Coral reefs worldwide are increasingly facing numerous threats from both global and local stressors (e.g. mass coral bleaching, ocean acidification, coastal development, sedimentation, etc.). Low-lying reef-lined nations are also vulnerable to coastal flooding under sea-level rise and extreme storm events. A full understanding of coral reef hydrodynamics is a crucial step towards interpreting the wide range of physical, chemical, and biological processes within reefs related to the above issues. This article reviews the state-of-the-art advances in modelling approaches for coral reef hydrodynamics based on an extensive literature review, and classifies them into physics-based models of four groups (coupled wave-current models, Boussinesq-type models, non-hydrostatic models, and Navier-Stokes-equation-based models) and AI-based models. The applications of each model group in the reef environments are extensively reviewed. Their strengths, limitations and future development directions are also discussed.
Hydrogen direct injection involves shock-containing, under-expanded jets for which Unsteady Reynolds–Averaged Navier–Stokes (URANS) predictions of penetration and mixture preparation are difficult to attribute to turbulence closure alone, because experimental, diagnostic, boundary-condition and numerical uncertainties can obscure model-form effects. This study assesses closure-model effects for a non-reacting, engine-representative high-pressure round [Formula: see text] jet through a controlled Large-Eddy Simulation (LES)–URANS numerical experiment with matched geometry, initial and boundary conditions, numerics and meshing strategy. A ten-realisation LES database, filtered within the dissipative range, is ensemble averaged using a Spherical Intersected Volume (SIV) operator to obtain statistically converged axisymmetric reference fields at approximately 3–5 times lower sampling cost than conventional ensemble sampling. Matched a-posteriori URANS calculations with STandarD (STD), Re-Normalisation Group (RNG), Rapid-Distortion RNG (RDRNG) and Generalised RNG (GRNG) k–ε closures quantify the coupled predictive response, while complementary a-priori frozen-field analyses isolate intrinsic momentum- and species-closure behaviour. With non-closure choices controlled, all tested URANS closures reproduce injection-phase penetration with low relative root-mean-square error of 2.56–[Formula: see text], indicating that larger URANS–experiment penetration discrepancies reported for similar configurations may include substantial non-closure contributions. Residual model-form errors are concentrated in mixture preparation: transport budgets identify turbulent scalar flux, rather than turbulent momentum transport, as the primary initiator of mass-fraction error, and error-propagation analysis shows that these scalar errors propagate through density, pressure and velocity coupling. Accordingly, the inferred LES-consistent turbulent diffusivity is space-, time- and model-dependent, with the corresponding [Formula: see text] spanning approximately 0.1–2, showing that a constant value cannot recover scalar transport consistently. The standard k–ε model gives the highest overall accuracy for the isolated round jet, whereas GRNG provides the most balanced RNG-family response and is therefore the more defensible compromise for engine-relevant applications requiring strain suppression. These results support closure-consistent, dynamically varying [Formula: see text] formulations, together with higher-fidelity realisable stress closures, pressure- and density-aware adaptive mesh refinement, and RNG-family extensions for future high-pressure gaseous-injection modelling.
