Combustion and Flame 228 (2021) 401–414

Contents lists available at ScienceDirect

Combustion and Flame

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

An experimental and kinetic modeling study of the ignition delay

characteristics of binary blends of ethane/propane and
ethylene/propane in multiple shock tubes and rapid compression
machines over a wide range of temperature, pressure, equivalence
ratio, and dilution
Sergio Martinez a, Mohammadreza Baigmohammadi a, Vaibhav Patel a,
Snehasish Panigrahy a,∗, Amrit B. Sahu a, Shashank S. Nagaraja a, Ajoy Ramalingam b,
A. Abd El-Sabor Mohamed a, Kieran P. Somers a, Karl A. Heufer b, Andrzej Pekalski c,
Henry J. Curran a,∗
a
Combustion Chemistry Centre, School of Chemistry, Ryan Institute, MaREI, NUI Galway, University Road, Galway H91 TK33, Ireland
b
Physico-Chemical Fundamentals of Combustion, RWTH Aachen University, 52056 Aachen, Germany
c
Shell Research Limited, Shell Centre London, London SE1 7NA, United Kingdom

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

Article history: In this work, the ignition delay time characteristics of C2 – C3 binary blends of gaseous hydrocarbons
Received 12 November 2020 including ethylene/propane and ethane/propane are studied over a wide range of temperatures (750 –
Revised 4 February 2021
20 0 0 K), pressures (1 – 135 bar), equivalence ratios (ϕ = 0.5 – 2.0) and dilutions (75 – 90%). A matrix of
Accepted 8 February 2021
experimental conditions is generated using the Taguchi (L9 ) approach to cover the range of conditions for
Available online 2 March 2021
the validation of a chemical kinetic model. The experimental ignition delay time data are recorded using
Keywords: low- and high-pressure shock tubes and two rapid compression machines (RCM) to include all of the
Ethane designed conditions. These novel experiments provide a direct validation of the chemical kinetic model,
Ethylene NUIGMech1.1, and its performance is characterized via statistical analysis, with the agreement between
Propane experiments and model being within ~ 26.4% over all of the conditions studied, which is comparable
Shock-tube with a general absolute uncertainty of the applied facilities (~ 20%). Sensitivity and flux analyses allow
Rapid compression machine
for the key reactions controlling the ignition behavior of the blends to be identified. Subsequent analyses
Ignition delay time
are performed to identify those reactions which are important for the pure fuel components and for
the blended fuels, and synergistic/antagonistic blending effects are therefore identified over the wide
range of conditions. The overall performance of NUIGMech1.1 and the correlations generated are in good
agreement with the experimental data.
© 2021 The Authors. Published by Elsevier Inc. on behalf of The Combustion Institute.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)

1. Introduction diesel, etc. are predicted to represent around 33% of energy con-
sumption, with natural gas at close to 30%, coal near 18%, with the
According to the U.S. Energy Information Administration (EIA) remaining 19% corresponding to nuclear, hydropower, and renew-
report 2019 [1], it is projected that global energy consumption will able sources [2].
increase by approximately 28% in 2050 compared to 2018 levels, The combustion of fossil fuels is the main sources of CO2 ,
with fossil fuels providing approximately 77% of the total energy SOx, and NOx emissions, among other pollutants. In this regard,
demand. Liquid fuels, natural gas, and coal are the most impor- natural gas is considered preferable to other fossil fuels including
tant sources amongst all fossil fuels. Liquid fuels, such as gasoline, liquid fuels and coal as it is a cleaner energy source, having the
highest hydrogen/carbon ratio among them. Widely used in the
∗
domestic, transportation, and industrial sectors, liquefied natural
Corresponding authors.
gas (LNG) is typically composed of methane (82–100%) but can
E-mail addresses: [email protected] (S. Panigrahy),
[email protected] (H.J. Curran). contain substantial amounts of ethane, propane and butane, while

https://doi.org/10.1016/j.combustflame.2021.02.009
0010-2180/© 2021 The Authors. Published by Elsevier Inc. on behalf of The Combustion Institute. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/)
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

Table 1
IDTs for C2 H4 , C2 H6 , C3 H8, and binary blends from the literature.

Facility Fuel pC / atm TC / K Year Reference

ST/RCM C2 H4 1 – 40 773 – 2200 1999 – 2020 [5, 8, 19-22]

ST/RCM C2 H6 1 – 40 830 – 1862 1971 – 2020 [5, 9, 23-25]
ST/RCM C3 H8 1 – 40 689 – 2615 1977 – 2013 [26-31]
ST/RCM C2 H4 /C2 H6 1 – 40 800 – 2000 2020 [6]

