Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Combustion and Flame 218 (2020) 189–204

Contents lists available at ScienceDirect

Combustion and Flame


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

Combustion of propanol isomers: Experimental and kinetic modeling


study
Gianluca Capriolo∗, Alexander A. Konnov
Division of Combustion Physics, Department of Physics, Lund University, P.O. Box 118, SE-221 00 Lund, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: In this work an experimental and kinetic modeling study on n-propanol and i-propanol combustion has
Received 5 October 2019 been performed. New burning velocity measurements were carried out using the heat flux method at 1
Revised 12 May 2020
atm over the temperature range of 323–393 K. Analysis of the temperature dependence was conducted
Accepted 14 May 2020
with to verify the data consistency of the new and available data from the literature. Important inconsis-
tencies were identified with the literature experiments performed using the spherical flame method and
Keywords: the nature of such inconsistencies was discussed. Moreover, a new kinetic mechanism, based on the most
Laminar burning velocity recent Konnov model and extended to include C3 alcohol isomers chemistry subset, was validated against
n-propanol new and all available literature data obtained at different combustion regimes. Rate constant parameters
i-propanol
were carefully selected by evaluating all experimental and theoretical sources. Moreover, Sarathy et al.
Modeling
(2014) detailed kinetic mechanism was also tested. Overall, both kinetic models reproduce experimental
data with good fidelity, but the presented model was found superior in representing ignition delay times
data performed at high-pressure conditions.
© 2020 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

1. Introduction propanol and i-propanol the fastest reaction rates for primary rad-
ical decomposition are attributable to the Cα -H bond cleavage. The
The depletion of fossil fuels, combined with the contemporary following studies on propyl alcohols were mainly focused on py-
awareness that their use has a detrimental impact on climate and rolysis [4–6], leading to different conclusions. Barnard and Hughes
air pollution, has led to a compulsive search for more sustainable [4], and Barnard [5], ascertained that the chain-initiating step for
transportation fuels. In this regard, small aliphatic alcohols can be n-propanol and i-propanol pyrolysis occurred via the scission of
valuable alternative fuels/fuel additives in SI engines, contributing Cα –Cβ and Cα –H bonds respectively. On the contrary Trenwith [6],
to improve engine combustion efficiency and performance as their investigating i-propanol pyrolysis with a static reactor at T = 721–
use increases antiknock index and burning velocities, if compared 801 K and P = 10–100 Torr came to a different conclusion, namely
to traditional gasoline. To this end, a recent review on alcohol com- that the reaction was initiated by scission of Cα –Cβ bond.
bustion chemistry performed by Sarathy et al. [1] provides a valu- Norton and Dryer [7] found out that the reactivity of n-
able explanation of C1 –C5 alcohols combustion characteristics and propanol was much higher than that of its isomer. The authors
why such hydroxylated compounds are suitable for internal com- [7] concluded that the difference between the two isomers was
bustion engine (ICE) applications. Alcohol fuels can be manufac- strictly related to the diverse reactivity of the major intermediate
tured sustainably, i.e., through fermentation of biomass feedstocks, species derived from their oxidations, i.e., propanal and acetone,
waste, and also from new alternative routes, such as metabolically in turn, related to the higher propensity of n-propanal to dehy-
engineered Escherichia Coli [2], which potentially would enable a drogenate than dehydrate. Assessment of the theoretical work of
high yield and volumetric productivity. Bui et al. [28] on the dehydration reaction rate of isopropanol was
Pyrolysis and oxidation kinetics of propanols has been exten- made by Heyne et al. [8]. The experimental results revealed that
sively studied in the literature, including static [3–6], flow [7–9], the dehydration reaction proceeds faster than the one predicted in
and well stirred reactors [10,11], as well as shock tubes 12–17], the study of Bui et al. [28].
burners [18–24] and laminar burning velocity (LBV) measurement Most recently, an experimental and kinetic work on C3 alcohols
apparatus 9–11,25–27]. Smith and Gordon [3] reported that, for n- was performed by Li et al. [9]. Results showed that the major oxy-
genated intermediates in n-propanol pyrolysis are C1 –C2 aldehy-
des and ethenol while, for i-propanol, C2 aldehydes, and ketones.

Corresponding author. Model analysis unveiled that n-propanol mainly reacts via the H-
E-mail address: gianluca.capriolo@forbrf.lth.se (G. Capriolo).

https://doi.org/10.1016/j.combustflame.2020.05.012
0010-2180/© 2020 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
190 G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204

abstraction and the Cα –Cβ bond dissociation reactions, while the A new kinetic model for propyl alcohols oxidation and pyrolysis
H2 O elimination and also the H abstraction reactions controlled the was also presented and assessed against new and literature ex-
i-propanol consumption. perimental data [7,13,18,20] and it showed an overall satisfactory
Combustion characteristics of n-propanol and i-propanol were agreement. An experimental and numerical study on extinction
experimentally studied, respectively, by Galmiche et al. [10] and and autoignition of C3 –C4 primary and secondary alcohols were
Togbé et al. [11] in a JSR and a spherical combustion chamber. Be- performed by Alfazi et al. [23]. The experimental results showed
sides, Galmiche et al. [10] also proposed a detailed kinetic mecha- that n-propanol had a higher propensity to auto-ignite and the
nism for the n-propanol oxidation, which showed good agreement lower to extinguish; on the contrary, i-propanol exhibits the lowest
with their experimental results, but significant inconsistencies with reactivity and extinction limits. Such behavior was attributed, sim-
the LBV data from [24]. In the study of Togbé et al. [11], the per- ilarly to previous studies, to the less reactive intermediates formed
formed experiments were compared with numerical simulations during the oxidation, such as acetone. Veloo and Egolfopoulos
using the mechanism by Johnson et al. [13]. The model showed [22] measured the LBV and the extinction strain rate of the two
poor performance in reproducing presented experimental data. Fi- propanol isomers at 343 K and at 1 atm and compared the results
nally, both studies stated that the effect of pressure on LBVs was with the prediction of the kinetic mechanism of Johnson et al. [13].
coherent with previous literature works on C1 –C6 alcohols. Numerical simulations were not able to reproduce experimental
Wang and Cadman [12] focused their study on soot and PAH data consistently. Authors [22] also attributed the higher reactiv-
formation during pyrolysis and oxidation of different hydrocar- ity of n-propanol to the differences in intermediate species’ forma-
bons, including n-propanol. It was found that, among the consid- tion between the two structural isomers. High-pressure measure-
ered fuels, n-propanol had a lesser sooting tendency. Johnson et ments of C1 –C4 primary alcohols LBV were performed by Beeck-
al. [13] investigated the ignition properties of propyl alcohols be- man et al. [25] and their experimental data were used to assess
tween 1350 and 20 0 0 K. A kinetic mechanism, based on Bourque the predictions of different models available from the literature.
et al. [29] model and extended to include C3 alcohol isomers reac- Specifically for n-propanol, the accuracy of the model of Johnson
tion subset, was also proposed. The authors highlighted the impor- et al. [13] was tested; numerical simulations failed in reproduc-
tant role of low hydrocarbons and hydrogen chemistry on ignition ing experimental results resulting in an important underestimation
delay time (IDT) predictions. High-temperature ignition behavior of the data over the whole investigated equivalence ratio range.
of i-propanol was also investigated by Akih-Kumgeh and Bergth- The authors [25] emphasized the need for additional studies of
orson [14]. Notable discrepancies were found when the model of the fuel-specific reactions that are pressure dependent. New LBV
Johnson et al. [13] was assessed against the experimental results. measurements of several C3 oxygenated were provided by Gong et
Noorani et al. [15] studied the ignition characteristics of C1 –C4 al. [26] A new kinetic mechanism, based on an updated version of
primary alcohols. The experimental results were compared with Aramco mech. 1.3 [30] and extended to include then-propanol sub-
different models from the literature. Particularly for n-propanol, mechanism from [13], was also tested against the experimental re-
the IDTs data were compared with Johnson et al. [13] mechanism. sults, resulting in a general overestimation of the experimental re-
Model prediction failed in reproducing experiments, resulting in a sults for propanols+air mixtures. Li et al. [27] measured the LBVs
general underestimation of the n-propanol reactivity. Lately, Man of C1 –C5 primary alcohol-isooctane blends and noted a monotonic
et al. [16] studied the IDTs of C3 alcohol isomers and also pro- relationship between LBV and the alcohol mole fraction present
posed a new kinetic model based on a modified version of John- into the fuel mixture and an empirical relationship was suggested.
son et al. [13] which was validated against new and literature data Katoch et al. [24] measured LBV at different temperatures and the
[10,11,15,22]. Modeling results showed good agreement in simulat- performance of six different models was also tested with the pre-
ing the IDTs and the JSR data, while tended to sensibly overpredict sented and the literature LBV measurements. Notable inconsisten-
the LBVs in rich mixtures. The model also showed that propanols cies among proposed and literature experimental data, as well as
combustion is mainly sensitive to small hydrocarbon chemistry. among numerical simulations, were found. The authors concluded
Jouzdani et al. [17] studied ignition and pyrolysis time scales of on the need for further investigation to understand the nature of
propyl alcohols and the experiments were compared with several model discrepancies among the tested mechanisms.
literature models [1,13,16]. Investigations endorsed what stated in In addition to the mentioned experimental investigations, ded-
other previous works, namely the highest reactivity of the primary icated modeling studies were also performed. Ranzi et al. [31] de-
alcohol under the tested conditions. Moreover, among tested mod- veloped a reaction scheme for hydrocarbon and oxygenated fuels
els, Sarathy mechanism [1] showed the closest agreement with the and used the LBV data from [22] to validate the propanol isomers
experimental data. subset, achieving very good agreement between the model and ex-
A counterflow diffusion flame burner was employed by Sinha periments. The detailed kinetic model of Man et al. [16] was em-
and Thomson [18] to determine profiles of several C3 oxygenated ployed by Liu et al. [32] to build a reduced PRF-alcohols mecha-
hydrocarbons and their products and stable intermediates. Partic- nism and validated against several literature datasets. With the in-
ularly, for i-propanol, the product compositions were found sim- tent to deepen the knowledge of alcohol combustion properties,
ilar to those reported by the previous literature studies [3,7,28]. Sarathy et al. [1] made a review on alcohol combustion chem-
Lately, Li et al. [19] studied the combustion chemistry of acetone istry, proposing also a new kinetic model for C1 –C5 alcohols. For
and propyl alcohols flames using synchrotron photoionization and the propanol isomers, the model was tested against experimental
molecular-beam mass spectrometry (MBMS); different structural data from [16] and [22]. For the IDTs data, good fidelity was found,
features of the oxygenated fuels were analyzed and discussed. Fur- while for the LBVs experiments there was a general trend in over-
thermore, the production routes of known and newly observed in- predicting the experiments in rich conditions. Authors also stated
termediates were found or proposed. Kasper et al. [20] employed that the lower LBV and higher IDT of the branched alcohol stem
two independent MSBM techniques, namely electron ionization from the more stable form of its combustion intermediates.
(EI) and vacuum UV light for single-photon ionization (VUV-PI), to In Table 1 are summarized the most recent kinetic works on
determine the products of n-propanol and i-propanol oxidation in propyl alcohols pyrolysis, oxidation and combustion as well as
stoichiometric and rich low-pressure flat flames. Profiles of major the literature data used for their validation. From the in-depth
species were found very similar for the two isomers, while larger analysis of the literature studies, it emerged as the existing LBV
differences were found for oxygenated intermediates. Flame struc- and kinetic studies are somewhat contradictory. In virtue of the
tures of propyl alcohols were investigated by Frassoldati et al. [21]. above, the aim of this study was twofold. First, to propose a new
G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204 191

