LES Data Research Articles

Assessment of Turbulence Models for Predicting the Interaction Region in a Wall Jet by Reference to LES Solution and Budgets

Highly-resolved LES and experimental data for a plane wall jet are used to study the characteristics of turbulence-closure proposals, mainly within the framework of second-moment-transport modelling. The study is motivated by the observed importance of diffusive Reynolds-stress transport in the interaction region between the outer shear layer and the near-wall layer of the wall jet, which gives the near-wall flow characteristics that are very different from those of a conventional boundary layer. Comparisons are presented for mean-flow quantities, second moments and budgets. Also included are a priori studies of approximations for the pressure-velocity interaction, pressure-fluctuation-driven transport and turbulent transport of the Reynolds stresses by triple correlations, the last observed to contribute significantly to the stress budgets. The study reveals major defects in the closure approximations for the pressure-velocity interaction terms, especially in the near-wall region. These defects result in a poor representation by the particular second-moment closures investigated of even the integral and mean-flow characteristics of the wall jet.

LES, T-RANS and hybrid simulations of thermal convection at high Ra numbers

Abstract The paper reports on application of different approaches to the simulations of thermal convection at high Rayleigh ( Ra ) numbers. Based on new well-resolved LES in 10 7 ⩽ Ra ⩽ 10 9 range, the performance of a T-RANS (using a low- Re three-equation 〈 k 〉–〈 e 〉–〈 θ 2 〉 ASM/AFM subscale model) and a hybrid approach (utilizing the concept of “seamless” RANS/LES merging) have been compared. Targeting accurate predictions of heat transfer at very high Ra numbers, the near-wall behaviour of the subscale turbulence contributions is analyzed in details. Whilst the application of the conventional LES on a coarse grid resulted in huge underprediction of the Nusselt number (50% at Ra = 10 9 ), thanks to a well-tuned subscale model, the T-RANS results showed excellent agreement for heat transfer with both the fine-resolved LES (for Ra ⩽ 10 9 ) and experimental data for over a ten-decades range of Ra (10 6 ⩽ Ra ⩽ 2 × 10 16 ). The visible absence of the fine-scale motion in the T-RANS, however, means that the T-RANS will perform poorly in flows where there are no strong large-scale forcing, as, e.g., in a side-heated vertical channel. In order to sensitize the T-RANS approach to high-frequency instabilities, different ways of hybrid RANS/LES merging based on “seamless” approach have been investigated. It is demonstrated that the hybrid approach is capable of capturing a significantly larger portion of the fine-structure spectrum than is possible with T-RANS, whilst also returning accurate predictions of heat transfer and turbulence statistics.

Large Eddy Simulation of turbulent thermal convection using a Mixed Scale Diffusivity Model

The Mixed Scale Diffusivity Model, originally developed in the case of the differentially heated cavity, is applied to compute turbulent Rayleigh-Benard flow in an infinite fluid layer at Pr = 0.71 for a large range of Rayleigh numbers (6.3 × 105 – 2 × 1011). The effect of this SGS modelling, which adjusts locally the SGS diffusivity to the thermal scales of the flow and results in variable PrSGS like the dynamic approach, is emphasised by comparison with LES and TRANS literature data. A single scaling regime is found in a range of Rayleigh numbers 6.3 × 105 – 2 × 1010, whose properties include the Ra0.302 scaling law for the Nusselt number and for the thermal boundary layer thickness, in agreement with the experimental correlation of Niemela et al. (2000). The first indication of a transition towards a new regime appears above Ra = 1011.

Large-eddy simulation and deduced scaling analysis of Rayleigh–Bénard convection up to Ra = 109**

Large-eddy simulation of turbulent Rayleigh–Bénard (RB) convection has been performed for a 6:1:6 open-ended domain for Rayleigh numbers ranging from 6.3 × 105 to 109 at Prandtl number of Pr = 0.71. The scaling analysis based on the LES data shows that the heat transfer follows a single relation of Nu = 0.162 Ra 0.286, which is consistent with the scaling law for the hard turbulence regime reported in several earlier experimental and DNS studies. The present LES also supports some earlier experimental and DNS findings that most of characteristic parameters can be scaled reasonably well with Ra number in the considered Ra number range using a single relation. Nonetheless, it is found that the scaling of several quantities shows a sensible offset from a single relation, and could be fitted better with the separate scaling relations for the soft and hard convective turbulence transitioned at about Ra = 4 × 107. It has been argued that the transition, reflected in the scaling relation, may be attributed to the increasing ‘containing effect’ of the plume leaving the horizontal wall on the plume approaching the wall at large Ra numbers in the near-wall region. **This paper is a modified version from the paper presented at the Forth International Symposium of Turbulence and Shear Flow Phenomena (Williamsburg, Virginia, 27–29 June 2005).

