Analysis of Mixing Models for use in Simulations of Turbulent Spray Combustion

The present study concerns the investigation of different mixing models for use in the transported probability density function (PDF) modeling of turbulent (reacting) spray flows. The modeling of the turbulent mixing and other characteristic scalar variables such as gas enthalpy using transported (joint) PDFs has become an important method to describe turbulent (reacting) spray flows since the evaporation process causes the PDF of the mixture fraction to deviate from the widely used β function, which is typically used in models for turbulent gas flows. In the PDF transport equation, the molecular mixing does not appear in closed form so that modeling strategies are required. For gas combustion, the interaction-by-exchange-with-the-mean (IEM) model, the modified Curl (MC) model, and the Euclidean minimum spanning tree (EMST) models are used. More recently, a new mixing model, the PSP model, which is based on parameterized scalar profiles has been developed. The present study focuses on the use and analysis of the IEM, MC and PSP models for turbulent spray flames. For this purpose, the models are reconsidered with respect to the evaporation process that must be included and evaluated if spray combustion is considered. For model evaluation, turbulent ethanol/air spray flames are simulated, and the results are compared to experimental data by A. Masri, University of Sydney, Australia.     

Presumed Mapping Functions for Eulerian Modelling of Turbulent Mixing

The mapping closure of Chen et al. [Phys. Rev. Lett., 63, 1989] is a transported probability density function (PDF) method that has proven very efficient for modelling of turbulent mixing in homogeneous turbulence. By utilizing a Gaussian reference field, the solution to the mapping function (in homogeneous turbulence) can be found analytically for a range of initial conditions common for turbulent combustion applications, e.g. for binary or trinary mixing. The purpose of this paper is to investigate the possibility of making this solution a presumed mapping function (PMF) for inhomogeneous flows. The PMF in turn will imply a presumed mixture fraction PDF that can be used for a wide range of models in turbulent combustion, e.g. flamelet models, the conditional moment closure (CMC) or large eddy simulations. The true novelty of the paper, though, is in the derivation of highly efficient, closed algebraic expressions for several existing models of conditional statistics, e.g. for the conditional scalar dissipation/diffusion rate or the conditional mean velocity. The closed form expressions nearly eliminates the overhead computational cost that usually is associated with nonlinear models for conditional statistics. In this respect it is argued that the PMF is particularly well suited for CMC that relies heavily on manipulations of the PDF for consistency. The accuracy of the PMF approach is shown with comparison to DNS of a single scalar mixing layer to be better than for the β-PDF. Not only in the shape of the PDF itself, but also for all conditional statistics models computed from the PDF.     

On the Relationship Between the Statistics of the Resolved and True Rate of Dissipation of Mixture Fraction

The relationship between the one-point probability-density-function (PDF) of the dissipation rate of mixture fraction fluctuations and the corresponding resolved quantity available in large eddy simulation (LES) is analyzed. The investigation pursues two fronts: an a priori study using direct numerical simulation (DNS), and an analytic development that, using common turbulence physics simplifications, relates the one-point statistics of the resolved and true scalar dissipations. Particularly, the analysis reveals the connection between the multi-point correlations of the mixture fraction gradient and the one-point PDF of the resolved scalar dissipation. A DNS of a temporally evolving shear layer with and without heat release is used to quantify the accuracy of the analytical result. It is verified, both by filtering the DNS and from the theory, that increasing the filter cutoff width reduces the magnitude of the resolved scalar dissipation fluctuations, as expected and observed experimentally. Comparison with DNS indicates that the analytical relationship predicts the behavior of the resolved scalar dissipation PDF well at the center planes of the shear layer, where turbulence is locally more isotropic and homogeneous. Large-scale anisotropy and inhomogeneities in the DNS degrade the accuracy of the approximate analytical result close to the edges of the shear layer. These results may be improved with future investigations to account fully for the missing statistics in LES, which have the potential to allow a more accurate quantification of finite-rate chemistry effects in reacting flows.     

