An analysis of natural convection in a horizontal, geometrically non-uniform slot exposed to spatially non-uniform heating has been carried out. The upper plate is smooth and isothermal, and the lower plate has sinusoidal corrugations with a sinusoidal temperature distribution. The distributions of the non-uniformities are characterized in terms of the wavenumber $\unicode[STIX]{x1D6FC}$ and their relative position is expressed in terms of the phase difference $\unicode[STIX]{x1D6FA}_{TL}$. The analysis is limited to heating conditions which do not give rise to secondary motions in the absence of the non-uniformities. The heating creates horizontal temperature gradients which lead to the formation of vertical and horizontal pressure gradients which drive the motion regardless of the intensity of the heating. When the hot spots (points of maximum temperature) overlap either with the corrugation tips or with the corrugation bottoms, convection assumes the form of pairs of counter-rotating rolls whose size is dictated by the heating/corrugation wavelengths. The formation of a net horizontal flow, referred to as thermal drift, is observed for all other relative positions of the hot spots and corrugation tips. Both periodic heating as well as periodic corrugations are required for the formation of this drift, which can be directed in the positive as well as in the negative horizontal directions depending on the phase difference between the heating and corrugation patterns. The most intense convection and the largest drift occur for wavelengths comparable to the slot height, and their intensities increase proportionally to the heating intensity as well as proportionally to the corrugation amplitude, with the drift being a very strong function of the phase difference. Convection creates forces at the plates which would cause horizontal displacement of the corrugated plate and deform the corrugations if such effects were allowed. Tangential forces generated by the uniform heating always contribute to the corrugation buildup while similar forces generated by the periodic heating contribute to the buildup only when the hot spots overlap with the upper part of the corrugation. The processes described above are qualitatively similar for all Prandtl numbers $Pr$, with the intensity of convection and the magnitude of the drift increasing with a reduction in $Pr$.

The existence of exact coherent structures in stably stratified plane Couette flow (gravity perpendicular to the plates) is investigated over Reynolds–Richardson number ($Re$–$Ri_{b}$) space for a fluid of unit Prandtl number $(Pr=1)$ using a combination of numerical and asymptotic techniques. Two states are repeatedly discovered using edge tracking – EQ7 and EQ7-1 in the nomenclature of Gibson & Brand (J. Fluid Mech., vol. 745, 2014, pp. 25–61) – and found to connect with two-dimensional convective roll solutions when tracked to negative $Ri_{b}$ (the Rayleigh–Bénard problem with shear). Both these states and Nagata’s (J. Fluid Mech., vol. 217, 1990, pp. 519–527) original exact solution feel the presence of stable stratification when $Ri_{b}=O(Re^{-2})$ or equivalently when the Rayleigh number $Ra:=-Ri_{b}Re^{2}Pr=O(1)$. This is confirmed via a stratified extension of the vortex wave interaction theory of Hall & Sherwin (J. Fluid Mech., vol. 661, 2010, pp. 178–205). If the stratification is increased further, EQ7 is found to progressively spanwise and cross-stream localise until a second regime is entered at $Ri_{b}=O(Re^{-2/3})$. This corresponds to a stratified version of the boundary region equations regime of Deguchi, Hall & Walton (J. Fluid Mech., vol. 721, 2013, pp. 58–85). Increasing the stratification further appears to lead to a third, ultimate regime where $Ri_{b}=O(1)$ in which the flow fully localises in all three directions at the minimal Kolmogorov scale which then corresponds to the Osmidov scale. Implications for the laminar–turbulent boundary in the ($Re$–$Ri_{b}$) plane are briefly discussed.

