Investigation of stoichiometric methane/air/benzene (1.5%) and methane/air low pressure flames

  • Published on

  • View

  • Download


  • Investigation of stoichiometric methane/air/benzene(1.5%) and methane/air low pressure flames

    Laurent Duponta, Abderrahman El Bakalia, Jean-Francois Pauwelsa, IsabelleDa Costab, Philippe Meunierb, Henning Richterc,*

    aUniversite des Sciences et Technologies de Lille, UMR 8522 Physicochimie des Processus de Combustion, CentredEtudes et de Recherches Lasers et Applications, F-59655 Villeneuve dAscq, France

    bGaz de France, Direction de la Recherche, B.P. 33, F-93211 Saint Denis la Plaine Cedex, FrancecDepartment of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139-4307, USA

    Received 28 October 2002


    Benzene depletion in a laminar premixed flat stoichiometric low-pressure methane/air/benzene (1.5%) flame wasinvestigated using a recently developed kinetic model that has been tested for low-pressure combustion of acetylene,ethylene, and benzene. Experimental flame structures of two stoichiometric methane/air flames (v34.2 cm s1, 5.33kPa) with and without the addition of 1.5% of benzene were measured previously by means of molecular beamsampling using mass spectrometry as well as gas chromatography coupled to mass spectrometry as analytic tools.Model computations were conducted using the Premix code within the Chemkin software package. Experimentaltemperature profiles were used as input. The analysis of rates of production of selected species allowed for theidentification of major formation and depletion pathways. The predictive capacility of the model was assessed in bothflames. Good to excellent agreements between predictions and measured mole fraction profiles were obtained forreactants, intermediates, and products such as methane, O2, methyl, H, OH, CO, CO2. and H2O. Benzene depletion andthe formation and consumption of intermediates such as cyclopentadiene were predicted correctly. According to themodel, methyl is exclusively formed by hydrogen abstraction from methane and subsequently oxidized by reactionwith O to formyl, (1) CH3 Ou HCO H2, and formaldehyde, (2) CH3 Ou CH2O H, the latter channelbeing the dominant one. Benzene consumption occurred mainly by hydrogen abstraction with OH and H as reactantbut also the contribution of its oxidation by O to phenoxy was significant. Phenol and phenoxy chemistries are stronglycoupled, unimolecular decay of phenoxy to cyclopentadienyl radicals and CO is the dominant consumption route.Small PAH are predicted to be formed in the reaction zone, followed by complete depletion in the postflame region.The pressure dependence of the dominant dimethylether formation (32) CH3 CH3Ou CH3OCH3 was found to besignificant and assessed by means of Quantum Rice-Ramsperger-Kassel (QRRK) analysis. 2003 The CombustionInstitute. All rights reserved.

    Keywords: Methane; Benzene; PAH

    1. Introduction

    The presence of fine particles generated by in-complete combustion processes in atmospheric aero-sols is of significant health concern due to their as-sociation with lung cancer and cardiopulmonarydisease shown in epidemiological studies [1,2]. Poly-

    * Corresponding author. Tel.: 1-617-253-6536; fax:1-617-258-5042.

    E-mail address: (H. Richter).

    Combustion and Flame 135 (2003) 171183

    0010-2180/03/$ see front matter 2003 The Combustion Institute. All rights reserved.doi: 1016/S0010-2180(03)00158-5

  • cyclic aromatic hydrocarbons (PAH) are thought tobe key precursors of soot [3]. PAH have been quan-tified in atmospheric aerosol samples [4] and many ofthem have been found to be mutagenic [5]. Theminimization of pollutants such as soot and PAH inthe exhaust of combustion devices requires the con-trol and therefore a detailed understanding of chem-ical reaction pathways. The presence of PAH andsoot in the exhaust of hydrocarbon combustion is theoverall result of competing processes leading to thegrowth of PAH of increasing size and ultimately tosoot as well as their depletion. Depending on thelocal conditions, PAH, soot and their precursors canbe oxidized within the flame before release. Kineticmodeling by means of complex reaction networksallows for the assessment of chemical processes in-herent to hydrocarbon oxidation and pyrolysis. De-tailed reaction mechanisms have been developed andtested for premixed combustion of different fuelssuch as methane [6], acetylene [7,8], ethylene[7,9,10], ethane [6], propane [11], 1,3-butadiene[12,13] and benzene [1418] by comparison to ex-perimental flame structure data. The quality of akinetic model can be assessed by its capability topredict correctly experimental data, particularly theevolution of concentrations of reactants, products andintermediates, over a large range of experimentalconditions such as fuel type and equivalence ratio.Thus, Lindstedt and Skevis [8] tested their kineticmechanism for six lean ( 0.12) to sooting (2.50) laminar, low-pressure acetylene flames.

