The integration of transfer learning with surrogate-assisted optimization represents a highly promising paradigm for complex engineering problems. However, it is fundamentally constrained by insufficient cross-domain adaptation and the absence of robust knowledge-fusion techniques. To address these limitations, this study proposes an adaptive knowledge-fusion method via dynamic stacking for surrogate-assisted transfer optimization. Domain adaptation is achieved through polynomial chaos expansion/augmented radial basis function surrogate modeling of source tasks, coupled with domain transformation coefficients derived from target task data. A cosine-similarity metric computed directly from surrogate predictions is introduced to quantify intertask relationships. This similarity measure is then integrated into an iterative updating framework, in which transfer-weight coefficients are dynamically computed to enable the adaptive construction of stacked surrogate models. Target task optimization is further refined via a balanced sampling criterion. The proposed method is validated through benchmark functions, a two-dimensional airfoil optimization case, and a multidisciplinary solid rocket motor design problem. Compared with existing transfer optimization approaches, the proposed method substantially improves optimization efficiency while achieving superior knowledge transfer capability and robustness.
The dynamic evolution of localized three-dimensional (3-D) stress distributions in thick-section thermal protection systems (TPSs) under harmonic excitation is investigated within complex thermal environments. Using a thicknesswise homogenization approach, a fourth-order polynomial warping function is established without ad hoc kinematic assumptions, transforming the 3-D TPS into an equivalent two-dimensional Reissner–Mindlin plate. This study elucidates the mechanism of thermally induced modal order swapping and its impact on dynamic stress redistribution. Through a global–local recovery framework, the research quantifies how temperature-dependent frequency shifts alter dynamic stress peaks. Results indicate that although qualitative stress patterns are predictable, response amplitudes and critical stress locations are highly sensitive to thermal–vibration coupling. Furthermore, the face-sheet material orientation is identified as a key variable governing failure modes. Although a 0 deg orientation maximizes compression failure susceptibility, rotating fiber alignment increases dynamic shear stress by up to 166.4%, significantly altering the failure envelope. Specifically, static thermal stress contributes nearly 50% of the peak load under steep temperature gradients, whereas dynamic components account for over 80% near resonance. These findings provide a high-fidelity foundation for predicting the structural integrity of lightweight TPSs under extreme vibrothermal environments.
High-resolution flowfield data are essential for turbulence research and engineering applications, yet their acquisition from numerical simulations and experiments remains prohibitively expensive. Super-resolution reconstruction offers a promising alternative, but conventional interpolation-based and generic image super-resolution methods often fail to recover turbulence-specific fine-scale features, such as coherent vortical structures, thin shear layers, and high-wavenumber fluctuations. To address this limitation, this study proposes a structure-aware deep-learning framework, termed the Coordinate-Attention Residual Super-Resolution Network (CARSR-Net), for reconstructing turbulent flows from severely underresolved inputs. Here, structure-aware means that the model is designed to preferentially preserve physically important flow structures and their multiscale statistical signatures. In particular, a turbulence-oriented composite loss function is introduced to emphasize high-gradient and vortical regions so that the reconstruction is constrained not only by pointwise field accuracy but also by the recovery of physically relevant small-scale structures. Lightweight attention and multiscale feature aggregation modules are further incorporated to improve the extraction of directional and multiscale flow information. The model is evaluated on wall-bounded turbulent channel flow and forced isotropic turbulence using both instantaneous fields and representative statistical quantities. The results show that CARSR-Net reconstructs high-resolution flowfields with lower relative [Formula: see text] error and more accurate energy spectra and vorticity probability density functions than baseline methods. In particular, it better preserves high-frequency content and the physical consistency of the reconstructed fields.
The influence of hydrogen injection configuration on combustion instability in arrayed microtube burners is investigated through experiments and large-eddy simulations. Three injector configurations, namely, baseline radial injection (nozzle A), upstream-shifted radial injection (nozzle B), and 45 deg inclined injection (nozzle C), are examined under different equivalence ratios and pressure-drop conditions. Chemiluminescence imaging and simulations show that nozzle A maintains a stable diffusion flame with a fixed heat-release region, whereas nozzle B enhances fuel–air premixing, increases flame liftoff, and strengthens the coupling between heat-release fluctuations and flow disturbances, leading to reduced stability. Nozzle C modifies flame anchoring through additional radial momentum and mitigates pressure oscillations relative to nozzle B. Pressure measurements, flame dynamics, recurrence analysis, and flame transfer function results consistently indicate that nozzle B and nozzle C are more sensitive to inlet velocity perturbations than nozzle A. Overall, the results demonstrate that hydrogen injection configuration plays an important role in regulating flame dynamics and thermoacoustic stability in arrayed microtube hydrogen combustors.