Joint PDF Closure of Turbulent Premixed Flames

In this paper, a novel model for turbulent premixed combustion in the corrugated flamelet regime is presented, which is based on transporting a joint probability density function (PDF) of velocity, turbulence frequency and a scalar vector. Due to the high dimensionality of the corresponding sample space, the PDF equation is solved with a Monte-Carlo method, where individual fluid elements are represented by computational particles. Unlike in most other PDF methods, the source term not only describes reaction rates, but accounts for “ignition” of reactive unburnt fluid elements due to propagating embedded quasi laminar flames within a turbulent flame brush. Unperturbed embedded flame structures and a constant laminar flame speed (as expected in the corrugated flamelet regime) are assumed. The probability for an individual particle to “ignite” during a time step is calculated based on an estimate of the mean flame surface density (FSD), latter gets transported by the PDF method. Whereas this model concept has recently been published [21], here, a new model to account for local production and dissipation of the FSD is proposed. The following particle properties are introduced: a flag indicating whether a particle represents the unburnt mixture; a flame residence time, which allows to resolve the embedded quasi laminar flame structure; and a flag indicating whether the flame residence time lies within a specified range. Latter is used to transport the FSD, but to account for flame stretching, curvature effects, collapse and cusp formation, a mixing model for the residence time is employed. The same mixing model also accounts for molecular mixing of the products with a co-flow. To validate the proposed PDF model, simulation results of three piloted methane-air Bunsen flames are compared with experimental data and very good agreement is observed.     

Joint Scalar versus Joint Velocity-Scalar PDF Simulations of Bluff-Body Stabilized Flames with REDIM

Two transported PDF strategies, joint velocity-scalar PDF (JVSPDF) and joint scalar PDF (JSPDF), are investigated for bluff-body stabilized jet-type turbulent diffusion flames with a variable degree of turbulence–chemistry interaction. Chemistry is modeled by means of the novel reaction-diffusion manifold (REDIM) technique. A detailed chemistry mechanism is reduced, including diffusion effects, with N ₂ and CO ₂ mass fractions as reduced coordinates. The second-moment closure RANS turbulence model and the modified Curl’s micro-mixing model are not varied. Radiative heat loss effects are ignored. The results for mean velocity and velocity fluctuations in physical space are very similar for both PDF methods. They agree well with experimental data up to the neck zone. Each of the two PDF approaches implies a different closure for the velocity-scalar correlation. This leads to differences in the radial profiles in physical space of mean scalars and mixture fraction variance, due to different scalar flux modeling. Differences are visible in mean mixture fraction and mean temperature, as well as in mixture fraction variance. In principle, the JVSPDF simulations can be closer to physical reality, as a differential model is implied for the scalar fluxes, whereas the gradient diffusion hypothesis is implied in JSPDF simulations. Yet, in JSPDF simulations, turbulent diffusion can be tuned by means of the turbulent Schmidt number. In the neck zone, where the turbulent flow field results deteriorate, the joint scalar PDF results are in somewhat better agreement with experimental data, for the test cases considered. In composition space, where results are reported as scatter plots, differences between the two PDF strategies are small in the calculations at hand, with a little more local extinction in the joint scalar PDF results.     

Direct Numerical Simulations of Localised Forced Ignition in Turbulent Mixing Layers: The Effects of Mixture Fraction and Its Gradient

The effects of mixture fraction value ξ and the magnitude of its gradient |∇ξ| at the ignitor location on the localised forced ignition of turbulent mixing layers under decaying turbulence is studied based on three-dimensional compressible Direct Numerical Simulations (DNS) with simplified chemistry. The localised ignition is accounted for by a spatial Gaussian power distribution in the energy transport equation, which deposits energy over a prescribed period of time. In successful ignitions, it is observed that the flame shows a tribrachial structure. The reaction rate is found to be greater in the fuel rich side than in stoichiometric and fuel-lean mixtures. Placing the ignitor at a fuel-lean region may initiate ignition, but extinction may eventually occur if the diffusion of heat from the hot gas kernel overcomes the heat release due to combustion. It is demonstrated that ignition in the fuel lean region may fail for an energy input for which self-sustained combustion has been achieved in the cases of igniting at stoichiometric and fuel-rich locations. It is also found that the fuel reaction rate magnitude is negatively correlated with density-weighted scalar dissipation rate in the most reactive region. An increase in the initial mixture fraction gradient at the ignition centre for the ignitor placed at stoichiometric mixture decreases the spreading of the burned region along the stoichiometric mixture fraction isosurface. By contrast, the mass of the burned region increases with an increase in the initial mixture fraction gradient at the ignition location, as for a given ignition kernel size the thinner mixing layer includes more fuel-rich mixture, which eventually makes the overall burning rate greater than that compared to a thicker mixing layer where relatively a smaller amount of fuel-rich mixture is engulfed within the hot gas kernel. Submitted as a full-length article to Flow Turbulence and Combustion.     

