Rich Flames Research Articles

Numerical Simulation of Propagation of Unsteady Tribrachial Flames in Laminar Non-Premixed Jets

The characteristics of propagating Tribrachial Flames in non-premixed laminar jets have been investigated computationally, using a two dimensional model in ANSYS FLUENT 16.0. The edge of the propagating flame in a laminar jet regime has a Tribrachial flame structure: a lean premixed flame, a rich premixed flame, and a diffusion flame, all extending from a single location. This is usually observed in non-premixed combustion such occurring in burners. Such flames are important in the interaction of heat and mass transfer with chemical reactions in gas turbines and commercial burners. A computational study has been performed to assess the laminar tribrachial flame propagation in methane jets and to compare the tribrachial flame propagation for various flow rates. With an increase in the fuel jet Reynolds number, the tribrachial point is found to move closer to the nozzle. Additionally the tribrachial point is found to become larger with an increase in the Reynolds number.

An Experimental and Detailed Chemical Kinetic Investigation of the Addition of C2 Oxygenated Species in Rich Ethylene Premixed Flames

ABSTRACTCurrent needs for reduced emissions and improved efficiency have motivated the study of the combustion of novel, oxygenated fuels. The scope of the present work is to experimentally and numerically analyse the effect of the addition of three C2 oxygenated species in rich ethylene flames and to quantify their propensity to inhibit pollutant formation. Acetaldehyde, ethanol and acetic acid are investigated, as representatives of three fuel classes. Novel experimental, speciation results from a low pressure flame configuration are reported. The mechanisms of both research groups co-authoring the study (NTUA and UCL) are analysed. Both appear to satisfactorily reproduce flame structure. The speciation measurements and the reaction path analysis indicate that the performance of oxygenated species with respect to molecular growth is related to their tendency to form ketene and other C2 species. In this context acetic acid seems to have the least beneficial behaviour towards heavier hydrocarbon formation.

Experimental analysis of oscillatory premixed flames in a Hele-Shaw cell propagating towards a closed end

An experimental study of methane, propane and dimethyl ether (DME) premixed flames propagating in a quasi-two-dimensional Hele-Shaw cell placed horizontally is presented in this paper. The flames are ignited at the open end of the combustion chamber and propagate towards the closed end. Our experiments revealed two distinct propagation regimes depending on the equivalence ratio of the mixture as a consequence of the coupling between the heat-release rate and the acoustic waves. The primary acoustic instability induces a small-amplitude, of around 8 mm, oscillatory motion across the chamber that is observed for lean propane, lean DME, and rich methane flames. Eventually, a secondary acoustic instability emerges for sufficiently rich (lean) propane and DME (methane) flames, inducing large-amplitude oscillations in the direction of propagation of the flame. The amplitude of these oscillations can be as large as 30 mm and drastically changes the outline of the flame. The front then forms pulsating finger-shaped structures that characterize the flame propagation under the secondary acoustic instability.The experimental setup allows the recording of the flame propagation from two different points of view. The top view is used to obtain accurate quantitative information about the flame propagation, while the lateral view offered a novel three dimensional perspective of the flame that gives relevant information on the transition between the two oscillatory regimes.The influence of the geometry of the Hele-Shaw cell and of the equivalence ratio on the transition between the two acoustic-instability regimes is analyzed. In particular, we find that the transition to the secondary instability occurs for values of the equivalence ratio ϕ above (below) a critical value ϕc for propane and DME (methane) flames. In all the tested fuels, the transition to the secondary instability emerges for values of the Markstein number M below a critical value Mc. The critical Markstein number varies with the gap size h formed by the two horizontal plates that bound the Hele-Shaw cell. As h is reduced, the critical Markstein number is shifted towards larger values.

