On The Validation of LES Applied to Internal Combustion Engine Flows: Part 1: Comprehensive Experimental Database

Improved understanding of in-cylinder flows requires knowledge from well-resolved experimental velocimetry measurements and flow simulation modeling. Engine simulations using large eddy simulations (LES) are making large progress and the need for well documented velocimetry measurements for model validation is high. This work presents velocimetry measurements from PIV, high-speed PIV, stereoscopic PIV, and tomographic PIV to extensively describe the in-cylinder flow field in a motored optical engine operating at 800 RPM. These measurements also establish a comprehensive database designed for LES model development and validation. Details of the engine, engine accessory components, and well-controlled boundary conditions and engine operation are presented. The first two statistical moments of the flow field are computed and show excellent agreement among the PIV database. Analysis of statistical moments based on limited sample size is presented and is important for modeling validation purposes. High-speed PIV resolved the instantaneous flow field throughout entire engine cycles (i.e. 719 consecutive crank-angles), while tomographic PIV images are further used to investigate the 3D flow field and identify regions of strong vortical structures identified by the Q-criterion. Principle velocity gradient components are computed and emphasize the need to resolve similar spatial scales between experimental and modeling efforts for suitable model validation.     

Detached eddy simulation of cyclic large scale fluctuations in a simplified engine setup

Computational Fluid Dynamics using RANS-based modelling approaches have become an important tool in the internal combustion engine development and optimization process. However, these models cannot resolve cycle to cycle variations, which are an important aspect in the design of new combustion systems. In this study the feasibility of using a Detached Eddy Simulation (DES) SST model, which is a hybrid RANS/LES model, to predict cycle to cycle variations is investigated. In the near wall region or in regions where the grid resolution is not sufficiently fine to resolve smaller structures, the two-equation RANS SST model is used. In the other regions with higher grid resolution an LES model is applied. The case considered is a geometrically simplified engine, for which detailed experimental data for the ensemble averaged and single cycle velocity field are available from Boreé et al. [Boreé, J., Maurel, S., Bazile, R., 2002. Disruption of a compressed vortex, Physics of Fluids 14 (7), 2543–2556]. The fluid flow shows a strong tumbling motion, which is a major characteristic for modern turbo-charged, direct-injection gasoline engines. The general flow structure is analyzed first and the extent of the LES region and the amount of resolved fluctuations are discussed. Multiple consecutive cycles are computed and turbulent statistics of DES SST, URANS and the measured velocity field are compared for different piston positions. Cycle to cycle variations of the velocity field are analyzed for both computation and experiment with a special emphasis on the useability of the DES SST model to predict cyclic variations.     

A Study of the Organized in-Cylinder Motion by a Dynamic Adaptive Scale-Resolving Turbulence Model

The feasibility of using a scale-adaptive turbulence model for the simulation of the unsteady flow in Internal Combustion (IC) Engines is investigated. The original theory of the Dynamic Length-Scale Resolution Model (DLRM), which includes an adaptive rescaling procedure for the modeled turbulent length and time scales, has been trimmed and applied to the simulation of the flow field of a geometrically simplified square-piston engine working under motored conditions. The flow field exhibits a strong tumbling motion, which is a major characteristic of modern turbo-charged, direct-injection gasoline engines. Multiple consecutive cycles have been computed and turbulent statistics, including the ensemble mean and the variance of the velocity field, have been compared with both experimental measurements and simulation results predicted by URANS and conventional LES with identical numerical setups.     