Table 1
Most recent kinetic models and experimental studies used for their validation.

detailed kinetic mechanism of C3 alcohol isomers combustion. equaled by the heat received by the inlet gasses flowing through
In this regard, new LBV measurements were also performed, for the preheated burner plate. The heat flux method provides flat
the first time, with the heat flux method. This method ensures flame configurations free from the stretch effects, which are well
robustness and reproducibility of the LBV measurements and known to be one of the major issues in LBV determination per-
therefore is an extremely useful tool for the model validation. formed with other methods, i.e., spherical and counterflow flames.
Then, the proposed kinetic mechanism is also validated against Method description, as well as the experimental setup and the data
LBV existing data [9–11,14,24–26], obtained with other different processing procedures, are extensively explained elsewhere [33],
methods. Second, to try to elucidate the discrepancies that have while the associated measurement uncertainties are discussed in
arisen between experimental and simulated LBV came out in [34]. New LBV of propyl alcohols +air mixtures, together with as-
previous studies [10,11,14,24–26], by checking and comparing sociated uncertainties, are available as Supplemental Material.
the consistency of the different burning velocity measurements
performed with different experimental methods. Additionally, the 3. Kinetic mechanism
proposed new kinetic model was further assessed against a large
literature dataset (see Table 1). Finally, the proposed model was The present kinetic model, hierarchically structured, consists
compared with Sarathy et al. [1] kinetic mechanism predictions. of 1787 reactions and 161 species and is based on the most
Sarathy mechanism, consisting of over 3600 reactions and more recent Konnov mechanism provided by Capriolo et al. [35] with
than 680 species, was tested in the current study since its C3 some important modification introduced by Konnov [36] i.e., the
alcohol isomers subsets are mainly based on the well-established integration of the termolecular reactions associated to hydrogen
mechanism [16] to which important low-temperature chemical oxidation, namely, H + O2 + R suggested by Klippenstein and
reactions have been added. Moreover, the model of Sarathy et Burke [37,38]; furthermore, the transport properties were updated
al. [1] was also tested in several recent publications [1,17,23], using calculations of Jasper et al. [39,40]. In this work, the mech-
showing good performance and also because it was found superior anism has been extended to update and extend the chemistry of
when compared with other kinetic models [23]. propanol isomers. All of the most important reactions associated
with n-propanol and i-propanol oxidation and pyrolysis, along
2. Experimental with their Arrhenius coefficients, are listed in Tables 2 and3 and
discussed in this section. All the rate constant expressions adopted
In this work, the heat flux method was applied to determine in the propyl alcohols submodel were carefully selected to assure
the LBVs of C3 alcohols at atmospheric pressure and over the a consistent modeling development; further details related to the
equivalence ratio range of  = 0.7–1.4. The low vapor pressure rate constants selections are available in the reaction datasheets in
of propyl alcohols did not allow performing measurements in the the Supplemental Material. Some rate constants of the pressure-
gas phase at standard conditions; as a result, the unburned mix- dependent reactions are incorporated using PLOG formulation;
ture temperature was set to 323 K; higher temperatures, i.e., 343 K for these reactions only the rate constants at 1 atm are listed in
and 393 K, were also explored. The test fuels were provided by Tables 2 and 3. The prresent detailed reaction mechanism with
Alfa Aesar with stated high-purity (99+% for 1-propanol and 99.5% the thermodynamic and the transport properties, is also provided
for 2-propanol); the oxidizer was provided by AGA with the fol- in the CHEMKIN format as Supplemental Material.
lowing composition: 21% O2 , 79% N2 . In the heat flux method According to the study of Oyeyemi et al. [41], performed using
a flat premixed flame is stabilized on a perforated burner under multireference averaged coupled-pair functional (MRACPF2), the
near-adiabatic conditions. The aforesaid conditions occur when the hyperconjugated Cα -Cβ and Cα -H (see Fig. 1) bonds in C1 –C4 al-
heat losses needed to stabilize the flame on the burner plate are cohols have the lowest bond dissociation energy values (BDEs). As
192 G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204

Table 2
Arrhenius parameters selected for n-propanol kinetic sub-mechanism. Units: A – cm3 mol−1 s−1 , Ea – cal mol−1 , T – K.

No Reaction A n Ea UF Ref.

