The secondary instability of linear transient growth (TG) in Couette flow is explored theoretically, utilizing an analytical representation of the TG based on four modes and their nonlinear interactions. The evolution of the secondary disturbance is derived using the multiple time scales method. The theoretical predictions are compared with direct numerical simulations and very good agreement with respect to the growth of the disturbance energy and associated vortical structures is observed, up to the final stage just before the breakdown to turbulence. The theoretical model enables us to perform a full parametric study, including TG symmetry type, various wavenumbers, initial energy, TG nonlinearity and Reynolds number, to find all possible routes to transition and the optimal parameters for each type of the secondary disturbance. It is found that the most dangerous secondary disturbances are associated with spanwise wavenumbers which generate the strongest inflection points, i.e. those having maximal shear, rather than with those maximizing the energy gain during the TG phase.

Self-sustained convective dynamos in planetary systems operate in an asymptotic regime of rapid rotation, where a balance is thought to hold between the Coriolis, pressure, buoyancy and Lorentz forces (the MAC balance). Classical numerical solutions have previously been obtained in a regime of moderate rotation where viscous and inertial forces are still significant. We define a uni-dimensional path in parameter space between classical models and asymptotic conditions from the requirements to enforce a MAC balance and to preserve the ratio between the magnetic diffusion and convective overturn times (the magnetic Reynolds number). Direct numerical simulations performed along this path show that the spatial structure of the solution at scales larger than the magnetic dissipation length is largely invariant. This enables the definition of large-eddy simulations resting on the assumption that small-scale details of the hydrodynamic turbulence are irrelevant to the determination of the large-scale asymptotic state. These simulations are shown to be in good agreement with direct simulations in the range where both are feasible, and can be computed for control parameter values far beyond the current state of the art, such as an Ekman number $E=10^{-8}$ . We obtain strong-field convective dynamos approaching the MAC balance and a Taylor state to an unprecedented degree of accuracy. The physical connection between classical models and asymptotic conditions is shown to be devoid of abrupt transitions, demonstrating the asymptotic relevance of classical numerical dynamo mechanisms. The fields of the system are confirmed to follow diffusivity-free, power-based scaling laws along the path.

A refined scaling analysis of the two-dimensional mean momentum balance (MMB) for the zero-pressure-gradient turbulent boundary layer (TBL) is presented and experimentally investigated up to high friction Reynolds numbers, $\unicode[STIX]{x1D6FF}^{+}$. For canonical boundary layers, the mean inertia, which is a function of the wall-normal distance, appears instead of the constant mean pressure gradient force in the MMB for pipes and channels. The constancy of the pressure gradient has led to theoretical treatments for pipes/channels, that are more precise than for the TBL. Elements of these analyses include the logarithmic behaviour of the mean velocity, specification of the Reynolds shear stress peak location, the square-root Reynolds number scaling for the log layer onset and a well-defined layer structure based on the balance of terms in the MMB. The present analyses evidence that similarly well-founded results also hold for turbulent boundary layers. This follows from transforming the mean inertia term in the MMB into a form that resembles that in pipes/channels, and is constant across the outer inertial region of the TBL. The physical reasoning is that the mean inertia is primarily a large-scale outer layer contribution, the ‘shape’ of which becomes invariant of $\unicode[STIX]{x1D6FF}^{+}$ with increasing $\unicode[STIX]{x1D6FF}^{+}$, and with a ‘magnitude’ that is inversely proportional to $\unicode[STIX]{x1D6FF}^{+}$. The present analyses are enabled and corroborated using recent high resolution, large Reynolds number hot-wire measurements of all the terms in the TBL MMB.

