Experimental and computational study of C2H2 and CO in a laminar axisymmetric methane–air diffusion flame

Experimental and computational study of C2H2 and CO in a laminar axisymmetric methane–air diffusion flame

Proceedings of the Combustion Institute Proceedings of the Combustion Institute 31 (2007) 997–1004 www.elsevier.com/locate/proci Experimental and ...

268KB Sizes 1 Downloads 28 Views

Proceedings of the

Combustion Institute

Proceedings of the Combustion Institute 31 (2007) 997–1004

www.elsevier.com/locate/proci

Experimental and computational study of C2H2 and CO in a laminar axisymmetric methane–air diffusion flame A.V. Mokhov a

a,*

, B.A.V. Bennett b, H.B. Levinsky

a,c

, M.D. Smooke

b

Laboratory for High Temperature Gas Kinetics, University of Groningen, Groningen, The Netherlands b Department of Mechanical Engineering, Yale Center for Combustion Studies, Yale University, New Haven, CT 06520-8284, USA c Gasunie Engineering and Technology, Groningen, The Netherlands

Abstract Raman measurements of C2H2 and CO mole fractions in a laminar axisymmetric methane–air diffusion flame are compared with numerical predictions. A high-repetition-rate, high-average-power laser is used to increase signal-to-noise ratio to measure these minor flame species. Computationally, the system of governing equations including detailed chemistry and transport is solved by a damped modified Newton’s method. The calculations predict the measured temperature and nitrogen mole fractions quantitatively. While there is agreement within experimental uncertainty between calculated and measured acetylene concentrations, the calculations predict sharper C2H2 gradients on the lean side of the radial profiles. Adjusting rate of the reaction between C2H2 and OH to values derived in recent experimental and theoretical studies has only a minor impact on the calculated C2H2 profiles. The numerical simulations describe the CO profiles qualitatively, underpredicting the measured CO mole fraction by 40%.  2006 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Spectroscopy; Raman; Flames; Diffusion; Methane

1. Introduction In last two decades, significant progress has been made towards understanding the behavior of axisymmetric laminar diffusion flames. Because of their relatively simple geometry these flames are a suitable object to model subtle interactions between heat and mass transfer and chemical reactions in multidimensional flows, while possessing many properties of flames in practical combustion systems. Combined experimental *

Corresponding author. Fax: +31 50 3634479. E-mail address: [email protected] (A.V. Mokhov).

and computational studies have provided insight into the structure [1–7], and NO [8,9] and soot formation [10–12] in these flames. In these studies, non-invasive laser diagnostics were used to measure temperature and the concentrations of major species and NO, while non-fuel hydrocarbons were measured by extractive probe techniques. Acetylene (C2H2) is one of the key intermediates in hydrocarbon combustion, which plays an important role in the formation of polycyclic aromatic hydrocarbons and soot [13–15]. In the previous studies of axisymmetric laminar diffusion flames, acetylene measurements were performed using extractive probe sampling, which has a serious potential drawback—disturbing the flame.

1540-7489/$ - see front matter  2006 The Combustion Institute. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.proci.2006.08.094