System identification of a large-scale swirled partially premixed combustor using LES and measurements

The forced response of a swirled partially premixed burner to acoustic excitations is computed with Large Eddy Simulations and compared to measurements performed at University of Karlsruhe. The configuration is a 1:1 representation of a prototype industrial burner provided by Siemens in which gases are injected through two complex-geometry swirlers. Thanks to hybrid meshes and parallel computations, the simulation integrates a large part of this complex geometry. The inlet of the largest swirler is numerically forced for three acoustic frequencies (80, 120 and 250 Hz) and the LES data is compared to experimental results in terms of combustion phase. LES also give access to the forced flow topology, revealing that the excitation creates large toroidal vortices which distort the flame surface and destabilise downstream of the flame. In addition to these toroidal vortices, LES also show that a precessing vortex core appears when excitation is started. Finally maps of combustion phase and response amplitude are extracted from LES and highlight the mechanisms controlling the flame behaviour under excitation. The phases obtained by LES are also compared to usual “convective phase” estimates used in Reynolds Averaged codes. This comparison shows that LES provide a better estimation of the phase between heat release and inlet velocity fluctuations.

Hybrid RANS/LES modelling with an approximate renormalization group. II: Applications

The hybrid RANS/LES model developed previously is extended with near-wall modifications. The model is validated, in its low-Reynolds form, for channel flow and for flow over a periodic hill. The results for the channel flow are (indirectly) compared with DES results for the same flow, and are qualitatively similar. The separated flow over the periodic hill shows good agreement with the reference LES data, although performed with about 20 times less grid points. Finally the model is applied in high-Reynolds form with wall functions to a sudden pipe expansion; good agreement with experimental data is also obtained.

표준영상을 이용한 2차원 PIV와 3차원 PIV 성능시험

Quantitative performance test on the conventional 2D-PIV and the hybrid angular 3D-PIV (Stereoscopic PIV) was earned out LES Data sets on an impinging jet winch are provided on the webpage(httpㆍ//www.vsj or jp/piv) for the PIV Standard Project were used for the generation of virtual images. The generated virtual images were used for the 2D-PIV and 3D-PlV measurements test It has been shown that the results obtained by 2D-PIV on average values are slightly closer to the LES data than those obtained by 3D-PIV, but the turbulent properties obtained by 2D-PIV are largely underestimated than those obtained by 3D-PIV.

Joint effects of geometry confinement and swirling inflow on turbulent mixing in model combustors: a second-moment closure study

An isothermal, incompressible, swirling flow in three generic combustor configurations featuring different inflow structure with respect to the circumferential velocity type (both configurations with annular and central swirling jet were considered), combustor confinement (in terms of expansion ratio ER) and swirl intensity (S) was studied computationally by means of Reynolds Averaged Navier-Stokes method (RANS), using second-Moment Closure (SMC) models. The work focuses on the investigation of the combined effects of the above-mentioned flow parameters on the mixing between a swirling annular jet (representing air stream) and the non-swirling inner jet (representing fuel) within a combustor. Both the basic high-Reynolds number SMC model, modified to account for the non-linearities in the pressure scrambling and dissipation processes, and a low-Reynolds number SMC model, accounting separately for the viscous and non-viscous wall blockage, were applied. Inflow conditions are computationally generated, rather than prescribed. In the course of the inflow data generation, a number of the equilibrium swirling flows in the concentric annulus and pipe geometries, from which coaxial and central swirling jets expand into a combustion chamber, are computed. The SMC results reproduced all important mean flow and turbulent features in good agreement with available experimental and LES data.

Large‐scale structures at various stages of separated boundary layer transition

AbstractLarge‐eddy simulation is employed to investigate separated boundary layer transition induced by a change of curvature of the surface. The geometry is a flat plate with a semi‐circular leading edge. The Reynolds number based on the uniform inlet velocity and the leading edge diameter is 3450. The simulated mean and turbulence quantities compare well with the available experimental data. The LES data have been comprehensively analysed to elucidate the transition process leading to breakdown to turbulence. Large‐scale structures have been observed at various stage of the transition process. Understanding the formation of these large‐scale structures, their evolution and eventual break‐up into smaller structures may help to shed light on the transition mechanism. Copyright © 2002 John Wiley & Sons, Ltd.

Development of Genetic Algorithm based 3D-PTV technique

A new 3D-PTV technique has been developed using GA (Genetic Algorithm). The measurement system consists of three CCD cameras, Ar-ion laser, an image grabber and a host computer. One-to-one correspondences are made by taking advantage of the combinatorial optimization of the genetic algorithm for the whole particles of the two images during the time interval of image frames. Two fitness functions are introduced in order to enhance the correspondences of the particles. One is based on the concept of continuum theory and the other one is based on the minimum error of three-dimensional distance. Performance of the developed algorithm is tested using a set of virtual images constructed by the use of LES data set on an impinging jet. The developed 3D-PTV system is successfully applied to the measurement of flow characteristics of the wake of a circular cylinder.

