In this study, we tackle the challenge of inferring the initial conditions of a Rayleigh–Taylor mixing zone for modelling purposes by analysing zero-dimensional (0-D) turbulent quantities measured at an unspecified time. This approach assesses the extent to which 0-D observations retain the memory of the flow, evaluating their effectiveness in determining initial conditions and, consequently, in predicting the flow’s evolution. To this end, we generated a comprehensive dataset of direct numerical simulations, focusing on miscible fluids with low density contrasts. The initial interface deformations in these simulations are characterised by an annular spectrum parametrised by four non-dimensional numbers. To study the sensitivity of 0-D turbulent quantities to initial perturbation distributions, we developed a surrogate model using a physics-informed neural network (PINN). This model enables computation of the Sobol indices for the turbulent quantities, disentangling the effects of the initial parameters on the growth of the mixing layer. Within a Bayesian framework, we employ a Markov chain Monte Carlo (MCMC) method to determine the posterior distributions of initial conditions and time, given various state variables. This analysis sheds light on inertial and diffusive trajectories, as well as the progressive loss of initial conditions memory during the transition to turbulence. Furthermore, it identifies which turbulent quantities serve as better predictors of Rayleigh–Taylor mixing zone dynamics by more effectively retaining the memory of the flow. By inferring initial conditions and forward propagating the maximum a posteriori (MAP) estimate, we propose a strategy for modelling the Rayleigh–Taylor transition to turbulence.

In soft porous media, deformation drives solute transport via the intrinsic coupling between flow of the fluid and rearrangement of the pore structure. Solute transport driven by periodic loading, in particular, can be of great relevance in applications including the geomechanics of contaminants in the subsurface and the biomechanics of nutrient transport in living tissues and scaffolds for tissue engineering. However, the basic features of this process have not previously been systematically investigated. Here, we fill this hole in the context of a one-dimensional model problem. We do so by expanding the results from a companion study, in which we explored the poromechanics of periodic deformations, by introducing and analysing the impact of the resulting fluid and solid motion on solute transport. We first characterise the independent roles of the three main mechanisms of solute transport in porous media – advection, molecular diffusion and hydrodynamic dispersion – by examining their impacts on the solute concentration profile during one loading cycle. We next explore the impact of the transport parameters, showing how these alter the relative importance of diffusion and dispersion. We then explore the loading parameters by considering a range of loading periods – from slow to fast, relative to the poroelastic time scale – and amplitudes – from infinitesimal to large. We show that solute spreading over several loading cycles increases monotonically with amplitude, but is maximised for intermediate periods because of the increasing poromechanical localisation of the flow and deformation near the permeable boundary as the period decreases.

The interaction between porous structures and flows with mean and oscillatory components has many applications in fluid dynamics. One such application is the hydrodynamic forces on offshore jacket structures from waves and current, which have been shown to give a significant blockage effect, leading to a reduction in drag forces. To better understand this, we derived analytical expressions that describe the effect of current on drag forces from large waves, and conducted experiments that measured forces on a model jacket in collinear waves and currents. We utilised symmetry and phase-inversion techniques, relying on the underlying physics of wave structure interaction, to separate Morison drag and inertia-type forces and to decompose these forces into their respective frequency harmonics. We find that the odd harmonics of the drag force mostly contain the loads from waves, while even harmonics vary much more rapidly with the current speed flowing through the jacket. At the time of peak force, these current speeds were estimated to be 40 % of the undisturbed current and 50 % of the industry-standard estimates, a result that has significant implications for design and re-assessment of jackets. At times away from the peak force, when there are no waves and only current, the blockage effects are reduced. Hence, the variation in blocked current speeds appears to occur on a relatively fast time scale similar to the compact wave envelope. These findings may be generalisable to any jacket-type structure in flows with mean and high Keulegan–Carpenter number oscillatory components.

Turbidity currents, which are stratified, sediment-laden bottom flows in the ocean or lakes, can run out for hundreds or thousands of kilometres in submarine channels without losing their stratified structure. Here, we derive a layer-averaged, two-layer model for turbidity currents, specifically designed to capture long-runout. A number of previous models have captured runout of only tens of kilometres, beyond which thickening of the flows becomes excessive, and the models without a lateral overspill mechanism fail. In our framework, a lower layer containing nearly all the sediment is a faster, gravity-driven flow that propels an upper layer, where sediment concentration is nearly zero. The thickness of the lower layer is controlled by competition between interfacial water entrainment due to turbulent mixing and water detrainment due to sediment settling at the interface. The detrainment mechanism, first identified in experiments, is the key feature that prevents excessive thickening of the lower layer and allows long-runout. Under normal flow conditions, we obtain an exact solution to the two-layer formulation, revealing a constant velocity and a constant thickening rate in each of the two layers. Numerical simulations applied

to gradually varied flows on both constant and exponentially declining bed slopes, with boundary conditions mimicking field observations, show that the predicted lower layer thickness after 200 km flow propagation compares with observed submarine channel depths, whereas previous models overestimate this thickness three- to fourfold. This formulation opens new avenues for modelling the fluid mechanics and morphodynamics of long-runout turbidity currents in the submarine setting.

