State Computational Fluid Dynamics Simulations Research Articles

End-to-end wind turbine wake modelling with deep graph representation learning

Wind turbine wake modelling is of crucial importance to accurate resource assessment, to layout optimisation, and to the operational control of wind farms. This work proposes a surrogate model for the representation of wind turbine wakes based on a state-of-the-art graph representation learning method termed a graph neural network. The proposed end-to-end deep learning model operates directly on unstructured meshes and has been validated against high-fidelity data, demonstrating its ability to rapidly make accurate 3D flow field predictions for various inlet conditions and turbine yaw angles. The specific graph neural network model employed here is shown to generalise well to unseen data and is less sensitive to over-smoothing compared to common graph neural networks. A case study based upon a real world wind farm further demonstrates the capability of the proposed approach to predict farm scale power generation. Moreover, the proposed graph neural network framework is flexible and highly generic and as formulated here can be applied to any steady state computational fluid dynamics simulations on unstructured meshes.

Computational fluid dynamics of sulfur dioxide and carbon dioxide capture using mixed feeding of calcium carbonate/calcium oxide in an industrial scale circulating fluidized bed boiler

The sulfur dioxide (SO2) and carbon dioxide (CO2) emissions from fuel combustion in a coal-fired power plant constitute a significant source of damage to the environment. Therefore, SO2 and CO2 should be captured before being released into the atmosphere. However, the competitiveness between SO2 and CO2 for calcium carbonate (CaCO3)/calcium oxide (CaO) solid sorbents is still unclear. In this study, unsteady state computational fluid dynamics simulation in a riser of an industrial scale circulating fluidized bed boiler integrated with heterogeneous combustion, carbonation, calcination, and desulfurization reactions using a mixed feeding of CaCO3/CaO solid sorbents was developed in a two-dimensional model to investigate the competition between SO2 and CO2 capture. Then, the effect of three operating variables, the mixed solid CaCO3/CaO sorbent particle size, feed position, and the proportion of inlet fuel velocity on the SO2 and CO2 capture were evaluated using a 23 factorial experimental design. The CaCO3/CaO solid sorbent particle size had a significant effect on the SO2 capture, while the interaction between CaCO3/CaO solid sorbent particle size and feed position had a significant effect on the CO2 capture. The reaction rate for CO2 capture was higher than that for SO2 capture. For SO2 capture, CaO reacted with SO2 faster than CaCO3 while, for CO2 capture, solid sorbents had higher carbonation rate than calcination rate. In addition, the overall level of SO2 and CO2 capture with a mixed CaCO3/CaO solid sorbent feed was higher than those with the conventional CaO solid sorbent.

Computer Analysis of a Methane Fired Crucible Furnace

Two design factors and one operation parameter of a methane fired crucible furnace are numerically explored in this work. These are the number of burners, the location of the exhaust gas exit, and the air-fuel ratio, respectively. Three dimensional steady state Computational Fluid Dynamics simulations are carried out in order to analyze the influence of the above factors on the mean cavity temperature in absence of thermal load, the methane content and the oxygen content of the exit gas.

Generation and validation of virtual nonwoven, foam and knitted filter (separator/coalescer) geometries for CFD simulations

Abstract Pivotal to efficient optimization of filter media using highly resolved pore-scale computational fluid dynamics (CFD), is the ability to generate virtual filter geometries with controllable geometric properties including packing density, fibre (or element) diameters, morphology and pore sizes. Yet, current design frameworks rely heavily on scans of available filters for representation of the pore-scale geometry in CFD simulations. This study demonstrates novel methodologies for fully customizable generation of realistic virtual nonwoven, open-cell foam and knitted filter geometries, using open-source tools in Gmsh, Blender, Python, OpenFOAM libraries and Fortran scripts. Further, a methodology for the generation and validation of a computational mesh for pore-scale CFD simulations, using open-source tools within OpenFOAM is described. The structure of the virtual filter media generated using the present techniques are validated by comparison against electron-microscopy (SEM) scans and photographic images of real media, as well as comparison of the mathematical structure and pore-scale statistics against the literature. Subsequently, steady state CFD simulations using OpenFOAM are performed to compare the predicted pressure drops across the virtual filters against both, measurements carried out for this study with similar filters, as well as theoretical models in the literature. It is found that there is good and consistent qualitative and quantitative agreement between the structures of the virtual media and real media, as well as between predicted and measured or theoretical pressure drops. The agreement found from the rigorous validations carried out in this study establish the suitability of the proposed methodology for efficient and cost-effective development and optimization of filter media.