liquified petroleum gas (LPG) includes mainly propane, and bu- mance with the experimental data. Additionally, a comparison of
tane. To reduce emissions, it is necessary to improve the efficiency the most important reactions for the pure fuels and their binary
of combustion systems for which a detailed understanding of blends are presented to determine the kinetics controlling the re-
the controlling chemistry is essential. The oxidation kinetics of activity of the blends.
small hydrocarbons plays an important role as the base of any
mechanism for alternative fuels. For these reasons, the combustion 2. Design of experiments and experimental approaches
community is interested in enhancing our understanding of the
chemistry controlling the oxidation of hydrocarbons to increase All of the measured IDTs collected and presented in this study
the efficiency of engines and to reduce emissions of pollutants were obtained using two different shock tubes (ST) and two
such as soot, NOx, UHCs (unburned hydrocarbons), and green- rapid compression machines (RCMs). For those experiments car-
house gases in general. A hierarchical [3–6] (bottom-up) strategy ried out at NUI Galway at pressures ranging from 1 to 40 bar
has proven to be a good way to develop reliable chemical kinetic and intermediate-to-high-temperatures (> 10 0 0 K), low- (LPST)
mechanisms and improve our understanding of the chemistry (pC = 1 bar) and high-pressure (pC ≥ 20 bar) shock tubes (HPSTs)
controlling pyrolysis and oxidation. were applied. The IDT experiments corresponding to the relatively
Combustion properties of fuels such as ignition delay time high-pressure (20 ≤ pC ≤ 40 bar) and low-temperature (< 10 0 0 K)
(IDT), speciation profiles, flame speed, among others become in- regimes were taken using a twin-piston RCM. Some experiments
valuable for the optimization of combustors. Relevant experiments at working pressures of 40 bar and greater were measured using a
and modeling studies for mono-fuels and some blends, such as single-piston RCM at the Physico-Chemical Fundamentals of Com-
ethylene, ethane, and propane, have been carried out with differ- bustion (PCFC)-RWTH [32,33] Aachen University to enhance the fi-
ent methods and are available in the literature [7–14]. delity of the experimental IDTs. Details of these facilities and their
Dagaut et al. [15–17] studied species profiles consumed and operating characteristics are available in the literature [6,34,35].
produced during the oxidation ethylene, ethane, and propane in a For the IDT experiments performed at NUIG, ethane, ethylene,
jet-stirred reactor (JSR) using fuel mixtures diluted with nitrogen, and propane gases with a purity of 99.95% were supplied by Air
at equivalence ratios (ϕ ) of 0.1 – 4.0, at pressures ranging from 1 to liquid UK. BOC Ireland provided all other gases with purities of
10 atm in the temperature range 800–1250 K. Their work showed 99.99% for oxygen, nitrogen, argon, and 99.96% for helium. At the
the importance of small molecule sub-mechanisms including CO2 , PCFC-RWTH Aachen University, the alkane/alkene gases were sup-
CH2 O, CH4 , C2 H2 , C2 H4 , and C3 H6 on the combustion of higher hy- plied by Westfalen AG with a 99.95% purity. All other gases were
drocarbons. supplied by Westfalen AG and Praxair with purities of oxygen ≥
Lowry et al. [18] measured laminar premixed flame speeds 99.995%, nitrogen ≥ 99.95%, and argon ≥ 99.996%.
of pure methane, ethane, propane, and their binary blends with To stochastically distribute the experimental IDTs, the experi-
methane, at ϕ = 0.7 – 1.3, in a constant-volume cylindrical vessel, mental conditions for this study were generated using the Taguchi
in the pressure range 1–10 atm, at room temperature (298 K). They [36] approach by applying an L9 matrix based on four parameters
highlighted the need to extensively study the synergistic effect of of propane concentration, pressure, equivalence ratio, and dilution
blends in comparison to pure fuels. and also three different levels for each parameter studied. This ap-
Baigmohammadi et al. [5,6] measured IDTs for pure ethylene, proach has already been described by Baigmohammadi et al. [5,6].
ethane, and propane, and binary alkane/alkene blends in a shock For the mixture conditions studied, the propane concentration
tube (ST) and in a rapid compression machine (RCM) at ϕ = 0.5– in the fuel blends varies from 10 to 50%, at pressures ranging from
2.0, at pressures ranging from 20 to 40 atm in the temperature 10 to 135 bar, for ϕ of 0.5 – 2.0 and at dilutions of 75 – 90% (75%
range of 80 0–20 0 0 K. They showed that the synergistic effect on N2 + 0–15% Ar). However, the ratio between the diluents were
the reactivity of the mixture is important not only based on the changed at low–temperature regime (RCM) depending on the de-
fuel blends but in each variable considered during combustion sired compressed gas temperature. A synopsis of the designed con-
such as pressure, temperature, dilution, etc. These previous studies ditions is presented in Figure 1 and Table 2.
[5,6,15–17] also used chemical kinetic mechanisms to predict the In the current study, most of the measured IDTs in the HPST
experimental data presented and identify the most relevant chem- and RCMs [10,33,37–42] are defined as the time between the end
ical reactions controlling the oxidation of these fuels. of compression and the maximum gradient in pressure ( ddtp ) behind
Despite the large amount of data available for pure ethylene, the reflected shock. However, we define the ignition event as the
∗
ethane, and propane fuels, there are comparatively fewer studies maximum gradient in CH∗ ( dCdtH ) behind the reflected shock in the
of their blends, Table 1. LPST measured by a photomultiplier and also when the test mix-
The current study aims to address the lack of data for mixtures ture is highly diluted in the HPST. The corresponding uncertain-
by providing IDT data for binary C2 H4 /C3 H8 and C2 H6 /C3 H8 blends ties involved in the measured IDTs are discussed by Baigmoham-
over a wide range of temperatures, pressures, equivalence ratios, madi et al. [5,6]. Based on the analysis, the uncertainties in com-
and dilutions relevant to engine and gas turbine conditions. It also pressed mixture temperatures (σ T c,5 ) and measured IDTs change
aims to validate a detailed chemical kinetic model using the novel for every individual experimental point depending on the initial
experiments and literature data. We first provide a summary of the temperature, pressure, and/or mixture composition. In this regard,
experimental conditions and approaches taken for this study, fol- the average uncertainties of the compressed temperatures and the
lowed by details of the modeling work. The results and discussion measured IDTs in NUIG-L/HPSTs are estimated to be approximately
section encompasses all of the comparisons of the model perfor- ± 10 K and ± 25%, respectively. However, the compressed temper-

402
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

Fig. 1. Experimental Taguchi [36] L9 matrix of conditions. For 90%/10%, 70%/30%, and 50%/50% ratios (a) red: binary C2 H4 /C3 H8 blends, blue: binary C2 H6 /C3 H8 blends. (b)
pressure, equivalence ratio, and dilution parameters. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
article.)

Table 2
C2 H4 /C3 H8 and C2 H6 /C3 H8 mixture compositions in% mole volume in the current study. Where keywords NUIG refers to ST/RCM facilities at C3 -NUIGalway, and PCFC refers
to RCM facility at PCFC-RWTH Aachen University, respectively.

