waves produced by the growth of the unstable waves themselves. An expression for the location of ligament formation onset is derived that can be compared with experiments. The model also explains changes in jet instability features away from the nozzle exit and for very short nozzles.

We investigate the fluctuations of thermodynamic state variables in compressible aerodynamic wall turbulence, using results of direct numerical simulation (DNS) of compressible turbulent plane channel flow. The basic transport equations governing the behaviour of thermodynamic variables (density, pressure, temperature and entropy) are reviewed and used to derive the exact transport equations for the variances and fluxes (transport by the fluctuating velocity field) of the thermodynamic fluctuations. The scaling with Reynolds and Mach number of compressible turbulent plane channel flow is discussed. Statistics and correlation coefficients of the thermodynamic fluctuations are examined. Finally, detailed budgets of the transport equations for the variances and fluxes of the thermodynamic variables are analysed. The implications of these results, leading both to the understanding of the thermodynamic interactions in compressible wall turbulence and to possible improvements in statistical modelling, are assessed. Finally, the required extension of existing DNS data to fully characterise this canonical flow is discussed.

A detailed analysis of the ‘turbulent inertia’ (TI) term (the wall-normal gradient of the Reynolds shear stress, $\mathrm{d} \langle -uv\rangle /\mathrm{d} y $), in the axial mean momentum equation is presented for turbulent pipe flows at friction Reynolds numbers $\delta ^{+} \approx 500$, 1000 and 2000 using direct numerical simulation. Two different decompositions for TI are employed to further understand the mean structure of wall turbulence. In the first, the TI term is decomposed into the sum of two velocity–vorticity correlations ($\langle v \omega _z \rangle + \langle - w \omega _y \rangle $) and their co-spectra, which we interpret as an advective transport (vorticity dispersion) contribution and a change-of-scale effect (associated with the mechanism of vorticity stretching and reorientation). In the second decomposition, TI is equivalently represented as the wall-normal gradient of the Reynolds shear stress co-spectra, which serves to clarify the accelerative or decelerative effects associated with turbulent motions at different scales. The results show that the inner-normalised position, $y_m^{+}$, where the TI profile crosses zero, as well as the beginning of the logarithmic region of the wall turbulent flows (where the viscous force is leading order) move outwards in unison with increasing Reynolds number as $y_m^{+} \sim \sqrt{\delta ^{+}}$ because the eddies located close to $y_m^{+}$ are influenced by large-scale accelerating motions of the type $\langle - w \omega _y \rangle $ related to the change-of-scale effect (due to vorticity stretching). These large-scale motions of $O(\delta ^{+})$ gain a spectrum of larger length scales with increasing $\delta ^{+}$ and are related to the emergence of a secondary peak in the $-uv$ co-spectra. With increasing Reynolds number, the influence of the $O(\delta ^{+})$ motions promotes viscosity to act over increasingly longer times, thereby increasing the $y^{+}$ extent over which the mean viscous force retains leading order. Furthermore, the TI decompositions show that the $\langle v \omega _z \rangle $ motions (advective transport and/or dispersion of vorticity) are the dominant mechanism in and above the log region, whereas $\langle - w \omega _y \rangle $ motions (vorticity stretching and/or reorientation) are most significant below the log region. The motions associated with $\langle - w \omega _y \rangle $ predominantly underlie accelerations, whereas $\langle v \omega _z \rangle $ primarily contribute to decelerations. Finally, a description of the structure of wall turbulence deduced from the present analysis and our physical interpretation is presented, and is shown to be consistent with previous flow visualisation studies.

