1D Raman/Rayleigh/CO-LIF line measurements of major and temperature in turbulent DME/air jet flame

Transcription

1 Paper # 7L-64 opic: urbulent Flames 8 th US National Combustion Meeting Organized by the Western States Section of the Combustion Institute and hosted by the University of Utah May 9-22, 23. D Raman/Rayleigh/-LIF line measurements of major and temperature in turbulent /air jet flame Frederik Fuest Robert S. Barlow 2 Gaetano Magnotti 2 Jeffrey A. Sutton Department of Mechanical and Aerospace Engineering, he Ohio State University, Columbus, OH, USA 2 Sandia National Laboratories, Livermore, CA, USA Dimethyl ether () is considered a promising alternative to diesel fuel in compression-ignition engines because of low particulate and NO x emissions along with high thermal efficiencies as well as improved auto-ignition characteristics due to its high cetane number. Recently, also has been selected as a prospective fuel candidate for the validation of turbulent combustion models organized within the NF Workshop, where recent focus has shifted to fuels with increasing complexity as compared to hydrogen and methane. In this work we present experimental results of seven major species mole fractions ( 2, O 2,, N 2,, H 2 O, H 2 ) and temperature determined from Raman/Rayleigh/-LIF line measurements in a series of piloted, partially-premixed /air turbulent jet flames with.2 mole fraction in the jet. Results are presented for Flame D, the lowest jet exit Reynolds number case of the series. Mean and RMS radial profiles were derived from the instantaneous measurements at various axial positions downstream of the nozzle exit of the Sydney/Sandia piloted jet burner. Insights into the turbulent flame structure are examined by conditioning the data on the mixture fraction and comparing the turbulent flame results to one-dimensional laminar flame calculations at varying strain rates. Introduction Beginning in 996, the validation of turbulent combustion models from benchmark experimental data sets has been a primary goal of the International Workshop on Measurement and Computation of urbulent Flames (NF) []. In order to provide experimental results on the flow field and the scalar structure of well-defined target flames, a variety of laser-based diagnostics have been applied. One important tool to gain information on the interaction between the turbulent flow field and the combustion chemistry is D Raman/Rayleigh scattering. Comprehensive data sets of temperature, major species ( 2, O 2,, N 2, CH 4, H 2 O, H 2 ), and derived quantities, such as mixture fraction, and their gradients in various turbulent flames which have been used to assess and validate turbulent combustion models [2]. In recent years, there have been efforts made to extend these diagnostic methods to fuels with increased complexity as compared to the predominately studied hydrogen and methane flames. Dimethyl ether () has been selected as a promising target fuel, primarily due to its low tendency to form soot and consequent accessibility by D Raman/Rayleigh measurements [3, 4]. Frank et al. [5] introduced a series of partially-premixed

