Direct numerical simulation of lean premixed CH4/air and H2/air flames at high Karlovitz numbers

Three-dimensional direct numerical simulation with detailed chemical kinetics of lean premixed CH₄/air and H₂/air flames at high Karlovitz numbers (Ka ∼ 1800) is carried out. It is found that the high intensity turbulence along with differential diffusion result in a much more rapid transport of H radicals from the reaction zone to the low temperature unburned mixtures (∼500 K) than that in laminar flamelets. The enhanced concentration of H radicals in the low temperature zone drastically increases the reaction rates of exothermic chain terminating reactions (e.g., H + O₂+M = HO₂ + M in lean H₂/air flames), which results in a significantly enhanced heat release rate at low temperatures. This effect is observed in both CH₄/air and H₂/air flames and locally, the heat release rate in the low temperature zone can exceed the peak heat release rate of a laminar flamelet. The effects of chemical kinetics and transport properties on the H₂/air flame are investigated, from which it is concluded that the enhanced heat release rate in the low temperature zone is a convection–diffusion-reaction phenomenon, and to obtain it, detailed chemistry is essential and detailed transport is important.     

Direct numerical simulation of hydrogen-enriched lean premixed methane-air flames

The effect of hydrogen blending on lean premixed methane-air flames is studied with the direct numerical simulation (DNS) approach coupled with a reduced chemical mechanism. Two flames are compared with respect to stability and pollutant formation characteristics—one a pure methane flame close to the lean limit, and one enriched with hydrogen. The stability of the flame is quantified in terms of the turbulent flame speed. A higher speed is observed for the hydrogen-enriched flame consistent with extended blow-off stability limits found in measurements. The greater flame speed is the result of a combination of higher laminar flame speed, enhanced area generation, and greater burning rate per unit area. Preferential diffusion of hydrogen coupled with shorter flame time scales accounts for the enhanced flame surface area. In particular, the enriched flame is less diffusive-thermally stable and more resistant to quenching than the pure methane flame, resulting in a greater flame area generation. The burning rate per unit area correlates strongly with curvature as a result of preferential diffusion effects focusing fuel at positive cusps. Lower CO emissions per unit fuel consumption are observed for the enriched flame, consistent with experimental data. CO production is greatest in regions which undergo significant downstream interaction. In these regions, the enriched flame exhibits faster oxidation rates as a result of higher levels of OH concentration. NO emissions are increased for the enriched flame as a result of locally higher temperature and radical concentrations found in cusp regions.     

Effects of hydrogen enrichment on CH4/Air turbulent swirling premixed flames in a cuboid combustor

Effects of H₂-enrichment on structures of CH₄/air turbulent swirling premixed flames affected by high intensity turbulence in a gas turbine model combsutor are investigated by conducting direct numerical simulations. Two stoichiometric mixture conditions, of which volume ratio of CH₄:H₂ = 50:50 and 80:20, are simulated by considering a reduced chemistry (25 species and 111 reactions). Results showed qualitatively different flame shapes and reaction zone characteristics between the cases. For the higher H₂-ratio case, the flame is stabilized both in the inner and outer shear layers. For the lower H₂-ratio case, the flame is stabilized only in the inner shear layer and extinction occurs in the outer shear layer. Comparison of the reaction zone characteristics with unstrained and strained laminar flames in phase space showed that H₂ mass fraction for the lower H₂-ratio case and reaction rate profiles for both cases deviate from the corresponding laminar values. Analysis of fuel species conservation equation suggests that the turbulent transports are substantially influential to determine local and global flame structures. These findings would be useful for designing practical H₂-enriched gas turbine combustor in the aspect of flame structures under high intensity turbulence.     

