Experimental and Modeling Study of the Oxidation Kinetics of n -Undecane and n -Dodecane in a Jet-Stirred Reactor

Experimental and Modeling Study of the Oxidation Kinetics ofn-Undecane and n-Dodecane in a Jet-Stirred ReactorAmir Mze-Ahmed,†,‡ Kamal Hadj-Ali,† Philippe Dagaut,*,† and Guillaume Dayma†,‡

†Institut des Sciences de l’Ingenierie et des Systemes (INSIS), Centre National de la Recherche Scientifique (CNRS), 1C,Avenue de la Recherche Scientifique, 45071 Orleans cedex 2, France‡Faculte des Sciences, 1 Rue de Chartres, Universite d’Orleans, BP 6759, 45067 Orleans cedex 2, France

*S Supporting Information

ABSTRACT: The kinetics of oxidation of two large n-alkanes (n-undecane and n-dodecane) was studied experimentally in a jet-stirred reactor (JSR) at high pressure (P = 10 bar), at temperatures ranging from 550 to 1150 K, at a constant residence time (τ)of 1 s, and for three equivalence ratios (ϕ = 0.5, 1.0, and 2.0). Chemical analyses by Fourier transform infrared (FTIR)spectrometry and gas chromatography allowed for the measurement of the mole fraction of reactants, stable intermediates(including substituted tetrahydrofurans), and final products as a function of the temperature. A similar behavior was observed forthe oxidation of n-undecane, n-dodecane, and Jet A-1 in a JSR. However, it was shown that the pure n-alkanes oxidized faster thanJet A-1 under cool-flame conditions and intermediately yielded more ethylene. A kinetic reaction mechanism based on previousstudies1,2 was developed and validated by a comparison to the present experimental results. The proposed reaction mechanismconsisted of 5864 reversible reactions involving 1377 species. Experimental data and simulation results obtained in the currentwork were compared to simulations performed with a literature model.3 Our model was also applied successfully to the modelingof the oxidation of n-dodecane under shock-tube conditions.4,5 Species time histories and ignition delay times representingvaluable complementary tests were simulated.

1. INTRODUCTIONAviation fuels are very complex mixtures of hydrocarbons.6,7

To simulate the oxidation of these fuels, we have to use simplemodel fuels (surrogate) consisting of a few representativehydrocarbons. The compounds identified in jet fuels (Jet A-1, JetA, and JP-8) at the highest levels of concentration are n-alkanes.7

Previous studies showed a strong similarity between theoxidation of kerosene and n-decane under the same initial experi-mental conditions (P = 10 bar; T = 550−1150 K; τ = 1 s; and ϕ =0.5, 1.0, and 2.0) in a jet-stirred reactor (JSR).6−10 However,whereas the chemical formula of Jet A-1 is close to C11H22,

7 then-decane formula is C10H22. Therefore, n-alkanes larger than C10

could be preferred to represent Jet A-1. Furthermore, syntheticparaffinic kerosene (SPK) has high concentrations of n-alkanes(∼20 vol %),1 which further increases the interest for studyingthe kinetics of oxidation of long-chain n-alkanes. The purpose ofthis study is to expand the work performed previously on thekinetics of oxidation of n-alkanes using a pressurized JSR. Todate, no data are available for the kinetics of oxidation of C11, C12,and C13−C15 n-alkanes under JSR conditions, whereas data areavailable for the oxidation of C10 and C16 n-alkanes and for thepyrolysis of n-dodecane.11 This work intends to provide new datato fill this gap and also to propose a validated kinetic scheme.Therefore, new experiments were performed for the oxidationof two large n-alkanes (n-undecane and n-dodecane). A detailedkinetic reaction mechanism based on the work by Dievart2 wasdeveloped for modeling the oxidation of these compounds. It wasvalidated by comparison to the present experimental results andcomplementary data taken from the literature.4,5

2. EXPERIMENTAL SECTIONThe experiments were performed using a JSR presented and usedearlier.1,12 The reactor is a small sphere of 33 cm3 in volume made offused silica to minimize wall catalytic reactions. The gas mixture isintroduced into the reactor through four nozzles (1mm inner diameter).The nozzles are opposite in pairs to make the gas mixture morehomogeneous. Two insulated heating elements surrounding the reactorallow for heating of the reaction zone to the desired temperature.A nitrogen flow of 100 L/h was used to dilute the fuel before admissionin the reactor. All gases were preheated before injection to minimizetemperature gradients inside the JSR. A high-performance liquidchromatography (HPLC) pump (Shimadzu LC10 ADVP) was usedto deliver the liquid fuel to an atomizer−vaporizer assembly maintainedat ca. 550 K. Therefore, the fuel was atomized and vaporized beforeinjection into the reactor. The reactants were diluted by a flow ofnitrogen (<50 ppm of O2, <1000 ppm of Ar, and <5 ppm of H2) andmixed at the entrance of the injectors. In these experiments, we usedhigh-purity oxygen (99.995% pure).

The sampling system assembly includes a low-pressure fused-silica sonic probe coupled to a mobile thermocouple (0.1 mm, Pt/Pt−Rh inside a thin-wall fused-silica tube). The mobility of this probeassembly allows for taking samples and measuring the temperaturealong the vertical axis of the reactor. Thermocouple measurementsshowed a good thermal homogeneity of the gas mixture (gradient< 3 K/cm).