2 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames /air flames following the well-documented methane/air flame series A-F in the Sydney/Sandia piloted jet burner geometry [6, 7]. In this work we present experimental results from D Raman/Rayleigh/-LIF measurements of seven major species ( 2, O 2,, N 2,, H 2 O, H 2 ) and temperature from turbulent flame -D, the lowest jet exit Reynolds number case of the new series. he experimental setup and details concerning accompanying laminar flame calculations are described in Section 2. Species and temperature profiles along the radial coordinate of the flame at different downstream locations are presented in addition to conditionally-averaged (on mixture fraction) results in Section 3 and the Appendix. A summary and concluding remarks are given in Section 4. 2 Experimental method Simultaneous D Raman/Rayleigh scattering and laser induced fluorescence (-LIF) were used to measure instantaneous species concentrations and temperature at the Combustion Research Facility of the Sandia National Laboratories [3, 8, 9]. Four sequentially-fired frequency-doubled Nd:YAG lasers operating at 532 nm were used to determine species mole fractions and temperature along a 6 mm line segment via spontaneous Raman/Rayleigh scattering. In addition, an Nd:YAG-pumped tunable dye laser was used for the simultaneous -LIF measurement. For the Raman/Rayleigh measurements optical pulse stretchers and reduced laser power were used to avoid optical breakdown at the probe volume, where a combined energy of J/pulse at 532 nm was focused by a 5 mm-lens to a projected beam waist of 2 µm as determined at /e 2 from the Rayleigh image. Laser energy fluctuations were monitored using a thermoelectric joule meter, yielding a precision in between.25% and % in the Rayleigh temperature measurement for room and flame temperatures, respectively. Within the spectrometer, the Raman, Rayleigh, and -LIF signals were separated by dichroic beam splitters. In order to reduce the crosstalk of depolarized broadband and C 2 -fluorescence interferences and chemiluminescence, the Raman-scattered light passed through a thin-film polarizer before dispersion via a high-transmission grating (2 lines/mm). A low-noise cryogenically cooled CCD camera (Roper Scientific, VersArray 3) in conjunction with custom-built rotating wheel shutters, were used to collect the dispersed Raman signals from 55-7 nm with exposure times as short as 3.9 µs (FWHM). he data was acquired using spectral and spatial hardware onchip binning to decrease camera readout noise and readout time. In the spatial direction a ten pixel on-chip binning was applied, yielding sixty spatial Raman superpixels along the line segment with a spatial resolution of 2.6 µm/pixel. he images were processed using the hybrid matrix inversion method with extensions for processing -data as outlined in [3, 4, 9]. One laminar flame calculation was used to derive Raman response and crosstalk curves and a temperature-dependent Rayleigh scattering cross section model to account for decomposition into smaller hydrocarbon molecules which were not measured separately. his adds additional uncertainties to the species and temperature measurements with a peak uncertainty around 4 K as shown in [3, 4, 9]. For example, the accuracy in the temperature measurement around 4 K is reduced from ±2% to ±5%. Air, cold gases diluted with nitrogen, and laminar methane/air flat flames were used for calibration of species signals and temperature. Piloted turbulent /air jet flames, as introduced previously by Frank et al. [] with a stoichio- 2

3 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames metric value of the mixture fraction of ξ st =.35 were investigated in the burner configuration of the Sydney/Sandia piloted flame series A-F [7, 4]. In the present work, results of flame D are presented. able gives the unburnt gas compositions and flow conditions for the main jet, pilot, and co-flow. able : Unburnt gas compositions of the turbulent /air flame D and the co-flow in mole fractions. C 2 H 2 H 2 N 2 O 2 Ar H 2 O 2 u (m/s) Jet Pilot Co-Flow Laminar flame calculations were conducted using CHEMD [5] and the reaction mechanism from Zhao et al. [6]. Experimental results are compared to calculations at various strain rates based on either complex species transport (multi-component) or the equal Lewis number assumption (Le = ) for all species. For all comparisons mixture fraction was derived consistently from species mole fractions for both the experimental data and the calculations. he important intermediate species, which were not measured in the experiment (CH 2 O, CH 4, C 2 H 2, C 2 H 4, C 2 H 6 and CH 3 ) were treated as outlined in [3, 4] to derive the adapted mixture fraction ξ6+(5) in order to provide a consistent comparison of the data from the experiment and the calculations. 3 Results and Discussion Data in the turbulent /air flame were taken at five different axial positions of x/d =, 5,, 2 and 4 and sufficient radial positions (at each downstream location) to cover the entire composition space of the flame and a part of the flame on the opposite side of the centerline to image the symmetry. he spatial distance between the reported data points along the radial coordinate is approximately. mm. 3 laser shots were taken at each radial position, using 3 mm radial steps, giving 5 percent overlap of the 6-mm probe volume. For the spatial mean and rms results shown in Figs. -5, the mean and rms values at overlapping spatial positions were interpolated and averaged after the data-processing in order to provide equidistant radial positions. Spatial profiles of N 2 are also part of the data set but are not shown in Figs. -5. At x/d =, the spatial domain is dominated by the products of the surrounding pilot flame. he unburnt /air mixture on the centerline extends to more than two-thirds of the jet nozzle diameter. Steeply increasing 2 and especially H 2 O profiles, as well as nonzero and H 2 levels within the shear layer between the main jet and the pilot, indicate the reaction of the /air mixture and its impact on the composition of the burnt gases. his observation is in accordance with the narrow rms profiles of all species, temperature and mixture fraction with widths of less than 2 mm. Although an inflection point in both the mean temperature and mean mixture fraction profiles is observed at r = 6.2 mm, no plateau of constant temperature nor mixture fraction is present between the end of the shear layer at about r = 5 mm and the end of the pilot at about r = 9 mm. 3