Effect of the equivalence ratio,Damköhler number,Lewis number and heat release on the stability of laminar premixed flames in microchannels

The effect of the equivalence ratio on the stability and dynamics of a premixed flame in a planar micro-channel with a step-wise wall temperature profile is numerically investigated using the thermo-diffusive approximation. To characterize the stability behavior of the flame, we construct the stability maps delineating the regions with different flame dynamics in the inlet mass flow rate m   vs. the equivalence ratio ?$?$ parametric space. The flame stability is analyzed for fuels with different diffusivity by changing the Lewis numbers in the range 0.3?_LeF?1.4$0.3?{\mathit{Le}}_F?1.4$. On the other hand, the Lewis number of the oxidizer is kept constant and equal to unity _LeO=1${\mathit{Le}}_O=1$. Our results show that, for very diffusive fuels, the stability of the flame varies significantly with the equivalence ratio, transitioning from stable flames for lean mixtures to highly unstable flames when ?>1$?>1$. As the fuel Lewis number approaches unity, the stability behavior of the flame for lean and rich mixtures becomes more similar to give, in the equidiffusional case _LeF=1${\mathit{Le}}_F=1$, a symmetric stability map around the stoichiometric mixture ?=1$?=1$. In all cases considered, the most stable flames are always found around the stoichiometric mixtures ?=1$?=1$, when the flame instabilities are completely suppressed for very diffusive fuels _LeF<1${\mathit{Le}}_F<1$, or are reduced to a narrow range of inflow velocities for fuel Lewis numbers equal or greater than unity.     

Direct numerical simulation of hydrogen addition in turbulent premixed Bunsen flames using flamelet-generated manifold reduction

Direct numerical simulations of a lean premixed turbulent Bunsen flame with hydrogen addition have been performed. We show the results for a case with equivalence ratio of 0.7 and a molar fractional distribution of 40% H₂ and 60% CH₄. The flamelet-generated manifold technique is used to reduce the chemistry; flamelets with different equivalence ratios and inflow temperature are used to account for stretch effects that are enhanced by preferential diffusion. The three-dimensional simulation clearly shows enhanced burning velocity in regions convex toward the reactants and reduced burning velocity with possible extinction in regions concave toward the reactants. To obtain these effects it was found to be necessary to include two three-dimensional transport equations with essentially different diffusivities. This point is illustrated by comparison of the results with cases in which either a single transport equation was used or two transport equations with minor differences in diffusivities were used. These latter cases incorporated preferential diffusion in the 1D flamelets (and thus in the manifold), but not in the three-dimensional transport. Thus the three-dimensional preferential diffusion effects are shown to enhance curvature and thereby to increase the turbulent burning velocity and reduce the mean flame height. In addition the turbulent burning velocity increases because hydrogen addition leads to a larger laminar flamelet consumption speed. To demonstrate this second effect, results of the cases mentioned above are compared to the results of simulations of the Bunsen flame with 0% hydrogen added to the fuel.     

Effects of Damköhler number on flame extinction and reignition in turbulent non-premixed flames using DNS

Results from a parametric study of flame extinction and reignition with varying Damköhler number using direct numerical simulation are presented. Three planar, non-premixed ethylene jet flames were simulated at a constant Reynolds number of 5120. The fuel and oxidizer stream compositions were varied to adjust the steady laminar extinction scalar dissipation rate, while maintaining constant flow and geometric conditions. Peak flame extinction varies from approximately 40% to nearly global blowout as the Damköhler number decreases. The degree of extinction significantly affects the development of the jets and the degree of mixing of fuel, oxidizer, and combustion products prior to reignition. The global characteristics of the flames are presented along with an analysis of the modes of reignition. It is found that the initially non-premixed flame undergoing nearly global extinction reignites through premixed flame propagation in a highly stratified mixture. A progress variable is defined and a budget of key terms in its transport equation is presented.     

