Large Eddy Simulation Of Flow Research Articles

Application of Different Sub-Grid Models Used in Large-Eddy Simulation of Flow Around a Cylinder at Re = 3900

Application of Different Sub-Grid Models Used in Large-Eddy Simulation of Flow Around a Cylinder at Re = 3900

Large eddy simulation of flow through an axisymmetric sudden expansion

This study aims at investigating the inlet flow conditions of flow through an axisymmetric sudden expansion with an expansion ratio of 2.0. A series of large eddy simulations with the WALE model were conducted for different inlet Reynolds numbers (Re) and turbulence intensities (urms/U¯m). The reattachment length, defined as the length measured downstream of the expansion where the flow direction is reversed adjacent to the wall (Lr), was measured for each case. For widely studied inlet turbulence intensity values (TI), the simulation results are in good agreement with the experimental and numerical results reported in the literature. Parametric studies revealed that turbulence intensity affects the critical Reynolds number, marking the transition between the laminar and transition regions and the reattachment length. The critical Reynolds number was found to decrease with increasing turbulence intensity. A correlation expression is proposed. Additional analysis with proper orthogonal decomposition was performed to enhance the understanding of complex flow structures downstream of the expansion. Finally, an overall correlation expression for the reattachment length was obtained for 500 ≤ Re ≤ 15 000 and 0.2 ≤ TI (%) ≤ 20. For a given turbulence intensity, the reattachment length can be expressed for laminar and turbulent regions as a function of the Reynolds number. The reattachment length in the transition region can be expressed as a fractional average of reattachment lengths for laminar and turbulent flows.

The spray distribution of a liquid jet in supersonic crossflow in the near-wall region

The gas–liquid mixing process of a liquid jet in supersonic crossflow with a gas–liquid momentum ratio of 7.67 in the near-wall region is investigated numerically. The two-phase flow large eddy simulation is based on the Eulerian–Lagrangian approach and considers the droplet–wall interaction. The results indicate the penetration depth and the lateral extension width, which are in good agreement with the experimental data. The Ω shape, especially the spray foot structure of spray in the cross-sectional plane, is captured well. The transport process of spray toward the wall and the formation of spray foot are systematically studied. Under the influence of the upper CVP (counter-rotating vortex pair), partial droplets in the center region of the spray are transported to the near-wall region and move toward both sides when encountering the wall CVP. Under the current gas–liquid momentum ratio, droplets collide with the wall mainly in the central region at the bottom, which will produce splashed droplets. Affected by the horseshoe vortex, the instantaneous distribution of droplets on both sides near the wall shows stripes shape. The spray foot structure forms the shape that is narrow on the top and wide on the bottom and is mainly formed by splashed droplets. Some splashed droplets in the low-speed boundary layer constitute the lower half of the spray foot; meanwhile, some splashed droplets enter mainstream and constitute the upper half of the spray foot. Moreover, the spray is mainly distributed in the core region, and the spray concentration is very sparse in the spray foot region.

A Coupled LES-Synthetic Turbulence Method for Jet Noise Prediction

Large-eddy simulation (LES)-based jet noise predictions do not resolve the entire broadband noise spectra, often under-predicting high frequencies that correspond to un-resolved small-scale turbulence. The coupled LES-synthetic turbulence (CLST) model is presented which aims to model the missing high frequencies. The CLST method resolves large-scale turbulent fluctuations from coarse-grid large-eddy simulations (CLES) and models small-scale fluctuations generated by a synthetic eddy method (SEM). Noise is predicted using a formulation of the linearized Euler equations (LEE), where the acoustic waves are generated by source terms from the combined fluctuations of the CLES and the stochastic fields. Sweeping and straining of the synthetic eddies are accounted for by convecting eddies with the large turbulent scales from the CLES flow field. The near-field noise of a Mach 0.9 jet at a Reynolds number of 100,000 is predicted with LES. A high-order numerical algorithm, involving a dispersion relation preserving scheme for spatial discretization and an Adams–Bashforth scheme for time marching, is used for both LES and LEE solvers. Near-field noise spectra from the LES solver are compared to published results. Filtering is applied to the LES flow field to produce an under-resolved CLES flow field, and a comparison to the un-filtered LES spectra reveals the missing noise for this case. The CLST method recovers the filtered high-frequency content, agreeing well with the spectra from LES and showing promise at modeling the high-frequency range in the acoustic noise spectrum at a reasonable expense.