998

A.V. Mokhov et al. / Proceedings of the Combustion Institute 31 (2007) 997–1004

For example, in [5] the diffusion flame was observed to attach to the probe at heights up to 1 cm above the burner exit. Further, a considerable shift in temperature and species profiles between probe and optical measurements was reported in premixed flames [16–18]. Estimating the magnitude of the disturbance arising, for example, from chemical reactions on the probe surface or acceleration of the combustion products into the probe orifice is difficult, and requires experimental verification of the probe results by an independent technique. Acetylene has been extensively investigated in both atmospheric- and low-pressure flat premixed flames. At atmospheric pressure, substantial discrepancies have been observed [19–21] between measured results and those calculated based on the C2H2 submechanism derived from Miller and Melius [22]. Recent ab initio calculations [23] yielded the rate coefficient for the C2H2 + OH reaction close to that derived from the experiments in the flat premixed flames. Because previous numerical simulations were performed with the Miller and Melius C2H2 submechanism, it is interesting to investigate the impact of the revised rate coefficients on the calculated C2H2 profiles in the diffusion flames. Recently, we have reported the measurement of native C2H2 in fuel-rich methane/air flames using spontaneous Raman scattering [18,21]. This method thus provides us with the means to obtain reliable experimental distributions of C2H2, the analysis of which can yield insight into C2H2 formation and destruction in diffusion flames. In this paper we compare the measured and computed profiles of C2H2 mole fractions in an axisymmetric laminar atmospheric-pressure methane/air flame. Towards this end, we use Raman scattering as a non-invasive method for measuring temperature and species concentrations. The computations are performed by solving the governing equations with detailed chemistry and transport. To assess the sensitivity of the calculated C2H2 profiles in this flame to choice of reaction rates, the calculations are performed with the three values for the C2H2 + OH reaction [21,23,24]. In addition to analyzing the distribution of acetylene, we also present results of non-invasive measurements of the distribution of CO in the same flame. Although it is the major intermediate species in hydrocarbon oxidation, there is a dearth of measurements of CO in laminar diffusion flames. That CO is difficult to measure by Raman methods prompted the use of LIF for CO measurement in turbulent flames [25]. Since the Raman frequency for CO is close to that of acetylene, the setup described below offers the possibility of the non-invasive measurement of CO with excellent signal-tonoise ratio.

2. Experimental approach The experimental setup is similar to that described in [7,18]. The burner is an upright stainless tube (455 mm length, o.d. 10 mm and i.d. 7 mm) surrounded by an air-coflow annulus (i.d. 95 mm). The coflow velocity profile is homogenized by flowing the air through a perforated plate after passing a settling chamber 100 mm high filled with glass beads. The exit plane of the fuel tube extends 5 mm above the perforated plate, to allow the small jets to relax to plug flow. The average fuel exit velocity of 130 mm/s has been used in the experiment. A parabolic velocity profile at the exit of the fuel tube was verified by hot-wire anemometry. The velocity of coflow air was set to the average velocity of the fuel jet. The fuel is composed of 40% CH4 and 60% N2. The flows of all gases were controlled by calibrated mass flow controllers (Alicat Scientific); the flow ranges of the controllers were selected to provide accuracy better than 5%. For calibration purposes, nitrogen with a known amount of added C2H2 and CO was flowed through the fuel tube instead of methane–nitrogen mixture. Concentrations and temperature profiles were obtained by moving the burner with a precision positioner (Parker). The Raman setup, used in our previous work [18] was upgraded with a Nd:YLF laser (Spectra Physics Empower) which provides 30 W average power at 528 nm and 5 kHz repetition rate. The advantages of this laser compared to the Nd:YAG lasers traditionally used for flame Raman spectroscopy are the high repetition rate (relatively low pulse energy with high average power) and the comparatively long pulse duration (350 ns). This allows the use of full laser power without causing optical breakdown, resulting in a substantial increase in the signal-to-noise ratio (SNR). The laser beam was focused into the center of the flame by a f = 500 mm lens. The scattered radiation was collected by an f = 300 mm, f/2.8 lens. A half-wave plate installed in rotating positioner was used to rotate the polarization of the laser beam by 90 for measuring the background (see below). The scattered radiation was dispersed by an f/4 spectrometer (Acton Research SpectraPro 2300i) with 2400 mm1 holographic grating, providing a reciprocal linear dispersion of 0.65 nm/mm at 600 nm. The spectrometer was oriented such that its entrance slit was parallel to the laser beam. To utilize the full resolution and throughput of the spectrometer, the sample volume was projected with 0.5 magnification onto the spectrometer entrance slit with 200 lm width. In the exit plane of the spectrometer, an intensified 1024 · 256 pixel CCD camera (Roper Scientific, 26 lm pixel size) was mounted that covered a spectral range of 11 nm and a distance of 13 mm along the laser beam. To increase the