We provide exact solutions of the Stokes equations for a squirming sphere close to a no-slip surface, both planar and spherical, and for the interactions between two squirmers, in three dimensions. These allow the hydrodynamic interactions of swimming microscopic organisms with confining boundaries, or with each other, to be determined for arbitrary separation and, in particular, in the close proximity regime where approximate methods based on point-singularity descriptions cease to be valid. We give a detailed description of the circular motion of an arbitrary squirmer moving parallel to a no-slip spherical boundary or flat free surface at close separation, finding that the circling generically has opposite sense at free surfaces and at solid boundaries. While the asymptotic interaction is symmetric under head–tail reversal of the swimmer, in the near field, microscopic structure can result in significant asymmetry. We also find the translational velocity towards the surface for a simple model with only the lowest two squirming modes. By comparing these to asymptotic approximations of the interaction we find that the transition from near- to far-field behaviour occurs at a separation of approximately two swimmer diameters. These solutions are for the rotational velocity about the wall normal, or common diameter of two spheres, and the translational speed along that same direction, and are obtained using the Lorentz reciprocal theorem for Stokes flows in conjunction with known solutions for the conjugate Stokes drag problems, the derivations of which are demonstrated here for completeness. The analogous motions in the perpendicular directions, i.e. parallel to the wall, currently cannot be calculated exactly since the relevant Stokes drag solutions needed for the reciprocal theorem are not available.

A drop impacting on a solid surface must push away the intervening gas layer before making contact. This entails a large lubricating air pressure which can deform the bottom of the drop, thus entrapping a bubble under its centre. For a millimetric water drop, the viscous-dominated flow in the thin air layer counteracts the inertia of the drop liquid. For highly viscous drops the viscous stresses within the liquid also affect the interplay between the drop and the gas. Here the drop also forms a central dimple, but its outer edge is surrounded by an extended thin air film, without contacting the solid. This is in sharp contrast with impacts of lower-viscosity drops where a kink in the drop surface forms at the edge of the central disc and makes a circular contact with the solid. Larger drop viscosities make the central air dimple thinner. The thin outer air film subsequently ruptures at numerous random locations around the periphery, when it reaches below 150 nm thickness. This thickness we measure using high-speed two-colour interferometry. The wetted circular contacts expand rapidly, at orders of magnitude larger velocities than would be predicted by a capillary–viscous balance. The spreading velocity of the wetting spots is ${\sim}0.4~\text{m}~\text{s}^{-1}$ independent of the liquid viscosity. This may suggest enhanced slip of the contact line, assisted by rarefied-gas effects, or van der Waals forces in what we call extreme wetting. Myriads of micro-bubbles are captured between the local wetting spots.

This work compares the enstrophy with the scalar dissipation rate, as well as the passive scalar variance with the turbulent kinetic energy, in the presence of coherent Kármán vortices in the intermediate wake of a circular cylinder. Measurements are made at $x/d=10$, 20 and 40, where $x$ is the streamwise distance from the cylinder axis and $d$ is the cylinder diameter, with a Reynolds number of $2.5\times 10^{3}$ based on the cylinder diameter and the free-stream velocity. A probe consisting of eight hot wires (four X-wires) and four cold wires is used to measure simultaneously the three components of the fluctuating velocity and vorticity vectors, as well as the fluctuating temperature gradient vector at nominally the same point in the plane of the mean shear. It is found that the enstrophy and scalar dissipation spectra collapse approximately at all wavenumbers except around the Kármán vortex street wavenumber for $x/d\geqslant 20$. The spectral similarity between the streamwise velocity fluctuation $u$ and the passive scalar $\unicode[STIX]{x1D703}$ is better than that between the velocity fluctuation vector $\boldsymbol{q}$ and $\unicode[STIX]{x1D703}$. This is closely related to the highly organized lateral velocity fluctuation $v$ in this flow. The present observations are fully consistent with the expectation that small scales of the velocity and temperature fields are more likely to exhibit a close relationship than scales associated with the bulk of the turbulent energy or scalar variance. The variation across the wake of the time scale ratio between scalar and velocity fields is significantly smaller than that of the turbulent Prandtl number.