Reynolds and dispersive shear stress in free-surface turbulent channel flow over square bars.

Reynolds and dispersive shear stresses in turbulent flow over spanwise-aligned square bars in an open channel flow are examined. Results of large-eddy simulation of flow over two different bar spacings corresponding to transitional and k-type (reattaching flow) roughness are analyzed. The Reynolds shear stress contribution to the momentum loss (or the friction factor, respectively) is greater than the dispersive shear stress contribution. By increasing the bars spacing, however, the contribution of the dispersive shear stress increases while the Reynolds shear stress contribution decreases, which is due to a standing wave at the water surface in the flow over k-type roughness which results in significant spatial variations in the time-averaged velocities. Strong sweep events take place and contribute to the friction coefficient. Investigating the dynamics of the flow reveals that there is momentum source below the crest of the bars and momentum sink above them, leading to acceleration or deceleration of flow, respectively. The contribution of dispersive shear stress is significant only in the deceleration of the flow near the crest of the bars and in the acceleration of the flow under the water surface. Quantification of the three components of total kinetic energy, i.e. mean, turbulent, and wake kinetic energy, reveals that the largest contribution is that of the mean flow in both geometries. By increasing the bar spacing, the contributions of turbulent and wake kinetic energy, which are localized at the bar height, increase, while the kinetic energy of the mean flow decreases.

Generative modeling of turbulence

We present a mathematically well-founded approach for the synthetic modeling of turbulent flows using generative adversarial networks (GAN). Based on the analysis of chaotic, deterministic systems in terms of ergodicity, we outline a mathematical proof that GAN can actually learn to sample state snapshots from the invariant measure of the chaotic system. Based on this analysis, we study a hierarchy of chaotic systems starting with the Lorenz attractor and then carry on to the modeling of turbulent flows with GAN. As training data, we use fields of velocity fluctuations obtained from large-eddy simulations (LES). Two architectures are investigated in detail: we use a deep, convolutional GAN (DCGAN) to synthesize the turbulent flow around a cylinder. We furthermore simulate the flow around a low-pressure turbine stator using the pix2pixHD architecture for a conditional DCGAN being conditioned on the position of a rotating wake in front of the stator. The settings of adversarial training and the effects of using specific GAN architectures are explained. We thereby show that GAN are efficient in simulating turbulence in technically challenging flow problems on the basis of a moderate amount of training data. GAN training and inference times significantly fall short when compared with classical numerical methods, in particular, LES, while still providing turbulent flows in high resolution. We furthermore analyze the statistical properties of the synthesized and LES flow fields, which agree excellently. We also show the ability of the conditional GAN to generalize over changes of geometry by generating turbulent flow fields for positions of the wake that are not included in the training data.

Large-Eddy Simulation of a wind turbine using a Filtered Actuator Line Model

When dealing with multirotor devices such as quadcopters or wind farms, the cost of blade-resolved large-eddy simulation (LES) becomes prohibitive. Combining LES with a family of lower-fidelity models, called actuator line models (ALMs), has grown in popularity in the past decade. ALM replaces full blade resolution with an array of actuator points or lines parameterized by aerodynamic lift/drag polar plots along the blades. Body forces computed based on these actuator points are then projected onto the LES flow mesh, mimicking the effect of rotating blades on the flow. However, the optimal projection radius and the associated LES grid size is often too restrictive for multirotor simulations. Recently, a new tip-correction-based filtered ALM (F-ALM) was proposed by Martínez-Tossas and Meneveau (2019), which allows coarser-than-optimal grids by avoiding the associated overprediction of thrust. In this work, F-ALM is implemented into a high-order, in-house LES code to simulate National Renewable Energy Laboratory Phase VI wind turbine. It is then followed by a comparison between the baseline ALM and the newly implemented F-ALM in terms of instantaneous and time-averaged flow fields and blade loads, revealing the advantages of F-ALM in preventing the overprediction of power on coarse grids. This encourages accurate and affordable simulations of multirotor devices in the future.