    The reaction network used in the present work hasbeen developed and tested by comparison of modelpredictions with experimental mole fraction profilesof stable and radical species measured in four acet-ylene ( 2.40), ethylene ( 0.75, 1.90) andbenzene (1.80) premixed flat low-pressure flamesusing molecular beam sampling coupled to massspectrometry (MBMS) [19].

    In the present work, the depletion of benzene in astoichiometric methane/air/benzene (1.5%) flamewas investigated. Combustion of methane, a majorcompound of natural gas, is of significant practicalimportance. Due to enhanced radiative heat release,the presence of soot particles can be desirable inpractical devices before their release from the flame

    while soot and PAH remaining in the exhaust gasrepresent a major environmental concern. Complexmixing patterns in practical systems can lead to alarge range of different local conditions, for instanceequivalence ratios and temperatures. Before the at-tempt of describing such complex systems, a detailedunderstanding of the chemical processes is required.For this purpose, the combustion of model com-pounds, representative for practical fuels, under well-defined conditions and setups has to be investigated.For instance, a quantitative understanding of the ox-idation of small aromatic species such as benzene isneeded in order to assess the competition between thedepletion of such compounds and their growth toPAH of increasing size and ultimately to soot.

    In the present work, a recently developed kineticmodel has been used for the analysis of benzenedepletion in a laminar premixed stoichiometric meth-ane/air low-pressure flame. The model has beentested for premixed low- pressure combustion ofacetylene, ethylene and benzene [19]. A good predic-tive capability was observed for reactants, productsand intermediates, including radical species [19]. Theoxidative depletion of all three fuels under fuel-richconditions was investigated and, in addition, a leanethylene flame was analyzed. Also, the formation ofsingle-ring aromatics was assessed quantitatively inthe case of aliphatic fuels, ie, acetylene and ethylene[19].

    2. Approach

    Flame structures of two laminar premixed flatstoichiometric methane/air (Flame I) and methane/air/benzene (Flame II) low-pressure flames (v 34.2cm s1, 5.33 kPa) were measured recently [20]. Inthe latter case, 1.5% of benzene was added to the coldgas mixture taking into account the contribution ofbenzene to the equivalence ratio, ie, the overall waskept at unity. Therefore, the mole fraction of methanein the initial mixture had to be reduced from 11.1% inthe reference flame to 6% in the seeded flame. Theparameters of both flames are summarized in Table 1.Sampling was conducted by molecular beam with

    Table 1Flame parameters. a) Flame I: laminar premixed stoichiometric methane/air flame b) Flame II: laminar premixedstoichiometric methane/air/benzene (1.5%) flameFlame Cold gas velocity v,

    25 C (cm s1)Pressure(kPa)

    XCH4 Xbenzene XO2 XN2

    Flame I 34.2 5.33 0.111 0.00 0.222 0.667Flame II 34.2 5.33 0.060 0.015 0.230 0.695

    172 L. Dupont et al. / Combustion and Flame 135 (2003) 171183

  • subsequent online analysis by mass spectrometry(MBMS) and, alternatively, by gas chromatographyusing thermal conductivity and flame ionization de-tectors (GC-TCD-FID) and a mass spectrometric de-tector (GC-MS). The use of two complementary an-alytical techniques allowed for the identification ofradical intermediates (by MBMS), essential for adetailed understanding of the combustion chemistry,and of species of identical mass such as cyclic andlinear C5H6 or C6H6 species (by GC-MS). Detectionlimits of both techniques were at mole fractions of5 107 for stable species. Due to the necessity tooperate the MBMS system at reduced ionization po-tentials to avoid fragmentation of parent molecules,detection limits of radical species are higher but de-pend on the specific conditions. Radical mole frac-tions of close to 1 105 could be detected in thepresent work. Temperature profiles were measuredby means of a coated Pt/Rh6%-Pt/Rh30% thermo-couple located 0.2 mm upstream the tip of the sam-pling probe. Radiative heat losses were taken intoaccount by electrical compensation. Details of theexperimental procedure were described previously[21,22].

    A recently developed kinetic model [19] was usedfor the investigation of reaction pathways in methanecombustion and benzene depletion under such con-ditions. Thermodynamic and kinetic property datawere updated using most recent literature, densityfunctional theory (DFT) and ab initio computationson a CBS-Q and CBS-RAD level. Quantum Rice-Ramsperger-Kassel (QRRK) analysis was conductedto determine pressure-dependent rate constants of

    chemically activated reactions [19,23]. The reactionmechanism and the corresponding thermodynamicproperty data, both carefully documented, are pro-vided in the supplementary information of [19] andare available electronically [19,24]. All model com-putations were conducted with the Premix codewithin the Chemkin software package [25] using ex-perimental temperature profiles (Fig. 1). The associ-ated postprocessor was used for the determination ofrates of production of selected species, ie, the contri-bution of specific reactions to their formation andconsumption. The model attempts the description ofthe studied combustion systems using the best avail-able thermodynamic and kinetic property data foreach species and individual reaction step; thereforeno adjustments were made in the present work. How-ever, pressure dependence of dimethylether forma-tion, not investigated previously, has been taken intoaccount in the present work.