湍流被动标量研究的最新进展

被动标量场的统计性质, 在湍流理论以及湍流燃烧、污染物防治等工程领域都有非常重要的意义. 最近十几年, 大量的实验测量、数值模拟和理论分析结果表明, 标量场具有很多自身独特的性质, 甚至有些性质并不依赖于速度场本身. 这就促使人们对传统经典理论进行重新认识、修正或者提出新的理论来取代. 本文对标量场的各向异性、标量和标量耗散率的概率密度函数(PDF)、标量场的空间结构及演化过程、还有标量小尺度混合模型等几个方面进行综述.      

条件矩模型模拟湍流扩散燃烧

对条件矩模型模拟湍流扩散燃烧进行了初步的研究.在条件矩模型中,标量的统计是以混合分数为条件的,条件平均使得非线性化学反应源项可以在一阶条件下被封闭.模拟结果和美国Sandia国家实验室的实验结果对比表明:对温度、主要组分浓度的预报结果是令人满意的,NO浓度的预报在趋势上也符合实验结果.误差分析表明,提高标量耗散率的预报精度和二阶条件矩模型都将有助于推动条件矩模型的发展.     

A Lagrangian-based flame index for the transported probability density function method

We propose a new flame index for the transported probability density function(PDF) method. The flame index uses mixing flux projections of Lagrangian particles on mixture fraction and progress variable directions as the metrics to identify the combustion mode, with the Burke-Schumann solution as a reference. A priori validation of the flame index is conducted with a series of constructed turbulent partially premixed reactors. It indicates that the proposed flame index is able to identify the combustion mode based on the subgrid mixing information. The flame index is then applied the large eddy simulation/PDF datasets of turbulent partially premixed jet flames. Results show that the flame index separate different combustion modes and extinction correctly. The proposed flame index provides a promising tool to analyze and model the partially premixed flames adaptively.     

A Comprehensive Study of Effects of Mixing and Chemical Kinetic Models on Predictions of n-heptane Jet Ignitions with the PDF Method

The composition Probability Density Function (PDF) model is coupled with a Reynolds-averaged k???ε turbulence model and three computationally efficient, yet widely used chemical mechanisms to simulate transient n-heptane spray injection and ignition in a high temperature and high density ambient fluid. Molecular diffusion is modelled by three mixing models, namely the interaction by exchange with the mean (IEM), modified Curl (MC) and Euclidean minimum spanning trees (EMST) models. The liquid phase is modelled by a discrete phase model (DPM). This represents among the first applications of the PDF method in practical diesel engine conditions. A non-reacting case is first considered, with the focus on the ability of the model to capture the spray structure, e.g., vapour penetration and liquid length, fuel mixture fraction and its variance. Reacting cases are then investigated to compare and evaluate the three different chemical mechanisms and the three mixing models. It is concluded that the EMST mixing model in conjunction with a reduced chemical kinetic model (Lu et al., Combust Flame 156(8):1542–1551, 2009) performs the best among the options considered. The sensitivity of the results to the choice of the mixing constant is also studied to understand its effect on the flame ignition and stabilisation. Finally, the PDF model is compared to a well-mixed model that assumes turbulent fluctuations are negligible, which has been widely used in the diesel spray combustion community. Significant structural differences in the modelled flame are revealed comparing the PDF method with the well-mixed model. Quantitatively, the PDF model exhibits excellent agreement with the measurements and shows much better results than the well-mixed model in all ambient O₂ and temperature conditions.     

LES/PDF Modeling of Turbulent Premixed Flames with Locally Enhanced Mixing by Reaction

Large-eddy simulations (LES) combined with the transported probability density function (PDF) method are carried out for two turbulent piloted premixed methane-air jet flames (flame F1 and flame F3) to assess the capability of LES/PDF for turbulent premixed combustion. The conventionally used model for the sub-filter scale mixing time-scale (or the mixing frequency) fails to capture the premixed flames correctly. This failure is expected to be caused by the lack of the sub-filter scale premixed flame propagation property in the sub-filter scale mixing process when the local flame front is under-resolved. It leads to slower turbulent premixed flame propagation and wider flame front. A new model for specifying the sub-filter scale mixing frequency is developed to account for the effect of sub-filter scale chemical reaction on mixing, based on past development of models for the sub-filter scale scalar dissipation rate in premixed combustion. The new model is assessed in the two turbulent premixed jet flames F1 and F3. Parametric studies are performed to examine the new model and its sensitivity when combined with the different mixing models. Significantly improved performance of the new mixing frequency model is observed to capture the premixed flame propagation reasonably, when compared with the conventional model. The sensitivity of the flame predictions is found be relatively weak to the different mixing models in conjunction with the new mixing frequency model.     