A.V. Mokhov et al. / Proceedings of the Combustion Institute 31 (2007) 997–1004

SNR, the pixels were binned in 8 · 8 groups, which resulted in a net spatial resolution of 0.4 mm. The spectral resolution (0.15 nm) is not significantly deteriorated by binning because the 200 lm entrance slit width of the spectrometer matches the group size. Since the spectral region covered by the spectrometer is not sufficient to measure the Raman scattered light from C2H2 (at 588 nm) and N2 (601 nm) simultaneously, at every height above the burner the measurements were performed twice: with the spectrometer centered at 588.5 and at 597.0 nm. Both spectrometer positions allowed the measurement of CO (595 nm) as well. The Raman spectra were obtained by subtracting the signals measured with the vertical and horizontal polarizations of the exciting laser beam. In this arrangement, depolarized fluorescence of C2 and polycyclic hydrocarbons is effectively removed, while the Raman signal remains practically unchanged due to the low depolarization ratios of the Raman transitions used [26,27]. Both signals were measured by taking 20 gated on every laser pulse accumulations with 25 s (C2H2) and 5 s (N2) exposure time, which resulted in a measurement time of approximately 30 min for one axial distance. The temperature was derived by fitting the measured Raman nitrogen spectrum. Pressure and Doppler broadening are neglected in the synthesized N2 spectrum, which is justified at the moderate resolution of the experimental setup. The parameters of the apparatus function were determined from measurements in dry air at room temperature. The spectral sensitivity of the spectrometer/detector combination was calibrated by a tungsten lamp with known color temperature. When fitting the N2 spectrum, the nitrogen mole fraction was also fitted. The species concentrations were derived from the Raman measurements by comparison of the frequency-integrated Raman intensity from the flame measurements with the integrated intensity from the known mole fractions of N2, C2H2 and CO at room temperature, applying density and Boltzmann corrections using the double-harmonic approximation for the intensity of the Raman vibrational transitions [28]. The difference in N2 concentrations derived by two methods was not larger than 3%, which is significantly less than the uncertainties in the measured CO and C2H2 mole fractions. The accuracy of the Raman measurements is estimated by comparing the measured and equilibrium temperature and concentrations in a near-adiabatic, fuelrich (u = 1.3) flat premixed methane/air flame. The comparisons for temperature and N2 concentration yielded differences of less than 50 K and 0.03 absolute mole fraction, respectively. The difference between the measured and equilibrium CO mole fractions is substantially higher, 20%. It is reasonable to assume the same accuracy for the measured acetylene mole fractions in diffusion

999

flames, where the impact of the estimated 10 times lower C2H2 mole fraction on the size of the Raman signal will be compensated by the almost 10 times higher C2H2 Raman cross-section. 3. Numerical model and method of solution The model considers an unconfined laminar flame in which a cylindrical fuel stream is surrounded by a coflowing oxidizer jet. At the burner surface (bottom domain boundary), the jet radii, compositions, and velocities are chosen to match the experiment; at the far radial and far axial boundaries, free-stream boundary conditions are employed. The gas-phase conservation equations (conservation of mass, momentum, energy, and individual species mass) are written using a vorticity–velocity formulation, resulting in a set of Nsp + 4 elliptic partial differential equations, where Nsp is the number of gas-phase species [2]. The chemical mechanism in the present study is GRI-Mech 3.0 with nitrogen chemistry removed, which has 35 species and 217 reversible reactions [24]. All thermodynamic, chemistry, and transport properties are evaluated using CHEMKIN [29] and TPLIB [30,31] routines, some of which have been rewritten and restructured for greater speed [32]. The effect of gas-phase radiation in the optically thin limit is included via a radiation submodel in which H2O, CO, and CO2 are the significant radiating species [33,34]. The details of the numerical solution method have been presented elsewhere [2], and only the essential features are outlined here. The governing equations and boundary conditions are discretized on a two-dimensional tensor-product grid using a finite difference technique on a nine-point stencil. The resulting set of strongly coupled, highly non-linear equations is written in residual form and is solved using a damped modified Newton’s method, in which the linearized equations within each Newton iteration are solved using preconditioned (block Gauss-Seidel) Bi-CGSTAB. The Jacobian matrix is only periodically evaluated, as determined by monitoring a set of theoretical estimates, thus increasing the overall efficiency of the algorithm. Pseudo-transient continuation is employed to ease convergence from an initial guess, with the pseudo-time steps chosen adaptively based on estimates of temporal discretization errors; after a specified number of pseudotime steps, the steady-state problem itself is solved. All computations were done on a non-uniform grid containing 107 grid lines in the radial direction (minimum spacing 0.02 cm) and 148 grid lines in the axial direction (minimum spacing 0.015 cm), for a total of 15,836 points. The size of the domain was 10 cm radially and 20 cm axially. To obtain this level of resolution on an equispaced grid, approximately 106 grid points