We have performed linearized direct numerical simulations (DNS) of flow induced by radial wall injection forced by white-noise Gaussian forcings. We have shown that the frequency spectrum of the flow exhibits low-frequency discrete peaks in the case of a spatial structure of the forcing that is large scale. On the other hand, we observed that the spectrum becomes smooth (with no discrete peaks) if the spatial structure of the forcing is of a smaller extent. We have then tried to analyse these results in the light of global stability analyses. We have first computed the eigenvalue spectrum of the Jacobian and shown that the computed eigenvalues in the frequency range of interest were strongly damped and extremely sensitive to numerical discretization choices if large domains in the axial direction were considered. If shorter domains are used, then the eigenvalues are more robust but still extremely sensitive to the location of the upstream and downstream boundaries. Analysis of the $\def \xmlpi #1{}\def \mathsfbi #1{\boldsymbol {\mathsf {#1}}}\let \le =\leqslant \let \leq =\leqslant \let \ge =\geqslant \let \geq =\geqslant \def \Pr {\mathit {Pr}}\def \Fr {\mathit {Fr}}\def \Rey {\mathit {Re}}\epsilon $-pseudo-spectrum showed that eigenvalues located in a region displaying ‘background’ values of $ \epsilon $ below $10^{-12}$ were extremely sensitive and confirmed that all values in this region of the spectrum were actually quasi-eigenvalues. The eigenvalues are therefore ill-behaved and cannot be invoked to explain the observed discrete frequency selection mechanism. We have then performed a singular value decomposition of the global resolvent matrix to compute the leading optimal gains, optimal forcings and optimal responses, which are robust quantities, insensitive to numerical discretization details. We showed that the frequency response of the flow with the large-scale forcing can accurately be reproduced by an approximation based on the leading optimal gain/forcing/response. Analysis of this approximation showed that it is the projection coefficient of the forcing onto the leading optimal forcing that is responsible for the discrete frequency selection mechanism in the case of the large-scale forcing. From a more physical point of view, such a discrete behaviour stems from the streamwise oscillations of the leading optimal forcings, whose wavelengths vary with frequency, in combination with finite extent forcings (which start or end at locations where the leading optimal forcings are strong). Experimental results in the literature are finally discussed in light of these findings.

The interaction of a planar shock wave with a polygonal $\def \xmlpi #1{}\def \mathsfbi #1{\boldsymbol {\mathsf {#1}}}\let \le =\leqslant \let \leq =\leqslant \let \ge =\geqslant \let \geq =\geqslant \def \Pr {\mathit {Pr}}\def \Fr {\mathit {Fr}}\def \Rey {\mathit {Re}}{\mathrm{N}}_2$ volume surrounded by ${\mathrm{SF}}_6$ is investigated experimentally and numerically. Three polygonal interfaces (square, triangle and diamond) are formed by the soap film technique developed in our previous work, in which thin pins are introduced as angular vertexes to connect adjacent sides of polygonal soap films. The evolutions of the shock-accelerated polygonal interfaces are then visualized by a high-speed schlieren system. Wave systems and interface structures can be clearly identified in experimental schlieren images, and agree well with the numerical ones. Quantitatively, the movement of the distorted interface, and the length and height of the interface structures are further compared and good agreements are achieved between experimental and numerical results. It is found that the evolution of these polygonal interfaces is closely related to their initial shapes. In the square interface, two vortices are generated shortly after the shock impact around the left corner and dominate the flow field at late stages. In the triangular and diamond cases, the most remarkable feature is the small ‘${\mathrm{SF}}_6$ jet’ which grows constantly with time and penetrates the downstream boundary of the interface, forming two independent vortices. These distinct morphologies of the three polygonal interfaces also lead to the different behaviours of the interface features including the length and height. It is also found that the velocities of the vortex pair predicted from the theory of Rudinger and Somers (J. Fluid Mech., vol. 7, 1960, pp. 161–176) agree with the experimental ones, especially for the square case. Typical free precursor irregular refraction phenomena and the transitions among them are observed and analysed, which gives direct experimental evidence for wave patterns and their transitions at a slow/fast interface. The velocities of triple points and shocks are experimentally measured. It is found that the transmitted shock near the interface boundary has weakened into an evanescent wave.

