Search Results (1,077)

Accurate prediction of the flow field is crucial to evaluating the aerodynamic performance of an aircraft. While traditional computational fluid dynamics (CFD) methods solve the governing equations to capture both global flow structures and localized gradients, they are computationally intensive. Deep learning-based surrogate models offer a promising alternative, yet often struggle to simultaneously model long-range dependencies and near-wall flow gradients with sufficient fidelity. To address this challenge, this paper introduces the Message-passing And Global-attention block (MAG-BLOCK), a graph neural network module that combines local message passing with global self-attention mechanisms to jointly learn fine-scale features and large-scale flow patterns. Building on MAG-BLOCK, we propose FLOW-GLIDE, a cross-architecture deep learning framework that learns a mapping from initial conditions to steady-state flow fields in a latent space. Evaluated on the AirfRANS dataset, FLOW-GLIDE outperforms existing models on key performance metrics. Specifically, it reduces the error in the volumetric flow field by 62% and surface pressure prediction by 82% compared to the state-of-the-art. Full article

Modern gas turbines for aeroengines operate at ever-increasing inlet temperatures to maximize thermal efficiency, power, output and thrust, subjecting turbine blades to severe thermal and mechanical stresses. To ensure component durability, effective cooling strategies are indispensable, yet they strongly influence the underlying aerothermal behavior, particularly in transonic regimes where shock–boundary layer interactions are critical. In this work, a comprehensive Reynolds-Averaged Navier–Stokes (RANS) investigation is carried out on the LS89 transonic turbine cascade, considering both adiabatic and cooled wall conditions. Three operating cases, spanning progressively higher outlet Mach numbers (0.84, 0.875, and 1.020), are analyzed using multiple turbulence closures. To mitigate the well-known model dependence of RANS predictions, a model-averaging strategy is introduced, providing a more robust prediction framework and reducing the uncertainty associated with single-model results. A systematic mesh convergence study is also performed to ensure grid-independent solutions. The results show that while wall pressure and isentropic Mach number remain largely unaffected by wall cooling, viscous near-wall quantities and wake characteristics exhibit a pronounced sensitivity to the wall-to-recovery temperature ratio. To support further research and model benchmarking, the complete RANS database generated in this work is released as an open-source resource and made publicly. Full article

The main objective of the study is to evaluate the recovery potential of the collapsed semi-anadromous pikeperch population (Sander lucioperca L.) in the Azov Sea during 2021–2030. We use a Ricker-based age-structured model that accounts for the effects of salinity and temperature on reproduction. In earlier work, the model predicted and explained the pikeperch stock collapse as the consequence of salinity and temperature exceeding the species’ tolerance limits. To assess the probability of stock recovery, we conducted a long-term retrospective validation and ran Monte Carlo projections under alternative climate scenarios with supplemental management actions. The results confirm that the dynamics of the pikeperch population in the Azov Sea are essentially environment-driven and negatively impacted by the large positive anomalies in both water temperature and salinity. Simulations suggest that either a substantial and persistent artificial restocking of juvenile recruits, or mostly unlikely scenarios of simultaneous reduction in salinity and temperature combined with additional restocking can provide conditions for the stock restoration within the decade considered. Based on these projections, we recommend a suite of urgent restoration measures to create the conditions required for future stock recovery. Full article

The high computational cost of high-fidelity CFD simulations forms a major bottleneck in aerodynamic design. This paper introduces a multi-fidelity transfer learning framework to rapidly predict airfoil aerodynamics with high accuracy. Our approach involves pre-training a deep fully connected neural network on a large dataset of low-fidelity Euler simulations. The pre-trained model is then fine-tuned using a limited set of high-fidelity RANS data, enabling efficient knowledge transfer from low- to high-fidelity domains. A specialized logarithmic-exponential normalization method is developed to handle the scale differences between aerodynamic coefficients. The framework demonstrates exceptional performance: after fine-tuning with only 700 high-fidelity samples, the model accurately predicts pressure distributions (lowest RMSE = 0.053) and force coefficients (R² > 0.947 for lift and drag). This method successfully bridges the gap between computational efficiency and high accuracy, providing a powerful data-driven surrogate model that can significantly accelerate the aerodynamic design and optimization process. Full article

