Combustion and Flame 157 (2010) 2–16

Contents lists available at ScienceDirect

Combustion and Flame

journal homepage: www.elsevier .com/locate /combustflame

Feature Article

An experimental and kinetic modeling study of n-propanoland iso-propanol combustion

Alessio Frassoldati a, Alberto Cuoci a, Tiziano Faravelli a, Ulrich Niemann b, Eliseo Ranzi a,*, Reinhard Seiser b,Kalyanasundaram Seshadri b

a Dipartimento di Chimica, Materiali e Ingegneria Chimica, Politecnico di Milano, 20133 Milano, Italyb Department of Mechanical and Aerospace Engineering, University of California at San Diego, 9500 Gilman Drive, Mail Code 0411, La Jolla, CA 92093-0411, USA

a r t i c l e i n f o

Article history:Received 25 June 2009Received in revised form 28 July 2009Accepted 3 September 2009Available online 6 October 2009

Keywords:Alcohol fuelsPropanol combustionDetailed kineticsCounterflow flames

0010-2180/$ - see front matter � 2009 The Combustdoi:10.1016/j.combustflame.2009.09.002

* Corresponding author.E-mail address: eliseo.ranzi@polimi.it (E. Ranzi).

a b s t r a c t

A kinetic model is developed to describe combustion of n-propanol and iso-propanol. It is validated bycomparing predictions made using this kinetic model with new experimental data on structures of coun-terflow non-premixed flames and previously reported data over a wide range of configurations and con-ditions. The elementary pyrolysis reactions of methanol and ethanol are well-known and were used as astarting point for extension to propanol. A detailed description leading to evaluation of rate constants forinitiation reactions, metathesis reactions, decomposition reactions, and four-center molecular dehydra-tion reactions are given. Decomposition and oxidation of primary intermediate products are describedusing a previously developed semi-detailed kinetic model for hydrocarbon fuels. The kinetic mechanismis made up of more than 7000 reactions among 300 species. The structures of counterflow non-premixedflames were measured by removing gas samples from the flame and analyzing the samples using a gaschromatograph. The flame structures were measured under similar conditions for both fuels to elucidatethe similarities and differences in combustion characteristics of the two isomers. The profiles measuredinclude those of formaldehyde, acetaldehyde, propanal, and acetone. These species are considered to bepollutants. Validation of the kinetic model was first performed by comparing predictions with experi-mental data reported in the literature obtained in flow reactors and shock tubes. In these configurations,combustion is not influenced by molecular transport. The agreement between the kinetic model andexperimental data was satisfactory. The predictions of the kinetic model were then compared withnew and previously reported experimental data on structures of counterflow non-premixed flames ofboth isomers. The agreement between the kinetic model and experimental data was again satisfactory.Satisfactory agreement was also obtained when the predictions of the kinetic model were compared withexperimental data obtained on low pressure burner stabilized premixed flames. The kinetic model wasthus validated over a wide range of temperatures (from 900 K to 2000 K), and configurations. In generalthe structures and overall combustion characteristics of n-propanol flames and iso-propanol flames aresimilar. Acetone is formed under all conditions in iso-propanol flames, while propanal is formed in n-pro-panol flames.

� 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

1. Introduction pounds, particulates, and soot. Ethanol is widely used as biofuel addi-

The need to limit the increase in greenhouse gas levels in the envi-ronment has motivated numerous studies on combustion of renew-able fuels. Oxygen containing biofuels, in particular alcohols, showconsiderable promise, because they are considered to be neutral inregard to net greenhouse gas emissions to the environment. As a con-sequence, numerous experimental and modeling studies have beencarried out to characterize combustion of methanol, ethanol, propa-nol (C3H7OH), and butanol [1–33]. Alcohols when used as additives tofossil fuels reduce the formation of polyaromatic hydrocarbon com-

ion Institute. Published by Elsevier

tive to gasoline. Recent work suggests that use of alcohols as fuels hassome deleterious effects on human health, due to high emission lev-els of toxic oxygenated by-products such as aldehydes [34]. More-over, high concentrations of acetone ((CH3)2CO) were detected inthe exhaust gas of a spark ignition engine upon addition of iso-propa-nol to a synthetic fuel [35]. The formation of oxygenated pollutants isa common feature of alcohol combustion that calls for detailedkinetic studies of the combustion chemistry of alcohol fuels [32].Here, a kinetic modeling study of combustion of the propanol iso-mers, n-propanol and iso-propanol is carried out.

Only few kinetic studies of oxidation of n-propanol and iso-pro-panol are reported in the literature; they include experimentalstudies using batch [24–26] and flow reactors [10], shock-tubes

Inc. All rights reserved.

A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16 3

[33], co-flow diffusion flames [1], counterflow diffusion flames[29], droplet burning [30] and low-pressure premixed flat flames[31,32]. The combustion and pyrolysis of n-propanol has been alsoinvestigated in reflected shock waves with particular attention tothe formation of soot and polycyclic aromatic hydrocarbons [23].Fundamental investigations include studies on thermal decompo-sition of iso-propanol [27] and structures of flames over liquidpools [28]. The structure of non-premixed iso-propanol flameswas measured previously employing the co-flow configuration[1] and the counterflow configuration [29]. The structure of diffu-sion flames depend on the stoichiometric mixture fraction and thestrain rate [36–38]. The previous measurements of the structure ofiso-propanol flames were made with the oxidizer stream made upof air enriched with oxygen [29]. The experimental conditionswere characterized by high values of the stoichiometric mixturefraction and low values of the strain rates. Combustion in practicalsystems are characterized by low values of the stoichiometric mix-ture fraction and both low and high values of the strain rates [39].To complement the previous studies new experimental data on thestructure of counterflow diffusion flames of n-propanol andiso-propanol are presented in the next sections of the paper. Thestructure was measured with air as the oxidizer. The experimentalconditions are characterized by low values of stoichiometric mix-ture fraction and moderately high values of strain rate. The newexperimental data allow comparison of the flame structures of pro-panol isomers. The new experimental data together with previousexperimental data are helpful in a further tuning and validation ofthe kinetic model of alcohol fuels.

2. Experimental measurements of flame structure

The structures of non-premixed flames of n-propanol and iso-propanol were measured employing the counterflow configura-tion. Fig. 1 shows a schematic illustration of the counterflowconfiguration. Steady, axisymmetric, laminar flow of two counter-flowing streams toward a stagnation plane is considered. In thisconfiguration a fuel stream made up of prevaporized fuel (n-propa-nol or iso-propanol) and nitrogen is injected from the fuel duct, andan oxidizer stream of air is injected from the oxidizer duct. Thesejets flow into the mixing layer between the two ducts. The exitof the fuel duct is called the fuel boundary and the exit of the oxi-dizer duct the oxidizer boundary. The mass fraction of fuel, thetemperature, and the component of the flow velocity normal tothe stagnation plane at the fuel boundary are represented by YF,1,T1, and V1, respectively. The mass fraction of oxygen, the tempera-

Fig. 1. Schematic illustration of the counterflow configuration. The figure shows thequartz microprobe used for measuring flame structure.

ture, and the component of the flow velocity normal to the stagna-tion plane at the oxidizer boundary are represented by YO2 ;2, T2,and V2, respectively. The distance between the fuel boundary andthe oxidizer boundary is represented by L = 10 mm.

The value of the strain rate, defined as the normal gradient ofthe normal component of the flow velocity, changes from the fuelboundary to the oxidizer boundary [40]. The characteristic strainrate on the oxidizer side of the stagnation plane a2 is given by [40]

a2 ¼2jV2j

L1þ jV1j

ffiffiffiffiffiffiq1p

jV2jffiffiffiffiffiffiq2p

� �ð1Þ

Here, q1 and q2 represent the density of the mixture at the fuelboundary and at the oxidizer boundary, respectively. The stoichi-ometric mixture fraction, Zst is [36–38]

Zst ¼ 1þ tYF;1=YO2 ;2� ��1 ð2Þ

where t is the stoichiometric mass ratio of oxygen to fuel.The profiles of concentration of stable species were measured

for YF,1 = 0.3, T1 = 353 K, YO2 ;2 = 0.233, T2 = 298 K, a2 = 97.5 s�1,V1 = 0.235 m/s, V2 = 0.25 m/s, and L = 10 mm. At these conditionsthe stoichiometric mixture fraction Zst = 0.2449. Concentrationsof stable species were measured by removing gas samples fromthe reaction zone using a heated quartz microprobe, and analyzingthem in a gas chromatograph. The microprobe has a tip with an in-ner diameter of 150 lm. To minimize disturbances to the flow-field, the tip of the microprobe was placed at a location of 5 mmoff the axis of symmetry as shown in Fig. 1. The location of thesampling probe in the flow-field was determined using a digitalphoto camera. The size of one pixel in the camera corresponds toa distance of approximately 20 lm in the flow-field. The mole frac-tions of various species in the sample were measured using an Agi-lent 3000microGC gas chromatograph. This instrument is equippedwith a 10 m long molecular sieve 5A column, a 8 m long Poraplot Ucolumn, a 8 m long Poraplot Q column, and a 8 m long OV-1column. The gas chromatograph has a built-in sample pump. Toensure equal sample sizes and thus comparable results across allmeasurements, a constant sample inlet pressure is required. There-fore species were sampled at a constant pressure of 600 mbar intoa sample vessel. Nitrogen was then introduced until the vessel at-tained a total pressure of 1150 mbar. After a waiting period of6 min, to allow sufficient mixing, the sample was drawn into thegas chromatograph. All lines from the sample probe to the gaschromatograph were heated to 373 K. The molecular sieve columnuses argon as a carrier gas. It was used to separate hydrogen (H2),oxygen (O2), nitrogen (N2), methane (CH4), and carbon monoxide(CO). All other columns use helium as a carrier gas. The PoraplotU column was used for separating carbon dioxide (CO2), ethylene(C2H4), ethane (C2H6), acetylene (C2H2), and formaldehyde(CH2O). The Poraplot Q column was used for separating water(H2O), propyne (C3H4), propylene (C3H6), propane (C3H8), acetalde-hyde (C2H4O), butene (C4H8), butadiene (C4H6), acetone (C3H6O)and propanal (C3H6O). The OV-1 column was used for separatingn-propanol and iso-propanol. The mole fractions of various specieseluting from the columns were measured using thermal-conduc-tivity detectors (TCD). The detectors were calibrated using samplesof known composition. For those species that are liquids at roomtemperature and atmospheric pressure a sample vessel is used.First the sample vessel is evacuated. Next the liquid is injectedwith a syringe through a septum in the wall of the sample vessel.It is then diluted with nitrogen. The species fraction is determinedfrom the pressure reading after evaporation. This procedure wasused for n-propanol, iso-propanol, water, formaldehyde, propional-dehyde, and acetone. For those species that are gases at room tem-perature and atmospheric pressure, the calibration is performedwith calibration gases of known composition. The peaks for all

