The viscoelastic instability of cross-slot flow was first observed experimentally almost half a century ago and reproduced numerically two decades ago, yet its physical origin remains unresolved. We revisit this problem for two-dimensional creeping flow of Oldroyd-B fluid by combining direct numerical simulations, global stability analysis, structural sensitivity analysis and energy-budget analysis. Our simulations reproduce the canonical pitchfork bifurcation, and the stability analysis consistently predicts the threshold and perturbation growth rates. The leading eigenmode consists of a chiral velocity–stress perturbation that tilts and rotates the birefringent strand generated by the extensional flow. Structural sensitivity and energy-budget analyses identify narrow high-extension-rate ridges within the extensional flow as both the spatial core and energetic source of the instability. In these ridges, the stress-based wavemaker colocalizes with large positive disturbance polymeric stress power density, indicating localized transfer of stored elastic energy to the disturbance flow. Analyses of cross-slot variants with rounded corners and with a centred cylinder further reveal that neither sharp corners nor a free central stagnation point is the essential destabilizing ingredient; rather, the instability originates from elastic-energy release in extension-dominated regions characteristic of cross-slot flow.

The stable thermohaline stratification of the ocean gives rise to complex generation, propagation and evolution characteristics of internal wave wakes produced by moving bodies. This study proposes a pycnocline injection approach to construct a stratified flow field representative of typical oceanic pycnoclines and develops a high-precision planar array conductivity measurement system to systematically capture the spatiotemporal evolution of internal wave amplitudes induced by a towed sphere. Using statistical analysis, spectral decomposition and filtering techniques, three distinct internal wave components with different generation mechanisms and propagation behaviours are identified. Their dominant Froude number ranges and spatial distribution characteristics are further clarified. Focusing on the predominant component at high Froude numbers, collapse-stage internal waves, this study examines how key parameters, including waveform, propagation speed, wavelength and wave amplitude, vary with the Froude number. Moreover, by incorporating oscillating-source theory, it is shown that the characteristic semicircular or circular concentric patterns of collapse-stage internal waves can be reasonably explained by pulsating disturbances in the turbulent wake and the resulting vertical restoring motions in the stratified fluid.

Sediment transport caused by large earthquakes produces distinctive deposits that help extend our records of these events. To understand the underlying mechanisms of seismically induced sediment transport, we perform particle-resolved simulations of sediment transport over vibrating beds, with earthquake vibrations modelled as sinusoidal horizontal and vertical oscillations. The simulations fully resolve both the carrier flow and the motion of individual grains, and phase-averaged statistics are used to characterise how oscillations modulate turbulence, stress and sediment motion. Analysis of the sediment-transport rate and its phase-averaged decomposition shows a rhythmic response to oscillations, enhanced transport, changes in the thickness of the mobile layer and episodes of bidirectional motion occurring during specific phases of the forcing. Our findings on the influence of oscillation conditions on sediment motions provide mechanistic insight in support of the idea that large earthquakes can induce sediment transport even on flat surfaces.

The spectral budget of the vertical heat flux $\langle w\theta \rangle$ is investigated in turbulent Rayleigh–Bénard convection by direct numerical simulation. Eight cases spanning three Rayleigh numbers ($\textit{Ra} = 1\times 10^5$, $6\times 10^6$ and $4\times 10^8$) and up to three geometric aspect ratios each are computed at $\textit{Pr} = 1$. The premultiplied spectra of $\langle w\theta \rangle$ reveal a two-peak structure when both $\textit{Ra}$ and $\varGamma$ are sufficiently large. Production of $\langle w\theta \rangle$ is dominated by temperature fluctuations; production by velocity fluctuations is confined to the near-wall region and the large-scale production that develops at sufficient $\varGamma$ is restricted to the bulk. At large $\varGamma$ an inverse inter-scale cascade transfers $\langle w\theta \rangle$ from intermediate to superstructure scales, the transfer towards the superstructure scale taking place away from the boundary layer. This upscale transfer is specific to the covariance, as the vertical kinetic energy and the temperature variance cascade towards small scales, so that the turbulence selectively transfers the correlation between vertical velocity and temperature towards the superstructure scale, consistent with the strong large-scale coherence of the superstructures. Pressure transport is the only mechanism that delivers large-scale $\langle w\theta \rangle$ towards the wall with appreciable magnitude, yet the pressure–strain term destroys the transported quantity so that the net pressure–temperature interaction is purely negative at all scales. No mechanism in the budget delivers large-scale spectral content into the boundary layer, providing a scale-by-scale explanation for the insensitivity of the Nusselt number to the aspect ratio beyond a moderate threshold.