We investigate single-fluid magnetohydrodynamics (MHD) with anisotropic viscosity, often referred to as Braginskii MHD, with a particular eye to solar coronal applications. First, we examine the full Braginskii viscous tensor in the single-fluid limit. We pay particular attention to how the Braginskii tensor behaves as the magnetic field strength vanishes. The solar corona contains a magnetic field with a complex and evolving topology, so the viscosity must revert to its isotropic form when the field strength is zero, e.g. at null points. We highlight that the standard form in which the Braginskii tensor is written is not suitable for inclusion in simulations as singularities in the individual terms can develop. Instead, an altered form, where the parallel and perpendicular tensors are combined, provides the required asymptotic behaviour in the weak-field limit. We implement this combined form of the tensor into the Lare3D code, which is widely used for coronal simulations. Since our main focus is the viscous heating of the solar corona, we drop the drift terms of the Braginskii tensor. In a stressed null point simulation, we discover that small-scale structures, which develop very close to the null, lead to anisotropic viscous heating at the null itself (that is, heating due to the anisotropic terms in the viscosity tensor). The null point simulation we present has a much higher resolution than many other simulations containing null points, so this excess heating is a practical concern in coronal simulations. To remedy this unwanted heating at the null point, we develop a model for the viscosity tensor that captures the most important physics of viscosity in the corona: parallel viscosity for strong fields and isotropic viscosity at null points. We derive a continuum model of viscosity where momentum transport, described by this viscosity model, has the magnetic field as its preferred orientation. When the field strength is zero, there is no preferred direction for momentum transport and viscosity reverts to the standard isotropic form. The most general viscous stress tensor of a (single-fluid) plasma satisfying these conditions is found. It is shown that the Braginskii model, without the drift terms, is a specialization of the general model. Performing the stressed null point simulation with this simplified model of viscosity reveals very similar heating profiles to those of the full Braginskii model. The new model, however, does not produce anisotropic heating at the null point, as required. Since the vast majority of coronal simulations use only isotropic viscosity, we perform the stressed null point simulation with isotropic viscosity and compare the heating profiles to those of the anisotropic models. It is shown that the fully isotropic viscosity can overestimate the viscous heating by an order of magnitude.

A novel linear stability analysis for the pattern formation of ripple-like corrugations, over the surface of stalactites due to calcite-laden falling films, is presented by using a modified version of the gradient expansion technique. A fully analytical theory combining thin film hydrodynamics, Kármán–Pohlhausen solution of the advection–diffusion equation for calcium concentration and an evolution equation for the shape of the liquid–solid interface is developed. Analytical formulas for the selected wavelength and the corresponding phase velocity are provided, which match previous numerical solutions. The obtained easy-to-use mathematical results have a potential for developing a novel set of paleo-reconstruction proxies based on the link between morphological patterns and paleo-flows.

In this work, we analyse the internal shear layer structures generated by the libration of an axisymmetric object in an unbounded fluid rotating at a rotation rate $\unicode[STIX]{x1D6FA}^{\ast }$ using direct numerical simulation and small Ekman number asymptotic analysis. We consider weak libration amplitude and libration frequency $\unicode[STIX]{x1D714}^{\ast }$ within the inertial wave interval $(0,2\unicode[STIX]{x1D6FA}^{\ast })$ such that the fluid dynamics is mainly described by a linear axisymmetric harmonic solution. The internal shear layer structures appear along the characteristic cones of angle $\unicode[STIX]{x1D703}_{c}=\text{acos}(\unicode[STIX]{x1D714}^{\ast }/(2\unicode[STIX]{x1D6FA}^{\ast }))$ which are tangent to the librating object at so-called critical latitudes. These layers correspond to thin viscous regions where the singularities of the inviscid solution are smoothed. We assume that the velocity field in these layers is described by the class of similarity solutions introduced by Moore & Saffman (Phil. Trans. R. Soc. Lond. A, vol. 264, 1969, pp. 597–634). These solutions are characterized by two parameters only: a real parameter $m$, which measures the strength of the underlying singularity, and a complex amplitude coefficient $C_{0}$. We first analyse the case of a disk for which a general asymptotic solution for small Ekman numbers is known when the disk is in a solid plane. We demonstrate that the numerical solutions obtained for a free disk and for a disk in a solid plane are both well described by the asymptotic solution and by its similarity form within the internal shear layers. For the disk, we obtain a parameter $m=1$ corresponding to a Dirac source at the edge of the disk and a coefficient $C_{0}\propto E^{1/6}$ where $E$ is the Ekman number. The case of a smoothed librating object such as a spheroid is found to be different. By asymptotically matching the boundary layer solution to similarity solutions close to a critical latitude on the surface, we show that the adequate parameter $m$ for the similarity solution is $m=5/4$, leading to a coefficient $C_{0}\propto E^{1/12}$, that is larger than for the case of a disk for small Ekman numbers. A simple general expression for $C_{0}$ valid for any axisymmetric object is obtained as a function of the local curvature radius at the critical latitude in agreement with this change of scaling. This result is tested and validated against direct numerical simulations.