4 A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16

species presented here show very good separation. Therefore, theexpected accuracy for the maximum concentrations of all speciesexcept H2O and CH2O is expected to be better than ±10%. Theexpected accuracy for H2O and CH2O is better than ±20%. The accu-racy for formaldehyde is based on the signal size compared to thebaseline noise. The calibration for formaldehyde showed very littledeviation and good repeatability.

The experimental data obtained here are compared with calcu-lated profiles. They are also compared with the experimental dataof Sinha and Thompson [29]. These previous experimental datawere obtained for YF,1 = 0.157, T1 = 318 K, YO2 ;2 = 0.422, T2 = 298 K,a2 = 20.0 s�1, V1 = 0.10 m/s, V2 = 0.10 m/s, and L = 20 mm. At theseconditions the stoichiometric mixture fraction Zst = 0.5283.

3. Detailed kinetic mechanism of n-propanol and iso-propanol

The pyrolysis and oxidation mechanisms of propanol isomersare very similar to those for hydrocarbon fuels. Therefore, thedevelopment of a complete set of the primary propagation reac-tions for these fuels requires the study and the definition of fewnew kinetic parameters for reactions involving bonds and H atomsnear to the OH group. The elementary pyrolysis and oxidation reac-tions of methanol and ethanol are reasonably well-known and havebeen revised recently [9,10,13–16]. The kinetic mechanism formethanol and ethanol are a useful starting point for the extensionto the kinetic schemes of n-propanol and iso-propanol. Initiationreactions are generally evaluated, by assuming a reference fre-quency factor with the activation energy equal to the bond energy,and microscopic reversibility based on the reverse radical recombi-nation reaction is applied. Metathesis reactions require defining thereactivity of the H atoms in hydroxyl position and the H atoms in aposition. Remaining H atoms are presumed to be unaffected by thepresence of the OH group. Decomposition reactions of thecorresponding alkoxy and parent radicals from alcohol fuels needfurther discussion. Isomerization reactions of these radicals are oflimited importance and are neglected here. Finally, the class ofthe four-center molecular dehydration reactions requires carefulattention. In this section, unimolecular reactions are discussed firstfollowed by metatheses reactions, decomposition reactions of pri-mary radicals from alcohol fuels, and four-center molecular dehy-dration reactions. The rate constant, ki for reaction i is written aski = AiT

niexp[�Ei/(RT)], where Ai is the frequency factor, Ei is the acti-vation energy in cal/mol, T the temperature, ni the temperatureexponent, and R is the gas constant.

3.1. Unimolecular initiation reactions

The activation energy of initiation reactions are evaluated fromthe strength of the C–C bond by defining the bond energy of pri-mary and secondary C atoms (Cp and Cs) with OH substitutions.Thus,

CH3CH2CH2OH ¡ C2H�5 þ �CH2OH

k ¼ 2:0� 1016 exp½�85000=ðRTÞ� ½s�1�

CH3CHOHCH3 ¡ CH�3 þ �CHOH—CH3

k ¼ 6:0� 1016 exp½�85000=ðRTÞ� ½s�1�

The following kinetic parameters are prescribed for the reactions

CH3CH2CH2OH ¡ 1—C3H�7 þ �OH

k ¼ 1:5� 1016 exp½�93200=ðRTÞ� ½s�1�

CH3CHOHCH3 ¡ 2—C3H�7 þ �OH

k ¼ 1:5� 1016 exp½�93200=ðRTÞ� ½s�1�

They are the same as that for the ethanol initiation reactionC2H5OH ¡ n-C2H�5 þ �OH. Note that these reference kinetic parame-ters are not affected by the nature of the C atom. More than100 kcal/mol is required to release the H atom from the OH group,therefore reactions of this type cannot contribute to fuel decompo-sition. On the contrary, the reverse reactions could affect flamepropagation. For this reason, all these reactions are included inthe overall mechanism with the same kinetic parameterk = 5.0 � 1010 [l mol�1 s�1]. Kinetic data for the remaining initiationreactions, that involve the splitting of the C–C bond, are similar tothose used for alkanes. Thus,

CH3CH2CH2OH ¡ CH�3 þ �CH2—CH2OH

k ¼ 2:0� 1016 exp½�88000=ðRTÞ� ½s�1�

3.2. Metathesis reactions

The energy of the O–H bond in the OH group is greater than100 kcal/mol and is similar to that of a C–H bond of primary Hatoms in alkyl radicals. As assumed in the formation of methoxyand ethoxy radicals from methanol and ethanol, the kinetic param-eters of this H-atom abstraction from the alcohol functional groupare assumed to be equal to those for the abstraction of a primary Hatom from a methyl group. Galano et al. [41] studied the gas phasereactions of alcohols with the OH radical employing a quantummechanical approach. Moving away from the previous values rec-ommended by Atkinson et al. [42,43], they concluded that the ratecoefficient corresponding to the a channel (ka) is larger than thoseof the other competing channels (kb,kc,k0), with the different chan-nels reactivity order: ka > kb > kc > k0. The a branching ratio (Ca) forn-propanol goes from �0.5 to 0.7, over the temperature range 290–500 K. These values are in line with the value 0.73 derived from thedata recommended by Atkinson et al. [42,43]. The branching ratioCb = 0.12 suggested by Dunlop and Tully [44] for iso-propanolagrees with the calculations developed by Galano et al. [41]. Theyconfirm a value of �0.10, thus supporting the higher reactivity ofthe H atoms of the a channel.

On the basis of these previous studies, we assume the followingkinetic parameters for the H-abstraction reactions of OH radicals(units are: cal, l, mole, s, K):

OHþ CH3CH2CH2OH ¡ H2Oþ CH3CH2CH2O�

k0 ¼ 400� T2 � exp½500=ðRTÞ�OHþ CH3CH2CH2OH ¡ H2Oþ CH3CH2CH�OH

ka ¼ 1200� T2 � exp½2260=ðRTÞ�OHþ CH3CH2CH2OH ¡ H2Oþ CH3CH�CH2OH

kb ¼ 800� T2 � exp½2260=ðRTÞ�OHþ CH3CH2CH2OH ¡ H2Oþ CH�2CH2CH2OH

kc ¼ 1200� T2 � exp½500=ðRTÞ�OHþ CH3CHOHCH3 ¡ H2Oþ CH3CHO�CH3

k0 ¼ 400� T2 � exp½500=ðRTÞ�OHþ CH3CHOHCH3 ¡ H2Oþ CH3C�OHCH3

ka ¼ 600� T2 � exp½3180=ðRTÞ�OHþ CH3CHOHCH3 ¡ H2Oþ CH�2CHOHCH3

kb ¼ 2400� T2 � exp½500=ðRTÞ�

We preferred to maintain, also for these H sites, our systematic ap-proach for metathesis reactions described elsewhere [45,46] andextensively used in the entire kinetic scheme [47]. Thus, kineticparameters of H abstractions from CH3–CH2–CH2–OH and from(CH3)2–CH–OH in a position are obtained by increasing by 50% thefrequency factors of secondary and tertiary H atoms, respectively.

+ H2O+ H2O

OH

H

OH

H

OH

H

+ H2O+ H2O

HOH

CH3CH2CH2OH

CH3CHOHCH3

Fig. 2. Dehydration reactions of n-propanol and iso-propanol to form propylene.

A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16 5

In this way, the a branching ratio for n-propanol is �0.5, and 0.6 foriso-propanol. Metathesis reactions of H, CH3 and other abstractingradicals are treated according to the same approach.