Hydrogen abstraction
1 nC3 H7 OH + OH  C3 H7 O + H2 O 5.88E+02 2.82E+00 −5.85E+02 2 [1]
2 nC3 H7 OH + OH  C2 H5 CHOH + H2 O 3.61E+03 2.89 −2291.0 2 [1]
3 nC3 H7 OH + OH  CH3 CHCH2 OH + H2 O 1.54E+00 3.70 −3736.0 2 [1]
4 nC3 H7 OH + OH  CH2 CH2 CH2 OH + H2 O 5.28E+09 9.70E-01 1.59E+03 2 [13]
5 nC3 H7 OH + H  C3 H7 O + H2 5.36E+04 2.53E+00 4.405E+03 2 [13]
6 nC3 H7 OH + H  C2 H5 CHOH + H2 8.79E+04 2.68E+00 2.92E+03 2 [47]
7 nC3 H7 OH + H  CH3 CHCH2 OH + H2 1.11E+05 2.40E+00 2.01E+03 2 [13]
8 nC3 H7 OH + H  CH2 CH2 CH2 OH + H2 6.66E+05 2.54E+00 6.76E+03 2 [13]
9 nC3 H7 OH + O2  C3 H7 O + HO2 1.00E+13 0.00E+00 4.82E+04 2 [13]
10 nC3 H7 OH + O2  C2 H5 CHOH + HO2 2.00E+13 0.00E+00 4.95E+04 2 [13]
11 nC3 H7 OH + O2  CH3 CHCH2 OH + HO2 2.44E+12 0.22 4.98E+04 2 [13]
12 nC3 H7 OH + O2  CH2 CH2 CH2 OH + HO2 3.00E+13 0.00E+00 5.23E+04 2 [13]
13 nC3 H7 OH + HO2  C3 H7 O + H2 O2 2.50E+12 0.00E+00 2.40E+04 2 [13]
14 nC3 H7 OH + HO2  C2 H5 CHOH + H2 O2 3.50E-05 5.26E+00 8.27E+03 2.5 [49]
15 nC3 H7 OH + HO2  CH3 CHCH2 OH + H2 O2 7.51E-03 4.52E+00 1.47E+04 2.5 [49]
16 nC3 H7 OH + HO2  CH2 CH2 CH2 OH + H2 O2 8.80E-02 4.31E+00 1.73E+04 2.5 [49]
17 nC3 H7 OH + O  C3 H7 O + OH 1.46E-03 4.73E+00 1.73E+03 2 [50]
18 nC3 H7 OH + O  C2 H5 CHOH + OH 1.45E+05 2.47E+00 8.76E+02 2 [50]
19 nC3 H7 OH + O  CH3 CHCH2 OH + OH 1.44E+05 2.61E+00 3.03E+03 2 [13]
20 nC3 H7 OH + O  CH2 CH2 CH2 OH + OH 9.81+05 2.43E+00 4.75E+03 2 [13]
21 nC3 H7 OH + CH3  C3 H7 O + CH4 2.04E+00 3.57 7721.0 2 [13]
22 nC3 H7 OH + CH3  C2 H5 CHOH + CH4 1.993E+01 3.37 7634.0 2 [13]
23 nC3 H7 OH + CH3  CH3 CHCH2 OH + CH4 8.02E+00 3.23 6461.0 2 [13]
24 nC3 H7 OH + CH3  CH2 CH2 CH2 OH + CH4 4.53E-01 3.65 7154.0 2 [13]
25 nC3 H7 OH + C2 H5  C3 H7 O + C2 H6 1.00E+11 0.00E+00 9.20E+03 2 [13]
26 nC3 H7 OH + C2 H5  C2 H5 CHOH + C2 H6 2.00E+11 0.00E+00 1.10E+04 2 [13]
27 nC3 H7 OH + C2 H5  CH3 CHCH2 OH + C2 H6 7.66E+00 3.11E+00 1.22E+04 2 [13]
28 nC3 H7 OH + C2 H5  CH2 CH2 CH2 OH + C2 H6 4.62E-01 3.65E+00 9.14E+03 2 [13]
29 nC3 H7 OH + CH2 OH  C3 H7 O + CH3 OH 1.20E+02 2.76E+00 1.08E+04 2 [13]
30 nC3 H7 OH + CH2 OH  C2 H5 CHOH + CH3 OH 6.00E+01 2.95E+00 1.20E+04 2 [13]
31 nC3 H7 OH + CH2 OH  1.53E+01 3.11E+00 1.22E+04 2 [13]
CH3 CHCH2 OH + CH3 OH
32 nC3 H7 OH + CH2 OH  1.01E+02 2.95E+00 1.40E+04 2 [13]
CH2 CH2 CH2 OH + CH3 OH
33 nC3 H7 OH + CH3 O  C3 H7 O + CH3 OH 2.30E+10 0.00E+00 2.90E+03 2 [13]
34 nC3 H7 OH + CH3 O  C2 H5 CHOH + CH3 OH 1.50E+00 2.95E+00 4.50E+03 2 [13]
35 nC3 H7 OH + CH3 O  CH3 CHCH2 OH + CH3 OH 3.02E+00 1.80E-01 4.70E+03 2 [13]
36 nC3 H7 OH + CH3 O  CH2 CH2 CH2 OH + CH3 OH 2.17E+11 0.00E+00 6.46E+03 2 [13]
37 nC3 H7 OH + CH3 O2  C3 H7 O + CH3 O2 H 2.50E+12 0.00E+00 2.40E+04 2 [13]
38 nC3 H7 OH + CH3 O2  C2 H5 CHOH + CH3 O2 H 6.00E+12 0.00E+00 1.60E+04 2 [13]
39 nC3 H7 OH + CH3 O2  1.56E+03 2.81E+00 1.43E+04 2 [13]
CH3 CHCH2 OH + CH3 O2 H
40 nC3 H7 OH + CH3 O2  2.38E+04 2.55E+00 1.65E+04 2 [13]
CH2 CH2 CH2 OH + CH3 O2 H
41 nC3 H7 OH + HCO  C3 H7 O + CH2 O 3.40E+04 2.50E+00 1.35E+04 2 [13]
42 nC3 H7 OH + HCO  C2 H5 CHOH + CH2 O 1.00E+07 1.90E+00 1.70E+04 2 [13]
43 nC3 H7 OH + HCO  CH3 CHCH2 OH + CH2 O 5.16E+05 2.25E+00 1.68E+04 2 [13]
44 nC3 H7 OH + HCO  CH2 CH2 CH2 OH + CH2 O 1.02E+05 2.50E+00 1.84E+04 2 [13]
Unimolecular decomposition
45 nC3 H7 OH  C3 H7 O + H 7.03E+14 7.40E−02 1.05E+05 2 [13]
46 nC3 H7 OH  C2 H5 CHOH + H 1.78E+16 −1.73E−01 9.53E+04 2 [13]
47 nC3 H7 OH  CH3 CHCH2 OH + H 3.52E+18 −7.28E−01 1.00E+05 2 [13]
48 nC3 H7 OH  CH2 CH2 CH2 OH + H 1.84E+17 −3.56E−01 1.01E+05 2 [13]
49 nC3 H7 OH (+M)  C3 H6 + H2 O (+M) 3.53E+13 0.00E+00 6.73E+04 2 [13]
Low P limit 1.99E+81 −1.86E+01 7.92E+04
Troe a = 0.13 T3 = 1 T1 = 8.94E+09 T2 = 9.33e+09
50 nC3 H7 OH (+M)  CH3 + PC2 H4 OH (+M) 8.81E+23 −2.12 8.98E+04 2 [13]
Low P limit 2.20E+64 −13.39 9.53E+04
Troe a = 0.307 T3 = 1110 T1 = 9.33E+09 T2 = 9.33e+09
51 nC3 H7 OH (+M)  C2 H5 + CH2 OH (+M) 1.728E+26 −2.72 86,207.0 2 [45]
Low P limit 1.6771E+89 −20.364 99,453.0
Troe a = 5.5845E−03 T3 = 358.2 T1 = 4.9325E+07 T2 = 2.4761E+03
52 nC3 H7 OH (+M)  C3 H7 + OH (+M) 5.26E+20 −1.34 9.46E+04 2 [13]
Low P limit 8.92E+57 −11.78 99,410
Troe a = 0.355 T3 = 942 T1 = 9.92E+09 T2 = 9.92E+09
Radical decomposition
53 C2 H5 CHOH + O2  C2 H5 CHO + HO2 5.28E+17 −1.638 839 2 [53]a
54 CH3 CHCH2 OH + O2  C3 H5 OH + HO2 1.50E+12 0.00E+00 5.00E+03 2 [21]
55 CH3 CHCH2 OH + O2  C2 H5 CHO + HO2 1.50E+12 0.00E+00 5.00E+03 2 [21]
56 CH2 CH2 CH2 OH + O2  C3 H5 OH + HO2 1.50E+12 0.00E+00 5.00E+03 2 [21]
57 C3 H7 O  C2 H5 + CH2 O 2.46E+35 −7.69 1.92E+04 2 [51]a
58 C3 H7 O  C2 H5 CHO + H 1.72E+19 −3.41 1.80E+04 2 [51]a
59 C3 H7 O  CH2 CH2 CH2 OH 7.46E+09 6.18E−01 2.08E+04 2 [52]
60 C3 H7 O  C2 H5 CHOH 2.72E+11 6.35E−01 2.92E+04 2 [52]
(continued on next page)
G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204 193

Table 2 (continued)

No Reaction A n Ea UF Ref.

61 C3 H7 O  CH3 CHCH2 OH 4.88E+10 5.62E−01 2.75E+04 2 [52]


62 CH2 CH2 CH2 OH  C2 H4 + CH2 OH 9.29E+38 −8.33 4.33E+04 2 [51]a
63 CH2 CH2 CH2 OH  C3 H5 OH + H 1.60E+12 0.427 4.08E+04 2 [52]
64 CH2 CH2 CH2 OH  C2 H5 + CH2 O 7.10E+41 −9.13 4.48E+04 2 [51]a
65 CH2 CH2 CH2 OH  C3 H6 + OH 3.16E+37 −8.17 4.19E+04 2 [51]a
66 C2 H5 CHOH  C2 H3 OH + CH3 5.67E+41 −8.8 4.19E+04 2 [51]a
67 C2 H5 CHOHC2 H5 CHO+H 1.03E+11 0.83 36,830 2 [52]
68 C2 H5 CHOH  CH3 CHCHOH + H 7.10E+41 −9.13 4.48E+04 2 [51]a
69 C2 H5 CHOH  CH3 CHCH2 OH 2.34E+10 8.38E-01 4.24 E+04 2 [52]
70 C2 H5 CHOH  C2 H5 + CH2 O 3.16E+37 −8.17 4.19E+04 2 [51]a
71 C2 H5 CHOH  CH2 CH2 CH2 OH 2.49E+09 8.79E-01 4.13 E+04 2 [52]
72 CH3 CHCH2 OH  C3 H6 + OH 1.51E+33 −6.46 3.24E+04 2 [51]a
73 CH3 CHCH2 OH  H+C3 H5 OH 1.36E+28 −5.24 3.83E+04 2 [51]a
74 CH3 CHCH2 OH  H + CH3 CHCHOH 1.79E+32 −6.51 3.84E+04 2 [51]a
Keto-enol isomerization
75 CH3 CHCHOH + HO2  C2 H5 CHO + HO2 1.49E+05 1.67E+00 6.81E+03 2 [57]
76 CH3 CHCHOH + HOCHO  2.81E−02 3.29E+00 −4.51E+03 2 [58]
C2 H5 CHO + HOCHO
a
Pressure-dependent reaction described using PLOG formulation.

