The world of cellular biology provides us with many fascinating fluid dynamical phenomena that lie at the heart of physiology, development, evolution and ecology. Advances in imaging, micromanipulation and microfluidics over the past decade have made possible high-precision measurements of such flows, providing tests of microhydrodynamic theories and revealing a wealth of new phenomena calling out for explanation. Here I summarize progress in four areas within the field of ‘active matter’: cytoplasmic streaming in plant cells, synchronization of eukaryotic flagella, interactions between swimming cells and surfaces and collective behaviour in suspensions of microswimmers. Throughout, I emphasize open problems in which fluid dynamical methods are key ingredients in an interdisciplinary approach to the mysteries of life.

We develop a posteriori ‘mechanical’ error estimators that are able to evaluate the solution discrepancy between two ice flow models. We first reformulate the classical shallow ice flow models by applying simplifications to the weak formulation of the Glen–Stokes model. This approach leads to a unified hierarchical formulation which relates the Glen–Stokes model, the Blatter model, the shallow ice approximation and the shallow shelf approximation. Based on this formulation and on residual techniques commonly used to estimate numerical errors, we derive three a posteriori estimators, each of which compares a pair of models using measures of the velocity field from the simpler (shallower) model. Numerical experiments confirm that these estimators can be used to assess the validity of the shallow ice models that are commonly used in glacier and ice sheet modelling.

We study the flow past a sphere rotating in the transverse direction with respect to the incoming uniform flow, and particularly consider the stability features of the wake as a function of the Reynolds number $Re$ and the sphere dimensionless rotation rate $\unicode[STIX]{x1D6FA}$. Direct numerical simulations and three-dimensional global stability analyses are performed in the ranges $150\leqslant \mathit{Re}\leqslant 300$ and $0\leqslant \unicode[STIX]{x1D6FA}\leqslant 1.2$. We first describe the base flow, computed as the steady solution of the Navier–Stokes equation, with special attention to the structure of the recirculating region and to the lift force exerted on the sphere. The stability analysis of this base flow shows the existence of two different unstable modes, which occur in different regions of the $Re/\unicode[STIX]{x1D6FA}$ parameter plane. Mode I, which exists for weak rotations ($\unicode[STIX]{x1D6FA}<0.4$), is similar to the unsteady mode existing for a non-rotating sphere. Mode II, which exists for larger rotations ($\unicode[STIX]{x1D6FA}>0.7$), is characterized by a larger frequency. Both modes preserve the planar symmetry of the base flow. We detail the structure of these eigenmodes, as well as their structural sensitivity, using adjoint methods. Considering small rotations, we then compare the numerical results with those obtained using weakly nonlinear approaches. We show that the steady bifurcation occurring for $Re>212$ for a non-rotating sphere is changed into an imperfect bifurcation, unveiling the existence of two other base-flow solutions which are always unstable.

We study the physics of steady detonation wave propagation in a two-dimensional circular arc via a Detonation Shock Dynamics (DSD) surface evolution model. The dependence of the surface angular speed and surface spatial structure on the inner arc radius ($R_{i}$), the arc thickness ($R_{e}-R_{i}$, where $R_{e}$ is the outer arc radius) and the degree of confinement on the inner and outer arc is examined. We first analyse the results for a linear $D_{n}$–$\unicode[STIX]{x1D705}$ model, in which the normal surface velocity $D_{n}=D_{CJ}(1-B\unicode[STIX]{x1D705})$, where $D_{CJ}$ is the planar Chapman–Jouguet velocity, $\unicode[STIX]{x1D705}$ is the total surface curvature and $B$ is a length scale representative of a reaction zone thickness. An asymptotic analysis assuming the ratio $B/R_{i}\ll 1$ is conducted for this model and reveals a complex surface structure as a function of the radial variation from the inner to the outer arc. For sufficiently thin arcs, where $(R_{e}-R_{i})/R_{i}=O(B/R_{i})$, the angular speed of the surface depends on the inner arc radius, the arc thickness and the inner and outer arc confinement. For thicker arcs, where $(R_{e}-R_{i})/R_{i}=O(1)$, the angular speed does not depend on the outer arc radius or the outer arc confinement to the order calculated. It is found that the leading-order angular speed depends only on $D_{CJ}$ and $R_{i}$, and corresponds to a Huygens limit (zero curvature) propagation model where $D_{n}=D_{CJ}$, assuming a constant angular speed and perfect confinement on the inner arc surface. Having the normal surface speed depend on curvature requires the insertion of a boundary layer structure near the inner arc surface. This is driven by an increase in the magnitude of the surface wave curvature as the inner arc surface is approached that is needed to meet the confinement condition on the inner arc surface. For weak inner arc confinement, the surface wave spatial variation with the radial coordinate is described by a triple-deck structure. The first-order correction to the angular speed brings in a dependence on the surface curvature through the parameter $B$, while the influence of the inner arc confinement on the angular velocity only appears in the second-order correction. For stronger inner arc confinement, the surface wave structure is described by a two-layer solution, where the effect of the confinement on the angular speed is promoted to the first-order correction. We also compare the steady-state arc solution for a PBX 9502 DSD model to an experimental two-dimensional arc geometry validation test.