4 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames MEAN Mole Fractions (-) O LIF MEAN Mixture Fraction (-) MEAN emperature (K) RMS Mole Fractions (-) O LIF RMS Mixture Fraction (-) RMS emperature (K) Figure : Spatial mean and rms values of species mole fractions, temperature, and mixture fraction at x/d =. his indicates a spatial overlap of the mixing and diffusion zones along the radial coordinate of co-flow air, pilot, and the main jet on the inner side of the pilot. A small peak in the rms temperature profile at the transition point from the pilot to the co-flow at r = 9 mm indicates small spatial fluctuations of the temperature profile around this position even though both pilot and co-flow are laminar flows. A mismatch between peak values from and -Raman is present throughout the entire data set gradually decreasing further downstream where higher absolute values of are found. However, since -Raman is prone to uncertainties from large interferences and concurrently there is little experience with -LIF measurements in the flames to date, particularly with respect to collisional quenching effects, this inconsistency cannot be resolved at this stage. In the data presented in this work, we use the values from -LIF measurements within the data-processing, species mole fraction normalization to unity and mixture fraction calculation. he influence from the pilot is no longer observable at x/d = 5, as shown in Fig. 2. Fuel consumption is evident up to about mm distance from the centerline. All mean profiles of species, temperature and mixture fraction are characteristic for a simple jet flame. he mixing layer between the main jet, and the co-flowing air is extended to about mm in width. Double-peak structures are observed in all rms profiles with the exception of and mixture fraction in accordance with their monotonically decreasing mean profiles. Compared to the x/d = results, the peak temperature increased slightly from 9 K to 95 K and moved by mm radially outwards. Peak values of the mean, H 2, and H 2 O profiles peak increased by factors of.5, 2, and.5, respectively, while the peak 2 value did not change. Figure 3 shows the spatial profiles at x/d =. At this axial position, the mixing layer extends further radially, with a total width of 6 mm. A small increase in the rms profiles of temperature and mixture fraction at r = indicates that turbulent mixing of the fuel stream, pilot products, and co-flowing air is starting to be observed along the entire radial profile up to the centerline although it is still too small to affect the mean values significantly. he peak in temperature has decreased 4

5 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames MEAN Mole Fractions (-) O LIF MEAN Mixture Fraction (-) MEAN emperature (K) RMS Mole Fractions (-) O LIF RMS Mixture Fraction (-) RMS emperature (K) Figure 2: Spatial mean and rms values of species mole fractions, temperature, and mixture fraction at x/d = 5. MEAN Mole Fractions (-) O LIF MEAN Mixture Fraction (-) MEAN emperature (K) RMS Mole Fractions (-) O LIF RMS Mixture Fraction (-) RMS emperature (K) Figure 3: Spatial mean and rms values of species mole fractions, temperature, and mixture fraction at x/d =. 5

6 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames MEAN Mole Fractions (-) O LIF MEAN Mixture Fraction (-) MEAN emperature (K) RMS Mole Fractions (-) O LIF RMS Mixture Fraction (-) RMS emperature (K) Figure 4: Spatial mean and rms values of species mole fractions, temperature, and mixture fraction at x/d = 2. to 9 K and moved radially outwards mm. In the rms profiles the second peaks at higher radial positions have become more prominent as compared to the lower axial positions of x/d = and 5. Significant mixing at x/d = 2 still is not present at the centerline position as can be seen in the mean mixture fraction profile shown in Fig. 4, which decreased by only a few percent as compared to the x/d = results. On the other hand, the mean profile has decreased significantly due to mixing. Both and H 2 levels increased further. he mean temperature peak value dropped down to about 8 K. Larger C 2 interferences, background luminosity, and increased effects of beam steering start impacting the measurement accuracy and discontinuities show up in some of the rms profiles. At this axial position, -Raman and -LIF agree to <%. At the furthest downstream location which was measured in this work (x/d = 4), both and the mixture fraction on the centerline have decreased significantly. Also observed is an increase in the mean centerline temperature from 65 K at x/d = 2 to 7 K and higher fluctuations in the mixture fraction and temperature rms profiles. Although the position of the peak mean temperature did not move further outwards radially, the width of the flame extended to r = 52 mm. At x/d = 4, the mean structure has changed from that of a flame burning around an unburnt jet core with two flame peaks on each side of the centerline, to a single flame across the centerline. Peak and H 2 levels increased by an additional 2% and -Raman and -LIF results are almost identical. Discontinuities due to the aforementioned distortions are now partly affecting the mean profiles. Conditional mean values in mixture fraction space are shown in Fig. 6 for x/d =, 5 and 4 in comparison to laminar flame calculations at different strain rates using both multi-component and Lewis number unity transport. he lowest strain rate in the calculation was a = s for both transport assumptions. he highest strain rates reflect the corresponding extinction limits. he experimental results show the flame crossing different strain rate and transport regimes depending on the axial location and mixture fraction range. In general, strain rate decreases with increas- 6

