Quality Of Large Eddy Simulation Research Articles

Large-eddy simulation of a full-scale glove box fire

Reactive large-eddy simulations (LES) of the stage of peak heat release rate of open atmosphere glove box fires are carried out. To the best of the authors’ knowledge, this is the first time that the LES of such a geometry is reported in the literature. A multi-fuel extension of the eddy-dissipation model (EDM) is used to simulate the combustion of the two distinct solid fuels, PMMA and Polycarbonate, relevant to this configuration. Three increasingly refined grids are considered. All of them lead to temperature values that are in good agreement with experimental data, and similar levels of heat release are obtained inside the glove box. From a more quantitative point of view, only the most refined grid leads to satisfactory values of LES quality criteria. The intermediate grid allows resolving a fraction of the inertial range dynamics, while the solution obtained on the coarsest grid exhibits very large turbulent structures with characteristic sizes of the order of the integral length scale. The analysis confirms that the choice of the grid cell size has a significant influence on the description of the local flow hydrodynamics, the location and thickness of the reactive zones.

Verification and validation of LES of a triple parallel jet flow in the context of a thermal striping investigation

Large Eddy Simulation (LES) consists in explicitly resolving the largest turbulent scales while the small-scale motions are taken into account by means of a subgrid-scale (SGS) model. Although it remains computationally expensive, LES seems to constitute an increasingly employed tool for engineering applications in fluid mechanics. In a nuclear context e.g., LES is a relevant approach for the numerical study and prediction of the thermal striping, which can occur during the functioning of sodium-cooled nuclear reactors and may cause some damage to the structures. Numerous experiments where three differentially heated jets mix inside a cavity were designed and conducted by the Japan Atomic Energy Agency (JAEA) to study these phenomena. In order to evaluate the quality of LES using the CEA in-house TrioCFD code, a Verification and Validation study is proposed. Several LES of triple parallel jets are conducted with various grid sizes and SGS models. Following the idea of Celik et al. (2009), the effective SGS kinetic energy is written as a polynomial function of the grid size Δ:keff,sgs=amodΔpmod+anumΔpnum, where the two terms in the right-hand side are respectively the contributions of the numerical scheme and of the subgrid-scale model. In the Verification part, the orders of convergence of these two contributions are assessed for the Smagorinsky and WALE models. In the Validation part, an evaluation of the results is proposed by comparing computed velocity profiles to experimental measurements. The effects of grid resolution, SGS model and boundary conditions are discussed.

Large Eddy Simulations with Conjugate Heat Transfer (CHT) modeling of Internal Combustion Engines (ICEs)

In this study, the effects of the thermal boundary conditions at the engine walls on the predictions of Large-Eddy Simulations (LES) of a motored Internal Combustion Engine (ICE) were examined. Two thermal boundary condition cases were simulated. One case used a fixed, uniform wall temperature, which is typically used in conventional LES modeling of ICEs. The second case utilized a Conjugate Heat Transfer (CHT) modeling approach to obtain temporally and spatially varying wall temperature. The CHT approach solves the coupled heat transfer problem between fluid and solid domains. The CHT case included the solid valves, piston, cylinder head, cylinder liner, valve seats, and spark plug geometries. The simulations were validated with measured bulk flow, near-wall flow, surface temperature, and surface heat flux. The LES quality of both simulations was also discussed. The CHT results show substantial spatial, temporal, and cyclic variability of the wall heat transfer. The surface temperature dynamics obtained from the CHT model compared well with measurements during the compression stroke, but the absolute magnitude was 5 K (or 1.4%) off and the prediction of the drop in temperature after top dead center suffered from temporal resolution limitations. Differences in the predicted flow and temperature fields between the uniform surface temperature and CHT simulations show the impact of the surface temperature on bulk behavior.