This paper presents an experimental study on internal waves emitted by a horizontally oscillating torus in a linearly stratified fluid. Two internal wave cones are generated with the kinetic energy focused at the apices of the cones above and below the torus where the wave amplitude is maximal. Their motion is measured via tracking of distortions of horizontal fluorescein dye planes created prior to the experiments and illuminated by a vertical laser sheet. The distortion of the dye planes gives a direct access to the Lagrangian displacement of local wave amplitudes and slopes, and in particular, allows us to calculate a local Richardson number. In addition particle image velocimetry measurements are used. Maximum wave slopes are found in the focal region and close to the surface of the torus. As the amplitude of oscillations of the torus increases, wave profiles in the regions of maximum wave slopes evolve nonlinearly toward local overturning. A theoretical approximation based on the theory of Hurley & Keady (J. Fluid Mech., vol. 351, 1997, pp. 119–138) is presented and shows, for small amplitudes of oscillation, a very reasonable agreement with the experimental data. For the focal region the internal wave amplitude is found to be overestimated by the theory. The wave breaking in the focal region is investigated as a function of the Keulegan–Carpenter number, $Ke=A/a$, with $A$ the oscillation amplitude and $a$ the short radius of the torus. A linear wave regime is found for $Ke<0.4$, nonlinear effects start at $Ke\approx 0.6$ and breaking for $Ke>0.8$. For large forcing, the measured wave amplitude normalized with the oscillation amplitude decreases almost everywhere in the wave field, but increases locally in the focal region due to nonlinear effects. Due to geometric focusing the amplitude of the wave increases with $\sqrt{\unicode[STIX]{x1D716}}$, with $\unicode[STIX]{x1D716}=b/a$ and $b$ is the mean radius of the torus. The relevance of wave focusing due to ocean topography is discussed.

An experimental investigation of pressure fluctuations generated by a single-stream compressible jet is carried out in an anechoic wind tunnel. Measurements are performed using a linear array of microphones installed in the near region of the jet and a polar arc of microphones in the far field. The main focus of the paper is on the analysis of the pressure fluctuations in the near field. Three novel signal processing techniques are presented to provide the decomposition of the near-field pressure into hydrodynamic and acoustic components. The procedures are all based on the application of the wavelet transform to the measured pressure data and possess the distinctive property of requiring a very simple arrangement to obtain the desired results (one or two microphones at most). The hydrodynamic and acoustic pressures are characterized separately in terms of their spectral and statistical quantities and a direct link between the acoustic pressure extracted from the near field and the actual noise in the far field is established. The analysis of the separated pressure components sheds light on the nearly Gaussian nature/intermittent behaviour of the acoustic/hydrodynamic pressure. The higher sensitivity of the acoustic component to the Mach number variation has been highlighted as well as the different propagation velocities of the two pressure components. The achieved outcomes are validated through the application to the same data of existing separation procedures evidencing the advantages and limitations of the new methods.

An experimental study of the inertial migration of neutrally buoyant spherical particles suspended in the Poiseuille flow through circular tubes has been conducted at Reynolds numbers $(Re)$ from 100 to 1100 for particle-to-tube diameter ratios of ${\sim}$0.1. The distributions of particles in the tube cross-section were measured at various distances from the tube inlet and the radial probability function of particles was calculated. At relatively high $Re$, the radial probability function was found to have two peaks, corresponding to the so-called Segre–Silberberg annulus and the inner annulus, the latter of which was first reported experimentally by Matas et al. (J. Fluid Mech. vol. 515, 2004, pp. 171–195) to represent accumulation of particles at smaller radial positions than the Segre–Silberberg annulus. They assumed that the inner annulus would be an equilibrium position of particles, where the resultant lateral force on the particles disappears, similar to the Segre–Silberberg annulus. The present experimental study showed that the fraction of particles observed on the Segre–Silberberg annulus increased and the fraction on the inner annulus decreased further downstream, accompanying an outward shift of the inner annulus towards the Segre–Silberberg annulus and a decrease in its width. These results suggested that if the tubes were long enough, the inner annulus would disappear such that all particles would be focused on the Segre–Silberberg annulus for $Re<1000$. At the cross-section nearest to the tube inlet, particles were absent in the peripheral region close to the tube wall including the expected Segre–Silberberg annulus position for $Re>700$. In addition, the entry length after which radial migration has fully developed was found to increase with increasing $Re$, in contrast to the conventional estimate. These results may be related to the developing flow in the tube entrance region where the radial force profile would be different from that of the fully developed Poiseuille flow and there may not be an equilibrium position corresponding to the Segre–Silberberg annulus.

