Kang Y. Huh

Application of the elliptic conditional moment closure model to a two-dimensional nonpremixed methanol bluff-body flame

Abstract A turbulent nonpremixed flame of methanol stabilized on a bluff body is simulated by the conditional moment closure (CMC) model. Full spatial variation of the conditional quantities is taken into account for the elliptic flow field. Comparison has shown good agreement for the conditional averages of the temperature and major species concentrations. Overprediction of OH near the stoichiometric mixture fraction is attributed to inaccurate prediction of the conditional scalar dissipation rate. Discrepancy for CO and H 2 may be due to several possible reasons such as the chemical kinetic mechanism or differential diffusion, although not clear yet. The results of the CMC model are in better agreement with conditional measurements than those of the stationary laminar flamelet model (SLFM), which is a quasi-steady form of the CMC without the convection term. The measurement data and CMC results are also in qualitative agreement for minor radial dependence of the conditional intermediate species concentrations. There is almost no noticeable difference in the unconditional Favre mean predictions by the CMC and SLFM. Discrepancy between prediction and measurement for the unconditional Favre average is due to inaccurate prediction of the pdf and mixing field by the k-ϵ turbulence model.

A new angular discretization scheme of the finite volume method for 3-D radiative heat transfer in absorbing, emitting and anisotropically scattering media

Abstract This paper presents a new angular discretization scheme, FT n , of the finite volume method (FVM) in three-dimensional radiative heat transfer. The FT n FVM is applied to absorbing, emitting and anisotropically scattering media with variable optical thickness in a rectangular enclosure. Results show that the FT n FVM performs better than the discrete ordinate method (DOM) and the FVM with N θ × N φ uniform angular discretization except near the optically thick diffusion limit. The FT n FVM closely reproduces the reference solutions by the Monte Carlo method for different scattering phase functions and optical thicknesses. It also turns out that anisotropic scattering has significant influence on radiative heat transfer with a symmetric boundary condition in a moderate optical thickness range as well as with a nonsymmetric boundary condition.

Use of the conditional moment closure model to predict NO formation in a turbulent CH4/H2 flame over a bluff-body

The first-order conditional moment closure (CMC) model is applied to a CH4/H2 bluff-body flame with emphasis on NO prediction. The flow and mixing fields are calculated by assuming fast chemistry and a beta function pdf for mixture fraction. Reacting scalar fields are calculated by elliptic CMC formulation and the three different chemical kinetic mechanisms, Miller-Bowman, GRI Mech 2.11 and 3.0. Calculation results show good agreement with the measured conditional mean temperature and mass fractions of major species, although with some discrepancy on the fuel rich side. The predicted conditional mean OH mass fractions satisfy the partial equilibrium assumption for the fast shuffle reactions between H2 and O2. The predicted conditional mean CO mass fractions are in good agreement with the Two-Photon Laser Induced Fluorescence (TPLIF) data. The GRI Mech 2.11 and Miller-Bowman mechanism show reasonable agreement with the measurements of NO while the GRI Mech 3.0 results are about twice as high. Effects of radiative heat loss on NO formation are shown to be of no significance in the recirculation and neck zone of the bluff-body flame. The unconditional Favre mean temperature and species mass fractions are also in reasonable agreement with measurements.

Development of a coherent flamelet model for a spark-ignited turbulent premixed flame in a closed vessel

Three-dimensional calculations of turbulent combustion which include ignition, laminar and turbulent flame propagation and quenching at the wall are performed by the coherent flamelet model (CFM). The existing CFMs in the literature are tested and new forms are proposed. A mean stretch factor I0 is introduced to consider the stretch and curvature effects of turbulence. Quenching at the wall is simulated by the simple wall flux model (SWFM) of the flame surface density. Two forms of the flame production term, CFM-1 and CFM-2, are tested to show the predictive capability of the CFM for turbulent burning velocity. CFM-1 has the flame production term given by the average rate of strain proposed by Cant et al. [20], while CFM-2 has the production term proportional to the rms turbulent velocity. It turns out that the turbulent burning velocity of CFM-2 is in reasonable agreement with the data of Checkel and Thomas and Bradley’s correlation with variation of the Karlovitz number, rms turbulent velocity, and integral length scale.

Numerical simulation of spray autoignition by the first-order conditional moment closure model

The ignition delay time of a n -heptane spray is calculated by the first-order conditional moment closure (CMC) model. At each time step the mixing field is calculated with the spray models in KIVA3. The CMC equations are then solved by the fractional step method, which sequentially considers transport and reaction terms. The evaporation terms in the variance equation of mixture fraction are treated in three different ways: no source, the model by Holman and Gutheil, and the one-droplet model. They all show similar spatial distributions with differences in the ignition delay time within 0.5%. The evaporation terms in the CMC equations are also treated in three different ways: no source, as a boundary flux on the fuel side, and the one-droplet model. They do not have any noticeable influence on the conditional profiles near the stoichiometric mixture fraction, where most reactions occur. A parametric study is performed to investigate the influence of initial temperature, drop size distribution, spray angle, and injected fuel quantity. The computed ignition delay times show a reasonable comparison with the measurements under different initial temperatures.

