Droplet behavior in high-speed and detonative flows is governed by tightly coupled aerodynamic, thermodynamic, and phase-change processes, yet a unified framework for classifying these interactions is still lacking. This work introduces a physics-based regime diagram using two dimensionless groups that quantify thermodynamic loading and the competition between phase-transition and aerodynamic time scales. The framework connects shock–droplet interactions, liquid-fueled detonations, and rotating detonation engines, identifying conditions that favor breakup, rapid phase transition, and sustained detonation coupling.

Although the onset of vortex shedding behind a fixed circular cylinder has been well established, free fall can change the threshold. This article demonstrates that the critical Reynolds number drops from 46.1 for a stationary circular cylinder to 42.3 for a freely falling cylinder slightly denser than the fluid. Transverse motion promotes instability, while rotation partly counteracts this effect. Across the motion constraints and density ratios studied, the maximum perturbation energy scales approximately linearly with the deviation from the instability threshold.

Capillary forces are usually viewed as a closed pathway that redistributes kinetic energy across scales in multiphase turbulence. We show that this picture breaks down in bubble-laden turbulence when short-range near-contact interactions become active. These interactions open an additional small-scale energy-transfer channel that compensates the residual capillary work, identifying an extended interfacial transfer as the proper energetic description.

The effects of free-stream conicity on flow over a sharp cone are investigated using theoretical methods and numerical simulations. The results show that the flow is significantly influenced by conicity. Free-stream conicity essentially rescales the nondimensional distributions of the flow properties, while strong viscous interaction modifies this behavior and reduces the accuracy of analytical predictions. The proposed framework can be qualitatively extended to other divergent free-streams by characterizing non-uniformity through the Mach number gradient.

This manuscript presents fully developed and experimentally validated research on a ternary multicomponent droplet system, revealing how interfacial transport, phase separation, and Marangoni effects fundamentally alter cavity collapse and jet formation. By establishing new scaling laws incorporating interfacial roughness and compositional heterogeneity, it provides significant fluid mechanical insight beyond classical single-component frameworks, substantially advancing the understanding of inertial-capillary singularities in complex fluids.

We carry out a second‑order perturbation analysis for the equilibrium deformation of a surfactant‑laden liquid column under a weak uniform transverse electric field. Our analytical solution shows that surfactant coverage and the elasticity number alter steady‑state deformation indirectly via the electric capillary number, while the surface Péclet number and effective Boussinesq number exert no effect on deformation up to second‑order in the electric capillary number. Nevertheless, all surfactant‑governed dimensionless parameters substantially modify interfacial characteristics and bulk flow circulation.

Near the pseudo-critical state, supercritical water exhibits strong thermophysical property variations with poorly understood stability consequences. Modal and non-modal analyses of plane Poiseuille flow show that non-ideality reshapes the neutral-stability landscape, producing a three-stage evolution and isolated instability islands in the trans-critical regime. Classical small-wavenumber transient-growth scaling persists outside this regime, and the fundamental lift-up mechanism remains shear driven. Non-ideal effects instead redistribute energy among shear production, thermodynamic, and viscous channels, providing a stability map and quantitative benchmarks for supercritical water flows.

Freezing becomes more complex when several immiscible fluids coexist and change volume as they solidify. We develop a reduction-consistent diffuse-interface model that couples phase-field and enthalpy formulations for N-phase flows with liquid-solid phase change. The model captures solidification-induced expansion and shrinkage while accounting for the interplay between fluid flow, heat transfer, and evolving interfaces. Simulations reveal how interfacial interactions and thermal diffusivity contrasts shape freezing fronts and solidified morphologies in compound droplets and impurity-laden systems.

Taylor–Aris theory describes the long-time dispersion reached after transverse mixing, but finite channels often operate before this asymptotic regime is established. We develop a two-anchor, dominant-mode closure that connects the exact short-time sampled-velocity variance to the long-time Taylor enhancement through a single buildup time. Spectral benchmarks and entropic lattice Boltzmann simulations show that the closure accurately captures reduced Brownian-rod transport across power-law channel flows and rotational Péclet numbers. The resulting development times and lengths provide practical criteria for deciding when long-time dispersion models can be trusted.

We describe a new idea for determining the self-similar profiles of the mean velocity and the second-order velocity moment profiles in a turbulent round jet. For this we use a Lie-symmetry theory inspired scheme to derive highly accurate approximations. In fact, most research about the turbulent round jets to date focused on the streamwise scaling of turbulent quantities. As part of the new analysis, we provide the first rigorous derivation of the Gaussian profile of the mean axial velocity. We validate the findings against high quality DNS data.

This article proposes the Reynolds stress transport nets (RSTnets) for reconstructing RANS solutions from the boundary data through embedding the Reynolds stress transport formulation via physics-informed neural networks. Different from previous works which either rely on the empirical modeling of all the unclosed terms or leave the Reynolds stress completely unmodeled, the proposed RSTnets are developed to reconstruct the flow fields in the domain based on the first- and second-order flow quantities at the domain boundary through incorporating as much physical information as possible while avoiding empirical modeling of the unclosed terms.

Navigating near obstacles is challenging for soft swimmers because walls reshape the surrounding flow. In two-dimensional simulations, a jellyfish-like swimmer combines local path guidance with deep reinforcement learning that uses hydrodynamic force and torque feedback. These mechanical cues enable the swimmer to exploit wall-induced flows for faster, more efficient maneuvers than a controller without this feedback. The framework also supports navigation among multiple obstacles and exploration of a model underwater cave.

Unexpected gusts can cause rapid changes in aerodynamic flows that are difficult to detect and reconstruct in real time. We combine sparse surface-pressure measurements with ensemble Kalman filtering in a learned low-dimensional space to detect disturbances and reconstruct the evolving flow field and aerodynamic loads. The framework provides fast, uncertainty-aware estimates and remains effective for unseen gust conditions and degraded sensor configurations.

In air-cushioned flat-plate water entry, the effect of plate size on peak impact load remains unresolved. A prevailing view holds that smaller plates entrap less air, thereby weakening the buffering effect and increasing the peak load. This work numerically investigates the plate-size effect in both two and three dimensions and finds, contrary to conventional expectations, that smaller plates entrap less air yet experience lower peak loads. Further analysis reveals that the load variation is dictated by the compression dynamics of the entrapped air layer, with air-layer compression weakening as plate size decreases.

We develop a data-driven framework to model the dynamics of turbulent flows with multiple dominant frequencies based on a set of oscillators. The framework extracts oscillators that capture the dominant dynamics from the flow field data using autoencoders arranged in parallel. The dynamics of coupled oscillators are modeled by leveraging a neural ordinary differential equation, which enables accurate prediction of the multi-frequency oscillatory behavior of turbulent flows assisted by observation-based correction. We demonstrate the capability of this oscillator-based model for a three-dimensional supersonic turbulent flow over a cavity.