Sunteți pe pagina 1din 36

This article was downloaded by: [Purdue University] On: 12 February 2014, At: 15:57 Publisher: Taylor &

Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Combustion Science and Technology


Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/gcst20

A Study of Flame Observables in Premixed Methane Air Flames


H. N. NAJM , O. M. KNIO , P. H. PAUL & P.S. WYCKOFF
a b a b a a

Sandia National Laboratories , Livermore, CA, 94551

The Johns Hopkins University , Baltimore, MD, 21218 Published online: 05 Apr 2007.

To cite this article: H. N. NAJM , O. M. KNIO , P. H. PAUL & P.S. WYCKOFF (1998) A Study of Flame Observables in Premixed Methane - Air Flames, Combustion Science and Technology, 140:1-6, 369-403, DOI: 10.1080/00102209808915779 To link to this article: http://dx.doi.org/10.1080/00102209808915779

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions

Combust, Sci, and Tech., 1998, Vol. 140, pp. 369-403

1998OPA (Overseas Publishers Association) N.V.


Published by license under the Gordon and Breach Science Publishers imprint. Printed in Malaysia.

Reprints availabledirectly from the publisher Photocopyingpermitted by licenseonly

A Study of Flame Observables in Premixed Methane - Air Flames


Downloaded by [Purdue University] at 15:57 12 February 2014
H. N. NAJM *, O. M. KNIO b , P.H. PAUL and P. S. WYCKOFF
Sandia Nationa/ Laboratories, Livermore, CA 94551; b The Johns Hopkins University, Baltimore, MD 21218

(Received 25 June 1998; /n tina/form 30 September 1998)

The use of particular experimental flame observables as flame markers, and as measures of flame burning and heat release rates requires the establishment of robust correlations between the particular observable and the rate in question. In this work, we use a compilation of results from numerical computations of the interaction of a premixed methane flame with a twodimensional counter-rotating vortex pair using detailed kinetics. The data set involves the use of two different chemical mechanisms, a two-fold variation in flow time scales, and the examination of both stoichiometric and rich methane flames. Correlations between a number of flame observables and heat release and burning rates are examined. We study HCO, '\l . v, OH, CH, CO, CH 3 , CH 20, CHi, and C 2H 2, as well as various concentration products (surrogates for production rates) including [OHJ(CH 20), (OHJ(CH4) , and [OHJ(CO). Other concentration products expected to relate to chemiluminescent observables such as CH., OH and CO; are also studied. HCO mole fraction is found to have the best correlation with flame burning and heat release rates for all cases studied. Results suggest that significant scatter due to flow unsteadiness is expected from correlations of peak 'V. v, CO mole fraction gradient, C2H 2 mole fraction, and CH with heat release. Changes in stoichiometry are found to adversely affect the correlation expected from peak CO, OH, OH gradient, CH, CH 3, and [OH][CH,O). Little scatter is observed in the [OH][CH 20) data, highlighting its utility in the absence of significant variation of reactants composition. We observe evidence of useful correlations of peak [OH][CH 4) and [OH][CO). Concentration products of the precursors of OW and COi are also found to correlate well with peak heal release rate. Peak CH,O data is found to have good correlation with peak burning and heat release rates, with small scatter, and little correlation shift due to changes in reactants composition.
Keywords: Flame; observables; methane; HCO; heat release; premixed

Address for correspondence: Sandia National Laboratories, P.O. Box 969, MS 9051, Livermore, CA 94551. Tel./Fax:. (925) 294-2054/2595, e-mail: hnnajm@ca.sandia.gov
369

370

H. N. NAJM et al.

INTRODUCTION

Proper evaluation of the performance of combustion systems requires accurate experimental determination of local flame burning and heat release rates. However, these rates are not directly measurable. Rather, specific observable flame parameters are typically measured and used to infer their values. This necessary approach assumes an implicit correlation between the particular flame observable and the rate quantity in question. Detailed computations of flame-flow interaction in unsteady multidimensional flow are useful for investigating the quality of these correlations, and for highlighting the utility of specific observables that are found to offer good correlations. This was done recently by Najm et al. [43], where the adequacy of certain flame observables as measures of flame burning and heat release rates was evaluated using both numerical and experimental data. Numerical results were based on a computed interaction between a freely propagating premixed methane-air flame and a two-dimensional (2D) counter-rotating vortex pair, using a C, chemical mechanism [64] under stoichiometric conditions. Experimental results were based on Planar Laser Induced Fluorescence (PLIF) measurements in a V-flame geometry, where a 2D vortex-pair is generated and allowed to impinge into one leg of the V-flame. One flame observable considered in [43] is the local flow dilatation rate V'. v. Mungal et al. [40] determined V'. v experimentally from 2D PlV velocity field measurements and used it as a flame marker. Peak V'. v has also been used as a measure of flame heat release rate [37]. Najm et al., demonstrated that peak V'. v is a good measure of peak heat release rate under steady state ID flame conditions. On the other hand, their results indicated that this correlation fails under conditions of high unsteadiness with significant strain-rate and curvature disturbances to internal flame structure. This is due to the contribution of the heat diffusion term to V'. v. Another commonly used flame observable is the mole fraction of OH. Using PLIF, with a judicious choice of excitation and detection wavelengths, it is possible to obtain PLIF images of OH that essentially represent mole fraction [46]. Computed peak OH mole fraction has been found to correlate reasonably with peak heat release rate Wr away from regions of high curvature [43]. In highly curved regions, however, it was practically insensitive to variations in Wr. Other commonly used measurements include PLIF of CH [46], and chemiluminescence measurements of the electronically excited states CH*, Ci, and OH* [5, 14, 15]. Based on (I) analysis of carbon flux along reaction pathways from CH 4 to CO 2 in one-dimensional (ID) flames using the