Investigation of thermal stratification in premixed homogeneous charge compression ignition engines: A Large Eddy Simulation study

The operating range of Homogeneous Charge Compression Ignition (HCCI) engines is limited to low and medium loads by high heat release rates. Negative valve overlap can be used to control ignition timing and heat release by diluting the mixture with residual gas and introducing thermal stratification. Cyclic variability in HCCI engines with NVO can result in reduced efficiency, unstable operation, and excessive pressure rise rates. Contrary to spark-ignition engines, where the sources of cyclic variability are well understood, there is a lack of understanding of the effects of turbulence on cyclic variability in HCCI engines and the dependence of cyclic variability on thermal stratification. A three-dimensional computational fluid dynamics (CFD) model of a 2.0L GM Ecotec engine cylinder, modified for HCCI combustion, was developed using Converge. Large Eddy Simulations (LES) were combined with detailed chemical kinetics for simulating the combustion process. Twenty consecutive cycles were simulated and the results were compared with individual cycle data of 300 consecutive experimental cycles. A verification approach based on the LES quality index indicated that this modeling framework can resolve more than 80% of the kinetic energy of the working fluid in the combustion chamber at the pre-ignition region. Lower cyclic variability was predicted by the LES model compared to the experiments. This difference is attributed to the resolution of the sub-grid velocity field, time averaging of the intake manifold pressure boundary conditions, and different variability in the equivalence ratio compared to the experimental data. Combustion phasing of each cycle was found to depend primarily on the bulk cylinder temperature, which agrees with established findings in the literature. Large cyclic variability of turbulent mixing and spatial distribution of temperature was predicted. However, both of these parameters were found to have a small effect on the cyclic variability of combustion phasing.

Investigation of Numerical Effects on the Flow and Combustion in LES of ICE

This work investigates the influence of numerical dissipation on the modeled combustion in Large Eddy Simulations (LES). It is well known that capturing the dynamics of the in-cylinder flow is crucial for engine simulations, as it strongly affects flame propagation. The flame propagation during the power-stroke highly depends on the turbulence level that is developed throughout compression. This turbulence level will be strongly influenced by the accuracy of the numerical schemes employed. Even a small extent of upwinding, filtering, low-order implicit time-stepping, cell-stretching or mapping between grids may affect the flow field, the turbulence level, and hence the turbulent flame speed and the pressure curve. To provide a reference, the LES in-house code PsiPhi is used, which ensures a minimum of dissipation due to high order explicit time-stepping, homogeneous and isotropic filters and cells. Good stability of the code permits the use of a second-order Central Differencing Scheme (CDS) for the transport of momentum, avoiding numerical dissipation. To analyse the effect of numerical dissipation, simulations of a fired engine are performed using different numerical schemes for the convection of momentum. Physical quantities including the total kinetic energy, the velocity gradient, the turbulent viscosity, the in-cylinder pressure, the flame propagation or the burning rate of different test cases are evaluated and compared to each other to show the numerical effects on combustion. Furthermore, the suitability of common LES quality criteria including an energy criterion and viscosity ratio is discussed based on the comparison of simulations with less and more accurate numerics. It is shown that these LES quality indicators can be highly misleading.

Influence of mesh deformation on the quality of large eddy simulations

SummaryThe influence of mesh motion on the quality of large eddy simulation (LES) was studied in the present article. A three‐dimensional, turbulent pipe flow (Reτ=360) was considered as a test case. Simulations with both stretching and static meshes were carried out in order to understand how mesh motion affects the turbulence statistics. The spatial filtering of static and moving mesh direct numerical simulation (DNS) data showed how an ideal LES would perform, while the comparison of DNS cases with static and moving meshes revealed that no significant numerical errors arise from the mesh motion when the simulation is fully resolved. The comparison of the filtered fields of the DNS with a moving mesh with the corresponding LES fields revealed different responses to mesh motion from different numerical approaches. A straightforward test was applied in order to verify that the moving mesh works consistently in LES: when the mesh is stretched in the streamwise direction, the moving mesh results should be in between the two extremal resolutions between which the mesh is stretched. Numerical investigations using four different LES approaches were carried out. In addition to the Smagorinsky model, three implicit LES approaches were used: linear interpolation (non‐dissipative), the Gamma limiter (dissipative), and the scale‐selective discretisation (slightly dissipative). The results indicate that while the Smagorinsky and the scale‐selective discretisation approaches produce results consistent with the resolution of the non‐static mesh, the implicit LES with linear interpolation or the Gamma scheme do not. Copyright © 2016 John Wiley & Sons, Ltd.