Effects of Lewis number on flame surface density transport in turbulent premixed combustion

The transport of flame surface density (FSD) in turbulent premixed flames has been studied using a database obtained from Direct Numerical Simulation (DNS). Three-dimensional freely propagating developing statistically planar turbulent premixed flames have been examined over a range of global Lewis numbers from 0.6 to 1.2. Simplified chemistry has been used and the emphasis is on the effects of Lewis number on FSD transport in the context of Reynolds-averaged closure modelling. Under the same initial conditions of turbulence, flames with low Lewis numbers are found to exhibit counter-gradient transport of FSD, whereas flames with higher Lewis numbers tend to exhibit gradient transport of FSD. Stronger heat release effects for lower Lewis number flames are found to lead to an increase in the positive (negative) value of the dilatation rate (normal strain rate) term in the FSD transport equation with decreasing Lewis number. The contribution of flame curvature to FSD transport is found to be influenced significantly by the effects of Lewis number on the curvature dependence of the magnitude of the reaction progress variable gradient, and on the combined reaction and normal diffusion components of displacement speed. The modelling of the various terms of the FSD transport equation has been analysed in detail and the performance of existing models is assessed with respect to the terms assembled from corresponding quantities extracted from DNS data. Based on this assessment, suitable models are identified which are able to address the effects of non-unity Lewis number on FSD transport, and new or modified models are suggested wherever necessary.     

Experimental study of Markstein number effects on laminar flamelet velocity in turbulent premixed flames

Effects of turbulent flame stretch on mean local laminar burning velocity of flamelets, , were investigated experimentally in an explosion vessel at normal temperature and pressure. In this context, the wrinkling, At/Al, and the burning velocity, ut, of turbulent flames were measured simultaneously. With the flamelet assumption the mean local laminar burning velocity of flamelets, , was calculated for different turbulence intensities. The results were compared to the influence of stretch on spherically expanding laminar flames. For spherically expanding laminar flames the stretched laminar burning velocity, un, varied linearly with the Karlovitz stretch factor, yielding Markstein numbers that depend on the mixture composition. Six different mixtures with positive and negative Markstein numbers were investigated. The measurements of the mean local laminar burning velocity of turbulent flamelets were used to derive an efficiency parameter, I, which reflects the impact of the Markstein number and turbulent flame stretch—expressed by the turbulent Karlovitz stretch factor—on the local laminar burning velocity of flamelets. The results showed that the efficiency is reduced with increasing turbulence intensity and the reduction can be correlated to unsteady effects.     

The interaction of high-speed turbulence with flames: Turbulent flame speed

Direct numerical simulations of the interaction of a premixed flame with driven, subsonic, homogeneous, isotropic, Kolmogorov-type turbulence in an unconfined system are used to study the mechanisms determining the turbulent flame speed, S_T, in the thin reaction zone regime. High intensity turbulence is considered with the r.m.s. velocity 35 times the laminar flame speed, S_L, resulting in the Damköhler number Da = 0.05. The simulations were performed with Athena-RFX, a massively parallel, fully compressible, high-order, dimensionally unsplit, reactive-flow code. A simplified reaction–diffusion model, based on the one-step Arrhenius kinetics, represents a stoichiometric H₂–air mixture under the assumption of the Lewis number Le = 1. Global properties and the internal structure of the flame were analyzed in an earlier paper, which showed that this system represents turbulent combustion in the thin reaction zone regime. This paper demonstrates that: (1) The flame brush has a complex internal structure, in which the isosurfaces of higher fuel mass fractions are folded on progressively smaller scales. (2) Global properties of the turbulent flame are best represented by the structure of the region of peak reaction rate, which defines the flame surface. (3) In the thin reaction zone regime, S_T is predominantly determined by the increase of the flame surface area, A_T, caused by turbulence. (4) The observed increase of S_T relative to S_L exceeds the corresponding increase of A_T relative to the surface area of the planar laminar flame, on average, by ≈14%, varying from only a few percent to as high as ≈30%. (5) This exaggerated response is the result of tight flame packing by turbulence, which causes frequent flame collisions and formation of regions of high flame curvature ?1/δ_L, or “cusps,” where δ_L is the thermal width of the laminar flame. (6) The local flame speed in the cusps substantially exceeds its laminar value, which results in a disproportionately large contribution of cusps to S_T compared with the flame surface area in them. (7) A criterion is established for transition to the regime significantly influenced by cusp formation. In particular, at Karlovitz numbers Ka ? 20, flame collisions provide an important mechanism controlling S_T, in addition to the increase of A_T by large-scale motions and the potential enhancement of diffusive transport by small-scale turbulence.     

