Performance Of Turbulence Models Research Articles

CFD simulations of two opposing plane wall jets in a generic empty airplane cabin: Comparison of RANS and LES

Accurate computational fluid dynamics (CFD) simulations of the ventilation flow in airplane cabins are important for ventilation design optimisation regarding passenger health, thermal comfort and energy efficiency. The complex unsteady flow phenomena inside the cabin are challenging in numerical modelling and therefore dedicated validation of the simulation outcomes is required. The majority of airplane ventilation studies performed steady Reynolds-averaged Navier-Stokes (RANS) simulations and focused mainly on global flow field characteristics. This paper presents both steady RANS and transient simulations, including unsteady RANS and large eddy simulations (LES), of the flow driven by opposing plane wall jets in a reduced-scale isothermal water-filled generic airplane cabin. A detailed analysis of the mean velocity and turbulence characteristics (including Reynolds stresses) of the cabin flow is outlined, encompassing a thorough investigation of the fundamental flow components such as the opposing jets and the merged jet. Specific attention is devoted to LES grid sensitivity and the performance of different RANS turbulence and LES subgrid-scale (SGS) models is assessed by comparison with particle image velocimetry (PIV) measurements. It is shown that LES in general performs much better than RANS, the latter being incapable of providing accurate mean flow characteristics within the interaction zone and merged jet due to the underlying dynamic opposing jet interaction. Unsteady RANS partly covers the unsteadiness, however, turbulence levels remain systematically underpredicted and LES is required for a correct representation. Differences between the LES SGS model predictions are limited to the SGS kinetic energy budget of the total turbulent kinetic energy.

Review and Comparison of Numerical Simulations of Secondary Flow in River Confluences

River confluences are a common feature in natural water resources. The flow characteristics in confluences are complicated, especially at junction areas between tributaries and the main river. One of the typical characteristics of confluences is secondary flow, which plays an important role in mixing, velocity, sediment transport, and pollutant dispersion. In addition to the experimental and field studies that have been conducted in this area, the development of computational fluid dynamics has allowed researchers in this field to use different numerical models to simulate turbulence properties in rivers, especially secondary flows. Nowadays, the hydrodynamics of flows in confluences are widely simulated by using three-dimensional models in order to fully capture the flow structures, as the flow characteristics are considered to be turbulent and three-dimensional at river junctions. Several numerical models have been recommended for this purpose, and various turbulence models have been used to simulate the flows at confluences. To assess the accuracy of turbulence models, flows have been predicted by applying different turbulence models in the numerical model and the results have been compared with other data, such as field, laboratory, and experimental data. The purpose behind these investigations was to find the suitable model for each case of turbulent flow and for different types of confluences. In this study, the performances of turbulence models for confluences are reviewed for different numerical simulation strategies.

Turbulence model performance for ventilation components pressure losses

This study looks to find a suitable turbulence model for calculating pressure losses of ventilation components. In building ventilation, the most relevant Reynolds number range is between 3×104 and 6×105, depending on the duct dimensions and airflow rates. Pressure loss coefficients can increase considerably for some components at Reynolds numbers below 2×105. An initial survey of popular turbulence models was conducted for a selected test case of a bend with such a strong Reynolds number dependence. Most of the turbulence models failed in reproducing this dependence and predicted curve progressions that were too flat and only applicable for higher Reynolds numbers. Viscous effects near walls played an important role in the present simulations. In turbulence modelling, near-wall damping functions are used to account for this influence. A model that implements near-wall modelling is the lag elliptic blending k-ε model. This model gave reasonable predictions for pressure loss coefficients at lower Reynolds numbers. Another example is the low Reynolds number k-ε turbulence model of Wilcox (LRN). The modification uses damping functions and was initially developed for simulating profiles such as aircraft wings. It has not been widely used for internal flows such as air duct flows. Based on selected reference cases, the three closure coefficients of the LRN model were adapted in this work to simulate ventilation components. Improved predictions were obtained with new coefficients (LRNM model). This underlined that low Reynolds number effects are relevant in ventilation ductworks and give first insights for suitable turbulence models for this application. Both the lag elliptic blending model and the modified LRNM model predicted the pressure losses relatively well for the test case where the other tested models failed.