Numerical Methods and Turbulence Modeling for LES of Piston Engines: Impact on Flow Motion and Combustion

In this article, Large Eddy Simulations (LES) of Spark Ignition (SI) engines are performed to evaluate the impact of the numerical set-upon the predictedflow motion and combustion process. Due to the high complexity and computational cost of such simulations, the classical set-up commonly includes “low” order numerical schemes (typically first or second-order accurate in time and space) as well as simple turbulence models (such as the well known constant coefficient Smagorinsky model (Smagorinsky J. (1963) Mon. Weather Rev. <b>91<b/>, 99-164). The scope of this paper is to evaluate the feasibility and the potential benefits of using high precision methods for engine simulations, relying on higher order numerical methods and state-of-the-art Sub-Grid-Scale (SGS) models. For this purpose, two high order convection schemes from the Two-step Taylor Galerkin (TTG) family (Colin and Rudgyard (2000) J. Comput. Phys. <b>162<b/>, 338-371) and several SGS turbulence models, namely Dynamic Smagorinsky (Germano et al. (1991) Phys. Fluids <b>3<b/>, 1760-1765) and sigma (Baya Toda et al. (2010) Proc. Summer Program 2010, Stanford, Center for Turbulence Research, NASA Ames/Stanford Univ., pp. 193-202) are considered to improve the accuracy of the classically used Lax-Wendroff (LW) (Lax and Wendroff (1964) Commun. Pure Appl. Math. <b>17<b/>, 381-398) - Smagorinsky set-up. This evaluation is performed considering two different engine configurations from IFP Energies nouvelles. The first one is the naturally aspirated four-valve spark-ignited F7P engine which benefits from an exhaustive experimental and numerical characterization. The second one, called Ecosural, is a highly supercharged spark-ignited engine. Unique realizations of engine cycles have been simulated for each set-up starting from the same initial conditions and the comparison is made with experimental and previous numerical results for the F7P configuration. For the Ecosural engine, experimental results are not available yet and only qualitative comparisons are performed to enforce the analysis and conclusions made on the F7P configuration. Regarding SGS models, only slight differences are found at the aerodynamic level even if sigma allows a better resolution of small structures of the velocity field. However, all results are in cycle-to-cycle variability envelopes from Granet (Granet et al. (2012) Combust. Flame <b>159<b/>, 1562-1575) and these single cycle computations don’t permit to distinguish clear improvements on macroscopic parameters such as resolved kinetic energy, heat release or mean in-cylinder pressure. Concerning numerical schemes, TTG schemes also allow a slighlty better resolution of small scale vortices but global quantities such as resolved kinetic energy and SGS viscosity are comparable. Nevertheless, clear differences appear between the different schemes in the combustion stroke. This is attributed to a better resolution of the flame-turbulence interaction process during the free flame propagation period, leading to an increase of the resolved part of heat release. It is also shown in this paper that an adjustment of the efficiency constant in the Thickened Flame (TF) model is compulsory to account for the over dissipation of the smallest resolved structures if LW is used. In the light of these conclusions an hybrid setup, called ES O<sub>2<sub/> (Engine Stroke Optimal Order), which consists in using TTGC during combustion and LW elsewhere is proposed and applied to the two engines configurations. Results are in good agreement with the ones obtained in the case of a full TTGC simulation, while the CPU (Central Processing Unit) cost increase is only about 10% compared to LW. The accuracy of LW seems therefore to be sufficient for pure aerodynamic phases, while the use of TTGC only during combustion permits an improvement in the LES quality. The hybrid ES O<sub>2<sub/> method thus appears as an attractive approach to improve further calculations accuracy without being greatly penalized by additional CPU costs in multi-cycle simulations.

Large-eddy simulation of the flow and acoustic fields of a Reynolds number 105 subsonic jet with tripped exit boundary layers

Large-eddy simulations (LESs) of isothermal round jets at a Mach number of 0.9 and a diameter-based Reynolds number ReD of 105 originating from a pipe are performed using low-dissipation schemes in combination with relaxation filtering. The aim is to carefully examine the capability of LES to compute the flow and acoustic fields of initially nominally turbulent jets. As in experiments on laboratory-scale jets, the boundary layers inside the pipe are tripped in order to obtain laminar mean exit velocity profiles with high perturbation levels. At the pipe outlet, their momentum thickness is δθ(0)=0.018 times the jet radius, yielding a Reynolds number Reθ=900, and peak turbulence intensities are around 9% of the jet velocity. Two methods of boundary-layer tripping and five grids are considered. The results are found to vary negligibly with the tripping procedure but appreciably with the grid resolution. Based on analyses of the LES quality and on comparisons with measurements at high Reynolds numbers, fine discretizations appear necessary in the three coordinate directions over the entire jet flow. The final LES carried out using 252×106 points with minimum radial, azimuthal, and axial mesh spacings, respectively, of 0.20, 0.34, and 0.40×δθ(0) is also shown to provide shear-layer solutions that are practically grid converged and, more generally, results that can be regarded as numerically accurate as well as physically relevant. They suggest that the mixing-layer development in the present tripped jet, while exhibiting a wide range of turbulent scales, is characterized by persistent coherent vortex pairings.