Hazard evaluation of difluoromethane cloud explosion

Traditional refrigerants (such as R22) contribute greatly to the greenhouse gas emissions and ozone layer depletion. The R32 (difluoromethane, CH2F2) is an environmental substitute for R22 and R410a.This study is aimed at evaluating the hazard of difluoromethane cloud explosion by varying the oxygen content and equivalence ratio. In the experiment, the flame morphologies are recorded using the soap bubble method and the far-field pressure is measured using microphone pressure transducer. In the numerical simulation, the adiabatic flame temperature, thermal diffusivity and sensitivity analysis are obtained to reveal the factors affecting laminar burning velocity. The results demonstrate that the instability appears at high-oxygen content conditions and the diffusional-thermal instability is the leading factor of the lean flame, while the hydrodynamic instability is the leading factor of the rich flame. With the increase of the oxygen content, the laminar burning velocity, maximum explosion pressure, maximum pressure rise rate, positive pressure impulse increase while the negative pressure impulse decreases. With the increase of the equivalence ratio, the laminar burning velocity, maximum explosion pressure, maximum pressure rise rate, positive pressure impulse increases first and then decreases while the negative pressure impulse decreases first and then increases. The adiabatic flame temperature play a more important role comparing to the thermal diffusivity as the factor affecting the laminar burning velocity. Sensitivity analysis is conducted to anaylze the chemical influence on the laminar burning velocity. Chemical reactions H + O2 < = > O + OH, CHF2 + O2 = > CF2:O + O + H, CHF2 + H = CHF + HF and CF2:O + H = CF:O + HF show significant effects on the laminar burning velocity.

Impinging premixed methane-air flame jet of tube burner: thermal performance analysis for varied equivalence ratios

Thermally efficient gas burners can be designed by optimising the parameters associated with combustion and heat transfer mechanisms. The current study presents the thermal analysis of the premixed methane-air flame jets of circular tube gas burner impinging on a target surface. The effect of flame jet parameters such as mixture Reynolds number, equivalence ratio and burner to target surface spacing on heat transfer characteristics is investigated. The thermal performance is quantified in terms of thermal efficiency. The lean and stoichiometric mixtures release maximum amount of thermal energy. However, in lean flames, a part of the energy released is used for rising the temperature of excess air. Though the combustion is incomplete for fuel rich flames, higher heat transfer is achieved because of higher flame height. The optimal thermal performance is observed when the mixture is near stoichiometric and the burner is spaced from the target surface such that the premixed cone tip just touches the surface.

Characterization of six clustered methane-air diffusion microflames through spectroscopic and tomographic analysis of CH* and C2* chemiluminescence

The current study presents an emission spectroscopic investigation of clustered microflames established on a burner consisting of six micro fuel nozzles surrounding a center air nozzle which has been designed for flame synthesis and characterization of microflames. We conducted spectrally resolved measurements of the chemiluminescence of C2* and CH* and compared these results to a tomographic method applied to filtered emission measurements of CH* emission which can yield spatially resolved three dimensional mapping of the flame front. The analysis of the spatial distribution of the integrated band emission of C2* and CH* showed that the emission of both species is generated in the same locations in the flame which are the thin flame sheets shown in the tomography results of CH*. The ratio of the C2* and the CH* emission from the emission spectroscopy measurements was used to determine a corresponding local equivalence ratio through empirically derived correlations for premixed flames reported in literature. The resulting equivalence ratio equals one when the flame zone established at the boundary between fuel and air, indicating that diffusion dominated flame zones are established. Under certain nozzle pitch and flow rate conditions, the equivalence ratio above the center air nozzle inside the merged flame became significantly greater than one, which indicates a local fuel rich premixed type flame zone being established.To determine distributions of rotational and vibrational temperatures throughout the entire flame, we used the spectrally resolved emission from C2* and CH*. The temperature associated with vibrational excitation was different for both species but remained constant throughout the entire flame region and was found no representation of the flame temperature but an artifact of the combustion reactions producing the chemiluminescence. The rotational temperature of di-atomic molecules, however, equilibrates rather quickly with the translational temperature and can accurately represent the gas-dynamic temperature of the flame. The temperature distributions found from the two species agreed qualitatively, the CH* temperature was consistently higher than the C2* temperature by about 200 K. Spatial resolution was assigned to the results from the spectroscopic measurements through a comparison with the tomographically analyzed images of the CH* chemiluminescence.

Growth of Soot Volume Fraction and Aggregate Size in 1D Premixed C2H4/Air Flames Studied by Laser-Induced Incandescence and Angle-Dependent Light Scattering

The growth of soot volume fraction and aggregate size was studied in burner-stabilized premixed C2H4/air flames with equivalence ratios between 2.0 and 2.35 as function of height above the burner using laser-induced incandescence (LII) to measure soot volume fractions and angle-dependent light scattering (ADLS) to measure corresponding aggregate sizes. Flame temperatures were varied at fixed equivalence ratio by changing the exit velocity of the unburned gas mixture. Temperatures were measured using spontaneous Raman scattering in flames with equivalence ratios up to ϕ = 2.1, with results showing good correspondence (within 50 K) with temperatures calculated using the San Diego mechanism. Both the soot volume fraction and radius of gyration strongly increase in richer flames. Furthermore, both show a nonmonotonic dependence on flame temperature, with a maximum occurring at ~1675 K for the volume fraction and ~1700 K for the radius of gyration. The measurement results were compared with calculations using two different semiempirical two-equation models of soot formation. Numerical calculations using both mechanisms substantially overpredict the measured soot volume fractions, although the models do better in richer flames. The model accounting for particle coagulation overpredicts the measured radii of gyration substantially for all equivalence ratios, although the calculated values improve at ϕ = 2.35.