Large-eddy Simulation of Motored Flow in a Two-valve Piston Engine: POD Analysis and Cycle-to-cycle Variations

Large-eddy simulation (LES) has been performed for a single-cylinder, two-valve, four-stroke-cycle piston engine through 70 consecutive motored cycles. Initial comparisons of ensemble-averaged velocity fields have been made between LES and experiment, and proper orthogonal decomposition (POD) has been used to analyze the complex in-cylinder turbulent flows. Convergence of POD modes has been quantified, several POD variants have been explored, and sensitivity of results to analyzing different subsets of engine cycles has been studied. In general, it has been found that conclusions that were drawn earlier from POD analysis of a simplified non-compressing piston-cylinder assembly with a fixed valve carry over to the much more complex flow in this motored four-stroke-cycle engine. For the cases that have been examined, the first POD mode essentially corresponds to the ensemble-averaged mean velocity. The number of engine cycles required to extract converged POD modes increases with mode number, and varies with phase (piston position). There is little change in the lower-order phase-invariant POD modes when as few as 24 phases per cycle (30° between samples) are used, and complex 3-D time-dependent in-cylinder velocity fields through full engine cycles can be reconstructed using a relatively small number of POD modes. Quantification of cycle-to-cycle variations and insight into in-cylinder flow dynamics can be extracted through analysis of phase-invariant POD modes and coefficients.     

Numerical Evaluation of Combustion Regimes in a GDI Engine

There is significant interest in the gasoline direct-injection engine due to its potential for improvements in fuel consumption but it still remains an area of active research due to a number of challenges including the effect of cycle-by-cycle variations. The current paper presents the use of a 3D-CFD model using both the RANS and LES turbulence modelling approaches, and a Lagrangian DDM to model an early fuel injection event, to evaluate the regimes of combustion in a gasoline direct-injection engine. The velocity fluctuations were investigated as an average value across the cylinder and in the region between the spark plug electrodes. The velocity fluctuations near the spark plug electrodes were seen to be of lower magnitude than the globally averaged fluctuations but exhibited higher levels of cyclic variation due to the influence of the spark plug electrode and the pent-roof geometry on the in-cylinder flow field. Differences in the predicted flame structure due to differences in the predicted velocity fluctuations between RANS and LES modelling approaches were seen as a consequence of the inherently higher dissipation levels present in the RANS methodology. The increased cyclic variation in velocity fluctuations near the spark plug electrodes in the LES predictions suggested significant variation in the relative strength of the in-cylinder turbulence and that may subsequently result in a thickening of the propagating flame front from cycle-to-cycle in this region. Throughout this paper, the numerical results were validated against published experimental data of the same engine geometry under investigation.     

The development of high-speed particle image velocimetry (20 kHz) for large eddy simulation code refinement in bluff body flows

Flow interaction with a bluff body generates a highly complex flow field and has been the subject of much experimental and theoretical analysis. It has been shown that large eddy simulation (LES) modelling provides a more realistic analysis of the flow for such situations where the large scales of turbulence must be resolved. The inherent small-scale spatial velocity averaging in particle image velocimetry (PIV) is commensurate with the sub-grid scale modelling of LES and, therefore, offers potential as a code refinement technique. To demonstrate this potential, however, PIV must be performed with a temporal resolution of typically kHz and a spatial resolution of sub-mm² to be relevant for the vast majority of flows of engineering interest. This paper reports the development of a high-speed PIV system capable of operating at 20 kHz with a spatial resolution of 0.9 mm². This is the combined highest speed, highest resolution PIV data reported to date. The experiment chosen to demonstrate the system is the study of the steady flow interaction with circular and square cross-section obstacles. A Reynolds number of 3,900 is chosen for the cylinder flow to extend the database used by Breuer M. (1998 Int J Heat Fluid 19:512–521) in his extensive LES modelling of this flow. Data presented include a sequence of two-dimensional velocity and vorticity fields, including flow streamlines. Importantly, the random error, inherent in a PIV measurement, is discussed and a formula presented which allows the error to be estimated and regions of the flow identified where LES comparisons would be uncertain.     