Controversial turbulent Schmidt number value in particle-laden boundary layer flows

When solid particles, such as sediments, are transported by a turbulent boundary layer flow, the wall-normal turbulent diffusion of particle concentration has been observed to be much higher than the turbulent diffusion of momentum (the eddy viscosity). This is characterized by a turbulent Schmidt number lower than unity. By using the few available existing direct measurements of Reynolds-averaged wall-normal particle flux and our two-phase flow large eddy simulations we demonstrate that the actual turbulent Schmidt number is indeed higher than unity. The misinterpretation may be explained by settling retardation of particles in turbulent flows compared with quiescent water.

Assessment of Wind Field Generation Methods on Predicted Wind Turbine Power Production Using a Free Vortex Filament Wake Approach

Abstract The generated power and thrust of a wind turbine strongly depend on the flow field around the turbine. In the present study, three different inflow methods, i.e., a time series (TS) from large eddy simulation (LES) of atmospheric boundary layer flow field, a synthetic turbulent flow field using the Mann model (MM), and a steady-state mean wind profile with shear, are integrated with the free vortex filament wake method to investigate the effect of wind field generation methods on the wind turbine performance where the impact of the turbine and the trailing wake vortices on the turbulent flow fields is ignored. For this purpose, an in-house vortex lattice free wake (VLFW) code is developed and used to predict the aerodynamic loads on rotor blades. The NREL 5-MW reference wind turbine is used for the VLFW simulations. For a fair assessment of different inflow generation methods on power production of a wind turbine, it is not sufficient that the generated wind fields employed in the TS and MM methods have the same streamwise mean velocity and turbulence intensity at hub height. Instead, the generated inflows must have equivalent power-spectral densities especially at low frequencies since the rotor blades essentially respond to the large-scale fluctuations (macroscopic scales) rather than small-scale fluctuations (microscopic scales). A faster energy decay rate of LES inflow leads to a higher energy content in the TS method at low frequencies (associated with the macroscopic dynamics of the rotor blades). This extra kinetic energy results in a slightly higher mean power production while using the TS method although the inflow conditions at hub height/rotor plane are the same for both the TS and MM methods. Moreover, the impact of simulation time (the length of time integration) on the power production of a wind turbine (exposed to an unsteady inflow) must be taken into account. A short simulation time remarkably affects the mean wind speed over the rotor area for identical turbulent inflows. For Taylor’s hypothesis application using a single LES flow field, the results show a significant difference in the mean powers corresponding to the different realizations due to large turbulent fluctuations.

LES of Turbulent Flow through Square Duct with Rounded Corners

In the present study, large eddy simulation (LES) for turbulent flows through square ducts with four rounded corners is performed using immersed boundary (IB) method. The geometrical shape of the duct is determined by an equation |y|α+ |z|α= Rα with a deformation parameter α. The distribution and magnitude of the mean secondary flow with α≧4 is almost close to those of the square-duct case. The center position of secondary circulation flows with α≦4 approaches the square-side bisector in contrast to the square-duct case. The profiles of the Reynolds stress and behaviour of the instantaneous turbulent structures in case of α≧4 are similar with those in square duct case.

Fragmentation and atomization characteristics of near-orifice liquid jet under transverse gas film