Fig. 1. BDEs of n-propanol (right) and i-propanol (left) expressed in kcal/mol from [40], depicting the different labels used in this work.

a result, the expected routes for fuel decomposition and metathe- nC3 H7 OH + H  C2 H5 CHOH + H2 (R6)
sis reactions are dominated by the scission of the aforementioned
bonds. nC3 H7 OH + H  CH3 CHCH2 OH + H2 (R7)

3.1. Unimolecular decomposition nC3 H7 OH + H  CH2 CH2 CH2 OH + H2 (R8)

For reaction (R6), the kinetic parameters were taken from


For the unimolecular decomposition of n-propanol, the rate
the experimental and theoretical work of Sivaramakrishnan [47].
constant expressions included in the current mechanism are taken
Channel-specific Arrhenius parameters were calculated using ab
from Johnson et al. [13] recommendations, in turn based on the
initio TST and successively adjusted to fit experimental data from
earlier studies of Tsang [42–44], and from Sarathy et al. [45]. For
their study and from Aders and Wagner [48], increasing the value
reactions (R100) and (R101), estimations from Man et al. [16] were
of each barrier by 0.8 kcal/ mol. The rate constants for reactions
accepted and included here.
(R5), (R7), and (R8) were adopted from the previous work on
propyl alcohols combustion of Johnson et al. [13].
3.2. Metathesis reactions
For i-C3 H7 OH, the direct H abstraction by H atom from occurs
via three channels:
Rate constant expressions needed to develop the C3 alcohol
isomers sub-model have been derived considering kinetic analo- iC3 H7 OH + H  iC3 H7 O + H2 (R80)
gies with published data for ethanol and butanol. Specifically, this
choice is made assuming that the H abstraction rates from the hy- iC3 H7 OH + H  tC3 H6 OH + H2 (R81)
droxyl group and Cα carbon are analogous to the ones that occur
in ethanol. On the other hand, the H atom transfer from Cβ and iC3 H7 OH + H  iC3 H6 OH + H2 (R82)
Cγ carbon to the specific radical is considered to be kinetically
identical to butanol. Such assumptions were made simply compar- Rate constants expressions related to the H abstraction from the
ing bond dissociation energy values of C1 –C4 alcohols calculated hydroxyl group, as well as from the CH3 group, were taken from
in [41]. the experimental and theoretical study on C2 H5 OH [34]. For reac-
For n-propanol and i-propanol, most of the rate constant ex- tion R81, the values recommended in the kinetic study of Johnson
pressions involving metathesis reactions were largely based on et al. [13] on propyl alcohols were accepted here.
Johnson [13] C3 alcohol subsets, which are in turn based on the For metathesis reactions of HO2 radical, the kinetic similarity
previous detailed chemical mechanism proposed by Black et al. with C4 H9 OH was applied for both C3 alcohol isomers. For (R14-
[46]. For n-C3 H7 OH, hydrogen atom transfer with H atom proceeds 16), the expressions suggested in the theoretical study of Zhou
via four channels: et al. [49], were adopted for the present model. Arrhenius parame-
ters in [49] were calculated using conventional TST with rigid-rotor
nC3 H7 OH + H  C3 H7 O + H2 (R5) harmonic oscillator approximations, over the temperature range of
194 G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204

Table 3
Arrhenius parameters selected for i-propanol kinetic sub-mechanism. Units: A – cm3 mol−1 s−1 , Ea – cal mol−1 , T – K.

No Reaction A n Ea UF Ref.

Hydrogen abstraction
77 iC3 H7 OH + OH  iC3 H7 O + H2 O 5.88E+02 2.82E+00 −5.85E+02 2 [16]
78 iC3 H7 OH + OH  tC3 H6 OH + H2 O 1.81E+03 2.89 −2611.0 2 [16]
79 iC3 H7 OH + OH  iC3H6 OH + H2 O 4.62E+00 3.70 −2946.0 2 [1]
80 iC3 H7 OH + H  iC3 H7 O + H2 9.45E+02 3.14E+00 8.70E+03 2 [47]
81 iC3 H7 OH + H  tC3 H6 OH + H2 2.30E+05 2.72 6260.0 2 [13]
82 iC3 H7 OH + H  iC3H6 OH + H2 5.31E+04 2.81E+00 7.50E+03 2 [47]
83 iC3 H7 OH + O2  iC3 H7 O + HO2 1.00E+13 0.00E+00 5.634E+04 2 [45]
84 iC3 H7 OH + O2  tC3 H6 OH + HO2 7.00E+12 0.00E+00 4.37E+04 2 [13]
85 iC3 H7 OH + O2  iC3H6 OH + HO2 4.20E+13 0.00E+00 5.20E+04 2 [13]
86 iC3 H7 OH + HO2  iC3 H7 O + H2 O2 2.50E+12 0.00 24,000.0 2.5 [49]
87 iC3 H7 OH + HO2  tC3 H6 OH + H2 O2 1.25E-05 5.26 7468.0 2 [16]
88 iC3 H7 OH + HO2  iC3H6 OH + H2 O2 2.26E-02 4.52 13,850.0 2 [1]
89 iC3 H7 OH + O  iC3 H7 O + OH 1.46E-03 4.73E+00 1.73E+03 2 [50]
90 iC3 H7 OH + O  tC3 H6 OH + OH 1.45E+05 2.47E+00 8.76E+02 2 [50]
91 iC3 H7 OH + O  iC3H6 OH + OH 9.70E+02 3.23 4660.0 2 [50]
94 iC3 H7 OH + CH3  iC3 H7 O + CH4 2.040E+00 3.57 −7721.0 2 [59]
93 iC3 H7 OH + CH3  tC3 H6 OH + CH4 9.56E−04 4.97 5710.0 2 [16]
94 iC3 H7 OH + CH3  iC3H6 OH + CH4 9.97E−12 7.05E+00 8.94E+03 2 [60]
95 iC3 H7 OH + CH3 O  tC3 H6 OH + CH3 OH 4.34E+11 0.00E+00 6.46E+03 2 [13]
96 iC3 H7 OH + CH3 O  iC3H6 OH + CH3 OH 3.00E+11 0.00E+00 7.00E+03 2 [13]
97 iC3 H7 OH + CH3 O2  tC3 H6 OH + CH3 O2 H 2.80E+12 0.00E+00 1.49E+04 2 [13]
98 iC3 H7 OH + CH3 O2  iC3H6 OH + CH3 O2 H 1.68E+13 0.00E+00 2.04E+04 2 [13]
Unimolecular decomposition
99 iC3 H7 O + H  iC3 H7 OH 1.00E+14 0.00E+00 0.00E+00 2 [45]
100 tC3 H6 OH + H  iC3 H7 OH 4.00E+13 0.00E+00 0.00E+00 2 [16]
101 iC3 H6 OH + H  iC3 H7 OH 4.00E+13 0.00E+00 0.00E+00 2 [16]
102 iC3 H7 OH (+M)  C3 H6 + H2 O (+M) 5.00E+13 0.00E+00 6.80E+04 2 [45]
Low P limit 4.82E+94 −22.464 8.35E+04
Troe a = 2.8246e-04 T3 = 362 T1 = 1.29E+04 T2 = 4.49E+03
103 iC3 H7 OH (+M)  CH3 + SC2 H4 OH (+M) 1.35E+24 −2.27E+00 8.70E+04 2 [13]
Low P limit 2.85E+69 −1.49E+01 9.47E+04
Troe a = 0.336 T3 =775 T1 =9.68E+09 T2 =8.21E+09
104 iC3 H7 OH (+M)  iC3 H7 + OH (+M) 6.43E+22 −1.865 9.62E+04 2 [45]
Low P limit 1.90E+78 −17.503 1.08E+05
Troe a = 8.6766e−05 T3 =2.93E+02 T1 =1.97E+04 T2 =1.67E+03
Radical decomposition
105 iC3 H7 O + O2  CH3 COCH3 + HO2 1.50E+12 0.00E+00 5.00E+03 2 [21]
106 tC3 H6 OH + O2  CH3 COCH3 + HO2 5.28E+17 −1.638 839 2 [53]a
107 iC3 H6 OH + O2  iC3 H5 OH + HO2 1.50E+12 0.00E+00 5.00E+03 2 [21]
108 iC3 H7 O  CH3 + CH3 CHO 8.10E+35 −7.88 1.89E+04 2 [51]a
109 iC3 H7 O  CH3 COCH3 + H 9.13E+21 −4.18 1.60E+04 2 [51]a
110 tC3 H6 OH  CH3 COCH3 + H 1.57E+40 −8.34 4.13E+04 2 [51]a
111 tC3 H6 OH  iC3 H5 OH + H 2.48E+38 −8.14 4.44E+04 2 [51]a
112 iC3 H6 OH  C3 H6 + OH 1.29E+35 −7.02 3.31E+04 2 [51]a
113 iC3 H6 OH  C2 H3 OH + CH3 2.27E+33 −6.63 3.58E+04 2 [51]a
Keto-enol isomerization
114 C2 H3 OH + HO2  CH3 CHO + HO2 1.49E+05 1.67E+00 6.81E+03 2 [57]
115 iC3 H5 OH  CH3 COCH3 6.957E+26 –4.138 57.344 2 [55]a , b
4.258E+56 –13.105 70.899 [55]a , c
116 iC3 H5 OH + HOCHO  CH3 COCH3 + HOCHO 2.81E-02 3.29E+00 −4.51E+03 2 [58]
117 iC3H5OH +HO2  CH3 COCH3 + HO2 1E−01+1.5E+00 3.1 + 3.0 16E+02+10E+03 2 [56]
a
Pressure-dependent reaction described using PLOG formulation.
b
Bath gas: N2.
c
Bath gas: Ar.

