Mingbo Sun

In this study, direct numerical simulations are performed to investigate inertial particle transport in supersonic turbulent flow in a straight-to-bent pipe. Three particle populations with viscous Stokes numbers ( St ⁺ ) of 0.9, 4 and 16 are considered. It is revealed that under the combined influence of wall confinement, flow compressibility and centrifugal forces, distinct turbulent structures emerge in different pipe regions, leading to inertia-dependent particle dynamics. In the straight section, particle streaks and near-wall accumulation are more significant for St ⁺ = 4 particles. When entering the curved section, Görtler vortices along the outer bend enhance the streak formation and the accumulation of St ⁺ = 16 particles. Further downstream, expansion waves suppress Görtler vortices while Dean vortices become dominant, transporting particles along the sidewalls towards the inner bend, thereby reducing outer-wall collisions and accumulation. Particles tend to escape the confinement of Dean vortices and accumulate on the sidewalls. A larger Stokes number results in an earlier detachment, accompanied by higher collision frequency, particle accumulation and slip velocity. In contrast, small inertia particles follow the vortical motions more closely, migrating into the inner low-speed region and resulting in a higher local particle concentration. These different behaviours stem primarily from differences in the centripetal force imparted by the flow in the streamwise and circumferential directions. The strongest centrifugal effect occurs in the areas on both lateral sides of the pipeline. These elucidations offer new physical insights for modelling and predicting particle-laden turbulent flows in curved geometries.

To investigate the effects of ignition position on flame propagation and combustion characteristics in a dual-cavity scramjet combustor, comparative experiments were conducted under upstream cavity ignition (UCI) and downstream cavity ignition (DCI) conditions in a direct-connect supersonic combustor facility at an inflow Mach number of 2.52. Synchronized schlieren and CH* chemiluminescence imaging systems were employed to visualize flow and flame dynamics during ignition and flameholding processes. For both ignition positions in the present study, the flame initially propagates downstream, then migrates upstream, and finally stabilizes above the upstream cavity. During inter-cavity propagation, the flame leading edge anchors in the near-wall low-velocity zone, and the flame propagation path within the unignited cavity follows the local recirculation direction. Combustion-induced backpressure shifts the shock train upstream; the resulting shock–flame interaction enhances flame propagation toward the unignited cavity and intensifies combustion during the first inter-cavity propagation in the DCI case, whereas this phenomenon is not observed in the UCI case. During flameholding, both cases exhibit cyclic oscillations with temporary flame weakening near the downstream cavity. Under UCI, the oscillation frequency progressively increases until premature flameout occurs, while under DCI, the frequency remains lower without acceleration; both cavities provide stronger flame stabilization, and the flame leading edge oscillates over a smaller range.

Passive fluidic thrust vectoring is attractive for detonation-based propulsion systems because it requires neither external gas supplies nor moving parts. This paper proposes a novel two-dimensional aerospike nozzle in which a throat-bypass passage is embedded within the central spike: the natural pressure difference across the nozzle drives a secondary jet through the passage, and its aerodynamic interference with the primary flow deflects the thrust vector. The concept is investigated through cold-flow wind tunnel experiments with schlieren diagnostics, complemented by steady compressible Reynolds-averaged Navier–Stokes simulations over nozzle pressure ratios (NPRs) of 3, 5, 10, and 15. The schlieren images provide qualitative experimental validation: with the upper bypass passage opened, the nozzle jet exhibits pronounced deflection and flow asymmetry at all tested NPRs, and a distinct Mach disk appears on the injection side at NPRs of 10 and 15, whereas the closed-bypass configuration remains symmetric. Validated against published benchmark experiments and the present schlieren images, the simulations reveal the underlying vectoring mechanism: the bypass jet deflects the throat sonic line and attenuates the Coanda-driven wall attachment, and the resulting differential expansion establishes a large-scale wall-pressure asymmetry over the thrust surface. Integrating the numerically predicted wall-pressure distributions yields normal-to-axial force ratios of 0.233, 0.431, and 0.359 at NPRs of 3, 5, and 10, respectively, corresponding to a peak vector angle of approximately 23.3° at an NPR of 5, while the axial-thrust penalty remains below 1.1% and the bypass mass flow fraction below 8%. These results demonstrate the feasibility and control efficiency of the proposed throat-bypass passive vectoring aerospike nozzle and provide a reference for passive thrust-vectoring design in future detonation engines.