The samples (≤50 mbar) were collected at steady residence time(τ = 1 s) and temperature. They were analyzed online by Fouriertransform infrared (FTIR) spectrometry and gas chromatography−mass spectrometry (GC−MS) and offline by GC after collection

Received: April 6, 2012Revised: June 11, 2012Published: June 18, 2012

Article

pubs.acs.org/EF

© 2012 American Chemical Society 4253 dx.doi.org/10.1021/ef300588j | Energy Fuels 2012, 26, 4253−4268

pubs.acs.org/EF

and storage at low pressure (ca. 40 mbar) in 1 L Pyrex bulbs. Indeed,to minimize adsorption of heavy components, the low-vapor-pressurespecies were analyzed online, whereas high-vapor-pressure chemicalsand permanent gases were analyzed offline. Gas chromatographsequipped with capillary columns (DB-5 ms, DB-624, Al2O3/KCl, andCarboplot-P7), a thermal conductivity detector (TCD), and a flameionization detector (FID) were used. Compound identification wasperformed analyzing the samples by GC−MS (Saturn 2000 ion-trapdetector and V1200 quadrupole, Varian), both operating in electronimpact ionization mode (70 eV). Online FTIR analyses (Nicolet Magna550, 1 cm−1 resolution) were used to quantify H2O, CO, CO2, CH2O,CH4, C2H2, and C2H4. For these measurements, the sampling probe was

Figure 1.Concentration profiles obtained from the oxidation of n-undecane in a JSR at 10 bar, τ = 1 s, and ϕ = 0.5. The initial mole fractions were XHC =0.1%, XO2

= 3.4%, and XN2= 96.5%. Experimental data (large symbols) are compared to the computations (lines and small symbols).

Table 1. Experimental Conditions for JSR Experiments at10 bar and 1 s

initial mole fractions

n-C11H24 n-C12H26 O2 N2 ϕ

0.001 0.03400 0.9650 0.50.001 0.01700 0.9820 10.001 0.00850 0.9905 2

0.001 0.03700 0.9620 0.50.001 0.01850 0.9805 10.001 0.00925 0.9898 2

Energy & Fuels Article

dx.doi.org/10.1021/ef300588j | Energy Fuels 2012, 26, 4253−42684254

connected to a temperature-controlled (413 K) gas cell (2 m path lengthand 0.5 bar) via a heated deactivated stainless-steal line (473 K and6.35 mm outer diameter). A good repeatability of the measurements,assessed by repeating experiments, and a reasonably good carbonbalance (typically 100 ± 12%) were obtained in this set of experiments.No oxygen balance could be computed because many oxygenatedintermediates could not be quantified. The experiments were carried outover the temperature range of 550−1150 K, at P = 10 bar, for variableequivalence ratios (ϕ = 0.5−2.0), and at a 1 s mean residence time.Ultimately, more than 20 species were identified and measured during

the two oxidations. The data are reported with an estimated uncertainlyof 12%.

3. MODELINGTo simulate the oxidation of n-undecane and n-dodecane in aJSR, we used the CHEMKIN package13 and the perfectly stirredreactor (PSR) computer code.14 It computes species concen-trations from the balance between the net rate of production ofeach species by chemical reactions and the difference between

Figure 2. Concentration profiles obtained from the oxidation of n-undecane in a JSR at 10 bar, τ = 1 s, and ϕ = 1. The initial mole fractions wereXHC = 0.1%, XO2




the input and output flow rates of species. The kinetic parameterswere provided in the form of three coefficients A (in cm3, mol,and s−1 units), n, and E (cal mol−1) appearing in the modifiedArrhenius equation.

= −⎜ ⎟⎛⎝

⎞⎠k AT

ERT

expn

(1)

Thermochemical properties of species in the reaction mecha-nism are provided using coefficients of NASA polynomials.15

The THERGAS program16 was used for calculating thermo-chemical properties of molecules and free radicals for which nodata were available. The chemical kinetic reaction mechanismused here is based on the work by Dievart;2 it consists of 1377species involved in 5864 reversible reactions. The oxidation sub-schemes of n-undecane and n-dodecane were consistentlywritten in terms of reactions and kinetics. They include initiationreactions (thermal decomposition via C−C and C−H bondbreaking and reaction with molecular oxygen) and metathesiswith simple radicals and atoms (H, O, OH, HO2, CH3, C2H5,

Figure 3. Concentration profiles obtained from the oxidation of n-undecane in a JSR at 10 bar, τ = 1 s, and ϕ = 2. The initial mole fractions wereXHC = 0.1%, XO2




HCO, CH3O, CH2OH, CH3O2, CH2HCO, CH3CO, 1-C3H7,1-C4H9, 1-C5H11, 1-C6H13, 1-C7H15, 1-C8H17, 1-C9H19, 1-C10H21,1-C11H23, 1-C12H25, and O2CHO). Alkyl radicals decompose byβ-scission and oxidize by reaction with O2. The low-temperatureoxidation sub-scheme follows the peroxidation pathways ofaccepted alkyl radicals (peroxidation, isomerization, and decom-position to form cyclic ethers and second peroxidation yielding todegenerate branching). The rate expressions for these reactions

were consistent with those used in previous modeling of theoxidation of alkanes and practical fuels.1,2,17