Coagulant dosage modelling has been widely investigated in the literature; however, existing methods often operate as black boxes, failing to provide clear practical explanations for decision-makers. To address this limitation, this study introduces an interpretable framework using SHAP and LIME to highlight the underlying practical problems. Furthermore, this is the first study of its kind in Algeria, bridging a critical regional gap in the literature. Yet, we investigate the application of boosting machine learning models to predict coagulant dose at the Taksebt drinking water treatment plant. Herein we compare between: (i) multiple linear regression (MLR), (ii) random forest regression (RFR), (iii) adaptive boosting (AdaBoost), (iv) categorical boosting (CatBoost), (v) gradient boosted regression trees (GBRT), (vi) Histogram Gradient Boosting (HistGBRT), (vii) Light Gradient Boosting Machine (LightGBM), (viii) Natural Gradient Boosting (NGBoost), and (ix) eXtreme Gradient Boosting (XGBoost). All models were trained using daily measured water quality variables, while aluminium sulphate (Al2(SO4)3·18H2O) was the coagulant dose. The predictive performance was evaluated using root-mean-square error (RMSE), mean absolute error (MAE), Nash-Sutcliffe efficiency (NSE), and the correlation coefficient (R). Global and local model interpretability were conducted using SHAP and LIME, thereby enabling feature importance ranking. To determine whether the observed differences among models were statistically meaningful, two complementary statistical tests were applied: the Diebold-Mariano test and the Kruskal-Walli’s test. The results indicate that CatBoost delivered the most accurate predictions, achieving R, NSE, RMSE, and MAE values of 0.922, 0.849, 2.498, and 1.702 mg/L, respectively, and significantly outperforming the MLR model, which yielded an R of 0.713, a NSE value of 0.507, RMSE of 4.510 mg/L, and MAE of 3.576 mg/L, respectively. Based on SHAP analysis, UV254 was identified as the most influential feature, contributing 24% to the model predictions, whereas COU showed a negligible effect, accounting for only 5%.HighlightsBoosting models for coagulant dosage prediction.CatBoost provided the best numerical performances.UV254 was the most influential raw water quality variable.SHAP and LIME for model interpretability.
The computational efficiency bottlenecks of existing random flow generators were analysed. To address these bottlenecks, a three-dimensional turbulent wind field random parallel generator based on GPU (Graphics Processing Unit) acceleration strategies was proposed. The proposed method employs a ‘GPU stream cubes’ strategy to enable efficient parallel synthesis of inlet turbulence. Multiple simulation scenarios were used to evaluate different acceleration schemes and verify the effectiveness of the proposed approach. The three-dimensional scheme achieves speedups exceeding 100× in most configurations, whereas the one-dimensional scheme provides only limited acceleration. For identical wind-field parameters, the GPU-based results agree well with the traditional CPU-based method, indicating that accuracy and stability are preserved. Four representative urban wind-field cases, from an idealised block to a WRF–CFD multiscale-coupled scenario, further demonstrate the applicability of the method to city-scale inflow turbulence synthesis. In addition, an efficient turbulence generation software tool for wind turbine simulation was developed based on the proposed method.
This study employs a Weather Research and Forecasting model with 3-dimensional variational data assimilation (WRF-3DVAR) to integrate multi-source meteorological data from satellites, Light Detection and Ranging (LiDAR), and surface weather stations, introducing multi-scale observational constraints to reproduce the urban wind field during typhoon Yagi. The research findings are as follows: satellite radiance assimilation calibrates the typhoon track at the mesoscale; surface weather stations add physical constraints on the near-surface wind field, compensating for deficiencies in the model’s description of underlying surface characteristics; and LiDAR-derived vertical wind profiles refine the vertical momentum exchange processes within the typhoon boundary layer, significantly improving the simulation accuracy of near-surface wind profiles. After independent validation using high wind speed stations (GI, NGP, NLS, PEN, TMS, and YTS) and LiDAR data from North District North East, demonstrated that the WRF-3DVAR assimilating multi-source data can accurately reproduce the near-surface wind speed and wind profile evolution in urban areas. The contribution of this study is to provide a new ground-space-sky data assimilation framework for reproducing urban 3D wind field, improving weather warning systems, and formulating long-term disaster prevention and mitigation strategies, effectively strengthening the city’s resilience against super typhoons.
This study proposes a novel asymmetric coaxial counter-rotating dual-rotor wind turbine (A-DRWT). Based on the asymmetric configuration of the A-DRWT, this study develops a Coupled-Dual Rotor (CDR) wake model by extending the Gaussian-distribution-based Jensen framework from a single-rotor representation to a coupled two-rotor formulation. The novelty of the model does not lie in introducing another Gaussian deficit function itself, but in separately resolving the wakes of the upwind rotor (UR) and downwind rotor (DR) and coupling them according to their asymmetric geometry and aerodynamic interaction. A dedicated correction is further introduced for the DR to account for the non-uniform inflow induced by the UR wake and the support struts. The model shows good agreement with the CFD results, with a maximum deviation of 5.39%, and indicates that the A-DRWT may allow a 30–40% reduction in downstream spacing compared with an equivalent SRWT. This work therefore provides a CFD-validated design concept and a low-order predictive framework for A-DRWT wake assessment. However, because wind-farm wake losses depend on the coupled effects of energy extraction, wake recovery rate and turbine spacing, further farm-scale assessment is required before drawing direct conclusions regarding net wind-farm energy gain or layout optimization.