We study the effect of anisotropy and inhomogeneity in the permeability of the porous layer on the stability of surface waves of an inclined fluid–porous double-layer system. The fluid is assumed to be Newtonian and the porous layer to be Darcian. The porous layer is saturated with the same fluid and the two layers are coupled at the interface via the Beavers–Joseph condition. Linear stability analysis is performed based on a long-wave approximation. The resulting eigenvalue problem is exactly solved up to third order in the wavenumber. The anisotropic behaviour of permeability, cross-stream component of permeability, surface tension and porosity are found to have only higher-order effects on the stability characteristics of the system. On the other hand, the inhomogeneous feature in the streamwise component of permeability play a dominant role in determining the stability of the gravity-driven surface waves; as do other system parameters such as the thickness of the fluid layer relative to that of the porous layer and the Beavers–Joseph coefficient.

There exists significant demand for improved Reynolds-averaged Navier–Stokes (RANS) turbulence models that are informed by and can represent a richer set of turbulence physics. This paper presents a method of using deep neural networks to learn a model for the Reynolds stress anisotropy tensor from high-fidelity simulation data. A novel neural network architecture is proposed which uses a multiplicative layer with an invariant tensor basis to embed Galilean invariance into the predicted anisotropy tensor. It is demonstrated that this neural network architecture provides improved prediction accuracy compared with a generic neural network architecture that does not embed this invariance property. The Reynolds stress anisotropy predictions of this invariant neural network are propagated through to the velocity field for two test cases. For both test cases, significant improvement versus baseline RANS linear eddy viscosity and nonlinear eddy viscosity models is demonstrated.

An analysis of flows in converging–diverging channels has been carried out with the primary goal of identifying geometries which result in increased mixing. The model geometry consists of a channel whose walls are fitted with spanwise grooves of moderate amplitudes (up to 10 % of the mean channel opening) and of sinusoidal shape. The groove systems on each wall are shifted by half of a wavelength with respect to each other, resulting in the formation of a converging–diverging conduit. The analysis is carried out up to Reynolds numbers resulting in the formation of secondary states. The first part of the analysis is based on a two-dimensional model and demonstrates that increasing the corrugation wavelength results in the appearance of an unsteady separation whose onset correlates with the onset of the travelling wave instability. The second part of the analysis is based on a three-dimensional model and demonstrates that the flow dynamics is dominated by the centrifugal instability over a large range of geometric parameters, resulting in the formation of streamwise vortices. It is shown that the onset of the vortices may lead to the elimination of the unsteady separation. The critical Reynolds number for the vortex onset initially decreases as the corrugation amplitude increases but an excessive increase leads to the stream lift up, reduction of the centrifugal forces and flow stabilization. The flow dynamics under such conditions is again dominated by the travelling wave instability. Conditions leading to the formation of streamwise vortices without interference from the travelling wave instability have been identified. The structure and the mixing properties of the saturated states are discussed.

This article considers the interaction of a rising bubble and a sedimenting fine particle in an incompressible viscous liquid under vibrations (ultrasound). The particle is subject to Stokes, Basset and buoyancy forces, and average force due to the inhomogeneity of the pulsating field. It is shown that the main contribution to the average force is made by interference of the external field and the field caused by the monopole mode of bubble oscillations. The interaction force is the attraction of the particle to the bubble. It is found that even weak vibrations lead to considerable increase of the effective cross-section of particle capture by the bubble. The evaluation of the efficiency of the flotation process exposed to an ultrasound action is discussed.