Buoyant displacement flows of two miscible fluids in rectangular channels are studied, theoretically and experimentally. The scenario considered involves the displacement of a fluid by a slightly heavier one at nearly horizontal channel inclinations, where inertial effects are weak and laminar stratified flows may be expected. In the theoretical part, a lubrication approximation model is developed to simplify the displacement flow governing equations and furnish a semi-analytical solution for the heavy and light fluid flux functions. Three key dimensionless parameters govern the fluid flow motion, i.e. a buoyancy number, the viscosity ratio and the channel cross-section aspect ratio. When these parameters are specified, the reduced model can deliver the interface propagation in time, leading and trailing front heights, shapes and speeds, cross-sectional velocity fields, etc. In addition, the model can be exploited to provide various classifications such as single or multiple fronts as well as main displacement flow regimes at long times such as no sustained backflows, stationary interface flows and sustained backflows. Focusing on the variation of the buoyancy number, a large number of iso-viscous displacement experiments are performed in a square duct and the results are compared with those of the lubrication model. Qualitative displacement flow features observed in the theory and experiments are in good agreement, in particular, in terms of the main displacement flow regimes. The quantitative comparisons are also reasonable for small and moderate imposed displacement flow velocities. However, at large flow rates, a deviation of the experimental results from the model results is observed, which may be due to the presence of non-negligible inertial effects.

The use of spanwise waviness in wings has been proposed in the literature as a possible mechanism for obtaining improved aerodynamic characteristics, motivated by the tubercles that cover the leading edge of the pectoral flippers of the humpback whale. We investigate the effect of this type of waviness on the incompressible flow around infinite wings with a NACA0012 profile, using direct numerical simulations employing the spectral/hp method. Simulations were performed for Reynolds numbers of $Re=10\,000$ and $Re=50\,000$, considering different angles of attack in both the pre-stall and post-stall regimes. The results show that the waviness can either increase or decrease the lift coefficient, depending on the particular $Re$ and flow regime. We observe that the flow around the wavy wing exhibits a tendency to remain attached behind the waviness peak, with separation restricted to the troughs, which is consistent with results from the literature. Then, we identify three important physical mechanisms in this flow. The first mechanism is the weakening of the suction peak on the sections corresponding to the waviness peaks. This characteristic had been observed in a previous investigation for a very low Reynolds number of $Re=1000$, and we show that this is still important even at $Re=50\,000$. As a second mechanism, the waviness has a significant effect on the stability of the separated shear layers, with transition occurring earlier for the wavy wing. In the pre-stall regime, for $Re=10\,000$, the flow around the baseline wing is completely laminar, and the earlier transition leads to a large increase in the lift coefficient, while for $Re=50\,000$, the earlier transition leads to a shortening of the separation bubble which does not lead to an increased lift coefficient. The last mechanism corresponds to a sub-harmonic behaviour, with the flow being notably different between subsequent wavelengths. This allows the wing to maintain higher lift coefficients in some portions of the span.