Investigation of Anisotropy-Resolving Turbulence Models by Reference to Highly-Resolved LES Data for Separated Flow

The predictive properties of several non-linear eddy-viscosity models are investigated by reference to highly-resolved LES data obtained by the authors for an internal flow featuring massive separation from a curved surface. The test geometry is a periodic segment of a channel constricted by two-dimensional (2D) `hills' on the lower wall. The mean-flow Reynolds number is 21560. Periodic boundary conditions are applied in the streamwise and spanwise directions. This makes the statistical properties of the simulated flow genuinely 2D and independent from boundary conditions, except at the walls. The simulation was performed on a high-quality, 5M-node grid. The focus of the study is on the exploitation of the LES data for the mean-flow, Reynolds stresses and macro-length-scale. Model solutions are first compared with the LES data, and selected models are then subjected to a-priori studies designed to elucidate the role of specific model fragments in the non-linear stress-strain/vorticity relation and their contribution to observed defects in the mean-flow and turbulence fields. The role of the equation governing the length-scale, via different surrogate variables, is also investigated. It is shown that, while most non-linear models overestimate the separation region, due mainly to model defects that result in insufficient shear stress in the separated shear layer, model forms can be derived which provide a satisfactory representation of the flow. One such model is identified. This combines a particular quadratic constitutive relation with a wall-anisotropy term, a high-normal-strain correction and a new form of the equation for the specific dissipation ω = ∈/k.

ASM, WETモデル, 渦粘性/渦拡散モデルのアプリオリテスト : 非等温室内気流のLESデータベースを用いた乱流モデルの評価

A priori test using the LES database created by the authors(2001)is carried out in order to evaluate the Algebraic Stress Model(ASM), the WET model and the linear eddy viscosity/diffusivity model. The accuracy and relative performance of various models are carefully examined in the flow field including supply jet, wall shear and impingement. ASM and the WET model reproduce well anisotropy features of Reynolds stresses and turbulent heat fluxes. However, normal stresses of ASM in the impinging region show slight discrepancy from the LES data. It is clarified quantitatively that this discrepancy is caused by the inaccuracy of the modeling pressure-strain term and algebraic approximation for convection and diffusion term. Furthermore, it is concluded that the WET model requires the use of appropriate wall damping function in order to preciously predict wall shear flows.

Lagrangian approaches for particle collisions: The colliding particle velocity correlation in the multiple particles tracking method and in the stochastic approach

Two different Lagrangian approaches for particle/particle collisions are described. The first model is based on the simultaneous tracking of several particles and suitable treatment is developed on particle pairs to detect collisions on each time step of the particle trajectory realization. The second method is based on a stochastic approach where one single particle is tracked, and successive random processes are applied to generate a fictitious partner of collision. In order to validate both approaches, simulations have been carried out in homogeneous isotropic turbulence and they have been compared with LES data. The particle/particle correlated motion through the surrounding fluid is proved to be a key parameter in a particle/particle collision process.

New Third-Order Moments for the Convective Boundary Layer

Abstract Turbulent convection is inherently a nonlocal phenomenon and a primary condition for a successful treatment of the convective boundary layer is a reliable model of nonlocality. In the dynamic equations governing the convective flux, the turbulent kinetic energy, etc., nonlocality is represented by the third-order moments (TOMs). Since the simplest form, the so-called down gradient approximation (DGA), severely underestimates the TOMs (up to an order of magnitude), a more physical model is needed. In 1994, an analytical model was presented that was derived directly from the dynamical equations for the TOMs. It considerably improved the DGA but was a bit cumbersome to use and, more importantly, it was based on the quasi-normal (QN) approximation for the fourth-order moments. Here, we present a new analytic expression for the TOMs that is structurally simpler than the 1994 expression and avoids the QN approximation. The resulting fit to the LES data is superior to that of the 1994 model.

Towards the development of a Reynolds-averaged algebraic turbulent scalar-flux model

Abstract In order to derive a possible direction for developing Reynolds-averaged algebraic turbulent scalar-flux models, a priori explorations are attempted by processing the LES data presently performed for channel flows under several flow-boundary conditions including shear-free boundaries. The present calibration has elucidated that the turbulent scalar-flux vectors obtainable from the simple generalized gradient-diffusion hypothesis (GGDH) hardly align with the simulation results in wall-shear flows at Pr ⩾0.71. However, the GGDH form returns a quite reasonable approximation in shear-free flow regions and/or lower Pr fluid cases. In the former flow cases, it has been found that an introduction of quadratic products of the Reynolds-stress tensor into the gradient diffusion model may improve the predictive performance.