No. % C2 H 6 % C2 H 4 % C3 H 8 % O2 Dilution ϕ pC / bar Facility

C2 H4 /C3 H8 1 0.000 1.40 1.400 22.20 75% N2 0.5 1 NUIG

50%/50% 2 0.000 1.50 1.500 12.00 75% N2 +10% Ar 1.0 20 NUIG
3 0.000 1.70 1.700 6.60 75% N2 +15% Ar 2.0 40 NUIG/ PCFC
C2 H4 /C3 H8 4 0.000 3.75 1.610 9.64 75% N2 +10% Ar 2.0 1 NUIG
70%/30% 5 0.000 0.85 0.360 8.790 75% N2 +15% Ar 0.5 20 NUIG
6 0.000 3.80 1.600 19.60 75% N2 1.0 40 NUIG
C2 H4 /C3 H8 7 0.000 2.10 0.200 7.70 75% N2 +15% Ar 1.0 1 NUIG
90%/10% 8 0.000 8.60 1.000 15.40 75% N2 2.0 20 NUIG
9 0.000 1.80 0.200 13.00 75% N2 +10% Ar 0.5 40 NUIG/ PCFC
C2 H6 /C3 H8 10 1.300 0.00 1.300 22.40 75% N2 0.5 1 NUIG
50%/50% 11 1.430 0.00 1.430 12.14 75% N2 +10% Ar 1.0 20 NUIG
12 1.600 0.00 1.600 6.80 75% N2 +15% Ar 2.0 40 NUIG/ PCFC
C2 H6 /C3 H8 13 3.530 0.00 1.510 9.96 75% N2 +10% Ar 2.0 1 NUIG
70%/30% 14 0.790 0.00 0.340 8.87 75% N2 +15% Ar 0.5 20 NUIG
15 3.535 0.00 1.515 19.95 75% N2 1.0 40 NUIG
C2 H6 /C3 H8 16 1.940 0.00 0.220 7.84 75% N2 +15% Ar 1.0 1 NUIG
90%/10% 17 8.000 0.00 0.900 16.10 75% N2 2.0 20 NUIG
18 1.600 0.00 0.200 13.20 75% N2 +10% Ar 0.5 40 NUIG/ PCFC
19 1.860 0.00 0.210 7.53 45.2% N2 +45.2% Ar 1.0 90 PCFC
20 2.520 0.00 0.280 20.40 76.8% N2 0.5 120 PCFC
21 1.860 0.00 0.210 7.53 65.4% N2 +25% Ar 1.0 135 PCFC

ature uncertainty and the measured IDT variation in the NUIG and previous publications [5,6,43,50], and thus are discussed in this
PCFC RCMs are evaluated to be about ± 5 – 15 K and ± 20% over study.
the entire range of conditions. The experimental results were simulated using Python scripts
based on the Cantera 2.4 [53] library and the CHEMKIN-Pro 18.2
3. Computational modeling [54] software. Cantera is suitable for automatization making data
manipulation faster; however, Chemkin-Pro is faster for simula-
In the current study, NUIGMech1.1 is used to simulate the ex- tions involving large mechanisms and thus is more suitable for
perimental targets. This mechanism comprises 2746 species and simulations when a full mechanism is required. As mentioned
11,270 reactions, which is developed based on series of recent above, the definition of IDT is taken as the maximum gradient of
experimental [4–6,43–47] and theoretical studies [48–50]. These pressure or radical concentration with respect to time for the ST
works are outcome of continuous evolution of the detailed NUIG- simulations. In the RCM simulations, facility effects are included
Mech1.1 model which is extensively validated in the prior studies using the volume-time profiles derived from non-reactive experi-
for oxidation of C1 – C2 hydrocarbons [5,6], natural gas mixtures mental pressure-time traces in which O2 is replaced by N2 in the
[44], propane/propene blends [47], propyne [45], iso-butene [51], mixture [55,56].
as well as auto-ignition and pyrolysis of C2 – C6 alkenes [4,46]. The global model uncertainties,  MAD and  MAPE, are calcu-
The current work is a part of simultaneous development of the lated based on the differences between the experimental data and
overall NUIGMecah1.1 mechanism. For the purpose of comparison, mechanism simulated data using the Mean Absolute Deviations
AramcoMech3.0 [52] is also utilized to perform simulations against (MAD), and the Mean Absolute Percentage Error (MAPE), Eqs. (1)
the IDT experimental data from this study. Modifications of the and (2), respectively. However, to analyze the data with an indi-
most important reactions explicit to ethane, ethylene and propane vidual error, the Relative Percentage Error (RPE), ε RPE, was used
chemistry in NUIGMech1.1 have not been mentioned in detail in (Eq. (3)) to generate the histograms presented in this work. The

403
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

Fig. 2. Comparisons of experimental ST and RCM data against model predictions using NUIGMech1.1 (solid lines) and AramcoMech3.0 (dashed lines) for (a) a 50% C2 H4 /50%
C3 H8 blend at 75% N2 (black symbols/lines), 75% N2 +10% Ar (red symbols/lines), and 75% N2 +15% Ar (blue symbols/lines), (b) a 70% C2 H4 /30% C3 H8 blend at 75% N2 +10%
Ar (black symbols/lines), 75% N2 +15% Ar (red symbols/lines), and 75% N2 (blue symbols/lines), and (c) a 90% C2 H4 /10% C3 H8 blend at 75% N2 +15% Ar (black symbols/lines),
75% N2 (red symbols/lines), and 75% N2 +10% Ar (blue symbols/lines). (For interpretation of the references to colour in this figure legend, the reader is referred to the web
version of this article.)

mathematical expressions used are the following: inhibiting reactivity (increasing IDT). Furthermore, rate of produc-
1 |IDT mod−IDT exp| tion (ROP) analyses were carried out to track the consumption of
 MAD = (1) the blends and the production of intermediate species.
n
A global correlation discussion based on constant volume (CV)
 
1  | IDT mod −IDT exp
| IDT simulations using NUIGMech1.1 is presented in the “Regres-
 MAPE = IDT exp
∗100 (2)
sion analysis” section (Section 4.5), together with general equa-
n
  tions sorted by various temperature and pressure conditions. The
IDT mod − IDT exp aim of these correlations is to provide an easy and quick way to
ε RPE = ∗ 100 (3)
IDT exp determine the IDT behavior of binary fuels. This does not require
any kind of software pre-set up, and the coefficients of interest can
where n is the total number of experimental measurements. Fur-
be directly substituted in the equations provided in the respective
ther details about the statistical analysis are provided as Supple-
section. A complete table of coefficient values and further details
mentary material.
are provided as Supplementary material.
To identify the reactions controlling IDTs, brute-force sensitivity
analyses were performed at the experimental conditions presented
in this study. The sensitivity coefficient (S; [57]) is defined as: 4. Results and discussions