4. RESULTS AND DISCUSSION4.1. JSR Results. In the present work, the oxidation of

n-undecane and n-dodecane was studied in a JSR. The experi-ments were carried out over the temperature range of 550−1150 K, at a high pressure (P = 10 bar), at a constant residence

Figure 4. Concentration profiles of cyclic ethers formed from the oxidation of n-undecane in a JSR at 10 bar, τ = 1 s, and (a) ϕ = 0.5, (b) ϕ = 1, and (c)ϕ = 2. Experimental data (large symbols) are compared to the computations (lines and small symbols). C11H22O−AD, 2-heptyl tetrahydrofuran;C11H22O−DG, 2-butyl-5-propyl tetrahydrofuran; C11H22O−CF, 2-ethyl-5-pentyl tetrahydrofuran; and C11H22O-BE, 2-methyl-5-hexyl tetrahydrofuran.



time (1 s), and for variable equivalence ratios (ϕ = 0.5−2.0).Table 1 gives the experimental conditions used in the JSR.Concentration profiles of reactants, stable intermediates, and

final products were measured versus the temperature for threeequivalence ratios. Combustion of these compounds showedthree regimes of oxidation (the cool-flame regime,∼550−750 K;the negative temperature coefficient (NTC) regime, ∼650−750 K; and the high-temperature regime, >750 K). We studiedthe evolution of the concentrations of stable species under theseconditions.

4.1.1. Results for n-Undecane. Results of the oxidation ofn-undecane (C11H24) are presented in Figures 1−4. More than20 species were identified and measured by GC−MS, FID, andTCD.Experimental concentration profiles were obtained for O2,

H2, H2O, CO, CO2, CH2O, CH3OH, CH4, C2H4, C2H6, C2H2,CH3CHO, C3H6, C2H3CHO, C2H5CHO, 1,3-C4H6, 1-C4H8,C3H7CHO, 1-C5H10, C6H6, 1-C6H12, 1-C7H14, 1-C8H16, C11H22Oisomers, and C11H24. The main C11H22O species measured herewere substituted tetrahydrofurans.

Figure 5. Concentration profiles obtained from the oxidation of n-dodecane in a JSR at 10 bar, τ = 1 s, and ϕ = 0.5. The initial mole fractions wereXHC = 0.1%, XO2




Carbonmonoxide is the main oxygenated intermediate speciesformed, followed by formaldehyde. Ethylene is the majorunburned hydrocarbon produced in the high-temperatureregime. There is a rapid consumption of the initial hydrocarbon(C11H24) at low temperatures between 560 and 670 K. Adecrease of reactivity (decrease of the mole fraction of CO andCO2) was observed between 670 and 730 K. This decrease ofreactivity corresponds to the NTC regime. The NTC was moremarked under the fuel-rich conditions (ϕ = 2), where the fuelconversion is close to 50% at 750 K (Figure 3), whereas it is

∼80% at the same temperature and 750 K at ϕ = 0.5 (Figure 1).The fuel oxidation increased again for temperatures above∼730 K. To validate the kinetic mechanism developed here forthe oxidation of n-undecane, experimental data were comparedto simulated results (Figures 1−4). For the three equivalenceratios considered (ϕ = 0.5, 1.0, and 2.0), similar agreement wasobtained between the modeling and experimental results for themain species (CO, CO2, H2O, O2, H2, CH4, C2H4, and C3H6).The model predictions in fuel-lean conditions (ϕ = 0.5) arepresented in Figure 1. The mole fractions of C11H24 were

Figure 6. Concentration profiles obtained from the oxidation of n-dodecane in a JSR at 10 bar, τ = 1 s, and ϕ = 1. The initial mole fractions were XHC =0.1%, XO2




reasonably well-represented by the model (Figures 1−3).The cool-flame regime (T < ∼750 K) and the high-temperatureregime (T > ∼750 K) are well-simulated. For the threeequivalence ratios, formaldehyde (CH2O) was overestimatedby the model at low and high temperatures and acetaldehyde(CH3CHO) was underestimated over the temperature rangeof 550−1150 K (Figures 1−3). Mole fractions of less abundantspecies were also well-predicted by the model. For cyclic ethersformed in the low-temperature range (R + O2 = RO2, RO2 =QOOH, followed by QOOH = cyclic ether + OH), the model

represents reasonably well the data, given the higheruncertainties associated with the experimental measurements(estimated±30% because of indirect calibration), as presented inFigure 4.

4.1.2. Results for n-Dodecane. Experimental and modelingresults for the oxidation of n-dodecane (C12H26) in a JSR arepresented in Figures 5−7. Chemical analyses by FTIR and gaschromatography (GC−FID−TCD−MS) allowed for identifica-tion and measurement of species concentrations during theoxidation of n-dodecane: H2, O2, CO, CO2, H2O, CH2O,

Figure 7. Concentration profiles obtained from the oxidation of n-dodecane in a JSR at 10 bar, τ = 1 s, and ϕ = 2. The initial mole fractions were XHC =0.1%, XO2