Downloaded by [Purdue University] at 15:57 12 February 2014

FLAME OBSERVABLES

371

GRImech1.2 [19] mechanism, and (2) the experimental findings of [43,46], sufficient reason has been found to question the utility ofCH, as wel1 as OH', and CH' mole fractions as measures of flame burning and heat release rates [43). On the other hand, numerical results in [43) indicated that the mole fraction of HCO exhibits excel1ent correlation with burning and heat release rates, which suggested its utility as a flame marker and as an experimental measure of these rates. HCO production is directly dependent on the concentration of CH 20, which is in turn directly dependent on the reaction CH 3 + 0 ~ CH 20 + H, which exhibits the largest fractional influence on changes in heat release rates. Experimental HCO PLIF measurement was also demonstrated in [43], based on the utilization of recent HCO spectroscopy data. More recently, Paul and Najm [49] found that both peak HCO concentration and the concentration product [OHJ[CH20] are good measures of flame heat release rate IVT at either stoichiometric or rich conditions. This concentration product provides an estimate of the production rate of HCO by the reaction CH 20 + OH ~ HCO + H 20. Given the fast removal rate of HCO, its mole fraction is roughly proportional to its production rate. The established link between HCO mole fraction and heat release rate leads to the good correlation of [OH][CH 20] with IVT. The relatively easier task of PLIF imaging of OH and CH 20 concentrations, versus that of HCO, highlights the utility of the concentration product approach. In the present work, we extend the above studies to investigate these and other flame observables under a wider range of flame, flow, and model parameters. We investigate correlations of HCO, 'V. v, OH, CH, CO, CH 3 , CH 20, CH;, and C 2H 2 with flame heat release and burning rates. We also evaluate correlations of peak concentration products [OH][CH 20], [OH][CO], [OH][CH 4 J, and others associated with the production of excited chemiluminescent species OH', CH', and CO; with peak IVT. The parameter range considered includes a two-fold variation in vortex-pair strength, and the utilization of stoichiometric and rich reactants mixtures. We use both C, and C\C 2 kinetic mechanisms to enable the investigation of the above ranges of species and stoichiometry, and to cross-compare predictions from the two mechanisms. The computations focus on the interaction of a freely propagating premixed methane-air flame with a 2D counter-rotating vortex pair. This flow has been investigated both experimentally [24,29,38,39,46,49, 53,55,58] and numerically [1,23, 33,41 - 45,52, 57]. The available experimental data includes flow-visualization of the interaction process between the vortexpair and the flame, quantitative PlY measurements of velocity, vorticity,

Downloaded by [Purdue University] at 15:57 12 February 2014

372

H. N. NAJM et al.

and dilatation, as well as PLIF measurements of OR, CR, CR 20 , and RCO. Available numerical results include flame response based on both single-step and detailed kinetics, along with computed flow and vorticity dynamics, and the generation of baroc1inic vorticity. In the present work we shall not re-examine the flow and flame dynamics observed in this flow, rather we focus on the investigation of flame observables. We proceed next to give a brief description of the model before discussing the results.

Downloaded by [Purdue University] at 15:57 12 February 2014

MODEL FORMULATION

The governing equations are presented in their non-dimensional form in 2D. The assumptions of zero bulk viscosity [62], negligible body forces, and low Mach number [32] give the conservative continuity and momentum equations:

al + V'.

ap

(pv) = 0

(I)

(2)

(3)

where p is the density, v = (u, v) is the velocity vector, P is the pressure, Re is the Reynolds number, and <I>x, <I>y are the viscous stress terms. We assume a detailed 'chemical reaction mechanism involving N species and M elementary reactions. The energy equation is developed allowing for variable transport properties, and a constant stagnation pressure Po, i.e., an open domain. We neglect Soret and Dufour effects [68], since they are not expected to play a significant role in hydrocarbon flames (versus say Hydrogen flames). We also neglect radiant heat transfer and soot formation. Minimal soot production is expected under the present Nrdiluted flame conditions. This, and the small size of the flow domain suggest a minor role for radiation. We also assume a perfect gas mixture, with individual species molecular weights, specific heats, and enthalpies offormation, using Fickian

FLAME OBSERVABLES

373

binary mass diffusion. The resulting low-Mach-number energy equation is: 8T = -v."\7T + I "\7. (>'"\7T) 8t RePr pCp

+ _I_Z. "\7T+ Da WT
ReSc
cp

(4)

pCp

Downloaded by [Purdue University] at 15:57 12 February 2014

'where T is the temperature, >. is the thermal conductivity, WT = h.w, is the chemical heat release rate source term, hi is enthalpy and Wi is the production rate of species i, cp is the mixture specific heat at constant pressure, Pr, Sc, and Da are the Prandtl, Schmidt, and Damkohler numbers respectively, and Z = 'L:I Cp,iDiN"\7Yi. The N-th species, here N z, is assumed dominant such that the diffusion velocity of any other species i in the mixture is approximated by Vi = -DiN "\7 Y i/ Y i, where DiN is the binary mass diffusion coefficient of species i into the N-th species, and Y i is the mass fraction of species i. VN is found from the identity YiV i = O. For computational efficiency, mixture transport properties (J.L, >') are set to those of the dominant species at the local temperature. Using GRlmechl.2 [19], we find that this simplified transport model, and the present numerical discretization, lead to less than 9% deviation in peak radical mole fractions in the ID freely-propagating stoichiometric 20% Nj-diluted methane-air flame, as compared to the solution using Chemkin [27,26]. The i-th species conservation equation, for i = 1, ... , N - I, is written as

'L:,

'L:,

8(pYi)

= -"\7. (pvY;) + ReSc"\7 (pDiN"\7Yi) + Da Wi,

(5)

and the mass fraction Y N is found from the identity Y; = I. The perfect gas state equation is: Po = pT/ W, where W = 1/ (Yi/ Wi), is the local effective molar mass of the mixture. The production rate for each species is given by the sum of contributions of elementary reactions [68], with Arrhenius rates r = AkTb'e-E,/RT, k = I, ... , M, including forward and backward rates, and third-body efficiencies [27]. Finally, for the purpose of the numerical implementation described below, the time rate of change of density is found by differentiating the state equation,

'L:,

'L:,

8p = p( 8t

_.!.. 8T _
T 8t

wt-I
;=1

8Yi) Wi 8t

(6)

with 8T/8t from Eq. (4), and 8Yi/8t = [8(pYi)/8t + Y i"\7 '(pv)]/p from Eq. (5).