Shock wave interaction with solid wedges has been an area of much research in past decades, but so far very few results have been obtained for shock wave reflection off liquid wedges. In this study, numerical simulations are performed using the inviscid Euler equations and the stiffened gas equation of state to study the transition angles, reflection patterns and triple point trajectory angles of shock reflection off solid and water wedges. Experiments using an inclined shock tube are also performed and schlieren photography results are compared to simulations. Results show that the transition angles for the water wedge cases are within 5.3 % and 9.2 %, for simulations and experiments respectively, compared to results obtained with the theoretical detachment criterion for solid surfaces. Triple point trajectory angles are measured and compared with analytic solutions, agreement within $1.3^{\circ }$ is shown for the water wedge cases. The transmitted wave in the water observed in the simulation is quantitatively studied, and two different scenarios are found. For low incident shock Mach numbers, $M_{s}=1.2$ and 2, no shock wave is formed in the water but a precursor wave is induced ahead of the incident shock wave and passes the information from the water back into the air. For high incident shock Mach numbers, $M_{s}=3$ and 4, precursor waves no longer appear but instead a shock wave is formed in the water and attached to the Mach stem at every instant. The temperature field in the water is measured in the simulation. For strong incident shock waves, e.g. $M_{s}=4$, the temperature increment in the water is up to 7.3 K.

The flow around a sphere descending at constant speed in a salt-stratified fluid is observed by particle image velocimetry. A unique characteristic of this flow is the appearance of a thin and high-speed rear jet whose maximum velocity can reach more than five times the sphere velocity. In this study we have investigated how the velocity distributions, especially those in the jet and in the boundary layer of the sphere, vary when the Froude number $Fr(=W^{\ast }/N^{\ast }a^{\ast })$ or the Reynolds number $Re(=W^{\ast }(2a^{\ast })/\unicode[STIX]{x1D708}^{\ast })$ ($W^{\ast }$: vertical velocity of the sphere, $N^{\ast }$: Brunt–Väisälä frequency, $a^{\ast }$: radius of the sphere, $\unicode[STIX]{x1D708}^{\ast }$: kinematic viscosity of the fluid) is changed. The results show that the radius of the jet and the thickness of the boundary layer are comparable, and they decrease for smaller Froude numbers and larger Reynolds numbers. Both of them are estimated at moderate Reynolds numbers by the primitive length scale of the stratified fluid ($l_{\unicode[STIX]{x1D708}}^{\ast }=\sqrt{\unicode[STIX]{x1D708}^{\ast }/N^{\ast }}$), or in non-dimensional form by $l_{\unicode[STIX]{x1D708}}^{\ast }/2a^{\ast }=(Fr/2Re)^{1/2}$. The overall velocity distribution in the lee of the sphere is measured to identify the internal wave patterns and their effect on the velocity variation along the jet. Corresponding numerical simulation results using the axisymmetry assumption are in agreement with the experimental results.

The conditions that maximize the propulsive efficiency of a heaving and pitching airfoil are analysed using a novel formulation for the thrust force within the linear potential theory. Stemming from the vortical impulse theory, which correctly predicts the decay of the thrust efficiency as the inverse of reduced frequency $k$ for large $k$ (Fernandez-Feria, Phys. Rev. Fluids, vol. 1, 2016, 084502), the formulation is corrected here at low frequencies by adding a constant representing the viscous drag. It is shown first that the thrust coefficient and propulsive efficiency thus computed agree quite well with several sets of available experimental data, even for not so small flapping amplitudes. For a pure pitching motion, it is found that the maximum propulsion efficiency is reached for the airfoil pitching close to the three-quarter chord point from the leading edge with a relatively large reduced frequency, corresponding to a relatively low thrust coefficient. According to the theory, this efficiency peak may approach unity. For smaller $k$, other less pronounced local maxima of the propulsive efficiency are attained for pitching points ahead of the leading edge, with larger thrust coefficients. The linear theory also predicts that no thrust is generated at all for a pitching axis located between the three-quarter chord point and the trailing edge. These findings contrast with the results obtained from the classical linear thrust by Garrick, with the addition of the same quasi-static thrust, which are also computed in the paper. For a combined heaving and pitching motion, the behaviour of the propulsive efficiency in relation to the pitching axis is qualitatively similar to that found for a pure pitching motion, for given fixed values of the feathering parameter (ratio between pitching and heaving amplitudes) and of the phase shift between the pitching and heaving motions. The peak propulsive efficiency predicted by the linear theory is for an airfoil with a pitching axis close to, but ahead of, the three-quarter chord point, with a relatively large reduced frequency, a feathering parameter of approximately $0.9$ and a phase shift slightly smaller than $90^{\circ }$.