An evaluation of different RANS turbulence models for simulating breaking waves past a vertical cylinder

The purpose of the present study is to evaluate the performance of different turbulence models for predicting the interaction between breaking waves and a vertical cylinder based on the volume of fluid (VOF) method. Six different models are investigated in the present study, i.e., no turbulence model, the k−ωSST turbulence model, the buoyancy-modified k−ωSST turbulence model, the stabilized k−ωSST turbulence model, the modified stabilized k−ωSST turbulence model and the realizable k−ε turbulence model. The vertical cylinder is installed at the edge of a 1:10 slope on the bottom of the numerical wave tank. The numerical simulations are conducted by solving the unsteady Reynolds-Averaged Navier-Stokes (RANS) equations using waves2Foam (a solver based on the open-source Computational Fluid Dynamic (CFD) software OpenFOAM). The present numerical results of the surface elevations and the breaking wave forces are compared with published experimental data. The kinetic characteristics beneath the free surface including averaged velocity, turbulent kinetic energy and turbulent kinematic viscosity are also investigated. It is observed that the stabilized k−ωSST turbulence (λ2=0.05, αβs=1.36) and the buoyancy-modified k−ωSST turbulence model (αβs=1.176) effectively reduce the turbulent kinetic energy before wave breaking, but the predicted breaking wave forces on the cylinder are smaller than that of the experimental data. The k−ωSST turbulence model shows good agreement with the experimental data in terms of the free surface elevation and the breaking wave force, but it over-predicts the turbulent kinetic energy. The realizable k−ε turbulence model does not give good predictions of both the free surface elevation and the breaking wave force as compared to the published experimental data.

A Review of Numerical Simulations of Secondary Flows in River Bends

River bends are one of the common elements in most natural rivers, and secondary flow is one of the most important flow features in the bends. The secondary flow is perpendicular to the main flow and has a helical path moving towards the outer bank at the upper part of the river cross-section, and towards the inner bank at the lower part of the river cross-section. The secondary flow causes a redistribution in the main flow. Accordingly, this redistribution and sediment transport by the secondary flow may lead to the formation of a typical pattern of river bend profile. It is important to study and understand the flow pattern in order to predict the profile and the position of the bend in the river. However, there are a lack of comprehensive reviews on the advances in numerical modeling of bend secondary flow in the literature. Therefore, this study comprehensively reviews the fundamentals of secondary flow, the governing equations and boundary conditions for numerical simulations, and previous numerical studies on river bend flows. Most importantly, it reviews various numerical simulation strategies and performance of various turbulence models in simulating the flow in river bends and concludes that the main problem is finding the appropriate model for each case of turbulent flow. The present review summarizes the recent advances in numerical modeling of secondary flow and points out the key challenges, which can provide useful information for future studies.

Evaluation of turbulence models for estimating the wake region of artificial reefs using particle image velocimetry and computational fluid dynamics

Many studies have evaluated the performance of turbulence models used for computational fluid dynamics (CFD) analysis of artificial reefs (ARs), but an optimal model remains elusive, particularly in terms of wake length and areal estimation. Only a few models were used in previous studies and no report has yet investigated wake shape. We compared five turbulence models and verified the CFD results using particle image velocimetry (PIV). A standard k-ε model, renormalization group k-ε model, k-ω model, shear stress transportation k-ω model, and Reynolds stress model (RSM) were used. A down-scaled half-ball-type AR was devised and CFD and PIV analyses were performed. Three different inlet velocities (three Reynolds numbers, Re) were considered in each model, and the CFD and PIV results were compared. Wake lengths obtained in the PIV experiments were 1.05 L (Re = 2632), 0.90 L (Re = 5655) and 0.85 L (Re = 8782), respectively. The RSM well-reflected this, especially when Re = 2632 (difference +1%) and Re = 5655 (difference −2%). PIV revealed that all wake regions had unique shapes, reflecting flow divergence (local upwelling) from the end of the wake. Considering such divergent flow, the RSM optimally predicted the overall characteristics of the wake region.