374

H. N. NAJM et al.

NUMERICAL SCHEME

Downloaded by [Purdue University] at 15:57 12 February 2014

The above equations are solved using a second-order predictor-corrector finite difference projection scheme. The projection method was first introduced by Chorin [10], and is discussed more recently by Kim and Moin [28]. McMurtry et al. [36] presented a formulation for reacting flow, using the continuity equation to update the density field. The present method follows more closely that of [31,57], in using the energy equation. Depending on the chemical mechanism considered, and the associated stiffness of the resulting governing equations, a stiff or non-stiff version of this scheme is used. The non-stiff version, which employs a second-order Runge Kutta (RK2) formulation in the corrector step, is presented in detail in [44], and has been used to study premixed methane-air flames, with C I kinetics, under stoichiometric conditions [43,44]. The stiff version, which employs an additive (non-split) semi-implicit formulation of the scalar conservation equations in the corrector step, is discussed in [45]. A brief outline of both implementations is provided below. , An open 2D rectangular domain is considered, and is overlaid by a uniform mesh. Velocity components are evaluated at cell edges, while other fields are evaluated at cell centers. Spatial derivatives are discretized using second-order central differences, and a second-order Adams-Bashforth (AB2) scheme is used for explicit time integration in the predictor. The numerical solution for each time step, from I" to I n + I, involves the following steps [45]: Predictor Evaluate predicted pn+1 and Y7+ ', using AB2 integration of Eqs. (5) and (6). Evaluate predicted T,,+I from the state equation. Evaluate predicted vn + 1 using a variable-density projection scheme [44]. 2 Corrector Evaluate scalar time gradients, the right-hand-sides of Eqs. (5) and (6), from the predicted fields at 1,,+ I. Non-stiff: Evaluate corrected p"+I, and Y7+ ' using an RK2 formulation of Eqs. (5) and (6), based on their right-hand-side values at In and those predicted at I n + I. Stiff: Evaluate corrected p,,+I, and Y7+ 1 using the stiff ODE integrator DVODE [6], operating on a semi-implicit discretization of Eqs. (5) and (6). This uses an RK2 formulation for the convection and diffusion

FLAME OBSERVABLES

375

terms based on their values at In and their predicted values at 1.+" added to an implicit formulation for the chemical source terms. Thus, for the scalar vector = {p,pYi } , the following ODE system is integrated in each computational cell from In to I n +"
(7)

Downloaded by [Purdue University] at 15:57 12 February 2014

where C, D, R, are the convection, diffusion, and reaction right-handside terms of the scalar conservation equations. Note that, except for R(), every term on the right-hand-side of Eq. (7) is a known constant for the purpose of ODE integration. Evaluate corrected rr:' from the state equation . Evaluate corrected v"+ I using a variable-density projection scheme [44].

COMPUTED FLOW

We study the interaction of a premixed methane-air flame, 20% N2"diluted, with a counter-rotating vortex pair in 2D, under atmospheric pressure conditions. The premixed reactants are at ambient temperature 298 K. Four flow-flame cases are considered, as follows: C I-1.0-F Using a C) mechanism [64], at stoichiometric conditions (equivalence ratio <P = 1.0), with a "Fast" vortex-pair (strong circulation). C)-1.0-S Using a C) mechanism [64], at <P= 1.0, with a "Slow" vortexpair with 1/2 the circulation in C)-1.0-F. C 1CZ-1.0 Using a C)C Z mechanism [19], at <P = 1.0, with the fast vortex pair of C)-1.0-F. C jC2"1.2 Using a C1CZ mechanism [19], at <P = 1.2, with the fast vortexpair of C)-1.0-F. The C) mechanism used involves 16 species and 46 reaction steps [64]. The dominant path of carbon in this mechanism under the present conditions, and the associated active species, are given by:

The C)C Z mechanism used is GRlmechl.2 [19] (32 species and 177 reactions), with the carbon flow diagram illustrated in Figure I. One-dimensional premixed flame comparisons between the two mechanisms may

376

H. N. NAJM et al.

Downloaded by [Purdue University] at 15:57 12 February 2014

FIGURE I A reaction pathway diagram for a freely propagating premixed methane-air


flame, based on the integrated reaction rates from computed ID flame profiles with a stoi-

chiometric 20% N,-diluted reactants mixture at room temperature, using Chemkin [26,27) and the GRImechl.2 C,C, [19] mechanism. Where convenient, several reaction paths have been lumped with relevant participating srcies listed. Again for convenience, reaction paths with
integrated rates below 10- 6 moles/em -s have been excluded from the schematic. Some reaction paths that fall below this threshold are included for completeness, and are delineated with

dashed lines. Excited species CH', OH', and CO; are not included in the mechanism, but are listed here based on their commonly accepted formation/consumption paths.

be found in [43]. Selected species peak mole fractions and profile widths for an atmospheric 10 freely propagating stoichiometric premixed methane-air flame computed using Chemkin [26,27] and GRImechJ.2 [19], with no N 2 dilution and under adiabatic conditions, are listed in Table I. This data is useful for assessing expected signal levels and spatial resolution requirements in experimental measurements of specific flame observables. The 20 computational domain is a 0.4 x 2.0cm 2 region for the slow C, case, and 0.4 x 1.6cm 2 for the other cases. The spatial discretization is based on a uniform cell-size of 15.6 urn in each direction. The flow evolution for case C,C 2-1.2 is illustrated in Figure 2. We impose symmetry boundary conditions in the horizontal x-direction, and outflow boundary conditions in the y-direction. The initial vorticity and temperature field contours are shown in the I = 0 frame in the figure. The vertical right edge of the domain is the centerline of the vortex pair under consideration, which is one member of an infinite periodic row of vortex pairs along the horizontal x-direction. This periodic construction is computationally convenient. It is of course expected to affect the vorticity dynamics of the flow. Allowing for these effects, the present work investigates the resulting flame response under the ensuing local strain-rate and curvature conditions.