The interaction between oscillating-grid turbulence and a solid, impermeable boundary (positioned below, and aligned parallel to, the grid) is studied experimentally. Instantaneous velocity measurements, obtained using two-dimensional particle imaging velocimetry in the vertical plane through the centre of the (horizontal) grid, are used to study the effect of the boundary on the root-mean-square velocity components, the vertical flux of turbulent kinetic energy (TKE) and the terms in the Reynolds stress transport equation. Identified as a critical aspect of the interaction is the blocking of a vertical flux of TKE across the boundary-affected region. Terms of the Reynolds stress transport equations show that the blocking of this energy flux acts to increase the boundary-tangential turbulent velocity component, relative to the far-field trend, but not the boundary-normal velocity component. The results are compared with previous studies of the interaction between zero-mean-shear turbulence and a solid boundary. In particular, the data reported here are in support of viscous and ‘return-to-isotropy’ mechanisms governing the intercomponent energy transfer previously proposed, respectively, by Perot & Moin (J. Fluid Mech., vol. 295, 1995, pp. 199–227) and Walker et al. (J. Fluid Mech., vol. 320, 1996, pp. 19–51), although we note that these mechanisms are not independent of the blocking of energy flux and draw parallels to the related model proposed by Magnaudet (J. Fluid Mech., vol. 484, 2003, pp. 167–196).

Sheltering of a perturbed vortical layer by a shear flow is a common method to control turbulence and transport in plasma physics. Despite the desire to exploit this phenomenon in wider engineering applications, shear sheltering has rarely been observed in general non-ionized fluids. In this study, we visualize this shear sheltering in a generic neutral-fluid situation in electromembrane desalination: electroconvection (EC) under the Hagen–Poiseuille flow initiated by ion concentration polarization. Our work is the first demonstration of shear sheltering in electrochemical systems. Experiment, numerical simulation and scaling analysis accurately capture the effect by pinpointing the threshold for shear suppression. Determined by balancing the velocity fluctuation (with EC vortices) and the flow shear (with no-slip walls), the threshold for shear suppression is scaled as the EC height. Stable EC with coherent unidirectional vortices occurs under the threshold height, whereas chaotic EC occurs beyond this height as the EC-induced vortical perturbation overwhelms the flow shear. Attractors in a time-delay phase space illustrate this sequence of steady–periodic (stable EC)–chaotic transitions precisely. Going one step further, the shear sheltering effect is decoupled from the shear-independent mechanism of vortex suppression, i.e. vortex sweeping by the mean flow. In the frequency domain, this shear-independent effect is negligible for stable EC (when shear sheltering dominates), whereas it can reduce the level of chaotic fluctuations of chaotic EC (when shear sheltering weakens). Lastly, taken together, we describe the EC-induced convective ion transport by the new scaling law for the electric Nusselt number as a function of the electric Rayleigh number and the Reynolds number. This work not only expands the scientific understanding of EC and the shear sheltering effect, but also affects a broad range of electrochemical applications, including desalination, energy harvesting and sensors.

In 1983, Tulin published a report proposing a framework for reducing the equations for gravity waves generated by moving bodies into a single nonlinear differential equation solvable in closed form (Proceedings of the 14th Symposium on Naval Hydrodynamics, 1983, pp. 19–51). Several new and puzzling issues were highlighted by Tulin, notably the existence of weak and strong wave-making regimes, and the paradoxical fact that the theory seemed to be applicable to flows at low speeds, ‘but not too low speeds’. These important issues were left unanswered, and despite the novelty of the ideas, Tulin’s report fell into relative obscurity. Now, 30 years later, we will revive Tulin’s observations, and explain how an asymptotically consistent framework allows us to address these concerns. Most notably, we demonstrate, using the asymptotic method of steepest descents, how the production of free-surface waves can be related to the arrangement of integration contours connected to the shape of the moving body. This approach provides a new and powerful methodology for the study of geometrically nonlinear wave–body interactions.