We investigate the effects of bottom roughness on bottom boundary-layer (BBL) instability beneath internal solitary waves (ISWs) of depression. Applying both two-dimensional (2-D) numerical simulations and linear stability theory, an extensive parametric study explores the effect of the Reynolds number, pressure gradient, roughness (periodic bump) height $h_b$ and roughness wavelength $\lambda _b$ on BBL instability. The simulations show that small $h_b$, comparable to that of laboratory-flume materials ($\sim$100 times less than the thickness of the viscous sublayer $\delta _v$), can destabilize the BBL and trigger vortex shedding at critical Reynolds numbers much lower than what occurs for numerically smooth surfaces. We identify two mechanisms of vortex shedding, depending on $h_b/\delta _v$. For $h_b/\delta _v \gtrapprox 1$, vortices are forced directly by local flow separation in the lee of each bump. Conversely, for $h_b/\delta _v \lessapprox 10^{-1}$ the roughness seeds perturbations in the BBL, which are amplified by the BBL flow. Roughness wavelengths close to those associated with the most unstable BBL mode, as predicted by linear instability theory, are preferentially amplified. This resonant amplification nature of the BBL flow, beneath ISWs, is consistent with what occurs in a BBL driven by surface solitary waves and by periodic monochromatic waves. Using the $N$-factor method for Tollmien–Schlichting waves, we propose an analogy between the roughness height and seed noise required to trigger instability. Including surface roughness, or more generally an appropriate level of seed noise, reconciles the discrepancies between the vortex-shedding threshold observed in the laboratory versus that predicted by otherwise smooth-bottomed 2-D spectral simulations.

Double-diffusive convection can arise when the fluid density is set by multiple species which diffuse at different rates. Different flow regimes are possible depending on the distribution of the diffusing species, including salt fingering and diffusive convection. Flows arising from diffusive convection commonly exhibit step-like density profiles with sharp density interfaces separated by well-mixed layers. The formation of density layers is also seen in stratified turbulence, where a framework based on sorted density coordinates (Winters & D’Asaro 1996 J. Fluid Mech. 317, 179–193) has been used to diagnose layer formation (Zhou et al. 2017 J. Fluid Mech. 823, 198–229; Taylor & Zhou 2017 J. Fluid Mech. 823, R5). In this framework, the evolution of the sorted density profile can be expressed solely in terms of the eddy diffusivity, $\kappa _e$. Here, we use the same framework to diagnose layer formation in two-dimensional simulations of double-diffusive convection. We show that $\kappa _e$ is negative everywhere, with the antidiffusive effect strongest in ‘well-mixed’ layers where a positive diffusion coefficient may be expected. By considering a decomposition of the budget of the square of the Brunt-Väisälä frequency $\partial N^2_*/\partial t$, we demonstrate that the density layers are maintained by fundamentally different processes than in single-component stratified turbulence. In more complicated flows where stratified turbulence and double-diffusive convection can coexist, this framework could provide a method to distinguish between the mechanisms responsible for generating density layers.

We study time-dependent density segregation of granular mixtures flowing over an inclined plane. Discrete element method (DEM) simulations in a periodic box are performed for granular mixtures of same size and different density particles flowing under the influence of gravity. In addition, a continuum model is developed to solve the momentum balance equations along with the species transport equation by accounting for the inter-coupling of segregation and rheology. The particle force-based density segregation theory has been used along with the $\mu {-}I$ rheology to predict evolution of flow properties with time for binary and multicomponent mixtures. The effect of particle arrangements on the transient evolution of flow properties for three different initial configurations is investigated using both continuum and DEM simulations. Continuum predictions for various flow properties of interest such as species concentration, velocity, pressure and shear stress at different time instants are compared with DEM simulations. The results from the discrete and continuum models are found to be in good agreement with each other for well-mixed and heavy-near-base initial configurations. Kinetic theory-based predictions of segregation evolution, however, show good quantitative agreement only for the heavy-near-base configuration with a much slower evolution for the well-mixed case. Interestingly, the continuum model is unable to predict the flow evolution for the light-near-base initial configuration. DEM simulations reveal the presence of an instability driven, quick segregation for this configuration which is not predicted by the one-dimensional model and requires generalisation to three dimensions.