Experimental Study on Propane/Air Flame Propagation Characteristics in a Disc-Like Gap Chamber

ABSTRACTA disc-like gap chamber was developed to investigate the microscale effects on flame propagation characteristics in closed chamber, which is very important to understand the combustion process in micro internal combustion engines. Under initially normal pressure and temperature, the morphology and evolution of outwardly propagating flame, the flammable limit, flame speed and flame instability of propane-air flames influenced by gap width and equivalence ratio were experimentally studied. The results show that, as the gap width nearly equals to the quenching distance, the sustained propagation was possible only with rich flames. The smooth flames and wrinkled flames were observed. It clearly shows that the flame wrinkles developed gradually instead of instantaneously. The flame speed was slower in the 2.0 mm gap chamber than in a conventional combustion bomb, and the flame speed decreased with the increase of flame radius. As a result, the flame acceleration was not observed in present experiments. With the smaller gap width, the wrinkled flame occurred early, and the flame speed was slower. Due to the geometry of disc-like gap chamber, simple theoretical analysis indicates that the heat loss and elevated pressure play important roles in reducing flame speed.

Structure of a stratified CH4 flame with H2 addition

To explore the effect of H2 addition (20 percent by volume) on stratified-premixed methane combustion in a turbulent flow, an experimental investigation on a new flame configuration of the Darmstadt stratified burner is conducted. Major species concentrations and temperature are measured with high spatial resolution by 1D Raman-Rayleigh scattering. A conditioning on local equivalence ratio (range from ϕ = 0.45 to ϕ = 1.25) and local stratification is applied to the large dataset and allows to analyze the impact of H2 addition on the flame structure. The local stratification level is determined as Δϕ/ΔT at the location of maximum CO mass fraction for each instantaneous flame realization. Due to the H2 addition, preferential diffusion of H2 is different than in pure methane flames. In addition to diffusing out of the reaction zone where it is formed, particularly in rich conditions, H2 also diffuses from the cold reactant mixture into the flame front. For rich conditions (ϕ = 1.05 to ϕ = 1.15) H2 mass fractions are significantly elevated within the intermediate temperature range compared to fully-premixed laminar flame simulations. This elevation is attributed to preferential transport of H2 into the rich flame front from adjacent even richer regions of the flow. Additionally, when the local stratification across the flame front is taken into account, it is revealed that the state-space relation of H2 is not only a function of the local stoichiometry but also the local stratification level. In these flames H2 is the only major species showing sensitivity of the state-space relation to an equivalence ratio gradient across the flame front.

Investigation of the influence of DMMP on the laminar burning velocity of methane/air premixed flames

This paper conducted experimental and numerical studies to evaluate the performance of dimethyl methylphosphonate (DMMP) in suppressing methane/air premixed flames. Laminar burning velocities were measured using the Bunsen flame method at an elevated temperature (373 K). As the concentration of DMMP increased, the laminar burning velocity initially decreased quickly and then gradually, which indicated the appearance of the saturation effect of the inhibition efficiency. The saturation effect was especially obvious for flames at an equivalence ratio of 1.2. Numerical calculations coupling with detailed chemical reaction mechanism were conducted to interpret the measurements. The newly constructed mechanism predicted well the laminar burning velocities. Sensitivity analyses revealed that as the equivalence ratio increased, reactions involving high-valence phosphorus had diminishing influence on the laminar burning velocity but the reactions involving low-valence phosphorus became more influential. By comparing the thermal and chemical effects of the main phosphorus-containing reactions in the rich flames (φ = 1.2), it was found that the thermal effect increased while the chemical effect decreased with the increase of DMMP addition. Therefore, the saturation effect of the inhibition efficiency of DMMP at high loadings was partially attributed to the heat release of the recombination reactions involving small P-bearing species.