Assessment of LES Subgrid-scale Models and Investigation of Hydrodynamic Behaviour for an Axisymmetrical Bluff Body Flow

This work is concerned with the investigation of fluid-mechanical behaviour and the performance of different subgrid-scale models for LES in the numerical prediction of a confined axisymmetrical bluff-body flow. Four subgrid-scale turbulence models comprising the Smagorinsky model, Dynamic Smagorinsky model, WALE model and subgrid turbulent kinetic energy model, are validated and compared directly against the experimental data. Two different mesh counts are used for the LES studies, one with a higher mesh resolution in the shear layer than the other. It is found that increasing the mesh resolution improves the time-averaged fluctuating velocity profiles, but has less effect on the time-averaged filtered velocity profiles. A comparison against experiment shows that the recirculation zone length is well predicted using LES. The accuracy of the four different subgrid scale models is then assessed by comparing the LES results using the dense mesh with the experiment. Comparisons with the time-averaged axial and radial velocity profiles demonstrate that LES displays good agreement with the experimental data, with the essential flow features captured both qualitative and quantitatively. The subgrid velocity also matches well with the experimental results, but a slight underprediction of the inner shear layer is observed for all subgrid models. In general, it is found that the Smagorinsky and WALE models are more dissipative than the Dynamic Smagorinsky model and subgrid TKE model. Comparison of the spectra against the experiment shows that LES can capture dominant features of the turbulent flow with reasonable accuracy, and weak spectral peaks related to the Kevin-Helmholtz instability and helical vortex shedding are present.     

On the Evolution of the Flow Field in a Spark Ignition Engine

The development of the turbulent flow field inside a spark ignition engine is examined by large-eddy simulation (LES), from the intake flow to the tumble break-down. Ten consecutive cold flow engine cycles on a coarse and twenty cycles on a fine grid are simulated and compared to experiments of the same engine. The turbulent subgrid scales are modeled by the standard Smagorinsky and by the recently developed Sigma model. A comparison of the intake flow is made against Particle Image Velocimetry (PIV) measurements along horizontal and vertical lines and to an LES simulation performed by the Darmstadt group. Furthermore, we show the first LES comparison to Magnetic Resonance Velocimetry (MRV conducted by Freudenhammer et al.) measurements, which provided the 3D flow field inside a full scale dummy of the entire upper cylinder head including the valve seat region, at a time which mimics inflow conditions of the corresponding engine. Our LES is in good qualitative and quantitative agreement with the simulation and the experiments, with the notable exception of the measured in-cylinder pressure, which is discussed in detail and compared to 0D simulations and simulations from other groups. A criterion is proposed for estimating the number of cycles needed in a simulation, if experimental data is available. We put emphasis on the flow in the valve seat region, where turbulence is generated, and discuss the formation of the large scale tumble motion, including a comparison of the radial velocity fields on rolled-up planes around the valve seat. Here, spots of high velocities were found in the under flow region, which cannot been seen by the ensemble averaged MRV measurement. Within the compression stroke, a 2D vortex center identification algorithm is applied on slices inside the combustion chamber, yielding a 3D visualization of the tumble vortex, which is found to have a “croissant-like” shape. The tumble vortex trajectory is plotted on the symmetry plane and compared to measurements. Finally, we consider a modified definition of the (turbulent) integral length scale that provided further insight to the tumble break-down process.     

Implicit large eddy simulation of a supersonic flat-plate boundary layer flow by weighted compact nonlinear scheme

The performance of implicit large eddy simulation (ILES) of a supersonic flat-plate turbulent boundary layer flow by weighted compact nonlinear scheme (WCNS) has been investigated. In view of features of WCNS and ILES, it was expected that ILES by WCNS could be an efficient approach to perform LES of supersonic turbulent flows. The flowfield calculated by WCNS was of lower turbulent intensity compared with an explicit LES data obtained by a numerical scheme of the same order of accuracy on a computational grid of similar resolution. It was concluded that the numerical dissipation inherent in WCNS is so large that applying WCNS to ILES of this flowfield is inefficient compared with explicit LES.     

