Journal of Hydrodynamics最新文献

High-fidelity simulation of incompressible turbulent channel flows at high Reynolds numbers remains computationally intensive due to their multi-scale nature, necessitating efficient strategies to reconstruct high-resolution (HR) fields from low-resolution (LR) data. This study introduces a physics-guided neural network (PGNN) framework that leverages the elliptic nature of the pressure field in incompressible flows to enhance super-resolution (SR) reconstruction of turbulent velocity fields. Inspired by the global correlation-capturing capabilities of attention mechanisms, we incorporate LR pressure data as auxiliary input features to guide neural networks in inferring local velocity components, thereby explicitly encoding physical constraints into the learning process. High-fidelity training datasets are generated via large-eddy simulations, with LR fields derived through spatial filtering using down-sampling ratios (DSR) of 4, 8. A physics-guided U-Net (pgU-Net) is developed, contrasting with a pure data-driven U-Net baseline. Results demonstrate that integrating pressure fields significantly improves reconstruction accuracy, particularly under challenging DSR of 8, the pgU-Net achieves an R² score of 0.8048, outperforming the data-driven model (0.6792) and bi-cubic interpolation (−0.1801). By leveraging pressure fields as physical knowledge encoding global flow correlations, our framework effectively addresses the multi-scale reconstruction challenges in turbulent flows while maintaining computational efficiency. By embedding elliptic pressure correlations as physical priors, this framework pioneers a hybrid approach that bridges data-driven learning and fluid physics, offering a robust approach to reduce computational costs while enhancing the fidelity of turbulent flow predictions.

The T-shaped hydrofoil, consisting of the surface-piercing strut and submerged hydrofoil, plays a crucial component of a hydrofoil system. This study investigates the hydrodynamic performance and flow patterns of the T-shaped hydrofoil under varying flow parameters and geometric characteristics, utilizing towing tests and large eddy simulation (LES) techniques. The computational and experimental results demonstrate a high degree of agreement. The analysis reveals that both lift and drag coefficients increase with the angle of attack. As the Reynolds number increases, the drag coefficient decreases, while the lift coefficient remains relatively stable. When the T-shaped hydrofoil is deeply submerged, the free-surface deformation primarily manifests as spray flows. Notably, the free surface effect becomes significant at submerged depths shallower than 3.5 times the chord length. As the submerged depth decreases, the free-surface depression in the wake is enhanced; however, both the lift and drag coefficients of the T-shaped hydrofoil decrease, governed by the S foil and strut, respectively. The incorporation of winglets plays a vital role in optimizing hydrodynamic performance by redistributing tip vortices away from the hydrofoil, thereby reducing the vorticity and spiral motion induced by the tip effect. Additionally, the interference effect from the strut enhances both lift and drag performance of the submerged hydrofoil, although it may also lead to an earlier cavitation inception. The increased lift and drag can be primarily attributed to the lateral spread of negative pressure arising from the separation region on the strut.

In soil water infiltration problems, the basic control equation, i.e., Richards equation is a nonlinear partial differential equation (PDE), and is difficult to solve. In this study, a finite difference lattice Boltzmann method (FDLBM), in which the D1Q5 model is employed as the lattice layout scheme, is developed to solve the 1-D Richards equation with water content as the main variable in unsaturated soil. The relationship between the lattice Boltzmann equation (LBE) and the Richards equation is established using a multiscale expansion technique. Numerical examples show that LBM is suitable to solve Richards equation in unsaturated soil water infiltration problems.

In this paper, the dynamic characteristics of a vertical vortex are examined by the theoretical method and the particle image velocimetry (PIV) technology. The results of the theoretical analysis validate that the main parameters of the vertical vortex are the Reynolds number Re, water level WL* and velocity circulation Γ_N. The experimental results indicate that, for the same Re, the low water levels are responsible for an intensive vortex intensity (maximum vorticity), but for a high-water level, water level is no longer a significant influencing factor on vortex intensity. The Γ_N number increases with the rising of the Reynolds number Re and drops with the height of measurement section enhance for a low water level case. For the higher WL*, Γ_N does not decrease significantly. The formation and evolution of the vertical vortex in the present paper can be utilized or restrained by controlling the relevant dynamic parameters. Obtained research results may provide an important reference for engineering applications with possible occurrence of vertical vortex phenomenon.