Dispersion Modeling of Particulate Matter in Different Size Ranges Releasing from a Biosolids Applied Agricultural Field Using Computational Fluid Dynamics

This paper proposes a methodology using computational fluid dynamics (CFD)-FLUENT to simulate the dispersion of particulate matter releasing from a biosolid applied agricultural field and predict the particulate concentrations for different ranges of particle sizes. The discrete phase model (Lagrangian-Eulerian approach) was used in combination with each of the four turbulence models: Standard kε (kε), Realizable kε (Rkε), Standard kω (kω), and Shear-stress transport k-ω (SST) to predict particulate matter size concentrations for distances downwind of the agricultural field. In this modeling approach, particulates were simulated as discrete phase and air as continuous phase. The predicted particulate matter concentrations were compared statistically with their corresponding field study observations to evaluate the performance of turbulence models. The statistical analysis concluded that among four turbulence models, the discrete phase model when used with Rkε performed the best in predicting particulate matter concentrations for low (u u u > 5 m/s) wind speeds, Rkε, kω, and SST showed similar performances. The discrete phase model using Rkε performed very well and modeled the best concentrations for the particle sizes (μm): 0.23, 0.3, 0.4, 0.5, 0.65, 0.8, 1, 1.6, 2, 3, 4, and 5. For particle sizes: 7.5 and 10, the performances of Rkε, kε, kω, and SST were similar.

Quality control of computational fluid dynamics (CFD) model of ozone reaction with human surface: Effects of mesh size and turbulence model

Indoor chemical reactions with human surfaces can meaningfully affect indoor pollutant dynamics and human exposure to reactants and products. Computational Fluid Dynamics models have been used to simulate ozone-human surface reactions in indoor environments. However, the model uncertainties due to near-human grid size and turbulence model are not well understood. The objective of this study is to assess performance of turbulence models and near-human surface grids in predicting local airflow and ozone transport near the human surface under a representative range of indoor airflow conditions. The results show that Large Eddy Simulation (LES) with a fine near-human grid size (e.g., y+ = 1) can yield reasonable results of airflow and mass transfer near the body. The SST k–ω model performs well for the simulation of near-human airspeed, whereas the standard k–ε model can predict lower airspeeds above the human head than the measurements. Both standard k–ε and SST k–ω models may underestimate the turbulent kinetic energy of the thermal plume and ozone deposition velocity at the human surface. The results also suggest that near-human grid sizes of y+ >10 can lead to unreasonable estimates of airflow and mass transfer rate near the human surface.

URANS simulations of a horizontal axis wind turbine under stall condition using Reynolds stress turbulence models

Over the last decade, a dramatic increase in the size of commercial wind turbines is noticeable. The optimal structural design and reliable fatigue-life prediction of these large-scale structures depend on the accurate modeling of the turbulent flow around the rotors. The present paper aims at extending the knowledge of the turbulent flow characteristics around horizontal axis wind turbines and assessing the performance of several URANS turbulence models, principally Reynolds stress turbulence (RST) models, in predicting the wind turbine aerodynamics under stall condition. The simulations are performed using the NREL phase VI wind turbine over a wide range of wind speeds. Three RST models and the linear and quadratic variants of the k−ω SST turbulence model are examined. The results demonstrate that the RST models generally provide more reliable predictions. An evaluation of the Boussinesq hypothesis questions the applicability of the turbulence models employing this hypothesis to the wind turbine aerodynamic problems. Finally, the elliptic blending RST model, which is one of the latest formulations of the RST models, appears to provide the best trade-off between accuracy and computational cost.