The effects of excitation on the coherent structures in a supersonic rectangular jet with a design Mach number of [Formula: see text] are investigated. Planar laser-sheet imaging and time-resolved schlieren imaging, combined with spectral proper orthogonal decomposition (SPOD), are used to investigate the jet over three flow regimes. Coherent structures within the jet shear layers are clearly visualized and are shown to dominate the first SPOD mode at both the natural screech frequency and the excitation frequency. The robust response of the jet is over a large range of frequencies consistent with the overlap between the jet column mode and the shear layer mode. Excitation at frequencies higher than the weak screech frequency of the design condition disrupts the naturally occurring coherent structures. Furthermore, excitation of different azimuthal modes alters both the organization and energy content of these structures within the jet. In the presence of screech staging, excitation can also change the dominant tonal peak within the jet. Excitation also changes the overall entrainment and mixing characteristics of the jet.
Revealing the dynamics of wall shear stress in high-speed turbulent boundary layers is critical for improving the flight performance of advanced vehicles. This study investigated extreme events of wall shear stress on the windward surface of a lifting body at a freestream Mach number of 6 and a 2 deg angle of attack, using direct numerical simulation data. It was found that the mean and fluctuating skin-friction coefficients were significantly higher in the side region. Based on wavelet analysis, a strong correlation was unraveled between the extreme events and localized bursts of high-frequency energy, highlighting the critical role of near-wall fine-scale structures. Using conditional volumetric averaging and skin-friction decomposition, it was shown that wall-normal momentum transport, driven by counter-rotating quasi-streamwise vortex pairs, was the primary physical origin of extreme events. To identify the underlying high-order structures, conditional spatial proper orthogonal decomposition was performed. Notably, two basic types of coherent structure were identified: the streamwise single-layer vortex pair and the streamwise stacked vortex pair, which induced near-wall sweep/ejection motions that directly contributed to the formation of extreme events. Furthermore, modal reconstruction demonstrated that the leading mode contributed to the majority of extreme events, although this contribution was not correlated with the mode energy.
The prediction of aircraft icing is conventionally performed using multishot simulation frameworks that fail to predict the progressive roughening of the ice surface. To understand roughness formation, we investigate droplet impingement on clean and laser-scanned rough ice shapes using a high-fidelity computational framework based on wall-modeled large-eddy simulations and Lagrangian particle tracking. This methodology is validated against experimental data for a NACA 23012 airfoil and a NACA 64A008 swept tail, accurately predicting collection efficiency and supercooled large droplet splashing. The framework is subsequently applied to laser-scanned rime ice geometries to quantify the impact of surface roughness on local impingement distributions. The results reveal that physical roughness induces a highly nonuniform collection efficiency, with droplet impingement intensely concentrated on upstream-facing surfaces of roughness elements, creating sheltered shadow zones immediately downstream. While the spanwise-averaged collection efficiency remains remarkably similar to that of an equivalent smooth body, idealized smooth surfaces completely suppress these localized impingement peaks. Ice accretion simulations demonstrate that this localized impingement creates a self-reinforcing feedback loop, actively amplifying existing roughness features over time. These findings provide a direct physical explanation for the formation of characteristic rime ice structures and highlight the critical role of local surface topology in the accretion process.
A method for predicting aerodynamic flows using learned kernel functions for the underlying boundary element problem is introduced. A formulation for the baseline potential flow kernels is presented, followed by the methodology of learning the kernel from a computational fluid dynamics dataset. These kernels are given an arbitrary formulation, and a gradient-based approach is used for the learning step, restricting the description to differentiable functions. The problem of relearning a potential flow vortex is presented, followed by learning steady 2D compressible flow around thick airfoils by using a neural network kernel. The resulting learned kernel solution yielded more accurate velocity and pressure distributions than the potential flow baseline. Lastly, the impact of enforcing rotational and translational invariance properties on the kernel definitions is investigated, which finds that more generalizable models can be created at the expense of accuracy.