Improvement of a Low-Reynolds-Number Second Moment Closure and Its Application to Rotating Turbulent Channel Flows.

A low-Reynolds-number second moment closure is improved with the aid of DNS data and applied to turbulent channel flows with a spanwise rotation. Shima’s pressure strain model is refined and a multiple source model is introduced for the exact e equation. Computational results for the turbulent plane channel flows rotating in orthogonal mode are compared with the DNS, LES and experimental data. The present second moment closure achieves a level of agreement with DNS data similar to those for the non-rotating flow cases. In particular, it yields the improved performance in predicting the distributions of e in the near wall sublayer.

Turbulence and laminar structures: can they co-exist?

Schwarzschild first suggested that the laminar structures observed in the high Reynolds number Re = UL/nu approx. = (10(exp 12)) solar photosphere are the result of turbulence rather than a proof of its absence. He reasoned that since turbulence generates large turbulent viscosities nu(sub t) much greater than nu, the effective Reynolds number Re = UL/nu(sub t) approx. = O(1). Schwarzschild's argument is, however, incomplete for it assumes that the entire role of the non-linear interactions is to enhance viscosity. While this is not true in general, we present a proof of how and why it may occur, thus completing Schwarzschild's argument. We further discuss the fact that the same model non-local turbulence models have been shown to reproduce LES data for a variety of flows pertaining to astrophysics, geophysics and laboratory situations (at a fraction of the time).

Numerical Simulation of Rotating Channel Flows Using a Second Moment Turbulence Closure

A low-Reynolds-number second moment turbulence closure is improved with the aid of DNS data. For the model coefficients of pressure-strain terms, we adopted Shima's model with some modification. Shin and Choi's new dissipation-rate equation is employed to simulate accurately the turbulence energy dissipation rate distribution in the near wall sublayer. The results of computations are compared with DNS, LES data and experimental data for turbulent plane channel flow with rotation about spanwise axis. The present second moment closure achieves a level of agreement similar to that for the non-rotating. In particular, it accurately captures the distribution of turbulence energy dissipation rate in the near wall region.

Stellar Turbulent Convection. I. Theory

The description of stellar turbulent convection requires a minimum of five coupled, time-dependent, nonlocal, differential equations for the five variables: turbulent kinetic energy, turbulent potential energy, turbulent pressure, convective flux, and energy dissipation. Any fewer number of equations makes the model local. In this paper, we present the following results: 1. We derive the five coupled equations using a new turbulence model. The physical foundations and the turbulence statistics on which the model was tested are discussed. The model is able to reproduce the high Rayleigh number laboratory and direct numerical simulation data corresponding to medium-to-high values of the Peclet number (a measure of the efficiency of convection). 2. One of the major difficulties for any stellar convective model is the description of the low-efficiency, low Pe number region in which the physical timescale is no longer the turbulent timescale but the radiative one. No previous turbulence model has been able to incorporate these multiple timescales within the same framework properly. The present model does. 3. Overshooting is an unsolved problem in stellar structure. Its solution requires not only the above ingredients, but an additional one, a nonlocal model. This is because in the stably stratified region where ? - ?ad < 0, the only source of energy is diffusion, a nonlocal process. We discuss why the expressions used thus far to describe diffusion terms are inadequate. We then present a model that was successfully tested against LES data on the convective planetary boundary layer. 4. We analyze the nonlocal models of Gough and Xiong and discuss the approximations that are required to derive them from the full set of equations. 5. We discuss a model that relates the up/down drafts filling factors found by DNS/LES to the skewness of the velocity field which can be computed from the turbulence model. The results from DNS/LES and this model can thus be cross-checked. 6. We show that the stationary, local limit of the model reproduces recent local models (independently derived) which have been successfully tested against a variety of astrophysical data. 7. We discuss the fact that if the dissipation is described by a local model with a mixing length l (as done by all authors thus far), the remaining nonlocal equations exhibit divergences which preclude a physical solution to be found. OV results based on this method may be a coincidence since they are arrived at by fine tuning a coefficient. 8. The role of compressibility is discussed.

Toward the Development of an Algebraic Turbulent Heat Flux Model with the Aid of LES Data.

In order to derive a possible direction for improvements of an algebraic heat flux model, a priori explorations were attempted by processing the LES data presently performed under various flow and heat transfer conditions including shear-free boundaries. The present calibration has elucidated that the turbulent heat flux vectors obtainable from the simple GGDH hardly align with the simulation results in wall-shear flows with a wall normal temperature gradient of fluids at Prg0.71. However, the GGDH returns a quite reasonable approximation for heat flux vectors in shear-free flow regions and/or lower Pr fluid cases. In the former flow cases, it has been found that a further introduction of a tensor expression, consisting of quadratic products of the Reynolds stress tensor, into the gradient diffusion model may improve the predictive performance.