We investigated experimentally the flow driven by a Lorentz force induced by an axial magnetic field $\boldsymbol{B}$ and a radial electric current $I$ applied between two fixed concentric copper cylinders. The gap geometry corresponds to a rectangular section with an aspect ratio of $\unicode[STIX]{x1D702}=4$ and we probe the azimuthal and axial velocity profiles of the flow along the vertical axis by using ultrasonic Doppler velocimetry. We have performed several runs at moderate magnetic field strengths, corresponding to moderate Hartmann numbers $M\leqslant 300$. At these forcing parameters and because of the geometry of our experimental device, we show that the inertial terms are not negligible and an azimuthal velocity that depends on both $I$ and $B$ is induced. From measurements of the vertical velocity we focus on the characteristics of the secondary flow: the time-averaged velocity profiles are compatible with a secondary flow presenting two pairs of stable vortices, as pointed out by previous numerical studies. The flow exhibited a transition between two dynamical modes, a high- and a low-frequency one. The high-frequency mode, which emerges at low magnetic field forcing, corresponds to the propagation in the radial $r$-direction of tilted vortices. This mode is consistent with our previous experiments and with the instability described in Zhao et al. (Phys. Fluids, vol. 23 (8), 2011, 084103) taking place in an elongated duct geometry. The low-frequency mode, observed for high magnetic field forcing, consists of large excursions of the vortices. The dynamics of these modes matches the first axisymmetric instability described in Zhao & Zikanov (J. Fluid Mech., vol. 692, 2012, pp. 288–316) taking place in an square duct geometry. We demonstrated that this transition is controlled by the inertial magnetic thickness $H^{\prime }$ which is the characteristic length we introduce as a balance between the advection and the Lorentz force. The key point here is that when the inertial magnetic thickness $H^{\prime }$ is comparable to one geometric characteristic length ($H/2$ in the vertical or $\unicode[STIX]{x0394}r$ in the radial direction) the corresponding mode is favoured. Therefore, when $H^{\prime }/(H/2)\approx 1$ we observe the high-frequency mode taking place in an elongated duct geometry, and when $H^{\prime }/\unicode[STIX]{x0394}r\approx 1$ we observe the low-frequency mode taking place in square duct geometry and high magnetic field.

This study introduces a new point-particle force model that attempts to account for the hydrodynamic influence of the neighbouring particles in an Eulerian–Lagrangian simulation. In previous point-particle models the force on a particle depends only on Reynolds number and mean volume fraction. Thus, as long as the mean local volume fraction is the same, the force on different particles will be estimated to be the same, even though the precise arrangement of neighbours can be vastly different. From direct numerical simulation (DNS) it has been observed that in a random arrangement of spheres that were distributed with uniform probability, the particle-to-particle variation in force can be as large as the mean drag. Since the Reynolds number and mean volume fraction of all the particles within the array are the same, the standard models fail to account for the significant particle-to-particle force variation within the random array. Here, we develop a model which can compute the drag and lateral forces on each particle by accounting for the precise location of a few surrounding neighbours. A pairwise interaction is assumed where the perturbation flow induced by each neighbour is considered separately, then the effects of all neighbours are linearly superposed to obtain the total perturbation. Faxén correction is used to quantify the force perturbation due to the presence of the neighbours. The single neighbour perturbations are mapped in the vicinity of a reference sphere and stored as libraries. We test the pairwise interaction extended point-particle (PIEP) model for random arrays at two different volume fractions of $\unicode[STIX]{x1D719}=0.1$ and 0.21 and Reynolds numbers in the range $16.5\leqslant Re\leqslant 170$. The PIEP model predictions are compared against drag and lift forces obtained from the fully resolved DNS simulations performed using the immersed boundary method. Although not perfect, we observe the PIEP model prediction to correlate much better with the DNS results than the classical mean drag model prediction.

