In this work, we perform weakly nonlinear analysis of the reflection process of a thin oscillating wave beam on a non-critical surface in a fluid rotating and stratified along the same vertical axis in the limit of weak viscosity, i.e. small Ekman number $E$. We assume that the beam has the self-similar viscous structure obtained by Moore & Saffman (Phil. Trans. R. Soc. A, vol. 264, 1969, pp. 597–634) and Thomas & Stevenson (J. Fluid Mech., vol. 54, 1972, pp. 495–506). Such a solution describes the viscous internal shear layers of width $O(E^{1/3})$ generated by a localized oscillating source. We first show that the reflected beam conserves at leading order the self-similar structure of the incident beam and is modified by an $O(E^{1/6})$ correction with a different self-similar structure. We then analyse the nonlinear interaction of the reflected beam with the incident beam of amplitude $\varepsilon$ and demonstrate that a second-harmonic beam and localized meanflow correction, both of amplitude $\varepsilon ^{2} E^{-1/3}$, are created. We further show that for the purely stratified case (respectively the purely rotating case), a non-localized meanflow correction of amplitude $\varepsilon ^{2} E^{-1/6}$ is generated, except when the boundary is horizontal (respectively vertical). In this latter case, the meanflow correction remains localized but exhibits a triple-layer structure with a large $O(E^{4/9})$ viscous layer.

The oscillatory Kelvin–Helmholtz (K–H) instability of a planar liquid sheet was experimentally investigated in the presence of an axial oscillating gas flow. An experimental system was initiated to study the oscillatory K–H instability. The surface wave growth rates were measured and compared with theoretical results obtained using the authors’ early linear method. Furthermore, in a larger parameter range experimentally studied, it is interesting that there are four different unstable modes: first disordered mode (FDM), second disordered mode (SDM), K–H harmonic unstable mode (KHH) and K–H subharmonic unstable mode (KHS). These unstable modes are determined by the oscillating amplitude, oscillating frequency and liquid inertia force. The frequencies of KHH are equal to the oscillating frequency; the frequency of KHS equals half the oscillating frequency, while the frequencies of FDM and SDM are irregular. By considering the mechanism of instability, the instability regime maps on the relative Weber number versus liquid Weber number (Werel–Wel) and the Weber number ratio versus the oscillating frequency (Werel/Wel–$\varOmega$s2) were plotted. Among these four modes, KHS is the most unexpected: the frequency of this mode is not equal to the oscillating frequency, but the surface wave can also couple with the oscillating gas flow. Linear instability theory was applied to divide the parameter range between the different unstable modes. According to linear instability theory, K–H and parametric unstable regions both exist. However, note that all four modes (KHH, KHS, FDM and SDM) corresponded primarily to the K–H unstable region obtained from the theoretical analysis. Nevertheless, the parametric unstable mode was also observed when the oscillating frequency and amplitude were relatively low, and the liquid inertia force was relatively high. The surface wave amplitude was small but regular, and the evolution of this wave was similar to that of Faraday waves. The wave oscillating frequency was half that of the surface wave.

Because the Navier–Stokes equations are dissipative, the long-time dynamics of a flow in state space are expected to collapse onto a manifold whose dimension may be much lower than the dimension required for a resolved simulation. On this manifold, the state of the system can be exactly described in a coordinate system parameterising the manifold. Describing the system in this low-dimensional coordinate system allows for much faster simulations and analysis. We show, for turbulent Couette flow, that this description of the dynamics is possible using a data-driven manifold dynamics modelling method. This approach consists of an autoencoder to find a low-dimensional manifold coordinate system and a set of ordinary differential equations defined by a neural network. Specifically, we apply this method to minimal flow unit turbulent plane Couette flow at $Re=400$, where a fully resolved solutions requires ${O}(10^5)$ degrees of freedom. Using only data from this simulation we build models with fewer than $20$ degrees of freedom that quantitatively capture key characteristics of the flow, including the streak breakdown and regeneration cycle. At short times, the models track the true trajectory for multiple Lyapunov times and, at long times, the models capture the Reynolds stress and the energy balance. For comparison, we show that the models outperform POD-Galerkin models with $\sim$2000 degrees of freedom. Finally, we compute unstable periodic orbits from the models. Many of these closely resemble previously computed orbits for the full system; in addition, we find nine orbits that correspond to previously unknown solutions in the full system.