50 0–20 0 0 K with an estimated uncertainty factor of 2.5. Product carbon is the dominant one at T<600 K (96%), while at T>2300 K,
branching ratio analysis showed that H abstraction from Cα car- H abstraction from Cβ becomes the most prominent (44%).
bon is the most favorable along the investigated temperatures, go-
ing from ~96% at 20 0 0 K to ~ 47% at 500 K. Note that for (R14),
3.3. Radical decomposition
as also suggested by Man et al. [16], a factor of 2.5 was applied
to the rate constant. For (R13), recommendations from [11] were
N-propanol and i-propanol H abstraction lead to the formation
adopted.
of metastable radical isomers (4 and 3, respectively) which in turn
For reactions (R17), (R18), and (R89-91), the kinetic similarity
mainly decompose via β -scission. For thermal decomposition of
with the lower alcohol homologous C2 H5 OH led to accept Wu et al.
the alkyl/alkoxy radicals, pressure and temperature-dependent rate
[50] recommendations. Rate constant expressions in [50] were de-
parameters were only evaluated by Zador et al. [51]; such values
rived using CVT/SCT over a temperature range of 30 0–30 0 0 K and
were accepted and included in the current mechanism; however,
compared with experimental results obtained using a diaphragm-
for reaction (R63) and (R67), the recent recommendations from
less shock tube over the temperature range of 782–1410 K. Authors
the theoretical study of Ferros-Costas et al. [52] were included,
observed that in the branching ratio analysis H abstraction from Cα
since that allowed to have a better agreement with LBV and IDT
G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204 195

experimental data. Calculations of dissociation and isomerization


of primary radicals of propyl alcohols were made using the RRKM-
based energy-resolved master equation (ME) using Variflex 2.03
[51]. Zador et al. [51] also found out that the preferred dissociation
route for tC3 H6 OH radical is the cleavage of OH bond, since its
activation energy was found 4.6 kcal/mol lower than the one for
the β -scission channel. Isomerization channels of propoxy radical,
as well as H-migration from α -hydroxypropyl radical are not
included in [51] study; however, at high temperatures, their con-
tribution becomes non-negligible, hence rate constants reported
in [52] were included in the current mechanism. Two different
approaches of variational transition-state theory (VTST) were used
in [52] to calculate thermal rate constants, namely one-well (1 W)
and the more accurate multipath (MP) formulation. The present
study accepted the rate constant values obtained using the MP
formulation of VTST.
For oxidation of hydroxypropyl/propoxy radical by O2 , kinetic
analogies with oxidation of α -hydroxyethyl were applied. The ex-
pressions recommended in the work on ethanol oxidation of Da
Silva et al. [53], also lately endorsed by Sarathy et al. [45] for butyl
alcohol isomers, were adopted for reaction (R53) and (R106), while
for all the other oxidation pathways, Arrhenius coefficients from
[21] were accepted.

3.4. Enol-keto isomerization

Enols are formed due to radical decomposition of propyl


alcohols and therefore play an important role in the pyrolysis
and oxidation of C3 alcohols. However, they tend to tautomerize
into more stable structures in the condensed phase [54], i.e.,
to ketones (R115-117) or aldehydes (R75-76, R114). For reaction
R117, recommendations from Grajales-González et al. [55] were
accepted here. Rate constant calculations were performed using
System-Specific Quantum Rice-Ramsperger-Kassel (SS-QRRK) the-
ory, over the temperature range of 20 0–30 0 0 K and pressure of
0.1 − 108 kPa. Authors [55] also found that, as generally expected,
keto form is largely dominant at the equilibrium under aforemen-
tioned temperature conditions. For reaction (R117), the only study
in the literature refers to the recent investigation performed by
Monge-Palacios et al. [56], and the suggested Arrhenius parame-
ters included in the proposed model. Calculations were performed
using CCSD(T)/aug-ccpVTZ//M06-2X/cc-pVTZ ab initio together
with the MS-VTST over the temperature range 20 0–30 0 0 K.
Regarding the other uncatalyzed reactions, kinetic parallelism
with the tautomerism occurring between vinyl alcohol and ac-
etaldehyde was applied. In the light of the above, recommenda-
tions of Da Silva [57,58] were accepted when enol-keto tautomer-
ization were catalyzed by perhydroxyl radical (R75, R114) and by
formic acid (R76, R116).

4. Modeling details

Cantera software [61] was used to model adiabatic, one dimen-


sional and freely propagating flames of propanols + air mixtures
at atmospheric pressure. Multicomponent transport and thermal
diffusion (Soret Effect) were enabled for the calculations; conver- Fig. 2. Laminar burning velocities of propanols + air flames. n-propanol: black;
gence was achieved for adaptive grid parameters GRAD = 0.015 i-propanol: light blue. Symbols: experiments, line: modeling. Open diamonds:
present work; half right-solid circles: [22]; half right-solid down triangles: [26];
and CURV = 0.1, which allowed generating grid-independent so-
solid spheres [24]. Solid line: present model; dash line: prediction from Sarathy
lutions consisted of 850–1200 points. et al. [1].
In the work of Man et al. [16], IDT was defined as the time
from the arrival of the reflected shock wave at the endwall and [16] and also in the later work of Sarathy et al. [1]. Shock tube
the ignition onset. This latter was calculated by extrapolating the experiments were modeled using a homogeneous reactor under
maximum rate of change of OH∗ emission to the baseline, while constant volume and adiabatic condition. JSR and flow reactor data
their modeled IDT was defined as the time when the maximum were modeled, respectively, using a perfect stirred reactor and
dT/dt is reached. These experimental datasets in the present work closed homogeneous configuration. Finally, flame structure data
were modeled adopting the same modeling criterion applied in were simulated using a premixed burner configuration adopting
196 G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204

Fig. 3. Laminar burning velocities of n-propanol + air as a function of inlet gas temperature at atmospheric pressure plotted in log-log scale. Symbols: experimental data,
line: present model.

the provided flame temperature profile, while the pyrolysis exper- reactivity of n-propanol is due to the propensity to form more re-
iments were modeled using the same criteria adopted in [9], i.e., active intermediates during its combustion process, as also stated
using the plug flow reactor module. in Veloo and Egolfopoulos study [22].
To check data consistency, the analysis of the temperature de-
5. Results and discussion pendence of LBV of n-propanol + air mixture was performed at
different  for present and literature data; the influence of the
5.1. Laminar burning velocity initial gas temperature was analyzed in logarithmic coordinates
as shown in Fig. 3 and in Fig. S2 . An excellent congruity of the
Figure 2 depicts new and available literature LBVs measure- presented data is observable at all the temperatures tested and,
ments for n-propanol and i-propanol + air flames at P = 1 atm more generally, strong methods consistency were found between
and at T = 323–393 K; other literature data are compared with the the heat flux, MDC and with data from [9] performed using spher-
modeling in Fig. S1 in the Supplemental Material. Overall uncer- ical flames configuration, as well as with counterflow technique
tainties in LBVs determination related to this work were evaluated up to =1.10 (Fig. 3), after which important deviations were en-
adopting Alekseev et al. [34] procedure and are represented with countered (8–27%). When compared with the data from [26], the
error bars and typically restricted in a range of less than ±1 cm/s. present LBV measurements were found higher by approximately
Whenever possible, present LBV results were compared with litera- 11%; a similar trend was observed in our recent study on propanal
ture data performed with a different experimental method, i.e. the kinetic and oxidation [35] when the two experimental methods
counterflow technique [22] and the externally heated mesoscale were compared. Important deviations were also found with the
diverging channel (MDC) [24] at 343 K and spherical flames ap- other measurements performed in spherical flames [10,27].
proach [9,10,26] at 393 and 423 K. For both fuels, the LBV peak Overall, it appears rather evident from Fig. 3 and Fig. S2 that
lies around =1.10, appearing in agreement with what was found the experiments carried out with the spherical flames approach
with lower alcohol homologous; the higher values of LBVs result- [9,10,26,27] are inconsistent with each other. In this regard, fur-
ing for the oxidation of n-propanol + air mixtures, are related to ther investigations are needed to comprehend the reasons for
the higher flame temperatures (~12 K) reached for the oxidation such inconsistencies. As previously stated [35], one of the possible
of the primary alcohol. When compared to i-propanol, the higher explanations might lies in the longer residence time required to
G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204 197

Fig. 4. IDTs for n-propanol + O2 + Ar from [16]. Symbols: experiments, lines: mod- Fig. 5. IDTs for i-propanol + O2 + Ar from [16]. Symbols: experiments, lines: mod-
eling. Solid lines: present model; dash lines: Sarathy mechanism [1]. eling. Solid lines: present model; dash lines: Sarathy mechanism [1].