Performances of Different Turbulence Models for Simulating Shallow Water Sloshing in Rectangular Tank

Liquid sloshing is a common phenomenon in the transportation of liquid-cargo tanks. Liquid waves lead to fluctuating forces on the tank walls. If these fluctuations are not predicted or controlled, for example, by using baffles, they can lead to large forces and momentums. The volume of fluid (VOF) two-phase numerical model in OpenFOAM open-source software has been widely used to model the liquid sloshing. However, a big challenge for modeling the sloshing phenomenon is selecting a suitable turbulence model. Therefore, in the present study, different turbulence models were studied to determine their sloshing phenomenon prediction accuracies. The predictions of these models were validated using experimental data. The turbulence models were ranked by their mean error in predicting the free surface behaviors. The renormalization group (RNG) k–e and the standard k–ω models were found to be the best and worst turbulence models for modeling the sloshing phenomena, respectively; moreover, the SST k–ω model and v2-f k-e results were very close to the RNG k–e model result.

Performance of turbulence models for single phase and liquid-solid slurry flows in pressurized pipe systems

In hydraulic engineering, turbulent flow is the most encountered flow regime in closed conduits. Because of the difficulty and complexity of the turbulence mechanism dominated by fluctuations, it is very important to model the turbulence mechanism correctly. Numerous models are developed by researchers in order to characterize turbulence. However, it is crucial to choose the model that gives the correct results. In this study, the performance of some well-known turbulence models is comparatively investigated. Extensive computational fluid dynamics (CFD) simulations and experimental analysis of water flow in the rough pipe, eccentric and concentric annulus, and liquid-solid slurry flow in the eccentric annulus are presented. The accuracy of turbulence models is verified with the experiments that have been conducted and presented data by Sorgun, 2010). Reasonable agreements are obtained between CFD simulations and experiments. Results show that the RNG κ-ϵ gives better results than the other turbulence models for the water flow through the rough pipe and eccentric annulus. For the water flow through the concentric annulus, the EARSM model outperforms the other turbulence models. All models except SSG showed similar performances in the estimation of the pressure difference for slurry flow.

New Prediction Models for Estimation of Aerodynamic Pressure Loss due to a Train of Mine Cars in an Underground Airway

Pressure losses in mine ventilation systems are estimated using friction factors and shock loss factors. The efficient planning and designing of mine ventilation systems depend on the accuracy of the friction factors as well as the shock loss factors that are taken into consideration. While there is a much more acceptable knowledge of what constitutes friction loss in mine roadways, the shock loss factors are in general readily not available for all possible contexts of flow separation phenomena in mines. One such special case of shock loss in mines is the train of mine cars in the airway. Mine cars are widely used for the transportation of material, and the information on shock loss factor for mine cars is limited. This paper examines the pressure loss occurring due to a train of mine cars in an airway using scale model studies and numerical investigations. Computational fluid dynamic (CFD) studies are conducted using four different turbulence models. The performance of different turbulence models is analyzed by comparing it with the experimental results. The study suggests that the boundary layer resolution method is significant in the simulation of shock loss. From the study, it is found that the shock loss factors from presently available model overestimate the pressure loss by 7 to 15 times. A better prediction model for the shock loss factor of a train of mine cars in the airway is proposed using the experimental results.

Conjugate Heat Transfer Predictions on Combined Impingement and Film Cooling of a Blade Leading Edge Model

In this paper, numerical simulation is performed to predict combined impingement and film cooling on a model of the turbine blade leading edge by the heat flow coupling method. The first-stage rotor blade leading edge of the GE-E3 engine high-pressure turbine is adopted for the simulation. The relative performances of turbulence models are compared with experimental data and the results show that standard k-ω model is the best, based on simulation accuracy. The standard k-ω model is adopted for the simulation. A grid independence study is also carried out. Five different mass flow ratios and seven different film cooling hole configurations are studied in detail. The results indicate that (1) the overall film cooling effectiveness on the leading edge surface and the Nusselt number on the target surface increase with increases of coolant mass flow ratio; (2) the blade leading edge temperature decreases with an increase in mass flow ratio; and (3) the area-averaged overall film cooling effectiveness increases with a decrease in spanwise film cooling injection angle, when the injection angle is lower than 25°, while it does not change much otherwise.