Assessing the mechanical behavior of the flow past cylinder arrays and the effect of environmental pollution is essential for layout design. This study presents an active learning surrogate modeling framework for predicting hydrodynamic and scalar transport properties in side-by-side and diamond cylinder arrays. Three independent Gaussian process regression models map the physical parameters, the Reynolds number, the layout angle, the gap ratio, and the Schmidt number to the mean drag coefficient, the root-mean-square lift coefficient, and the maximum entropy of the pollutant field, respectively. The models are embedded in a sequential design loop that seeds the parameter space with samples and iteratively selects new simulations from a candidate pool using a max-variance acquisition rule. This strategy focuses queries on high-uncertainty regions, reduces predictive confidence intervals, and avoids redundant runs. Across both layouts, the loop consistently lowers the negative log-likelihood and cumulative error, consistent response surfaces with calibrated uncertainty that appropriately inflates near extrapolation. The resulting surrogates enable fast evaluation of coupled fluid and pollutant behavior and provide a data-efficient pipeline for scientific design under uncertainty.
Air–water interactions in deep tunnel systems can cause safety risks. However, the multiphase flow dynamics under continuous air entrainment from upstream and release from downstream, which are the typical case in engineering practice, remain insufficiently understood. In this study, a three-dimensional computational fluid dynamics model was developed and validated based on an experimental study to investigate the periodic air–water behaviors in a dropshaft-tunnel system. With the numerical model, the features of air entrainment and release across various flow regimes were quantified, and the evolution of air–water flow corresponding to different stages of pressure fluctuations was clarified, which could hardly be measured from experiments. The simulation results revealed that the periodic pressure fluctuations were driven by the cyclic process of slug flow development in the tunnel, and the amplitude was mainly governed by the amount of water rushing into the air release shaft. Furthermore, the effects of different layout designs on the pressure fluctuations were explored numerically. It was found that increasing the distance between the air entrainment and release points could significantly amplify the pressure amplitude and prolong the fluctuation period. These findings contribute to new knowledge on air–water flow dynamics and design considerations of deep tunnel systems.
As a specialized pump with electromechanical integration characteristics, full tubular pump (FTP) features the flow region and motor region that operate in concert and are mutually coupled. The stator-rotor clearance flow (SRCF), as the key medium connecting these two regions, influencing both hydraulic energy dissipation and motor heat transfer. To investigate the energy characteristics of FTP, based on the model experiment and numerical simulation, this paper established an electromechanical energy transfer model considering the stator-rotor heat conduction effect (SRHC). Within the framework of fluid thermal coupling (FTC), the SRCF flow characteristics, motor dissipation and the entropy generation distribution were systematically analyzed under different rotational speeds and heat transfer conditions. The results show that SRHC changes the energy transfer and dissipation pathways of SRCF by regulating the relationship among different loss mechanisms. After considering SRHC in FTP system, the temperature rise of SRCF reduces its dynamic viscosity and enhances its flow capacity. On one hand, this weakens the shear action of SRCF on the rotor outer ring wall and reduces mechanical friction loss. On the other hand, the increase in flow inertia promotes near-wall turbulence development, causing part of the energy dissipation to shift from wall-shear dissipation to turbulence dissipation. Concurrently, convective heat transfer via SRCF significantly reduces the temperature levels of the stator and rotor, thereby reducing motor loss and ultimately improving the overall operating efficiency of FTP by up to 7.56% under the low-flow condition of 0.7Qd. This study reveals the SRCF-mediated FTC mechanism in FTP, coupling its hydraulic performance with motor performance and provides a theoretical basis for energy characteristics and efficiency optimization of electromechanical integrated pump systems.