perform some of the experiments using spherical flames con- effects dependent on the chamber size, both reducing the flame
figuration, which could result in a partial oxidation of the fuel speed propagation and so the LBV values.
prior to its ignition. Moreover other possible error sources in the At all tested conditions from this study, i.e. 323 K, 343 K and
LBV determination could be due to stretch extrapolation of the 393 K, the general trend of LBV of n-propanol + air mixtures is
expanding flame, as well as radiation losses and compression well captured by the present and Sarathy et al. [1] models, while

Fig. 6. Sensitivity analysis of the ignition of n-propanol (black) and i-propanol (gray) at T = 1428 K, P = 1.2 bar and at =2.0.
198 G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204

Fig. 7. Mole fraction profiles of reactants and major products of n-propanol oxidation in a JSR at P = 10 atm and τ = 0.7 s. Symbols: experiments [10], lines: modeling. Solid
line: present model; dash line: Sarathy mechanism [1].

both slightly underestimate (~5%) the LBV of i-propanol + air mix- reflected shock waves with P = 1.2, 4 and 16 atm, T = 1100–
ture near stoichiometric conditions. One may note that both model 1500 K, at = 2.0; for both investigation the fuel concentration
behaviors are close to each other. When compared with the liter- was set to 0.75%. Overall, both mechanisms exhibit good perfor-
ature results, the models show satisfactory performance in repro- mances in reproducing the experimental results. However, for n-
ducing data from [9,11,24], while discrepancies arise when com- propanol-based experiments, the proposed mechanism was found
pared with [10,25,26,27] studies. Particularly, where simulations superior to the Sarathy model at 4 and 12 bar, while this latter
overestimate the experimental data, namely in [10,26,27] studies, shown a closer agreement to the experiments at 1.2 bar. It has to
the inconsistencies can be related to the abovementioned oxidative be noted that both models sensibly overpredict the fuel mixture
and thermal stability issues of the reactant mixture; a similar de- reactivity under rich conditions. When compared to i-propanol
viation from data of [18] is also observed at 3 bar (Fig.S1 ). On the IDTs data, the present mechanism exhibits better performance in
contrary, at high-pressure conditions, models considerably under- comparison to the Sarathy model [1] at 1.2 and 16 bar while it
predict values from [25] up to 20% (Fig.S1 ), while less noticeable tends to underpredict the fuel mixture reactivity at 4 bar.
differences are found when the models are compared with Li et To emphasize the difference between the models’ behavior,
al. [9] data. Such discrepancies might be explained, on the exper- brute-force sensitivity analysis of the simulated IDT’s was per-
imental side, with the possible presence of Darrieus-Landau insta- formed with both models at T = 1428 K, P = 1.2 atm and at
bilities, resulting in higher values of the LBV, and/or, on the model =2.0 and the most sensitive reactions were ranked and de-
side, to the deficiencies of fuel-specific pressure-dependent reac- picted in Fig. 6. A negative value indicates that a reaction de-
tions, as also suggested in [25], resulting in an underestimation of creases the temperature of the system, causing a longer ignition
the experimental values. delay time. For the ignition of n-propanol, the reaction (R8) is a
reactivity-inhibiting step for the present model, while it enhances
5.2. Ignition delays the propensity to ignite the primary alcohol in the Sarathy mecha-
nism, resulted of the different branching ratio for the nC3 H7 OH+H
In Figs. 4 and 5 the ignition characteristics of n- and i- reaction present in the two models. For the ignition of i-propanol,
propanol + O2 + Ar mixtures from Man et al. [16] are compared the only difference between the two models is related to the re-
with the numerical simulations performed with the present and action CH3 COCH3 C2 H6 +CO, which inhibits the reactivity of the
Sarathy model. The IDTs measurements were carried out behind branched isomer and is not present in Sarathy mechanism.
G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204 199

Fig. 8. Mole fraction profiles of reactants and major products of i-propanol oxidation in a JSR at P = 10 atm and τ = 0.7 s. Symbols: experiments [11], lines: modeling. Solid
line: present model; dash line: Sarathy mechanism [1].

5.3. Well stirred reactor

Figures 7 and 8 represent, respectively, experimental and pre-


dicted mole fraction profiles of the major species in stoichiometric
and rich n-propanol+O2 +N2 and i-propanol+O2 +N2 mixtures from
[10,11], together with some key intermediates involved in their
oxidation process. Similar comparisons for other mixtures studied
in the well-stirred reactor [10,11] are shown in Figs. S5 – S10 of
the Supplemental Material. Generally, as for the previous compar-
ison made at different regimes, both models reveal similar perfor-
mances. Particularly, the fuel reactivity and also the major species
formation is well reproduced by both models at all investigated
conditions, while some discrepancies arise in representing minor
species chemistry. However, for the oxidation of n-propanol fuel
mixtures, the proposed model shows a closer fidelity in reproduc-
ing the concentrations of H2 and C1 –C2 saturated and unsaturated
hydrocarbons, while the formaldehyde formation is systematically
underpredicted. For the oxidation of i-propanol fuel mixtures, in-
termediates like acetone, 2-propenal, propene, and ethylene forma- Fig. 9. Mole fraction profiles of major products of n-propanol oxidation in a flow
tions are well reproduced by the present model, and good predic- reactor at P = 1 atm, T = 1084 K, and = 0.64. Symbols: experiments [7], lines:
tions are even shown for CO2 product; however, numerical simu- modeling. Solid line: present model; dash line: Sarathy mechanism [1].
lation from the present model tends to underestimate H2 O con-
centration in the high-temperature region. A tendency to underes- 5.4. Flow reactor
timate the formation of non-oxygenated compounds is shown by
both models; such difference becomes more relevant as the exper- In Figs. 9 and 10 the experimental mole fraction profiles dur-
imental conditions become richer. ing flow reactor oxidation of C3 alcohols experimentally studied by
200 G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204

Fig. 10. Mole fraction profiles of major products of i-propanol oxidation in a flow
reactor at P = 1 atm, T = 1121 K, and = 0.78. Symbols: experiments [7], lines:
modeling. Solid line: present model; dash line: Sarathy mechanism [1].

Fig. 12. Mole fraction profiles of the major and minor species for and 3% i-propanol
pyrolysis at 1atm. Symbols: experiments [9], lines: modeling. Solid line: present
model; dash line: Sarathy mechanism [1].

Norton and Dryer [7] are compared with the present and Sarathy
model predictions. As for many fuel oxidation studies, to match
the point of 50% experimental fuel consumption, model predictions
were adjusted using time-shifting. For n-propanol oxidation, both
models show good agreement with experimental results; however,
in the present model simulations, CO and CO2 maxima are sensi-
bly shifted in time by ~20 ms. For i-propanol oxidation both mod-
els shown similar predictions, i.e., they tend to underestimate fuel
reactivity under the investigated conditions as well as some mi-
nor species formations, such as CO, C2 H4 and C3 H6 . On the con-
trary, CH4 formation is instead constantly overpredicted under the
all simulation condition range.
The effect of temperature on n-propanol and i-propanol py-
rolysis [9] is represented in Figs. 11 and 12; the experimental
mole fraction profiles of the major species are compared with
the present and Sarathy kinetic model predictions. Experiments
were performed in a flow reactor with 3% fuel in argon, at a
stated flow rate of 10 0 0 sccm. Comparison of the n-propanol ex-
perimental data with models predictions show that both models
Fig. 11. Mole fraction profiles of the major and minor species for and 3% n-propanol
well capture n-propanol conversion and formation of some ma-
pyrolysis at 1atm. Symbols: experiments [9], lines: modeling. Solid line: present jor species, such as H2 O, CH4 , CO, and C2 H4 . However, they tend
model; dash line: Sarathy mechanism [1]. to underestimate the formation of some minor species profiles
G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204 201

Fig. 13. Sensitivity analysis of the pyrolysis of n-propanol (black) and i-propanol (gray) at 50% of fuel consumption.

such as CH3 CHO and C2 H5 CHO; while C3 H6 concentration is bet- 5.5. Flame structure
ter represented by the present model. Regarding i-propanol pyrol-
ysis, models show a satisfactory performance in reproducing fuel In Figs. 14 and 15 and S11 –S12 are depicted the experimental
conversion and formation of major species while some discrep- mole fraction profiles of major intermediates measured in low-
ancies arise with the CH3 COCH3 profile, where experimental data pressure flames of C3 alcohol isomers [20], together with the
are overestimated and maxima shifted to higher temperature (by present and Sarathy model predictions. As previously mentioned,
~100 K). experiments were carried out using 2 different MBMS techniques,
Figure 13 shows sensitivity analyses performed at 50% fuel i.e., EI and VUV-PI, at the experimental condition of P = 3333–
decomposition point for both C3 alcohols pyrolysis; reactions 4666 Pa and =1.00–1.94. In this work, experiments conducted
with negative sensitivity coefficients promote fuel depletion. using EI are compared with model predictions. Simulation pro-
Both models show that the main contribution in promot- files were shifted by ~2 mm to better represent experimental data
ing/inhibiting n-propanol decomposition is due to the fall-off from [20]; a similar approach was also employed by Frassoldati et
reactions CH2 OH(+M)CH2 O+H(+M) and (R51), respectively. al. [21]. Once more, present and Sarathy mechanisms show sim-
For the present model an important contribution to decreas- ilar performances, and major species profiles of n-propanol and
ing fuel depletion is also given by the recombination reaction i-propanol are well captured by both; on the contrary, simulated
C2 H5 CO+CH3 O CH2 O+C2 H5 CHO, which is not included in the CO and H2 concentration profiles systematically underestimate ex-
Sarathy mechanism, whereas in this latter, the reactions (R8) perimental data. It has to be noted that simulations were per-
and CH2 OH+CH2 O=CH3 OH+HCO show a higher relevance than formed using the unperturbed flame temperature profiles provided
in the present model. For the pyrolysis of i-propanol, no major by Kasper et al. [20], since in their work the nozzle effect of the
differences are appreciable between the tested models for the sample probe on the flame temperature profile was not quantified
most sensitive reactions. and this might sensibly affect the measured species profiles [62].
202 G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204