    Quantitative measurements, particularly of spe-cies with concentrations close to the detection limitand of radicals, are challenging. Despite the excellentreproducibility found in the present work, uncertain-ties of up to 30% for some species, such as hydrogenradicals, cannot be excluded. Experimental tempera-ture profiles were used for all model calculations,therefore, the impact of possible uncertainties wasassessed. For instance, even an increase of 100 K ofall temperatures led only to an insignificant increaseof predicted maximum mole fractions while mostprofiles were shifted by 1 to 1.5 mm toward theburner.

    Fig. 1. Experimental temperature profiles [20]. stoichiometric methane/air flame (Flame I). ... stoichioimetric methane/air/benzene (1.5%) flame (Flame II).

    173L. Dupont et al. / Combustion and Flame 135 (2003) 171183

  • 3. Results

    3.1. Methane combustion

    Meaningful investigation of benzene depletion ina seeded premixed methane/air flame necessitates theaccurate description of methane oxidation chemistry.A high level of predictive capability for methanecombustion is required from kinetic models due tothe practical importance of methane and its formationas intermediate in the case of other fuels. Modelpredictions for reactants, products, and intermediateswere compared to mole fraction profiles measured inthe unseeded reference flame. Measured and com-puted mole fractions profiles of methane, O2, methyl,H, H2, OH, CO, CO2, and propane (C3H8) are shownin Fig. 2 to 4. Original experimental data pointssubsequent to calibration are given. No fitting wasperformed in order to allow the assessment of thereproducibility of the experimental procedure. Nearlycomplete methane consumption beyond approxima-tively 0.9 cm from the burner was correctly predictedby the model and a computed oxygen mole fractionof about 0.02 remaining in the postflame zone is inexcellent agreement with the experimental findings.The determination of rates of production showedhydrogen abstraction to methyl to be the only signif-icant methane consumption route. Relative contribu-tions decrease from reactions with OH and H (both40% of the maximum consumption rate) to thatwith O. Shape, peak location and value of the methylmole fraction profile was well predicted (Fig. 2). Itsrates of production showed its exclusive formationfrom methane while (1) CH3 Ou HCO H2, and(2) CH3 O u CH2O H were found to be theonly significant consumption pathways. Using ki-netic data based on the total rate constant measuredby Lim and Michael [26] and the branching ratiodetermined by Marcy et al. [27], formation of form-aldehyde (CH2O) was identified as dominant productchannel. Subsequently, formaldehyde is depleted byhydrogen abstraction to HCO via reaction with H(75%) and OH (20%). Unimolecular hydrogenloss and hydrogen abstraction by H, O2, OH, andCH3 to CO are the only significant HCO consump-tion and CO formation routes. CO consumption andCO2 formation were found to occur exclusively via(3) CO OHu CO2 H. The equilibrium constantof reaction (3) depends strongly on the temperature.Consistent with the encouraging quality of the com-puted H and OH profiles (Fig. 3), the good agree-ments for both CO and CO2 model predictions withexperimental data (Fig. 4), indicate the accuracy ofthe temperature profile used as input for the modelcomputations. Hydrogen abstraction reactions with Hradicals from methane, CH2O and HCO were found

    to play an important role. The good agreement be-tween prediction and experiment for molecular hy-drogen (Fig. 3) provides evidence of the correct ther-modynamic description of these hydrogen abstractionreactions because of the non-negligible contributionof the reverse reaction under certain conditions.

    Measured and computed mole fraction profiles ofpropane (C3H8) are given in Fig. 4 in order to illus-trate the formation and consumption of larger hydro-carbons in the case of stoichiometric methane com-bustion. (4) C2H5 CH3 u C3H8 was found to bethe only major propane formation pathway. Ethyl(C2H5) is formed by hydrogen abstraction fromethane (C2H6) by H, OH and O radicals. Self-com-bination of methyl, ie, (5) CH3 CH3u C2H6, wasidentified as dominant ethane formation route. Thedepletion of propane occurs by hydrogen abstractionwith H and OH to n- and i-propyl. Unimoleculardecay of both propyl isomers to methyl and ethylene(C2H4) and of i-propyl to propene (C3H6) are domi-nant subsequent reactions.

    3.2. Benzene de...


View more >