We study axisymmetric, free oscillations driven by gravity and surface tension at the interface of two viscous, immiscible, radially unbounded fluids, analytically and numerically. The interface is perturbed as a zeroth-order Bessel function (in space) and its evolution is obtained as a function of time. In the linearised approximation, we solve the initial value problem (IVP) to obtain an analytic expression for the time evolution of wave amplitude. It is shown that a linearised Bessel mode temporally evolves in exactly the same manner as a Fourier mode in planar geometry. We obtain novel analytical expressions for the time varying vorticity and pressure fields in both fluids. For small initial amplitudes, our analytical results show excellent agreement with those obtained from solving the axisymmetric Navier–Stokes equations numerically. We also compare our results with the normal mode approximation and find the latter to be an accurate representation at very early and late times. The deviation between the normal mode approximation and the IVP solution is found to increase as a function of viscosity ratio. The vorticity field has a jump discontinuity at the interface and we find that this jump depends on the viscosity and the density ratio of the two fluids. Upon increasing the initial perturbation amplitude in the simulations, nonlinearity produces qualitatively new features not present in the analytical IVP solution. Notably, a jet is found to emerge at the axis of symmetry rising to a height greater than the initial perturbation amplitude. Increasing the perturbation amplitude further causes the jet to undergo end pinch off, giving birth to a daughter droplet. This can happen either for an advancing or a receding jet, depending on the viscosity ratio. A relation is found between the maximum jet height and the perturbation amplitude. Hankel transform of the interface demonstrates that at large perturbation amplitudes higher wavenumbers emerge, sharing some of the energy of the lowest mode. When these additional higher modes are present, the interface has pointed crests and rounded troughs.

Ritchmyer–Meshkov instability on an air/SF$_{6}$ interface is experimentally studied in a coaxial converging shock tube by a high-speed laser sheet imaging technique. An unperturbed case is first examined to obtain the characteristics of the converging shock and the shocked interface. For sinusoidal interfaces, the wave pattern and the interface morphology of the whole process are clearly observed. It is quantitatively found that the perturbation amplitude first decreases due to the shock compression, then experiences a rapid growth to a maximum value and finally drops quickly before the reshock. The reduction of growth rate is ascribed to the Rayleigh–Taylor stabilization caused by the interface deceleration motion that is present in the converging circumstance. It is noted that the influence of the wavenumber on the amplitude growth is very little before the reshock, but becomes significant after the reshock.