Pulsatile rough-wall turbulent pipe flow is compared against its non-pulsatile counterpart using data obtained from direct numerical simulation. Results are presented at a mean friction Reynolds number of 540 for a set of three geometrically scaled roughness topographies at a single forcing condition, which, based on existing classifications, falls into the current-dominated very-high-frequency regime. By comparing the pulsatile data against an equivalent non-pulsatile dataset (Chan et al., J. Fluid Mech., vol. 854, 2018, pp. 5–33), the key differences (and similarities) between the forced and unforced configurations are identified. A major finding of this study is that the flow in the outer region retains its self-similar functional form under pulsatile rough-wall conditions, and, as a result, Townsend’s outer-layer similarity hypothesis holds for the roughness-forcing combinations considered here. On the other hand, the unsteady cases exhibit a rich array of flow physics in the region beneath the roughness crests not observed in the steady case. These differences are examined using a Moody chart, which encapsulates how the hydraulic properties of pulsatile rough-wall pipe flow differ from their non-pulsatile counterpart.

The purpose of the paper is to identify Mach-number effects on pressure fluctuations $p'$ in compressible turbulent plane channel flow. We use data from a specifically constructed $(Re_{\tau ^\star },\bar {M}_{{CL}_x})$-matrix direct numerical simulation (DNS) database, with systematic variation of the centreline streamwise Mach number $0.32\leqslant \bar {M}_{{CL}_x}\leqslant 2.49$ and of the HCB (Huang et al., J. Fluid Mech., vol. 305, 1995, pp. 185–218) friction Reynolds number $66\leqslant Re_{\tau ^\star }\lessapprox 1000$. Strong $\bar {M}_{{CL}_x}$ effects (enhanced by the increasingly cold-wall condition) appear for $\bar {M}_{{CL}_x}\gtrapprox 2$, for all $Re_{\tau ^\star }$, very close to the wall ($y^\star \lessapprox 15$). Compared with incompressible flow at the same $Re_{\tau ^\star }$, the wall root-mean-square $[p'_{rms}]^+_w$ (in wall-units, i.e. scaled by the average wall shear stress $\bar {\tau }_w$) strongly increases with $\bar {M}_{{CL}_x}$. In contrast, the peak level across the channel, $[p'_{rms}]^+_{PEAK}$, slightly decreases with increasing $\bar {M}_{{CL}_x}$. In order to study the near-wall coherent structures we introduce a new wall-distance-independent non-local system of units, based for all $y$ on wall friction and the extreme values of density and dynamic viscosity, namely, for cold walls $\{\bar {\tau }_w,\min _y\bar {\rho },\max _y\bar {\mu }\}$. The average spanwise distance between streaks, scaled by this length-unit, is nearly independent of $\bar {M}_{{CL}_x}$ at constant $Re_{\tau ^\star }$. Using the in-plane (parallel to the wall) Laplacian $\nabla ^2_{xz}p'$ we find that the $(+/-)\text {-}p'$ wave-packet-like structures appearing inside the low-speed streaks ($y^\star \lessapprox 15$) with increasing $\bar {M}_{{CL}_x}\gtrapprox 2$ are part of a more complex wave system with spanwise extent over several streaks, whose spatial density decreases rapidly with decreasing $\bar {M}_{{CL}_x}$ or increasing $y^\star$. These $p'$ wave packets appear to be collocated with strong $(+/-)$-$v'$ events and could be responsible for compensating towards 0 the negative incompressible-flow correlation coefficient $c_{p'v'}$, with increasing $\bar {M}_{{CL}_x}$ very near the wall.

This investigation characterises the time response and the transient turbulence dynamics undergone by rapidly decelerating turbulent pipe flows. A series of direct numerical simulations of decelerating flows between two steady Reynolds numbers were conducted for this purpose. The statistical analyses reveal that rapidly decelerating turbulent flows undergo four coherent, unambiguous transitional stages: inertial (stage I), a dramatic change of sign in the viscous force associated with the decay of the viscous shear stress at the wall together with a mild turbulence decay in the viscous sublayer; friction recovery (stage II), a recovery in viscous force and progressive decay in the turbulent inertia at the near-wall region; turbulence decay (stage III), a balanced decay in both turbulent inertia and viscous force at the near-wall and overlap regions; core relaxation (stage IV), slow turbulence decay at the core region. The FIK identity derived by Fukagata, Iwamoto and Kasagi (Phys. Fluids, vol. 14, 2002, L73–L76) was used to understand further how the flow dynamics influence the time response of the skin friction coefficient ($C_f$). The results show that although $C_f$ plateaus during the fourth stage, the turbulent contribution keeps decaying, undershoots and finally recovers to attain its final steady value. The time evolution of the azimuthal vorticity ($\omega _\theta$) flux reveals that as the flow is decelerated, a layer of negative $\omega _\theta$ is produced at the wall during the flow excursion. As time progresses, this negative vorticity propagates in the wall-normal direction, attenuating the pre-existing vorticity and producing a decay in the turbulence levels.