Relevance of the Bray number in the small-scale modeling of turbulent premixed flames

The present study is devoted to the analysis of the influence of expansion phenomena on turbulent small scales in premixed reactive flows. It is shown that, under certain conditions, the expansion that takes place across wrinkled laminar flamelets can be sufficient to control the fluctuating velocity gradients and associated dissipation rate functions. These conditions are fixed by the respective values of a set of nondimensional parameters, namely the turbulence Reynolds number ReT, the Bray number, and the ratio between integral length scale of turbulence and thermal flame front thickness. A new criterion is introduced that makes it possible to delineate the influence of expansion phenomena on small-scale turbulent premixed reactive flows. The relevance of this criterion is analyzed in the light of experimental results represented in the classical diagram of combustion regime. The present analysis confirms that special care is required to represent and include the influence of expansion phenomena when using either RANS or LES closures to model turbulent premixed combustion.     

Modelling and simulation of lean premixed turbulent methane/hydrogen/air flames with an effective Lewis number approach

Recent work on reaction modelling of turbulent lean premixed combustion has shown a significant influence of the Lewis number even at high turbulence intensities, if different fuels and varied pressure is regarded. This was unexpected, as the Lewis number is based on molecular transport quantities (ratio of molecular thermal diffusivity to mass diffusivity), while highly turbulent flames are thought to be dominated from turbulent mixing and not from molecular transport. A simple physical picture allows an explanation, assuming that essentially the leading part of the wrinkled flame front determines the flame propagation and the average reaction rate, while the rear part of the flame is of reduced importance here (determining possibly the burnout process and the flame brush thickness but not the flame propagation). Following this argumentation, mostly positively curved flame elements determine the flame propagation and the average reaction rate, where the influence of the preferential molecular diffusion and the Lewis number can easily seen to be important. Additionally, an extension of this picture allows a simple derivation of an effective Lewis number relation for lean hydrogen/methane mixtures. The applicability and the limit of this concept is investigated for two sets of flames: turbulent pressurized Bunsen flames, where hydrogen content and pressure is varied (from CNRS Orléans), and highly turbulent pressurized dump combustor flames where the hydrogen content is varied (from PSI Baden). For RANS simulations, comparison of flame length data between experiment and an effective Lewis number model shows a very good agreement for all these flames with hydrogen content of the fuel up to 20 vol.%, and even rather good agreement for 30% and 40% hydrogen.     

Error analysis of large-eddy simulation of the turbulent non-premixed sydney bluff-body flame

A computational error analysis is applied to the large-eddy simulation of the turbulent non-premixed Sydney bluff-body flame, where the error is defined with respect to experimental data. The error-landscape approach is extended to heterogeneous compressible turbulence, which is coupled to combustion as described by a flamelet model. The Smagorinsky model formulation is used to model the unknown turbulent stresses. We introduce several measures to quantify the total simulation error and observe a striking ‘valley-structure’ in the error that arises as function of the spatial resolution and the Smagorinsky length parameter. The optimal refinement strategy that can be extracted from this error-landscape is reminiscent of that for non-reacting turbulent flow.     

Turbulent flame topology and the wrinkled structure characteristics of high pressure syngas flames up to 1.0 MPa

