Nan-Sheng Liu

Direct numerical simulation is performed to investigate the three-dimensional turbulent boundary layer on the leeward side of the Hypersonic Transition Research Vehicle at Mach 12 and a 6° angle of attack. Under the combined influence of bypass transition induced by blowing/suction and three-dimensional effects, two spatially separated regions of enhanced skin friction and wall heat flux (WHF) are identified, denoted as regions A and B, whose formation mechanisms are fundamentally distinct. In region A, the high WHF is predominantly driven by steepened gradients induced by pressure gradients, whereas the enhanced skin friction is further driven by the unsteady separation and reattachment resulting from an adverse-to-favorable pressure gradient transition, exhibiting banded quasi-streamwise high-momentum flows. Region B, in contrast, is governed by the roll-up and breakdown of quasi-streamwise vortices, which sustain a fully developed turbulent state populated by fine-scale transverse vortical structures and azimuthally distributed heat-flux streaks. Compressibility effects are pronounced in both regions, as evidenced by the emergence of shocklets extending into the outer boundary layer. The two regions further exhibit contrasting turbulence signatures: large-scale structures driven by blowing/suction generate outer peaks in the Reynolds stresses upstream, whereas intense turbulence production in region B gives rise to distinct inner peaks. These findings highlight the critical role of geometric three-dimensionality and compressibility-induced structural changes in shaping the aerodynamic and aerothermal environment of high-speed lifting-body configurations.

Turbulence amplification in shock wave/turbulent boundary layer interactions critically affects flow unsteadiness and wall loading in supersonic aerodynamic configurations. Direct numerical simulations (DNS) are performed at shock generator angles of β=8°, 10°, and 12°. At the two smaller angles, turbulence amplification is confined to the vicinity of the separation point. As β increases to 12°, however, a pronounced secondary amplification emerges near the reattachment point, giving rise to a bimodal streamwise distribution of turbulent kinetic energy (TKE). Spectral analysis of the TKE production terms reveals that this secondary amplification is predominantly driven by shear production associated with large-scale vortical structures. Global stability analysis (GSA) of the mean flow further demonstrates that all unstable eigenmodes are intrinsically tied to the separation bubble, and the most unstable spanwise wavenumber predicted by GSA agrees closely with the dominant scale observed in the DNS. Reduced-order variational mode decomposition identifies an energetically dominant, low-frequency three-dimensional mode in the reattachment region for the β=12° case, characterized by coherent Görtler-like vortices whose spanwise wavelength is consistent with both the DNS and GSA results. These results suggest that the secondary turbulence amplification originates from the modal instability of the mean flow associated with the large-scale separation bubble and the coherent structures it induces.

Accurate prediction of the hydrodynamic coefficients of non-spherical particles in wall-confined flows is crucial for understanding particle–fluid interactions and reliable modelling of particle motion. Under strong wall confinement, the hydrodynamic coefficients exhibit a highly nonlinear dependence on the Reynolds number, wall distance and particle orientation – posing significant modelling challenges. In this study, we propose a multi-stage physics-informed machine-learning (MSPIML) framework for modelling the drag, lift and pitching torque coefficients of a wall-bounded prolate spheroid over the explored parameter space. In the first stage, a physics-informed mixture-of-experts (PIMoE) model predicts the drag coefficient by intelligently blending empirical correlations with a data-driven statistical expert. The resulting high-fidelity drag coefficient is then injected as an auxiliary input to a second-stage model, either a deep neural network (DNN) or an additional MoE, that predicts lift and pitching torque coefficients, thereby leveraging the strong physical coupling among the three coefficients. Trained on a comprehensive dataset of 720 direct numerical simulations covering wide ranges of Reynolds number, wall distance and particle orientation, the optimal PIMoE–DNN and PIMoE–MoE configurations achieve relative errors below 2.2 % for drag, 11.4 % for lift and 7.0 % for pitching torque while maintaining excellent generalisation across the entire parameter space. Moreover, the Shapley additive explanations analysis confirms that the MSPIML framework correctly captures the physical dependencies: dominant influence of Reynolds number and strong pitching torque dependence on the drag coefficient. The MSPIML framework provides an interpretable and efficient approach to the prediction of hydrodynamic coefficients and offers substantial potential for dynamic modelling of non-spherical particles in multiphase flows.

It is of great importance for fields such as implosion dynamics and fusion research to understand the dynamics of ejecta transport in converging gases. In this paper, the evolution of particulate flow within a cylindrically imploding system is investigated experimentally and numerically. The ejecta particles are emitted from the inner surface of a roughened Sn liner into vacuum, He and Ar gases. Dynamic images of liner implosion and ejecta transport are obtained with X-ray radiographs and multi-frame optical schlieren images. The transport of ejecta particles is simulated with a four-way coupled multiphase flow model, including modelling of gas–particle coupling and inter-particle collisions. Results reveal that the ejecta transport in shock-induced converging gases differs significantly from that in planar systems, primarily due to features such as interaction with the rebounding gas shock wave and continuous compression by the imploding liner. After being generated from the inner surface, the ejecta width undergoes an ‘expansion–compression’ variation. According to mechanisms governing ejecta–gas coupling, three distinct stages of ejecta evolution are identified: (i) post-shock transport dominated by drag and particle breakup; (ii) shock-particle interaction leading to quick reduction in particle size and rapid deceleration of the ejecta front; and (iii) dense ejecta compression governed by inter-particle collisions. Leveraging particle motion and size predictions at the ejecta front, combined with the self-similar converging shock solution, a theoretical model is established to estimate the three-stage evolution of ejecta width in a cylindrically converging system.

Compressible wall-bounded turbulent flows exhibit complex mean profiles because of the pronounced compressibility effects and heat transfer. We propose a hybrid transformation framework to collapse compressible mean velocity and temperature profiles onto incompressible forms through scaling each layer by its effective transformation, with the underlying mapping functions discovered via a physics-informed symbolic regression (PISR) method. The hybrid velocity transformation incorporates an intrinsic compressibility correction for the buffer layer and a PISR-derived mapping function for the logarithmic layer. For temperature, we introduce a hybrid transformation that integrates the Mach-invariant-type transformation in the viscous sublayer and a novel PISR-derived scaling in the logarithmic layer. The performance of these transformations is evaluated across compressible turbulent boundary layers with free-stream Mach numbers ranging from 0.5 to 8 and wall-to-recovery-temperature ratios ranging from 0.25 to 1. The hybrid velocity transformation outperforms Griffin–Fu–Moin transformation for the transformed mean velocity profiles, with the mean integrated percent error across the dataset decreasing from 1.67 % to 0.96 %. The hybrid temperature transformation performs better than the Mach-invariant-type and Trettel–Larsson-type transformations for mean temperature profiles. Moreover, the inverse hybrid velocity and temperature transformations can effectively predict the compressible mean velocity and temperature profiles with only wall conditions.