FLAME OBSERVABLES

377

TABLE I Computed peak mole fractions and profile thicknesses of various species in a stoichiometric ID freely propagating premixed adiabatic methane-air flame, using Chernkin [26,27] and GRlmech1.2 [19] at atmospheric pressure, with no N, dilution. Profile thickness is Full Width at Half Maximum (FWHM) except where indicated
Species Peak mole fraction

Profile thickness (mm)

OH

0, H,
H,O HO, H,O, CH. CH 3 CH,O CH,OH CH 30 CH 30H HCO CO CO,

Downloaded by [Purdue University] at 15:57 12 February 2014

CHi CH, CH C,H, C,H. C,H3


C,H, C,H 6

C,H

7.7 X 10- 3 3.1 X 10- 3 6.5 X 10- 3 0.19 18 x 10- 3 0.18 0.15 x 10- 3 18 X 10- 6 0.095 2.1 x 10- 3 1.3 X 10- 3 5.5 X 10- 6 19 X 10- 6 0.14 X 10- 3 62 X 10- 6 0.048 0.085 5.4 x 10- 6 58 X 10- 6 7.1 X 10- 6 1.2 X 10- 3 47 X 10- 6 0.69 X 10- 3 16 X 10- 6 0.14 X 10- 3 0.081 X 10- 6

0.21' 0.18' 0.22' 0.37' 0.50' 0.39' 0.30 0.27 0.3" 0.15 0.24 0.13 0.13 0.23 0.15 0.26' 0.52' 0.13 0.13 0.11 0.18 0.13 0.19 0.11 0.17 0.11

Profile thickness based on maximum gradient, 6= (XrnaK - Xmin)/ldXjdn!mu'

The initial condition for each flow-flame case is a superposition of the velocity (u, v) field due to the x-row of vortex pairs, and the (T, p, Y j ) fields in the y-direction from a ID premixed flame solution based on the selected kinetic mechanism using Chemkin [26,27]. The Chernkin solution is relaxed on a ID uniform grid prior to its use in the 20 model. The vorticity field corresponding to each initial vortex is a second-order Gaussian. The initial flame thickness (defined, along with burning speed, in the Appendix) is in the range 6; = [0.067 - 0.084Jem depending on the particular flow case. The initial flame burning speed is in the range SL = [13.2 - 17.7Jcm/s. The maximum rotational velocity in the domain is around IOm/s for the slow case and 20 m/s for the fast cases, giving a maximum Mach number in the range 0.03 - 0.06, a small value as required by the present formulation. The Darnkohler number (Da), defined as the ratio of the vortex-pair time scale to the flame time scale, is in the range 0.02-0.05. With Oa I, the flow is faster than the flame, and significant flame contortion is expected.

Downloaded by [Purdue University] at 15:57 12 February 2014

FLAME OBSERVABLES

379

Downloaded by [Purdue University] at 15:57 12 February 2014

Using data from ID normal flame cuts at various time instances during each set of results, we study peak-to-peak correlations between relevant fields. Point-to-point correlations are also used where appropriate. It should be noted that this latter approach must be used with caution, as a simple shift of any field normal to the flame would result in a very poor point-to-point self-correlation, even when the peak values at each flame cut are perfectly correlated. When the object is comparison of spatially distinct components of flame structure between different spatial locations or over time intervals, the peak-to-peak approach is more relevant. On the other hand, when two fields are not spatially shifted within the flame structure, then point-to-point correlations provide a more powerful and general statement. Peak-to-peak correlation data is generated by examining normal cuts through the flame along the whole flame length for the four flow-flame cases considered. Data is included at 0.2 ms time intervals from the fast flow cases (I ms total flow time), and at 0.4 ms from the slow case (4 ms flow time). In order to generate a correlation plot between two specific flame fields, e.g., OH mole fraction and heat release rate, the peak values of both quantities in the flame are extracted from each normal cut, and used to determine one point on the plot. The two peak values are not necessarily at the same physical location, but are within the flame structure. The number of normal flame cut samples is determined by the overall length of the flame within the computational domain, i.e., the length of the above 10% CH 4 mass fraction contour line. One cut is selected in each computational cell that intersects this contour, in the direction of the local flame normal R. The cut location is selected at the contour-point closest to the cell center. These peak-topeak correlations exclude regions of high flame curvature (defined in the Appendix), where, 111;18; > I, because of the ambiguity in defining the local "flame front" and associated peaks in regions of high curvature.

RESULTS

We now proceed to examine correlations between various experimental flame observables and burning and heat release rates. Note that the peak burning rate (burning rate is defined here as the rate of consumption of methane, lVCH,) and peak heat release rate lVr are very well correlated together for all the cases considered, as seen in Figure 3. Thus, good correlation with peak heat release rate can also be inferred to indicate a similar correlation with peak CH 4 consumption rate. In the following,

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

FLAME OBSERVABLES

383

high (C,C 2-I.O) and low (C 1C2-1.2) WT regions, values of 'V . v spanning the full range of the plot are evident. Moreover, the effect of flow unsteadiness is evident by comparing the degree of vertical scatter in the 'V . v data between the fast and slow C 1 cases. The faster flow leads to a higher degree of unsteadiness at the flame, resulting in the increased scatter of the data in Figure 6 for case C,-1.0-F versus that of C I-1.0-S.
OH

The mole fraction of OH is commonly measured in flames using PUF techniques [39,46). With a judicious choice of excitation and detection wavelengths, measured PUF signals are essentially representative of mole fraction with little dependence on temperature and composition of the collisional bath [46). A reasonable model for OH UF signal, using A-X excitation and nonresonant fluorescence detection, can be developed owing to the availability of collisional quenching and vibrational relaxation cross sections for a number of collision partners over a wide range of temperatures [47,48). It is noteworthy that OH is relatively stable at high temperature, and exists in significant concentrations in the hot products. Accordingly, Nguyen and Paul [46) suggest that the use of OH as a flame marker must rely on absolute signal levels or gradients. The utility of peak OH as a flame marker was studied in [43]. Experimental evidence [46) suggests that as the vortexpair propagates into the rich flame (1) = 1.2), a four-fold increase in OH mole fraction is observed on the vortex-pair centerline, coincident in time with the disappearance of CH. No such rise is observed at stoichiometric conditions. As indicated in [42) this behaviour is not observed numerically with GRlmech1.2 [19). Rather, and as found in the present results (not shown), peak OH decays monotonically on the vortex-pair centerline for all cases considered. Moreover, no penetration of OH into the cold reactants (reported in [46)) is observed using the present flow, chemical, and transport models. Peak OH mole fraction XO H and peak OH gradient normal to the flame dXOH/dn are shown plotted against peak WT in Figures 7 and 8 respectively. The scatter in both figures is an indication of the effect of unsteady strainrate and curvature as well as flame stoichiometry on these correlations. The stoichiometric C, and C 1C2 data are not coincident, but are generally close. On the other hand, the effect of stoichiometry is clearly evident. While each of the stoichiometric and rich flame data may be used to construct a rough local correlation, no general correlation is feasible. Therefore, the use of