This study investigates the inundation depths of urban floods induced by real storm events, focusing on the development and assessment of super-resolution model based on ensemble learning methods. Unlike traditional deep neural networks which require extensive training and high parameterization, this study utilizes ensemble learning model to reconstruct high-resolution flood predictions from low-resolution hydrodynamic simulations. Hydrodynamic modeling results of real pluvial flood event at various spatial resolution are used for constructing datasets and for training and testing the point-based super-resolution model. Influencing factors related to urban terrain, subsurface, rainfall inputs and the hydrodynamic modeling results at coarser resolutions are used as features in the super-resolution model on basis of Random Forest, in which hyperparameters are tuned with Bayesian optimization method. The trained super-resolution models effectively reconstruct high-resolution inundation conditions from 30 m to 5 m coarse resolution inputs, highlighting an increase in correlation coefficients and a decrease in root mean squared error (RMSE) as resolution improves. Dominant influencing factors in the super-resolution models are identified together with variances in their contributions to the model performance. Two optimization approaches are applied to enhance accuracy and mitigate overestimation at coarse resolutions for the super-resolution models. The first integrates outputs from various coarse resolution models as features, notably reducing overestimation, especially with finer 5 m resolutions. The second employs ensemble modeling with super-resolution models from different datasets, which improves the performance across all tested resolutions, demonstrating the robustness of combining multiple predictive models for better flood forecasting in urban environments.

For Francis turbines, frequent operations under extremely low load conditions result in significant noise and pressure fluctuation issues. These issues may cause vibration and fatigue damage to the unit, accompanied by difficulties in connecting to the grid and reductions in the power generation efficiency of renewable energy. However, there is limited research on the relationship between pressure fluctuations and the induced noise of Francis turbines during extreme operations. In the present study, an acoustic numerical simulation based on the Ffowcs Williams-Hawkings equation and large eddy simulation is used to analyze the acoustic performances of Francis turbines. In the current study, for evaluating the acoustic characteristics under such terrible conditions, the results of variable flow rate and guide vane opening conditions are compared. Results indicated that Francis turbine noise is mostly due to pressure fluctuations brought on by rotor-stator interference and corkscrew-shaped vortices. The blade passing frequency (BPF) of 130.00 Hz and the low frequency of 0.33 f_n (where f_n denotes the rotating frequency) are the key factors affecting pressure and noise fluctuations. The influence of low frequency is reduced as the flow rate rises, whereas the influence of BPF gradually increases. Besides, the hydrodynamic noise of Francis turbines is primarily low-frequency, with discrete and broad-band features. The rotating noise with distinct peak values and the turbulence noise produced by large-scale vortices (corkscrew-shaped vortices) make up the majority of low-frequency noise. Therefore, reducing pressure fluctuations is a key strategy for lowering flow-induced noise radiation.