We report an anomalous capillary phenomenon that reverses typical capillary trapping via nanoparticle suspension and leads to a counterintuitive self-removal of non-aqueous fluid from dead-end structures under weakly hydrophilic conditions. Fluid interfacial energy drives the trapped liquid out by multiscale surfaces: the nanoscopic structure formed by nanoparticle adsorption transfers the molecular-level adsorption film to hydrodynamic film by capillary condensation, and maintains its robust connectivity, then the capillary pressure gradient in the dead-end structures drives trapped fluid motion out of the pore continuously. The developed mathematical models agree well with the measured evolution dynamics of the released fluid. This reversing capillary trapping phenomenon via nanoparticle suspension can be a general event in a random porous medium and could dramatically increase displacement efficiency. Our findings have implications for manipulating capillary pressure gradient direction via nanoparticle suspensions to trap or release the trapped fluid from complex geometries, especially for site-specific delivery, self-cleaning, or self-recover systems.

A purely elastic linear instability was recently reported for viscoelastic plane Poiseuille flow in the limit of ultra-dilute (solvent to solution viscosity ratio $\beta \gt 0.99$), highly elastic (Weissenberg number $W \sim 1000$) polymer solutions, within the framework of the Oldroyd-B model (Khalid et al., Phys. Rev. Lett., vol. 127, 2021, pp. 134–502). This is the first instance of a purely elastic instability in a strictly rectilinear shearing flow, with the phase speed of the unstable ‘centre mode’ being close to the base-state maximum velocity at the channel centreline. Subsequently, Buza, Page and Kerswell (J. Fluid Mech., vol. 940, 2022, A11) have shown, using the FENE-P model, that the centre-mode instability persists down to moderate elasticities ($W \sim O (100)$), the reduction in threshold evidently due to the finite extensibility of the polymer molecules. In this work, we augment this latter finding and provide a comprehensive account of the effect of finite extensibility on the centre-mode instability in viscoelastic channel flow, using the FENE-P and FENE-CR models, in both the absence and presence of fluid inertia. In both these models, finite extensibility causes a decrease in the polymer relaxation time at high shear rates, and the resulting weakening of elastic stresses would seem to indicate a stabilising effect. The latter trend has been demonstrated by earlier analyses of hoop-stress-driven instabilities in curvilinear flows, and is indeed borne out for the FENE-CR case, where finite extensibility has a largely stabilising influence on the centre-mode instability. In stark contrast, for the FENE-P model, finite extensibility plays a dual role – a stabilising one at lower values of the elasticity number $E$, but, surprisingly, a destabilising one at higher $E$ values. Further, the centre-mode instability is predicted over a significantly larger domain of the $Re$–$E$–$\beta$ parameter space, compared to the Oldroyd-B model, making it more amenable to experimental observations.

The study examines supersonic square jets in a twin nozzle configuration with the aim of identifying and characterising emergent instability modes during overexpanded operation. Unlike screeching rectangular jets that undergo strong fluctuations normal to the wider jet dimension, the equilateral nature of the exit geometry in square nozzles leads to multiple instability states dictated by shock–turbulence interactions and nozzle operating conditions. Furthermore, strong coupling modes between the jets were identified that led to either phase locked or out of phase interactions of the inner shear layers. Results from experimental studies were examined using spatial and temporal decomposition techniques based on spectral methods to identify the resultants from triadic shock–turbulence interactions. The primary instability mode across both operating conditions were driven by optimal interactions while the harmonics were found to be associated with the suboptimal shock–turbulence interactions.

Although active flow control based on deep reinforcement learning (DRL) has been demonstrated extensively in numerical environments, practical implementation of real-time DRL control in experiments remains challenging, largely because of the critical time requirement imposed on data acquisition and neural-network computation. In this study, a high-speed field-programmable gate array (FPGA) -based experimental DRL (FeDRL) control framework is developed, capable of achieving a control frequency of 1–10 kHz, two orders higher than that of the existing CPU-based framework (10 Hz). The feasibility of the FeDRL framework is tested in a rather challenging case of supersonic backward-facing step flow at Mach 2, with an array of plasma synthetic jets and a hot-wire acting as the actuator and sensor, respectively. The closed-loop control law is represented by a radial basis function network and optimised by a classical value-based algorithm (i.e. deep Q-network). Results show that, with only ten seconds of training, the agent is able to find a satisfying control law that increases the mixing in the shear layer by 21.2 %. Such a high training efficiency has never been reported in previous experiments (typical time cost: hours).