In(τ +/τ − ) In(τ +/τ − )

S = = All of the experimental results for the ethane/propane
In(k+/k− ) In(2.0/0.5) (C2 H6 /C3 H8 ) and ethylene/propane (C2 H4 /C3 H8 ) blends are pre-
The sensitivity coefficient S calculated using the brute force sented in Section 4.1 together with simulations using NUIGMech1.1
method is based on the IDT (τ ), with the pre-exponential factor and AramcoMech3.0 [52]. Henceforth, in all figures, the open sym-
in the Arrhenius equations for each reaction perturbed in the sen- bols represent experimental LPST and/or HPST data, and the solid
sitivity analysis. The sensitivity coefficient can be negative or pos- symbols represent the experimental low-temperature RCM data.
itive, where a negative value refers to a reaction promoting reac- Sections 4.2 – 4.4 present results for the effects of blend composi-
tivity (decreasing IDT), while a positive value refers to a reaction tion, pressure, and equivalence ratio using NUIGMech1.1 and their

404
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

Fig. 3. Comparisons of experimental ST and RCM data against model predictions using NUIGMech1.1 (solid lines) and AramcoMech3.0 (dashed lines), for (a) a 50% C2 H6 /50%
C3 H8 blend at 75% N2 (black symbols/lines), 75% N2 +10% Ar (red symbols/lines), and 75% N2 +15% Ar (blue symbols/lines), (b) a 70% C2 H6 /30% C3 H8 blend at 75% N2 +10%
Ar (black symbols/lines), 75% N2 +15% Ar (red symbols/lines), and 75% N2 (blue symbols/lines), (c) a 90% C2 H6 /10% C3 H8 blend at 75% N2 +15% Ar (black symbols/lines),
75% N2 (red symbols/lines), and 75% N2 +10% Ar (blue symbols/lines) and (d) a 90% C2 H6 /10% C3 H8 blend at 45.2% N2 +45.2% Ar (magenta symbols/lines), 76.8% N2 (green
symbols/lines), and 65.4% N2 +25% Ar (orange symbols/lines). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of
this article.)

corresponding correlations. Finally, Section 4.5 discusses the corre- of MAPE calculated over the entire dataset using NUIGMech1.1
lation performance. was 26.4%, while that for AramcoMech3.0 is 31.9%, indicating the
greater accuracy of NUIGMech1.1. As it can accurately predict the
IDT data measured over a wide range of temperatures, pressures
4.1. Ethylene/propane and ethane/propane blends
and equivalence ratios, CV simulations are performed using NUIG-
Mech1.1 to understand the effects of these operating conditions on
Figures 2 and 3 present experimental data and model pre-
the IDTs of the pure fuels and their binary blends.
dictions of IDTs over the range of conditions studied for the bi-
nary C2 H4 /C3 H8 and C2 H6 /C3 H8 blends. Figures 2 and 3 show that
NUIGMech1.1 is in better agreement than AramcoMech3.0 with the 4.2. Synergistic/antagonistic effect of blends
experimental data. Statistical analyses were conducted using the
IDTs from the experiments, and those calculated using both NUIG- First, the ignition behavior of the pure fuels is analyzed to
Mech1.1 and AramcoMech3.0. A total sample of 328 IDTs was used determine the important reactions controlling the reactivity. In
to determine the mean, standard deviation (σ ), mean absolute de- Fig. 4, the IDT predictions for C2 H4 /air, C2 H6 /air, and C3 H8 /air mix-
viation (MAD), relative percentage error (RPE), and mean absolute tures at fuel-lean conditions, at pC = 40 bar and TC in the range
percentage error (MAPE). Figure S10(a) and S10(b), with “S” no- 740 − 1660 K are shown. At lower temperatures (TC < 1050 K),
tation referring to the Supplementary material, provide the RPE C3 H8 is the fastest fuel to ignite, however, the trend reverses at
frequency distribution for NUIGMech1.1 and AramcoMech3.0 rela- higher temperatures, and propane is the slowest to react compared
tive to the IDT experiments. It can be inferred that the differences to both C2 H4 and C2 H6 . The reactivity of C2 H4 is observed to be
between NUIGMech1.1 and AramcoMech3.0 are a consequence of higher than C2 H6 at all temperatures studied here.
the poor predictions of AramcoMech3.0 in the low-temperature To explore the controlling chemistry at high-temperature con-
regime for the C2 H4 /C3 H8 blends. Furthermore, the absolute value ditions, ROP analyses for C2 H4 /air, C2 H6 /air, and C3 H8 /air mixtures