This paper employs a combined experimental and numerical approach to investigate the influence of airflow characteristics—specifically air velocity ν_a and air density ρ_a—on the evolution of water-entry cavities at low Froude numbers (Fr<13). A custom-designed test platform enables control over air density ρ_a and water-entry initial velocity V₀. The velocity V₀ influences the cavity expansion rate, which in turn determines the air inflow velocity ν_a into the cavity. Based on the experimental results, a critical condition for surface seal is proposed: ρ^*·Fr ^2.62_c =315, where ρ^*=ρ_a/ρ₀, ρ₀ is the ambient density. For constant air density, deep seal dynamics exhibit negligible direct sensitivity to airflow when the Fr < Fr_c, aligning with classical inertial theories and a 1/2-power scaling law during radial collapse. As ρ^* or Fr increases beyond the critical threshold, the cavity closure mode transitions from deep seal to surface seal. Numerical simulations, based on finite volume method, reveals that when the flow field is approximately uniform, the ratio of air flow velocity ν_a to projectile velocity ν is ~1.5. Furthermore, the airflow-induced pressure difference basically satisfies Δp = ρ_au_a²/2, driving inward splash motion. Neglecting the change in projectile velocity, there is Δp ∝ ρ_aV₀²/2. When the splash is about to close, the flow field distribution is relatively complex, and the pressure relationship no longer holds consistently. Surface seal blocks the connection between the cavity and the external atmosphere, directly impacting the internal pressure dynamics and further influencing the deep seal characteristics.

The underwater launch of an axisymmetric body involves complex cavity-structure interactions. Studying the evolution of cavity pressure around an axisymmetric body is crucial for researching its motion stability. In this work, we propose a deep neural network model for cavity pressure distribution recovery, called CPDR-net. This model can reconstruct the full-domain distribution of surface pressure based solely on the local pressure distribution. The CPDR-net model was trained using numerical simulation data with different launch depths and initial velocities, and subsequently tested on two simulation datasets under new conditions. Both training and testing datasets are obtained from the ventilated cavitating flow over an underwater axisymmetric vehicle. Results demonstrated that CPDR-net can accurately predict the pressure distribution along each longitudinal line of the axisymmetric body and provide the pressure evolution over time for each point on the surface. Thus, we can obtain the evolution of surface pressure distribution throughout the entire voyage process based on the CPDR-net model. The findings from this study may provide a valuable reference for subsequent research on underwater launches.

The floating horizontal-axis tidal turbine (FHATT) stands out as the most commercially viable tidal energy device. This paper reviews recent literature on FHATT and summarizes experimental and computational fluid dynamics (CFD) methods employed in FHATT research. Based on this foundation, the coupling effects of wave and platform motion (pitch/roll) on FHATT hydrodynamic performance were investigated through flume experiments and CFD simulations. The variations of the power coefficient (C_P) and thrust coefficient (C_T) are analyzed under different platform motion periods, amplitudes, wave periods, and wave heights. The results demonstrate that under the coupling of waves and pitch motion, C_P, C_T exhibit dual-frequency oscillations based on the pitch period, with oscillation amplitudes increasing with both pitch frequency (wave frequency) and pitch amplitude (wave height). Within the working conditions of this study, the maximum mean output power under the coupling of pitch motion and waves increases by 26.1%. The maximum fluctuation amplitude of C_P reaches 349.8%. When waves and roll motion are coupled, wave parameters dominate, while the influence of roll motion can be ignored. Moreover, the hydrodynamic fluctuations induced by waves and platform motion can couple with each other. This coupling effect not only amplifies the fluctuation amplitude of hydrodynamic coefficients but also has the potential to offset each other. These findings provide insights into the structural design and system control of FHATT, serving as valuable references for FHATT development.

Vortex-induced vibration (VIV) of cylindrical structures is a critical fluid-structure interaction (FSI) phenomenon in ocean engineering. Simulating VIV accurately can be computationally expensive. This study presents a graphics processing unit (GPU)-accelerated simulation model for VIV utilizing the immersed boundary lattice Boltzmann method (IB-LBM), aiming to reduce computational costs while preserving accuracy. The program is developed using machine learning library JAX, which enables parallelism on GPU and multi-GPU platforms. The model incorporates multi-GPU parallelization and multi-block grid refinement strategies to enhance computational efficiency. Validation against existing high-fidelity simulation data demonstrates good agreement. Performance tests show significant speed-ups with GPU acceleration compared to traditional CPU-based approaches. These results underscore the potential of the developed simulator as an efficient and reliable tool for in-depth parametric studies and practical engineering analysis of VIV, facilitating more rapid design iterations and risk assessments for offshore structures.