Virtual flow-diverter treatment planning: The effect of device placement on bifurcation aneurysm haemodynamics.

Bifurcation aneurysms account for a large fraction of cerebral aneurysms and often present morphologies that render traditional endovascular treatments, such as coiling, challenging and problematic. Flow-diverter stents offer a potentially elegant treatment option for such aneurysms, but clinical use of these devices remains controversial. Specifically, the deployment of a flow-diverter device in a bifurcation entails jailing one or more potentially vital vessels with a low-porosity mesh designed to restrict the flow. When multiple device placement configurations exist, the most appropriate clinical decision becomes increasingly opaque. In this study, three bifurcation aneurysm geometries were virtually treated by flow-diverter device. Each aneurysm was selected to offer two possible device deployment positions. Flow-diverters similar to commercially available designs were deployed with a fast-deployment algorithm before transient and steady state computational fluid dynamics simulations were performed. Reductions in aneurysm inflow, mean wall shear stress and maximum wall shear stress, all factors often linked with aneurysm treatment outcome, were compared for different device configurations in each aneurysm. In each of the three aneurysms modelled, a particular preferential device placement was shown to offer superior performance with the greatest reduction in the flow metrics considered. In all the three aneurysm geometries, substantial variations in inflow reduction (up to 25.3%), mean wall shear stress reduction (up to 14.6%) and maximum wall shear stress reduction (up to 12.1%) were seen, which were all attributed to device placement alone. Optimal device placement was found to be non-trivial and highly aneurysm specific; in only one-third of the simulated geometries, the best overall performance was achieved by deploying a device in the daughter vessel with the highest flow rate. Good correspondence was seen between transient results and steady state computations that offered a significant reduction in simulation run time. If accurate steady state computations are combined with the fast-deployment algorithm used, the modest run time and corresponding hardware make a virtual treatment pipeline in the clinical setting a meaningful possibility.

The consideration of atmospheric stability within wind farm AEP calculations

The annual energy production of an existing wind farm including thermal stratification is calculated with two different methods and compared to the average of three years of SCADA data. The first method is based on steady state computational fluid dynamics simulations and the assumption of Reynolds-similarity at hub height. The second method is a wake modelling calculation, where a new stratification transformation model was imposed on the Jensen an Ainslie wake models. The inflow states for both approaches were obtained from one year WRF simulation data of the site. Although all models underestimate the mean wind speed and wake effects, the results from the phenomenological wake transformation are compatible with high-fidelity simulation results.

Modeling heat transfer of the electrothermal reactor for magnesium production

Thermal analysis of high temperature electrothermal reactor for magnesium production was carried out in full 3D configuration using computational fluid dynamics (CFD). As the temperature inside the reactor reaches to 1200–1500 K, it becomes difficult to understand and control the heat transfer inside the reactor as the thermocouples were embedded in the insulating layers, much away from the core. The present analysis provides a useful tool to correlate the core temperature with experimental thermocouple readings. The transient state CFD simulation is carried out for the actual pilot scale design of the reactor, considering all the modes of heat transfer-conduction, convection and radiation and actual temperature dependent physical properties of the insulating materials. The heat flux and the spatial temperature profile of the various interfaces of insulation layers were also quantified.

Quasi-Steady-State scheme and application on prewhirl flow and heat transfer in aeroengine

For solving three-dimensional complex unsteady flow to predict aerodynamic characteristics of turbine disc cavity in aeroengine, a quasi-steady-state scheme is proposed. To validate and get reliable turbulence model, three-dimensional numerical studies, of turbine disc cavity prewhirl flow and heat transfer, are conducted. For comparison, unsteady state computational fluid dynamics simulations of prewhirl flow are also performed. Five turbulence models, including standard k-a model, RNG k-a model, Realizable k-a model, SST k-u model and RSM model, are applied both quasi-steady-state scheme and unsteady CFD. Then, reliable quasi-steady-state scheme is obtained and the feasibility, instead of unsteady-state CFD simulation for reducing time resource, is verified. Furthermore, investigations of prewhirl flow and heat transfer in turbine disc cavity of aeroengine are conducted by the proposed quasi-steady-state scheme. It was found that the rotation Reynolds number should be increased to strengthen the blade cooling and weaken the heat transfer in turbine disc cavity. All the results are applied in the design of aeroengine. Also, it proves that quasi-steady-state scheme is effective in the prediction of aerodynamic characteristics in aeroengine.