1000

A.V. Mokhov et al. / Proceedings of the Combustion Institute 31 (2007) 997–1004

would be needed. The computations, which required approximately 2 GB of RAM, were performed on a 2.0-GHz Opteron processor. 4. Results and discussion The overall flame structure is illustrated by two-dimensional false-color plots of computed temperature, C2H2 and CO distributions in Fig. 1. As can be seen, the flame is stabilized close to the burner exit, with the typical high-temperature ‘‘wishbone’’ structure of jet flames [1,7] resulting from deficiency of oxygen in the fuel. The two high-temperature regions of the wishbone, with maximum temperatures of 1800 K, merge at 3.5 cm above the burner. Beyond this point, the axial temperature gradually falls to 800 K at axial distance z = 10 cm. We point out that this maximum temperature is far below the adiabatic temperature of 2075 K calculated for the stoichiometric mixture of 40% CH4/60% N2 and air; this indicates that the rate of mixing of the combustion products with the coflow air is faster than the rate of heat release during burnout of residual H2 and CO. Acetylene and carbon monoxide are formed in the high-temperature zone, and are completely consumed by z = 3 cm. Although fast enough to keep the hot gases from reaching the adiabatic stoichiometric temperature, the mixing rate is apparently not so fast as to impair the burnout of C2H2 and CO. Because of the relatively small axial extent of the region in which measurable quantities of C2H2 and CO are expected, the measurements were performed only at z = 9, 14, 19, 24 and 29 mm. As an example, false-color images of the Raman spectra measured at z = 14 mm are presented in Fig. 2, where the left and right images show the intensity distributions when the spectrometer was centered at 588.5 and 597.0 nm, respectively. In these images, the vertical direction represents the spatial coordinate, while the spec-

10

T

8

1826

10

C2H2

8

1571

10

0.0035 0.0030

CO

8

1062

6

4

807

4

2

553

2

0.0020 0.0015

6 0.015 4 0.010

0.0010 2

0.005

0 -2 -1 0 1 2 R, cm

0.0

0.0005 0 -2 -1 0 1 2 R, cm

298

0 -2 -1 0 1 2 R, cm

0.0

0.025 0.020

Z, cm

6

Z, cm

Z, cm

0.0025

0.030

Fig. 1. Two-dimensional false-color plots of computed temperature, C2H2 and CO distributions in a portion of the computational domain.

Fig. 2. Two-dimensional false-color plots of measured Raman spectra at z = 14 mm. Spectrometer is tuned to 588.5 (left image) and 597.0 nm (right image). Figures below images show spectral intensity distribution along white stripe at R = 0.

tral coordinate extends horizontally. The spectral intensity distributions at the burner axis are shown below the images. The right image is dominated by the N2 Raman spectrum, while the feature corresponding to the CO Raman transition is barely distinguishable. Since temperatures are rarely derived from the shape of the N2 Raman spectrum, we trace qualitatively the form of the spectrum along a radial profile of temperature from the calculations presented in Fig. 1, as an illustration. It is narrow and intense outside the flame (the bottom of the figure), where the temperature is low, then becomes less intense and broadens when passing though the high-temperature region in the ‘‘wishbone’’, and becomes more intense and narrows at the centerline, reflecting the fact that at this height the temperature maximum is positioned off axis. The N2 (v 0 = 1  v00 = 2) hot-band transition is clearly seen in the figure, showing that the spectrometer resolution is sufficient to derive temperatures from the measured Raman spectrum. In the left image, which was acquired with a 5 times longer exposure time, the two spectral features corresponding to CO and C2H2 are clearly visible. Both features are spectrally broader at a radial position r = 2 mm than in the center of the flame, indicating the elevated temperature in this zone, similar to that observed in the N2 spectra. As can be seen from the intensity distribution at the axial centerline, the CO spectrum shows typical features of the high-temperature Raman spectrum of a diatomic molecule while the C2H2 Raman feature is substantially broader. The radial profiles of temperature and N2 mole fraction derived from the nitrogen Raman spectra are presented in Fig. 3, together with the results of the computation. When evaluating the experimental data, the temperature was also derived from the measured CO Raman spectra. These temperatures (not shown to avoid clutter in figures) were very close to those derived from the nitrogen spectrum, but substantially noisier. The sequence of measured temperature profiles as a function of