In the limit of a large yield stress, or equivalently at the initiation of motion, viscoplastic flows can develop narrow boundary layers that provide either surfaces of failure between rigid plugs, the lubrication between a plugged flow and a wall or buffers for regions of predominantly plastic deformation. Oldroyd (Proc. Camb. Phil. Soc., vol. 43, 1947, pp. 383–395) presented the first theoretical discussion of these viscoplastic boundary layers, offering an asymptotic reduction of the governing equations and a discussion of some model flow problems. However, the complicated nonlinear form of Oldroyd’s boundary-layer equations has evidently precluded further discussion of them. In the current paper, we revisit Oldroyd’s viscoplastic boundary-layer analysis and his canonical examples of a jet-like intrusion and flow past a thin plate. We also consider flow down channels with either sudden expansions or wavy walls. In all these examples, we verify that viscoplastic boundary layers form as envisioned by Oldroyd. For each example, we extract the dependence of the boundary-layer thickness and flow profiles on the dimensionless yield-stress parameter (Bingham number). We find that, while Oldroyd’s boundary-layer theory applies to free viscoplastic shear layers, it does not apply when the boundary layer is adjacent to a wall, as has been observed previously for two-dimensional flow around circular obstructions. Instead, the boundary-layer thickness scales in a different fashion with the Bingham number, as suggested by classical solutions for plane-parallel flows, lubrication theory and, for flow around a plate, by Piau (J. Non-Newtonian Fluid Mech., vol. 102, 2002, pp. 193–218); we rationalize this second scaling and provide an alternative boundary-layer theory.

The steady and transient response of a laminar separation bubble to flow disturbances is examined experimentally. Wind tunnel experiments are performed on a NACA 0012 aerofoil at a chord Reynolds number of 130 000 and angle of attack of $2^{\circ }$. Under the investigated conditions, a laminar separation bubble forms on the suction side of the aerofoil in the unperturbed flow. Periodic disturbances are introduced into the boundary layer just upstream of separation by means of a surface-mounted dielectric barrier discharge plasma actuator. Two-component, time-resolved particle image velocimetry measurements are performed to characterise both quasi-steady and transient response of the flow to periodic disturbances. The results show that the dynamics of the laminar separation bubble is dominated by the periodic shedding of shear layer vortices, forming upstream of the mean reattachment location due to the amplification of unstable flow disturbances. Introducing the controlled perturbations leads to significant changes in separation bubble topology and the characteristics of the dominant coherent structures, with the effect dependent on both amplitude and frequency of disturbances. Linear stability analysis demonstrates that the induced changes to the mean bubble topology affect the stability characteristics, reducing the maximum growth rate and the frequency of the most amplified disturbances by 35 % and 20 %, respectively, when the bubble length is reduced by up to 40 %. The observed changes in stability characteristics are shown to correlate with the attendant variations in the shape factor. The transient response of the bubble is associated with significant changes in the separation bubble dynamics, with significant differences observed between the relative duration (${\approx}45\,\%$) of the transients flow response associated with the introduction and removal of the controlled disturbances. A detailed analysis of the results offers new insight into the response of laminar separation bubbles to changes in the disturbance environment.

An ultra-thin film of a carrier liquid containing nanosize ferromagnetic particles sitting on a solid substrate and surrounded by an ambient gas phase can be acted upon, apart from viscous and capillary forces, by attractive van der Waals forces which may promote instability leading to film rupture and substrate dewetting. In this article we show that the collective rotation of the particles on the liquid–gas interface, due to a magnetic torque, competes against the instability induced by the van der Waals forces that tend to deepen depressions of the liquid. The competition between the two effects (forcing and instability) leads to the generation of a permanent nonlinear interfacial viscous–capillary wave. Thus, film rupture and substrate dewetting are both averted. This is a general effect that may also be employed to suppress other types of instabilities such as the Rayleigh–Taylor and thermocapillary instabilities. This effect has yet to be observed in experiment.