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

Downloaded by [Purdue University] at 15:57 12 February 2014

FLAME OBSERVABLES

397

C;

Downloaded by [Purdue University] at 15:57 12 February 2014

Chemiluminescence from C z d 3 ng(C;) has been a widely applied diagnostic for the study of premixed flames. It is generally accepted that C; is produced by reaction of CH with C atom or CH. Bleekrode and Nieuwpoort [3] studied CH' and C; production and spectral signatures in a low pressure flame. Based on measured concentrations, they concluded against production of C; via a reaction of CH with CH because the rate constant would need to be unrealistically high (since C atom is present in even lower concentration the reaction C + CH ~ C; + H would require an even higher rate coefficient). This conclusion was further supported by the subsequent studies of Bulewicz et al. [7]. Vanpee and Quang [67] restated support for 2CH ~ C; + Hz, but this relied on the presumption that their measured CH' signal reflected ground state CH concentration which cannot be substantiated given the recently reported mechanism for CH' production [14, 15]. At present there is no C; production mechanism which appears to explain the reported signal levels and spectral signatures. However if either of the above mechanisms is correct then the C; signal would be linear or quadratic in CH concentration and would be thus expected to suffer the same difficulties and even accentuate the undesirable characteristics of CH as a measure of heat release. Finally C; exhibits a maximum in signal at even richer conditions (lower heat release) than CH' signal, suggesting a complicated relation with heat release as a function of local stoichiometry.

CONCLUSIONS

We have presented a numerical study of methane-air flame observables in an unsteady 2D flow environment. We focused on a flame-flow interaction involving a freely propagating premixed methane-air flame and a counterrotating vortex pair. Four flow-flame cases were considered using two different chemical mechanisms. Flow conditions spanned a two-fold variation of vortex-pair strength, while flame conditions included stoichiometric and rich (g> = 1.2) conditions. Results were analyzed and used to examine correlations between various flame observables and heat release rate. Correlations were studied using both point-to-point plots involving all points in the computational domain independent of flame structure, and peak-to-peak plots based on the observed peak values of relevant fields observed in normal cuts of flame structure. The latter technique was limited

398

H. N. NAJM et al.

to flame regions where the radius of curvature is smaller than the initial flame thickness. We demonstrated excellent peak-to-peak correlation between CH 4 consumption rate and peak Wr. Thus, all statements regarding the quality of correlations with peak Wr may be extended to peak CH 4 consumption rate. For all the cases studied, HCO mole fraction was found to have an excellent correlation with heat release rate Wr, based both on point-to-point and peak-to-peak studies. HCO is a good flame marker and a quantitative measure of heat release and burning rates. Peak flow dilatation rate '\l. v was found to have generally poor correlation with peak Wr under unsteady flow conditions, in agreement with earlier findings [43]. Peak OH, CO, and CH were found to have some positive correlation with peak Wr for a given reactants mixture composition. Here again, large scatter does exist, however, due to flow disturbances. On the other hand, allowing for variations in mixture composition leads to essentially uncorrelated data. Thus, in a turbulent reacting flow, where local mixture stoichiometry may not be known, it is generally not possible to use peak OH, CO, or CH as local measures of flame burning or heat release rate. On the other hand, the present data suggests that one exception may be the use ofCH at low burning rate, as the flame approaches extinction, where meaningful overlap of the rich and stoichiometric flame data is evident. We recall however earlier experimental data [43,46] that suggests continued flame burning after CH has completely disappeared, both at rich and stoichiometric conditions. This is clearly not evident here. Further studies are necessary with other chemical mechanisms, broader ranges of parameters, and slower vortices, to allow more extensive comparison with the experimental data and examine this discrepancy. Peak gradient of OH mole fraction in the flame was also found to be affected significantly by mixture composition, thus rendering it a questionable measure of peak Wr in a general turbulent flow. Worse performance was evident from the peak CO mole fraction gradient, where both extensive scatter due to flow disturbances, and mixture composition lead to essentially uncorrelated data. Both CH 3 and CH 20 data exhibited strong dependence on the chemical mechanism used in the study, for the same flow-flame conditions. Significantly higher CH 3 and CH 20 mole fractions were predicted by the C\ versus the C\C 2 mechanism. We observe no useable general correlation for peak CH 3 with peak Wr, allowing for variations in mixture composition. On the other hand, composition had no significant influence on the observed correlation of CH 20 with peak heat release rate, as predicted by the present

Downloaded by [Purdue University] at 15:57 12 February 2014

FLAME OBSERVABLES

399

C, C z kinetics. A good, albeit non-linear, correlation is evident, suggesting the adequacy of peak CHzO as a measure of flame burning and heat release rates. Peak CHi mole fraction was also found to exhibit reasonable correlation with peak WT. albeit with significant scatter. This suggests that measurement of CHi ('CH z) mole fraction may be a good measure of peak burning and heat release rates in this flame. Poor correlation was found between peak CzH z mole fraction and heat release rate, primarily due to data scatter. What correlation does exist reveals an inverse relationship with peak WT, consistent with earlier ID flame studies for the present equivalence ratio range. On the other hand, both lower CzH z and WT would be expected from leaner mixtures with cJ> < I. Concentration products were also studied to examine their utility as measures of flame burning and heat release rates. The peak [OH][CH 4 ] and [OH][CO] data was found to give useful correlations with peak WT, with little systematic effect due to stoichiometry or flow disturbances, but sizeable scatter. On the other hand, despite the remarkably low scatter in the peak [OH][CHzO] data, this particular correlation reveals a systematic shift due to mixture composition. Finally, as possible respective measures of the production rates of CH', OH', and COi (which are not included in the present kinetics), we studied the concentration products [CzH][O]. [CHI[Oz], and [CO][O]. Large scatter was observed in the [CzH][O] data, while very low scatter and a compact linear correlation was observed between peak [CH][Oz] and peak WT. Useful correlation was also observed for peak [CO][O], with no systematic effect of flow disturbances or stoichiometry, although with significant scatter. We also discussed Ci chemiluminescence and the relationship between Ci and CH. Further studies on the chemiluminescence and kinetics of CH', OH', Ci, and COi are necessary to establish their relationships with flame burning and heat release rates. In closing, it is important to note that these findings are necessarily specific to the range of parameters investigated. Additional work is necessary to evaluate their extension to wider ranges of operating conditions, or to other fuels and mixture compositions.
Acknowledgments

Downloaded by [Purdue University] at 15:57 12 February 2014

This work was supported by the US Department of Energy (DOE), the DOE Office of Basic Energy Sciences, Chemical Sciences Division, and the DOE Defense Programs Accelerated Strategic Computing Initiative (ASCI).