This study compares enhanced turbulence models in a natural river channel 3D simulation under extreme hydrometeorological conditions. Using ANSYS Fluent 2024 R1 and the Volume of Fluid scheme, five RANS closures were evaluated: realizable k–ε, Renormalization-Group k–ε, Shear Stress Transport k–ω, Generalized k–ω, and Baseline-Explicit Algebraic Reynolds Stress model. A segment of the Santa Catarina River in Monterrey, Mexico, defined the computational domain, which produced high-energy, non-repeatable real-world flow conditions where hydrometric data were not yet available. Empirical validation was conducted using surface velocity estimations obtained through high-resolution video analysis. Systematic bias was minimized through mesh-independent validation (<1% error) and a benchmarked reference closure, ensuring a fair basis for inter-model comparison. All models were realized on a validated polyhedral mesh with consistent boundary conditions, evaluating performance in terms of mean velocity, turbulent viscosity, strain rate, and vorticity. Mean velocity predictions matched the empirical value of 4.43 [m/s]. The Baseline model offered the highest overall fidelity in turbulent viscosity structure (up to 43 [kg/m·s]) and anisotropy representation. Simulation runtimes ranged from 10 to 16 h, reflecting a computational cost that increases with model complexity but justified by improved flow anisotropy representation. Results show that all models yielded similar mean flow predictions within a narrow error margin. However, they differed notably in resolving low-velocity zones, turbulence intensity, and anisotropy within a purely hydrodynamic framework that does not include sediment transport. Full article

In this study, the modeling of rough surfaces by eddy-viscosity-based roughness models is investigated, specifically focusing on surfaces representative of deterioration in aero-engines. In order to test these models, experimental measurements from a rough T106C blade section at a Reynolds number of 400 K are adopted. The modeling framework is based on the k–ω–SST with Dassler’s roughness transition model. The roughness model is recalibrated for the k–ω–SST model. As a complement to the available experimental data, a high-fidelity test rig designed for scale-resolving simulations is built. This allows us to examine the local flow phenomenon in detail, enabling the identification and rectification of shortcomings in the current RANS models. The scale-resolving simulations feature a high-order flux-reconstruction scheme, which enables the use of curved element faces to match the roughness geometry. The wake-loss predictions, as well as blade pressure profiles, show good agreement, especially between LES and the model-based RANS. The slight deviation from the experimental measurements can be attributed to the inherent uncertainties in the experiment, such as the end-wall effects. The outcomes of this study lend credibility to the roughness models proposed. In fact, these models have the potential to quantify the influence of roughness on the aerodynamics and the aero-acoustics of aero-engines, an area that remains an open question in the maintenance, repair, and overhaul (MRO) of aero-engines. Full article

Harnessing Europe’s strong wave energy could support net-zero emissions goals, but extreme ocean loads still make wave energy expensive and delay the rollout of commercial wave-energy converters (WECs). To address this, the twin-floater CECO WEC has been redesigned into a single-pivot device called the Pivoting WEC (PWEC), which includes a passive duck diving survival mode to reduce extreme wave impacts. Its performance is evaluated using detailed wave simulations based on Reynolds-Averaged Navier–Stokes (RANS) equations and the Volume-of-Fluid (VoF) method in OpenFOAM-olaFlow, which is validated with data from small-scale (1:20) wave tank experiments. Extreme non-breaking and breaking waves are simulated based on 100-year hindcast data for the case study site of Matosinhos (Portugal) using a modified Miche criterion. These are validated using data of surface elevation and force sensors. Wave height errors averaged 5.13%, and period errors remain below 0.75%. The model captures well major wave loads with a root mean square error down to 47 kN compared to a peak load of 260 kN and an R² up to 0.80. The most violent plunging waves increase peak forces by 5 to 30% compared to the highest non-breaking crests. The validated numerical approach provides accurate extreme load predictions and confirms the effectiveness of the PWEC’s passive duck diving survival mode. The results contribute to the development of structurally resilient WECs, supporting the progress of WECs toward higher readiness levels. Full article

Long-span cable-supported bridges, such as cable-stayed and suspension bridges, are highly sensitive to wind-induced effects due to their flexibility, low damping, and relatively light weight. Aerodynamic analysis is therefore essential in their design and safety assessment. This study examines the aeroelastic stability of the Oued Dib cable-stayed bridge in Mila, Algeria, with emphasis on vortex shedding, galloping, torsional divergence, and classical flutter. A finite element modal analysis was carried out on a three-dimensional model to identify natural frequencies and mode shapes. A two-dimensional deck section was then analyzed using Computational Fluid Dynamics (CFD) under a steady wind flow of U = 20 m/s and varying angles of attack (AoA) from −10° to +10°. The simulations employed a RANS k-ω SST turbulence model with a wall function of Y+ = 30. The results provided detailed airflow patterns around the deck and enabled the evaluation of static aerodynamic coefficients—drag (C_D), lift (C_L), and moment (C_M)—as functions of AoA. Finally, the bridge’s aeroelastic performance was assessed against the four instabilities. The findings indicate that the Oued Dib Bridge remains stable under the design wind conditions, although fatigue due to vortex shedding requires further consideration. Full article