Numerical simulation of flow over NACA 0015 airfoil with different turbulence models

Airfoils in various types are widely used in many devices subjected to fluid flows such as aircrafts, vehicles, turbines etc. Therefore, analyzing the fluid flow around an airfoil is one of the important subjects in fluid mechanics. In this study, the conservation equations of two dimensional compressible flow over standard airfoils were solved by using different numerical techniques. After a mesh independence study, applied mathematical model, numerical techniques and obtained results are confirmed with experimental results given in literature. Three different turbulence models, namely the k-w spalarat almaras and the reynolds stress models were used in the solutions. The performances of turbulence models were evaluated under the results obtained. The verified numerical model was also applied to the flow over different types of blades, including a special airfoil design. Velocity and pressure fields obtained around these airfoils were analyzed, and their aerodynamic performances in terms of the lift and drag coefficients were compared to each other at different angles of attack.

Application of flux vector splitting methods with SST turbulence model to wall-bounded flows

An explicit 3D Reynolds Averaged Navier-Stokes solver SURFS3D has been developed and run on a structured grid. A 4-stage Runge-Kutta (RK) method is used for the time integration and OpenMP is used to parallelize the solver. This is used as a platform to compare Van Leer’s Flux Vector Splitting Method (VLFVSM) to a variant of Liou’s Advection Upstream Splitting with velocity and pressure diffusion (AUSM+-up2). The code results are validated for wall bounded flows. Menter’s k−ω Shear Stress Transport (SST) turbulence model is employed and validated for turbulent flows. The SST turbulence model is selected due to its capability to predict the shock location and separation location in adverse pressure gradient accurately. The validation cases are laminar and turbulent incompressible flows and turbulent supersonic flow over flat plate, transonic flow over an axisymmetric bump, and supersonic flow over a blunt bi-conic configuration. It is observed that VLFVSM+SST lacks accurate viscous prediction capability for the wall-bounded flows at low speeds compared to AUSM +-up2+SST, which works well at low speed. In transonic flow conditions, both schemes perform well. AUSM +-up2 scheme exhibits mild carbuncle problem for Mach 3.0 flow over a blunt biconic configuration, whereas, VLFVSM is free from this problem. Overall, the SST turbulence model performance is good for all the validation cases studied.

Numerical Investigation on Heat Transfer of Supercritical Water With a Variable Turbulent Prandtl Number Model

Abstract Turbulent Prandtl number (Prt) has a great impact on the performance of turbulence models in predicting heat transfer of supercritical fluids. Unrealistic treatment of Prt may lead to large deviations of the prediction results from experimental data under supercritical conditions. In this study, the effect of Prt on heat transfer of supercritical water was extensively studied by using shear stress transport (SST) k–ω turbulence model, and the results suggested that using the existing Prt models would lead to failures in predicting the heat transfer characteristics of supercritical water under deteriorated heat transfer (dht) conditions. A new variable Prt model was proposed with the Prt varied with pressure, turbulent viscosity ratio, and molecular Prandtl number. The new model was validated by comparing the numerical results with the corresponding experimental data, and it was found that the new variable Prt model exhibited better performance on reproducing the dht of supercritical water in vertical tubes than those of the existing Prt models.