7 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames MEAN Mole Fractions (-) O LIF MEAN Mixture Fraction (-) MEAN emperature (K) RMS Mole Fractions (-) O LIF RMS Mixture Fraction (-) RMS emperature (K) Figure 5: Spatial mean and rms values of species mole fractions, temperature, and mixture fraction at x/d = 4. ing distance from the jet nozzle. his is evident for six of the seven species and the temperature profiles. Although the same trend is apparent in the 2 profile, the differences are less obvious. Whereas the experimental 2 profiles at x/d = 5 and 4 still follow the trend of decreasing strain rate with increasing axial position, the profile at x/d = shows higher 2 levels around stoichiometric mixture fraction. his may be attributed to the influence of the pilot which also is bending this profile around ξ =.2 notably. Furthermore, due to the presence of the pilot, experimental profiles from, N 2, H 2 and -LIF at x/d = are significantly shifted beyond the highest strain rate and cannot be matched by a laminar flame calculation. In addition, a small amount of fuel is left around the stoichiometric contour. urbulent transport and differential diffusion is affecting the flame differently depending on the axial location as well as on the position in mixture fraction space. here are distinctive features observable in the laminar flame calculations which can indicate differential diffusion effects in the experimental results: Mixture fraction values greater than one only are possible if differential diffusion takes place, mostly from hydrogen diffusing preferentially into fuel-rich mixtures. If the amount of hydrogen around ξ = is not near zero, this is a clear indication of differential diffusion. In the mixture fraction range between < ξ <.5, the most sensitive parameters indicating differential diffusion effects are the molecular diffusion parameter z and the H 2 peak value at lower strain rate conditions. Near the stoichiometric mixture fraction ξ st =.35, differential diffusion effects are less apparent and laminar flame calculations from both transport models cross each other in this region. Only the 2 peak values at low strain rate conditions differ according to the calculations. he most sensitive parameter on the fuel-lean side is the atom ratio C /H. When turbulent transport dominates C /H is constant over mixture fraction space with a steep gradient close to ξ =. When differential diffusion is present, a distinct gradient in C /H appears across the stoichiometric mixture fraction into the lean region. Although no calculation can match the experimental profiles at x/d =, due to the presence of the 7

8 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames X +Cx H y multi-component Le = x/d = x/d = 5 x/d = 4 5 s - (multi.), 23 s - (Le = ) X X N X a = s - X O.2.5. X Ray (K) X Ram (K) X -LIF Ray / Ram C 2 Interf C*/H* N/O* z* = H - C Figure 6: Conditional mean values of species mole fractions, temperature, atom ratios, interference level, and molecular diffusion parameter in comparison at x/d =, 5, and 4. Lines in between points are plotted to guide eyes. Laminar flame calculations from CHEMD are denoted by solid lines. Strain rates of the calculations are s and 25 s for Le= (red) and s and 75 s for multi-component (blue) transport. Data at x/d = and 2 were omitted for the sake of clarity. However, they follow the general trend observed for the transition from x/d = to x/d = 4 consistently. 8