Experimental and Numerical Study of the Aerodynamic Characteristics of an Archimedes Spiral Wind Turbine Blade

A new type of horizontal axis wind turbine adopting the Archimedes spiral blade is introduced for urban-use. Based on the angular momentum conservation law, the design formula for the blade was derived using a variety of shape factors. The aerodynamic characteristics and performance of the designed Archimedes wind turbine were examined using computational fluid dynamics (CFD) simulations. The CFD simulations showed that the new type of wind turbine produced a power coefficient (Cp) of approximately 0.25, which is relatively high compared to other types of urban-usage wind turbines. To validate the CFD results, experimental studies were carried out using a scaled-down model. The instantaneous velocity fields were measured using the two-dimensional particle image velocimetry (PIV) method in the near field of the blade. The PIV measurements revealed the presence of dominant vortical structures downstream the hub and near the blade tip. The interaction between the wake flow at the rotor downstream and the induced velocity due to the tip vortices were strongly affected by the wind speed and resulting rotational speed of the blade. The mean velocity profiles were compared with those predicted by the steady state and unsteady state CFD simulations. The unsteady CFD simulation agreed better with those of the PIV experiments than the steady state CFD simulations.

Experimental study and three-dimensional (3D) computational fluid dynamics (CFD) analysis on the effect of the convergence ratio, pressure inlet and number of nozzle intake on vortex tube performance–Validation and CFD optimization

Energy separation procedure of vortex tube can be improved by using convergent nozzle. In the experimental investigation, the parameters are focused on the convergence ratio of nozzle, inlet pressure and number of nozzle intakes. The effect of the convergence ratio of nozzle is investigated in the range of 1–2.85. The most objective of this investigation is the demonstration of the successful use of computational fluid dynamics (CFD) in order to develop a design tool that can be utilized with confidence over a range of operating conditions and geometries, thereby providing a powerful tool that can be employed to optimize vortex tube design as well as assess its utility in the field of new applications and industries. A computational fluid dynamics model was developed to predict the performances of the vortex tube system. The numerical investigation was carried out by full three-dimensional (3D) steady state CFD simulation using FLUENT 6.3.26. This model utilizes the k–ɛ turbulence model to solve the flow equations. Experiments were also conducted to validate results obtained for the simulation. First purpose of numerical study in this case was validation with experimental data to confirm these results and the second was the optimization of experimental model to achieve the highest performance.

Aerodynamic Design and Numerical Simulation Analysis of a Passenger Car’s Defrosting Duct

Aerodynamic design of airflow duct has become an important issue of the automobile Heating, Ventilation and Air Conditioning (HVAC) system. In this research, the HVAC defrosting airflow duct of a passenger car was designed using Computational Fluid Dynamics (CFD) method. The HVAC defrosting outlets’ sizes and positions were considered as design variables. The mass flow rate distribution and the cabin flow structure were considered as design targets. The steady Reynolds Averaged Navier-Stokes (RANS) method was adopted because it is an efficient way to help select possible good designs in the early stage of new product development without time consuming and huge computational cost of the unsteady methods. Various duct design configurations were evaluated by analyzing the defrosting mass flow distribution of each flow outlet and by visualizing the flow structure near the windshield and the front left side window. The CFD results showed that the total area of the outlets near the rain wipers was a decisive parameter for mass flow distribution in this duct design. The defrosting flow structure near side windows were difficult to be improved only by enlarging the area of the outlet. The effective flow structure was realized by choosing proper angles of the vanes skirt of the outlets to defrost the windshield region. The overall performance of HVAC system, such as defrosting time and cabin temperature distribution, could not only be predicted based on the results of the steady RANS method. It was shown that the important parameters including the mass flow distribution of each outlets and the flow structure near the windshield and side windows could be quickly evaluated from the steady state CFD simulation results in the early design stage.