CH3OH, CH4, C2H4, C2H2, CH3CHO, C2H6, C3H6, C2H5CHO,1-C4H8, 1,3-C4H6, C3H7CHO, 1-C5H10, C4H9CHO,C5H11CHO, C6H6, 1-C6H12, 1-C7H14, 1-C8H16, 1-C9H18, andn-C12H26. Acetone, butanone, and pentanone were also detectedin small quantities. Concentration profiles of 17 species versusthe temperature obtained during the oxidation of n-dodecaneare shown in Figures 5−7. As for the oxidation of n-undecane,we observed three regimes of oxidation: the cool-flame regime(T < ∼650 K), the NTC regime (∼650−740 K), and the high-temperature regime (T > ∼740 K).For the kinetic modeling of n-dodecane oxidation, the same

trends as for the oxidation of n-undecane were observed. Also,whereas the kinetic model agrees with the experimental trends, atϕ = 1 and 2, the computed rates of consumption of n-dodecanewere overestimated by up to a factor of 2.Measured and simulated concentration profiles showed that

the fuel conversion at the end of the NTC region decreases asthe initial concentration of oxygen decreases. Good agreement

was obtained between the model and experiment for the mainproducts (CO, CO2, CH2O, C6H6, CH4, C2H4, and 1,3-C4H6)resulting from the oxidation of n-dodecane. However, for thethree equivalence ratios (ϕ = 0.5, 1, and 2), acetaldehyde wasunderestimated over the temperature range of 550−1150 Kand formaldehyde (CH2O) was overestimated by the model(Figures 5−7). Nevertheless, JSR data obtained for the oxidationof n-dodecane were consistent with the simulated results. Table 2presents other species that have been identified by GC−MS atlow concentrations.To identify reactions responsible for the consumption of the fuel

and the formation of products, a kinetic analysis of the reaction pathsduring the stoichiometric oxidation of n-undecane at a low tem-perature (T =650K) and high pressure (P =10 bar)was performed.It indicated that the overall oxidation of the fuel occurs via

metathesis reactions with OH radicals to form undecyl radicals(Figure 8). Two degradation pathways of undecyl radicals wereobserved. The main reaction path proceeds via peroxidation ofalkyl radicals, followed by isomerization to produce alkylhydroperoxy radicals that decompose to OH and a cyclic etheror further peroxidize to form O2QOOH radicals that, in turn,yield OH and carbonyl compounds after isomerization anddecomposition.Figure 9 shows reaction paths at a high temperature (T =

1030 K), where bond breaking reactions of alkyl radicals prevail,in contradiction with what was delineated at 650 K. Nevertheless,the fuel was still mainly consumed by metathesis reactionswith OH and H radicals to form undecyl radicals. Thereafter,these radicals mainly undergo C−C bond breaking reactions to

Table 2. Minor Species Identified by GC−MS

oxygenated species large alkenes

propanal 1-hexenebutanal 1-heptenepentanal 1-octenehexanal 1-noneneacetonebutanonepentanone

Figure 8.Main reaction paths for the oxidation of n-undecane in a JSR at ϕ = 1 and T = 650 K (P = 10 bar and τ = 1 s). Species identified by GC−MSappear in red. The thickness of arrows indicates the importance of the reactions.



produce 1-olefins and 1-alkyl radicals. In summary, at a lowtemperature, the oxidation mechanism takes places viaperoxidation−isomerization processes, and at a high temper-ature, bond breaking reactions predominate.The present sensitivity analyses showed (Figure 10) that

the metathesis reaction CH2O +OH⇌HCO+H2O (S = 0.277;the rate constant taken from Tsang and Hampson18 has anuncertainty of <5) slows the consumption of n-undecane at alow temperature because it consumes OH radicals at the expenseof C11H24 + OH ⇌ C11H23 + H2O (the rate constants werecalculated on the basis of revised tabulations by Curran et al.19).Other less sensitive reactions, such as 2HO2 ⇌ H2O2 + O2 (S =0.061), also lead to lower consumption of the fuel. Indeed, H2O2

is stable at this temperature and does not contribute to chainbranching over the NTC regime. This explains the observeddecrease of reactivity in the NTC.Metathesis reactions between the fuel and OH radicals

accelerate the consumption of n-undecane and increase theglobal reactivity (n-C11H24 + OH⇌ H2O + A−DC11H23, with asensitivity coefficient S = from−0.07 to−0.184). In n-undecane,there are two equivalent carbon atoms named “A−E” and onlyone carbon named “F” (in the center). Therefore, the importanceof hydrogen abstraction on carbon F is half that on the othersecondary carbon atoms “B−E”. The reaction of primary C−H isslower than that on secondary C−H because of stronger C−H

Figure 9.Main reaction paths for the oxidation of n-undecane in a JSR at ϕ = 1 and T = 1030 K (P = 10 bar and τ = 1 s). Species identified by GC−MSappear in red. The thickness of arrows indicates the importance of the reactions.