In order to understand the fragmentation and atomization characteristics of the liquid jet in transverse gas film, a pintle injection element using air and water as simulants is designed. The two-phase flow large eddy simulation and backlight imaging are used to study the liquid-jet breakup process and spray-field dynamic characteristics in the nearorifice area of pinte injection element under the atmospheric environment. The primary fragmentation process of the liquid jet dominated by surface wave is obtained by large eddy simulation, which reveals the establishment process of the spray field in near-orifice area of the gas-liquid pintle injector. After the subsonic airflow leaves the slit, it expands and accelerates into supersonic state. Then the deceleration and pressurization phenomenon occurs once the supersonic airflow passes through the detached bow shock upstream of the liquid jet. The liquid jet bends downstream due to the difference in pressure between upstream and downstream, and the Rayleigh-Taylor (R-T) unstable surface wave appears on the jet windward surface. As the surface wave develops, the penetration of the wave trough by airflow causes the continuous liquid jet to fragment. Proper orthogonal decomposition (POD) method can effectively reconstruct spray snapshot. The POD mode shows that the low-frequency spray oscillation in near-orifice area is caused by the overall expansion/contraction process of the spray field, while the high-frequency one is due to the “impact wave” movement of the liquid block or liquid mist group on the windward side. The latter is produced by the R-T unstable surface wave before the jet breakup, and can be categorized as traveling wave structure. The dimensionless traveling wave wavelength has a power-law relationship with Weber number.

Numerical simulations and analysis of the turbulent flow field in a practical gas turbine engine combustor

The turbulent flow field in a practical gas turbine combustor is very complex because of the interactions between various flows resulting from components like multiple types of swirlers, dilution holes, and liner effusion cooling holes. Numerical simulations of flows in such complex combustor configurations are challenging. The challenges result from (a) the complexities of the interfaces between multiple three-dimensional shear layers, (b) the need for proper treatment of a large number of tiny effusion holes with multiple angles, and (c) the requirements for fast turnaround times in support of engineering design optimization. Both the Reynolds averaged Navier–Stokes simulation (RANS) and the large eddy simulation (LES) for the practical combustor geometry are considered. An autonomous meshing using the cut-cell Cartesian method and adaptive mesh refinement (AMR) is demonstrated for the first time to simulate the flow in a practical combustor geometry. The numerical studies include a set of computations of flows under a prescribed pressure drop across the passage of interest and another set of computations with all passages open with a specified total flow rate at the plenum inlet and the pressure at the exit. For both sets, the results of the RANS and the LES flow computations agree with each other and with the corresponding measurements. The results from the high-resolution LES simulations are utilized to gain fundamental insights into the complex turbulent flow field by examining the profiles of the velocity, the vorticity, and the turbulent kinetic energy. The dynamics of the turbulent structures are well captured in the results of the LES simulations.

Secondary vortex, laminar separation bubble and vortex shedding in flow past a low aspect ratio circular cylinder

Large eddy simulation of flow past a circular cylinder of low aspect ratio ( $AR=1$ and $3$ ), spanning subcritical, critical and supercritical regimes, is carried out for $2\times 10^3 \le Re \le 4\times 10^5$ . The end walls restrict three-dimensionality of the flow. The critical $Re$ for the onset of the critical regime is significantly lower for small aspect ratio cylinders. The evolution of secondary vortex (SV), laminar separation bubble (LSB) and the related transition of boundary layer with $Re$ is investigated. The plateau in the surface pressure due to LSB is modified by the presence of SV. Proper orthogonal decomposition of surface pressure reveals that although the vortex shedding mode is most dominant throughout the $Re$ regime studied, significant energy of the flow lies in a symmetric mode that corresponds to expansion–contraction of the vortex formation region and is responsible for bursts of weak vortex shedding. A triple decomposition of the time signals comprising of contributions from shear layer vortices, von Kármán vortex shedding and low frequency modulation due to the symmetric mode of flow is proposed. A moving average, with appropriate size of window, is utilized to estimate the component due to vortex shedding. It is used to assess the variation, with $Re$ , of strength of vortex shedding as well as its coherence along the span. Weakening of vortex shedding in the high subcritical and critical regime is followed by its rejuvenation in the supercritical regime. Its spanwise correlation is high in the subcritical regime, decreases in the critical regime and improves again in the supercritical regime.