An innovative air-layer establishment method is developed and experimentally investigated, where superhydrophobic transverse grooves are introduced to modulate localised minimal-flux ventilation. The results demonstrate that, owing to the modulation by the transverse grooves, the characteristics of multiphase flow undergo a fundamental transition from discrete bubbles to a continuous air layer. This modulation is strictly governed by both the free-stream velocity and the groove size, expressed as Reynolds number ($\textit{Re}_D$) and dimensionless groove size ($\delta ^*$), respectively. With a systematic variation in $\textit{Re}_D$ and $\delta ^*$, the macroscopic topology of the air layer evolves through three distinct regimes: (i) continuous air layer (CAL), (ii) partial air layer (PAL) and (iii) bubble shedding (BS). The boundaries for these regimes are delineated within the regime map constructed by $\textit{Re}_D$ and $\delta ^*$. It is established that size effects are negligible for the CAL–PAL transition, whereas the transition from PAL to BS exhibits a pronounced dependence on $\delta ^*$. The two regime transitions therefore follow distinct scaling relations. Neither transition boundary shows a discernible dependence on the solid fraction of the grooves. The dependence of regime transitions on ventilation flux is further examined, and the Weber number ($We_j$) is introduced to characterise the competition of the dynamic pressure of injected gas and the capillary force within grooves. The drag-reduction effectiveness of the groove-modulated air layers was also demonstrated. The present findings elucidate the mechanisms of air-layer establishment and provide a robust framework for the design of superhydrophobic grooved surfaces in air-layer drag-reduction applications.

The dynamics of laterally unconstrained flexible filaments in an inclined uniform flow is investigated numerically, inspired by the self-propelled behaviour of droplets and filaments in a crosswind. It is found that in addition to conventional downwind drifting, the filament can also undergo counterintuitive upwind motion, arising from a supercritical pitchfork bifurcation. Depending on the Reynolds number, bending stiffness and inflow inclination angle, the filament exhibits multiple wake and motion states, including static, quasi-static, vibrational and translational modes. In the reference frame translating laterally with the incoming flow, the upwind and downwind drifting states induced by different initial perturbations form a pair of dynamically equivalent, symmetry-related branches. Force decomposition shows that both the streamwise drag and lateral thrust predominantly originate from the normal force, and the corresponding scaling relations are identified. Stability analysis further reveals that the hydrodynamic forces near the equilibrium states behave as restoring forces. The resulting stable equilibria can be characterised by a generalised hydrodynamic potential, whose potential wells correspond to the stable states of the system. These findings provide insight into the dynamics of flexible structures in uniform flow.

We investigate the gas flow induced in a closed, two-dimensional square cavity by imposed differences in the temperatures of its walls. The problem is studied over the entire range of gas rarefaction rates. Building upon the known free-molecular description, we derive an asymptotic solution at high yet finite Knudsen numbers. The latter is derived based on the Bhatnagar–Gross–Krook (BGK), Shakhov and full hard-sphere (HS) Boltzmann collision operator. These are complemented by direct simulation Monte Carlo and particle scheme calculations at arbitrary rarefaction rates, together with a thermodynamic ‘pure conduction’ continuum solution at vanishing $\textit{Kn}$. The gas is found at an expectedly macroscopically quiescent state in both limiting states: the collisionless free-molecular limit, where heat is transported ballistically, and the leading-order Navier–Stokes–Fourier continuum limit, in which heat is transported by pure conduction. At large yet finite Knudsen numbers, however, even a small number of molecular collisions suffices to generate vortical flow, which magnitude vanishes as the free-molecular limit is approached. Analysis of the BGK asymptotic correction identifies the microscopic origin and rotation direction of this circulation. The asymptotic results are found in excellent agreement with the numerical calculations through $O(1)$ Knudsen numbers. The variation of the gas total kinetic energy with gas rarefaction is analysed, exhibiting a maximum at $\sim O(10^{-1})$ Knudsen numbers. The maximum location and magnitude are quantified. The quantitative differences between the HS, BGK and Shakhov model predictions are discussed.