400

H. N. NAJM et al.

References
[I] Ashurst, W. T. and McMurtry, P. A. (1989). Flame Generation of Vorticity: Vortex Dipoles from Monopoles, Combustion Science and Technology, 66, 17-37.

[2J Bauerle, B., Warnatz, J. and Behrendt, F. (1996). Time Resolved Investigation of Hot Spots in the End Gas of an S.1. Engine by Means of 2D Double-Pulse L1F of Formaldehyde, In: Twenty-Sixth Symposium (lnternationa/) on Combustion, pp. 26192626, The Combustion Institute. [3] B1eekrode, R. and Nieuwpoort, W. C. (1965). Absorption and Emission Measurements of C, and CH Electronic Bands in Low-Pressure Oxyacetylene Flames, J. Chern. Phys., 43, 3680-3687. [4J Bowman, C. T., Hanson, R. K. Gardiner, W. c., Lissianski, V., Frenklach, M., Goldberg, M. and Smith, M. P. (1997). GRI-Mech2.II-An Optimized Detailed Chemical Reaction Mechanism for Methane Combustion and NO Reburning, Topical Report GRI-97/0020, Gas Research Inst., Chicago. [5] Bowman, C. T. and Seery, D. G. (1968). Chemiluminescence in the High-Temperature Oxidation of Methane, Combustion and Flame, 12,611-614. [6] Brown. P. N., Byrne, G. D. and Hindmarsh, A. C. (1989). VODE: A Variable Coefficient ODE Solver, SIAM J. Sci. Stat. Comput., 10, 1038-1051. [7] Bulewicz, E. M., Padley, P. J. and Smith, R. E. (1970). Spectroscopic Studies ofC" CH, and OH Radicals in Low Pressure Acetylene + Oxygen Flames, Proc. Royal Soc. London, A315, 129-148. [8] Candel, S. M. and Poinsot, T. J. (1990). Flame Stretch and the Balance Equation for the Flame Area, Combustion Science and Technology, 70, 1-15. [9] Cappelli, M. A. and Paul, P. H. (1990). An Investigation of Diamond Film Deposition in a Premixed Oxyacetylene Flame, J. Appl. Phys., 67(5), 2596- 2602. [10] Chorin, A. J. (1967). A Numerical Method for Solving Incompressible Viscous Flow Problems, J. Comput, Phys., 2, 12-26. [I I] Clyne, M. A. A. and Thrush, B. A. (1962). Mechanism ofChemiluminescent Combination Reactions Involving Oxygen Atoms, Proc. Royal Soc. London A, 269, 404-418. [12] Cool, T. A., Bernstein, J. S., Song, X.-M., and Goodwin, P. M. (1998). Profiles of HCO and CH, in CH./O, and C,H./O, Flames by Resonance Ionization, In: Twenty-Second Symposium (lnternational) on Combustion, p. 1421. The Combustion Institute. [I 3] Desgroux, P., Gasnot, L., Crunelle, B. and Pauwels, J. F. (1996). CH, Detection in Flames using Photodissociation-induced Fluorescence, In: Twenty-Sixth Symposium (lnternationa/) on Combustion, pp. 967-974, The Combustion Institute. [14] Dcvriendt, K. and Peeters, J. (1997). Direct Identification of C,H + O'P-CH (A'~) + CO Reaction as the Source of CH (A'~ - x-rn Chemiluminescence in C,H,/ O/H Atomic Flames, J. Phys. Chern. A, 101, 2546-2551. [15] Devriendt, K., Van Look, H., Ceursters, B. and Peeters, J. (1996). Kinetics of Formation of Chemiluminescent CH(A'~) by the Elementary Reactions of C,H(X'E+) with O('P) and O,(X'E;): A Pulse Laser Photolysis Study, Chern. Phys. LeIlS., 261, 450-456, [I 6] DiRosa, M. D. and Farrow, R. L., Cross Sections of Photo ionization and Stark Shift for 2 + I REM PI of CO, AIAA-98-2477, June 1998 AIAA 29th Plasma Dynamics and Laser Conference, Albuquerque, NM. [17] Etzkorn, T., Fitzer, J., Muris, S. and Wolfrum, J. (1993). Determination of Absolute Methyl-Radical and Hydroxyl-Radical Concentrations in a Low-Pressure. Methane Oxygen Flame, Chern. Phys. Lett., 208, 307-310. [18] Farrow, R. L., Bui-Pham, M. N. and Sick, V. (1996). Degenerate Four-Wave Mixing Measurements of Methyl Radical Distributions in Hydrocarbon Flames: Comparison with Model Predictions, In: Twenty-Sixth Symposium (International) on Combustion, pp. 975-983, The Combustion Institute. [19] Frenklach, M.. Wang, H., Goldenberg, M., Smith, G. P., Golden, D. M., Bowman, C. T.. Hanson, R. K., Gardiner, W. C. and Lissianski, V., GRlmech -an Optimized Detailed Chemical Reaction Mechanism for Methane Combustion, Topical Report GRI-95/0058, Gas Research Institute, November 1995.