3.3. Decomposition reactions of primary radicals from propanol

The expressions for the rate constants for C1–C4 alkyl andalkoxy radicals decomposition via b-scission have been revised re-cently [48]. Recommended rate parameters of the b-decompositionreactions are calculated using microscopic reversibility based onthe reverse addition reactions of H or an alkyl radical to an olefinor carbonyl species. Recommended rate constants for C1–C4 alkoxyradical decompositions are based on an extensive study of availableexperimental data. Thus, ethoxy radical can decompose to formeither formaldehyde and methyl radical, or acetaldehyde and ahydrogen atom, with the following suggested kinetic rates:

CH3CH2O� ¡ CH2Oþ CH�3 k ¼ 5:0� 1013 exp½�18700=ðRTÞ� ½s�1�

CH3CH2O� ¡ CH3CHOþH� k ¼ 5:0� 1013 exp½�22200=ðRTÞ� ½s�1�

These rate expressions on one hand indicate a very high reactivityof alkoxy radicals, and on the other hand show a limited selectivitytoward the dehydrogenation channel. In the temperature range700–1000 K the first reaction channel that gives formaldehyde is fa-vored by about a factor of ten. Even lower selectivity towards dehy-drogenation paths are observed for C3 alkoxy radicals than thoseindicated above for C2 radical. Thus

CH3CH2CH2O�¡ CH2Oþ C2H�5k ¼ 5:0� 1013 exp½�14700=ðRTÞ� ½s�1�

CH3CH2CH2O� ¡ CH3CH2CHOþH�

k ¼ 5:0� 1013 exp½�23100=ðRTÞ� ½s�1�CH3CHO�CH3 ¡ CH3CHOþ CH�3

k ¼ 5:0� 1013 exp½�13000=ðRTÞ� ½s�1�CH3CHO�CH3 ¡ CH3COCH3 þH�

k ¼ 5:0� 1013 exp½�20700=ðRTÞ� ½s�1�

These considerations could lead to simplifications of the overallmechanism and the kinetic model. Alkoxy radicals do not signifi-cantly interact with the reacting system and could be consideredinstantaneously transformed into their final decomposition prod-ucts. This limits the number of intermediate radical species in-volved in the overall kinetic scheme.

Kinetic data for decomposition reactions of alkyl radicals are thesame as those used in the case of alkanes. The activation energy ofreactions that involve the splitting of C–C or C–H bonds in a posi-tion of the OH group requires a careful analysis. Following previouswork on ethanol decomposition, the following decomposition ratesare assumed for n-propanol radicals:

CH�2CH2CH2OH ¡ C2H4 þ CH2OH�

k ¼ 3:0� 1013 exp½�30000=ðRTÞ� ½s�1�CH3CH�CH2OH ¡ C3H6 þ OH�

k ¼ 3:0� 1013 exp½�36000=ðRTÞ� ½s�1�CH3CH�CH2OH ¡ C3H6OþH�

k ¼ 6:0� 1013 exp½�36000=ðRTÞ� ½s�1�CH3CH2CH�OH ¡ CH3CHOþ CH�3

k ¼ 3:0� 1013 exp½�32500=ðRTÞ� ½s�1�CH3CH2CH�OH ¡ C3H6OþH�

k ¼ 6:0� 1013 exp½�36000=ðRTÞ� ½s�1�

Similarly, the following decomposition reactions of iso-propanolradicals are postulated:

CH�2CHOHCH3 ¡ CH3COCH3 þH�

k ¼ 6:0� 1013 exp½�35000=ðRTÞ� ½s�1�CH�2CHOHCH3 ¡ CH3CHOþ CH�3

k ¼ 3:0� 1013 exp½�36000=ðRTÞ� ½s�1�CH�2CHOHCH3 ¡ C3H6 þ OH�

k ¼ 3:0� 1013 exp½�37000=ðRTÞ� ½s�1�CH3C�OHCH3 ¡ CH3COCH3 þH�

k ¼ 6:0� 1013 exp½�36000=ðRTÞ� ½s�1�

In these dehydrogenation reactions, the methyl-vinyl alcoholformed is presumed to be instantaneously transformed into ace-tone, via keto-enol tautomerism. The isomerization reactions ofthese radicals can be reasonably neglected, due to their very limitedweight in respect of the corresponding decomposition reactions.Thus, the isomerization reaction of the primary radical of n-propa-nol to form the alkoxy radical, via a five membered ringintermediate:

CH�2CH2CH2OH ¡ CH3CH2CH2O�

k ¼ 1:0� 1011 exp½�20600=ðRTÞ� ½s�1�

could become significant only at temperatures lower than 800 K.This is further enhanced for the reverse reaction of the alkoxy rad-ical. The remaining isomerization reactions would require 4-mem-bered ring intermediates and are therefore negligible.

3.4. Four-center molecular dehydration reactions

This class of reactions involves a four-center cyclic transitionstate with the formation of parent alkenes and H2O, as shown inFig. 2. Thus both propanol isomers form propylene with a fourmembered ring intermediate. Several kinetic parameters havebeen suggested for this class of reactions [12,13,19,20,27,49]. Thefollowing kinetic parameters are assumed for the dehydrationreactions of the two propanol isomers:

CH3CH2CH2OH ¡ C3H6 þH2O

k ¼ 2:0� 1014 exp½�67100=ðRTÞ� ½s�1�CH3CHOHCH3 ¡ C3H6 þH2O

k ¼ 2:0� 1014 exp½�67100=ðRTÞ� ½s�1�

These values for the rate constants are consistent with the recentkinetic analysis of n-butanol dehydration reactions [21]. This reac-tion is endothermic by 8.25 kcal/mol with an associated activationenthalpy of �67.5 kcal/mol. Fig. 3 is a schematic illustration of n-propanol and iso-propanol decomposition. The detailed sub-mecha-nism of n-propanol and iso-propanol is reported in Table 1. Furtherdecomposition and/or oxidation reactions of primary intermediateproducts are described in a semi-detailed oxidation mechanismfor hydrocarbon fuels up to C16 developed in previous studies[47,50]. The overall kinetic scheme is based on hierarchical

6 A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16

modularity and is made up of more than 7000 reactions among 300species. Thermochemical data for most species was obtained fromthe CHEMKIN thermodynamic database [51,52]. For those speciesfor which thermodynamic data is not available in the literature,the group additive method was used to estimate these properties[53]. The complete mechanism, with thermodynamic and transportproperties, is available online in CHEMKIN format [54].

4. Numerical methods and simulations

The kinetic model described in the previous section was vali-dated by comparing the results of numerical simulations withnew and previous experimental data on counterflow non-pre-mixed flames, and available experimental data obtained employingshock tube and flow reactor.

The DSMOKE code was used to numerically solve the system ofequations that describe various aspects of combustion in shocktube and ideal flow reactor [55]. To predict aspects of combustionthat include molecular transport and chemical reactions a 1-D lam-inar flame model was employed. A detailed description of this codeis given elsewhere [56]. This code includes multicomponent diffu-sion and thermal diffusion. Discretization of the differential equa-tions is carried out using conventional finite differencingtechniques for non-uniform mesh spacing. The numerical problemcorresponds to a large system of differential–algebraic equations(DAE). The specifically conceived methods and solver routines ofBzzMathLibrary [57,58] are used to handle the complexity of thisnumerical problem. More than 300 grid points are used to ensure

-H

-H2O

-H -H

OH

OH

-CH3 -H

O OH

O OH

-H-H

CH2O

-C2H5

OH-H2O

-H -H

OH

OH OHOHOH

-H -CH3

OOOO

-CH3

OOO

-OH

n-propa

iso-prop

Fig. 3. Reaction mechanism of n-propan

grid-insensitive results. This 1-D laminar flame code is used hereto predict the structure of counterflow non-premixed flames andpremixed flames. In premixed burner-stabilized flames experi-ments there are often significant heat losses to the burner whichare difficult to estimate. Therefore, in the numerical simulationsthe measured temperature profile is used as an input parameter.As a consequence, for premixed flames, the energy conservationequation is not included in the system of equations that are solved.The boundary conditions employed in flame calculations are iden-tical to those in the experiments.

5. Model predictions and comparisons with experimental data

Norton and Dryer [10] studied the oxidation of propanol isomersin a flow reactor at atmospheric pressure, temperatures of 1000–1100 K and residence time of �0.1 s. They investigated propanoloxidation under fuel lean conditions with high N2 dilution. Recentlyignition delay times have been measured for n-propanol and iso-propanol in shock tubes [33]. New experimental data on the struc-ture of counterflow non-premixed n-propanol and iso-propanolflames have been obtained and are discussed later. Similar non-premixed flames of pure iso-propanol and different mixtures withpropane were investigated by Sinha and Thomson [29].

Kasper et al. [32] studied the combustion of the two propanolisomers in low-pressure, premixed flat flames using two indepen-dent MBMS techniques. For each alcohol, a set of three flames at�2000 K with different stoichiometries was measured, providingan extensive data base. Profiles of stable and intermediate species,

-H

-H

OHOHOH

-H

OHOHOH

-OH -CH2OH

OHOHOH

-H

-H

O

-H

OH

-H-H

OH

nol

anol

ol and iso-propanol decomposition.

Table 1Sub-mechanisms of n- and iso-propanol decomposition.