The wake dynamics and regime transitions of a streamwise-rotating angular particle in an inertial flow at moderate Reynolds numbers are investigated by particle-resolved direct numerical simulations, with particular focus on a rotating cube. We first identify the mechanisms of vortex generation on the front edges and the rear face, showing that rotation-induced vortex merging governs several key regime transitions. A previously unreported spearhead regime is uncovered, characterised by a helical near-wake and a steady filamentary far-wake structure. We demonstrate that it originates from the partial merging of edge vortices and the central rotation-induced vortex. Streamwise rotation is found to moderately stabilise the wake, and to induce a quadratic scaling of both the drag coefficient $\overline {C_d}$ and the streamwise torque $\overline {T_x}$ with the rotation rate $\varOmega$. The evolution of the lift coefficient closely mirrors that of the recirculation length $L_r$. For the rotating cube, the streamwise torque data collapse onto a single curve when expressed in terms of the rotation Reynolds number ${\textit{Re}}_{\varOmega }$. Frequency analyses reveal that the fundamental transverse-force frequency is set by the number of rotational symmetry folds $n_{\varOmega }$, while higher harmonics reflect the nonlinear nature of the flow–particle interaction. We further extend the analysis to general angular particles in their dual angular positions. Particles in dual configurations (sRCV, denoting a rotating cube with one vertex facing the incoming flow, and sROF, denoting a rotating octahedron with a front face facing the incoming flow) can display nearly identical regime transitions, as they share the same rotational symmetry folds and comparable vortex-pair strengths. Hydrodynamic responses remain strongly dependent on particle shape and orientation. Their frequency signatures likewise collapse when organised by $n_{\varOmega }$.

Linear-model-based estimation provides an equation-based route to reconstruct turbulent fluctuations from limited measurements, but its performance in compressible turbulent boundary layers depends critically on both the inter-layer coherence and the modelling of nonlinear forcing. This study systematically extends and assesses linear-model-based estimation for velocity and temperature fluctuations in compressible turbulent boundary layers, using a logarithmic-region measurement plane at Mach numbers $M=2$, 4 and 6 under both adiabatic- and cold-wall conditions. With the direct numerical simulation databases, the linear coherence spectra are first analysed. Results show that the recoverable information is mainly confined to large scales especially when the estimation plane is located near the wall, whereas the coherent scale range broadens markedly as the estimation location moves towards the measurement plane. The temperature coherence exhibits stronger sensitivity to Mach number and wall thermal condition than the streamwise velocity coherence, especially in cold-wall cases where the non-monotonic mean temperature profile alters the inter-layer thermal correlation. Based on these coherence characteristics, large-scale velocity and temperature snapshots are reconstructed using three levels of linear-model-based estimation. A controlled hierarchy comprising the baseline linear model with white forcing, an eddy-viscosity-corrected operator and a combined eddy-viscosity and forcing model is used to distinguish the effects of operator correction and forcing closure. Finally, a single-plane-calibrated estimation of wall-normal fluctuation-energy profiles is performed using the response cross-spectral density estimated by the Lyapunov formulation. The principal physical insight is that inter-layer coherence determines which scales are recoverable, the eddy-viscosity correction primarily improves the reconstructed response structure and the forcing model modifies the fluctuation amplitude and wall-normal energy redistribution. These roles are distinct in instantaneous and statistical reconstruction and are particularly important for temperature fluctuations under cold-wall conditions.

We study stirring and dispersion in steady flow through two-dimensional heterogeneous porous media. A fluid is stirred by the porous medium structure that leads to spatial flow variability and the deformation of fluid elements. While dispersion measures the extension of a solute distribution, stretching underpins solute mixing, chemical reactions and biological activity. Despite this, much less is known about fluid stretching in porous media. Its relation to the dispersion process on the one hand and the flow and medium properties on the other remain open questions. To close this gap, we probe the heterogeneity mechanisms that are at the origin of stretching and dispersion of fluid elements, and quantify the upscaled laws that govern their dynamics. Thus we perform detailed numerical simulations of Darcy flow in two-dimensional porous media characterised by a spatially varying hydraulic conductivity. Dispersion is analysed in terms of first-passage time distributions of fluid elements and the temporal evolution of their displacement mean and variance, while deformation is studied in terms of the probability density function and average elongation of fluid elements. We find that the dispersion and stretching behaviours are closely related. The exponents that govern the decay of the first-passage time distributions and evolution of the displacement variances also determine the evolution of the mean elongation. This dynamics is quantified in a single upscaled theoretical framework that bridges between Eulerian flow structure, large-scale dispersion and stretching of fluid elements.