Non-spherical particles are known to orient preferentially in near-wall turbulence, but rod-like and disk-like particles align themselves differently relative to the mean vorticity direction. To uncover the mechanism that gives rise to such preferential particle orientations in anisotropic turbulence, Lagrangian statistics from a channel-flow simulation have been analysed. Ni et al. (J. Fluid Mech., vol. 743, 2014, R3) showed that the fluid vorticity and long rods independently aligned with the Lagrangian fluid stretching direction in isotropic turbulence. Following their approach, we deduced the left Cauchy–Green strain tensor along Lagrangian trajectories of tracer spheroids in channel-flow turbulence. The results showed that the alignment of the fluid vorticity vector with the strongest Lagrangian stretching direction in the channel centre, just as in isotropic turbulence, vanished in the vicinity of the walls. The analysis revealed that the directions of the strongest Lagrangian stretching and compression in near-wall turbulence are in the streamwise and wall-normal directions, respectively. All over the channel we found that the symmetry axis of prolate spheroids aligned with the direction of strongest Lagrangian stretching whereas oblate spheroids oriented with the direction of Lagrangian compression. This finding is apparently universal since the same trends were found in highly anisotropic wall turbulence as well as in isotropic turbulence. Contrary to the prevailing view, we have shown for the first time that the preferential orientation of the symmetry axis of long rods in the streamwise direction and of flat disks in the wall-normal direction is caused by Lagrangian stretching and not by fluid rotation. This finding fills a gap in our understanding of orientation and rotation of tracer spheroids in anisotropic wall turbulence.

We analyse flow of non-Newtonian fluids in a Hele-Shaw cell, subjected to spatially non-uniform electro-osmotic slip. Motivated by their potential use for increasing the characteristic pressure fields, we specifically focus on power-law fluids with wall depletion properties. We derive a $p$-Poisson equation governing the pressure field, as well as a set of linearized equations representing its asymptotic approximation for weakly non-Newtonian behaviour. To investigate the effect of non-Newtonian properties on the resulting fluidic pressure and velocity, we consider several configurations in one and two dimensions, and calculate both exact and approximate solutions. We show that the asymptotic approximation is in good agreement with exact solutions even for fluids with significant non-Newtonian behaviour, allowing its use in the analysis and design of microfluidic systems involving electrokinetic transport of such fluids.

We study in the present work a Mach 2.5 flow over a forward-facing step. The focus of the work is the flow ahead of the step, in particular, the unsteady interactions between the shock, the boundary layer and the separation bubble. The primary geometrical parameter in the problem is the ratio of the step height to the incoming boundary layer thickness, $h/\unicode[STIX]{x1D6FF}$, which is kept fixed at 2. Results are presented from detailed particle image velocimetry (PIV) measurements in two orthogonal planes to obtain a reasonable picture of the whole flow field. The mean velocity field in the central cross-stream or wall-normal ($x$–$y$) plane shows that the incoming boundary layer separates upstream of the step forming a large separation bubble ahead of the step, which can be relatively well resolved in PIV measurements compared to the compression ramp cases. Wall pressure fluctuation spectra close to the separation location show a dominant frequency ($f$) that is two orders of magnitude smaller than the characteristic frequency of the incoming boundary layer ($U_{\infty }/\unicode[STIX]{x1D6FF}$), consistent with low-frequency motions of the shock that have received a lot of recent attention ($U_{\infty }$ $=$ free-stream velocity, $\unicode[STIX]{x1D6FF}$ $=$ boundary layer thickness). PIV measurements in the wall-normal plane show large variations in shock position with time. The shock position measured from velocity data outside the boundary layer is found to be well correlated with the reverse flow area ahead of the step, and weakly correlated to structures in the incoming boundary layer. In contrast, the shock foot, determined from velocity data within the boundary layer, is found to be well correlated to the low- and high-speed streaks in the incoming boundary layer, in addition to the reverse flow area ahead of the step. Instantaneous velocity fields in the spanwise ($x$–$z$) plane parallel to the lower wall show that the shock is broadly two-dimensional with small spanwise ripples, while the recirculation region has very large spanwise variations. The spanwise-averaged shock location is found to be well correlated to the most upstream location of the recirculation region over a spanwise length ($x_{r,min}^{sp}$). Instantaneous velocity fields show that when some part of the recirculation region is far upstream, the corresponding nearly two-dimensional shock is also far upstream. On the other hand, when $x_{r,min}^{sp}$ is relatively downstream, the resulting shock is also found to be downstream. Hence, the present results suggest that for the forward-facing step configuration, the large-scale streamwise motions of the shock are mainly correlated to the most upstream point of the recirculation region, which has large spanwise variations.