Engine fan noise contains multiple radial modes, and the transmission loss (TL) of acoustic liners in such cases is influenced by both the power ratio among these radial modes and the phase difference between them. The objective of this paper is to develop a method for the rapid and accurate calculation of the attainable range of TL. Based on the transfer element method (TEM), a transmission matrix can be established, which directly relates the modal coefficient between the incident sound waves and the transmitted waves at the liner outlet. Following a mathematical processing of the transmission matrix, the Rayleigh quotient theorem is applied to analyze its eigenvalues, facilitating rapid and accurate calculation of the extreme values (maximum and minimum) of TL. TEM-based TL contour maps for two- and three-mode incidence, obtained from varying power ratios and phase differences, validate the predicted extreme values. For the scenario involving two radial modes, it is observed that the power ratios at the maximum and minimum TLs are reciprocals, which are characterized by a phase difference of 180 deg, and the underlying mechanisms are elucidated.
In this article, we present a comparison of experimental measurements of density fluctuations over a hollow cylinder to direct numerical simulation (DNS) data at similar conditions and wall-normal positions. The experimental observations are made using a closely spaced multiple-beam-pair focused laser differential interferometry (FLDI) flow diagnostic, allowing data to be acquired above 1 MHz. The power spectral density (PSD), autocorrelation and cross-correlation, and coherence spectra computed from FLDI and DNS data are presented for comparison. The PSD spectra obtained from experimental results and numerical analysis show good agreement in the energy-containing and inertial subrange, with a minor deviation in the higher frequencies of the dissipation range. A comparison of the cross-correlation maintains the agreement in the energy-containing and inertial ranges, but shows slightly higher correlation values for DNS data until it asymptotically approaches zero. The coherence spectra consistently show that the DNS data remain better correlated at higher frequencies. However, both datasets show steep reductions in correlation with increasing probe separation distances, extending into the lower frequencies of the energy-containing and inertial ranges.
The interaction between a grazing turbulent boundary layer at Mach 0.3 and tonal acoustic waves with an amplitude equal to 145 dB over a single-degree-of-freedom acoustic liner is investigated experimentally using laser Doppler velocimetry. Streamwise and wall-normal velocity profiles are measured along the liner, and the acoustic-induced velocity component is extracted from time-resolved measurements via a turbulence-rejection technique, allowing the separation of coherent acoustic and turbulent velocity fluctuations within the boundary layer. The influence of the acoustic source position, upstream or downstream of the liner, is examined. Results show that acoustic excitation near liner resonance frequency reduces the near-wall streamwise mean velocity while enhancing velocity fluctuations, producing a localized hump in second-order statistics. The spatial distribution and amplitude of the acoustic-induced velocity above the liner are strongly dependent on source position and exhibit significant distortions within the near-wall region, revealing a pronounced distortion of the plane wave acoustic field. These measurements provide direct experimental evidence that the turbulent boundary layer modifies the structure of the propagating acoustic wave. These observations highlight that source orientation significantly modulates the transfer of acoustic energy to the turbulent boundary layer, providing new insights into the mechanisms governing acoustic–turbulence interactions over lined surfaces.
This study investigates the free vibration, buckling response, and optimal design of functionally graded sandwich plates with a reentrant auxetic core. A Ritz-based formulation is utilized, combining global in-plane admissible functions with a spectral through-thickness representation. The auxetic core is modeled as a homogenized orthotropic continuum. The results show that natural frequency monotonically decreases with the increase of the relative density, indicating the dominance of the mass effects on the dynamic response. On the other hand, the critical buckling load increases with density, confirming the stiffness-dominated nature of the stability behavior. The effect of material gradation is strongly dependent on the chosen law: power-law and exponential distributions have a great impact on both responses, while the sigmoid distribution gives almost invariant results, showing more stable structural behavior. The geometric parameters of the auxetic core lead to significant nonlinear effects, with the strut length exhibiting a non-monotonic behavior and the reentrant angle always improving the buckling resistance. The optimization results show a clear tradeoff between mass, frequency, and buckling performance, with feasible solutions forming a narrow Pareto region. The best compromise is found at an intermediate density level ([Formula: see text]) with a power-law distribution.