A.V. Mokhov et al. / Proceedings of the Combustion Institute 31 (2007) 997–1004

0.90

1400

0.85

1200 1000

0.80 0.75

800

0.70

T Exp. T Calc. GRI Mech 3.0

600

0.65

N2 Exp. N2 Calc. GRI Mech 3.0

400

0.60

1800 1600

0.95 0.90

Z = 24 mm

1400

0.85

1200 1000

0.80 0.75

800

0.70

T Exp. T Calc. GRI Mech 3.0

600 400

0.65 0.60

N2 Exp. N2 Calc. GRI Mech 3.0

200

0.55 0.95

Z = 19 mm

1600

0.90

1400

0.85

1200

0.80

1000

0.75

800

0.70

T Exp. T Calc. GRI Mech 3.0

600

0.65

N2 Exp. N2 Calc. GRI Mech 3.0

400

Mole fraction

Temperature, K

1800

0.60

200

0.55

1800 Temperature, K

Mole fraction

0.55

0.95

Z = 14 mm

1600

0.9

1400

0.85

1200

0.8

1000

0.75

800

0.7

T Exp. T Calc. GRI Mech 3.0 N2 Exp. N2 Calc. GRI Mech 3.0

600 400

Mole fraction

Temperature, K

200

0.65 0.6

200

0.55

1800 Temperature, K

Mole fraction

0.95

Z = 29 mm

1600

0.95

Z = 9 mm

1600

0.9

1400

0.85

1200

0.8

1000

0.75

800

T Exp.

0.7

600

T Calc. GRI Mech 3.0

0.65

400

N2 Exp.

0.6

N2 Calc. GRI Mech 3.0

200 0

1

2

3

4

5

6

Mole fraction

Temperature, K

1800

0.55 7

8

9

Distance, cm

Fig. 3. Radial temperature and N2 profiles at different axial distances. Lines denote calculations; symbols denote measured temperature (squares) and nitrogen mole fraction (triangles).

axial distance follows the above description of the temperature field well. The mixing process is also illustrated by the growth in the N2 mole fraction along the centerline (r = 0). At z = 9 mm, the N2 mole fraction of 0.65 is still close to that in the fuel mixture; at z = 14 mm, the mole fraction is above 0.7, while the gas temperature is still only