Measurements of Scalar Variance, Scalar Dissipation, and Length Scales in Turbulent Piloted Methane/Air Jet Flames

One-dimensional (line) measurements of mixture fraction, temperature, and scalar dissipation in piloted turbulent partially premixed methane/air jet flames (Sandia flames C, D, and E) are presented. The experimental facility combines line imaging of Raman scattering, Rayleigh scattering, and laser-induced CO fluorescence. Simultaneous single-shot measurements of temperature and the mass fractions of all the major species (N₂, O₂, CH₄, CO₂, H₂O, CO, and H₂) are obtained along 7 mm segments with a nominal spatial resolution of 0.2 mm. Mixture fraction, ξ, is then calculated from the measured mass fractions. Ensembles of instantaneous mixture fraction profiles at several streamwise locations are analyzed to quantify the effect of spatial averaging on the Favre average scalar variance, which is an important term in the modeling of turbulent nonpremixed flames. Results suggest that the fully resolved scalar variance may be estimated by simple extrapolation of spatially filtered measurements. Differentiation of the instantaneous mixture fraction profiles yields the radial contribution to the scalar dissipation, χ _r = 2D _ξ(?ξ/?r)², and radial profiles of the Favre mean and rms scalar dissipation are reported. Scalar length scales, based on autocorrelation of the spatial profiles of ξ and χ _r , are also reported. These new data on this already well-documented series of flames should be useful in the context of validating models for sub-grid scalar variance and for scalar dissipation in turbulent flames.     

Stochastic Shell Model for Turbulent Mixing of Multiple Scalars with Mean Gradients and Differential Diffusion

In this paper, we develop a shell model for the velocity and scalar concentrations that, by design, is consistent with the eddy damped quasi-normal Markovian (EDQNM) model for multiple mixing scalars. We review the realizable form of the EDQNM model derived by Ulitsky and Collins (J Fluid Mech 412:303–329, 2000), which forms the basis for the shell model. The equations governing the velocity and scalar within each shell are stochastic ordinary differential equations with drift and diffusion terms chosen so that the velocity variance, velocity–scalar cross correlations, and scalar–scalar cross correlations within each shell precisely match the EDQNM model predictions. Consequently, shell averages can be thought of as a representation of the discrete three-dimensional spectrum. An advantage the shell model has over the original EDQNM equations is that the sum of each realization over the shells is a model for the fine-grained, joint velocity/scalar probability density function (PDF). Indeed, this provides some of the motivation for the development of the model. We cannot exploit this feature in the present study of the mixing of two scalars with uniform mean gradients, as the PDF is a joint Gaussian throughout (and hence the correlation matrix completely defines the distribution). The model is capable of predicting Lagrangian correlation functions for the scalar, scalar dissipation and velocity. We find the predictions of the model are in good qualitative agreement with direct numerical simulations by Yeung (J Fluid Mech 427:241–274, 2001). Eventually we will apply the shell model to scalars that are initially highly non-Gaussian (e.g., double delta function) and observe the relaxation towards a Gaussian. As the shell model contains information on the spectral distribution of the scalar field, the relaxation rate will depend upon the length and time scales of the turbulence and the scalar fields, as well as the molecular diffusivities of the species. The full capabilities of the PDF predictions of the model will be the subject of a future publication.     

Pre-Chamber Ignition Mechanism: Simulations of Transient Autoignition in a Mixing Layer Between Reactants and Partially-Burnt Products

The structure of autoignition in a mixing layer between fully-burnt or partially-burnt combustion products from a methane-air flame at ? = 0.85 and a methane-air mixture of a leaner equivalence ratio has been studied with transient diffusion flamelet calculations. This configuration is relevant to scavenged pre-chamber natural-gas engines, where the turbulent jet ejected from the pre-chamber may be quenched or may be composed of fully-burnt products. The degree of reaction in the jet fluid is described by a progress variable c (c = taking values 0.5, 0.8, and 1.0) and the mixing by a mixture fraction ξ (ξ = 1 in the jet fluid and 0 in the CH₄-air mixture to be ignited). At high scalar dissipation rates, N₀, ignition does not occur and a chemically-frozen steady-state condition emerges at long times. At scalar dissipation rates below a critical value, ignition occurs at a time that increases with N₀. The flame reaches the ξ = 0 boundary at a finite time that decreases with N₀. The results help identify overall timescales of the jet-ignition problem and suggest a methodology by which estimates of ignition times in real engines may be made.     