With the motivation of determining the critical roughness size, a global stability and sensitivity analysis of a three-dimensional Falkner–Skan–Cooke (FSC) boundary layer with a cylindrical surface roughness is performed. The roughness size is chosen such that breakdown to turbulence is initiated by a global version of traditional secondary instabilities of the cross-flow (CF) vortices instead of an immediate flow tripping at the roughness. The resulting global eigenvalue spectra of the systems are found to be very sensitive to numerical parameters and domain size. This sensitivity to numerical parameters is quantified using the $\unicode[STIX]{x1D700}$-pseudospectrum, and the dependency on the domain is analysed through an impulse response, structural sensitivity analysis and an energy budget. It is shown that while the frequencies remain relatively unchanged, the growth rates increase with domain size, which originates from the inclusion of stronger CF vortices in the baseflow. This is reflected in a change in the rate of advective energy transport by the baseflow. It is concluded that the onset of global instability in a FSC boundary layer as the roughness height is increased does not correspond to an immediate flow tripping behind the roughness, but occurs for lower roughness heights if sufficiently long domains are considered. However, the great sensitivity results in an inability to accurately pinpoint the exact parameter values for the bifurcation, and the large spatial growth of the disturbances in the long domains eventually becomes larger than can be resolved using finite-precision arithmetic.

We report an experimental investigation of the separating/reattaching flow over a descending ramp with a $25^{\circ }$ expansion angle. Emphasis is given to mass entrainment through the boundaries of the separated shear layer emanating from the upper edge of the ramp. For this purpose, the turbulent/non-turbulent interface and the separation line inferred from image-based analysis are used respectively to mark the upper and lower bounds of the separated shear layer. The main objective of this study is to identify the physical parameters that scale the development of the separated shear layer, by giving a specific emphasis to the investigation of mass entrainment. Our results emphasise the multiscale nature of mass entrainment through the separated shear layer. The recirculation length $L_{R}$, step height $h$ and free-stream velocity $U_{\infty }$ are the dominant scales that organise the separated flow (and related large-scale quantities as pressure distribution or shear layer growth rate) and set mean mass fluxes. However, local viscous mechanisms seem to be responsible for most of local mass entrainment. Furthermore, it is shown that large-scale mass entrainment is driven by incoming boundary layer properties, since $L_{R}$ scales with $Re_{\unicode[STIX]{x1D703}}$, and in particular by its turbulent state. Surprisingly, the relationships evidenced in this study suggest that these dependencies are established over a large distance upstream of separation and that they might also extend to small scales, at which viscous entrainment is dominant. If confirmed by additional studies, our findings would open new perspectives for designing effective separation control systems.

We present here a new stochastic modelling approach in the constitution of fluid flow reduced-order models. This framework introduces a spatially inhomogeneous random field to represent the unresolved small-scale velocity component. Such a decomposition of the velocity in terms of a smooth large-scale velocity component and a rough, highly oscillating component gives rise, without any supplementary assumption, to a large-scale flow dynamics that includes a modified advection term together with an inhomogeneous diffusion term. Both of those terms, related respectively to turbophoresis and mixing effects, depend on the variance of the unresolved small-scale velocity component. They bring an explicit subgrid term to the reduced system which enables us to take into account the action of the truncated modes. Besides, a decomposition of the variance tensor in terms of diffusion modes provides a meaningful statistical representation of the stationary or non-stationary structuration of the small-scale velocity and of its action on the resolved modes. This supplies a useful tool for turbulent fluid flow data analysis. We apply this methodology to circular cylinder wake flow at Reynolds numbers $Re=100$ and $Re=3900$. The finite-dimensional models of the wake flows reveal the energy and the anisotropy distributions of the small-scale diffusion modes. These distributions identify critical regions where corrective advection effects, as well as structured energy dissipation effects, take place. In providing rigorously derived subgrid terms, the proposed approach yields accurate and robust temporal reconstruction of the low-dimensional models.