Downloaded by [Purdue University] at 15:57 12 February 2014

FLAME OBSERVABLES

401

Downloaded by [Purdue University] at 15:57 12 February 2014

[20] Harrington, J. E. and Smyth, K. C. (1993). Laser-Induced Fluorescence Measurements of Formaldehyde in a Methane Air Diffusion Flame, Chern, Phys. Lett., 202, 196-202. [21] Haworth, D. C. and Poinsot, T. J. (1992). Numerical simulations of Lewis number effects in turbulent premixed flames, J. Fluid Mechanics, 244, 405-436. [22] Herzberg, G. (1996). Electronic Spectra of Polyatomic Molecules Van Nostrand, NY. [23] Hilka, M., Veynante, D., Baum, M. and Poinsot, T. J., Simulation of flame-vortex interactions using detailed and reduced chemical kinetics, In: Tenth Symp. on Turb. Shear Flows, 2, 19/19-24, University Park, PA, August 1995, Penn. State Univ. [241 Jarosinski, J., Lee, J. H. and Knystautas, R. (1988). Interaction of a Vortex Ring and a Laminar Flame, In: Twenty-Second Symposium (International) on Combustion, pp. 505514, The Combustion Institute. [25] Jefferies, J. 8., Crosley, D. R., Wysong, I. J. and Smith, G. P. (1990). Laser-Induced Fluorescence Detection of HCO in a Low Pressure Flame, In: Twenty-Third Symposium (International) on Combustion, pp. 1847-1854, The Combustion Institute. [26] Kee, R. J., Grear, J. F., Smooke, M. D. and Miller, J. A., A Fortran Program for Modeling Steady Laminar One-Dimensional Premixed Flames, Sandia Report SAND8S8240, Sandia National Labs., Livermore, CA., September 1993. [27] Kee, R. J., Rupley, F. M. and Miller, J. A., Chemkin-Il: A Fortran Chemical Kinetics Package for the Analysis of Gas Phase Chemical Kinetics, Sandia Report SAND898009B, Sandia National Labs., Livermore, CA., August 1993. [28] Kim, J. and Moin, P. (1985). Application ofa Fractional-Step Method to Incompressible Navier-Stokes Equations, J. Comput. Phys., 59, 308-323. [29] Lee, T.-W., Lee, J. G., Nye, D. A. and Santavicca, D. A. (1993). Local Response and
Surface Properties of Premixed Flames During Interactions with Karman Vortex Streets, Combustion and Flame, 94, 146-160.

[30] Luque, J., Smith, G. P. and Crosley, D. R. (1996). Quantitative CH Determinations in Low-pressure Flames, In: Twenty-Sixth Symposium (International) on Combustion, pp. 959-966, The Combustion Institute. [31] Mahalingam, S., Cantwell, B. J. and Ferziger, J. H. (1990). Full Numerical Simulations of Coflowing, Axisymmetric Jet Diffusion Flames, Physics of Fluids A, 2, 720-728. [32] Majda, A. and Sethian, J. (1985). The Derivation and Numerical Solution of the Equations for Zero Mach Number Combustion, Comb. Sci. and Technology, 42, 185-205. [33] Mantel, T. (1994). Fundamental mechanisms in premixed flame propagation via vortexflame interactions - numerical simulations, Annual research briefs, Center for Turbulence

Research, Stanford University/NASA Ames Research Center. [34] Markus, M. W., Roth, P. and Just, T. (1996). A Shock-Tube Study of the Reactions ofCH with CO, and 0" Int. J. Chem. Kinet., 28, 171. [35] Masri, A. R., Dibble, R. W. and Barlow, R. S. (1996). The Structure of Turbulent Nonpremixed Flames Revealed by Raman-Rayleigh-LlF Measurements, Prog. Energy Combust. Sci., 22, 307- 362. [36] McMurtry, P. A., Jou, W.-H., Riley, J. J. and Metcalfe, R. W. (1986). Direct Numerical Simulations of a Reacting Mixing Layer with Chemical Heat Release, AIAA J., 24(6), 962-970. [37] Mueller, C. J., Driscoll, J. F., Reuss, D. L. and Drake, M. C. (1996). Effects of Unsteady Stretch on the Strength of a Freely-Propagating Flame Wrinkled by a Vortex, In: Twenty-Sixth Symposium (International) on Combustion, pp. 347-355, The Combustion Inst. [38) Mueller, C. J., Driscoll, J. F., Reuss, D. L., Drake, M. C. and Rosalik, M. E. (1998).
Vorticity Generation and Attenuation as Vortices Convect through a Premixed Flame, Combustion and Flame, 112(3), 342-358. [39] Mueller, C. J., Driscoll, J. F., Sutkus, D. J., Roberts, W. L., Drake, M. C. and Smooke,

M. D. (1995). Effect of Unsteady Stretch Rate on OH Chemistry during a Flame-Vortex Interaction: To Assess Flamelet Models, Combustion and Flame, 100, 323- 331. [40) Mungal, M. G., Lourenco, L. M., and Krothapalli, A. (1995). Instantaneous Velocity Measurements in Laminar and Turbulent Premixed Flames using On- Line PIV, Combustion Science and Technology, 106, 239-265.

402

H. N. NAJM et 01.

[41J Najm, H. N. (1996). A Conservative Low Mach Number Projection Method for Reacting Flow Modeling, In: Chan, S. H. (Ed.), Transport Phenomena in Combustion, 2, 921-932, Taylor and Francis, Wash. DC. [42] Najrn, H. N., Knio, O. M. and Wyckoff, P. S. (1988). Response of Stoichiometric and Rich Methane-Air Flames to Unsteady Strain-Rate and Curvature, Combustion Theory and Modelling, 1998 submitted. [43] Najm, H. N., Paul, P. H., Mueller, C. J. and Wyckoff, P. S. (1998). On the Adequacy of
Certain Experimental Observables as Measurements of Flame Burning Rate, Combustion

Downloaded by [Purdue University] at 15:57 12 February 2014

and Flame, 113(3),312-332. [44] Najm, H. N. and Wyckoff, P. S. (1997). Premixed Flame Response to Unsteady StrainRate and Curvature, Combustion and Flame, 110(1-2),92-112. 145J Najm, H. N., Wyckoff, P. S. and Knio, O. M. (1998). A Semi-Implicit Numerical Scheme for Reacting Flow. I. Stiff Chemistry, Journal of Computational Physics, 143(2), 381-402. [46J Nguyen, Q.-V. and Paul, P. H. (1996). The Time Evolution of a Vortex-Flame Interaction Observed via Planar Imaging of CH and OH, In: Twenty-Sixth Symposium (International) on Combustion, pp. 357 - 364, The Combustion Institute. [47] Paul, P. H. (1994). A Model for Temperature-Dependent Collisional Quenching of OH A'E+. J. Quan/. Spectrosc. Radiat. Transfer, 51, 511-524. [48] Paul, P. H. (1995). Vibrational-Energy Transfer and Quenching of OH A'E+(v = I) Measured at High Temperatures in a Shock Tube, J. Phys. Chern., 99, 8472-8476. [49J Paul, P. H. and Najm, H. N. (1998). Planar Laser-Induced Fluorescence Imaging of Flame
Heat Release Rate, In: Twenty-Seventh Symposium (International) on Combustion, The Combustion Institute, in press.