Ensemble data assimilation methods integrate measurement data and computational flow models to estimate the state of fluid systems in a robust, scalable way. However, discretization errors in the dynamical and observation models lead to biased forecasts and poor estimator performance. We propose a low-rank representation for this bias, whose dynamics is modelled by data-informed, time-correlated processes. State and bias parameters are simultaneously corrected online with the ensemble Kalman filter. The proposed methodology is then applied to the problem of estimating the state of a two-dimensional flow at modest Reynolds number using an ensemble of coarse-mesh simulations and pressure measurements at the surface of an immersed body in a synthetic experiment framework. Using an ensemble size of 60, the bias-aware estimator is demonstrated to achieve at least 70 % error reduction when compared to its bias-blind counterpart. Strategies to determine the bias statistics and their impact on the estimator performance are discussed.

Nonlinear machine learning for turbulent flows can exhibit robust performance even outside the range of training data. This is achieved when machine-learning models can accommodate scale-invariant characteristics of turbulent flow structures. This study presents a data-driven approach to reveal scale-invariant vortical structures across Reynolds numbers that provide insights for supporting nonlinear machine-learning-based studies of turbulent flows. To uncover conditions for which nonlinear models are likely to perform well, we use a Buckingham-Pi-based sparse nonlinear scaling to find the influence of the Pi groups on the turbulent flow data. We consider nonlinear scalings of the invariants of the velocity gradient tensor for an example of three-dimensional decaying isotropic turbulence. The present scaling not only enables the identification of vortical structures that are interpolatory and extrapolatory for the given flow field data but also captures non-equilibrium effects of the energy cascade. As a demonstration, the present findings are applied to machine-learning-based super-resolution analysis of three-dimensional isotropic turbulence. We show that machine-learning models reconstruct vortical structures well in the interpolatory space with reduced performance in the extrapolatory space revealed by the nonlinearly scaled invariants. The present approach enables us to depart from labelling turbulent flow data with a single parameter of Reynolds number and comprehensively examine the flow field to support training and testing of nonlinear machine-learning techniques.

Following the findings by Wangsawijaya et al. (J. Fluid Mech., vol. 894, 2020, A7), we re-examine the turbulent boundary layers developing over surfaces with spanwise heterogeneous roughness of various roughness half-wavelengths $0.32 \leq S/\bar {\delta } \leq 3.63$, where $S$ is the width of the roughness strips and $\bar {\delta }$ is the spanwise-averaged boundary-layer thickness. The heterogeneous cases induce counter-rotating secondary flows, and these are compared with the large-scale turbulent structures that occur naturally over the smooth wall. Both appear as meandering elongated high- and low-momentum streaks in the instantaneous flow field. Results based on the triple decomposed velocity fluctuations suggest that the secondary flows are spanwise-locked turbulent structures, with $S/\bar {\delta }$ governing the strength of the turbulent structures and the efficacy of the surface in locking the structures in place (most effective when $S/\bar {\delta } \approx 1$). In terms of unsteadiness, we find additional evidence from conditional averages of the fluctuating velocity fields showing that the secondary flows exhibit maximum unsteadiness (or meandering) when $S/\bar {\delta } \approx 1$. The conditional averages of both spanwise heterogeneous and smooth-wall cases result in structures that are reminiscent of those proposed for the streak-vortex instability model for the inner cycle of wall-bounded turbulence. However, in this case these structures are larger and do not necessarily share the same formation mechanism with the inner cycle. Secondary flows and large-scale structures coexist in the limits where either $S/\bar {\delta } \gg 1$ or $S/\bar {\delta } \ll 1$, where the secondary flows scale on $\delta$ or $S$, respectively. When $S/\bar {\delta } \gg 1$, the secondary flows are locked about the roughness transition, while relatively unaltered large-scale structures occur further from the transition. In the case where $S/\bar {\delta } \ll 1$, $S$-scaled secondary flows are confined close to the surface, coexisting with unaltered larger-scale turbulent structures that penetrate much deeper into the layer.

The present work investigates the mechanical behaviour of finite-size, elastic and inertial fibres freely moving in a homogeneous and isotropic turbulent flow at moderate Reynolds number. Four-way coupled, direct numerical simulations, based on a finite difference discretisation and the immersed boundary method, are performed to mutually couple the dynamics of fibres and fluid turbulence, allowing us to account for the backreaction of the dispersed phase to the carrier flow. An extensive parametric study is carried out in zero-gravity condition over the characteristic properties of the suspension, i.e. fibre's linear density (from iso-dense to denser-than-the-fluid fibres), length (from short fibres comparable with the dissipative scale to long fibres comparable with the integral scale) and bending stiffness (from highly flexible to almost rigid fibres), as well as the concentration (from dilute to non-dilute suspensions). Results reveal the existence of a robust turbulence modulation mechanism that is primarily controlled by the mass fraction of the suspension (with only a minor influence of the fibre's bending stiffness), which is characterised in detail by means of a scale-by-scale analysis in Fourier space. Despite such alteration with respect to the single-phase case due to the non-negligible backreaction, fibres experience only two possible flapping states (previously identified in the very dilute condition) while being transported and deformed by the flow. In addition, we show that the maximum curvature obeys different scaling laws that can be derived from the fibre dynamical equation. Finally, we explore the clustering and preferential alignment of fibres within the flow, highlighting the peculiar role of inertia and elasticity.