Reactiona A n Ea

n-Propanol1 nC3H7OH ¡ CH3 + C2H4OH 2.0 � 1016 0 88,0002 nC3H7OH ¡ C2H5 + CH2OH 2.0 � 1016 0 85,0003 nC3H7OH ¡ OH + nC3H7 1.5 � 1016 0 92,3004 nC3H7OH ¡ CH2CH2CH2OH + H 3.6 � 1019 �1 102,8005 nC3H7OH ¡ CH3CHCH2OH + H 1.8 � 1019 �1 99,6006 nC3H7OH ¡ CH3CH2CHOH + H 7.7 � 1018 �1 97,4007 R + nC3H7OH ? RH + CH2CH2CH2OH 3HPrim

8 R + nC3H7OH ? RH + CH3CHCH2OH 2HSec

9 R + nC3H7OH ? RH + CH3CH2CHOH 2HOHSec

10 R + nC3H7OH ? RH + CH3CH2CH2O 1HPrim

11 CH3CH2CH2O ¡ CH2O + C2H5 5.0 � 1013 0 14,70012 CH2CH2CH2OH ¡ C2H4 + CH2OH 3.0 � 1013 0 30,00013 CH2CH2CH2OH ¡ H+C3H5OH 3.0 � 1013 0 36,00014 CH3CHCH2OH ¡ C3H6 + OH 3.0 � 1013 0 36,00015 CH3CHCH2OH ¡ H + C2H5CHO 6.0 � 1013 0 36,00016 CH3CH2CHOH ¡ CH3CHO + CH3 3.0 � 1013 0 32,50017 CH3CH2CHOH ¡ H + 0.5C2H5CHO + 0.5C3H5OH 6.0 � 1013 0 36,00018 O2 + CH2CH2CH2OH ¡ HO2 + C2H5CHO 1.5 � 109 0 500019 O2 + CH3CHCH2OH ¡ HO2 + 0.5C2H5CHO + 0.5C3H5OH 1.5 � 109 0 500020 O2 + CH3CH2CHOH ¡ HO2 + C3H5OH 1.5 � 109 0 500021 nC3H7OH ¡ C3H6 + H2O 2.0 � 1014 0 67,100

iso-Propanol1 iC3H7OH ¡ CH3 + CH3CHOH 6.0 � 1016 0 85,0002 iC3H7OH ¡ OH + iC3H7 1.5 � 1016 0 93,2003 iC3H7OH ¡ CH3COHCH3 + H 7.0 � 1018 �1 97,8004 iC3H7OH ¡ CH3CHOHCH2 + H 3.3 � 1019 �1 103,2005 R + iC3H7OH ? RH + CH3CHOCH3 1HPrim

6 R + iC3H7OH ? RH + CH3CHOHCH2 6HPrim

7 R + iC3H7OH ? RH + CH3COHCH3 2HOHTert

8 CH3CHOCH3 ¡ CH3 + CH3CHO 5.0 � 1013 0 13,0009 CH3CHOCH3 ¡ CH3COCH3 + H 5.0 � 1013 0 20,700

10 CH3COHCH3 ¡ CH3COCH3 + H 6.0 � 1013 0 36,00011 CH3CHOHCH2 ¡ CH3 + CH3CHO 3.0 � 1013 0 36,00012 CH3CHOHCH2 ¡ CH3COCH3 + H 6.0 � 1013 0 35,00013 CH3CHOHCH2 ¡ C3H6 + OH 3.0 � 1013 0 37,00014 O2 + CH3COHCH3 ¡ HO2 + CH3COCH3 1.5 � 109 0 500015 O2 + CH3CHOHCH2 ¡ HO2 + CH3COCH3 1.5 � 109 0 500016 iC3H7OH ¡ C3H6 + H2O 2.0 � 1014 0 67,100

a k = A�Tn�exp(�Ea/RT). Units are: mole, l, s, K and cal.

A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16 7

including several radicals, were measured as a function of heightabove the burner. Finally, Li et al. [31] investigated lean and richpremixed flames of n-propanol and iso-propanol at low pressure,aiming at a better understanding of the combustion chemistry ofC3 alcohols. They used the synchrotron photoionization andmolecular-beam mass spectrometry (PI-MBMS) techniques tomeasure radical intermediates and stable species in the flame[31]. Validation of the kinetic mechanism was carried out over awide range by comparing predictions with experimental data ob-tained: (1) in flow reactors [10], (2) ignition delay times measuredin shock tube [33], (3) new and previous [29] measurements ofstructures of counterflow non-premixed flames, and (4) low-pres-sure premixed flames [31,32]. While the complete set of predictedresults and relevant comparisons with the experimental measure-ments is reported in the Supplemental material, the key and crucialfeatures are described in the following sections.

5.1. Flow reactor at atmospheric pressure

Figs. 4 and 5 compare predictions of the kinetic model withexperimental data obtained in flow reactor [10]. Fig. 4 shows com-parison for n-propanol and Fig. 5 for iso-propanol. The agreementbetween model predictions and experimental measurements aresatisfactory. Note that, following usual practice employed for com-parison with flow reactor data [16], predicted profiles are shiftedby �20–30 ms in order to account for non-ideal reactant mixing.

Moreover, to better match the fuel conversion, predicted profilesare obtained with the input temperature profile increased by�5–10 K above the measured profile for both the fuels. Experimen-tal data clearly confirm the higher reactivity of n-propanol with re-spect to iso-propanol. Propylene (C3H6) yields from iso-propanolare higher than 35% while less than one half is obtained fromn-propanol. The larger propylene formation from iso-propanol ismainly due to the molecular dehydration reaction of the fuel. Asshown in Fig. 3, H-abstraction reactions form three different inter-mediate radicals C3H7O whose major decomposition product isacetone. n-Propanol is the most reactive of the two isomers, dueto the larger importance of radical pathways with the significantformation of propanal (C2H5CHO). Acetone yield from iso-propanolpeaks to �20%, while the maximum yield of propanal from n-pro-panol is �10%. Radical decomposition paths prevail on the n-pro-panol decomposition. As usual, H-abstraction reactions of OH andH radicals are the dominant ones (�65% and �25%, respectively),with a minor role of methyl (�7%) and oxygen radicals. On the con-trary, molecular dehydration of iso-propanol is the main source ofpropylene and it accounts for more than 50% of the overall conver-sion in these conditions. It seems relevant to underline that ethyl-ene (C2H4) is only a secondary product of iso-propanol, its scarceformation moves through the H addition to propylene, the forma-tion of 1-propyl radical and the successive de-methylation reac-tion. On the contrary the formation of ethane (C2H6), due to therecombination reactions of methyl radicals, is over-predicted.

0.0

0.2

0.4

0.6

0.8

1.0

Time [ms]

Mol

e fr

actio

n [%

]n-propanol ×3COCO2

0

100

200

300

400

Time [ms]

Mol

e fr

actio

n [p

pm]

C3 H6

CH3CHOC2 H5CHOC2 H2

0.00

0.04

0.08

0.12

0.16

0.20

0 20 40 60 0 20 40 600 20 40 60Time [ms]

Mol

e fr

actio

n [%

]

C2 H4

C2 H6

C2 H4

CH4

Fig. 4. Comparison of predictions of the kinetic model with experimental data obtained in flow reactor for n-propanol. The symbols represent experimental data [10] and thecurves are predictions of the kinetic model.

C3 H4

CO2 × 0.5

C2CH CHO

iso -propanolCO × 0.5

C3 H6

C2 H4C2 H6

CH4

C3 H6 O

0

40

80

120

160

200

Time [ms]M

ole

frac

tion

[ppm

]

0

400

800

1200

1600

Time [ms]

Mol

e fr

actio

n [p

pm]

0

400

800

1200

1600

2000

0 20 40 60 80 1000 20 40 60 80 1000 20 40 60 80 100Time [ms]

Mol

e fr

actio

n [p

pm]

C3H4

CO2 × 0.5

C2H2CH3 CHO

iso -propanolCO × 0.5

C3 H6

C2 H4C2 H6

CH4

C3 H6 O

Fig. 5. Comparison of predictions of the kinetic model with experimental data obtained in flow reactor for iso-propanol. The symbols represent experimental data [10] andthe curves are predictions of the kinetic model.

8 A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16

Methane is well predicted and this seems to indicate a reasonableconcentration of methyl radicals. The large formation of ethylene,and model under-prediction, could also induce perplexity on theaccuracy of ethane–ethylene separation.

Other minor products, such as propadiene (C3H4) and allene andC4 species are also well predicted. The experimental data of ther-mal decomposition of iso-propanol in a batch reactor at 0.058 barand 766 K (Trenwith [26]) allow to better validate the radicalpaths. In fact, molecular dehydration reaction is not significant atthese conditions. The 30% iso-propanol decomposition observedand properly predicted by the model after �400 s is only due tothe radical reaction mechanism. Pyrolysis of both n-propanol andiso-propanol at temperatures of 800–900 K in a closed batch reac-tor was also investigated by Barnard and Hughes [24] and Barnard[25]. Model prediction underestimates propanol conversion by afactor of �2 for both the fuels. According to the experimental mea-surements, model prediction confirms that pyrolysis of iso-propa-nol is faster than the corresponding one of n-propanol, in theseconditions. The formation of tar components and the large deficitin the C/H/O balances observed in experimental data does not al-low detailed comparisons of model predictions with experimentaldata.

5.2. Ignition delay times in shock tube

Recently, ignition delay times of reactive mixtures of eithern-propanol or iso-propanol with oxygen and argon have been mea-sured behind reflected shock waves at high temperatures (1350–2000 K) and atmospheric pressure [33]. Fuel-lean, stoichiometricand fuel-rich mixtures were considered. Pressure measurementsand CH* emissions were used to determine ignition delay times.The influence of temperature and stoichiometry on ignition delay

times has been characterized and also compared to the predictionsof a detailed kinetic mechanism currently under development [33].This study and experimental data [33] shows the similar behaviorof the two fuels confirming the greater reactivity of n-propanol rel-ative to corresponding mixtures of iso-propanol only at low tem-peratures. The kinetic model developed here was also used topredict the ignition delay times. Fig. 6 compares the predictionsof ignition delay times [ls] as a function of 1000/T, with experi-mental data. Here T is the initial temperature of the reactive mix-ture. The numerical simulations were performed for values ofequivalence ratio, U equal to 0.5, 1.0, and 2.0. For U = 1.0, andU = 2.0, the fuel mass fraction in the reactive mixture was 0.005,and for U = 0.5 the fuel mass fraction was 0.0025. The predictionsof the kinetic model agrees well with experimental data at lowtemperatures. Larger deviations between predictions and dataare mainly observed at temperatures higher than 1700 K. The rel-evant curvature at high temperature is not reproduced by the mod-el. The apparent activation energies predicted are in the order of45–50 kcal/mol, while those calculated using the measured dataare lower than 35 kcal/mol. This clearly points to further experi-mental and kinetic modeling analysis.