Figure 10. Sensitivity spectrum for n-undecane during the oxidation ofundecane in a JSR at ϕ = 1 and T = 650 K (P = 10 bar and τ = 1 s).




bonds in −CH3 than in −CH2−. However, there are sixabstractable primary hydrogen atoms in the two methyl groups.Thus, the importance of hydrogen abstraction in n-undecane isin the order B ∼ C ∼ D ∼ E > A ∼ E. This is true at low andhigh temperatures. However, at a high temperature, the systemis mostly sensitive to H + O2 ⇌ OH + O, whereas metathesisreactions become less influential, as observed in Figure 11.The reactions HO2 + OH ⇌ H2O + O2 (S = 0.193; the rate

constant taken from Keyser20 has an uncertainty of <2) andCH2O + OH ⇌ HCO + H2O (S = 0.123; the rate constanttaken from Hidaka et al.21 has an uncertainty of ca. 2) consumeOH radicals needed for the oxidation of the fuel and theformation of the final products. These reactions reduce the rateof consumption of the fuel. The reaction CH3 + HO2 ⇌ CH4 +O2 (S = 0.294) strongly slows the fuel consumption rate. Themain branching reaction H + O2 ⇌ OH + O was the mostinfluential (with a sensitivity coefficient S = −0.441; the rateconstant taken from the review of Baulch et al.22 has an uncertaintyof <2) because it promotes the production of hydroxyl radicals thatare mostly responsible for the consumption of n-undecane viametathesis reactions. Hydroxyl radicals are also a product of thereaction CH3 +HO2⇌CH3O +OH, with a sensitivity coefficientS = −0.419.Sensitivity analyses were also performed for n-undecane at

ϕ = 2 (P = 10 bar and τ = 1 s). Figure 12 shows a sensitivityspectrum at a low temperature (T = 650 K). Similar results wereobserved with ϕ = 1 at the same temperature. The metathesisreaction between formaldehyde andOH radicals CH2O +OH⇌HCO + H2O (S = 0.258) slows the consumption of the fuel at alow temperature. The metathesis reactions n-C11H24 + OH ⇌H2O + A−DC11H23, with a sensitivity coefficient S = from−0.069 to −0.163, accelerate the consumption of the fuel andincrease reactivity.Figure 13 presents the sensitivity spectrum at a high temp-

erature (T = 1030 K) under the fuel-rich conditions (ϕ = 2, P =10 bar, and τ = 1 s). At ϕ = 2, the reaction CH3 + HO2⇌ CH4 +O2 (S = 0.106) has a great influence on the decrease in fuel con-sumption. The reactions that consume OH radicals also reducethe rate of consumption of n-undecane (H2 + OH⇌ H2O + H,



Figure 14. Ignition delay times for n-undecane/O2/Ar mixtures (P = 1.5bar; Xn‑undecane, 0.056%; XO2

, 0.944%; and XAr, 99%). Comparison

between experimental results from Rotavera and Petersen24 (opensymbols) and computation data (black solid line, current study; reddashed line, JetSurf, version 2.025 simulation; and blue dashed anddotted line, LLNL3 simulation).

Figure 15. Ignition delay times for n-dodecane/air mixtures.Experimental data (open symbols)4 are compared to the computations(solid line, ϕ = 0.5; dashed line, ϕ = 1); 15 < P (bar) < 34; ϕ = 0.5(Xdodecane, 1.123%; XO2

, 20.77%; and XN2, 78.10%); and ϕ = 1 (Xdodecane,

0.565%; XO2, 20.89%; and XN2

, 78.55%).



with a sensitivity coefficient S = 0.092; CH4 + OH ⇌ CH3 +H2O, with a sensitivity coefficient S = 0.090). The reactionsCH3 +HO2⇌CH3O+OH (S =−0.120; the rate constant taken

from Jasper et al.23 has an uncertainty of <2) and H + O2 ⇌OH+O (S =−0.099) accelerate the consumption of the fuel at ahigh temperature.

Figure 16. Oxidation of n-dodecane in the shock tube. Comparison between experimental results from Davidson et al.5 (black dotted line) andcomputation data (black solid line, current study; red dashed line, JetSurf, version 1.025 simulation; and blue dashed and dotted line, LLNL3 simulation).



4.2. Shock-Tube Studies. Rotavera and Petersen24 studiedthe oxidation and ignition of n-undecane in a shock tube. Inaddition, the oxidation of n-dodecane was studied in a shock tubeby Vasu et al.4 and Davidson et al.5 To further test the validity ofthe proposed mechanism under shock-tube conditions, theseexperimental data obtained in heated shock tubes4,5,24 were

compared to our modeling results. A comparison between resultsfrom our model, JetSurF, version 225 simulations, LawrenceLivermoreNational Laboratory (LLNL)3 simulations, and experi-mental data by Rotavera and Petersen24 is shown in Figure 14.Overall, the computed ignition delay times from the current studyare in a good agreement with the data by Rotavera and Petersen.However, the computed ignition delays are longer thanmeasured.However, at high temperatures (T > 1512 K), the present modelbetter represents the data than JetSurF, version 2.025 and LLNL3

models. Furthermore, in contradiction with what was observedwith the literature models,3,25 the presently computed globalactivation energy is in very good agreement with the experimentalfindings.Figure 15 presents a comparison between the experimental

data of Vasu et al.4 and our modeling. Ignition delay times wereoverestimated between 747 and 1000 K. However, at hightemperatures (T > 1000 K), ignition times were well-simulated.Low-temperature data are not well-simulated here. That couldresult from uncertainties in the data associated with possiblepreoxidation of the fuel or inaccuracies in the low-temperaturesub-scheme.Concentration time histories for five species (C12H26, CO2,

H2O, OH, and C2H4) and ignition times were measured duringn-dodecane oxidation in the shock tube by Davidson et al.5

These experiments were performed between 1300 and 1600 K,at pressures close to 2 bar and at ϕ = 1. Species time history data

Figure 17. Ignition delay times for n-dodecane/O2/Ar mixtures.Experimental data (open symbols)5 are compared to the computations(closed symbols); 2.06 < P (bar) < 2.48; and ϕ = 1.