Analysis of In-cylinder Flow Field Anisotropy in IC Engine using Large Eddy Simulation

The objective of the analysis presented in this work is to investigate the distribution of the anisotropy invariants of the in-cylinder flow to examine the characteristics of the turbulence during an engine cycle and its interactions with the in-cylinder coherent structures. For this purpose, both the results of Large Eddy Simulations of the flow in an internal combustion engine under motored conditions and measurements obtained in the same configuration have been employed. Investigations include an analysis of the in-cylinder flow properties in the engine’s cross plane by means of first and second order statistical moments. Additionally, the behaviour of the flows anisotropy tensor has been analysed. It has been observed that during the intake stroke the in-cylinder turbulence shows a rather anisotropic structure, whereas during the compression stroke it tends to be more isotropic. Furthermore, the degree of anisotropy increases again in correspondence of vortex breakdown, during which particularly high velocity fluctuations have been identified also by experimental investigation.     

Proper orthogonal decomposition analysis for cycle-to-cycle variations of engine flow. Effect of a control device in an inlet pipe

This paper aims to investigate cycle-to-cycle variations of non-reacting flow inside a motored single-cylinder transparent engine in order to judge the insertion amplitude of a control device able to displace linearly inside the inlet pipe. Three positions corresponding to three insertion amplitudes are implemented to modify the main aerodynamic properties from one cycle to the next. Numerous particle image velocimetry (PIV) two-dimensional velocity fields following cycle database are post-treated to discriminate specific contributions of the fluctuating flow. We performed a multiple snapshot proper orthogonal decomposition (POD) in the tumble plane of a pent roof SI engine. The analytical process consists of a triple decomposition for each instantaneous velocity field into three distinctive parts named mean part, coherent part and turbulent part. The 3rd- and 4th-centered statistical moments of the proper orthogonal decomposition (POD)-filtered velocity field as well as the probability density function of the PIV realizations proved that the POD extracts different behaviors of the flow. Especially, the cyclic variability is assumed to be contained essentially in the coherent part. Thus, the cycle-to-cycle variations of the engine flows might be provided from the corresponding POD temporal coefficients. It has been shown that the in-cylinder aerodynamic dispersions can be adapted and monitored by controlling the insertion depth of the control instrument inside the inlet pipe.     

Numerical investigation of flow structures around a sphere

A numerical investigation of flow around a sphere is performed and compared with previous studies. Here, a second-order accurate, finite volume method is used in order to predict the instantaneous and time-averaged flow characteristics using large eddy simulation (LES) on the multi-block grid system. Namely, the objectives of this article are: (i) the presentation of flow structures in the wake region downstream of the sphere with a wide variety of flow properties such as the distribution of velocity vectors, patterns of streamlines, Reynolds stress correlations, root mean square of velocity components and other time-averaged flow data in order to reveal the vortical flow structures in detail and (ii) to demonstrate the abilities of computational methods in simulation of vortical flow data. Finally, it has been concluded that there are good agreements between the experimental results and numerical predictions.     

Mesh adaptation for large‐eddy simulations in complex geometries

Large‐eddy simulation (LES) consists in explicitly simulating the large scales of the fluid motion and in modeling the influence of the smallest scales. Thanks to the steady growth of computational resources, LES can now be used to simulate realistic systems with complex geometries. However, when LES is used in such complex geometries, an adequate mesh has to be determined to perform valid LES. In this work, a strategy is proposed to assess the quality of a given mesh and to adapt it locally. Two different criteria are used as mesh adaptation criteria. The first criterion is defined to ensure a correct discretization of the mean field, whereas the second criterion is defined to ensure enough explicit resolution of turbulent scales motions. The use of both criteria is shown in canonical flow cases. As a second part of this work, a numerical strategy for mesh adaptation in high‐performance computing context is proposed by coupling the flow solver, YALES2, and the remeshing library, MMG3D, for massively parallel computations. This coupling enables an efficient and parallel remeshing of grids alleviating any memory or performance issues encountered in sequential tools. This strategy is finally applied to the simulation of the isothermal flow in a complex meso‐combustor to demonstrate the applicability of the adaptation methodology to complex turbulent flows. Copyright © 2015 John Wiley & Sons, Ltd.     