405
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

radical (ĊH2 CHO) through the chain branching reaction Ċ2 H3 + O2

↔ ĊH2 CHO + Ö. Oxygen atoms further react with ethylene greatly
promoting reactivity by generating Ḣ atoms through two different
channels, C2 H4 + Ö ↔ ĊH2 CHO + Ḣ (18.1%) and C2 H4 + Ö ↔
ĊH2 + CH2 O (16%) followed by ĊH2 + O2 ↔ CO2 + Ḣ + Ḣ. For
the reaction of oxygen atoms with ethylene the total rate constant
and the branching fractions through the various product channels
(ĊH3 + HĊO, ĊH2 CHO + Ḣ, ĊH2 + CH2 O, CH2 CO + H2 ) are taken
from the calculations by Li et al. [58]. These are in good agree-
ment available experimental data, as shown in Fig. 6(a). Figure 6(b)
compares the rate constants for the individual pathways associated
with the C2 H4 + Ö system. AramcoMech3.0 used the rate con-
stants for C2 H4 + Ö producing ĊH3 + HĊO and ĊH2 CHO + Ḣ based
on the Baulch et al. [59] recommendation. The pathway producing
ĊH2 + CH2 O was not included in AramcoMech3.0, and its inclusion
in NUIGMech1.1 significantly increases the predicted reactivity. The
effect of updating the rate constant for the C2 H4 + Ö → products
reactions on IDT predictions for C2 H4 /air mixtures is shown in Fig.
S19 of the Supplementary material. The ĊH2 CHO radical formed
here further decomposes to produce ketene and Ḣ atom, Fig. 5(a).
Fig. 4. IDT predictions of pure fuels, 90% C2 H4 /10% C3 H8 and 90% C2 H6 /10% C3 H8
The formation of substantial concentrations of Ḣ atoms is respon-
binary blend in air. The corresponding derived correlation predictions are marked sible for the faster ignition of C2 H4 /air mixtures at higher temper-
as dotted lines for pure fuels and dotted-dashed for binary blends. atures.
Ethylene is an important intermediate formed during C2 H6 ox-
idation. At 1430 K, C2 H6 consumption is initiated by H-atom ab-
are illustrated in Fig. 5 at TC = 1430 K and pC = 40 bar. The ROP straction primarily by ȮH and Ḣ forming ethyl (Ċ2 H5 ) radicals.
analyses are performed following an elemental carbon (C) balance. There has been a wide variety of experimental investigations for
The percentage value above the arrow refers to the percentage of these rate constants, as shown in Fig. S13. NUIGMech1.1 has an
the fuel proceeding through that pathway. The reaction paths rep- updated rate constant for H-atom abstraction by ȮH based on the
resent the promoting (red color) and inhibiting (blue color) chan- fit recommended by Krasnoperov and Michael [69]. For H-atom
nels of the corresponding fuels. At high temperatures, the reactiv- abstraction by Ḣ atoms, we have adopted the theoretical calcula-
ity of all fuels is governed by the dominating chain branching re- tions from Sivaramakrishnan et al. [70]. Ċ2 H5 radicals decompose
action Ḣ + O2 ↔ Ö + ȮH, which depends on the concentrations promptly to C2 H4 and Ḣ atoms, which undergo chain branching
of Ḣ atoms and O2 . In the case of C2 H4 /air ignition, at 1430 K, the by reacting with O2 via Ḣ + O2 ↔ Ö + ȮH, promoting reactiv-
fuel mainly undergoes H-atom abstraction by ȮH and Ḣ producing ity. However, at 1430 K, approximately 15% of Ċ2 H5 radicals re-
vinyl (Ċ2 H3 ) radicals. This radical reacts with O2 to generate vinoxy act with O2 to form C2 H4 via H-atom abstraction that competes

Fig. 5. Flux analyses of pure (a) C2 H4 , (b) C2 H6 , and (c) C3 H8 fuel ignition for TC = 1430 K, pC = 40 bar and ϕ = 0.5, at the time of 15% fuel consumption.

406
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

Fig. 6. Comparisons for experimental and theoretical determinations for (a) the total reaction rate constant of C2 H4 + Ö [58, 60-68] and (b) product pathways for the
reaction C2 H4 + Ö.

with Ċ2 H5 radical decomposition. The subsequent reaction path- tively, at 790 K. To interpret the influence of C3 H8 addition on the
ways associated with the C2 H6 consumption flux are governed by ignition of the C2 H4 /air and C2 H6 /air mixtures, sensitivity analy-
the high-temperature chemistry of C2 H4, as discussed in the pre- ses were performed at 790 K, Fig. 7. Moreover, Figure 8 illustrates
vious paragraph. At higher temperature conditions, Ċ2 H5 + O2 ↔ the flux analyses performed for these mixtures in the same con-
C2 H4 + HȮ2 , as well as the H-atom abstraction by Ḣ from the fuel dition. The black color represents the flux for the pure C2 H4 /air
which competes with the major chain branching reaction Ḣ + O2 or C2 H6 /air mixtures, and the red color represents the flux for
↔ Ö + ȮH, are responsible for the lower reactivity of C2 H6 com- the C3 H8 blended binary mixtures. The flux analysis presented in
pared to C2 H4 . Fig. 8 shows that adding propane to the mixture does not alter the
Similar to ethane and ethylene, propane oxidation is mainly reaction pathways of ethylene and ethane chemistry nor does it
driven by H-atom abstraction by ȮH radicals and Ḣ atoms, generat- significantly affect their flux values.
ing primary (nĊ3 H7 ) and secondary propyl (iĊ3 H7 ) radicals. Due to At 790 K, for both pure C2 H4 and the 90% C2 H4 /10% C3 H8 blend,
the importance of H-atom abstraction by ȮH from propane, there ethylene is primarily consumed by the addition of ȮH radical to
have been a large number of measurements performed, Fig. S14. the double bond forming hydroxyethyl radicals, which accounts for
The rate constants adopted in this work are best fits from the more approximately 70% of the overall C2 H4 consumption. These radi-
recent direct measurements by Sivaramakrishnan et al. [71], who cals add to molecular oxygen producing hydroxyethyl-peroxy rad-
investigated the branching fraction for the abstraction of the sec- icals (Ȯ2 C2 H4 OH), which subsequently decompose, producing two
ondary C–H bond in the temperature range 927 – 1146 K, together formaldehyde molecules and an ȮH radical or form vinyl alcohol
with the measurement by Droege and Tully [72] over the temper- and HȮ2 radicals, the former being the most favorable product
ature range 298–900 K (Fig. S14). At 1430 K, approximately 15% of channel promoting reactivity for the C2 H4 /air mixture, Fig. 8(a).
the C3 H8 is consumed by unimolecular decomposition producing Besides ȮH addition, HȮ2 addition to ethylene producing oxirane
Ċ2 H5 and methyl (ĊH3 ) radicals, Fig. 5(c). Substantial concentra- (C2 H4 O1–2) and an ȮH radical and this reaction also has a large
tions of ĊH3 radicals are also formed from the β -scission of nĊ3 H7 promoting effect on the reactivity of ethylene at low temperatures,
radicals. Methyl radicals are consumed by reaction with HȮ2 to especially for fuel-rich conditions.
produce methoxy radicals through the chain branching reaction The importance of the C2 H4 + HȮ2 ↔ C2 H4 O1–2 + ȮH and
ĊH3 + HȮ2 ↔ ĊH3 O + ȮH, which promotes reactivity. The route Ȯ2 C2 H4 OH → products reaction systems on C2 H4 oxidation is
through the chain-terminating reaction ĊH3 + HȮ2 ↔ CH4 + O2 shown in Fig. 9(a), which also presents the performance of the
inhibits reactivity. This competition between chain branching and current mechanism and AramcoMech3.0 as can be seen by the
termination significantly influences IDT predictions for C3 H8 . The red solid lines and black dashed lines, respectively for the 90%
rate constants for these reactions are taken from the theoretical C2 H6 /10% C3 H8 mixtures at pC = 20 atm, and ϕ = 2.0. Aram-
calculations of Jasper et al. [73] and Zhu and Lin [74] respectively. coMech3.0 severely under-predicts the IDTs, particularly in the low
The rate constants and the branching ratio of the two ĊH3 + HȮ2 temperature region in the range 80 0–90 0 K. AramcoMech3.0 im-
channels agree well with the most recent experimental measure- plemented a reaction rate for C2 H4 + HȮ2 ↔ C2 H4 O1–2 + ȮH
ments by Hong et al. [75] (Fig. S15). The self-recombination of ĊH3 based on the recommendation by Zsély et al. [11]. Recently Zá-
radicals producing C2 H6 further contributes to a reduction in the dor et al. [77] and Klippenstein [76] studied the potential en-
reactivity of propane. The presence of high concentrations of ĊH3 ergy surfaces of the C2 H5 Ȯ2 system using high-level quantum
radicals ultimately decreases the reactivity of C3 H8 compared to chemistry calculations. NUIGMech1.1 has adopted the rate con-
C2 H6 at high-temperature conditions. stant for C2 H4 + HȮ2 from Klippenstein [76], which is approx-
The effects on IDTs of the addition of C3 H8 to C2 H4 /air and imately a factor of three lower than the rate constant recom-
C2 H6 /air mixtures are presented in Fig. 4. The reactivities of the mended by Zsély et al. [11] at 800 K, Fig. 9(b). Updating this rate
mixtures increase significantly for the 90% C2 H4 /10% C3 H8 and 90% constant in AramocMech3.0 leads to a significant improvement in
C2 H6 /10% C3 H8 binary blends at lower temperatures in the range IDT predictions as depicted by the dashed-dotted line in Fig. 9(a).
740–10 0 0 K. The addition of only 10% C3 H8 to the C2 H4 /air and Another important reaction pathway controlling ethylene IDT is
C2 H6 /air mixtures shortens IDTs by a factor of 2.8 and 2.0 respec- the consumption of Ȯ2 C2 H4 OH radicals through the Waddington