Direct numerical simulations of rich premixed turbulent n-dodecane/air flames at diesel engine conditions

Rich premixed turbulent n-dodecane/air flames at diesel engine conditions are analyzed using direct numerical simulations. The conditions correspond to a parametric variation of the Engine Combustion Network Spray A (pressure 60 atm; oxidizer oxygen level and temperature 21% and 900 K, respectively; fuel temperature 363 K). Three simulations with equivalence ratios of 3, 5, and 7 are performed with a Karlovitz number (Ka, based on flame time) of order 100 to match the estimated Ka of the rich premixed combustion region in Spray A. At these conditions, the reference laminar flames exhibit a complex structure which involves both low-temperature chemistry (LTC) and high-temperature chemistry over a wide range of length scales. In the presence of turbulence, the flame structure is strongly affected in physical space and the reaction zone exhibits a very complex structure in which broken, distributed, and thin regions co-exist, especially for the leanest case. However, the contribution of the LTC pathway is only weakly affected by turbulence. In progress variable space, the mean flame structure, including the chemical source terms, is found to match remarkably well that of the corresponding unity Lewis number laminar flame, particularly for the ϕ= 3 and 5 cases. This behavior is attributed to the strong turbulent mixing occurring throughout the flames/reaction zones, which suppresses differential diffusion effects. Nevertheless, large conditional fluctuations around the mean chemical source terms are identified. These are found to correlate very well with radical species mass fractions such as OH. In addition, a similar functional dependence is obtained from counterflow laminar flames. As such, it appears from these results that laminar flame models have a potential to be used to represent the thermochemical state of rich premixed turbulent flames under diesel engine conditions.

Effect of H-atom concentration on soot formation in premixed ethylene/air flames

Soot formation is a major challenge in the development of clean and efficient combustion systems based on hydrocarbon fuels. Fundamental understanding of the reaction mechanism leading to soot formation can be obtained by investigating the role of key reactive species such as atomic hydrogen taking part in soot formation pathways. In this study, two-dimensional laser induced incandescence (LII) measurements using λ = 1064 nm laser have been used to measure soot volume fraction (fV) in a series of rich ethylene (C2H4)/air flames, stabilized over a McKenna burner fitted with a flame stabilizing metal disc. Moreover, a comparison of UV (λ = 283 nm), visible (λ = 532 nm) and IR (λ = 1064 nm) laser excited LII measurements of soot is discussed. Recently developed, femtosecond two-photon laser-induced fluorescence (fs-TPLIF) technique has been applied for obtaining spatially resolved H-atom concentration ([H]) profiles under the same flame conditions. The structure of the flames has also been determined using hydroxyl radical (OH) planar laser induced fluorescence (PLIF) imaging. The results indicate an inverse dependence of fV on [H] for a range of C2H4/air rich flames up to an equivalence ratio, Φ = 3.0. Although an absolute relationship between [H] and fV cannot be easily derived owing to the multiple steps involving H and other intermediate species in soot formation pathways, the present study demonstrates the feasibility to couple [H] and fV obtained using advanced optical techniques for soot formation studies.

Examination of a soot model in premixed laminar flames at fuel-rich conditions

The primary objective of the present work is to collect and review the vast amount of experimental data on rich laminar premixed flames of hydrocarbon fuels reported in recent years, and to analyze them by using a detailed kinetic mechanism, identifying aspects of the mechanism of PAH and soot formation requiring further revisions. The kinetic assessment was hierarchically conducted, with the progressive extension of the core C0–C2 NUIG mechanism up to the CRECK kinetic mechanism of PAH and soot formation. This mechanism is here adopted to evaluate and analyze the extensive amount of experimental data collected. Therefore, it provides a kinetic guideline, useful to critically compare and unify flames involving similar fuels and/or conditions from different sources. The relevant effect of soot particles formation on heavy PAHs concentration is also discussed, together with the kinetic analysis highlighting systematic deviations and critical issues still existing in the present model. The model performances were evaluated using the Curve Matching approach (Bernardi et al., 2016). Considering the challenges of quantitative PAH measurements and associated uncertainties, this extensive database is a further value of this paper and is beneficial for improving reliability of kinetic models in a wide range of conditions.

Applicability of Femtosecond Laser Electronic Excitation Tagging in Combustion Flow Field Velocity Measurements.

Femtosecond laser electronic excitation tagging (FLEET) is a molecular tagging velocimetry technique that can be applied in combustion flow fields, although detailed studies of its application in combustion are still needed. We report the applicability of FLEET in premixed CH4-air flames. We found that FLEET can be applied in all of the combustion areas (e.g., the unburned region, the burned region and the reaction zone). The FLEET signal in the unburned region is significantly higher than that in the burned region. This technique is suitable for both lean and rich CH4-air combustion flow fields and its performance in lean flames is better than that in rich flames.