This study investigates the influence of rarefaction on hypersonic flow over compression corners in the near-continuum regime and evaluates the validity of improved Navier–Stokes models for such flows. Numerical simulations are performed using the direct simulation Monte Carlo (DSMC) method, Navier–Stokes (NS) equations, and a data-improved NS (DiNS) model with constitutive corrections. The rarefaction effects are systematically examined over a range of Knudsen and Mach numbers, with particular emphasis on flow structure, surface aerothermodynamic quantities, and separation behavior. The results show that rarefaction induces coupled modification of interaction region, which is characterized by an upstream shift and augmentation on interaction-induced peaks in coefficients of skin friction and heat flux. Increasing the Knudsen number weakens shock-wave/boundary-layer interaction and suppresses flow separation. A separation criterion map in [Formula: see text] space is constructed, which identifies a regime where rarefaction significantly impacts the prediction of separation. Model comparisons demonstrate that improved NS models provide significant improvement over the NS method. Although the velocity slip enhances prediction accuracy, constitutive corrections in the DiNS model further improve agreement with DSMC results. These findings highlight the critical role of rarefaction in hypersonic compression corner flows and demonstrate that both slip boundary conditions and nonequilibrium transport effects are essential for accurate predictions in the near-continuum regime.
Experiments investigated the interaction of a 45-deg-pitched, finite-span, rectangular synthetic jet (SJ) upstream of a low-aspect-ratio circular cylinder (pin) with and without a jet-assisted surface-mounted actuator (JASMA) in a laminar boundary layer over a flat plate. The SJ generated a train of hairpin vortices that convected into the vicinity of the pin/JASMA downstream. The interaction was quantified using particle image velocimetry and stereoscopic particle image velocimetry at multiple planes, which were aggregated into a flowfield volume. The interaction of the SJ with the pin resulted in a breakup of the SJ vortices. Moreover, the centerline flowfield was similar to the wake of the pin alone, but outboard of the pin, the SJ locked the horseshoe vortices to its driving frequency. The interaction of the SJ with either the pin or the JASMA resulted in increased turbulent kinetic energy, prolonged streamwise vorticity, and increased downwash downstream. Additionally, the SJ/JASMA combination increased momentum injection into the boundary layer compared to the case when only the SJ was actuated. Results of this study are encouraging for application of the JASMA as a low-cost actuator to utilize naturally occurring turbulent vortical structures for more efficient separation control.
A gust mitigation technique is investigated for a NACA 0012 airfoil configured with a jet functioning as a fluidic spoiler, which is located on the upper surface near the trailing edge. Large-eddy simulations test the wing-jet configuration at a zero angle of attack with Mach = 0.1 and Reynolds number = 50,000 freestream conditions, encountering a discrete one-minus cosine transverse gust of length-to-chord ratio of [Formula: see text]. Two gusts are tested, which are defined by peak effective angles of attack of [Formula: see text] and 27 deg, imparting undesirable lift forces. The first gust is characterized by gradual separation and turbulent transition of the laminar boundary layer, whereas the more challenging [Formula: see text] gust is dominated by a substantial leading-edge vortex (LEV). The jet is modeled as a velocity boundary condition activated by a proportional feedback controller with the objective of reducing the gust induced lift. A supplementary parametric study determined the optimal jet variable to be a near-vertical blowing angle of [Formula: see text], which was measured relative to the downstream direction, and a momentum coefficient of [Formula: see text]. These parameters were carried forward to detailed aerodynamic studies examining the unsteady pressure, vorticity, and skin friction, comparing the baseline and controlled gust responses. The results show the jet spoiler works by stagnating the incoming flow to create a high pressure on the upper surface while simultaneously augmenting the trailing-edge vortex, accelerating the flow underneath the wing and decreasing the lower-surface pressure, the combined effects of which mitigate the imparted lift from the gust. This control mechanism is robust and only diminishes locally by the low pressure of strong vortical structures, which are exacerbated in the LEV gust case.