Figure 18.Comparison between experimental data (large symbols) and modeling (solid line with small symbols, current study; thick line, the model byWestbrook et al.3) for the oxidation of n-undecane; ϕ = 1; XHC = 0.1%; XO2

= 1.7%; and XN2= 98.2%.



are compared to our computations in Figure 16. Overall, themodel represents this data set fairly well. Indeed, the con-sumption of n-dodecane was simulated correctly by the model.Good agreement was obtained between the model and experi-ment for CO2 and H2O. The decrease in the concentrationof C2H4 was predicted tardily by the mechanism (Figure 16e).This can be explained by the late consumption of hydroxyl radicalOH in the modeling (Figure 16d).The comparison between experimental data from Davidson

et al.5 and simulation results from JetSurf, version 1.025 andLLNL C8−C16

3 mechanisms used by Davidson et al.5 waspresented in Figure 16. Figure 16 shows that our mechanismdeveloped in the present study is close to the JetSurF, version 1.0model.Ignition delay times measured by Davidson et al.5 are

illustrated in Figure 17. Davidson et al.5 defined their ignitiondelay times as the time between the passage of the incident shockwave and 50% of the maximum peak emissions of OH and H2O.The same definition was used in our modeling. Experiment datawere correctly simulated by the model at high temperatures.However, the reactivity was underestimated by the model at lowtemperatures.The experimental results obtained on the oxidation of

n-undecane and n-dodecane in a JSR allowed for validationof the kinetic mechanism developed in the current study. Themodel was also validated using shock-tube ignition delay timesof n-dodecane at lower pressures and higher temperatures bymodeling the experiments of Vasu et al.12 and Davidson et al.5

4.3. Comparison of the Model Developed, ModelTaken from the Literature, and Experimental DataObtained in a JSR at ϕ = 1. This section presents acomparison between our experimental results and computedresults obtained using the mechanism developed here and thatproposed earlier by Westbrook et al.3 The current experimentaldata are defined by large symbols without lines. The results formain products of the oxidation of n-undecane (CO, CH2O,C2H4, and 1,3-C4H6) are shown in Figure 18. A similar agree-ment between data and computed results was observed for CO.This comparison showed that the model developed earlier3

predicts the mole fractions of CH2Owell but overestimates thoseof 1,3-butadiene. Also, ethylene, the main olefin formed from theoxidation of n-alkanes, was underestimated from the model byWestbrook et al.3

4.4. Comparative Study of the Oxidation of Jet A-126,n-Undecane, and n-Dodecane at ϕ=1.The results obtainedexperimentally in a JSR allowed for the comparison of theoxidation of Jet A-126 and the oxidation of two large n-alkanes,n-undecane (C11H24) and n-dodecane (C12H26), studied here.To compare the kinetics of oxidation of these three fuels, weconsidered the measured concentration profiles of CO, CO2,and CH2O obtained as a function of the temperature during theoxidation of Jet A-1, n-undecane, and n-dodecane conductedunder the same conditions. The mole fractions were scaled to11 000 ppmof carbon corresponding to 1000 ppm of n-undecaneor jet A-1 (there are 11 carbon atoms in undecane and jet A-1,

Figure 19.Normalized concentrations of CO, CO2, and CH2O during the oxidation of Jet A-1,26 C11H24, and C12H26 in a JSR (P = 10 bar; ϕ = 1; τ = 1 s;and XC = 0.11%). The data were scaled to 11 000 ppm of carbon.



whereas there are 12 carbon atoms in dodecane). The molefraction profiles are presented in Figure 19.As shown in this figure, n-dodecane and n-undecane are more

reactive at low temperatures than Jet A-1. The measured molefractions of carbon monoxide and formaldehyde formed in thecool flame were higher during the oxidation of the two n-alkanes.This can be explained by the presence of aromatics in Jet A-1.2

Aromatics have difficulties to ignite at low temperatures and tendto reduce the overall rate of oxidation of the fuel. A similarreactivity was observed between the three fuels in the high-temperature oxidation regime (T > 750 K). Overall, results showa similar oxidation behavior for the three fuels under the sameJSR conditions.Then, we compared the concentrations of several intermediate

species (CH4, C2H4, 1,3-C4H6, and CH3CHO) measured duringthe oxidations of the three fuels. These species are considered aspotential pollutants emitted by incomplete combustion of fuels.In Table 3, we report the maximum mole fractions of thesepollutants for three equivalence ratios (ϕ = 0.5, 1.0, and 2.0).Similar maximummole fractions were observed for most of thesepollutants. However, the maximum concentration of ethylenewas ∼40% less in the case of Jet A-1, because of the presenceof hydrocarbons yielding less ethylene via β-scissions thann-alkanes.