1001

1200 K. At z = 29 mm, where the temperature is close to its maximum value, the N2 mole fraction is already close to that in air. The predictive capability of the numerical approach and chemical mechanism is illustrated by the quantitative comparisons with the experimental results shown in the figure. The computed nitrogen profiles reproduce the experimental results nearly quantitatively. Very good agreement is observed between the measured and calculated temperatures as well, with the exception that the calculations predict slightly higher temperatures (<100 K) on the ‘‘lean side’’ (towards larger radial distance) of the maximum of the temperature profiles. It is interesting to note that a similar result was observed in hydrogen flames [7] in the same burner arrangement. While this discrepancy is observed at all heights, the equivalent spatial discrepancy in the radial coordinate (thus the error in the width of the temperature profile) is roughly on the order of the spatial resolution of the measurements. As an intermediate in the oxidation of the fuel, we expect acetylene to be found in the high-temperature regions of the flame, and thus to follow the development of the temperature field. The measured acetylene profiles do so, as shown in Fig. 4. At z = 9 and 14 mm, the C2H2 profiles show the same non-monotonic behavior as the temperature, with local minima at the center, and maxima at r = 3.5 mm and r = 3.0 mm for 9 and 14 mm axial distance, respectively. As expected, these maxima are closer to the centerline (and thus more fuel rich) compared to the maximum radial temperatures, as can be seen by comparison with Fig. 3. At z = 19 mm the measured C2H2 profile has rather a Gaussian form around the centerline, with a maximum mole fraction of 3000 ppm. The maximum C2H2 grows further downstream to 3500 ppm at z = 24 mm, and then decreases rapidly. At z = 29 mm, the C2H2 concentration is equal to or even less than the detectability limit (500 ppm) of the present experimental setup. Figure 4 also shows acetylene profiles calculated with different rates of reaction between C2H2 and OH. As described above, the calculations were first performed with the unchanged rates from Miller and Melius [22], which are incorporated in GRI-Mech 3.0. In the second run the rate coefficient of the C2H2 + OH fi H + CH2CO reaction, derived from the measurements of acetylene in flat premixed flames [21], was used. Lastly, the reaction rates calculated in [23] for this reaction were used. As can be seen from Fig. 4, these variations have relatively little impact on the calculated profiles. This behavior can be explained from the C2H2 production rate analysis. In calculations using GRI Mech 3.0, the reaction between C2H2 and O atoms is fast enough to be the main channel for acetylene oxidation, while the reaction

A.V. Mokhov et al. / Proceedings of the Combustion Institute 31 (2007) 997–1004 500 450 400 350 300 250 200 150 100 50 0

Z = 29 mm Mole fraction, ppm

Mole fraction

1002

4000

Z = 24 mm

Z = 24 mm

35000

3000

Mole fraction, ppm

Mole fraction

Z = 29 mm

40000

3500 2500 2000 1500 1000 500 0

30000 25000 20000 15000 10000 5000 0

3500 3000

40000

Z = 19 mm

Z = 19 mm

35000

2500

Mole fraction, ppm

Mole fraction

20000 18000 16000 14000 12000 10000 8000 6000 4000 2000 0

2000 1500 1000 500 0

30000 25000 20000 15000 10000 5000 0

3500

40000

Z = 14 mm

35000

2500

Mole fraction, ppm

Mole fraction

3000

2000 1500 1000 500 0

Z = 14 mm

30000 25000 20000 15000 10000 5000 0

2500

Z = 9 mm

40000

Z = 9 mm

35000

2000 1500 1000 500 0 0

1

2

3

4

5

6

7

8

9

Distance, mm

Mole fraction, ppm

Mole fraction

3000

30000 25000 20000 15000 10000 5000 0

Fig. 4. Radial profiles of C2H2 at different axial distances. Squares denote measured mole fractions; lines denote calculations with different rates of C2H2 + OH reaction (solid, [24]; dashed, [21]; dotted, [23]).

between C2H2 and OH is of little impact. Increasing the rate of C2H2 + OH reaction can only increase an already fast rate of C2H2 removal. All three calculations show good agreement with measured profiles, when considering the experimental uncertainty. As discussed, the discrepancy

0

1

2

3

4

5

6

7

8

9

Distance, mm

Fig. 5. Radial profiles of CO at different heights above the burner. Squares denote measured mole fractions; solid lines denote calculations.

at z = 29 mm is clearly due to the detection limit of the experimental method. At z = 9, 14 and 19 mm, the calculations show a steeper C2H2 gradient on the lean side of the flame than the experiments. Evidence that the more gradually