Accurate simulation of wind-induced effects on building clusters under typhoon conditions is fundamental to the construction and operation of modern resilient cities. This paper develops an integrated computational framework coupling the Weather Research and Forecasting (WRF) model with an urban canopy model (UCM), large-eddy simulation (LES), and multi-degree-of-freedom (MDOF) structural models to simulate urban wind fields, surface wind pressures, and wind-induced dynamic responses of building clusters. The framework includes both mesoscale–urban canopy coupling and mesoscale–microscale coupling methods. Within this framework, the WRF-UCM coupling is used to reproduce typhoon-scale wind fields and provide local wind profiles, wind directions, and turbulence information for building-scale simulations. Subsequently, an LES model is used to resolve fluctuating wind fields and unsteady pressure loads within the target urban area. By integrating building data from geographic information systems (GIS) with MDOF structural models, regional-scale time-history analysis is conducted to evaluate the dynamic responses of the building cluster. The framework was applied to the building cluster surrounding the Galaxy Twin Towers in Shenzhen during Typhoon Saola in 2023. The WRF-UCM model reasonably reproduced the typhoon track and intensity, while the LES results captured representative building-induced flow and pressure features within the dense high-rise cluster. The LES-derived unsteady pressure loads were then transferred to MDOF structural models to assess the dynamic responses of multiple buildings. The results demonstrate that the proposed framework provides a continuous computational route from typhoon-scale atmospheric forcing to building-scale wind loading and cluster-level structural response assessment. The presented methodology can serve as a reference for refined simulation of wind-induced effects and wind-risk assessment of urban building clusters under typhoon conditions.
The rapid expansion of renewable energy systems and electric vehicles has increased demand for batteries. Effective thermal management is essential because of the significant effect of temperature on the lifespan, performance, and safety of batteries. This study aimed to enhance the cooling performance of a battery thermal management system through targeted structural modifications of the cooling system. A series of design configurations were evaluated and analyzed using Computational Fluid Dynamics (CFD) simulations to improve heat transfer and reduce the maximum temperature within a battery stack. The proposed modifications were developed based on a detailed analysis of the flow field and the identification of nonuniformities in the temperature of the preceding designs. It was found that the use of a deep notch combined with filleted branches of cooling systems for proper directing of the fluid flow provided the best case among the evaluated configurations. The results of the study revealed that the temperature uniformity of the battery stack reduces from [Formula: see text] in the reference case to [Formula: see text] in the optimal case. Furthermore, maximum temperature of the battery was reduced by 2.76°C in the optimal case in comparison with the reference case. Finally, a sensitivity analysis was performed on the impact of coolant inlet temperature and velocity on the maximum temperature of the stack, and it was observed that the effect of inlet temperature was more significant. According to the obtained results, it is concluded that a systematic structural modification of the thermal management unit can significantly enhance the cooling performance without increasing the complexity of the system.
Cavitation-induced erosion is a critical challenge in U-shaped notch (USN) hydraulic spool valves. However, the coupled influence of notch depth and valve opening on cavitation morphology remains unclear. In this study, time-averaged cavitation characteristics in USN valve ports is investigated systematically over multiple notch depths and valve openings. A geometric effective flow area analysis is combined with experimentally validated numerical simulations. Specifically, the effective flow area is quantified as a function of valve opening and used to interpret the migration of the controlling throttling section along the notch. Three cavitation flow regimes are examined: shallow-notch, deep-notch and valve-orifice migration flow. Results indicate that cavitation is governed by the interaction between shear layers and low-pressure vortices. At small openings, shallow notches promote outlet-induced low-pressure vortices associated with a wall-attached jet. In contrast, deep notches favour bend-induced separation near the upstream bottom region, leading to stable attached cavitation. With increasing valve opening, the effective throttling section migrates. This migration reorganizes the internal pressure gradient. It suppresses vortex-dominated regions, enhances boundary-layer separation and wall shear, and shifts cavitation inception from the notch bottom toward the upper throttling edge. These findings provide a unified physical interpretation of cavitation regime transitions in USN spool valves and offer guidance for notch design aimed at cavitation control.