5.3. Counterflow non-premixed flame of n-propanol and iso-propanol

Experimental measurements on counterflow non-premixedflames discussed in Section 2 are compared with model predictionsin Figs. 7 and 8. Fig. 7 shows the structure of a non-premixed iso-propanol flame while Fig. 8 shows the structure of a non-premixedn-propanol flame. These figures show the mole fraction of variousspecies as a function of distance from the fuel boundary. The sym-bols in these figures represent experimental data and the lines aremodel prediction. The profiles in these figures were obtained at a

Fig. 6. Comparison of predictions of ignition delay times [ls], as a function of the reciprocal of temperature, obtained using the kinetic model with experimental dataobtained in shock tube [33] for n-propanol and iso-propanol. The symbols represent experimental data [33] and the lines are predictions of the kinetic model.

A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16 9

strain rate a2 = 97.5 s�1, and the stoichiometric mixture fractionZst = 0.2449. There are a number of similarities between the flamestructure of these isomers. The measured profile of ethylene showsthat the concentration of this compound is higher in the n-propa-

×

0.000

0.002

0.004

0.006

0.008

0.011

0.013

Exit from fuel duct [cm]

Mol

e Fr

actio

n

0

0.05

0.1

0.15

0.2

0.25

0.2 0.3 0.4 0.5 0.6 0.7

0.2 0.3 0.4 0.5 0.6 0.7Exit from fuel duct [cm]

Mol

e Fr

actio

n

CO2

iso-propanol

CO

O2

H2O

×

CH4

C4H6 × 5H2

Fig. 7. Profiles of mole fraction of various species as a function of distance from thea2 = 97.5 s�1, and the stoichiometric mixture fraction Zst = 0.2449. The symbols represen

nol flame in comparison to its concentration in the iso-propanolflame. This large difference in ethylene peaks is properly predicted.As already mentioned, ethylene is only a secondary product in iso-propanol decomposition. This fact is well evident in Fig. 9, which

0

0.002

0.004

0.006

0.008

Exit from fuel duct [cm]

Mol

e Fr

actio

n

0

0.002

0.004

0.006

Exit from fuel duct [cm]

Mol

e Fr

actio

n

0.2 0.3 0.4 0.5 0.6 0.7

0.2 0.3 0.4 0.5 0.6 0.7

C2H2C2H4C2H6nC4H8

C3H4

C3H6

CH3CHOCH3COCH3

fuel boundary for non-premixed iso-propanol flames at a value of the strain ratet experimental data and the lines are model predictions.

×

×

0

0.05

0.1

0.15

0.2

0.25

Exit from Fuel Duct [mm]

Mol

e Fr

actio

n

CO2

n-propanol

CO

O2

H2O

×

0

0.004

0.008

0.012

0.016

0.02

Exit from Fuel Duct [mm]

Mol

e Fr

actio

n

×C2H5CHO

×

CH4

C4H6 × 20H2

0

0.002

0.004

0.006

0.008

Exit from Fuel Duct [mm]

Mol

e Fr

actio

nC3H8

C3H6

CH2OC3H4

CH3CHO

×

0

0.002

0.004

0.006

0.008

0.01

0.2 0.3 0.4 0.5 0.6 0.7

0.2 0.3 0.4 0.5 0.6 0.7 0.2 0.3 0.4 0.5 0.6 0.7

0.2 0.3 0.4 0.5 0.6 0.7Exit from Fuel Duct [mm]

Mol

e Fr

actio

n

×

C2H2C2H4C2H6nC4H8 × 2

Fig. 8. Profiles of mole fraction of various species as a function of distance from the fuel boundary for non-premixed n-propanol flames at a value of the strain ratea2 = 97.5 s�1, and the stoichiometric mixture fraction Zst = 0.2449. The symbols represent experimental data and the lines are model predictions.

10 A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16

shows the reaction flux analysis of the two flames. Ethylene forma-tion from iso-propanol is only due to the dehydrogenation of ethylradicals (formed via recombination of methyl radicals leading toethane and successive H abstractions) and to the de-methylationof n-propyl radical (formed via H addition reaction on propylene).On the contrary, primary ethylene formation from n-propanol isalso sustained by the b-decomposition reaction of the�CH2CH2CH2OH radical (see Fig. 3). In iso-propanol and n-propanolflames, ethylene consumption is mainly due to the H-abstraction

→

-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6

Rea

ctio

n ra

te [m

ol/l/

s]

Exit from Fuel duct [cm]

C2H4 Net rate of formation

C2H4+H+M=C2H5+M

C3H7=CH3+C2H4

C3H7O=C2H4+CH2OH

R+C2H4 →RH+C2H3

O+C2H4=CH3+HCO

n-propanol flame

Fig. 9. Reaction flux analysis of ethylene form

reactions to form vinyl radical in the fuel region and to the additionreaction of O radical in the flame front:

Oþ C2H4 ¡ CH3 þHCO

The kinetic model satisfactorily predicts the profile of acetone inn-propanol flame (Fig. 8) and propanal in iso-propanol flame(Fig. 7). Fig. 10 shows the structure of a non-premixed iso-propanolflame measured by Sinha and Thompson [29] at a strain ratea2 = 20 s�1, and the stoichiometric mixture fraction Zst = 0.5283.

→

0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6Exit from Fuel duct [cm]

C2H4 Net rate of formation

C2H4+H+M=C2H5+M

C3H7=CH3+C2H4

C2H6=C2H4+H2

R+C2H4 →RH+C2H3

O+C2H4=CH3+HCO

iso -propanol flame

ation in propanol flames of Figs. 7 and 8.

A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16 11

The symbols represent experimental data [29] and the lines aremodel predictions. Comparison of Figs. 7 and 10 shows the influ-ence of strain rate and stoichiometric mixture fraction on flamestructure. The agreement between the predicted and measured pro-files of oxygenated and C2 species is satisfactory. Fig. 7 shows thatthe predicted profile of propylene agrees with experimental data,while the predicted profile of propylene in Fig. 10 exceed the exper-imental data approximately by a factor of 2. These data are usefulboth in order to confirm the experimental measurements and toverify the possible systematic deviations between model predic-tions and experiments.

5.4. Low-pressure premixed flames of n-propanol and iso-propanol

Kasper et al. [32] carried out an experimental study of combus-tion of n-propanol and iso-propanol in low-pressure, premixed flatflames using two complementary molecular-beam mass spectrom-etry (MBMS) techniques: electron ionization (EI) and photon ioni-zation (PI). The equivalence ratio of the reactive mixture, U,considered was 1, 1.5 and 1.9.

Profiles of various species were calculated employing thekinetic model described here. Predicted profiles were shifted by1.5–2 mm to better match flame structure and maxima of interme-diate species. These values are within the experimental uncertain-ties for the measured temperature profiles. For all flames tested

0.00

0.02

0.04

0.06

0.08

0.10

Exit from fuel duct [cm]

Mol

e Fr

actio

n

COCO2

iso -propanol

0.0000

0.0010

0.0020

0.0030

0.0040

0.0050

0.0060

Exit from fuel duct [cm]

Mol

e Fr

actio

n

C3H6CH4

0.0 0.5 1.0 1.5 2.0

0.2 0.4 0.6 0.8 1.0 1.2

Fig. 10. Profiles of mole fraction of various species as a function of distance from thea2 = 20 s�1, and the stoichiometric mixture fraction Zst = 0.5283. The symbols representpredictions.

good agreement was obtained between model predictions andexperimental data for major species. As observed by Kasper et al.[32], the structures of n-propanol flame and iso-propanol flameare similar. Differences were mainly observed in the primary oxy-genated products as well as in ethylene that is a secondary productin the iso-propanol flames. Further detail of model comparisonswith the experimental measurements of both sets of n-propanoland iso-propanol flames at U = 1.0, 1.5 and 1.9 are available inthe Supplemental material.

Figs. 11 and 12 show comparison of predicted profiles of majorintermediate species with experimental data for U = 1.0, andU = 1.9. Fig. 11 shows comparison for iso-propanol flame andFig. 12 for n-propanol flame. Experimental measurements shownin these figures include those obtained by Kasper et al. [32] usingelectron EI and PI techniques. Figs. 13 and 14 show similar compar-ison of methyl (CH3) and propargyl (C3H3) radicals, for both thefuels. Benzene production in the rich flames is well predicted bythe model and is similar for both the fuels, although it is possibleto observe a larger formation from iso-propanol flame.

Fig. 15 reports the sensitivity analysis to benzene formation.Benzene is mainly formed from propargyl radicals in both thecases, thus all the reactions favoring propylene, allene as well asallyl and propargyl radicals formation show a positive sensitivitycoefficient. On the contrary, all the propagation reactions favorbenzene depletion. It is also noteworthy that benzene peak is high-

0.0000

0.0002

0.0004

0.0006

0.0008

0.0010

0.0012

0.0014

Exit from fuel duct [cm]

Mol

e Fr

actio

n

C3H4CH3CHOCH3COCH3

0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.2 0.4 0.6 0.8 1.0 1.2

0.2 0.4 0.6 0.8 1.0 1.2Exit from fuel duct [cm]

Mol

e Fr

actio

n

C2H2C2H4C2H6

fuel boundary for non-premixed iso-propanol flames at a value of the strain rateexperimental data obtained by Sinha and Thompson [29] and the lines are model

Fig. 11. Profiles of mole fraction of various intermediate species as a function of distance from the fuel duct. The fuel tested is iso-propanol. The equivalence ratio is U = 1, andU = 1.9. The symbols represent experimental data measured by Kasper et al. [32] using electron ionization (EI) and photon ionization (PI) molecular-beam mass spectrometry.The curves are model predictions.