A.V. Mokhov et al. / Proceedings of the Combustion Institute 31 (2007) 997–1004

changing measured profiles at these heights are not the result of poor spatial resolution is provided by the profile at z = 24 mm, where the calculated and measured gradients are similar. The milder gradients of the measured C2H2 profiles imply that the mixing zone with high concentrations of oxygen atoms and hydroxyl is broader than predicted by the calculations. Measurements of OH and O are needed to clarify this issue. The differences between the calculated and measured CO profiles are larger than those for C2H2. As can be seen from Fig. 5, the calculations follow the measured CO profiles qualitatively, but underpredict the measured values by roughly 40%. We consider it unlikely that these differences arise solely from experimental uncertainty in the CO measurements. An extra verification of the reliability of the measurements was performed by determining CO mole fractions by two different methods. In one method, they were based on the calibration in cold N2 with known CO concentration. The other method derived the CO mole fractions from the ratio of the N2 and CO Raman intensities in the flame. Both methods yielded the same CO mole fractions. A similar discrepancy between numerical simulations and Raman measurements was observed in a flame with a fuel composition close to that used in present work [5]. The reason for the discrepancy is unclear, and requires further investigation.

5. Conclusions Mole fractions of C2H2 and CO in an axisymmetric laminar methane–air diffusion flame have been measured by Raman spectroscopy and computed by solving the governing equations including detailed chemistry and transport. Using a high-repetition-rate, high-average-power laser yielded sufficient signal-to-noise ratio to perform non-invasive measurements of acetylene in this flame. Agreement between measured and computed temperature and N2 mole fractions is excellent. Experimental/computational agreement to within the experimental uncertainty is observed for C2H2 concentrations, with the calculations predicting sharper C2H2 gradients on the lean side of the radial profiles. Although the calculations underpredict measured CO mole fraction by 40%, they predict the qualitative trends in CO correctly.

Acknowledgments A.V.M. and H.B.L. are thankful to the Dutch fund for Ecology, Economy and Technology (EET) for financial support of this work. B.A.V.B. and M.D.S. gratefully acknowledge the

1003

support of the US Department of Energy Office of Basic Energy Sciences, under Contract DEFG02-88ER13966.

References [1] M.D. Smooke, P. Lin, K. Lam, M.B. Long, Proc. Combust. Inst. 23 (1990) 575–582. [2] M.D. Smooke, A. Ern, M.A. Tanoff, B.A. Valdati, R.K. Mohammed, D.F. Marran, M.B. Long, Proc. Combust. Inst. 26 (1996) 2161–2170. [3] B.A.V. Bennett, J. Fielding, R.J. Mauro, M.B. Long, M.D. Smooke, Combust. Theory Model. 3 (4) (1999) 657–687. [4] B.A.V. Bennett, C.S. McEnally, L.D. Pfefferle, M.D. Smooke, Combust. Flame 123 (4) (2000) 522–546. [5] C.S. McEnally, L.D. Pfefferle, A.M. Schaffer, M.B. Long, R.K. Mohammed, M.D. Smooke, M.B. Colket, Proc. Combust. Inst. 28 (2000) 2063– 2070. [6] B.A.V. Bennett, C.S. McEnally, L.D. Pfefferle, M.D. Smooke, M.B. Colket, Combust. Flame 127 (1–2) (2001) 2004–2022. [7] V.V. Toro, A. Mokhov, H.B. Levinsky, M.D. Smooke, Proc. Combust. Inst. 30 (2005) 485–492. [8] N. Sullivan, A. Jensen, P. Glarborg, M.S. Day, J.F. Grcar, J.B. Bell, Combust. Flame 131 (3) (2002) 285–298. [9] J.B. Bell, M.S. Day, J.F. Grcar, W.G. Bessler, C. Schulz, P. Glarborg, A.D. Jensen, Proc. Combust. Inst. 29 (2002) 2195–2202. [10] M.D. Smooke, C.S. McEnally, L.D. Pfefferle, R.J. Hall, M.B. Colket, Combust. Flame 117 (1–2) (1999) 117–139. [11] K.T. Walsh, J. Fielding, M.D. Smooke, M.B. Long, Proc. Combust. Inst. 28 (2000) 1973–1979. [12] M.D. Smooke, M.B. Long, B.C. Connelly, M.B. Colket, R.J. Hall, Combust. Flame 143 (4) (2005) 613–628. [13] J. Warnatz, H. Bockhorn, A. Mozer, H.W. Wenz, Proc. Combust. Inst. 19 (1982) 197–209. [14] P. Lindstedt, Proc. Combust. Inst. 27 (1998) 269– 285. [15] H. Richter, J.B. Howard, Prog. Energy Combust. Sci. 26 (4–6) (2000) 565–608. [16] R.J. Cattolica, S. Yoon, E.L. Knuth, Combust. Sci. Technol. 28 (5–6) (1982) 225–239. [17] A.T. Hartlieb, B. Atakan, K. Kohse-Ho¨inghaus, Combust. Flame 121 (4) (2000) 610–624. [18] A.V. Mokhov, S. Gersen, H.B. Levinsky, Chem. Phys. Lett. 403 (4-6) (2005) 233–237. [19] E.W. Kaiser, J. Phys. Chem. 94 (11) (1990) 4493– 4499. [20] E.W. Kaiser, T.J. Wailington, M.D. Hurley, J. Platz, H.J. Curran, W.J. Pitz, C.K. Westbrook, J. Phys. Chem. A 104 (35) (2000) 8194–8206. [21] S. Gersen, A. Mokhov, H.B. Levinsky, Combust. Flame 143 (3) (2005) 333–336. [22] J.A. Miller, C.F. Melius, Proc. Combust. Inst. 22 (1988) 1031–1039. [23] J.P. Senosiain, S.J. Klippenstein, J.A. Miller, J. Phys. Chem. A 109 (27) (2005) 6045–6055. [24] G.P. Smith, D.M. Golden, M. Frenklach, et al., available at .