9 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames pilot flame condition at mixture fraction ξ =.2, it is evident that turbulent transport is preventing differential diffusion on the fuel-rich side of the flame. However, a trend towards differential diffusion can be observed in between ξ =.5 and.3 in the C /H profile. owards smaller ξ values this trend is interrupted by the mixing with the pilot gases of a fixed C /H ratio, most prominent at the peak at ξ =.2. his picture of no differential diffusion on the fuel-rich side but differential diffusion on the fuel-lean side inverts continuously towards larger downstream locations. At x/d = 4 evidence for differential diffusion is no longer found on the fuel-lean side and the experimental results follow clearly the Le = calculation. However, at ξ =.95, the experimental results depart from the Le = calculation, and begin to follow the multi-component calculation. Since the results from x/d = 4 are as well indicating effects of pyrolysis and/or low temperature premixed reactions at the centerline, as no longer samples at cold gas temperatures were found, the impact of differential diffusion cannot be addressed without ambiguity at this stage. Experimental results from x/d = and 2 follow the trends described above. Accordingly, the observations fall in between the results from x/d = 5 and 4, and hence are closer to the results from x/d = 4 than to the results from x/d =. he experimental results at the downstream locations x/d = 5 and (shown in the Appendix) clearly follow the multi-component calculation on the fuellean side. However, there is no indication for differential diffusion or low temperature premixed reactions apparent at x/d = 2 on the rich side of the flame (shown in the Appendix). In contrast to the results at x/d = 4, these effects on the fuel-rich side are still very small and cannot be determined from the experimental results. Regarding the two competing transport mechanisms and the observed trends described above makes x/d = 2 the likely location of transition from differential diffusion on the fuel-lean side to differential diffusion and/or low temperature premixed reactions on the fuel-rich side. At the same time this typically is the location within the flame where the classical assumption of unity Lewis number (used in a majority of current LES turbulent combustion models) is satisfied within the entire mixture fraction range. 4 Summary and Conclusions Results from measurements obtained in a partially premixed piloted turbulent /air jet with.2 mole fraction in the jet were presented in both physical space and mixture fraction space. Five different axial positions were discussed and compared to each other. he spatial mean and rms profiles of major species, temperature, and mixture fraction showed the influence of the pilot and radial positions of the shear layer and flame brush as well as the mixing progress and increasing and H 2 levels towards higher axial locations. First insights into the flame structure of the investigated /air flame were derived from conditional mean values in mixture fraction space in comparison to laminar flame calculations of various strain rates. he lowest, measured axial location, x/d =, showed the impact of the pilot by a shift of the measured species and temperature profiles at the fuel-rich side beyond the laminar calculations with the highest strain rate. he comparison between different axial position showed the trend towards smaller strain rates with increasing axial position. he relative importance of differential diffusion was investigated by comparing the scalar profiles to laminar calculations based on multi-component and unity Lewis number assumption. A decreasing relevance of differential 9

10 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames diffusion with increasing axial position was found on the fuel-lean side. Up to x/d = 2, no indication of differential diffusion was found on the fuel-rich side. he conditional average picture at x/d = 4 revealed effects of pyrolysis and/or low temperature premixed reactions at the centerline. herefore, at this stage differential diffusion effects could not be isolated without ambiguity and will be the source of future work. Acknowledgements he authors kindly acknowledge financial support by the Combustion Energy Frontier Research Center funded by the US department of Energy, Office of Science. Work performed at Sandia was supported by the Division of Chemical Sciences, Geosciences and Biosciences, Office of Basic Energy Sciences, US Department of Energy. Sandia National Laboratories is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under contract DE-AC4-94-AL85. he help of R. Harmon during the experiments is gratefully acknowledged. References [] International workshop on measurement and computation of turbulent nonpremixed flames (NF), [2] R. S. Barlow. Proc. Combust. Inst., 3 (27) [3] F. Fuest. PhD thesis, echnische Universität Darmstadt, Germany, 2, [4] F. Fuest, R. S. Barlow, J.-Y. Chen, and A. Dreizler. Combust. Flame, 59 (22) [5] J. H. Frank., A. G. Hsu, and J. Kuhl. Proceedings of NF, (2) journal. [6] R. S. Barlow, J. H. Frank, and G. J. Fiechtner. Spring Meeting of the Western States Section/Combustion Institute, (998) journal. [7] R. S. Barlow, J. H. Frank, A. N. Karpetis, and J.-Y Chen. Combust. Flame, 43 (25) [8] R. S. Barlow, M. J. Dunn, M. S. Sweeney, and S. Hochgreb. Combust. Flame, 59 (22) [9] F. Fuest, R. S. Barlow, D. Geyer, F. Seffrin, and A. Dreizler. Proc. Combust. Inst., 33 (2) [] R. S. Barlow and J. H. Frank. Proc. Combust. Inst., 27 (998) [] C. Schneider, A. Dreizler, J. Janicka, and E. P. Hassel. Combust. Flame, 35 (23) [2] A. N. Karpetis and R. S. Barlow. Proc. Combust. Inst., 29 (22)

