The study presents a series of Wall-Modelled Large Eddy Simulations (WMLES) using the high-resolution GPU-accelerated CABARET method to investigate boundary layer tripping effects on a NACA 0012 airfoil. Simulations are performed at zero angle of attack for chord-based Reynolds numbers of 1.0 × 106 and 1.5 × 106, corresponding to the BANC III experimental conditions. For far-field noise prediction, the LES solutions are coupled with Amiet theory and the semi-empirical Brooks model. The numerical framework is validated against experimental data for pressure and skin friction coefficients, boundary layer flow and turbulence profiles, as well as wall pressure and far-field noise spectra. Having validated the baseline model, the trip geometry is systematically varied to evaluate its effect on the flow, turbulence, and pressure fluctuations in the boundary layer and the wake region around the trailing edge using statistical analyses, including Proper Orthogonal Decomposition and conditional averaging. Compared to the meanflow and turbulence details, the pressure fluctuations around the trailing edge are found to be insensitive to the trip parameters within the experimental error bar, indicating that the dominant coherent pressure structures responsible for noise scattering remain little affected by the upstream transition-triggering mechanism.
Fast and accurate simulations are essential for engineering design and optimisation of large, coupled vibroacoustic finite element systems, where model complexity and parameter dependence often make full-order model solutions computationally prohibitive. To address this challenge, this work presents a systematic comparison of parametric model order reduction (PMOR) methods within a dynamic substructuring framework. Global and local PMOR approaches are combined with component mode synthesis, comparing modal and Krylov-based reduction techniques. The methods are applied to structural optimisation tasks on two test models of different scale and complexity: a plate–cavity system, and a water-filled container comprising two coupled vibroacoustic substructures. Results show that integrating substructuring and adaptive sampling with PMOR reduces the computational cost of optimisation by up to 91% while maintaining accuracy, and that the local approach based on interpolation of reduced matrices, combined with Krylov-based component mode synthesis, offers the best efficiency–accuracy trade-off.
In the present study, the crack propagation process in pre-notched concrete beams subjected to three-point bending was monitored by tracking variations in resonant frequencies. The tests were conducted under displacement-controlled conditions until final failure, using an incremental loading procedure. At each load step, the beam was subjected to impulse excitation, and the response was recorded by sensors positioned along the beam. A signal-processing methodology based on the Fast Fourier Transform (FFT) was then applied to extract the resonant frequencies from the recorded data. To complement the experimental analysis, a numerical investigation was performed. A version of the Lattice Discrete Element Method (LDEM) was implemented to simulate damage progression. The damage state obtained from the LDEM simulation was subsequently imported into a Finite Element Method (FEM) solver to perform modal analysis, providing additional dynamic characteristics such as mode shapes and resonant frequencies for direct comparison with the experimental results. The changes in resonant frequencies and mode shapes obtained from both the numerical and experimental analyses were compared. The results demonstrate that these dynamic characteristics are effective for monitoring damage evolution in the structure and predicting final failure. Emphasis is placed on the role of numerical simulation in supporting the interpretation of the experimental damage process.
This study investigates the dynamic response mechanisms of spatial mechanisms under the combined effects of frictional clearance joints and flexible components. Dynamic models were developed for both pure rigid and rigid-flexible coupling configurations of a variable stator vane (VSV) mechanism with clearance joints, using an improved contact force model and a friction model calibrated through experiments. The analysis shows that frictionless joints cause slippage, compromising stability, while friction improves it. Among the five friction models, the Brown-McPhee (B-M) model was found to be the most stable and efficient. The friction coefficient's effect on stability is non-monotonic, with an optimal kinetic friction coefficient of 0.355 for the VSV mechanism. When the clearance exceeds 0.25 mm, the system becomes chaotic with increasing clearance. While flexible components effectively suppress vibrations, overly flexible designs are not optimal. Additionally, a friction parameter identification method based on material experiments is proposed. The effects of friction and flexibility on the mechanism's dynamic response were validated through VSV experiments. Results align with theoretical analysis, confirming the crucial roles of friction, clearance, and flexibility in optimizing mechanism performance.
In a typical adaptive filter used for active noise control (ANC), the coefficient vector is sometimes highly sparse, but the existing ANC algorithms do not utilize this characteristic, leading to suboptimal noise reduction performance. Additionally, prolonged reverberation time causes existing ANC algorithms to exhibit overwhelming computational complexity. To overcome the two disadvantages mentioned above, we present a correntropy-induced metric constrained set-membership filtered-x normalized LMS (CIM-SM-FxNLMS) algorithm, which adds a sparse penalty and a constraint to the cost function of the FxNLMS to exploit the coefficient vector’s sparsity and lower the computational complexity. This paper provides a detailed description and derivation of the CIM-SM-FxNLMS. An experimental setup was built to measure the actual secondary and primary acoustic paths. Drawing upon the measured secondary and primary acoustic paths, the performance of the proposed algorithm is analyzed through numerical simulation. According to the simulation analysis, the proposed CIM-SM-FxNLMS demonstrates superior noise attenuation capabilities compared to existing ANC algorithms. Furthermore, the proposed CIM-SM-FxNLMS offers a significant reduction in computational complexity compared to the correntropy-induced metric constrained FxNLMS (CIM-FxNLMS).
Honeycomb structures have wide applications in the aerospace field due to their excellent structural properties. However, achieving high-resolution in-service inspection imaging without couplant remains a challenge. In this paper, a high-resolution imaging method based on the low frequency localized resonance mode in honeycomb structures is proposed. By analyzing the displacement eigen-fields of the dispersion relation, it is revealed that the skin’s localized resonance mode is caused by the restriction of the bond joint. Due to the low frequency of this mode, it can be easily excited in the structure without couplant and can propagate over long distances. Experimental results demonstrate that when this mode is excited in the structure, the signal amplitude response difference can reach about 25 dB at the honeycomb hole and core-skin bond joint, and therefore the hexagon cell (with a side length of 4 mm and a core wall thickness of 0.1 mm) of the intact area can be clearly observed in the imaging result. In contrast, these differences disappear in the defective area due to core-skin debonding. Further experiments confirm that the proposed method can obtain a high-resolution image, which is comparable to the result of water immersion focusing ultrasonic C-scan at a center frequency of 20 MHz. This method can also be integrated with laser vibrometers to achieve non-contact, high-precision imaging over large areas, which is more suitable for in-service inspection.