1004

A.V. Mokhov et al. / Proceedings of the Combustion Institute 31 (2007) 997–1004

[25] R.S. Barlow, A.N. Karpetis, Proc. Combust. Inst. 29 (2002) 1929–1936. [26] C.M. Penney, L.M. Goldman, M. Lapp, Nat-Phys. Sci. 235 (58) (1972) 110–112. [27] F. Orduna, C. Domingo, S. Montero, W.F. Murphy, Mol. Phys. 45 (1) (1982) 65–75. [28] D.A. Long, The Raman Effect, Wiley, Chichester, 2002. [29] R.J. Kee, F.M. Rupley, J.A. Miller, CHEMKIN II: A Fortran Chemical Kinetics Package for the Analysis of Gas-Phase Chemical Kinetics, Sandia National Laboratories, 1989. [30] R.J. Kee, J. Warnatz, J.A. Miller, A fortran computer code package for the evaluation of gas-phase viscosities, conductivities, and diffusion

[31]

[32]

[33] [34]

coefficients, Sandia National Laboratories, 1983. R.J. Kee, G. Dixon-Lewis, J. Warnatz, M.E. Coltrin, J.A. Miller, A Fortran Computer Package for the Evaluation of Gas-Phase, Multicomponent Transport Properties, Sandia National Laboratories, 1986. V.M. Giovangigli, N. Darabiha, in: C.-M. Brauner, C. Schmidt-Laine (Eds.), Mathematical Modelling in Combustion and Related Topics, Nijhoff, Dordrecht, 1988, p. 491. R.J. Hall, J. Quant. Spectrosc. Radiat. Transfer 49 (5) (1993) 517–523. R.J. Hall, J. Quant. Spectrosc. Radiat. Transfer 51 (4) (1994) 635–644.

Comments T.S. Cheng, Chung Hua University, Taiwan. The Raman signal of C2H2 is quite low compared to O2 and N2 Raman signals. What is the S/N ratio of C2H2 measurement? Reply. In the present Raman setup we could reach signal-to-noise ratio of 5 at acetylene mole fraction of 3000 ppm and gas temperature of 1800 K. At lower acetylene concentrations (<500 ppm) the signal and noise were of the same order. The signal-to-noise ratio can be significantly improved by using more powerful laser source.

d

Russ Fitzgerald, Saint Louis University, USA. How good is the assumption that the media is optically thin? Perhaps, this is the reason for large discrepancy in CO temperatures and concentrations? It would be advisable to at least measure the absorption to verify the optically firm limit. Reply. The small size of our diffusion flame justifies treatment of radiative heat transfer in the optically thin limit. The excellent agreement we observe between measured and calculated flame temperatures further supports this assumption.