Non-canonical wakes of two-dimensional elliptic cylinders are analysed numerically for their near- and far-wake characteristics. The governing equations are solved using an immersed boundary method based projection scheme. The wakes are then classified into three distinct types according to diverse flow and thermal properties. An unexpected mean temperature evolution along the centreline of the wake is observed for certain wake states. In order to explain this unusual variation, novel heat transport models are constructed based on the vortex dynamics. These models are derived by considering vorticity is acted by flow, which has shear and swirl. Mechanisms of the primary vortex street breakdown and formation of the secondary vortex street are also proposed based on these models. A new phenomenon namely ‘dual near-wall instantaneous recirculation’ is observed, and its appearance is found to be a function of length of the primary von Kármán vortex street. The same phenomenon is also found to be responsible for the secondary peak in the Nusselt number variation along the circumference of the cylinder. Despite varied differences between the wake types, it is observed that the transitions occur through a supercritical Hopf bifurcation in all of them, at least in the von Kármán region of the wake. Low-frequency unsteadiness observed in the far wakes is examined through a signal decomposition method. Our results show that the secondary low frequency is resulting from the transition region which has a negative instability slope. Finally, onset of the primary vortex street breakdown and its scale in terms of Reynolds number is computed.

We study the dynamics of an air bubble bouncing at a liquid/liquid/gas interface, which we refer to as a compound interface. When a bubble interacts with a thin layer of oil on top of bulk water, the oil layer modifies the interfacial properties and thus the entire process of bouncing and bubble bursting. The influence on the bubble motion is experimentally and numerically investigated. Based on the coefficient of restitution and the damping rate of the bubble velocity profile, the damping increases with the oil layer thickness and viscosity. In addition, the effect of the oil layer thickness is more prominent for high-viscosity oil. Furthermore, a reduced-order mass–spring–damper model is proposed to describe the bubble bouncing at the compound interface, which predicts the time of the first contact of the bubble with the interface and agrees well with the experimental results. Such a model also captures the general experimental trends of the coefficient of restitution for the multiphase system. Our work contributes to a further understanding of the collision and coalescence of bubbles with a compound interface.

Only a few fundamental studies on the dynamics and interactions of supersonic streamwise vortices have been conducted so far despite the recognized potential of these structures to enhance supersonic mixing. In an effort to shed light on this largely unexplored field, multiple experimental campaigns were conducted in a Mach 2.5 flow to probe the dynamics of turbulence decay in complex flows originating from selected modes of supersonic streamwise vortex interaction. The first part of the manuscript presents the detailed study of two vortex interaction scenarios: one selected to obtain merging of co-rotating vortices and the other to prevent vorticity amalgamation. In the second part, data from three additional vortex merging cases are used to substantiate the findings of the first part of the study and characterize the decay of turbulence. Stereoscopic particle image velocimetry was employed to probe the resulting flow fields at different downstream stations. It was found that these complex vortex interactions measurably affect both the morphology and the magnitude of the streamwise vorticity and turbulent kinetic energy as well as the associated decays. Particularly, while the turbulent kinetic energy across each vorticity patch undergoes an initial production before decreasing monotonically in both scenarios, its content in the coalesced structure is roughly double that of the isolated vortices. The manuscript also presents the analysis of the turbulence data from 27 supersonic vortical structures differing in shape, strength and modes of interaction, acquired within a range of vortex Reynolds numbers of almost one order of magnitude. Dimensional analysis was then used to correlate the spatial decay of turbulent kinetic energy with the vortex Reynolds number. For all the cases considered here, where the fluctuating Mach number was found to be subsonic, the form of the resulting law was similar to that reported in previous scholarly publications, despite the complexity of the vortex dynamics considered in this work.

The paper investigates the dissipation tensor $\unicode[STIX]{x1D700}_{ij}$ in wall turbulence. Available direct numerical simulation (DNS) data are examined to illustrate the differences in the anisotropy of the dissipation tensor $\unicode[STIX]{x1D700}_{ij}$ with respect to the anisotropy of the Reynolds stresses $\unicode[STIX]{x1D633}_{ij}$. The budgets of the transport equations of the dissipation tensor $\unicode[STIX]{x1D700}_{ij}$ are studied using novel DNS data of low Reynolds number turbulent plane channel flow with spatial resolution sufficiently fine to accurately determine the correlations of products of two-derivatives of fluctuating velocities $u_{i}^{\prime }$ and pressure $p^{\prime }$ which appear in various terms. Finally, the influence of the Reynolds number on the diagonal components of $\unicode[STIX]{x1D700}_{ij}$ ($\unicode[STIX]{x1D700}_{xx}$, $\unicode[STIX]{x1D700}_{yy}$, $\unicode[STIX]{x1D700}_{zz}$) and on the various terms in their transport equations is studied using available DNS data of Vreman and Kuerten (Phys. Fluids, vol. 26, 2014b, 085103).