Experimental investigation on cellular breakup of a planar liquid sheet from an air-blast nozzle

The cellular breakup phenomenon is investigated experimentally for a planar liquid sheet from an air-blast nozzle. The dominant sinuous wave growing spatially downstream forms complicated cellular structures of perforated thin films and surrounding ligaments. Several characteristic parameters are measured from photographic images and compared with linear temporal analysis. The dominant wavelength is proportional to the inverse square of the relative velocity between air and liquid. The estimated breakup time matches the growth time of the most unstable wave, while the breakup length corresponds to a product of breakup time and liquid velocity. Numerical simulation shows a substantially reduced mean effective velocity near flow reattachment region of the air stream. Air turbulence seems to play a major role on initial perturbations of cellular breakup in the given nozzle configuration. The measured spatial growth rates are always less than linear predictions due to deviation from the linear regime at higher amplitudes.

Modeling autoignition of a turbulent methane jet by the conditional moment closure model

Autoignition of a turbulent methane jet has been studied by the first-order conditional moment closure (CMC) model with the detailed chemical reaction mechanism GRI Mech 3.0. Methane was injected into hot air in a constant volume chamber under various initial temperatures and pressures. The flow and mixing field were calculated by the transient SIMPLE algorithm with the κ-e-g turbulence model. The CMC equations were solved by the fractional step method, which sequentially treats the transport and chemical reaction terms in each time step. The stiff ordinary differential equation solver was used for chemical reaction steps. The calculated ignition delays are in good agreement in both magnitudes and major trends of variation in the measurements. The ignition delay decreases significantly as the initial air temperature increases. The chamber pressure has only a minor effect on the ignition delay, which tends to decrease slightly at a higher ambient pressure. There is a slight decrease in the ignition delay of methane/ethane mixture as the fraction of ethane increases. Comparison with the homogeneous CMC, which ignores spatial variation of the conditional moments, shows that the spatial dependence should be taken into account for accurate prediction of the ignition delays. It is shown that autoignition occurs on the sides of a fuel jet, where the most reactive mixture fraction is combined with a low conditional mean scalar dissipation rate.

ASSESSMENT OF THE FINITE-VOLUME METHOD AND THE DISCRETE ORDINATE METHOD FOR RADIATIVE HEAT TRANSFER IN A THREE-DIMENSIONAL RECTANGULAR ENCLOSURE

The finite-volume method (FVM) and the discrete ordinate method (DOM) are implemented to assess their capability to predict radiative heat transfer in a three-dimensional enclosure. A varying optical thickness and a nonuniform temperature profile are assumed to reproduce a typical furnace. Results show that the FVMperforms better than the DOMin optically thin media, while they show comparable accuracy in optically thick media. The lower-order FVM and DOM may lead to erroneous results due to the ray effect in optically thin media with a highly nonuniform temperature profile. The 8 12 FVM and the S8 DOM have shown good accuracy in all the test cases in this article.

Measurement and analysis of flame surface density for turbulent premixed combustion on a nozzle-type burner

Abstract The flame surface density for turbulent premixed combustion on a nozzle-type burner is measured by planar laser-induced fluorescence (PLIF) and image processing techniques. The maximum flame surface density tends to show linear dependence on the K-factor given as a function of the integral length scale and u′0/SL. The flame surface density shows an asymmetric profile in c space with the peak location correlated in terms of the dimensionless parameter, NB, which represents the degree of gradient or countergradient diffusion by turbulence. At values of NB close to unity the peak occurs at a value of c of about 0.7. As NB increases above unity, the peak moves to a lower value in c space, approaching a symmetric profile. The thickness of a turbulent flame brush nondimensionalized by the integral length scale tends to show linear dependence on the H-factor which is obtained by integrating the first moment equation of the reaction progress variable. The flame surface density increases at a higher ambient pressure due to decrease in the laminar burning velocity and the length scales of flame wrinkling.

Second-order conditional moment closure modeling of local extinction and reignition in turbulent non-premixed hydrocarbon flames

A second-order conditional moment closure (CMC) model is applied to turbulent non-premixed hydrocarbon flames with significant local extinction and reignition. Combustion of hydrocarbon fuel is described by a systematically reduced two-step mechanism. Second-order correction is made to the conditional mean reaction rates by assumed probability density function (PDF) method and Taylor expansion method. In the assumed PDF method, the conditional joint PDF of the reaction progress variables is assumed to be the maximum entropy PDF. Second-order CMC predictions show good agreement with direct numerical simulation (DNS) data for all test cases, while first-order CMC tends to overpredict the intermediate species and reaction of fuel. Results show that the intermediate species is more sensitive to accuracy of the predicted reaction rates than the major species. The predicted conditional variances and convariances of the reaction progress variables turn out to be in good agreement with DNS results during an extinction process. The onset of reignition is predicted to occur too early due to a linear fit of chemical reaction rates in the Taylor expansion method. Although the assumed PDF method generally gives better results than the Taylor expansion method, the latter may be a promising, computationally inexpensive approach for engineering problems with more complex chemistry.