The turbulent flame topology characteristics of the model syngas with two different hydrogen ratios were statistically investigated, namely CO/H₂ ratio at 65/35 and 80/20, at equivalence ratio of 0.7. The combustion pressure was kept at 0.5 MPa and 1.0 MPa, to simulate the engine-like condition. The model syngas was diluted with CO₂ with a mole fraction of 0.3 which mimics the flue gas recycle in the turbulent combustion. CH₄/air flame with equivalence ratio of 1.0 was also tested for comparison. The flame was anchored on a premixed type Bunsen burner, which can generate a controllable turbulent flow. Flame front, which is represented by the sharp increased interface of the OH radical distribution, was measured with OH-PLIF technique. Flame front parameters were obtained through image processing to interpret the flame topology characteristics. Results showed that the turbulent flames possess a wrinkled character with smaller scale concave/convex structure superimposed on a larger scale convex structure under high pressure. The wrinkled structure of syngas flame is much finer and more corrugated than hydrocarbon fuel flames. The main reason is that scale of wrinkled structure is smaller for syngas flame, resulting from the unstable physics. Hydrogen in syngas can increase the intensity of the finer structure. Moreover, the model syngas flames have larger flame surface density than CH₄/air flame, and hydrogen ratio in syngas can increase flame surface density. This would be mainly attributed to the fact that the syngas flames have smaller flame intrinsic instability scale l_i than CH₄/air flame. S_T/S_L of the model syngas tested in this study is higher than CH₄/air flames for both pressures, due to the high diffusivity and fast burning property of H₂. This is mainly due to smaller L_M and l_i. V_f of the two model syngas is much smaller than CH₄/air flames, which suggests that syngas flame would lead to a larger possibility to occur combustion oscillation.     

Stability characteristics of lifted turbulent partially premixed jet flames

The stability characteristics of partially premixed turbulent lifted methane flames have been investigated and discussed in the present work. Mixture fraction and reaction zone behavior have been measured using a combined 2-D technique of simultaneous Rayleigh scattering, Laser Induced Predissociation Fluorescence (LIPF) of OH and Laser Induced Fluorescence (LIF) of C₂H_x. The stability characteristics and simultaneous mixture fraction-LIPF-LIF measurements in three lifted flames with originally partially premixed jets at different mean equivalence ratio and Reynolds number are presented and discussed in this paper. Higher stability of partially premixed flames as compared to non-premixed flames has been observed. Lifted, attached, blow-out and blow-off regimes have been addressed and discussed in this work. The data show that the mixture fraction field on approaching the stabilization region is uniquely characterized by a certain level of mean and rms fluctuations. This suggests that the stabilization mechanism is likely to be controlled by premixed flame propagation at the stabilization region. Triple flame structure has been detected in the present flames, which is likely to be the appropriate model at the stabilization point.     

The interaction of high-speed turbulence with flames: Global properties and internal flame structure

We study the dynamics and properties of a turbulent flame, formed in the presence of subsonic, high-speed, homogeneous, isotropic Kolmogorov-type turbulence in an unconfined system. Direct numerical simulations are performed with Athena-RFX, a massively parallel, fully compressible, high-order, dimensionally unsplit, reactive flow code. A simplified reaction-diffusion model represents a stoichiometric H₂-air mixture. The system being modeled represents turbulent combustion with the Damköhler number Da=0.05 and with the turbulent velocity at the energy injection scale 30 times larger than the laminar flame speed. The simulations show that flame interaction with high-speed turbulence forms a steadily propagating turbulent flame with a flame brush width approximately twice the energy injection scale and a speed four times the laminar flame speed. A method for reconstructing the internal flame structure is described and used to show that the turbulent flame consists of tightly folded flamelets. The reaction zone structure of these is virtually identical to that of the planar laminar flame, while the preheat zone is broadened by approximately a factor of two. Consequently, the system evolution represents turbulent combustion in the thin reaction zone regime. The turbulent cascade fails to penetrate the internal flame structure, and thus the action of small-scale turbulence is suppressed throughout most of the flame. Finally, our results suggest that for stoichiometric H₂-air mixtures, any substantial flame broadening by the action of turbulence cannot be expected in all subsonic regimes.     