During intaglio (gravure) printing, a blade wipes excess ink from the engraved plate with the object of leaving ink-filled cells defining the image to be printed. That objective is not completely attained. Capillarity draws some ink from the cell into a meniscus connecting the blade to the substrate, and the continuing motion of the engraved plate smears that ink over its surface. By examining the limit of vanishing capillary number ($Ca$, based on substrate speed), we reduce the problem of determining smear volume to one of hydrostatics. Using numerical solutions of the corresponding free-boundary problem for the Stokes equations of motion, we show that the hydrostatic theory provides an upper bound to smear volume for finite $Ca$. The theory explains why polishing to reduce the tip radius of the blade is an effective way to control smearing.

The low-frequency unsteadiness in the direct numerical simulation of a Mach 2.9 shock wave/turbulent boundary layer interaction with mean flow separation is analysed using dynamic mode decomposition (DMD). The analysis is applied both to three-dimensional and spanwise-averaged snapshots of the flow. The observed low-frequency DMD modes all share a common structure, characterized by perturbations along the shock, together with streamwise-elongated regions of low and high momentum that originate at the shock foot and extend into the downstream flow. A linear superposition of these modes, with dynamics governed by their corresponding DMD eigenvalues, accurately captures the unsteadiness of the shock. In addition, DMD analysis shows that the downstream regions of low and high momentum are unsteady and that their unsteadiness is linked to the unsteadiness of the shock. The observed flow structures in the downstream flow are reminiscent of Görtler-like vortices that are present in this type of flow due to an underlying centrifugal instability, suggesting a possible physical mechanism for the low-frequency unsteadiness in shock wave/turbulent boundary layer interactions.

Temporal dynamics of gas-filled spherical bubbles is often described using the Rayleigh–Plesset equation, a special case of the Navier–Stokes equations that describes the oscillations of a spherical cavity in an infinite incompressible fluid. While analytical approximations and numerical simulations have previously been given in some parameter regimes, we are able to completely classify all possible dynamics exactly, in terms of only the model parameters. We present an analytical study of the solutions to the Rayleigh–Plesset equation in any number of spatial dimensions, and we demonstrate that the possible behaviours of solutions include bubbles of constant radius, bubbles with temporally oscillating radius and bubbles with finite time collapse. Each of these behaviours can be predicted solely in terms of the spatial dimension, pressures acting on the bubble and initial strain. In the case of oscillating bubbles, we give the amplitude and period of these oscillations in terms of an integral which is a function of the aforementioned parameters, while when the bubble collapses, we can similarly give the time of collapse in terms of these parameters. We give a systematic study of all possible behaviours, and capture special case solutions presented numerically or asymptotically in the literature. We also discuss the influence of both surface tension and viscosity when these terms are included in the Rayleigh–Plesset dynamics.

Hyperbolic points and their unsteady generalization – hyperbolic trajectories – drive the exponential stretching that is the hallmark of nonlinear and chaotic flow. In infinite-time steady or periodic flows, the stable and unstable manifolds attached to each hyperbolic trajectory mark fluid elements that asymptote either towards or away from the hyperbolic trajectory, and which will therefore eventually experience exponential stretching. But typical experimental and observational velocity data are unsteady and available only over a finite time interval, and in such situations hyperbolic trajectories will move around in the flow, and may lose their hyperbolicity at times. Here we introduce a way to determine their region of influence, which we term a hyperbolic neighbourhood, that marks the portion of the domain that is instantaneously dominated by the hyperbolic trajectory. We establish, using both theoretical arguments and empirical verification from model and experimental data, that the hyperbolic neighbourhoods profoundly impact the Lagrangian stretching experienced by fluid elements. In particular, we show that fluid elements traversing a flow experience exponential boosts in stretching while within these time-varying regions, that greater residence time within hyperbolic neighbourhoods is directly correlated to larger finite-time Lyapunov exponent (FTLE) values, and that FTLE diagnostics are reliable only when the hyperbolic neighbourhoods have a geometrical structure that is ‘regular’ in a specific sense.

In order to better understand the mechanism governing segregation in dense granular flows, the force experienced by a large particle embedded in a granular flow made of small particles is studied using discrete numerical simulations. Accurate force measurements have been obtained in a large range of flow parameters by trapping the large particle in a harmonic potential well to mimic an optical tweezer. Results show that positive or negative segregation lift forces (perpendicular to the shear) exist depending on the stress inhomogeneity. An empirical expression of the segregation force is proposed as a sum of a term proportional to the gradient of pressure and a term proportional to the gradient of shear stress, which both depend on the local friction and particle size ratio.