In view of the instability characteristics of self-sustained combustion in a Ma 2.5 dual-cavity supersonic combustor, an active control technique based on gliding-arc plasma assistance is used to suppress combustion oscillations. Through analyzing flame spontaneous emissions, Schlieren images, wall pressure taps, and OpenFOAM-based three-dimensional simulations, the flame development characteristics and plasma-assisted combustion (PAC) mechanisms are analyzed. It is found that the gliding-arc plasma continuous assistance at the bottom of the rear cavity forms a stable global flame more rapidly, improving the flame distribution and enhancing the flame spontaneous emission by 10%–24%. In contrast, the enhancement of the gliding-arc plasma assistance in the front cavity ranges from 4% to 14%, with a longer delay time in forming stable combustion. The flow mechanism of PAC lies in increasing the turbulence intensity and heat release in local regions within cavities. Thus, the interaction between the mainstream flame and the hot flow is facilitated continuously by the plasma, forming stable resident flames in the vicinity of cavities. During the plasma action, excited species such as N2(C3Πu) and O(1D) act on the local plasma region through thermal and kinetic effects, inducing the continuous formation of initial flames and heat release for temperature rise. Meanwhile, excited species with longer relaxation time, such as N2(v1) and N2(v2), are transported into the mainstream region. Outside the plasma-actuated region, these long-life excited species do not directly engage in the dominant heat-release reactions, but catalytically facilitating the redox reactions via transport effects and kinetic effects, thereby achieving combustion stabilization and enhancement.

To mitigate severe aerodynamic heating at the rudder/wing leading edges of high-speed vehicles, opposing-jet active thermal protection is investigated. Three-dimensional Reynolds-averaged Navier–Stokes simulations at Mach 8 and 30 km altitude examine single-orifice, dual-orifice, and spanwise multi-orifice jets over variable-bluntness leading edges under a constant mass flow rate constraint. The results show that the restriction on radial jet expansion caused by a reduced orifice diameter can be compensated by the enhanced expansion associated with an increased jet total pressure ratio (PR). Consequently, macroscopic shock structures and wall thermal loads remain highly consistent for different combinations of orifice diameter and PR. This compensation behavior is maintained under both strong and weak dual-orifice interference. For spanwise multi-orifice arrays, the constant mass flow rate per orifice strategy provides more uniform thermal protection than the constant PR strategy and gradually establishes a downstream dynamic equilibrium region with stable flow structures. Under constant PR, the mass flow rate and jet strength decrease with orifice size, causing pronounced increases in downstream local pressure and heat flux peaks. The present study verifies the applicability of mass-flow rate-based parameter compensation predicted by classical jet scaling relations to complex flows involving three-dimensional sweep, multi-orifice interactions, and variable-bluntness leading edges, and provides a design basis for orifice parameter matching and mass flow distribution in multi-orifice opposing-jet thermal protection systems.

This paper presents a G-Agent, a large language model (LLM)-driven multi-agent framework that automates the optimization workflow for an axisymmetric cavity-based scramjet combustor. The framework integrates a supersonic compressible reacting flow solver with the Qwen3-VL-2B-Instruct model. It consists of four specialized agents: Interactor, Runner, Corrector, and Optimizer. The Interactor parses user requirements and generates simulation input files. The Runner manages mesh generation, solver configuration, and simulation execution. The Corrector diagnoses failed cases by analyzing fault logs and applies corrective actions. The Optimizer coordinates the optimization process using a trust-region response surface method to maximize thrust. The framework employs Latin hypercube sampling (LHS) to explore the design space defined by three key geometric parameters: isolator length, cavity depth, and fuel injector position. Without manual intervention beyond initial specifications, the G-Agent successfully conducted multiple optimization iterations. Starting from a baseline thrust of 489.4 N, the optimized configuration achieved 523.5 N after three iterations, reaching a relative improvement of 6.97%. Our proposed agent achieves closed-loop control of the entire optimization process, significantly lowering the technical barrier and human effort in scramjet combustor design.