Two air entrainment mechanisms driven by vortex instability are reported in the unstable relaxation of a horizontal shear layer below a free surface. This flow is experimentally investigated by means of planar laser-induced fluorescence (PLIF) and particle image velocimetry (PIV) coupled with surface profilometry. PLIF identifies counter-rotating vortex pairs (CRVP) emanating from the surface following the growth of high steepness two-dimensional millimetre-size waves for Reynolds and Weber numbers based on the momentum thickness of 177 to 222 and 7.59 to 13.9, respectively. High spatio-temporal resolution PIV reveals the role of surface-generated vorticity and flow separation in the highly curved trough of the waves on the injection of a CRVP. Air bubbles are entrapped in the wake of these CRVPs at Reynolds number above 190. PIV data and spanwise PLIF images show two initiation mechanisms: primary vortex instability modulating the spanwise location where the flow separates, resulting in the pinch off of an air ligament, and secondary vortex instability turning a CRVP into $\unicode[STIX]{x1D6FA}$-shaped loops pulling the surface down. Instability wavelengths agree with linear stability analysis, and models for these new air entrainment mechanisms are proposed.

Using extensive direct numerical simulations, the dynamics of laminar–turbulent fronts in pipe flow is investigated for Reynolds numbers between $Re=2000$ and 5500. We here investigate the physical distinction between the fronts of weak and strong slugs both by analysing the turbulent kinetic energy budget and by comparing the downstream front motion to the advection speed of bulk turbulent structures. Our study shows that weak downstream fronts travel slower than turbulent structures in the bulk and correspond to decaying turbulence at the front. At $Re\simeq 2900$ the downstream front speed becomes faster than the advection speed, marking the onset of strong fronts. In contrast to weak fronts, turbulent eddies are generated at strong fronts by feeding on the downstream laminar flow. Our study also suggests that temporal fluctuations of production and dissipation at the downstream laminar–turbulent front drive the dynamical switches between the two types of front observed up to $Re\simeq 3200$.

The effect of an axial temperature gradient on the flow profile and the induced streaming potential of a pressure-driven symmetric electrolyte in a slit channel is investigated. Based on the non-isothermal Nernst–Planck equations, as well as the Poisson equation in the lubrication approximation, expressions for the ion distribution in the electric double layer (EDL) are derived. It is found that thermophoretic ion motion and a temperature-dependent electrophoretic ion mobility increase the local EDL thickness with temperature, whereas a temperature-dependent permittivity shrinks the EDL. Within the Debye–Hückel approximation, the Navier–Stokes equation with the corresponding electric body force terms is solved. Analytical expressions for the flow profile and the induced (streaming) field under non-isothermal conditions are derived. It is shown that for such a situation the induced electric field is the linear superposition of at least seven individual contributions. For very wide channels, only the thermoelectric field typically present in bulk electrolytes when subjected to a temperature gradient (Soret equilibrium) as well as the conventional pressure-induced streaming field are of importance. Counterintuitively, for the latter, while still being affected by the temperature dependence of the dielectric permittivity and local salt concentration, the temperature dependencies of the viscosity, Fickian diffusion coefficients and ion electromobilities exactly cancel each other. For narrow channels, five additional contributions become relevant, which – similar to the Soret voltage – do not vanish in the case that the externally applied pressure gradient is removed. The first is caused by selective thermo-electromigration driven by the interplay between the temperature-dependent electrophoretic ion mobility and the interaction of the ions with the surface wall charge. This non-advective effect is at its maximum under extreme confinement. For channels whose widths are of the same order as the EDL thickness, four thermoosmotic effects become significant. Besides the well-known thermoosmosis due to the temperature dependence of the dielectric permittivity in the (extended) Korteweg–Helmholtz force, it is demonstrated that – by contrast to isothermal conditions – a thermal gradient renders the ion cloud in the EDL out of mechanical equilibrium. In this context it is shown that a thermophoretic ion motion (i.e. the intrinsic Soret effect of the ions) and a temperature-dependent ion electromobility as well as a temperature-dependent permittivity not only cause an axial gradient of the EDL potential, but simultaneously lead to a pressure of thermal origin, which sets the fluid into an advective motion. Corresponding phenomena were not previously discussed in the literature and may be interpreted as an apparent, thermally induced slip velocity within the EDL. Subsequently, the ion advection affiliated with such thermoosmotic flow may induce a thermoelectric field of a similar order of magnitude to that caused by more conventional thermal effects.