A millimetric droplet may bounce and self-propel across the surface of a vertically vibrating liquid bath, guided by the slope of its accompanying Faraday wave field. The ‘walker’, consisting of a droplet dressed in a quasi-monochromatic wave form, is a spatially extended object that exhibits many phenomena previously thought exclusive to the quantum realm. While the walker dynamics can be remarkably complex, steady and periodic states arise in which the energy added by the bath vibration necessarily balances that dissipated by viscous effects. The system energetics may then be characterised in terms of the exchange between the bouncing droplet and its guiding or ‘pilot’ wave. We here characterise this energy exchange by means of a theoretical investigation into the dynamics of the pilot-wave system when time-averaged over one bouncing period. Specifically, we derive simple formulae characterising the dependence of the droplet’s gravitational potential energy and wave energy on the droplet speed. Doing so makes clear the partitioning between the gravitational, wave and kinetic energies of walking droplets in a number of steady, periodic and statistically steady dynamical states. We demonstrate that this partitioning depends exclusively on the ratio of the droplet speed to its speed limit.

A single particle representation of a self-propelled microorganism in a viscous incompressible fluid is derived based on regularised Stokeslets in three dimensions. The formulation is developed from a limiting process in which two regularised Stokeslets of equal and opposite strength but with different size regularisation parameters approach each other. A parameter that captures the size difference in regularisation provides the asymmetry needed for propulsion. We show that the resulting limit is the superposition of a regularised stresslet and a potential dipole. The model framework is then explored relative to the model parameters to provide insight into their selection. The particular case of two identical particles swimming next to each other is presented and their stability is investigated. Additional flow characteristics are incorporated into the modelling framework with in the addition of a rotlet double to characterise rotational flows present during swimming. Lastly, we show the versatility of deriving the model in the method of regularised Stokeslets framework to model wall effects of an infinite plane wall using the method of images.

The hydrodynamic analysis of motion of small particles (e.g. proteins) within lipid bilayers appears to be naturally suitable for the framework of two-dimensional Stokes flow. Given the Stokes paradox, the problem in an unbounded domain is ill-posed. In his classical paper, Saffman (J. Fluid Mech., vol. 73, 1976, pp. 593–602) proposed several possible remedies, one of them based upon the finite extent of the membrane. Considering a circular boundary, that regularisation was briefly addressed by Saffman in the isotropic configuration, where the particle is concentrically positioned in the membrane. We investigate here the hydrodynamic problem in bounded membranes for the general case of eccentric particle position and a rectilinear motion in an arbitrary direction. Symmetry arguments provide a representation of the hydrodynamic drag in terms of ‘radial’ and ‘transverse’ coefficients, which depend upon two parameters: the ratio $\lambda$ of particle to membrane radii and the eccentricity $\beta$. Using matched asymptotic expansions we obtain closed-form approximations for these coefficients in the limit where $\lambda$ is small. In the isotropic case ($\beta = 0$) we find that the drag coefficient is $4\pi /(\ln ({1}/{\lambda })- {1})$, contradicting the value $4\pi /(\ln ({1}/{\lambda })- {1}/{2})$ obtained by Saffman. We explain the oversight in Saffman’s argument.

We conducted a series of pore-scale numerical simulations on convective flow in porous media, with a fixed Schmidt number of 400 and a wide range of Rayleigh numbers. The porous media are modeled using regularly arranged square obstacles in a Rayleigh–Bénard (RB) system. As the Rayleigh number increases, the flow transitions from a Darcy-type regime to an RB-type regime, with the corresponding $Sh$–$Ra_D$ relationship shifting from sublinear scaling to the classical 0.3 scaling of RB convection. Here, $Sh$ and $Ra_D$ represent the Sherwood number and the Rayleigh–Darcy number, respectively. For different porosities, the transition begins at approximately $Ra_D = 4000$, at which point the characteristic horizontal scale of the flow field is comparable to the size of a single obstacle unit. When the thickness of the concentration boundary layer is less than approximately one-sixth of the pore spacing, the flow finally enters the RB regime. In the Darcy regime, the scaling exponent of $Sh$ and $Ra_D$ decreases as porosity increases. Based on the Grossman–Lohse theory (J. Fluid Mech. vol. 407, 2000, pp. 27–56; Phys. Rev. Lett. vol. 86, 2001, p. 3316), we provide an explanation for the scaling laws in each regime and highlight the significant impact of mechanical dispersion effects during the development of the plumes. Our findings provide some new insights into the validity range of the Darcy model.