12 A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16

er and it occurs later in the iso-propanol flame, due to the lowerreactivity and the larger presence of methyl radicals.

Studies on premixed laminar flames of iso-propanol and n-pro-panol at low pressure were recently carried out by Li et al. [31]. Thecomplete simulation results are reported in the Supplementalmaterial. Table 2 shows a comparison between measured and pre-dicted peak positions and corresponding maximum mole fractionsin both the rich flames (U = 1.8). Predicted profiles are shifted ofabout 4–5 mm to match measured flame structure and maximaof intermediate species. It is noteworthy to observe again thatthere exists strong similarity between the two flames. Moreoveris of interest to observe the similarity between these flames andthe corresponding ones studied by Kasper et al. [32].

Table 2 shows that although the predicted peak values and posi-tions of several species, including acetone and propanal, agree wellwith the experimental measurements, large discrepancies are alsoevident for some species and they require some comments. Theavailability of two other sets of independent experimental mea-surements, resulting from the work of Kasper et al. [32] and Liet al. [31], allows to better evaluating the meaning and importanceof these deviations. The following summarizes some observationsof the comparisons shown in Table 2:

Acetylene: The observed over-prediction of acetylene for the iso-propanol flame is not consistent with the slight under-predictionreported in Fig. 12.

Vinyl radical: This species is largely over-predicted for bothfuels. These experimental measurements do not agree with thecorresponding measurements, both with EI and PI techniques ofKasper [59].

Formyl radical: This species is largely over-predicted for bothfuels. These experimental measurements do not agree with thecorresponding measurements with PI technique of Kasper [59].Ethyl radical, with atomic mass 29, is �1 � 10�5 is completely cov-ered by formyl radical.

Allyl radical: This species is again 10 times over-predicted forboth fuels. The measurements of Kasper [59] employing the PItechnique indicate that for the iso-propanol flame the measuredvalues of allyl radical are closer to model predictions.

Propylene: This species is over-predicted for both fuels. Compar-isons with the experimental data of Kasper et al. [32] do not con-firm this deviation.

1-Butene and butadiene: These species are under-predicted by afactor 2 in n-propanol flame, while a closer agreement is observedin iso-propanol flame.

Diacetylene and vinylacetylene: These species are largely under-predicted in all the flames and C4H5 radical shows the samedeviation.

Benzene: The predicted peak value of this species is �one half ofthe measured value in n-propanol flame, while the predictions aredouble of the measured value in the iso-propanol flame. The pre-dicted values are consistent with those measured by Kasper et al.[32] employing the PI technique, but both are under-predictedwhen compared with those measured employing the EI technique.

Pentadiene and cyclopentadiene: Predicted values of these spe-cies show satisfactory agreement with experimental data.

The availability of this double set of very accurate experimen-tal data is of course of extreme value for the kinetic modelingactivity. In addition, model results could be of potential interest

Fig. 12. Profiles of mole fraction of various intermediate species as a function of distance from the fuel duct. The fuel tested is n-propanol. The equivalence ratio is U = 1, andU = 1.9. The symbols represent experimental data measured by Kasper et al. [32] using electron ionization (EI) and photon ionization (PI) molecular-beam mass spectrometry.The curves are model predictions.

Fig. 13. Profiles of mole fraction of methyl (CH3) radical and propargyl (C3H3) radical as a function of distance from the fuel duct. The fuel tested is iso-propanol. Theequivalence ratio is U = 1, and U = 1.9. The symbols represent experimental data measured by Kasper et al. [32] using electron ionization (EI) and photon ionization (PI)molecular-beam mass spectrometry. The curves are model predictions.

Fig. 14. Profiles of mole fraction of methyl (CH3) radical and propargyl (C3H3) radical as a function of distance from the fuel duct. The fuel tested is n-propanol. Theequivalence ratio is U = 1, and U = 1.9. The symbols represent experimental data measured by Kasper et al. [32] using electron ionization (EI) and photon ionization (PI)molecular-beam mass spectrometry. The curves are model predictions.

A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16 13

nC3H7OH = C3H6 + H2OH + O2 = OH + O

H + CH2CHCH2 = H2 + aC3H4R + C3H6 = RH + CH2CHCH2

C3H3 + OH = C2H3 + HCOHCO + M = CO + H + M

C3H3 + C3H3 = C6H5 + H

nC3H7OH = C2H5 + CH2OHC3H3 + O = CH2O + C2HaC3H4 → pC3H4

R + C6H6 = RH + C6H5aC3H4 + H = C3H3 + H2

HCO + H = H2 + COHCO + CH3 → CO + CH4

O2 + C6H5 = C6H5O + OC6H6 = C6H5 + H

C3H3 + H = C3H2 + H2

C3H3 + C3H3 + M = C6H6 + M

n-propanol

iC3H7OH = C3H6 + H2O

H + O2 = OH + OH + CH2CHCH2 = H2 + aC3H4

R + C3H6 = RH + CH2CHCH2

C3H3 + OH = C2H3 + HCOHCO + M = CO + H + M

C3H3 + C3H3 = C6H5 + H

iC3H7OH = CH3 + CH3CHOHC3H3 + O = CH2O + C2H

aC3H4 → pC3H4R + C6H6 = RH + C6H5

aC3H4 + H = C3H3 + H2

O2 + C6H5 = C6H5O + O

C6H6 = C6H5 + H

C3H3 + H = C3H2 + H2

C3H3 + C3H3 + M = C6H6 + M

CH3 + CH3 = H + C2H5

O + CH3 = CH2O + H

iso-propanol

nC3H7OH = C3H6 + H2OH + O2 = OH + O

H + CH2CHCH2 = H2 + aC3H4R + C3H6 = RH + CH2CHCH2

C3H3 + OH = C2H3 + HCOHCO + M = CO + H + M

C3H3 + C3H3 = C6H5 + H

nC3H7OH = C2H5 + CH2OHC3H3 + O = CH2O + C2HaC3H4 → pC3H4

R + C6H6 = RH + C6H5aC3H4 + H = C3H3 + H2

HCO + H = H2 + COHCO + CH3 → CO + CH4

O2 + C6H5 = C6H5O + OC6H6 = C6H5 + H

C3H3 + H = C3H2 + H2

C3H3 + C3H3 + M = C6H6 + M

n-propanol

nC3H7OH = C3H6 + H2OH + O2 = OH + O

H + CH2CHCH2 = H2 + aC3H4R + C3H6 = RH + CH2CHCH2

C3H3 + OH = C2H3 + HCOHCO + M = CO + H + M

C3H3 + C3H3 = C6H5 + H

nC3H7OH = C2H5 + CH2OHC3H3 + O = CH2O + C2HaC3H4 → pC3H4

R + C6H6 = RH + C6H5aC3H4 + H = C3H3 + H2

HCO + H = H2 + COHCO + CH3 → CO + CH4

O2 + C6H5 = C6H5O + OC6H6 = C6H5 + H

C3H3 + H = C3H2 + H2

C3H3 + C3H3 + M = C6H6 + M

n-propanol

iC3H7OH = C3H6 + H2O

H + O2 = OH + OH + CH2CHCH2 = H2 + aC3H4

R + C3H6 = RH + CH2CHCH2

C3H3 + OH = C2H3 + HCOHCO + M = CO + H + M

C3H3 + C3H3 = C6H5 + H

iC3H7OH = CH3 + CH3CHOHC3H3 + O = CH2O + C2H

aC3H4 → pC3H4R + C6H6 = RH + C6H5

aC3H4 + H = C3H3 + H2

O2 + C6H5 = C6H5O + O

C6H6 = C6H5 + H

C3H3 + H = C3H2 + H2

C3H3 + C3H3 + M = C6H6 + M

CH3 + CH3 = H + C2H5

O + CH3 = CH2O + H

iso-propanol

iC3H7OH = C3H6 + H2O

H + O2 = OH + OH + CH2CHCH2 = H2 + aC3H4

R + C3H6 = RH + CH2CHCH2

C3H3 + OH = C2H3 + HCOHCO + M = CO + H + M

C3H3 + C3H3 = C6H5 + H

iC3H7OH = CH3 + CH3CHOHC3H3 + O = CH2O + C2H

aC3H4 → pC3H4R + C6H6 = RH + C6H5

aC3H4 + H = C3H3 + H2

O2 + C6H5 = C6H5O + O

C6H6 = C6H5 + H

C3H3 + H = C3H2 + H2

C3H3 + C3H3 + M = C6H6 + M

CH3 + CH3 = H + C2H5

O + CH3 = CH2O + H

iso-propanol

Fig. 15. Sensitivity coefficients for benzene formation in n-propanol and iso-propanol flames [32] a U = 1.9. Sensitivity coefficients refer to the axial location corresponding tothe benzene maxima in the flames. Dark bars correspond to positive sensitivity coefficients.

14 A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16

to further improve the experimental accuracy and/or experimentalmethods.