407
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

Fig. 7. Sensitivity analyses to IDT at 790 K, pC = 40 atm, ϕ = 0.5, for (a) C2 H4 and 90% C2 H4 /10% C3 H8 , in air, and (b) C2 H6 and 90% C2 H6 /10% C3 H8 in air.

Fig. 8. Flux analyses for (a) pure C2 H4 (black) and 90% C2 H4 /10% C3 H8 (red), (b) pure C2 H6 (black) and 90% C2 H6 /10% C3 H8 (red) mixtures ignition for 790 K and at pC = 40
atm, and ϕ = 0.5. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

[79] mechanism Ȯ2 C2 H4 OH → 2CH2 O + ȮH and the HȮ2 elimina- hyde, CO, and Ḣ atoms. Some Ċ2 H3 radicals also produce ĊH2 CHO
tion channel producing C2 H3 OH which inhibits reactivity. In NUIG- and Ö atoms increasing the reactivity of ethylene ignition, as
Mech1.1 the rate constant for the dissociation of Ȯ2 C2 H4 OH radi- shown in Fig. 8(a). For the C2 H6 /air mixture, the fuel is mainly
cals is adopted from Zádor et al. [78]. AramcoMech3.0 utilized an consumed by H-atom abstraction by ȮH and HȮ2 radicals form-
estimated rate constant for the Waddington pathway that is an or- ing Ċ2 H5 radicals. These react with O2 to produce ethylperoxy
der of magnitude higher than the rate determined by Zádor et al. (C2 H5 Ȯ2 ) radicals, which subsequently decompose to C2 H4 and
[78], while surprisingly, the HȮ2 elimination channel was not in- HȮ2 radicals. Figure 7(b) shows that the concerted elimination re-
cluded in the mechanism. The last agreement represented by the action C2 H5 Ȯ2 ↔ C2 H4 + HȮ2 is important in inhibiting the au-
solid black line in Fig. 9(a) is attained by updating both C2 H4 +HȮ2 toignition of C2 H6 .
and dissociations of Ȯ2 C2 H4 OH reactions in AramcoMech3.0 that Figure 7(a) and 7(b) shows that C3 H8 specific reactions become
leads to significant improvement in the agreement of the simula- important when propane is added to the C2 H4 /air and C2 H6 /air
tions compared to the experimental measurements. mixtures. At 790 K, H-atom abstraction from C3 H8 by ȮH produc-
As seen in Fig. 8(a), ȮH radicals can abstract hydrogen atoms ing nĊ3 H7 and H2 O is the most sensitive reaction promoting re-
from ethylene producing Ċ2 H3 radicals. These add to O2 generating activity, while abstraction leading to iĊ3 H7 radicals inhibits reac-
vinyl-peroxy radicals, which subsequently dissociate to formalde- tivity. At 790 K, ~14.8% (C2 H4 /C3 H8 blend) and ~11.9% (C2 H6 /C3 H8

408
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

Fig. 9. (a) Effect of changing the rate constant for C2 H4 + HȮ2 ↔ C2 H4 O1–2 + ȮH and Ȯ2 C2 H4 OH → products on IDT predictions for 90%C2 H4 /10%C3 H8 mixtures, –––
NUIGMech1.1, –––––– AramcoMech3.0, –– AramcoMech3.0 plus updated rate constant [76] for C2 H4 + HȮ2 ↔ C2 H4 O1–2 + ȮH, ––– AramcoMech3.0 plus updated rate
constant for C2 H4 + HȮ2 ↔ C2 H4 O1–2 + ȮH and Ȯ2 C2 H4 OH → products [77]. (b) Comparison of current rate constant [75] for C2 H4 + HȮ2 ↔ C2 H4 O1–2 + ȮH against the
study by Zsély et al. [11] and Zádor et al. [78].