11 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames [3] R. S. Barlow and A. N. Karpetis. Flow urbul. Combust., 72 (24) [4] R. S. Barlow, H. C. Ozarovsky, A. N. Karpetis, and R. P. Lindstedt. Combust. Flame, 56 (29) [5] CHEMD a one-dimensional laminar flame code. Eindhoven University of echnology. [6] Z. Zhao, M. Chaos, A. Kazakov, and F. L. Dryer. Int. J. Chem. Kinet., 4 (28) 8. Appendix

12 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames X +Cx H y multi-component Le = x/d = X X N X a X O.2.5. X Ray (K) X Ram (K) X -LIF Ray / Ram C 2 Interf C*/H* N/O* z* = H - C Figure 7: Conditional mean and rms values denoted as error bars of species mole fractions, temperature, atom ratios, interference level, and molecular diffusion parameter at x/d =. Laminar flame calculations from CHEMD are denoted by solid lines. Strain rates of the calculations are s and 25 s for Le= (red) and s and 75 s for multi-component (blue) transport. 2

13 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames X +Cx H y multi-component Le = x/d = 5 X X N X a X O.2.5. X Ray (K) X Ram (K) X -LIF Ray / Ram C 2 Interf C*/H* N/O* z* = H - C Figure 8: Conditional mean and rms values denoted as error bars of species mole fractions, temperature, atom ratios, interference level, and molecular diffusion parameter at x/d = 5. Laminar flame calculations from CHEMD are denoted by solid lines. Strain rates of the calculations are s and 25 s for Le= (red) and s and 75 s for multi-component (blue) transport. 3

14 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames X +Cx H y multi-component Le = x/d = X X N X a X O.2.5. X X.5..5 X -LIF.5..5 C 2 Interf Ray (K) Ram (K) Ray / Ram C*/H* N/O* z* = H - C Figure 9: Conditional mean and rms values denoted as error bars of species mole fractions, temperature, atom ratios, interference level, and molecular diffusion parameter at x/d =. Laminar flame calculations from CHEMD are denoted by solid lines. Strain rates of the calculations are s and 25 s for Le= (red) and s and 75 s for multi-component (blue) transport. 4

15 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames X +Cx H y multi-component Le = x/d = 2 X X N X a X O.2.5. X X.5..5 X -LIF.5..5 C 2 Interf Ray (K) Ram (K) Ray / Ram C*/H* N/O* z* = H - C Figure : Conditional mean and rms values denoted as error bars of species mole fractions, temperature, atom ratios, interference level, and molecular diffusion parameter at x/d = 2. Laminar flame calculations from CHEMD are denoted by solid lines. Strain rates of the calculations are s and 25 s for Le= (red) and s and 75 s for multi-component (blue) transport. 5

16 8 th US Combustion Meeting Paper # 7L-64 opic: urbulent Flames X +Cx H y multi-component Le = x/d = 4 X X N X a X O.2.5. X Ray (K) X Ram (K) X -LIF Ray / Ram C 2 Interf C*/H* N/O* z* = H - C Figure : Conditional mean and rms values denoted as error bars of species mole fractions, temperature, atom ratios, interference level, and molecular diffusion parameter at x/d = 4. Laminar flame calculations from CHEMD are denoted by solid lines. Strain rates of the calculations are s and 25 s for Le= (red) and s and 75 s for multi-component (blue) transport. 6