6. Summary and conclusions

A kinetic mechanism that describes the primary reactions ofcombustion of n-propanol and iso-propanol is developed. This pri-mary mechanism is appended to a previously developed detailedscheme of pyrolysis and oxidation of hydrocarbon fuels. The maineffect arising from the presence of hydroxyl group in alcohols wason the a position in the H-abstraction reactions and radical decom-

Table 2Comparison of predicted peak position and maximum mole fractions (Xmax) of the rich flamthose of Li et. al. [31].

Formula Species Rich n-propanol flame

Position (cm) Xmax

Pred.a Exp Pred.

CH3 Methyl radical 0.76 0.85 6.1 � 10�03

C2H2 Acetylene 0.87 0.85 1.9 � 10�02

C2H3 Vinyl radical 0.78 0.85 1.6 � 10�04

C2H4 Ethylene 0.72 0.75 1.4 � 10�02

HCO Formyl radical 0.8 0.7 3.2 � 10�04

H2CO Formaldehyde 0.69 0.7 6.7 � 10�03

CH3OH Methanol 0.59 0.5 1.1 � 10�03

C3H3 Propargyl radical 0.81 0.85 5.1 � 10�04

C3H4 Propyne 0.82 0.8 4.1 � 10�04

C3H4 Allene 0.77 0.75 3.6 � 10�04

C3H5 Allyl radical 0.82 0.8 1.4 � 10�03

C2H2O Ketene 0.72 0.7 2.4 � 10�03

C3H6 Propylene 0.72 0.7 7.5 � 10�03

C2H4O Acetaldehyde 0.63 0.65 4.5 � 10�03

C4H2 Diacetylene 0.92 0.95 8.0 � 10�05

C4H4 Vinylacetylene 0.77 0.85 1.8 � 10�05

C4H5 But-2-yn-1-yl radical 0.63 0.7 4.2 � 10�07

C4H6 1,3-Butadiene 0.72 0.75 4.3 � 10�05

C4H8 1-Butene 0.67 0.7 6.2 � 10�05

C3H6O Propanal 0.64 0.55 2.9 � 10�03

C3H6O Acetone 0.97 nd 2.2 � 10�05

C5H6 1,3-Cyclopentadiene 0.74 0.8 1.1 � 10�05

C5H8 1,3-Pentadiene 0.76 0.7 4.6 � 10�06

C6H6 Benzene 0.74 0.7 7.6 � 10�06

a Predicted values are shifted of 0.52 cm.b Predicted values are shifted of 0.44 cm.

positions. Employing a systematic approach for the metathesisreactions and some lumping procedure for fast intermediate radi-cals, a small subset of new primary reactions were developed fordescribing combustion of n-propanol and iso-propanol. Four-cen-ter molecular dehydration reaction to form propylene is mainlyrelevant for iso-propanol. New experimental data on counterflownon-premixed flames were obtained for mechanism validation.Flame structures were measured under similar conditions for bothfuels to elucidate the similarities and differences in combustioncharacteristics of the two isomers. The profiles measured includethose of formaldehyde, acetaldehyde, propanal, and acetone. The

es of n-propanol and iso-propanol with experimental data. The experimental data are

Rich iso-propanol flame

Position (cm) Xmax

Exp Pred.b Exp Pred. Exp

6.8 � 10�03 0.84 0.85 8.1 � 10�03 4.9 � 10�03

1.8 � 10�02 0.98 0.9 1.9 � 10�02 5.6 � 10�03

3.0 � 10�05 0.91 0.95 1.1 � 10�04 1.1 � 10�05

1.9 � 10�02 0.84 0.8 1.0 � 10�02 5.7 � 10�03

4.1 � 10�05 0.91 0.85 2.5 � 10�04 2.0 � 10�05

6.9 � 10�03 0.82 0.65 5.0 � 10�03 2.8 � 10�03

8.1 � 10�04 0.57 0.5 6.2 � 10�04 2.9 � 10�04

3.3 � 10�04 0.91 0.95 7.3 � 10�04 3.4 � 10�04

5.6 � 10�04 0.89 0.85 6.5 � 10�04 4.6 � 10�04

2.5 � 10�04 0.85 0.85 6.0 � 10�04 1.8 � 10�04

1.4 � 10�04 0.89 0.9 1.8 � 10�03 1.4 � 10�04

1.2 � 10�03 0.77 0.8 3.8 � 10�03 3.9 � 10�03

2.5 � 10�03 0.77 0.8 1.4 � 10�02 3.9 � 10�03

5.2 � 10�03 0.74 0.65 3.8 � 10�03 1.1 � 10�03

4.7 � 10�04 1 1 9.8 � 10�05 2.4 � 10�04

1.3 � 10�04 0.87 0.9 3.9 � 10�05 1.6 � 10�04

4.9 � 10�06 0.6 0.95 3.3 � 10�07 9.9 � 10�06

9.1 � 10�05 0.79 0.85 1.2 � 10�04 1.4 � 10�04

1.2 � 10�04 0.72 0.75 2.2 � 10�04 1.7 � 10�04

2.2 � 10�03 0.64 nd 1.0 � 10�05 ndnd 0.67 0.75 6.5 � 10�03 5.3 � 10�03

1.5 � 10�05 0.82 0.85 3.3 � 10�05 2.8 � 10�05

6.0 � 10�06 0.84 0.75 8.5 � 10�06 9.4 � 10�06

1.4 � 10�05 0.86 0.85 2.3 � 10�05 1.3 � 10�05

A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16 15

validation of the kinetic model was extended to different sets ofexperimental data obtained by other investigators in very differentoperative conditions and reaction configurations. The agreementbetween the kinetic model and experimental data was generallysatisfactory, in terms of reactivity and selectivity in major productsand minor species. The flame structures and overall combustioncharacteristics of n-propanol and iso-propanol are found to be sim-ilar. Modeling shows that ethylene is only a secondary product ofiso-propanol combustion and its amount is significantly lower.Acetone is formed under all conditions in iso-propanol flames,while propanal is formed in n-propanol flames. The kinetic modeldeveloped here is expected to be the starting point for extension tohigher alcohols.

Acknowledgments

The authors acknowledge the valuable discussions with Profes-sor M. Dente and the useful work of R. Grana. The Authors furtheracknowledge Professor M. J. Thompson, T. Kasper and H. Curran forproviding detailed information and further insight on their exper-imental data. The authors appreciate the use of the shock tubeexperimental data of Fig. 6 by Johnson et al. prior to their publica-tion. The research at the University of California at San Diego issupported by UC Discovery/Westbiofuels, Grant # GCP06-10228.The research activity of Politecnico was supported by the EuropeanCommunity in the behalf of the DREAM Project (FP7-211861) andby CNR-DET (MAP/MSE Biofuel Program).

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.combustflame.2009.09.002.

References

[1] S.R. Smith, A.S. Gordon, Studies of diffusion flames. II. Diffusion flames of somesimple alcohols, Journal of Physical Chemistry 60 (8) (1956) 1059–1062.

[2] K. Seshadri, Structure and extinction of laminar diffusion flames abovecondensed fuels with water and nitrogen, Combustion and Flame 33 (1978)197–215.

[3] A. Hamins, K. Seshadri, The influence of alcohols on the combustion ofhydrocarbon fuels in diffusion flames, Combustion and Flame 64 (1986) 43–54.

[4] A. Hamins, K. Seshadri, The structure of diffusion flames burning pure, binary,and ternary solutions of methanol, heptane, and toluene, Combustion andFlame 68 (1987) 295–307.

[5] S.C. Li, F.A, Williams, Experimental and numerical studies of two-stagemethanol flames, in: Proceedings of the Combustion Institute, vol. 26, 1996,pp. 1017–1024.

[6] T. Held, F. Dryer, A comprehensive mechanism for methanol oxidation,International Journal of Chemical Kinetics 118 (1998) 805–830.

[7] O.L.Gülder, Laminar burning velocities of methanol, ethanol and isooctane–airmixtures, in: Proceedings of the Combustion Institute, 1982, pp. 275–281.

[8] N. Marinov, A detailed kinetic model for high temperature ethanol oxidation,International Journal of Chemical Kinetics 31 (1999) 183–220.

[9] R. Seiser, S. Humer, K. Seshadri, E. Pucher, Experimental investigation ofmethanol and ethanol flames in nonuniform flows, in: Proceedings of theCombustion Institute, vol. 31, 2007, pp. 1173–1180.

[10] T.S. Norton, F.L. Dryer, The flow reactor oxidation of C1–C4 alcohols and MTBE,in: Proceedings of the Combustion Institute, vol. 23, 1990, pp. 179–185.

[11] F. N. Egolfopoulos, D. X. Lu, and C. K. Law, A study on ethanol oxidation kineticsin laminar premixed flames, flow reactors and shock tubes, in: Proceedings ofthe Combustion Institute, vol. 24, 1992, pp. 833–841.

[12] W. Tsang, Energy transfer effects during the multichannel decomposition ofethanol, International Journal of Chemical Kinetics 36 (8) (2004) 456–465.

[13] P. Saxena, F.A. Williams, Numerical and experimental studies of ethanolflames, in: Proceedings of the Combustion Institute, vol. 31, 2007, pp. 1149–1156.

[14] P. Dagaut, C. Togbé, Experimental and modeling study of the kinetics ofoxidation of ethanol–gasoline surrogate mixtures (E85 surrogate) in a jet-stirred reactor, Energy & Fuels 22 (2008) 3499–3505.

[15] M. Abián, C. Esarte, A. Millera, R. Bilbao, M.U. Alzueta, Oxidation of acetylene–ethanol mixtures and their interaction with NO, Energy & Fuels 22 (2008)3814–3823.