Fig. 10. Effect of pressure for (a) 50% C2 H4 /50% C3 H8 (solid lines) binary blend and pure ethylene (dashed lines), (b) 50% C2 H6 /50% C3 H8 (solid lines) binary blend and pure
ethane (dashed lines). The derived correlation predictions for binary blends are represented by dotted lines.

blend) of iĊ3 H7 radicals react with O2 to form C3 H6 and HȮ2 radi- temperature conditions. The self-recombination of methyl radicals
cals, which reduces reactivity. However, ~38.3% (C2 H4 /C3 H8 blend) is responsible for the lower reactivity of the propane blended mix-
and ~37.8% (C2 H6 /C3 H8 blend) of nĊ3 H7 radicals add to O2 form- tures as discussed in Section 4.2. In the case of the C2 H4 /C3 H8
ing n-propyl-peroxy (nĊ3 H7 O2 ) radicals which undergo isomeriza- blend at 1250 K, as the pressure decreases to 1 atm the ĊH3 + ĊH3
tion generating hydroperoxyl-propyl (Ċ3 H6 OOH1–3) radicals. These (+M) ↔ C2 H6 (+M) reaction more strongly inhibits reactivity, ac-
then further add to O2 producing hydroperoxyl-propyl peroxy rad- counting for 35% of the total flux through methyl radicals, while at
icals (Ċ3 H6 OOH1–3O2 ), which isomerize to produce carbonylhy- 40 bar this reaction contributes only 12% to ĊH3 consumption. Fur-
droperoxides and ȮH radicals. The carbonylhydroperoxide can fur- thermore, from Fig. 10 it is observed that the overall reactivity of
ther dissociate, producing a carbonyl-alkoxy radical and a second the system increases with pressure due to the corresponding in-
ȮH radical, which is a chain branching pathway, resulting in higher crease in the concentration of the reactants. At 800 K, upon in-
reactivity of the C3 H8 blended mixtures compared to the pure creasing the pressure from 1 to 20 atm, there is an order of mag-
C2 H4 /air or C2 H6 /air mixtures. nitude increase in reactivity observed for both the C2 H4 /C3 H8 and
C2 H6 /C3 H8 mixtures. There is approximately a four-fold increase in
reactivity by increasing the pressure from 20 to 40 atm. To deter-
4.3. Effect of pressure on ignition
mine the reactions controlling IDT predictions at these conditions,
sensitivity analyses are presented in Fig. 11 and Fig. S16 for the
Figure 10 presents the influence of pressure on the IDTs for the
binary mixtures at pC = 1, 20, and 40 atm.
50% C2 H4 /50% C3 H8 and 50% C2 H6 /50% C3 H8 binary blended mix-
Figure 11 shows that at low temperature (800 K) and high-
tures as well as for pure C2 H4 and C2 H6 at ϕ = 0.5 and 75% N2
pressure conditions (20 and 40 atm), the reactivity of the binary
dilution. The model predicts that the reduction in reactivity due to
blends is mainly controlled by H-atom abstraction from C3 H8 by
the addition of C3 H8 with C2 H4 at 1 atm is more than for the cor-
ȮH radicals, with the formation of nĊ3 H7 radicals promoting reac-
responding 20 atm and 40 atm cases at intermediate and higher

409
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

Fig. 11. Sensitivity analyses to IDT predictions as function of pressure at ϕ = 0.5, 50% C2 H4 /50% C3 H8. (a) 80 0 K and (b) 160 0 K.

Fig. 12. Flux analyses at 800 K, ϕ = 0.5, p = 1 (black), and 40 atm (red), with 75% N2 as diluent for (a) 50% C2 H4 /50% C3 H8 , and (b) 50% C2 H6 /50% C3 H8 . (For interpretation
of the references to colour in this figure legend, the reader is referred to the web version of this article.)

tivity and iĊ3 H7 radicals inhibiting reactivity. However, at 800 K of pressure, as seen in Figs. 11(b) and S16. The reactions that con-
and 1 atm, H-atom abstraction from C3 H8 no longer influences sume Ḣ atoms such as, C3 H8 + Ḣ ↔ nĊ3 H7 +H2 , C3 H8 + Ḣ ↔
IDT predictions, but rather the competition between the reactions iĊ3 H7 +H2 , C2 H6 + Ḣ ↔ Ċ2 H5 +H2 , CH2 O + Ḣ ↔ HĊO + H2 and
generating and consuming hydrogen peroxide, via HȮ2 + HȮ2 ↔ Ḣ + O2 (+M) ↔ HȮ2 (+M) compete with Ḣ + O2 ↔ Ӧ + ȮH and
H2 O2 + O2 and H2 O2 (+M) ↔ ȮH + ȮH (+M), respectively control thus inhibit the reactivity of the binary mixtures (Fig. 11).
the reactivity of the binary blends. The flux analyses presented in
Figure 12 show that, at 40 atm pressure, the percentage contribu- 4.4. Effect of equivalence ratio on ignition
tion of nĊ3 H7 radical β -scission forming C2 H4 and ĊH3 reduces,
while the importance of nĊ3 H7 radical addition to O2 increases Figure 13 presents the effect of equivalence ratio on IDTs for
compared to the 1 atm case. Since the addition of nĊ3 H7 radi- the pure fuels, 50% C2 H4 /50% C3 H8 and 50% C2 H6 /50% C3 H8 binary
cals to O2 and the subsequent chain branching channels produces mixtures at pC = 20 atm, 75% N2 , and at ϕ = 0.5, 1.0, and 2.0. It
two reactive ȮH radicals and thus increases reactivity, the forma- is observed that at temperatures above 1250 K, the reactivities of
tion of nĊ3 H7 radicals, and other low-temperature reactions those both the pure fuels and the binary mixtures are fastest for the fuel-
are not favorable at low pressures become significant in controlling lean mixtures and slowest for the fuel-rich mixtures. However, at a
the overall reactivity of the binary mixtures at higher pressures . At temperature below 1250 K, fuel-rich mixtures are fastest to ignite,
the higher temperature of 1600 K, the reactivity is only controlled and the fuel-lean mixtures are slowest. To determine the governing
by the chain branching reaction, Ḣ + O2 ↔ Ö + ȮH irrespective chemistry under these conditions, sensitivity analyses were per-