Traditional machine learning approaches for 5G network management relieve data from operational networks, which are often noisy and confounded, making it difficult to identify key influencing factors. This research addresses the critical gap between correlation-based prediction and interpretable, data-driven explanation. To this end, a software-defined standalone 5G architecture was developed using srsRAN and Open5GS to support multi-user scenarios. A multi-user environment was then simulated with GNU Radio, from which the initial dataset was collected. This dataset was further generated using a Conditional Tabular Generative Adversarial Network (CTGAN) to improve diversity and balance. Several machine learning models, including Linear Regression, Decision Tree, Random Forest, Gradient Boosting, and XGBoost, were trained and evaluated for predicting network performance. Among them, XGBoost achieved the best results, with an R² score of 0.998. To interpret the model, we conducted a SHAP (SHapley Additive exPlanations) analysis, which revealed that the download-to-upload bitrate ratio (dl_ul_ratio) and upload bitrate (brate_ul) were the most influential features. By leveraging a controlled experimental 5G environment, this study demonstrates how machine learning can move beyond predictive accuracy to uncover the fundamental principles governing 5G system performance, providing a robust foundation for future network optimization. Full article

Low Reynolds number flows are central to the performance of airfoils used in small unmanned aerial vehicles (UAVs), micro air vehicles (MAVs), and aerodynamic platforms operating in rarefied atmospheres. Consequently, a deep understanding of airfoil behavior and accurate prediction of aerodynamic performance are essential for the optimal design of such systems. The present study employs Computational Fluid Dynamics (CFD) simulations to analyze the aerodynamic performance of a cambered plate at a Reynolds number of 10,000. Two Reynolds-Averaged Navier–Stokes (RANS) turbulence models, $\gamma$–$R{e}_{\theta }$ and $k\text{-}{k}_L\text{-}\omega$, are utilized, along with the Unsteady Navier–Stokes (UNS) equations. The simulation results are compared against experimental data, with a focus on lift, drag, and pressure coefficients. The models studied perform moderately well at small angles of attack. The $\gamma$–$R{e}_{\theta }$ model yields the lowest lift and drag errors (below 0.17 and 0.04, respectively), while the other models show significantly higher discrepancies, particularly in lift prediction. The $\gamma$–$R{e}_{\theta }$ model demonstrates good overall accuracy, with notable deviation only in the prediction of the stall angle. In contrast, the $k\text{-}{k}_L\text{-}\omega$ model and the UNS equations capture the general flow trend up to stall but fail to provide reliable predictions beyond that point. These findings indicate that the $\gamma$–$R{e}_{\theta }$ model is the most suitable among those tested for low Reynolds number transitional flow simulations. Full article

This study presents a morphing Non-Uniform Rational B-Spline (NURBS) optimization method for enhancing sports car aerodynamics, with performance evaluation conducted in the vehicle’s symmetry plane. The morphing approach enables precise, smooth deformations of rear-end and spoiler geometries while preserving shape continuity, allowing controlled aerodynamic modifications suitable for comparative analysis. Flow simulations were carried out in ANSYS Fluent 2022 using the Reynolds-Averaged Navier–Stokes (RANS) equations with the standard k-ε turbulence model, selected for its stability and accuracy in predicting boundary-layer evolution, wake behavior, and flow separation in external automotive flows. Three configurations were assessed: the baseline model, a spoiler-equipped version, and two NURBS-morphed designs. The symmetry-plane evaluation ensured bilateral balance across all variants, enabling direct comparison of drag and lift performance. The results show that the proposed morphing strategy achieved notable lift reduction and favorable drag-to-lift ratios while maintaining manufacturability. The findings demonstrate that combining NURBS-based morphing with symmetry-plane aerodynamic assessment offers an efficient, reliable framework for vehicle aerodynamic optimization, bridging geometric flexibility with robust computational evaluation. Full article