Experiments on the collapse of non-colloidal and neutrally buoyant particles suspended in a Newtonian fluid column are presented, in which the initial volume fraction of the suspension $\unicode[STIX]{x1D719}$ , the viscosity of the interstitial fluid $\unicode[STIX]{x1D707}_{f}$ , the diameter of the particles $d$ and the mixing protocol, i.e. the initial preparation of the suspension, are varied. The temporal evolution of the slumping current highlights two main regimes: (i) an inertial-dominated regime followed by (ii) a viscous-dominated regime. The inertial regime is characterized by a constant-speed slumping which is shown to scale as in the case of a classical inertial dam-break. The viscous-dominated regime is observed as a decreasing-speed phase of the front evolution. Lubrication models for Newtonian and power-law fluids describe most of situations encountered in this regime, which strongly depends on the suspension parameters. The temporal evolution of the propagating front is used to extract the rheological parameters of the fluid models. At the early stages of the viscous-dominated regime, a constant effective shear viscosity, referred to as an apparent Newtonian viscous regime, is found to depend only on $\unicode[STIX]{x1D719}$ and $\unicode[STIX]{x1D707}_{f}$ for each mixing protocol. The obtained values are shown to be well fitted by the Krieger–Dougherty model whose parameters involved, say a critical volume fraction $\unicode[STIX]{x1D719}_{m}$ and the exponent of divergence, depend on the mixing protocol, i.e. the microscale interaction between particles. On a longer time scale which depends on $\unicode[STIX]{x1D719}$ , the front evolution is shown to slightly deviate from the apparent Newtonian model. In this apparent non-Newtonian viscous regime, the power-law model, indicating both shear-thinning and shear-thickening behaviours, is shown to be more appropriate to describe the front evolution. The present experiments indicate that the mixing protocol plays a crucial role in the selection of a shear-thinning or shear-thickening type of collapse, while the particle diameter $d$ and volume fraction $\unicode[STIX]{x1D719}$ play a significant role in the shear-thickening case. In all cases, the normalized effective consistency of the power-law fluid model is found to be a unique function of $\unicode[STIX]{x1D719}$ . Finally, an apparent viscoplastic regime, characterized by a finite length spreading reached at finite time, is observed at high $\unicode[STIX]{x1D719}$ . This regime is mostly observed for volume fractions larger than $\unicode[STIX]{x1D719}_{m}$ and up to a volume fraction $\unicode[STIX]{x1D719}_{M}$ close to the random close packing fraction at which the initial column remains undeformed on opening the gate.

In the analysis of the interaction between external periodic excitation and unsteady separated flow in controlling the flow separation, a new nonlinear approximate model has been established. This model is used to describe the typical chaotic and coherent characteristics of a separated flow such as small- or large-scale vortices, the injection, and the dissipation of the kinetic energy based on a simulation of a simplified cross-direction motion of free shear flows. This study presents an appropriate treatment to simulate the external periodic excitation and uses the maximum Lyapunov exponent to evaluate the degree of flow ordering in the different control states. The results of the nonlinear model are compared with experimental and numerical results, showing that the nonlinear model could be used to effectively explain the behaviours of chaotic flows and investigate the rules for controlling separated flows. In addition, as shown in the nonlinear approximate model, the self-synchronization of unsteady flow separation and periodic excitation has been analysed. Initially, the research provided an explanation of the self-synchronization mechanism, which cites that the effects of the separated flow control are independent of the phase difference between the periodic excitation and the unsteady flow. The characteristics of unsteady separated flow control have also been presented in this model, where the corresponding large eddy simulation (LES) was used for separated flows in a curved diffuser. The proper orthogonal decomposition (POD) method was used to analyse the difference between separated vortical structures with or without periodic excitation. The results showed that the model and the simulation had the same mechanism of flow control as for the separated flows. The periodic excitation transforms the original chaotic flow into a relatively ordered flow and decreases the magnitude of the chaotic unstable vortices, rather than completely eliminating the vortices, while flow mixing is reduced, inducing less energy loss.