Error-Landscape Assessment of Large-Eddy Simulations: A Review of the Methodology

A review is presented of the error-landscape methodology. This approach evaluates the error-response surface of large-eddy simulations (LES) to essential model and numerical parameters by a systematic variation of these parameters. Using an error landscape constructed for LES of decaying homogeneous isotropic turbulence, it is shown that the determination of LES quality based on one error measure alone, can lead to misleading results, related to underlying error-balancing mechanisms. This problem can be avoided by considering a range of errors simultaneously, emphasizing different scales in the solution. Subsequently, the error-landscape method is further illustrated by comparing different numerical discretizations for Smagorinsky LES. Finally, a more complex case, i.e. a high (infinite) Reynolds number boundary layer, is considered.

An improved method to assess the quality of large eddy simulations in the context of implicit filtering

Large eddy simulation (LES) quality assessment is very important in view of predictive LES applications, but it is more complex than the verification of solutions from the Reynolds averaged Navier–Stokes equations. One reason is that the numerical discretization error and the subgrid scale model contribution depend on the grid resolution and that both terms interact. Recently, a method has been proposed to evaluate these error contributions by a systematic grid and model variation, assuming that the numerical error and the modeling error scale like a power of the grid spacing resp. filter width. A second-order dissipation error has been assumed in that work. However, theoretical arguments suggest that the true subgrid stresses scale like Δ2/3. The determination of the suitability and generality of this assumption is part of the present work. It will be shown that smaller values of the scaling exponent for the modeling error, i.e. m≈ 2/3, seem to be more appropriate. Furthermore, a more conservative measure for estimating the uncertainty is proposed. Besides a wall bounded flow (channel flow) and a shear flow (plane jet) the new procedure will be applied to a third important flow category, swirling flows.

An Attempt to Assess the Quality of Large Eddy Simulations in the Context of Implicit Filtering

While methods for assessing the uncertainty of Reynolds–Averaged–Navier–Stokes (RANS) simulations have been well established in the past, the verification of Large Eddy Simulations (LES) is more difficult. One reason is that the numerical discretization error as well as the subgrid scale model contribution depend on the grid resolution and that both terms interact. In the present paper the accuracy of single-grid estimators to assess the amount of the unresolved turbulent kinetic energy is studied first. In the second part of the paper the sensitivity of the simulation results on the modeling error as well as the numerical error will be investigated in the context of LES with implicit filtering. This will be achieved by performing a systematic grid and model variation. The analysis is applied to an isothermal, turbulent, plane jet and a turbulent channel flow.

Index of Resolution Quality for Large Eddy Simulations

In the light of rapidly increasing applications of large-eddy simulations (LES), it is deemed necessary to impose some quality assessment measures for such studies. The verification of LES calculations is difficult because of the fact that both the subgrid scale (SGS) model contribution and numerical discretization errors are functions of the grid resolution. In this study, various indexes of quality measures, hereafter referred to as LES̱IQ, are proposed. The recommended LES̱IQ is based on the Richardson extrapolation concept. This method has been applied to various cases and the calculated LES̱IQ results are compared with the relative total experimental and direct numerical simulation (DNS) error, defined as IQ̱EX and IQ̱DNS, respectively. It is postulated that in practical applications of LES, numerical dissipation will always be a significant part of the overall dissipation, and it must be accounted for in any assessment of the quality of LES. It is further suggested that LES̱IQ of 75% to 85% can be considered adequate for most engineering applications that typically occur at high Reynolds numbers; the proposed index is an indicator of good resolution (i.e., verification), but not necessarily a good or accurate model (i.e., validation).

Towards Technical Application of Large Eddy Simulation

AbstractThe paper is concerned with the computation of high Reynolds number circular cylinder flow (Re = 3900/140,000) based on the large eddy simulation (LES) technique. Because this flow involves a variety of complex flow features encountered in technical applications, successful simulations for this test case, especially at high Reynolds numbers, can be considered as the first step to real world applications of LES. Based on an efficient finite‐volume LES code, a detailed study on different aspects influencing the quality of LES results was carried out. In the present paper, some of the results are presented and compared with experimental measurements available.