Upon drop impact on a surface of comparable size to that of the drop, a sheet is produced that evolves freely in the air, bounded by a rim from which ligaments and droplets are continuously shed. This process is a canonical example of unsteady sheet fragmentation. The sheet dynamics is coupled with continuous ligament and drop shedding (Wang & Bourouiba, J. Fluid Mech., vol. 848, 2018b, 946–967; Wang & Bourouiba, J. Fluid Mech., vol. 910, 2021b, A39) and is governed by a nonlinear non-Galilean Taylor–Culick law (Wang & Bourouiba, 2022 (in press)). Here, we report the results of a combined theoretical and experimental study of the partition and temporal evolution of mass, momentum and energy in each part of the system composed of sheet, rim, ligaments and drops. We elucidate and derive analytical predictions, without fitting parameters, of the temporal evolution of the fractions of volume/mass, momentum and energy in each sub-part of the system: from sheet, to rim to fluid shed. We show that their temporal evolution and partitioning are independent of impact conditions. Interestingly this implies, for example, that the fraction of initial drop volume shed from an impacting drop is independent of the initial energy (or Weber number) of impact. We validate our predictions against precise measurements. Finally, we show that the partition laws for this unsteady fragmentation system are robust to changes of fluid properties (viscosity, surface tension and density). We provide the ranges of validity of our partition law on a Weber–Reynolds numbers regime map.

The atomization of a water column by a gas jet flow (Reynolds number $\sim O(10^{4}\unicode{x2013}10^{5})$) issued from a two-stage annular nozzle is investigated experimentally. Varying the nozzle geometry, the momentum flux ratio of the upper and lower jets, and the water flow rate, we measure the processes of atomization with high-speed imaging, analysed analytically into four regimes. In the bulk atomization regime, the atomization is driven by the lower jet, but it is forced to occur earlier by the stronger upper jet before the water column reaches the lower jet in the droplet atomization regime. Interestingly, the size of the atomized droplets remains unaffected by the momentum flux ratio of upper to lower jets. The atomization process is governed by the Rayleigh–Taylor instability, by which the estimated droplet size agrees well with the measurement. In the backflow regime, a strong reverse flow is induced to force a substantial portion of atomized droplets to be drawn backward to the nozzle; a floating liquid column regime is captured transitionally, i.e. the column stagnates near the lower nozzle when the water flow rate is very low. To understand the mechanisms of each regime, the single-phase jet flow is measured separately using particle image velocimetry, and implemented into the control volume analysis with which we predicted analytically and validated the conditions for the occurrence of each regime. It is found that the acceleration of gas flow (velocity gradient) experienced by the falling water is the key parameter to drive the atomization.

We study a fifty-year-old problem of fast acoustic streaming, that is, the generation of moderate or large hydrodynamic Reynolds number ($\textit {Re}$) acoustic streaming (or steady flow) by the convection of momentum in an acoustic wave (or another periodic flow), while the latter is simultaneously altered by the former. The intrinsic disparity of length and time scales makes a brute-force solution of the full Navier–Stokes and continuity equations a formidable problem. Circumventing this difficulty, we split the problem into a time-averaged system of equations for the steady flow component and a dynamic system of equations for its quasi-periodic flow counterpart. The latter system of equations is obtained by subtracting the time-averaged Navier–Stokes equation from its original dynamic form, and is rendered a nonlinear wave equation using the continuity equation and an adiabatic connection between density and pressure. The resulting equations are compatible with the theory by Eckart for small $\textit {Re}$ flow, and capture large-$\textit {Re}$ effects. Scaling analysis and a case study show that acoustic streaming is weak and does not contribute to the acoustic wave close to the wave source, relevant to many microfluidic systems. At small $\textit {Re}$, the streaming magnitude is proportional to an inverse Strouhal number, a small quantity in experiments. Moderate and large $\textit {Re}$ render the streaming magnitude comparable to the pre-attenuating periodic flow (or particle velocity of the wave) at approximately a wave attenuation length away from the wave source or further; the wave is altered by the streaming that it generates, and the streaming dominates the flow far from the wave source.