The process of initial ice formation in brine is a highly complex problem. In this paper, we propose a mathematical model that captures the dynamics of nucleation and development of ice inclusions in brine. The primary emphasis is on the interaction between ice growth and salt diffusion, subject to external forcing provided by temperature. Within this setting two freezing regimes are identified, depending on the rate of change of the temperature: a slow freezing regime where a continuous ice domain is formed; and a fast freezing regime where recurrent nucleation appears within the fluid domain. The second regime is of primary interest, as it leads to fractal-like ice structures. We analyse the critical threshold between the slow and fast regimes by identifying the explicit rates of external temperature control that lead to self-similar salt-concentration profiles in the fluid domains. Subsequent heuristic analysis provides estimates of the characteristic length scales of the fluid domains depending on the time-variation of the temperature. The analysis is confirmed by numerical simulations.

An investigation on the flow around three side-by-side square prisms can provide a better understanding of complicated flow physics associated with multiple, closely spaced structures in which more than one gap flow is involved. In this paper, the flow around three side-by-side square prisms at a Reynolds number $Re=150$ is studied systematically at $L/W=1.1{-}9.0$, where $L$ is the prism centre-to-centre spacing and $W$ is the prism width. Five distinct flow structures and their ranges are identified, viz. base-bleed flow ($L/W<1.4$), flip-flopping flow $(1.4<L/W<2.1)$, symmetrically biased beat flow $(2.1<L/W<2.6)$, non-biased beat flow $(2.6<L/W<7.25)$ and weak interaction flow $(7.25<L/W<9.0)$. Physical aspects of each flow regime, such as vortex structures, vortex dynamics, gap-flow behaviours, shedding frequencies and fluid forces, are discussed in detail. A secondary (beat) frequency other than the Strouhal frequency (primary frequency) is observed in the symmetrically biased and non-biased beat flows, associated with the beat-like modulation in $C_{L}$-peak or amplitude, where $C_{L}$ is the lift force coefficient. Here we reveal the generic and intrinsic origin of the secondary frequency, establishing its connections with the phase lag between the two shear-layer sheddings from the two sides of a gap. When the two sheddings are in phase, no viscous force acts at the interface (i.e. at the centreline of the gap) of the two sheddings, resulting in the largest fluctuations in streamwise momentum, streamwise velocity and pressure; the maximum $C_{L}$ amplitude thus features the in-phase shedding. Conversely, when the two sheddings are antiphase, a viscous force exists at the interface of the two sheddings and restricts the momentum fluctuation through the gap, yielding a minimum $C_{L}$ amplitude. When the phase relationship between the two sheddings changes from in phase to antiphase, the extra viscous force acting at the interface becomes larger and causes the $C_{L}$ amplitude to change from a maximum to a minimum.

Wentzel–Kramers–Brillouin theory was employed by Grimshaw (Geophys. Fluid Dyn., vol. 6, 1974, pp. 131–148) and Achatz et al. (J. Fluid Mech., vol. 210, 2010, pp. 120–147) to derive modulation equations for non-hydrostatic internal gravity wave packets in the atmosphere. This theory allows for wave packet envelopes with vertical extent comparable to the pressure scale height and for large wave amplitudes with wave-induced mean-flow speeds comparable to the local fluctuation velocities. Two classes of exact travelling wave solutions to these nonlinear modulation equations are derived here. The first class involves horizontally propagating wave packets superimposed over rather general background states. In a co-moving frame of reference, examples from this class have a structure akin to stationary mountain lee waves. Numerical simulations corroborate the existence of nearby travelling wave solutions under the pseudo-incompressible model and reveal better than expected convergence with respect to the asymptotic expansion parameter. Travelling wave solutions of the second class also feature a vertical component of their group velocity but exist under isothermal background stratification only. These waves include an interesting nonlinear wave–mean-flow interaction process: a horizontally periodic wave packet propagates vertically while draining energy from the mean wind aloft. In the process it decelerates the lower-level wind. It is shown that the modulation equations apply equally to hydrostatic waves in the limit of large horizontal wavelengths. Aside from these results of direct physical interest, the new nonlinear travelling wave solutions provide a firm basis for subsequent studies of nonlinear internal wave instability and for the design of subtle test cases for numerical flow solvers.