Accurate prediction of the speed of sound in electrolyte solutions is important for understanding physicochemical behavior and improving industrial process design. In this study, several advanced machine learning models, including Decision Tree (DT), AdaBoost, Random Forest (RF), K-Nearest Neighbors (KNN), Gradient Boosting (GB), Ensemble Learning (EL), Support Vector Machine (SVM), XGBoost, and CatBoost, were developed and evaluated to predict the speed of sound in electrolyte solutions. Eight physicochemical parameters, namely molar mass, density, cation radius, anion radius, lattice energy, hydration energy, electrical conductivity, and molarity, were used as input variables based on a dataset of 125 experimental observations. Pearson and Spearman correlation analyses showed that lattice energy and molarity had the strongest positive relationships with the speed of sound, whereas hydration energy, conductivity, and density exhibited inverse relationships. Model performance was assessed using the correlation coefficient (R), RMSE, MAE, MBE, DR, and SI. Among all models, Gradient Boosting demonstrated the most balanced predictive performance during the testing stage with R2 = 0.9624, RMSE = 4.4884, and MAE = 2.4064, indicating excellent prediction accuracy and generalization capability. SHAP analysis revealed that lattice energy was the most influential parameter affecting the predicted speed of sound, followed by molarity and anion radius. Overall, the proposed machine learning framework provides an accurate and interpretable data-driven tool for estimating the speed of sound in electrolyte solutions and offers valuable insight into the physicochemical factors governing acoustic behavior in aqueous electrolyte systems.
The Smoothed Particle Hydrodynamics (SPH) method has demonstrated significant advantages in many different engineering problems. However, it suffers from great computational cost and needs a mature and flexible multi-resolution technique similar to those of mesh-based approaches to further improve computational efficiency. A previous SPH framework with anisotropic adaptive spatial resolution (SPH-AASR) has successfully integrated an anisotropic kernel with spatially variable particle spacing. However, it relied on predefined particle spacing and therefore lacked true adaptivity. In this paper, we propose a substantial enhancement to the SPH-AASR framework for free-surface flows. In the new approach, free-surface particles are assigned smaller spacing in the normal direction of the free surface, and larger spacing in the tangential direction. Meanwhile, the particle spacing gradually increases from the free surface towards the fluid interior. Several challenging benchmark cases, including dam breaking, water entry of cylinders, liquid sloshing, and standing wave, are conducted to validate the method. The results show that, when the adaptive number Cr ≤ 1.1 and the anisotropic ratio r ≤ 3, the AASR results closely match the reference solutions. Under these conditions, compared with uniformly distributed particles, the calculation efficiency can increase to 2.0–3.5 times.
Recently, biomimetic porous structures have garnered significant attention due to promising application in thermal management systems. However, flexible design and performance optimization remain challenges to the in-depth investigation of cold plate heat transfer enhancement. Triply Periodic Minimal Surfaces (TPMS), mathematically defined as implicit surfaces with minimal mean curvature, exhibit superior structure interconnectivity and a high surface-area-to-volume ratio, making TPMS structures particularly promising for advanced fluid cooling applications. In cold plate design, the geometric configuration determines the fluid flow direction, which in turn strongly influences the heat transfer pathway. This study advances the design and optimization of TPMS cold plates, with particular attention to Gyroid and Diamond structures. A high-degree-of-freedom modelling framework is developed to generate TPMS variants with different periods and shell thicknesses using a small set of control parameters. The qualitative and quantitative relationships among geometric characteristics are systematically investigated. A material interpolation model is constructed to realize meshless numerical calculations and validated for accuracy. Subsequently, a multi-objective optimization problem is formulated based on the non-gradient NSGA-II algorithm, targeting both the average surface temperature of the heating source (overall thermal performance) and standard deviation of temperature (temperature uniformity). The adaptively optimized designs reveal that employing a small and uniform period along the mainstream direction enhances overall heat transfer performance, whereas a gradually decreasing period contributes to improved temperature uniformity. Besides, a gradually increasing shell thickness benefits both thermal metrics. Compared to the regular TPMS designs, the optimized Gyroid structure achieved a maximum reduction of 7.80 K (2.24%) in mean temperature and 3.78 K (32.98%) in standard deviation of temperature, whereas the optimized Diamond structures yielded respective reductions of 1.86 K (0.55%) and 5.09 K (46.53%). The proposed method effectively reduces the geometrical modelling and numerical analysis costs of flexible design and performance optimization for TPMS structures, and further extends their application potential.