Turbulence Quantification in Supercritical Nitrogen Injection: An Analysis of Turbulence Models

In Liquid Rocket Engines, higher combustion efficiencies come at the cost of the propellants exceeding their critical point conditions and entering the supercritical domain. The term fluid is used because, under these conditions, there is no longer a clear distinction between a liquid and a gas phase. The non-conventional behavior of thermophysical properties makes the modeling of supercritical fluid flows a most challenging task. In the present work, a RANS computational method following an incompressible but variable density approach is devised on which the performance of several turbulence models is compared in conjunction with a high accuracy multi-parameter equation of state. Also, a suitable methodology to describe transport properties accounting for dense fluid corrections is applied. The results are validated against experimental data, becoming clear that there is no trend between turbulence model complexity and the quality of the produced results. For several instances, one- and two- equation turbulence models produce similar and better results than those of Large Eddy Simulation (LES). Finally, considerations about the applicability of the tested turbulence models in supercritical simulations are given based on the results and the structural nature of each model.
 Keywords: Supercritial fluids, RANS turbulence modeling, Liquid rocket engines

Performance of turbulence models for flow prediction in a mould of continuous steel caster

The flow of molten steel in a mould during continuous casting process is nowadays most often studied using numerical approach. With such an approach, appropriate numerical tools have to be selected to achieve physically sound results. Within this work we focus on performance of different turbulence models for the case of casting of steel billets with single SEN port oriented in vertical, downward direction. Only fluid dynamics aspect of the casting process is considered, while all other phenomena are disregarded at this stage. Simulations were performed with commercial CFD package ANSYS Fluent with k-ε, k-ω and SAS turbulent models. The geometric model, material properties and process conditions are closely matched to those from our experimental water model studies, which is typically not done. Hence, the difference between the experiments and the numerical simulation originates only due to turbulence model performance. The experimental PIV data from the water model experimental system is used to benchmark the performance of various turbulence models. The models are compared based on lateral dispersion of the liquid jet exiting submerged entry nozzle (SEN) and formation of recirculation zones around the SEN. The results show that while all three models are capable of capturing general flow features, only the SAS model predicts finer turbulent structures and reasonably captures the flow behaviour in temporal domain.

Evaluating the Impact of Turbulence Closure Models on Solute Transport Simulations in Meandering Open Channels

River meanders form complex 3D flow patterns, including secondary flows and flow separation. In particular, the flow separation traps solutes and delays their transport via storage effects associated with recirculating flows. The simulation of the separated flows highly relies in the performance of turbulence models. Thus, these closure schemes can control dispersion behaviors simulated in rivers. This study performs 3D simulations to quantify the impact of the turbulence models on solute transport simulations in channels under different sinuosity conditions. The 3D Reynolds-averaged Navier-Stokes equations coupled with the k − ε , k − ω and SST k − ω models are adopted for flow simulations. The 3D Lagrangian particle-tracking model simulates solute transport. An increase in sinuosity causes strong transverse gradients of mean velocity, thereby driving the onset of the separated flow recirculation along the outer bank. Here, the onset and extent of the flow separation are strongly influenced by the turbulence models. The k − ε model fails to reproduce the flow separation or underestimates its size. As a result, the k − ε model yields residence times shorter than those of other models. In contrast, the SST k − ω model exhibits a strong tailing of breakthrough curves by generating more pronounced flow separation.

Turbulence Modeling Insights into Supercritical Nitrogen Mixing Layers

In Liquid Rocket Engines, higher combustion efficiencies come at the cost of the propellants exceeding their critical point conditions and entering the supercritical domain. The term fluid is used because, under these conditions, there is no longer a clear distinction between a liquid and a gas phase. The non-conventional behavior of thermophysical properties makes the modeling of supercritical fluid flows a most challenging task. In the present work, a Reynolds Averaged Navier Stokes (RANS) computational method following an incompressible but variable density approach is devised on which the performance of several turbulence models is compared in conjunction with a high accuracy multi-parameter equation of state. In addition, a suitable methodology to describe transport properties accounting for dense fluid corrections is applied. The results are validated against experimental data, making it clear that there is no trend between turbulence model complexity and the quality of the produced results. For several instances, one- and two-equation turbulence models produce similar results. Finally, considerations about the applicability of the tested turbulence models in supercritical simulations are given based on the results and the structural nature of each model.