Numerical and experimental investigation on droplet dynamics and dispersion of a jet engine injector

In the case of turbine combustors operating with liquid fuel the combustion process is governed by the liquid fuel atomization and its dispersion in the combustion chamber. By highly unsteady flow field conditions the transient interaction between the liquid and the gaseous phase is of interest, because it results in a temporal variation of air–fuel ratio which leads to a fluctuating temperature distribution. The objective of this research was the investigation of transient flow field phenomena (e.g. large coherent structures) on droplet dynamics and dispersion of an isothermal flow (of inert water droplets) as a necessary first step towards a full analysis of spray combustion in real-life devices. The advanced injector system for lean jet engine combustors PERM (Partial Evaporated Rapid Mixing) was applied, generating a dilute polydispersed spray in a swirled flow field. Experiments were performed using Phase Doppler Anemometry (PDA) and a patternator to determine the droplet polydispersity, concentration maps, and velocity profiles in the flow. An important finding is the effect of large-scale coherent structures due mainly to the precessing of the vortex core (PVC) of the swirling air jet on the particle dispersion patterns. The experimental results then serve as reference data to assess the accuracy of the Eulerian–Lagrangian computations using a Large Eddy Simulation (LES), a Unsteady Reynolds-Average Navier–Stokes Simulation (URANS) and two simplified (steady-state) simulations. There, a simplified droplet injection model was used and the required boundary conditions of injected droplet sizes were obtained from measurements. Important transient effects of deterministic droplet separation observed during experiments, could be perfectly replicated with this injection model. It is convincingly shown, through extensive computations, that the resolution of instantaneous vortical structures is indeed crucial; hence the LES, or a reasonably-well resolved URANS are preferred over the steady-state solutions with additional, stochastic-type, turbulent dispersion models.     

Analysis of high-order velocity moments in a strained channel flow

In the current study, model expressions for fifth-order velocity moments obtained from the truncated Gram-Charlier series expansions model for a turbulent flow field probability density function are validated using data from direct numerical simulation (DNS) of a planar turbulent flow in a strained channel. Simplicity of the model expressions, the lack of unknown coefficients, and their applicability to non-Gaussian turbulent flows make this approach attractive to use for closing turbulent models based on the Reynolds-averaged Navier-Stokes equations. The study confirms validity of the model expressions. It also shows that the imposed flow deformation improves an agreement between the model and DNS profiles for the fifth-order moments in the flow buffer zone including when the flow reverses its direction. The study reveals sensitivity of particularly odd velocity moments to the grid resolution. A new length scale is proposed as a criterion for the grid generation near the wall and in the other flow areas dominated by high mean velocity gradients when higher-order statistics have to be collected from DNS.     

Experimental study of an impinging jet with different swirl rates

A stereo PIV technique using advanced pre- and post-processing algorithms is implemented for the experimental study of the local structure of turbulent swirling impinging jets. The main emphasis of the present work is the analysis of the influence of swirl rate on the flow structure. During measurements, the Reynolds number was 8900, the nozzle-to-plate distance was equal to three nozzle diameters and the swirl rate was varied from 0 to 1.0. For the studied flows, spatial distributions of the mean velocity and statistical moments (including triple moments) of turbulent pulsations were measured.

The influence of the PIV finite spatial resolution on the measured dissipation rate and velocity moments was analyzed and compared with theoretical predictions. For this purpose, a special series of 2D PIV measurements was carried out with vector spacing up to several Kolmogorov lengthscales.