Large-eddy simulation of flow around an isolated building: A step-by-step analysis of influencing factors on turbulent statistics

We designed and exclusively performed a wind tunnel experiment to obtain high-quality and high-resolution measurement results for three components of mean and fluctuating velocities around a 1: 1: 2 isolated building to establish a benchmark case for large-eddy simulation (LES) as a part of the activities of the Architectural Institute of Japan. Subsequently, we conducted LES to investigate the effect of computational grid arrangement on turbulent statistics by changing three grid arrangements. Finally, we performed a comparison by changing spatial discretization schemes, sub-grid scale (SGS) turbulence models, and convergence criteria. A grid in which the width of a building was discretized into 20 grids yielded sufficient resolution to reproduce the mean flow and second-order turbulent statistics. Under this appropriate grid arrangement, in the condition that the blending ratio of the first-order upwind interpolation scheme with the second-order linear interpolation scheme from 5 to 10%, the appropriate second-order statistics of wind velocity around the isolated building was obtained and numerical oscillation was effectively removed. Then, with this grid, any SGS model yielded appropriate results. In addition, the convergence criteria had little effects on the LES results if stable solution was obtained. Because the convergence criteria greatly affect the total calculation cost, weaker appropriate criteria should be set.

Large Eddy Simulation of the By-pass Transition Process under Different Inlet Turbulence Conditions

The transition process of the boundary layer developing over a flat plate with elevated inlet Free Stream Turbulence Intensity (FSTI) has been studied by means of Large Eddy Simulation (LES). To this purpose, four cases with different inflow disturbances have been tested varying the magnitude and the length scale of turbulence. LES has been performed by using the finite-volume ANSYS Fluent code. The computational domain, which was constituted by a rectangular domain with a zero thickness plate, was based on an ERCOFTAC test case in order to provide a validation with a well-known set of data by comparing the boundary layer integral parameters and mean and fluctuating streamwise velocity profiles.The four cases were discussed within the paper by looking at classical statistical properties as well as advanced post-processing tools. It was shown that the decrease in the free stream turbulence level postpones the transition location, whereas the variation of the integral length scale has a very low influence on the distribution of the time-mean flow properties. Proper Orthogonal Decomposition (POD) has been applied to the instantaneous LES flow fields in order to provide a statistical representation of the structures responsible for transition and their response to free-stream turbulence intensity and length scale. The presence of vortical filaments parallel to the wall, typically referred as boundary layer streaks, is clearly identified; their characteristic dimensions and how they change as a function of FSTI properties were analyzed within the paper.

A divergence-free inflow turbulence generator using spectral representation method for large-eddy simulation of ABL flows

The generation of inflow turbulence is the first necessary step to conduct a successful large-eddy simulation (LES) of atmospheric boundary layer (ABL) flows. It is required that the generated inflow turbulence is divergence-free and satisfies essential target statistical properties, including the two-point statistics of the flow described by the coherency function and the spatial correlation. In this paper, a computationally efficient synthetic inflow turbulence generation technique with explicitly defined two-point flow statistics is proposed based on the spectral representation method. The divergence-free condition is imposed on the generated turbulence through a posteriori procedure embedded in the Pressure-Implicit with Splitting of Operators (PISO) solver of OpenFOAM. The efficacy of the proposed method to represent one-point and two-point statistics is demonstrated by comparing the generated turbulence with wind field measurements taken at the Boundary Layer Wind Tunnel Laboratory of Western University. The proposed method is then applied to LES of ABL flows for three exposure conditions, and the wind profiles downstream of the inlet are examined. The mean velocity, turbulence intensity, integral length scale profiles, and the velocity spectra are in good agreement with the target experimental data, thus, indicating the potential of the method for computational wind engineering applications such as wind load evaluation studies. The implementation of the proposed method is available at https://github.com/GBitsuamlak/DFSR.

Subgrid-scale helicity equation model for large-eddy simulation of turbulent flows