Combined wind tunnel and CFD analysis for indoor airflow prediction of wind-driven cross ventilation

This study presents a method for predicting wind-induced cross ventilation using wind tunnel testing coupled with Computational Fluid Dynamics (CFD). First, in a boundary layer wind tunnel, facade pressures near openings of a test house were recorded for multiple incident wind angles using a 1:25 scale building model. The wind tunnel pressure data was then converted into appropriate boundary conditions for an indoor CFD airflow model of the test house. Both steady state and transient CFD simulations were conducted to capture averaged flow and instantaneous flow rates of the cross ventilation flow. The CFD simulation result was compared to the full scale experimental data from a previous study. The steady state result shows that it is possible for this combined wind tunnel-CFD method to predict the averaged cross ventilation through small openings adequately. Similarly, the transient result shows the fluctuation of the flow at the openings could be predicted at a frequency as high as 0.1 Hz. As many buildings design processes now include wind tunnel testing as a part of structural analysis, this proposed method could utilize the wind tunnel pressure data for assessments of a building's cross ventilation potential.

Effects of Particle Blockage and Eccentricity in Location on the Non-Newtonian Fluid Hydrodynamics around a Sphere

The flow of an incompressible non-Newtonian fluid over an eccentrically located sphere confined in a circular tube was investigated using three-dimensional steady state Computational Fluid Dynamics simulations. Pseudoplastic fluids with different power-law indices of 0.57, 0.76, and 0.94 were considered. The effects of ratio of particle diameter to tube diameter (blockage ratio, BR) and ratio of offset of the sphere position from the tube axis to the tube radius (eccentricity, Ec) on the hydrodynamic phenomena around the sphere are reported over a range of particle Reynolds numbers (Rep). The simulations were carried out in the range (0.1 ≤ Rep ≤ 40, 0.01 ≤ BR ≤ 0.5, and 0.0 ≤ Ec ≤ 0.6). The drag coefficient predictions for an unconfined sphere were found to be sensitive to the value of the consistency index parameter (K) in the viscous flow regime, especially at lower n values. At lower particle Reynolds numbers and centrally located sphere, the enhancement in drag coefficient due to blockage ratio was felt least by the fluid with the lowest n value. Even at higher Rep, higher blockage ratios still could cause significant enhancements in the drag coefficients (∼20%) for centrally located spheres. Irrespective of the power-law index, eccentric location of the sphere caused a decline in the overall drag coefficient due to the dominant influence of the lower hemisphere which was closer to the tube wall. At the highest particle Reynolds numbers, eccentricity, and blockage ratio, asymmetric fluid flow distribution caused opposing effects by decreasing the viscous drag coefficient and increasing the form drag coefficient relative to those obtained with the centrally confined sphere. Sharp variations in the shear rate dependent viscosity at the sphere surface could be associated with boundary layer separation. Wall effects at higher blockage ratio suppressed the boundary layer separation in the case of the fluid with the highest power-law index. Eccentricity also caused accelerated boundary layer separation at the upper hemisphere and absence of boundary layer separation along the lower hemisphere. At lower blockage ratios a maximum in the lift coefficient versus particle Reynolds number was usually observed. This study will be relevant in applications such as aseptic food processing, petroleum well drilling, fluidization, and particle transport in microfluidic devices.

Compartmental Models for Continuous Flow Reactors Derived from CFD Simulations

Reactor modeling is of major interest in environmental technology. In this context, new contaminants with higher degradation requirements increase the importance of reactor hydraulics. CFD (Computational Fluid Dynamics) may meet this challenge but is expensive for everyday use. In this paper, we provide research and practice with a methodology designed to automatically reduce the complexity of such a high-dimensional flow model to a compartmental model. The derivation is based on the concentration field of a reacting species which is included in the steady state CFD simulation. While still capturing the most important flow features, the compartmental model is fast, easy to use, and open for process modeling with yet unknown compounds. The inherent overestimation of diffusion by compartmental models has been corrected by locally adjusting turbulent fluxes. We successfully applied the methodology to the ozonation process and experimentally verified it with tracer experiments. The loss of information was quantified as a deviation from CFD performance prediction for different reactions. With increasing discretisation of the compartmental model, these deviations diminish. General advice on the necessary discretisation is given.