Over the last decades, Computational Fluid Dynamics (CFD) has become a fundamental tool for the design of gas turbine combustors, partly making up for the costs and duration issues related to the experimental tests involving high-pressure reactive processes. Nevertheless, high-fidelity simulations of reactive flows remain computationally expensive, particularly for conjugate heat transfer (CHT) analyses aimed at predicting liner metal temperatures and characterising wall heat losses. This work investigates the robustness of a cost-effective numerical setup for CHT simulations, focusing on the prediction of cold-side thermal loads in industrial combustor liners under realistic operating conditions. The proposed approach is tested using both Reynolds-Averaged Navier–Stokes (RANS) and unsteady Stress-Blended Eddy Simulation (SBES) turbulence models for the combustor flame tube, coupled via a time desynchronisation strategy with transient heat conduction in the solid domain. Cold-side heat transfer is modelled using a 1D correlation-based tool, runtime coupled with the CHT simulation to account for cooling-induced thermal loads without explicitly resolving complex cooling passages. The methodology is applied to a single periodic sector of the NovaLT^TM16 annular combustor, developed by Baker Hughes and operating under high-pressure conditions with natural gas. Validation against experimental data demonstrates the methodology’s ability to predict liner metal temperatures accurately, account for modifications in cooling geometries, and support design-phase evaluations efficiently. Overall, the proposed approach offers a robust trade-off between computational cost and predictive accuracy, making it suitable for practical engineering applications. Full article

This paper presents a mathematical modeling and numerical analysis of fluid-thermal processes in a two-stage steam turbine cascade, focusing on the application and comparative assessment of turbulence models in computational fluid dynamics (CFD) simulations. Using the finite volume method implemented in the ANSYS CFX-Task Flow (ANSYS CFX 2022 R2) workflow, the study investigates the performance of standard k-ε, k-ω, and SST turbulence models in predicting flow structures, pressure fields, and velocity distributions within the turbine flow passages. The governing equations, including the Reynolds-Averaged Navier–Stokes (RANS) equations and associated energy and constitutive relations, are solved in conservative form under compressible flow conditions. Experimental data from turbine tests performed at the Institute of Fluid Machinery at Lodz University of Technology are used for validation. Results demonstrate that turbulence modeling significantly influences the accuracy of predicted flow phenomena. The study identifies strengths and limitations of the models in capturing complex three-dimensional flow structures and provides quantitative error margins and practical guidance for their application in industrial turbine flow simulations. Full article

To mitigate the impact of hydroelectric power plants on downstream fish migration, fish-friendly intakes, combining a low bar spacing rack and several bypasses, are implemented. There are still sites that can be improved thanks to a better bypass design. For this purpose, Computational Fluid Dynamics (CFD) can be a useful tool, even if such devices are still uncommon. This paper investigates the use of a 3D model based on the Reynolds-Averaged Navier–Stokes (RANS) equation for a single phase to simulate the flow in a real-scale water intake equipped with an inclined bar rack and three surface bypasses. The results of numerical simulations are compared to in situ measurements of flow velocities at four cross-sections along the rack, gauging the discharge flowing into the bypasses. The simulated velocities are in accordance with the velocities measured in situ, with a mean square error for the longitudinal velocity ($v_x$) of 0.034 (m²/s²) for the initial simulation and 0.021 (m²/s²) for the improved simulation. The split of the total bypass discharge between the three bypass entrances was satisfyingly predicted by the simulation with the true inlet velocity condition, showing the significant influence of upstream flow non-uniformity. Full article

Chest X-ray (CXR) imaging is essential for diagnosing thoracic diseases, and computer-aided diagnosis (CAD) systems have made substantial progress in automating the interpretation of CXR images. However, some existing methods often overemphasize local features while neglecting global context, limiting their ability to capture the broader pathological landscape. Moreover, most methods fail to model label correlations, leading to insufficient utilization of prior knowledge. To address these limitations, we propose a novel multi-label CXR image classification framework, termed the Label Masking-enhanced Residual Attention Network (LMeRAN). Specifically, LMeRAN introduces an original label-specific residual attention to capture disease-relevant information effectively. By integrating multi-head self-attention with average pooling, the model dynamically assigns higher weights to critical lesion areas while retaining global contextual features. In addition, LMeRAN employs a label mask training strategy, enabling the model to learn complex label dependencies from partially available label information. Experiments conducted on the large-scale public dataset ChestX-ray14 demonstrate that LMeRAN achieves the highest mean AUC value of 0.825, resulting in an increase of 3.1% to 8.0% over several advanced baselines. To enhance interpretability, we also visualize the lesion regions relied upon by the model for classification, providing clearer insights into the model’s decision-making process. Full article