Influences of stoichiometry on steadily propagating triple flames in counterflows

Most studies of triple flames in counterflowing streams of fuel and oxidizer have been focused on the symmetric problem in which the stoichiometric mixture fraction is 1/2. There then exist lean and rich premixed flames of roughly equal strengths, with a diffusion flame trailing behind from the stoichiometric point at which they meet. In the majority of realistic situations, however, the stoichiometric mixture fraction departs appreciably from unity, typically being quite small. With the objective of clarifying the influences of stoichiometry, attention is focused on one of the simplest possible models, addressed here mainly by numerical integration. When the stoichiometric mixture fraction departs appreciably from 1/2, one of the premixed wings is found to be dominant to such an extent that the diffusion flame and the other premixed flame are very weak by comparison. These curved, partially premixed flames are expected to be relevant in realistic configurations. In addition, a simple kinematic balance is shown to predict the shape of the front and the propagation velocity reasonably well in the limit of low stretch and low curvature.

Experimental and kinetic studies on laminar flame characteristics of acetone-butanol-ethanol (ABE) and toluene reference fuel (TRF) blends at atmospheric pressure

Acetone–Butanol–Ethanol (ABE), as an alternative for butanol, can be used either in neat form or in blends with fossil fuels. Yet the flame characteristics of ABE blending gasolines have not been reported to date. To explore the effects of ABE addition on the laminar flame speed of gasoline, the spherical propagating flames of ABE, gasoline (represented by toluene reference fuel, TRF), as well as ABE-TRF mixture were measured at an initial condition of 1 bar, 400 K, and an equivalence ranging from 0.8 to 1.6. The results indicated that with an addition of 20% ABE fuel, the laminar flame speeds of ABE-TRF flame showed apparent increments at stoichiometric and ϕ=1.2 flames. The normalized increment of laminar burning velocity Iv raised monotonically with the increase of equivalence ratio. The three ABE-TRF mixtures showed small increment of laminar burning velocity than TRF flame in the order of un.ABE20631<un.ABE20361<un.ABE20163, suggesting a suppressing effect of acetone addition on the mixture burning velocity of ABE-TRF mixtures. ABE fuel with more alcohol contents extended the range of stable flame while acetone addition diminished the flame diffusional thermal stability. The kinetic analysis showed that the rate-limiting reactions shifted from heavier molecular reactions to lighter ones by blending the ABE fuel into TRF, as well as the rate of production (ROP) of OH radical increased by 5.5%. The sensitivity reactions for rich ABE-TRF flames are also more dependent on light fractions compared to lean flames.

Effect of hydrogen addition on laminar burning velocity of CH4/DME mixtures by heat flux method and kinetic modeling

Laminar burning velocities were measured by the heat flux method for premixed methane dimethyl-ether mixtures (CH4/DME) with air, at various mixture fractions from 100% CH4 to 100% DME with hydrogen addition of 0%, 20% and 40%. Experiments were carried out at atmospheric pressure and room temperature within an equivalence ratio range between 0.6 and 1.7. Four chemical kinetic mechanisms specifically created for DME combustion were validated against measured laminar burning velocity data, with Zhao’s mechanism providing accurate prediction for both CH4 and DME. Decreasing the DME/CH4 ratio in the fuel decreased the laminar burning velocity but the decreasing pattern behaved differently in rich flames. Enhancement of the laminar burning velocity by hydrogen addition was greater for 100% CH4 fuel than for 100% DME by 20–60% depending on the equivalence ratio and showed a non-monotonic behavior in rich CH4/DME flames. Reaction path and sensitivity analyses point at the increasing importance of CH3 than H and OH in rich flames with H2 addition. A strong linear correlation was indicated between laminar burning velocity and maximum concentration of [H + OH + CH3] radicals. This result indicates the importance of the CH3 radical in promoting combustion at rich equivalence ratios which is the reason for the non-monotonic laminar burning velocity enhancement by hydrogen addition.