A new one-equation eddy-viscosity model based on subgrid-scale (SGS) helicity is introduced in this paper for large-eddy simulation (LES) of turbulent flows. First, the governing equation of SGS helicity is deduced from the incompressible Navier–Stokes equations, and it reflects the transfer of the small-scale helicity that has been filtered out. We deduce a certain functional relation between the eddy viscosity and SGS helicity based on the kinetic energy and helicity spectra in the homogeneous and isotropic helical turbulence. For improving the accuracy, each unclosed term in the governing equation of SGS helicity is modeled independently, and the coefficients of these unclosed terms are constants or are determined dynamically. The new one-equation eddy-viscosity model is first tested and validated in the simulation of the homogeneous and isotropic helical turbulence. The a priori tests from the direct numerical simulation of forced homogeneous and isotropic turbulence show that the energy and helicity fluxes exhibit scale invariance in the inertial subrange. Additionally, the a posteriori tests demonstrate that the constant-coefficient and dynamic SGS helicity equation models can predict both the energy and helicity spectra more precisely than the common SGS models. For the LES of channel flow, the SGS helicity equation model can accurately predict the mean velocity, the turbulent stress, and the viscous shear stress and supply more abundant flow structures than the compared SGS model under the same grid resolution.

Source term estimation in complex urban environments based on Bayesian inference and unsteady adjoint equations simulated via large eddy simulation

Establishing an accurate source–receptor relationship is essential for identifying unknown sources of air pollution in complex urban environments. Existing source term estimation (STE) methods determine this relationship via steady simulations of adjoint equations with ensembled-averaged flow fields. However, the effect of turbulent diffusion on the dispersion of pollutants is simplified according to the gradient dispersion hypothesis, which may result in considerable STE errors, especially in the complex urban areas. Therefore, we developed a new STE method by embedding unsteady adjoint equations modeled via large eddy simulation (LES) into a Bayesian inference framework. The tremendous storage requirement of the unsteady simulation was mitigated using a wavelet-based compression method. The performance of the proposed method was then evaluated based on dispersion measurements collected in a wind tunnel experiment for a regular, block-arrayed building group model. The estimation results were compared with those derived from an existing method, in which the steady simulation of adjoint equations was employed based on the mean LES flow field. Our results suggest that the LES-based approach significantly improved the modeling accuracy of the adjoint equations and the resulting STEs. If the 50th percentile of the posterior probability was regarded as the point estimate, the absolute errors in source location and strength were reduced by 89% and 99%, respectively, when compared to the existing method.

Validation of large eddy simulation of flow behind a circular cylinder

The flow in the wake behind a circular cylinder in a cross-flow at Reynolds number of 4815 was studied both experimentally and via mathematical modeling. The mathematical model was performed as a Large Eddy Simulation (LES), while the experiments were carried out using the time-resolved variant of the Particle Image Velocimetry (PIV) method. Both the simulation and experiment took into account the dynamical aspects of the studied phenomenon, which enabled a detailed validation of the mathematical model. The overall statistical properties of the simulated flow were validated via comparing the time-averaged measured and computed velocity and vorticity fields. To validate the dynamical behavior, the velocity spectra were examined first. Next, the Proper Orthogonal Decomposition (POD) of the spatio-temporal velocity data was performed on both the experimental and numerical data and a comparison of the obtained energetic modes was carried out. All the performed validations have shown a satisfactory agreement between the simulation and the experiment.

Turbulent flow and heat flux analysis from validated large eddy simulations of flow past a heated cylinder in the near wake region

We conduct non-isothermal large eddy simulations (LESs) of flow past a heated cylinder (Re = 3900) to investigate flow physics throughout the wake region and develop a foundation upon which future heat flux wall models can be built (both for wall-modeled LES and other lower fidelity models) for mathematical closure of the energy equation. A rigorous validation of the mesh is made under isothermal conditions with results showing a closer match to experimental data than any other LES studies to date. The insights gained into the mesh design and approach are discussed. Simulation of non-isothermal flow is performed on the validated mesh for temperature differences between the cylinder surface and the freestream of 25 K and 300 K. The mesh design and realistic (temperature-dependent) thermodynamic property variations play key roles in predicting delayed separation, larger re-circulation zones, and enhanced turbulence intensity for the higher temperature difference case. The effect of both temperature differences on the flow is analyzed, and a new scaling of the flow domain is proposed to gain further insight into non-isothermal flow physics. Key scaling variables, friction temperature and friction velocity, are able to reduce nearly all of the temperature dependence of first and second order flow statistics, including turbulent heat fluxes. This leads to the finding that the turbulent heat flux in the wake region scales with the wall heat flux irrespective of the temperature difference in the flow.