Density fronts are common features of ocean and atmosphere boundary layers. Field observations and numerical simulations have shown that the sharpening of frontal gradients, or frontogenesis, can spontaneously generate inertia–gravity waves (IGWs). Although significant progress has been made in describing frontogenesis using approximations such as quasi-geostrophy (Stone, J. Atmos. Sci., vol. 23, 1966, pp. 455–565, Williams & Plotkin J. Atmos. Sci., vol. 25, 1968, pp. 201–206) semi-geostrophy (Hoskins, Annu. Rev. Fluid Mech., vol. 14, 1982, pp. 131–151), these models omit waves. Here, we further develop the analytical model of Shakespeare & Taylor (J. Fluid Mech., vol. 736, 2013, pp. 366–413) to describe the spontaneous emission of IGWs from an initially geostrophically balanced front subjected to a time-varying horizontal strain. The model uses the idealised configuration of an infinitely long, straight front and uniform potential vorticity (PV) fluid, with a uniform imposed convergent strain across the front, similar to Hoskins & Bretherton (J. Atmos. Sci., vol. 29, 1972, pp. 11–37). Inertia–gravity waves are generated via two distinct mechanisms: acceleration of the large-scale flow and frontal collapse. Wave emission via frontal collapse is predicted to be exponentially small for small values of strain but significant for larger strains. Time-varying strain can also generate finite-amplitude waves by accelerating the cross-front flow and disrupting geostrophic balance. In both cases waves are trapped by the oncoming strain flow and can only propagate away from the frontal zone when the strain field weakens sufficiently, leading to wave emission that is strongly localised in both time and space.

The stability of $\def \xmlpi #1{}\def \mathsfbi #1{\boldsymbol {\mathsf {#1}}}\let \le =\leqslant \let \leq =\leqslant \let \ge =\geqslant \let \geq =\geqslant \def \Pr {\mathit {Pr}}\def \Fr {\mathit {Fr}}\def \Rey {\mathit {Re}}\gamma _2$ travelling waves at the surface of a film flow down an inclined plane is considered experimentally and numerically. These waves are fast, one-humped and quasi-solitary. They undergo a three-dimensional secondary instability if the flow rate (or Reynolds number) is sufficiently high. Rugged or scallop wave patterns are generated by the interplay between a short-wave and a long-wave instability mode. The short-wave mode arises in the capillary region of the wave, with a mechanism of capillary origin which is similar to the Rayleigh–Plateau instability, whereas the long-wave mode deforms the entire wave and is triggered by a Rayleigh–Taylor instability. Rugged waves are observed at relatively small inclination angles. At larger angles, the long-wave mode predominates and scallop waves are observed. For a water film the transition between rugged and scallop waves occurs for an inclination angle around 12°.

The logarithmic law for the mean velocity in turbulent boundary layers has long provided a valuable and robust reference for comparison with theories, models and large-eddy simulations (LES) of wall-bounded turbulence. More recently, analysis of high-Reynolds-number experimental boundary-layer data has shown that also the variance and higher-order moments of the streamwise velocity fluctuations $\def \xmlpi #1{}\def \mathsfbi #1{\boldsymbol {\mathsf {#1}}}\let \le =\leqslant \let \leq =\leqslant \let \ge =\geqslant \let \geq =\geqslant \def \Pr {\mathit {Pr}}\def \Fr {\mathit {Fr}}\def \Rey {\mathit {Re}}u^{\prime +}$ display logarithmic laws. Such experimental observations motivate the question whether LES can accurately reproduce the variance and the higher-order moments, in particular their logarithmic dependency on distance to the wall. In this study we perform LES of very high-Reynolds-number wall-modelled channel flow and focus on profiles of variance and higher-order moments of the streamwise velocity fluctuations. In agreement with the experimental data, we observe an approximately logarithmic law for the variance in the LES, with a ‘Townsend–Perry’ constant of $A_1\approx 1.25$. The LES also yields approximate logarithmic laws for the higher-order moments of the streamwise velocity. Good agreement is found between $A_p$, the generalized ‘Townsend–Perry’ constants for moments of order $2p$, from experiments and simulations. Both are indicative of sub-Gaussian behaviour of the streamwise velocity fluctuations. The near-wall behaviour of the variance, the ranges of validity of the logarithmic law and in particular possible dependencies on characteristic length scales such as the roughness length $z_0$, the LES grid scale $\Delta $, and subgrid scale mixing length $C_s\Delta $ are examined. We also present LES results on moments of spanwise and wall-normal fluctuations of velocity.