Hysteresis loops of methane catalytic partial oxidation for hydrogen production under the effects of varied Reynolds number and Damköhler number

Hysteresis loops of catalytic partial oxidation of methane (CPOM) for hydrogen production under the effects of varied Reynolds number and Damköhler number are investigated numerically in this study. The physical phenomena are predicted using the indirect mechanism, which consists of the total oxidation (or combustion), steam reforming and CO₂ reforming of methane in a catalyst bed. Numerical results reveal that, when the Damköhler number is relatively low, a hysteresis loop of CPOM from varying Reynolds number develops. Increasing the Damköhler number leads to the loop shifting toward the regime of high Reynolds number. However, once the Damköhler number is large to a certain extent, the chemical reactions are always exhibited for the Reynolds number less than 2000. A closed loop is thus not observed. Alternatively, for a given Reynolds number, an ignited Damköhler number and an extinguished Damköhler number can be obtained. Accordingly, three different regions in the plot of Damköhler number versus Reynolds number are identified. Physically, when the role played by Damköhler number on CPOM is much more important than by the Reynolds number (Region I), the thermal effect governs the chemical reactions. In contrast, if the Reynolds number plays a key role in determining the CPOM (Region III), the chemically frozen flow prevails over the catalyst bed. When the residence times of the total oxidation and convection in the catalyst bed are in an equivalent state (Region II), CPOM is characterized by a dual-solution, rendering the hysteresis loops. From the distributions of ignited and extinguished Damköhler numbers, the catalytic reactor and operation of partial oxidation of methane and other fuels can be designed accordingly.     

High resolution dual-plane stereo-PIV for validation of subgrid scale models in large-eddy simulations of turbulent premixed flames

A measurement strategy for the experimental validation of subgrid scale (SGS) models for large-eddy simulations (LES) of turbulent premixed flames is presented. The approach is based on a dual-plane stereo-PIV technique. The measurement of the flow field is performed in two parallel planes which allows the determination of velocity gradients in all three directions. The flame front position in the PIV images is deduced from the clearly observable step in the particle number density between burnt and unburnt gases. This facilitates the single-shot based evaluation of important quantities for reacting flows, e.g., the density weighted rate-of-strain tensor. Also filtered quantities like the SGS scalar flux of the reaction progress variable can be directly determined by spatial averaging over several regions of interest which reproduces the application of the filter function in LES. Moreover, the measured data allows the direct interpretation of SGS model formulations since besides the filtered values, also resolved data are generated. Thus, statistical a-priori tests of SGS models are possible. The measurement strategy is explained, a statistical evaluation of the density weighted rate-of-strain tensor is given, and exemplarily an instantaneous distribution of the measured radial SGS scalar flux is compared with predictions of two models, the gradient diffusion model and the Clark model. Starting from a reference operation point of a turbulent V-shaped flame the following three parameters - Reynolds number, fuel-air ratio and fuel type - have been varied independently. First results show that the gradient diffusion model fails completely, while the Clark model predictions show a high degree of correlation to the directly determined flux components, especially in the reactant zone. More advanced modeling, however, may be needed, to incorporate for instance heat-release effects more closely.     

The structure of partially premixed methane flames in high-intensity turbulent flows

Direct numerical simulations (DNS) are conducted to study the structure of partially premixed and non-premixed methane flames in high-intensity two-dimensional isotropic turbulent flows. The results obtained via “flame normal analysis” show local extinction and reignition for both non-premixed and partially premixed flames. Dynamical analysis of the flame with a Lagrangian method indicates that the time integrated strain rate characterizes the finite-rate chemistry effects and the flame extinction better than the strain rate. It is observed that the flame behavior is affected by the “pressure-dilatation” and “viscous-dissipation” in addition to strain rate. Consistent with previous studies, high vorticity values are detected close to the reaction zone, where the vorticity generation by the “baroclinic torque” was found to be significant. The influences of (initial) Reynolds and Damköhler numbers, and various air-fuel premixing levels on flame and turbulence variables are also studied. It is observed that the flame extinction occurs similarly in flames with different fuel-air premixing. Our simulations also indicate that the CO emission increases as the partial premixing of the fuel with air increases. Higher values of the temperature, the OH mass fraction and the CO mass fraction are observed within the flame zone at higher Reynolds numbers.