Flow separation in rocket nozzles generates intense side loads that can lead to structural damage affecting space flight safety. Recent studies have hypothesised that a screech-like resonance loop may be responsible for such oscillations in the free shock separation (FSS) regime. In this work, we aim to confirm if screech is indeed present in this flow. Five thrust-optimised parabolic nozzles with different length and area ratios were examined at a nozzle pressure ratio range where FSS was expected to occur. Spectral proper orthogonal decomposition (SPOD) revealed resonance for both $m=0$ and $\,m=1$ azimuthal modes. Further decomposition of the SPOD modes allowed for the identification of downstream- and upstream-travelling waves, whose characteristics match the description of the Kelvin–Helmholtz (KH) instability wave and the guided jet mode (GJM), respectively. Linear stability analysis performed on Reynolds-averaged Navier–Stokes mean-flow fields confirms the existence of these two waves for the conditions studied and provides insight into their dispersion relation. A well-validated screech predictive model was employed and accurately predicted the resonant frequencies. The accurate predictions, combined with the observation of the KH and GJM waves, confirm that screech is active in this flow.

A novel experimental finding of vortex splitting and merging (VSM) route to turbulence is reported in the counter-rotation regime of the Taylor–Couette flow of a Newtonian fluid, and the role of neutrally buoyant inertial particles on this transition pathway is analysed up to a particle volume fraction of $\phi = 0.3$ over a range of rotation ratios $-0.35\leqslant \varOmega =\omega _o/\omega _i\lt 0$ (where $\omega _i$ and $\omega _o$ are the angular speeds of the inner and outer cylinders, respectively). The onset of VSM is found to coincide with the transition of a quasi-periodic state (‘doubly’ modulated wavy vortex [DMWV], characterised by two incommensurate frequencies) to a chaotic state (chaotic wavy vortex [CWV], in the background of VSM) at $\phi \leqslant 0.15$, giving rise to a new state of ‘VSM + CWV’ in which the VSM lives within a distinct CWV state. With further increasing particle loading, however, the VSM appears via MWV$\to$DMWV$\to$CWV$_{\mathrm{B}}\to$VSM + CWV$\to$CWV at $\phi = 0.2$ and is found to be absent at $\phi = 0.3$; another variant (CWV$_{\mathrm{B}}$) of CWV, characterised by an additional broad low-frequency peak, bifurcates from the quasi-periodic (DMWV) state at $\phi \geqslant 0.2$. The VSM events are shown to be temporally uncorrelated, with a multimodal skewed probability density function for the interevent time interval, implying that the VSM-induced chaos (VSM + CWV) is stochastic. The final transition to turbulent Taylor vortex (TTV) is found to occur via a propagating state of turbulent wavy vortices (TWV). The power spectrum of TTV is flat and featureless with a broad noisy tail, whereas the flat spectrum of TWV contains a sharp peak that represents a large-scale wavy vortex mode, whose disappearance marks the onset of TTV. It is shown that the propagation frequency of the TWV mode decreases with increasing shear Reynolds number (${\textit{Re}}_s(\phi )$) and follows a power-law scaling in terms of ${\textit{Re}}_s(\phi )$ and the relative viscosity of the suspension. Our results establish that the Ruelle–Takens route to chaos (Ruelle & Takens, Commun. Math. Phys., 1971, vol. 20, pp. 167–192) holds in counter-rotating suspension Taylor–Couette flow over a range of particle loading $\phi \leqslant 0.3$ at $-0.35\leqslant \varOmega \lt 0$.

Transonic shock buffet is a typical flow instability induced by shock-wave/boundary-layer interaction. It exhibits sustained shock oscillations and lift fluctuations, thereby restricting the flight envelope of aircraft. Previous numerical studies have revealed that evident hysteresis envelopes exist between the onset and extinction boundaries of transonic buffet during angle-of-attack sweeps. Hence accurate characterisation of buffet boundaries under representative dynamic manoeuvres is critical for assessing flight safety and expanding the flight envelope. In this study, a time-varying autoregressive with exogenous input model is established by integrating reduced-order and time-varying modelling to predict the buffet boundary under nonlinear transonic dynamic conditions. Specifically, the nonlinear flow evolution during ascending and descending angle-of-attack sweeps is modelled via piecewise linearisation, and the evolutionary trajectory of system stability is derived from state-space representation to characterise when buffet occurs. The influences of pitching rate and sweep direction on the hysteresis characteristics of the buffet onset boundary are investigated. For rate-dependent hysteresis, a lower pitching rate advances the buffet onset boundary under ascending angle-of-attack sweep, and the boundary gradually converges as the pitching rate further decreases. For path-dependent hysteresis, the ascending sweep yields a higher buffet onset angle of attack than the descending sweep, and such discrepancy diminishes gradually with the reduction of pitching rate. In summary, lowering the pitching rate can effectively suppress both types of dynamic hysteresis characteristics. The proposed method offers engineering guidance for monitoring and early warning of time-varying stability in aerospace and mechanical vibration applications.