Chemical Explosive Mode Analysis for a Jet-in-Hot-Coflow burner operating in MILD combustion

Large Eddy Simulations (LES) of Moderate and Intense Low oxygen Dilution (MILD) combustion of a Jet-in-Hot-Coflow (JHC) burner were performed using detailed chemistry. On the contrary to traditional flames, where heat release is occurring in very thin fronts, MILD combustion occurs in the distributed reaction regime where the reaction zone is broad, thus, this paper applies a direct Arrhenius closure with detailed chemistry to resolve important details of the fuel oxidation reactions. Comparisons of LES results are in good agreement with experiments, demonstrating that the simulations capture the intermediate species and finite reaction rate effects. A Chemical Explosive Mode Analysis (CEMA) was used to determine the flame structure and to detect the pre- and post-ignition regions, including the contributions to the CEMs analyzing the Explosion Index (EI) and Participation Index (PI). To the best of our knowledge, a detailed study of CEMA on MILD or flameless regime has never been reported. The flame structure was clearly visualized with CEMA, as well as the lean and the rich flame fronts. Different flame zones close to the anchoring points of these turbulent lifted flames were selected and the analysis demonstrates the contributions of dominant chemical species, such as HO2 and O. The reactions related to the dominant local CEM were obtained to highlight the nature of the stabilization in these highly diluted operating conditions.

Autoignited lifted flames of dimethyl ether in heated coflow air

Autoignited lifted flames of dimethyl ether (DME) in laminar nonpremixed jets with high-temperature coflow air have been studied experimentally. When the initial temperature was elevated to over 860 K, an autoignition occurred without requiring an external ignition source. A planar laser-induced fluorescence (PLIF) technique for formaldehyde (CH2O) visualized qualitatively the zone of low temperature kinetics in a premixed flame. Two flame configurations were investigated; (1) autoignited lifted flames with tribrachial edge having three distinct branches of a lean and a rich premixed flame wings with a trailing diffusion flame and (2) autoignited lifted flames with mild combustion when the fuel was highly diluted. For the autoignited tribrachial edge flames at critical autoignition conditions, exhibiting repetitive extinction and re-ignition phenomena near a blowout condition, the characteristic flow time (liftoff height scaled with jet velocity) was correlated with the square of the ignition delay time of the stoichiometric mixture. The liftoff heights were also correlated as a function of jet velocity times the square of ignition delay time. Formaldehydes were observed between the fuel nozzle and the lifted flame edge, emphasizing a low-temperature kinetics for autoignited lifted flames, while for a non-autoignited lifted flame, formaldehydes were observed near a thin luminous flame zone.For the autoignited lifted flames with mild combustion, especially at a high temperature, a unique non-monotonic liftoff height behavior was observed; decreasing and then increasing liftoff height with jet velocity. This behavior was similar to the binary mixture fuels of CH4/H2 and CO/H2 observed previously. A transient homogeneous autoignition analysis suggested that such decreasing behavior with jet velocity can be attributed to partial oxidation characteristics of DME in producing appreciable amounts of CH4/CO/H2 ahead of the edge flame region.

Modelling of diesel spray flames under engine-like conditions using an accelerated Eulerian Stochastic Field method

This paper aims to simulate diesel spray flames across a wide range of engine-like conditions using the Eulerian Stochastic Field probability density function (ESF-PDF) model. The ESF model is coupled with the Chemistry Coordinate Mapping approach to expedite the calculation. A convergence study is carried out for a number of stochastic fields at five different conditions, covering both conventional diesel combustion and low-temperature combustion regimes. Ignition delay time, flame lift-off length as well as distributions of temperature and various combustion products are used to evaluate the performance of the model. The peak values of these properties generated using thirty-two stochastic fields are found to converge, with a maximum relative difference of 27% as compared to those from a greater number of stochastic fields. The ESF-PDF model with thirty-two stochastic fields performs reasonably well in reproducing the experimental flame development, ignition delay times and lift-off lengths. The ESF-PDF model also predicts a broader hydroxyl radical distribution which resembles the experimental observation, indicating that the turbulence–chemistry interaction is captured by the ESF-PDF model. The validated model is subsequently used to investigate the flame structures under different conditions. Analyses based on flame index and formaldehyde distribution suggest that a triple flame, which consists of a rich premixed flame, a diffusion flame and a lean premixed flame, is established in the earlier stage of the combustion. As the combustion progresses, the lean premixed flame weakens and diminishes with time. Eventually, only a double-flame structure, made up of the diffusion flame and the rich premixed flame, is observed. The analyses for various ambient temperatures show that the triple-flame structure remains for a longer period of time in cases with lower ambient temperatures. The present study shows that the ESF-PDF method is a valuable alternative to Lagrangian particle PDF methods.