All terms of the axial mean momentum and the turbulent kinetic energy budget equations were obtained for the cross-section located one nozzle diameter from the impinging plate. For the TKE budget, the dissipation term was directly calculated from the instantaneous velocity fields, thereby allowing the pressure diffusion term to be found as a residual one. It was found that the magnitude of pressure diffusion decreased with the growth of the swirl rate. In general, the studied swirling impinging jets had a greater spread rate and a more rapid decay in absolute velocity when compared to the non-swirling jet.     

Construction of three-dimensional flow structure out of two-dimensional steady flow field velocity measurements

 Two-dimensional particle image velocimetry (PIV) is used to obtain a set of parallel vector maps in spanwise direction over the delta wing configuration ELAC. The out-of-plane velocity component is then constructed by application of continuity equation. This yields the whole three-dimensional separated flow field over the leeward side of the model. The spatial resolution of the measurements enables a detailed examination of the three-dimensional flow structure. The growth and the helical structure of primary vortex as well as smaller flow structures caused by secondary separation can be observed. Accuracy of the constructed velocity component is estimated with help of a numerically obtained three-dimensional dataset of the flow field around this configuration. The reconstruction procedure was applied to this data set taking the experimental uncertainty and the grid spacing of the PIV measurements into consideration. A comparison of reconstructed out-of-plane component and data of the numerical solution of Navier–Stokes equations results in a promising low error. A statistical analysis of different procedures allows interpretation of reconstruction capabilities. Received: 15 April 1998 / Accepted: 15 September 1998     

Experimental study and large eddy simulation for the turbulent flow around four cylinders in an in-line square configuration

The turbulent flows around four cylinders in an in-line square configuration with different spacing ratios of 1.5, 2.5, 3.5 and 5.0 have been investigated experimentally at subcritical Reynolds numbers from 11,000 to 20,000. The mean and fluctuating velocity distributions were obtained using the laser Doppler anemometry (LDA) measurement. The digital particle image velocimetry (DPIV) was employed to characterize the full field vorticity and velocity distributions as well as other turbulent quantities. The experimental study indicated that several distinct flow patterns exist depending on the spacing ratio and subcritical Reynolds number for turbulent flow. The three-dimensional numerical simulations were also carried out using the large eddy simulation (LES) at Reynolds number of 15,000 with the spacing ratio of 1.5 and 3.5. The results show that the LES numerical predictions are in good agreement with the experimental measurements. Therefore, the three-dimensional vortex structures and the full field instantaneous and mean quantities of the flow field such as velocity field, vorticity field, etc., which are very difficult to obtain experimentally, can be extracted from the simulation results for the deepening of our understanding on the complex flow phenomena around four cylinders in in-line configuration.     

A Hybrid-Filter Approach to Turbulence Simulation

Germano (Theor Comput Fluid Dyn 17:225–331, 2004) proposed a hybrid-filter approach, which additively combines an LES-like filter operator (F) and a RANS-like statistical operator (E) using a blending function k: H?=?kF?+?(1???k)E. Using turbulent channel flow as an example, we first conducted a priori tests in order to gain some insights into this hybrid-filter approach, and then performed full simulations to further assess the approach in actual simulations. For a priori tests, two separate simulations, RANS (E) and LES (F), were performed using the same grid in order to construct a hybrid-filtered field (H). It was shown that the extra terms arising out of the hybrid-filtered Navier–Stokes (HFNS) equations provided additional energy transfer from the RANS region to the LES region, thus alleviating the need for the ad hoc forcing term that has been used by some investigators. The complexity of the governing equations necessitated several modifications in order to render it suitable for a full numerical simulation. Despite some issues associated with the numerical implementation, good results were obtained for the mean velocity and skin friction coefficient. The mean velocity profile did not have an overshoot in the logarithmic region for most blending functions, confirming that proper energy transfer from the RANS to the LES region was a key to successful hybrid models. It is shown that Germano’s hybrid-filter approach is a viable and mathematically more appealing approach to simulate high Reynolds number turbulent flows.