Accurate prediction of hydraulic transients in axial-flow pump stations during start-up, shutdown and flow reversal requires complete four-quadrant pump characteristics. However, such data are rarely available and are often replaced by incomplete or borrowed Suter curves. This study combines large-scale experiments and computational modelling to construct, validate and apply complete four-quadrant performance maps for axial-flow pumps. First, static four-quadrant tests were performed for several model pumps spanning different specific speeds and blade setting angles, yielding head-discharge-torque datasets over pump, braking and turbine regimes. The measurements were converted into periodic, dimensionless Suter curves that can be directly embedded in one-dimensional method-of-characteristics solvers. Second, a CFD-assisted three-dimensional internal-characteristic workflow was developed to sample operating points in Suter space, switch boundary conditions consistently across quadrants, and reconstruct continuous four-quadrant curves; the CFD-derived curves agree well with the experiments. Third, the measured curves were coupled to a prototype pump-station transient model and validated against three-dimensional CFD and field start-up/shutdown measurements. With the authentic ZM25 four-quadrant curve, the one-dimensional model reproduces the start-up transient with an overall feature error of 7.7% and captures the main timing and magnitude trends of shutdown flow reversal. Sensitivity analyses show that curve fidelity strongly controls transient predictions: blade-angle-mismatched curves (+/- 4 degrees), specific-speed-mismatched curves, and single-quadrant curves can introduce large peak reverse-flow errors, non-physical oscillations and wrong backflow durations. In particular, substituting higher- and lower-specific-speed curves leads to peak reverse-flow deviations of approximately 50% and up to 80%, respectively. Completing the missing quadrants with the CFD-derived curve markedly improves numerical robustness and reduces the start-up error by about 12 percentage points relative to an incomplete curve. The resulting dataset and verified workflow provide a practical route for constructing reliable pump curves for transient simulations and a benchmark for validating reduced-order and CFD methods for hydraulic machinery across all four quadrants.
The dihedral angle, an inherent geometric characteristic between the suction surface of the compressor blade and the endwall, significantly influences corner separation. To deeply investigate the impact of the size and variation of the dihedral angle on corner separation, this paper designs physical models of varying dihedral angle diffusers equivalent to compressors. These models leverage the deceleration and pressurisation effects of diffusers on airflow. Using these models, a study is conducted employing large eddy simulation (LES) to investigate the impact of varying dihedral angles on corner separation under adverse pressure gradients. The research results indicate that in varying dihedral angle diffusers, where the dihedral angle decreases axially, the following three strategies can be employed to enhance aerodynamic performance: (1) Keeping the inlet dihedral angle constant while increasing the outlet dihedral angle. (2) Keeping the difference in dihedral angles between the inlet and outlet constant while increasing the inlet dihedral angle. (3) Keeping the outlet dihedral angle constant while increasing the inlet dihedral angle. Among these strategies, the first and second can delay corner separation and promote flow reattachment. While the third strategy can also delay corner separation, it increases the axial variation rate of the dihedral angle, exacerbating corner separation at the end of the expansion section. Additionally, secondary flow in the diffuser develops from the center toward the periphery, forming oppositely rotating vortex pairs near the corner and wall center, with increasing intensity downstream. The highest Reynolds stresses are concentrated in the corner region, where the flow exhibits strong anisotropy. Overall, these findings improve understanding of corner separation flow under varying dihedral angle conditions and provide new insights and data to support the design of modern compressors.