This study presents a comprehensive analysis of sloshing dynamics in rectangular tanks equipped with flexible antisloshing devices (ASDs), focusing on how varying baffle elasticity influences fluid–structure interaction (FSI). Three configurations, differing in baffle submergence, are investigated using a two-way partitioned numerical approach. This framework integrates the finite element method for structural modeling with the finite volume method and a volume of fluid approach for free-surface reconstruction. Numerical decay tests are conducted to evaluate the system’s damping characteristics and sloshing frequencies. Furthermore, a reduced-order acoustic-structural model is introduced to efficiently predict fundamental frequency shifts across different flexibilities and submergence levels. This reduced-order model demonstrates strong agreement with the full-order FSI simulations, yielding a maximum relative error of approximately 4% when its foundational hypotheses hold, thereby confirming its value as a computationally inexpensive design tool. Crucially, the study reveals that the elastic restoring motion of the flexible baffle enhances coherent vortex formation. This phenomenon strongly aligns with the evolution of the fluid’s energy dissipation rate, indicating that increased vorticity is critical for effective damping. Ultimately, these findings demonstrate that flexible ASDs can match or exceed the damping performance of rigid counterparts while significantly reducing structural weight.
Pyroshock, generated by pyrotechnic explosions, poses a severe threat to spacecraft by potentially causing fatal damage to critical components. Reliable ground simulation techniques are consequently essential for evaluating the resistance of spacecraft-borne instruments to pyroshocks. Specifically, laser-induced shock is believed to have significant potential to simulate real pyroshock environments, but direct experimental validation is still lacking. This study performs a comparative experimental investigation of pyroshock and laser shock responses (LSRs). Multiple detonating cord-based experiments were performed to investigate pyroshocks with varying explosive charge masses and shock wave propagation distances. The involved explosive train structures were characterized to identify the primary impact sources. Various LSR experiments were performed to evaluate the laser-induced shock response with diverse laser pulse energies. Water confinement was employed to enhance the shock amplitude. Results demonstrate that the shock response spectrum amplitudes scale with both explosive charge mass and laser energy, reaching near-field pyroshock environment levels. A close similarity is observed between the two types of shock responses. This work provides, for the first time, direct experimental verification of the feasibility of using laser-induced shock to simulate near-field pyroshock environments.
To address the tradeoff among high force, fast response, and compact volume in active magnetic dry friction dampers, this paper proposes a systematic multi-objective optimization framework centered on an interpretable global design surface. A parametric analytical model mapping the electromechanical couplings is established and validated via ANSYS simulations. To solve the high-dimensional conflict, an improved genetic algorithm integrating an adaptive mutation strategy and a random immigration mechanism is developed, significantly enhancing global convergence and population diversity. The framework progresses from bi-objective to tri-objective optimization, executing high-precision explicit mathematical regressions on the generated Pareto fronts. Furthermore, a system-level transient dynamic validation under sudden unbalance is implemented on a benchmark rotor platform, rigorously verifying that the optimized actuator parameters (300 N, 2 ms) successfully suppress the initial peak amplitude by 20.39% and curtail the settling time by 73.58% compared to nonoptimized configurations. The results demonstrate that the proposed framework successfully transforms discrete optimization datasets into an invertible and continuous engineering decision surface, providing a reliable design paradigm for aerospace electromagnetic actuators.
An experimental study was conducted on a NACA 0025 airfoil at [Formula: see text] to investigate the interaction between disparate forcing frequencies on separation control and coherent flow dynamics. Using a dual-row microblower array, forcing frequencies were set to [Formula: see text] and [Formula: see text] at two distinct chordwise locations. While single-row low-frequency forcing promotes lift recovery through the formation of large-scale coherent vortices, it introduces significant periodic unsteadiness. In contrast, dual-frequency forcing reveals a strong coupling mechanism where high-frequency actuation dominates the global response. Proper orthogonal decomposition analyses demonstrate that high-frequency forcing effectively modifies the large-scale structures into smaller, more localized vortices, redistributing fluctuation energy into higher-order modes. Results indicate that the sequence of actuation is critical: upstream low-frequency entrainment combined with downstream high-frequency stabilization yields the most favorable lift-drag tradeoff among the tested cases. This multifrequency strategy effectively preserves lift recovery while mitigating the deleterious unsteadiness associated with large-scale separation.