Modeling of the Joint PDF of a Scalar and Its Gradient in Turbulent Mixing

A joint probability distribution function of a conservative scalar (mixture fraction) and its gradient is predicted numerically. Statistical moments of this function are compared to their approximations, direct numerical simulation data, and also to the results obtained by simplified models for a conditional rate of scalar dissipation, the surface density function, and the one-point PDF of scalar fluctuation under homogeneous isotropic turbulence. The results allow to evaluate the performance of existing statistical micromixing models.     

Noise Correction and Length Scale Estimation for Scalar Dissipation Rate Measurements in Turbulent Partially Premixed Flames

A recently developed conditional sampling-based method for correcting noise effects in scalar dissipation rate measurements and for estimating the extent of resolution of the dissipation rate is employed to analyze the data obtained in turbulent partially premixed (Sandia) flames. The method uses conditional sampling to select instantaneous fully resolved local scalar fields, which are analyzed to determine the measurement noise and to correct the Favre mean, conditional, and conditionally filtered dissipation rates. The potentially under-resolved local scalar fields, also selected using conditional sampling, are corrected for noise and are analyzed to examine the extent of resolution. The error function is used as a model for the potentially under-resolved local scalar to evaluate the scalar dissipation length scales and the percentage of the dissipation resolved. The results show that the Favre mean dissipation rate, the mean dissipation rate conditional on the mixture fraction, and dissipation rate filtered conditionally on the mixture fraction generally are well resolved in the flames. Analyses of the dissipation rates filtered conditionally on the mixture fraction and temperature show that the length scale increases with temperature, due to lower dissipation rate and higher diffusivity. The dissipation rate is well resolved for temperatures above 1,300 K but is less resolved at lower temperatures, although the probability of very low temperature events is low. To fully resolve these rare events the sample spacing needs to be reduced by approximately one half. The present study further demonstrates the effectiveness of the new noise correction and length scale estimation method.     

A turbulent time scale based k–ε model for probability density function modeling of turbulence/chemistry interactions: Application to HCCI combustion

Homogenous Charge Compression Ignition (HCCI) engine technology is known as an alternative to reduce NO_x and particulate matter (PM) emissions. As shown by several experimental studies published in the literature, the ideally homogeneous mixture charge becomes stratified in composition and temperature, and turbulent mixing is found to play an important role in controlling the combustion progress. In a previous study, an IEM model (Interaction by Exchange with the Mean) has been used to describe the micromixing in a stochastic reactor model that simulates the HCCI process. The IEM model is a deterministic model, based on the principle that the scalar value approaches the mean value over the entire volume with a characteristic mixing time. In this previous model, the turbulent time scale was treated as a fixed parameter. The present study focuses on the development of a micro-mixing time model, in order to take into account the physical phenomena it stands for. For that purpose, a (k–ε) model is used to express this micro-mixing time model. The turbulence model used here is based on zero dimensional energy cascade applied during the compression and the expansion cycle; mean kinetic energy is converted to turbulent kinetic energy. Turbulent kinetic energy is converted to heat through viscous dissipation. Besides, in this study a relation to calculate the initial heterogeneities amplitude is proposed. The comparison of simulation results against experimental data shows overall satisfactory agreement at variable turbulent time scale.     

Direct Numerical Simulation of Inert Droplet Effects on Scalar Dissipation Rate in Turbulent Reacting and Non-Reacting Shear Layers

Three-dimensional direct numerical simulation has been performed to investigate the effects of inert evaporating droplets on scalar dissipation rate χ in temporally-developing turbulent reacting and non-reacting mixing layers with the Reynolds number based on the vorticity thickness up to 8000 and the number of traced Lagrangian droplets up to 10⁷. The detailed instantaneous field analysis and ensemble-averaged statistics reveal complex interactions among combustion, droplet dynamics and evaporation, all of which have a considerable influence on χ. The presence of inert evaporating droplets promotes χ in both non-reacting and reacting mixing layers. In the latter, combustion reduces χ, so when combustion is suppressed by evaporating droplets, χ is enhanced. The transport equation of χ has been analyzed to investigate the various effects on χ in detail. The terms in the equation contain explicitly the evaporation rate and its spatial derivative, acting as a sink and a source for χ, respectively. On the whole, the net effect of the evaporation-rate terms is to promote χ. However, the production and dissipation terms are the dominant source and sink terms, respectively.