Fig. 14. Mole fraction profiles of major products of n-propanol oxidation in a low- Fig. 15. Mole fraction profiles of major products of i-propanol oxidation in a low-
pressure burner at P = 0.033 atm, and = 1.03. Symbols: experiments [20], lines: pressure burner at P = 0.033 atm, and = 1.00. Symbols: experiments [20], lines:
modeling. Solid line: present model; dash line: Sarathy mechanism [1]. modeling. Solid line: present model; dash line: Sarathy mechanism [1].

The selection of the rate constants employed to build the n-


6. Conclusions
propanol and i-propanol subset was also presented. Models com-
parison shows that both kinetic mechanisms successfully repro-
This study provides a new kinetic model for propyl alcohols py-
duce new and literature data. However, the newly developed
rolysis and oxidation, together with new LBV measurements car-
model shows a closer fidelity in representing IDTs data obtained
ried out with the heat flux method at T = 343–393 K, P = 1 atm
at high-pressure conditions.
and =0.7–1.4. Moreover, all the available LBV data from the lit-
erature are used to validate the proposed model and the litera-
ture model from [1], as well as to check experimental data consis-
tency. In addition, the kinetic models were also tested against the Acknowledgments
literature experiments performed at different combustion regimes
(IDT, well-stirred and flow reactors, flame structure). Excellent The authors would like to acknowledge the financial support
agreement was found with experiments performed with the MDC from the Swedish Energy Agency through the centre for Combus-
method, as well as with the data from [9] obtained using spher- tion Science and Technology (Project KC-CECOST 22538-4), and the
ical flame configuration and with counterflow technique up to Swedish Research Council (VR) via project 2015-04042. The au-
slightly rich conditions. Important inconsistencies appeared with thors further acknowledge the valuable discussions with Dr. V. A.
other experiments performed in spherical flames, as also previ- Alekseev and Dr. T. Methling.
ously observed in our recent work [35] and, more generally, spher-
ical flame measurements were found inconsistent with each other.
Once again such differences were attributed to the possible oxida- Declaration of Competing Interest
tive phenomena occurring to the combustible mixture prior to its
ignition. None.
G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204 203