[16] A. Frassoldati, A. Cuoci, T. Faravelli, E. Ranzi, Kinetic modeling of the oxidationof ethanol and gasoline surrogate mixtures, Combustion Science andTechnology, 2009, in press.

[17] C. McEnally, L. D. Pfefferle, Fuel decomposition and hydrocarbon growthprocesses for oxygenated hydrocarbons: butyl alcohols, in: Proceedings of theCombustion Institute, vol. 30, 2005, pp. 1363–1370.

[18] B. Yang, P. Oßwald, Y. Li, J. Wang, L. Wei, Z. Tian, F. Qi, K. Kohse-Höinghaus,Identification of combustion intermediates in isomeric fuel-rich premixedbutanol–oxygen flames at low pressure, Combustion and Flame 148 (2007)198–209.

[19] P. Dagaut, C. Togbé, Oxidation kinetics of butanol–gasoline surrogate mixturein a jet-stirred reactor: experimental and modeling study, Fuel 87 (2008)3313–3321.

[20] J.T. Moss, A.M. Berkowitz, M.A. Oehlschlaeger, J. Biet, A. Warth, P.A. Glaude, F.Battin-Leclerc, An experimental and kinetic modelling study of the oxidationof four isomers of butanol, Journal of Physical Chemistry 112 (2008) 1084–10855.

[21] J. Moc, J.M. Simmie, H.J. Curran, The elimination of water from aconformationally complex alcohol: a computational study of the gas phasedehydration of n-butanol, Journal of Molecular Structure 928 (2009) 149–157.

[22] S.M. Sarathy, M.J. Thomson, C. Togbé, P. Dagaut, F. Halter, C. Mountaim-Rousselle, An experimental and kinetic modeling study of n-butanolcombustion, Combustion and Flame 156 (2009) 852–864.

[23] R. Wang, P. Cadman, Soot and pah production from spray combustion ofdifferent hydrocarbons behind reflected shock waves, Combustion and Flame112 (1998) 359–370.

[24] J.A. Barnard, H.W.D. Hughes, The pyrolysis of n-propanol, Transactions of theFaraday Society 56 (1) (1960) 64–71.

[25] J.A. Barnard, The pyrolysis of iso-propanol, Transactions of the Faraday Society56 (1) (1960) 72–79.

[26] A.B. Trenwith, Thermal decomposition of isopropanol, Journal of the ChemicalSociety, Faraday Transactions 1: Physical Chemistry in Condensed Phases 71(1975) 2405–2412.

[27] B.H. Bui, R.S. Zhu, M.C. Lin, Thermal decomposition of iso-propanol: first-principles prediction of total and product-branching rate constants, Journal ofChemical Physics 117 (24) (2002) 11188–11195.

[28] J. Cai, F. Li, W.A. Sirignano, Three-dimensional structures of flames over liquidfuel pools, Combustion Science and Technology 175 (2003) 2113–2139.

[29] A. Sinha, M.J. Thomson, The chemical structures of opposed flow diffusionflames of C3 oxygenated hydrocarbons (iso-propanol, dimethoxymethane, anddimethyl carbonate) and their mixtures, Combustion and Flame 136 (2004)548–556.

[30] B.D. Shaw, J.B. Wei, Propanol droplet flammability and combustion in air–diluent environments under normal and reduced gravity, Combustion Scienceand Technology 179 (2007) 1205–1223.

[31] Y. Li, L. Wei, Z. Tian, B. Yang, J. Wang, T. Zhang, F. Qi, A comprehensiveexperimental study of low-pressure premixed C3-oxygenated hydrocarbonflames with tunable synchrotron photoionization, Combustion and Flame 152(2008) 336–359.

[32] T. Kasper, P. Oßwald, U. Struckmeier, K. Kohse-Höinghaus, C.A. Taatjes, J.Wang, T.A. Cool, M.E. Law, A. Morel, P.R. Westmorelend, Combustionchemistry of propanol isomers—investigated by electron ionization andVUV-photoionization molecular-beam mass spectrometry, Combustion andFlame 156 (2009) 1181–1201.

[33] M.V. Johnson, S.S. Goldsborough, E. Larkin, G. OMalley, Z. Serinyel, P. OToole,H.J. Curran, Ignition delay measurements of n-propanol and iso-propanol in ashock tube, in: Proceedings of the 6th US National Combustion Meeting, Paper# 22C5, The University of Michigan, Ann Arbor, Michigan, USA, May 17–20,2009, The Combustion Institute, 2009.

[34] M.Z. Jacobson, Effects of ethanol (E85) versus gasoline vehicles on cancer andmortality in the United States, Environmental Science and Technology 41 (11)(2007) 4150–4157.

[35] E. Zervas, X. Montagne, J. Lahaye, Emission of alcohols and carbonylcompounds from spark ignition engine influence of fuel and air/fuelequivalence ratio, Environmental Science and Technology 36 (2002) 2414–2421.

[36] F.A. Williams, Combustion Theory, second ed., Addison-Wesley PublishingCompany, Redwood City, CA, 1985.

[37] A. Liñán, F.A. Williams, Fundamental Aspects of Combustion, volume 34 ofOxford Engineering Science Series, Oxford University Press, New York, 1993.

[38] N. Peters, Local quenching due to flame stretch and non-premixed turbulentcombustion, Combustion Science and Technology 30 (1983).

[39] N. Peters, Turbulent Combustion, Cambridge University Press, Cambridge,England, 2000.

[40] K. Seshadri, F.A. Williams, Laminar flow between parallel plates with injectionof a reactant at high Reynolds number, International Journal of Heat and MassTransfer 21 (2) (1978) 251–253.

[41] A. Galano, J.R. Alvarez-Idaboy, G. Bravo-Perez, M.E. Ruiz-Santoyo, Gas phasereactions of C1–C4 alcohols with the OH radical: a quantum mechanicalapproach, Physical Chemistry Chemical Physics Journal 4 (2002) 4648–4662.

[42] R. Atkinson, D.L. Baulch, R.A. Cox, R.F. Hampson Jr., J.A. Kerr, M.J. Rossi, J. Troe,Evaluated kinetic, photochemical and heterogeneous data for atmosphericchemistry: supplement V. IUPAC subcommittee on gas kinetic data evaluationfor atmospheric chemistry, Journal of Physical and Chemical Reference Data 26(1997) 521–1011.

16 A. Frassoldati et al. / Combustion and Flame 157 (2010) 2–16

[43] R. Atkinson, D.L. Baulch, R.A. Cox, R.F. Hampson Jr., J.A. Kerr, M.J. Rossi, J. Troe,Evaluated kinetic and photochemical data for atmospheric chemistry, organicspecies: supplement VII, Journal of Physical and Chemical Reference Data 28(1999) 191–394.

[44] J.R. Dunlop, F.P. Tully, Catalytic dehydration of alcohols by hydroxyl: 2-propanol;an intermediate case, Journal of Physical Chemistry B 97 (1993) 6457–6464.

[45] E. Ranzi, M. Dente, T. Faravelli, G. Pennati, Prediction of kinetic parameters forhydrogen abstraction reactions, Combustion Science and Technology 95(1983) 1–50.

[46] E. Ranzi, T. Faravelli, P. Gaffuri, G.C. Pennati, A. Sogaro, A wide range modelingstudy of propane and n-butane oxidation, Combustion Science and Technology100 (1994) 299–330.

[47] E. Ranzi, M. Dente, A. Goldaniga, G. Bozzano, T. Faravelli, Lumping proceduresin detailed kinetic modeling of gasification, pyrolysis, partial oxidation andcombustion of hydrocarbon mixtures, Progress in Energy and CombustionScience 27 (2001) 99–139.

[48] H.J. Curran, Rate constant estimation for C1 to C4 alkyl and alkoxyl radicaldecomposition, International Journal of Chemical Kinetics 38 (4) (2006) 250–275.

[49] C.G. Newman, H.E. O’Neal, M.A. Ring, F. Leska, N. Shipley, Kinetics andmechanism of the silane decomposition, International Journal of ChemicalKinetics 11 (1979) 1167–1182.

[50] E. Ranzi, T. Faravelli, A. Frassoldati, S. Granata, Wide range kinetic modelingstudy of pyrolysis and combustion of heavy n-alkanes, Industrial andEngineering Chemistry Research 44 (14) (2005) 5170–5183.

[51] R.J. Kee, F. Rupley, J.A. Miller, The Chemkin thermodynamic data base, Reportsand86-8215b, Sandia National Laboratories, Livermore, California,1987.

[52] R.J. Kee, F. Rupley, J.A. Miller, The Chemkin thermodynamic data base,Technical report, Sandia National Laboratories, Livermore, California, 1989.

[53] S.W. Benson, Thermochemical Kinetics, second ed., Wiley, New York, 1996.[54] <http://www.chem.polimi.it/CRECKModeling>.[55] D. Manca, G. Buzzi-Ferraris, T. Faravelli, E. Ranzi, Numerical problems in the

solution of oxidation and combustion models, Combustion Theory andModelling 5 (2) (2001) 185–199.

[56] A. Cuoci, A. Frassoldati, T. Faravelli, E. Ranzi, Frequency response of counterflow diffusion flames to strain rate harmonic oscillations, Combustion Scienceand Technology 180 (2008) 767–784.

[57] G. Buzzi-Ferraris, Building Numerical Libraries the Object-Oriented Way,Addison Wesley/Longman, 1993.

[58] G. Buzzi-Ferraris, Numerical libraries in c++. Politecnico di Milano – Freelyavailable at the web address: <http://www.chem.polimi.it/homes/gbuzzi>.

[59] T. Kasper, Personal communication, 2009.