Due to attractive inter-particle forces, cohesive particles suspended in turbulence undergo a complex process of aggregation, breakup and restructuring. Despite a growing body of knowledge on the `flocculation’ of cohesive granular materials suspended in homogeneous isotropic turbulence, little focus has so far been placed on wall-bounded flows where turbulence and shear are inhomogeneous. This study presents a first investigation of a fully developed wall-bounded flow of resolved cohesive particles. Five direct numerical simulations of turbulent channel flows laden with finite-sized particles at successively increasing cohesive strength are performed. A population balance equation (PBE) framework is used to analyse aggregate dynamics. When integrated over the full domain, the PBE is closed by aggregation and breakup alone. However, this balance is found to not hold locally in the wall-normal direction, where regions of net aggregate production and depletion are identified. This imbalance is shown to be compensated by the size-dependent wall-normal transport of aggregates, revealing a mean circulation: larger aggregates are preferentially produced in the channel centre and migrate toward the wall where they break, while smaller aggregates are transported away from the wall, grow and re-enter the cycle.

Model-based active flow control requires predictive models that are accurate, stable and fast enough for real-time optimisation. In controlled wake flows, this is often achieved through reduced-order models (ROMs) that first compress high-dimensional velocity snapshots into a latent space and then learn a time-stepping predictor for the dynamics in the latent space. Here, we study how the choice of the spatial encoder affects the predictability of the resulting latent coordinates for wake flows under control inputs. Using two actuated two-dimensional wake configurations, a simplified truck wake and the fluidic pinball, we compare proper orthogonal decomposition (POD) against nonlinear convolutional autoencoders (CAEs) and two types of variational autoencoders for compression, and evaluate several temporal predictors based on long short-term memory networks. The CAEs achieve higher compression efficiency and sharper short-term reconstructions, but they produce a latent dynamics that is more irregular and with broadband spectral content. As a consequence, long-horizon forecasts degrade faster and show a higher probability of catastrophic divergence than POD-based models. The POD yields smoother latent trajectories that are easier to learn and extrapolate, leading to more reliable predictions beyond the short-term regime. These results reveal a clear trade-off between compactness and forecast accuracy, and suggest that the stability of the latent dynamics prediction can outweigh maximal compression. This is particularly relevant for control strategies rooted in forecasts of the dynamics, such as model predictive control and reinforcement learning. The findings provide practical guidance for designing actuation-aware, hardware-feasible predictive ROMs for real-time flow control.

Mean flow and instability measurements of a nominally axisymmetric, transitional separation bubble arising from a flare-induced shock-wave/boundary-layer interaction were conducted on a cone–flare geometry in hypersonic flow. Simultaneous infrared thermography and high-speed, high-resolution background-oriented schlieren measurements provided a global, time-resolved characterisation of both the mean and unsteady flow. Experiments were performed in the AFOSR–Notre Dame Large Mach-6 Quiet Tunnel under conventional noise at free-stream unit Reynolds numbers ranging from $5.8\times 10^6$ to $12.3\times 10^6$ m$^{-1}$. The model comprised a $7^\circ$ half-angle circular cone, a $20^\circ$ half-angle flare and three nose radii. Complementary laminar and turbulent base flows, convective and global stability analyses (GSAs) and direct numerical simulations (DNSs) were carried out at the experimental conditions. With inflow forcing of disturbances, the DNS reproduces the experimentally observed flare-heating trends. The flow exhibits both convective and global instabilities, with the latter giving rise to stationary and oscillatory three-dimensional disturbances within the recirculation bubble. Thermal streaks observed in global heating measurements provide experimental evidence of these instabilities, which is further corroborated by the onset of three-dimensionality in DNS. Quantitative agreement of streak wavelengths and their spatial prevalence across experiments, GSAs and DNSs confirms the recirculation bubble’s global instability is the primary driver of the reattachment streaks in this flow, providing direct experimental evidence of this mechanism.