5. CONCLUSIONThe kinetics of oxidation of n-undecane and n-dodecane wasexperimentally studied in a JSR under the same initial conditions(T = 550−1150 K; P = 10 bar; equivalence ratios of 0.5−2;1000 ppm of fuel; and at a constant mean residence time τ = 1 s).A similar behavior was observed for the oxidation of

n-undecane, n-dodecane, and a conventional Jet A-1 in a JSRover the high-temperature oxidation regime. However, the puren-alkanes were more reactive than Jet A-1 under cool-flameconditions.A chemical kinetic reaction mechanism involving 1377 species

and 5865 reversible reactions was proposed on the basis ofprevious work on the kinetics of oxidation of n-alkanes.Computed results were compared to the experimental dataobtained here for the oxidation of n-undecane and n-dodecane ina JSR and for the ignition of these fuels in shock tubes.12,13

Computations showed that the detailed kinetic mechanismdeveloped here allows for correctly simulating the present JSRexperiments, whereas the model overestimates the ignitiondelays of n-dodecane under cool-flame conditions. Overall, theproposed model seems to perform better than previouslyproposed models. Sensitivity and reaction path analyses wereused to interpret the results. At low temperatures, the oxidationof the fuels proceeds via peroxidation−isomerization routes, andat high temperatures, bond breaking reactions predominate.

■ ASSOCIATED CONTENT*S Supporting InformationKinetic model used here in CHEMKIN format. This material isavailable free of charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*Telephone: +33-238-255-466. Fax: +33-238-696-004. E-mail:[email protected].

NotesThe authors declare no competing financial interest.T

able3.Maxim

umMoleFraction

s(X,p

pm)of

theMainPollutantsandUnb

urned,

Form

edatϕ=0.5,1.0,and2.0in

theJSR

Xmax(T

max)

JetA-1a

n-undecane

n-dodecane

unburned

HC

ϕ=0.5

ϕ=1.0

ϕ=2.0

ϕ=0.5

ϕ=1.0

ϕ=2.0

ϕ=0.5

ϕ=1.0

ϕ=2.0

CH

4194(870

K)

321(870

K)

852(1030K)

205(850

K)

321(900

K)

1214

(1075K)

136(870

K)

339(870

K)

911(1030K)

C2H

4323(870

K)

530(870

K)

761(945

K)

595.5(850

K)

918.8(850

K)

1209

(1075K)

640(870

K)

902.5(870

K)

1803

(945

K)

1,3-C4H

68.4(800

K)

12(870

K)

22(870

K)

12(810

K)

18(810

K)

31.3(850

K)

12(800

K)

16(800

K)

25(870

K)

CH

3CHO

89(650

K)

64(620

K)

60(800

K)

142(610

K)

134(670

K)

118(640

K)

128(602

K)

139(680

K)

105(650

K)

80(800

K)

83(800

K)

65(800

K)

170(700

K)

96(810

K)

99(810

K)

166(710

K)

122(770

K)

99(800

K)

aVolum

etric

compositio

n:13.5%

n-alkanes,28.4%

isoalkanes,2

4%cycloalkanes,and

30%

arom

atics.



http://pubs.acs.org

mailto:[email protected]

■ ACKNOWLEDGMENTS

AmirMze-Ahmed is grateful to the FrenchMinistry of Educationand Research for a doctoral grant and for an ATER position atthe University of Orleans (2011−2012). The authors thankDr. Pascal Dievart for his help.