Ultrasonic cleaning systems generate acoustic fields using high-frequency transducer vibrations. While transducer parameters can be tuned, a unified and efficient model for predicting field behavior across diverse cleaning conditions remains unavailable. In this work, we present Ultrasonic-Net, a Physics-Informed Neural Network (PINN) developed to predict ultrasonic fields, specifically targeting cleaning applications. The JAX-Fluids solver is used to generate ground truth for evaluation. The complex system, comprising the ultrasonic transducers (acoustic sources) and the target devices, is modeled using a sharp interface method. The design of Ultrasonic-Net is based on a synthesis of detailed physics and established knowledge. In the proposed architecture, Multi-scale Convolutional Neural Networks (Multi-Scale CNN) and Fast Kolmogorov-Arnold Networks (FastKAN) have been integrated into the Spline-based PINN. The Attention Fusion Layer integrates features from the multi-scale local and global branches by performing adaptive channel-wise recalibration on their concatenated representations. During training, the wave equation is employed as a physical constraint. The vibration frequency, along with the variations in transducers, and the configuration of the devices to be cleaned, is incorporated through initial and boundary condition (IBC) constraints. The decomposition method combined with vectorization is designed to facilitate the enforcement of these constraints. The combination is motivated by ultrasonic cleaning-specific physical requirements. Systematic ablation studies across seven PINN variants and multiple hyperparameter groups confirm that the proposed architecture and training configuration achieve the best accuracy-efficiency trade-off. Once trained, Ultrasonic-Net efficiently predicts the spatiotemporal evolution of acoustic waves under varying conditions, closely matching ground truth while significantly reducing computational cost. An extended experimental comparison is provided to further validate the model's performance. The model's practical utility is demonstrated through application to acoustically driven bubble dynamics. Finally, a systematic evaluation confirms its accuracy and generalization capacity across diverse conditions.
Coastal and offshore protective structures are often loaded by free-surface flows that carry sediment. This study develops a three-phase numerical framework that couples Computational Fluid Dynamics (CFD), the Discrete Element Method (DEM), and the Finite Element Method (FEM) for fluid-particle-structure interaction. The framework combines a semi-resolved CFD-DEM method that is independent of particle shape and mesh size, an Arbitrary Lagrangian-Eulerian (ALE) formulation with Interface Quasi-Newton with Inverse Least Squares (IQN-ILS) and Radial Basis Function (RBF) interpolation for fluid-structure coupling, and an adaptive wall-update algorithm that reduces non-physical force oscillations at the FEM-DEM interface. Non-spherical particle clusters are represented with a bonded multi-sphere model. After benchmark validation, the model is applied to seepage-induced erosion beneath a flexible barrier and sediment-laden dam-break impact on a flexible baffle. Cubic particles produce smoother and lower collision forces because of interlocking, whereas spherical particles generate high-frequency impacts. As structural stiffness increases, the dominant load changes from particle impact to hydrodynamic pressure. The results show that the CFD-DEM-FEM framework can capture particle-shape effects, sediment shielding, and stiffness-dependent loading transitions in sediment-laden free-surface flows.
Solar air heaters provide a well-established means for harnessing solar thermal energy, but their efficiency is fundamentally limited by the low convective heat transfer coefficient of air. To address this limitation, various enhancement techniques, such as artificial surface roughness and jet impingement, have been extensively investigated. Wavy channels have also been explored as an alternative enhancement strategy. However, existing studies have predominantly focused on fully corrugated absorber plates. Configurations in which the flow path alone is wavy while the absorber plate remains flat have received considerably less attention. Moreover, most investigations have focused on sinusoidal waves, while other wave shapes have received considerably less attention. This study addresses these gaps by evaluating a solar air heater configuration comprising a flat absorber plate and a wavy bottom channel. Four periodic wavy profiles, including rectangular, triangular, trapezoidal, and sinusoidal, were systematically evaluated using computational fluid dynamics (CFD) simulations to determine which geometry offers the most favorable thermo-hydraulic performance. The study parametrically analyzed the influence of wave amplitude and frequency for each profile, revealing substantial performance differences among the geometries. The triangular wavy channel emerged as the most effective configuration for heat transfer, achieving Nusselt numbers 2.1 and 3.1 times greater than that of a straight channel at Reynolds numbers of 15,000 and 10,000, respectively. In contrast, sinusoidal channel exhibited the lowest friction factors across all tested conditions, thereby minimizing pumping power requirements. Most notably, both triangular and sinusoidal designs delivered superior overall thermo-hydraulic performance. At a Reynolds number of 15,000, the triangular channel achieved an optimal thermo-hydraulic performance parameter (THPP) of 1.15, while at a Reynolds number of 10,000, the sinusoidal channel attained a peak THPP of 1.73. A THPP greater than unity confirms that the net energy benefit from enhanced heat transfer outweighs the energy penalty of their increased flow resistance.