[50J Peeters, J. and Mahnen, G. (1973). Reaction Mechanism and Rate Constants of
Elementary Steps in Methane-Oxygen Flames, In: Fourteenth Symposium (International)

on Combustion, pp. 133-146, The Combustion Institute. 151] Poinsot, T., Echekki, T. and Mungal, M. G. (1991). A Study of the Laminar Flame Tip and Implications for Premixed Turbulent Combustion. Combustion Science and Technology, 81, 45 -73. [52J Poinsot, T., Veynante, D. and Candel, S. (1991). Quenching Processes and Premixed Turbulent Combustion Diagrams, J. Fluid Mechanics, 228, 561-606. [53] Roberts, W. L., Driscoll, J. F., Drake, M. C. and Goss, L. P. (1993). Images of the Quenching of a Flame by a Vortex-To quantify Regimes of Turbulent Combustion, Combustion and Flame, 94, 58-69. [54] Rohrig, M., Petersen, E. L., Davidson, D. F., Hanson, R. K. and Bowman, C. T. (1997). Measurement of the Rate Coefficient of the Reaction CH + 0, =} Products in the Temperature Range 2200 to 2600 K, Int. J. Chem. Kinet., 29, 781. [55J Rolon, J. C., Aguerre, F. and Candel, S. (1995). Experiments on the Interaction between a Vortex and a Strained Diffusion Flame, Combustion and Flame, 100(3),422-429. [56J Rutland, c., Ferziger, J. H. and Cantwell, B. J., Effects of Strain, Vorticity, and Turbulence on Premixed Flames, Report TF-44, Thermosciences Div., Mech. Eng., Stanford Univ., Stanford, CA, October 1989. 157] Rutland, C. J. and Ferziger, J. H. (1991). Simulations of Flame-Vortex Interactions, Combustion and Flame, 84; 343-360. [58J Samaniego, J.-M. (1993). Stretch-induced quenching in flame-vortex interactions, Annual research briefs, Center for Turbulence Research, Stanford University/NASA Ames
Research Center.

[59] Samaniego, J.-M., Egolfopoulos, F. N. and Bowman, C. T., COi Chemiluminescence in Flames. Combustion Science and Technology, 109, 183- 203. [60] Sappey, A. D., Crosley, D. R. and Copeland, R. A. (1990). Laser-Induced Fluorescence Detection of Singlet CH 2 in Low-Pressure Methane/Oxygen Flames, Appl. Phys. B, 50, 463-472. [61] Scherer, J. J. and Rakestraw, D. (1997). Cavity Ringdown Laser Absorption Spectroscopy Detection of Formyl (HCO) in a Low Pressure Flame, Chern. Phys. Letts., in press. [62] Schlichting, H. (1979). Boundary-Layer Theory McGraw-Hili, New York, 7th edition, 1979.

FLAME OBSERVABLES

403

[63] Seitzman, J. M., Haumann, J. and Hanson, R. K. (1987). Quantitative 2-Photon L1F Imaging of Carbon-Monoxide in Combustion Gases, Applied Optics, 26, 2892-2899. [64) Smooke, M. D., Puri, I. K. and Seshadri, K. (1986). A Comparison Between Numerical Calculations and Experimental Measurements of the Structure ofa Counterflow Diffusion
Flame Burning Diluted Methane in Diluted Air, In: Twenty-First Symposium (Interna-

tional) on Combustion, pp. 1783-1792, Pittsburgh, PA, The Combustion Institute. [65] Smyth, K. C. and Taylor, P. H. (1985). Detection of the Methyl Radical in a Methane Air Diffusion Flame by Multiphoton Ionization Spectroscopy, Chem. Phys. Leu., 122, 518-522. [66] Trouve, A. and Poinsot, T. (1994). The Evolution Equation for Flame Surface Density, J. Fluid Mechanics, 278, 1- 3 I. [671 Vanpee, M. and Quang. I. N. (1979). A Study of Premixed Hydrocarbon-Fluorine Flames, In: Seventeenth Symposium (International) on Combustion, pp. 881-890, The Combustion
Institute.

Downloaded by [Purdue University] at 15:57 12 February 2014

[68] Williams, F. A. (1985). Combustion Theory Addison-Wesley, New York, 2nd edition. [69] Woiki, D., Votsmeier, M., Davidson, D. F., Hanson, R. K. and Bowman, C. T. (1998). CH-Radical Concentration Measurements in Fuel-Rich CH./O,/Ar and CH./O,jNO/Ar Mixtures Behind Shock Waves, Combustion and Flame, 113,624-626. [70] Zalicki, P., Ma, Y., Zare, R. N., Wahl, E. H., Owano, T. G. and Kruger, C. H. (1995). Measurement of the Methyl Radical Concentration Profile in a Hot-Filament Reactor, Appl. Phys. Leu., 67, 144-146.

APPENDIX

Following the notation in the Flame Diagnostics section, the tangential strain rate at the flame, 7" is evaluated from the strain-rate tensor ~ [68]: 7{ = t-e-t. Flame curvature [8], K., is defined positive when the flame is convex to the reactants, in 20: K. = I/R = V' . n, where R is the radius of curvature. The flame displacement speed [8,51,66] is defined as the speed of displacement of a given contour-level of a scalar field variable relative to the local flow velocity. Thus, at the 10% methane contour level, the displacement speed is defined, positive for propagation into the reactants, as:

The laminar burning speed relative to the unburnt fluid is a density-weighted displacement speed defined as SL = PSd/Pu, where P is the gas density at the same location as Sd, and Pu is the unburnt reactants density. We also define the flame thermal thickness as:

aT/an is the

where Ti; Tu are the burnt and unburnt gas temperatures respectively, and temperature gradient along the flame normal direction.

S-ar putea să vă placă și