This paper uses direct numerical simulations to systematically investigate supersonic flows in a bent pipe with a developing turbulent boundary layer and a core flow region. The results reveals that boundary layers experience intricate flow patterns at different azimuthal angles: secondary flows that drives the streaks move from the lower to the upper side; separation triggered by the combined effects of the adverse pressure gradient and flow deceleration on the upper wall; and the increase of the G\"ortler instability on the lower wall inducing the clustering and uplift of low-momentum fluids, the presence of G\"ortler-like vortices, and the baroclinic effect of turbulent transport processes.

Capturing the highly nonlinear, transient temporal dynamics of nucleate boiling heat flux imposes severe computational burdens, particularly when involving complex inhomogeneous surfaces where intense liquid–vapor interactions induce chaotic thermal fluctuations. To deeply reveal these underlying phase-change mechanisms, this study establishes a coupled framework integrating the hybrid pseudopotential lattice Boltzmann method with deep learning architectures. Systematic analysis reveals that the transient surface heat flux exhibits pronounced high-frequency oscillations, which are fundamentally driven by the periodic physical cycles of vapor blanketing, bubble departure, and rapid surface rewetting. While conventional neural networks (such as multilayer perceptrons and convolutional neural networks) structurally disrupt the temporal continuity of this thermal history, the recurrent neural network (RNN) uniquely aligns with the physical periodicity of the boiling cycle. The internal memory cells of the RNN function analogously to the thermal inertia of the heating wall, successfully capturing the cyclic destruction and reconstruction of the thermal boundary layer and yielding superior predictive accuracy. Furthermore, on inhomogeneous surfaces with isosceles trapezoidal microstructures, the fundamental periodicity of boiling is physically disrupted. Intense contact line pinning and premature lateral bubble coalescence drastically prolong localized vapor blanketing, shifting the multiphase heat transfer from a stable cyclic process to a highly chaotic, nonlinear dynamic regime. The proposed framework effectively tracks these complex transitions, proving that structural micro-confinement profoundly modulates the localized thermal fluctuations.

Flash evaporation, a liquid-to-gas phase transition phenomenon in real fluids, is prevalent in aerospace propulsion systems. To elucidate the wave structures of flash evaporation in the equilibrium limit and provide theoretical benchmarks for computational fluid dynamics simulations, this paper formalises the flash evaporation Riemann problem (FeRP) characterised by the expansion branch crossing the saturation line, within the framework of homogeneous equilibrium and vapour–liquid equilibrium assumptions. An exact solution framework that analytically resolves all thermodynamic derivatives of equilibrium two-phase fluids is established for arbitrary two-parameter equations of state. By evaluating the Landau fundamental derivative, the non-classical wave structures arising in the FeRP are analysed, for which a stable iterative solution strategy incorporating the Chapman–Jouguet condition as an outer constraint is proposed. Furthermore, the FeRP framework is extended to the Riemann problem (RP) with a given thermodynamic non-equilibrium speed of sound, enabling a comprehensive evaluation of how this non-equilibrium closure affects the RP solution. Results indicate that such a non-equilibrium model alters the definition of the two-phase mixture entropy in the Euler equations, introducing a different isentropic path characterised by non-physical entropy decrease and density lag relative to the equilibrium path. Consequently, as the given speed of sound shifts from the complete equilibrium limit towards Wood’s mechanical equilibrium, the intermediate pressure, velocity and vaporisation extent in the RP solution decrease.

This study elucidates the scale effect on auto-ignition characteristics in cavity-stabilized scramjet combustors. Both experimental and numerical investigations are conducted on ethylene-fueled combustors at different geometric scales. Experimental observations indicate that the flame kernel consistently first emerges near the cavity rear wall across all scaled combustors. However, under the present experimental conditions, a scale-related delay in kernel formation is observed with increasing combustor scale, with an averaged apparent delay of approximately 80 ms in the √2-scale combustor compared with the reference-scale combustor. Cold-flow simulations are then performed to interpret this behavior. The results reveal that scaling does not alter the potential auto-ignition locations or the primary flow structures. Instead, the time delay is attributed to prolonged processes of fuel enrichment and heat accumulation. In 2-scale combustors, the fuel transportation path is extended, which reduces the efficiency of forming fuel-rich pockets. Concurrently, effective heat feedback within the recirculation zone is attenuated. As a result, more time is required for local fuel/oxidizer mixing to reach critical auto-ignition conditions. It is indicated that the scale effect during the auto-ignition stage manifests as a time delay in mixing and energy accumulation, rather than a fundamental change in the auto-ignition mechanism.