■ REFERENCES(1) Mze-Ahmed, A.; Hadj-Ali, K.; Dievart, P.; Dagaut, P. Kinetics ofoxidation of a synthetic jet fuel in a jet-stirred reactor: Experimental andmodeling study. Energy Fuels 2010, 24, 4904−4911.(2) Dievart, P. Oxidation and combustion under ultra-lean conditionsof diesel-relevant fuels: Experimental study in a jet-stirred reactor andmodeling. Ph.D. Thesis, Universite Lille 1, Lille, France, 2008.(3)Westbrook, C. K.; Pitz, W. J.; Herbinet, O.; Curran, H. J.; Silke, E. J.A comprehensive detailed chemical kinetic reaction mechanism forcombustion of n-alkane hydrocarbons from n-octane to n-hexadecane.Combust. Flame 2009, 156 (1), 181−199.(4) Vasu, S. S.; Davidson, D. F.; Hong, Z.; Vasudevan, V.; Hanson, R.K. n-Dodecane oxidation at high pressures: Measurements of ignitiondelay times and OH concentration time histories. Proc. Combust. Inst.2009, 32 (1), 173−180.(5) Davidson, D. F.; Hong, Z.; Pilla, G. L.; Farooq, A.; Cook, R. D.;Hanson, R. K. Multi-species time-history measurements during n-dodecane oxidation behind reflected shock waves. Proc. Combust. Inst.2011, 33 (1), 151−157.(6) Dagaut, P.; El Bakali, A.; Ristori, A. The combustion of kerosene:Experimental results and kinetic modelling using 1- to 3-componentsurrogate model fuels. Fuel 2006, 85 (7−8), 944−956.(7) Dagaut, P.; Cathonnet, M. The ignition, oxidation, and combustionof kerosene: A review of experimental and kinetic modeling. Prog. EnergyCombust. Sci. 2006, 32 (1), 48−92.(8) Dagaut, P.; Reuillon, M.; Cathonnet, M. High-pressure oxidation ofliquid fuels from low to high temperature. 3. n-Decane. Combust. Sci.Technol. 1994, 103 (1−6), 349−359.(9) Dagaut, P.; Reuillon, M.; Cathonnet, M.; Voisin, D. High-pressureoxidation of normal decane and kerosene in dilute conditions from lowto high temperature. J. Chim. Phys. Phys.-Chim. Biol. 1995, 92 (1), 47−76.(10) Dagaut, P. On the kinetics of hydrocarbons oxidation from naturalgas to kerosene and diesel fuel. Phys. Chem. Chem. Phys. 2002, 4 (11),2079−2094.(11) Herbinet, O.; Marquaire, P.-M.; Battin-Leclerc, F.; Fournet, R.Thermal decomposition of n-dodecane: Experiments and kineticmodeling. J. Anal. Appl. Pyrolysis 2007, 78 (2), 419−429.(12) Le Cong, T.; Dagaut, P.; Dayma, G. Oxidation of natural gas,natural gas/syngas mixtures, and effect of burnt gas recirculation:Experimental and detailed kinetic modeling. J. Eng. Gas Turbines Power2008, 130 (4), No. 041502-10.(13) Kee, R. J.; Rupley, F. M.; Miller, J. A. CHEMKIN-II: A FortranChemical Kinetics Package for the Analysis of Gas-Phase Chemical Kinetics;Sandia National Laboratories: Livermore, CA, 1989; SAND89-8009.(14) Glarborg, P.; Kee, R. J.; Grcar, J. F.; Miller, J. A. PSR: A FORTRANprogram for modeling well-stirred reactors; Sandia National Laboratories:Livermore, CA, 1986; SAND86-8209.(15) Gordon, S.; McBride, B. J. Computer Program for Calculation ofComplex Chemical Equilibrium Compositions, Rocket Performance,Incident and Reflected Shocks and Chapman−Jouguet Detonations;National Aeronautics and Space Administration (NASA): Washington,D.C., 1971.(16) Muller, C.; Michel, V.; Scacchi, G.; Come, G. M. THERGAS: Acomputer program for the evaluation of thermochemical data ofmolecules and free radicals in the gas phase. J. Chim. Phys. Phys.-Chim.Biol. 1995, 92 (5), 1154−1178.(17) Ramirez L., H. P.; Hadj-Ali, K.; Dievart, P.; Moreac, G.; Dagaut, P.Kinetics of oxidation of commercial and surrogate diesel fuels in a jet-stirred reactor: Experimental and modeling studies. Energy Fuels 2010,24 (3), 1668−1676.

(18) Tsang, W.; Hampson, R. F. Chemical kinetic database forcombustion chemistry. 1. Methane and related compounds. J. Phys.Chem. Ref. Data 1986, 15 (3), 1087−1279.(19) Curran, H. J.; Gaffuri, P.; Pitz, W. J.; Westbrook, C. K. Acomprehensive modeling study of n-heptane oxidation. Combust. Flame1998, 114 (1−2), 149−177.(20) Keyser, L. F. Kinetics of the reaction hydroxyl + hydroperoxo→water + oxygen from 254 to 382 K. J. Phys. Chem. 1988, 92 (5), 1193−1200.(21) Hidaka, Y.; Taniguchi, T.; Kamesawa, T.; Masaoka, H.; Inami, K.;Kawano, H. High temperature pyrolysis of formaldehyde in shockwaves. Int. J. Chem. Kinet. 1993, 25 (4), 305−322.(22) Baulch, D. L.; Cobos, C. J.; Cox, R. A.; Esser, C.; Frank, P.; Just, T.;Kerr, J. A.; Pilling, M. J.; Troe, J.; Walker, R. W.; Warnatz, J. Evaluatedkinetic data for combustion modeling. J. Phys. Chem. Ref. Data 1992, 21(3), 411−734.(23) Jasper, A. W.; Klippenstein, S. J.; Harding, L. B. Theoretical ratecoefficients for the reaction of methyl radical with hydroperoxyl radicaland for methylhydroperoxide decomposition. Proc. Combust. Inst. 2009,32 (1), 279−286.(24) Rotavera, B.; Petersen, E. L. Shock-wave induced ignition ofnormal undecane (n-C11H24) and comparison to other high-molecular-weight n-alkanes. In 28th International Symposium on Shock Waves;Kontis, K., Ed.; Springer-Verlag: Berlin, Germany, 2012; pp 783−788.(25) Wang, H.; Dames, E.; Sirjean, B.; Sheen, D. A.; Tangko, R.; Violi,A.; Lai, J. Y. W.; Egolfopoulos, F. N.; Davidson, D. F.; Hanson, R. K.;Bowman, C. T.; Law, C. K.; Tsang, W.; Cernansky, N. P.; Miller, D. L.;Lindstedt, R. P. A high-temperature chemical kinetic model of n-alkane(up to n-dodecane), cyclohexane, and methyl-, ethyl-, n-propyl- and n-butyl-cyclohexane oxidation at high temperatures. JetSurF, Version 2.0;Sept 19, 2010; http://melchior.usc.edu/JetSurF/JetSurF2.0.(26) Mze-Ahmed, A. Experimental andModeling Study of Combustion ofHigh Alkanes, Reformulated Kerosenes and Surrogate Fuels: PolluantsFormation; University of Orleans: Orleans, France, 2011.



http://melchior.usc.edu/JetSurF/JetSurF2.0

Documents

Experimental and Modeling Study of the Oxidation Kinetics of n -Undecane and n -Dodecane in a Jet-Stirred Reactor