The Maxey–Riley–Gatignol equation for the flow around a sphere at low particle Reynolds number tells us that the fluid–particle interaction force decomposes into a contribution from the local flow disturbance caused by the particle’s boundary – consisting of the drag, virtual-mass and history forces, and their Faxén corrections – and another contribution from the stress of the background flow, termed generalized-buoyancy force. There is also a consensus that, for general disperse two-phase flow, the interfacial force density, coupling the average fluid phase and dispersed-phase momentum balances, decomposes in a like manner. However, there has been a long-standing controversy about the physical closure separating the generalized buoyancy from the interfacial force density, especially whether or not pseudostresses, such as the Reynolds stress, should be attributed to the background flow. Furthermore, most existing propositions for this closure involve small-particle approximations. Here, we show that all existing buoyancy closures are inconsistent with particle-resolving numerical simulations and/or at least one of two simple thought experiments designed to determine the roles of pseudostresses and small-particle approximations. We then derive the unique consistent closure. It requires no approximation and implies that all stresses and pseudostresses in the average fluid phase momentum balance, except the Reynolds stress, fully contribute to the background flow responsible for buoyancy. Remarkably, it exhibits a low-pass filter property, attenuating buoyancy at short wavelengths, that prevents it from causing Hadamard instabilities, constituting a first-principles-based solution to the long-standing ill-posedness problem of two-fluid models. When employing the derived closure, even simplistic two-fluid models are linearly well-posed.

Wall-bounded turbulent flows are chaotic and multiscale, rendering fast prediction at high Reynolds numbers computationally prohibitive in applications such as wind farms. Classical data assimilation is based on repeated solutions of the governing equations and, thus, inherits this cost. Generative models learn the probability distribution of flow states, enabling scalable probabilistic reconstruction. Our generative framework couples a variational autoencoder with a diffusion transformer to generate four-dimensional spatiotemporal samples. Bayesian conditioning enables data assimilation without retraining and allows statistical constraints to be imposed through sampling. The framework is applied to a subdomain of turbulent plane Couette flow, where the corresponding direct numerical simulations in this generation region requires $O(10^6)$ spatial degrees of freedom. Using $O(10)$ latent spatial degrees of freedom, the model achieves a compression ratio of $O(10^5)$, which is one to two orders of magnitude above prior reports. It reproduces single-point statistics up to fourth order and the energy spectra, as well as the intermittency and phase-sensitive structure captured by velocity-increment and $Q$–$R$ statistics. Two assimilation scenarios demonstrate that, when observations are statistically consistent with the prior, conditional diffusion models with the proposed sampling strategy preserve complex turbulent statistics in the posterior. However, enforcing these constraints while preserving physical fidelity and sample diversity introduces an inherent trade-off. Excessive conditioning can distort the learned prior, paralleling limitations of classical ensemble-based data assimilation, where this can likewise degrade the prior covariance. These results highlight both the promise of diffusion models as probabilistic surrogates and the challenges of conditioning them.

Symmetry-breaking in the near-wake of a wall-proximity sphere is a claussic problem. Yet a clear understanding of how the gap ratio ($G/D$) governs the evolution and suppression of spanwise flow asymmetry across different Reynolds numbers ($Re$) is still lacking. Here, direct numerical simulations combined with proper orthogonal decomposition and vorticity transport analysis are employed over $420 \leqslant Re \leqslant 600$ and $0.1 \leqslant G/D \leqslant 1.5$ to elucidate the wall-induced suppression of spanwise asymmetry. As $G/D$ decreases, four regimes are identified: asymmetric unsteady II, asymmetric unsteady I, symmetric unsteady and symmetric steady. Notably, the spanwise wake deflection reverses direction between the two asymmetric regimes, signalling a structural transition regulated by vortical competition. This corresponds to the transition from the alternating shedding of upper and lower hairpin vortices to the formation of longitudinal vortices (LV) that interact with wall-secondary vortices (WSV). Symmetry restoration is associated with merging and the emergence of the second harmonic of the dominant shedding frequency. Crucially, the wall effect homogenises the streamwise pressure gradient and velocity-gradients in the spanwise direction. This gradient homogenisation attenuates the asymmetric tilting of spanwise vorticity, thereby reducing the feedback that maintains the spanwise asymmetry of the shear layer. In asymmetric regimes, wake deflection is sustained by dominant spanwise-asymmetric vorticity tilting into streamwise and wall-normal components. In the symmetric regime, suppression of the spanwise deflection of the shear layer supports wake-symmetry recovery. These findings provide insights into symmetry-breaking in wall-proximity bluff-body wakes, and offer guidance for active flow control.