Supplementary materials [28] B.H. Bui, R.S. Zhu, M.C. Lin, Thermal decomposition of iso-propanol: first-prin-
ciples prediction of total and product-branching rate constants, J. Chem. Phys.
117 (24) (2002) 11188–11195.
Supplementary material associated with this article can be [29] G. Bourque, D. Healy, H. Curran, C. Zinner, D. Kalitan, J. de Vries, C. Aul,
found, in the online version, at doi:10.1016/j.combustflame.2020. E. Petersen, Ignition and flame speed kinetics of two natural gas blends with
05.012. high levels of heavier hydrocarbons, Proc. ASME Turbo Expo 3 (2008) 1051–
1066.
[30] W.K. Metcalfe, S.M. Burke, S.S. Ahmed, H.J. Curran, A hierarchical and compar-
ative kinetic modeling study of c1-c2 hydrocarbon and oxygenated fuels, Int. J.
Chem. Kinet. 45 (2013) 638–675.
References [31] E. Ranzi, A. Frassoldati, R. Grana, A. Cuoci, T. Faravelli, A.P. Kelley, C.K. Law, Hi-
erachical and comparative kinetic modeling of laminar flame speeds of hydro–
[1] S.M. Sarathy, P. Oßwald, N. Hansen, K. Kohse-Höinghaus, Alcohol combustion carbon and oxygenated fuels, Prog. Energy Combust. Sci. 38 (2012) 468–501.
chemistry, Prog. Energy Combust. Sci. 44 (2014) 40–102. [32] X. Liu, H. Wang, Z. Zheng, J. Liu, R.D. Reitz, M. Yao, Devel-
[2] V. Koppolu, V.K.R. Vasigala, Role of Escherichia coli in biofuel production, Mi- opment of a combined reduced primary reference fuel-alcohols
crobiol. Insights 9 (2016) 29–35. (methanol/ethanol/propanols/butanols/n-pentanol) mechanism for engine
[3] S.R. Smith, A.S. Gordon, Studies of diffusion flames. Diffusion flames of some applications, Energy 114 (2016) 542–558.
simple alcohols, Phys. Chem. J. 60 (8) (1956) 1059–1062. [33] V.A. Alekseev, M. Christensen, E. Berrocal, E.J.K. Nilsson, A.A. Konnov, Laminar
[4] J.A. Barnard, H.W.D. Hughes, The pyrolysis of n-propanol, Trans. Faraday Soc premixed flat non-stretched lean flames of hydrogen in air, Combust. Flame
56 (1960) 64–71. 162 (2015) 4063–4074.
[5] J.A. Barnard, The pyrolysis of isopropanol, Trans. Faraday Soc 56 (1960) 72–79. [34] V.A. Alekseev, J.D. Naucler, M. Christensen, E.J.K. Nilsson, E.N. Volkov, L.P.H. de
[6] A.B. Trenwith A, Thermal decomposition of isopropanol, J. Chem. Soc. Farad. Goey, A.A. Konnov, Experimental uncertainties of the heat flux method for
Trans. 1 71 (1975) 2405–2412. measuring burning velocities, Combust. Sci. Technol. 188 (2016) 853–894.
[7] T.S. Norton, F.L. Dryer, The flow reactor oxidation of C1-C4 alcohols and MTBE, [35] G. Capriolo, V.A. Alekseev, A.A. Konnov, An experimental and kinetic study of
Symp. (Int.) Combust. 23 (1991) 179–185. propanal oxidation, Combust. Flame 197 (2018) 11–21.
[8] J.S. Heyne, S. Dooley, Z. Serinyel, F.L. Dryer, H. Curran, Decomposition stud- [36] A.A. Konnov, Yet another kinetic mechanism for hydrogen combustion, Com-
ies of isopropanol in a variable pressure flow reactor, Z. Phys. Chem. 229 (6) bust. Flame 203 (2019) 14–22.
(2015) 881–907. [37] S.J. Klippenstein, From theoretical reaction dynamics to chemical modeling of
[9] W. Li, Y. Zhang, Y. Li, C. Cao, J. Zou, J. Yang, Z. Cheng, Experimental and ki- combustion, Proc. Combust. Inst 36 (2017) 77–111.
netic modeling study of n-propanol and i-propanol combustion: flow reactor [38] M.P. Burke, S.J. Klippenstein, Ephemeral collision complexes mediate chemi-
pyrolysis and laminar flame propagation, Combust. Flame 207 (2019) 171–185. cally termolecular transformations that affect system chemistry, Nat. Chem. 9
[10] B. Galmiche, C. Togbé, P. Dagaut, F. Halter, F. Foucher, Experimental and de- (2017) 1078–1082.
tailed kinetic modeling study of the oxidation of 1-propanol in a pressurized [39] A.W. Jasper, E. Kamarchik, J.A. Miller, S.J. Klippenstein, First-principles binary
jet-stirred reactor (JSR) and a combustion bomb, Energy Fuels 25 (5) (2011) diffusion coefficients for H, H2, and four normal alkanes+ N2, J. Chem. Phys.
2013–2021. 141 (2014) 124313.
[11] C. Togbé, P. Dagaut, F. Halter, F. Foucher, 2-propanol oxidation in a pressur- [40] A.W. Jasper, J.A. Miller, Lennard–Jones parameters for combustion and chem-
ized jet-stirred reactor (JSR) and combustion bomb: experimental and detailed ical kinetics modeling from full-dimensional intermolecular potentials, Com-
kinetic modeling study, Energy Fuels 25 (2011) 676–683. bust. Flame 161 (2014) 101–110.
[12] R. Wang, P. Cadman, Soot and PAH production from spray combustion of dif- [41] V.B. Oyeyemi, J.A. Keith, E.A. Carter, Trends in bond dissociation energies of
ferent hydrocarbons behind reflected shock waves, Combust. Flame 112 (1998) alcohols and aldehydes computed with multireference averaged coupled-pair
359–370. functional theory, J. Phys. Chem. A 118 (2014) 3039–3050.
[13] M.V. Johnson, S.S. Goldsborough, Z. Serinyel, P. O’Toole, E. Larkin, G. O’Malley, [42] W. Tsang, Thermal stability of alcohols, Int. J. Chem. Kinet. 8 (2) (1976)
H.J. Curran, A shock tube study of n- and iso-propanol ignition, Energy Fuels 173–192.
23 (2009) 5886–5898. [43] W. Tsang, Chemical kinetic data base for combustion chemistry. Part 3:
[14] B. Akih-Kumgeh, J.M. Bergthorson, Ignition of C3 oxygenated hydrocarbons and propane, J. Phys. Chem. Ref. Data 17 (1988) 887–952.
chemical kinetic modeling of propanal oxidation, Combust. Flame 158 (2011) [44] W. Tsang, Rate constants for the decomposition and formation of simple alka-
1877–1889. nes over extended temperature and pressure ranges, Combust. Flame 78 (1989)
[15] K.E. Noorani, B. Akih-Kumgeh, J.M. Bergthorson, Comparative high tempera- 71–86.
ture shock tube ignition of C1–C4 primary alcohols, Energy Fuels 24 (2010) [45] S.M. Sarathy, S. Vranckx, K. Yasunaga, M. Mehl, P. Oßwald, W.K. Metcalfe,
5834–5843. C.K. Westbrook, W.J. Pitz, K. Kohse-Höinghaus, R.X. Fernandes, H.J. Curran, A
[16] X. Man, C. Tang, J. Zhang, Y. Zhang, L. Pan, Z. Huang, C.K. Law, An experimen- comprehensive chemical kinetic combustion model for the four butanol iso-
tal and kinetic modeling study of n-propanol and i-propanol ignition at high mers, Combust. Flame 159 (2012) 2028–2055.
temperatures, Combust. Flame 161 (2014) 644–656. [46] G. Black, H.J. Curran, S. Pichon, J.M. Simmie, V. Zhukov, Bio-butanol: com-
[17] S. Jouzdani, A. Zhou, B. Akih-Kumgeh, Propanol isomers: investigation on igni- bustion properties and detailed chemical kinetic model, Combust. Flame 157
tion and pyrolysis time scales, Combust. Flame 176 (2017) 229–244. (2010) 363–373.
[18] A. Sinha, M.J. Thomson, The chemical structures of opposed flow diffusion [47] R. Sivaramakrishnan, M.C. Su, J.V. Michael, S.J. Klippenstein, L.B. Harding,
flames of C3 oxygenated hydrocarbons (isopropanol, dimethoxymethane, and B. Ruscic, rate constants for the thermal decomposition of ethanol and its bi-
dimethyl carbonate) and their mixtures, Combust. Flame 136 (2004) 548–556. molecular reactions with OH and D: reflected shock tube and theoretical stud-
[19] Y. Li, L. Wei, Z. Tian, B. Yang, J. Wang, T. Zhang, F. Qi, A comprehensive ies, J. Phys. Chem. A 114 (2010) 9425–9439.
experimental study of low-pressure premixed C3-oxygenated hydrocarbon [48] W.K. Von Aders, H.G. Wagner, Berichte der bunsengesellschaft fuer physikalis-
flames with tunable synchrotron photoionization, Combust. Flame 152 (2008) che chemie, Ber. Bunsenges. Phys. Chem 77 (9) (1973) 712–718.
336–359. [49] C.W. Zhou, J.M. Simmie, H.J. Curran, Rate constants for hydrogen abstraction
[20] T. Kasper, P. Oßwald, U. Struckmeier, K. Kohse-Höinghaus, C.A. Taatjes, J. Wang, by HȮ2 from n-butanol, Int. J. Chem. Kinet 44 (3) (2012) 155–164.
T.A. Cool, M.E. Law, A. Morel, P.R. Westmoreland, Combustion chemistry of [50] C.W. Wu, Y.P. Lee, S. Xu, M.C. Lin, Experimental and theoretical studies of rate
the propanol isomers — investigated by electron ionization and VUV-pho- coefficients for the reaction O(3P) + C2H5OH at high temperatures, J. Phys.
toionization molecular-beam mass spectrometry, Combust. Flame 156 (2009) Chem. A 111 (2007) 6693–6703.
1181–1201. [51] J. Zador, J.A. Miller, Unimolecular dissociation of hydroxypropyl and propoxy
[21] A. Frassoldati, A. Cuoci, T. Faravelli, U. Niemann, E. Ranzi, R. Seiser, K. Seshadri, radicals, Proc. Combust. Inst. 34 (2013) 519–526.
An experimental and kinetic modeling study of n-propanol and iso-propanol [52] D. Ferro-Costas, E. Martínez-Núñez, J. Rodriguez-Otero, E.M. Cabaleiro-Lago,
combustion, Combust. Flame 157 (2010) 2–16. C.M. Estevez, B. Fernandez, A. Fernandez-Ramos, S.A. Vazquez, Influence of
[22] P.S. Veloo, F.N. Egolfopoulos, Studies of n-propanol, iso-propanol, and propane multiple conformations and paths on rate constants and product branching
flames, Combust. Flame 158 (2011) 501–510. ratios. Thermal decomposition of 1-propanol radicals, J. Phys. Chem. A 122
[23] A. Alfazazi, U. Niemann, H. Selim, R.J. Cattolica, S.M. Sarathy, Effects of substi- (2018) 4790–4800.
tution on counterflow ignition and extinction of C3 and C4 Alcohols, Energy [53] G. da Silva, J. Bozzelli, L. Liang, J Farrell, Ethanol oxidation: kinetics of the
Fuels 30 (2016) 6091–6097. r-hydroxyethyl radical + O2 reaction, J. Phys. Chem. A 113 (2009) 8923–
[24] A. Katoch, A. Chauhan, S. Kumar, Laminar burning velocity of n-propanol 8933.
and air mixtures at elevated mixture temperatures, Energy Fuels 32 (2018) [54] K. Yasunaga, T. Mikajiri, S.M. Sarathy, T. Koike, F. Gillespie, T. Nagy, J.M. Sim-
6363–6370. mie, H.J. Curran, A shock tube and chemical kinetic modeling study of the
[25] J. Beeckmann, L. Cai, H. Pitsch, Experimental investigation of the laminar burn- pyrolysis and oxidation of butanols, Combust. Flame 159 (6) (2012) 2009–
ing velocities of methanol, ethanol, n-propanol, and n-butanol at high pres- 2027.
sure, Fuel 117 (2014) 340–350. [55] E. Grajales-González, M. Monge-Palacios, S.M. Sarathy, Theoretical kinetic
[26] J. Gong, S. Zhang, Y. Cheng, Z. Huang, C. Tang, J. Zhang, A comparative study study of the unimolecular keto-enol tautomerism Propen-2-ol ↔ Acetone.
of n-propanol, propanal, acetone, and propane combustion in laminar flames, Pressure effects and implications in the pyrolysis of tert- and 2-butanol, J.
Proc. Combust. Inst. 35 (2015) 795–801. Phys. Chem. A 122 (2018) 3547–3555.
[27] Q. Li, W. Jin, Z. Huang, Laminar flame characteristics of C1–C5 primary alco- [56] M. Monge-Palacios, E. Grajales-González, S. Mani Sarathy, Ab Initio, Tran-
hol-isooctane blends at elevated temperature, Energies 9 (2016) 511. sition state theory, and kinetic modelling study of the HO2-assisted Ke-
204 G. Capriolo and A.A. Konnov / Combustion and Flame 218 (2020) 189–204

to-enol Tautomerism Propen-2-ol + HO2 ⇔ Acetone + HO2 under combus- [60] D. Katsikadakos, C.W. Zhou, J.M. Simmie, H.J. Curran, P.A. Hunt, Y. Hardalupas,
tion, atmospheric, and interstellar conditions, J. Phys. Chem. A 122 (51) (2018) A.M.K.P. Taylor, Rate constants of hydrogen abstraction by methyl radical from
9792–9805. n-butanol and a comparison of CanTherm, MultiWell and Variflex, Proc. Com-
[57] G. da Silva, J.W. Bozzelli, Role of the α -hydroxyethylperoxy radical in the reac- bust. Inst. 34 (2013) 483–491.
tions of acetaldehyde and vinyl alcohol with HO2 , Chem. Phys. Lett. 483 (2009) [61] D.G. Goodwin, H.K. Moffat, R.L. Speth, Cantera: an object-oriented software
25–29. toolkit for chemical kinetics, thermodynamics, and transport processes, http:
[58] G. da Silva, Carboxylic acid catalyzed keto-enol tautomerizations in the gas //www.cantera.org, Version 2.2.0 (2015).
phase, Angew. Chem. Int. Ed. 49 (2010) 7523–7525. [62] A.T. Hartlieb, B. Atakan, K. Kohse-Höinghaus, Effects of a sampling quartz noz-
[59] Z.F. Xu, J. Park, M.C. Lin, Thermal decomposition of ethanol. III. A computa- zle on the flame structure of a fuel-rich low-pressure propene flame, Combust.
tional study of the kinetics and mechanism for the CH3 +C2 H5 OH reaction, J. Flame 121 (4) (20 0 0) 610–624.
Chem. Phys. 120 (2004) 6593.

You might also like