This study investigates the suppression of the sound produced when a jet, issued from a circular nozzle or hole in a plate, goes through a similar hole in a second plate. The sound, known as a hole tone, is encountered in many practical engineering situations. The mean velocity of the air jet $\def \xmlpi #1{}\def \mathsfbi #1{\boldsymbol {\mathsf {#1}}}\let \le =\leqslant \let \leq =\leqslant \let \ge =\geqslant \let \geq =\geqslant \def \Pr {\mathit {Pr}}\def \Fr {\mathit {Fr}}\def \Rey {\mathit {Re}}u_0$ was $6\text {--}12\ \mathrm{m}\ {\mathrm{s}}^{-1}$. The nozzle and the end plate hole both had a diameter of 51 mm, and the impingement length $L_{im}$ between the nozzle and the end plate was 50–90 mm. We propose a novel passive control method of suppressing the tone with an axisymmetric obstacle on the end plate. We find that the effect of the obstacle is well described by the combination ($W/L_{im}$, $h$) where $W$ is the distance from the edge of the end plate hole to the inner wall of the obstacle, and $h$ is the obstacle height. The tone is suppressed when backflows from the obstacle affect the jet shear layers near the nozzle exit. We do a direct sound computation for a typical case where the tone is successfully suppressed. Axisymmetric uniformity observed in the uncontrolled case is broken almost completely in the controlled case. The destruction is maintained by the process in which three-dimensional vortices in the jet shear layers convect downstream, interact with the obstacle and recursively disturb the jet flow from the nozzle exit. While regions near the edge of the end plate hole are responsible for producing the sound in the controlled case as well as in the uncontrolled case, acoustic power in the controlled case is much lower than in the uncontrolled case because of the disorganized state.

Gas flow through fractured nanoporous shale formations is complicated by a hierarchy of structural features (ranging from nanopores to microseismic and hydraulic fractures) and by several transport mechanisms that differ from the standard viscous flow used in reservoir modelling. In small pores, self-diffusion becomes more important than advection; also, slippage effects and Knudsen diffusion might become relevant at low densities. We derive a nonlinear effective diffusion coefficient that describes the main transport mechanisms in shale-gas production. In dimensionless form, this coefficient depends only on a geometric factor (or dimensionless permeability) and on the kinetic model that describes the gas. To simplify the description of the complex structure of fractured shales, we observe that the production rate is controlled by the flow from the shale matrix (which has the smallest diffusivity) into the fracture network, which is assumed to produce instantaneously. Therefore, we propose to model the flow in the shale matrix and estimate the production rate with a simple bundle-of-dual-tubes model (BoDTM), in which each tube is characterized by two diameters (one for transport and the other for storage). The solution of a single tube is approximately self-similar at early time, but not at late time, when the gas flux decays exponentially owing to the finite length of the tube. To construct a BoDTM, a reliable estimate of the joint statistics of the matrix-porosity parameters is required. This can be either inferred from core measurements or postulated on the basis of some a priori assumptions when information from laboratory and field measurements is scarce. By comparison with field production data from the Barnett shale-gas field, we demonstrate that BoDTM can be calibrated to estimate structural parameters of the shale formation and to predict the cumulative production of shale gas. Our framework has enough flexibility to construct models of increasing complexity that can be employed in the presence of a complex dataset or when more information is available.

For inhomogeneous turbulent flows at high Reynolds number, it is shown that the redistribution term in Reynolds-stress turbulence models can be determined from the velocity–acceleration correlation. It is further shown that the drift coefficient in the generalized Langevin model (which is used in probability density function (PDF) methods) can be determined from the Reynolds stresses and the velocity–acceleration correlation. These observations are valuable, since the second moments of velocity and acceleration can be measured in experiments, in direct numerical simulations and in well-resolved large-eddy simulations (LES), and hence these turbulence-model quantities can be determined. The redistribution is closely related to the pressure–rate-of-strain, and the unknown in the PDF equation is closely related to the conditional mean pressure gradient (conditional on velocity). In contrast to the velocity–acceleration moments, these pressure statistics are much more difficult to obtain, and our knowledge of them is quite limited. It is also shown that the generalized Langevin model can be re-expressed to provide a direct connection between the drift term and the fluid acceleration. All of these results are first obtained using the constant-property Navier–Stokes equations, but it is then shown that the results are simply extended to variable-density flows.