410
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

Fig. 13. Effect of equivalence ratio in (a) 50% C2 H4 /50% C3 H8 (solid lines) binary blend and pure ethylene (dashed lines), (b) 50% C2 H6 /50% C3 H8 (solid lines) binary blend
and pure ethane (dashed lines). The derived correlation predictions for binary blends are represented by dotted lines.

formed, the results of which are presented as a function of equiva- ,204,65

0.113 0.413 −1.344 0.131
lence ratio in Figs. S17 and S18 at 800 K and 1600 K. At high tem- τ corr = 10−9.402 e ,T,C
[C 2 H 6 ] [C 3 H 8 ] [oxidizer] [diluent]
peratures (> 1250 K), IDTs are mainly controlled by the concentra- (6)
tion of O2 in the binary blends through the main chain branching
reaction Ḣ + O2 ↔ Ӧ + ȮH, and its influence increases as the mix- For 1100 ≤ TC ≤ 1500 K:
ture become leaner. Thus, fuel-lean mixtures are fastest to ignite at
high temperatures. However, at low temperatures (< 1250 K), the
reactivity is mainly governed by the addition of the fuel derived τ corr
,192,20.,37
−0.491 −0.056 −0.447 0.149
alkyl radicals to O2 and the following low-temperature chemistry = 10−9.89 e ,T,C
[C 2 H 4 ] [C 3 H 8 ] [oxidizer] [diluent]
leading to chain branching, which is limited by the fuel concen- (7)
tration through H-atom abstraction from propane by ȮH radicals.
Thus, at a lower temperature, the dependence on the equivalence
τ corr
ratio is reversed, with fuel-rich mixtures being the most reactive. ,190,65.,65
−0.408 −0.169 −0.330 0.062
= 10−9.79 e ,T,C
[C 2 H 6 ] [C 3 H 8 ] [oxidizer] [diluent]
4.5. Regression analysis (8)

Global regression equations have been developed using NUIG- For 800 ≤ TC ≤ 1100 K:
Mech1.1 with approximately 17,280 CV simulations for each
blend mixture spanning over five parameters: pC (1 – 40 atm),
τ corr
TC (80 0 – 20 0 0 K), ϕ (0.2 – 2.0), dilution (75% – 90%) and fuel , 14,,136 ,.9
−0.44 −0.392 −0.427 −0.017
ratio composition (50% C2 H4 or C2 H6 /50% C3 H8 , and 70% C2 H4 = 10−7.217 e T,C [C2 H4 ] [C 3 H 8 ] [oxidizer] [diluent]
or C2 H6 /30% C3 H8 ). The regression equations developed using the (9)
predictions are compared with the ST experimental data in Fig.
S20 and S21 of the Supplementary material. The expression (τ corr ) τ corr
used is analogous to the Arrhenius rate expression and is defined , 18,,630 ,.7
−0.145 −0.555 −0.186 −0.145
= 10−9.52 e T,C [C 2 H 6 ] [C3 H8 ] [oxidizer] [diluent]
as shown in Eq. (4) below:
(10)
B
τ corr = 10 e [C2 H4 ] [C2 H6 ] [C3 H8 ] [oxidizer] [diluent]
A TC C D E F G
(4) At high temperatures (150 0–20 0 0 K), the coefficient for ethy-
where A represents the pre-exponential factor coefficient, B repre- lene is strongly negative, while those for ethane and propane are
sents the activation energy, and C – G are the concentration depen- positive. This is because at high temperatures increasing the ethy-
dencies of ethylene, ethane, propane, oxidizer, and dilution, respec- lene concentration increases the concentration of vinyl radicals,
tively. A synopsis of the derived correlations for the binary fuels which react with O2 (Ċ2 H3 + O2 ↔ ĊH2 CHO + Ö) in a reaction
studied in the pressure range 20 ≤ pC ≤ 40 atm over three regimes which is chain branching. On the other hand, both ethane and
of temperature is presented below. However, details of the coeffi- propane compete with O2 for Ḣ atoms, and if their concentrations
cients of the derived correlations along with their corresponding increase, the rate of Ḣ + O2 ↔ Ö + ȮH decreases, reducing reactiv-
χ 2 and R2 for the C2 H4 /C3 H8 and C2 H6 /C3 H8 mixtures are pre- ity. For ethane/propane mixtures, both coefficients are positive, but
sented in Table S6 and S7, respectively. it is less positive for ethane than for propane, as the oxidation of
ethane leads to higher concentrations of ethylene. Thus, increasing
For 1500 ≤ TC ≤ 2000 K: the concentration of ethane relative to propane will increase reac-
tivity and vice versa.
At low temperatures (80 0–110 0 K), the coefficients associated
τ corr
, 21,,386 ,.6 with ethylene, ethane, and propane become negative, with ethane
−0.502 0.463 −1.080 0.354
= 10−10.34 e T,C [C 2 H 4 ] [C3 H8 ] [oxidizer] [diluent] being less negative than propane. At low temperatures, propane
(5) promotes reactivity through the addition of n-propyl radicals to

411
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

O2 that proceeds to chain branching through the low-temperature Acknowledgments

reaction sequence that generates two highly active ȮH radicals.
Thus, for the ethane/propane mixtures, increasing the propane The authors would like to express their gratitude to Sci-
concentration will increase mixture reactivity. However, for ethy- ence Foundation Ireland (SFI) via their Research Centre Program
lene/propane mixtures, the coefficient of ethylene is comparable to through project numbers 15/IA/3177 and 16/SP/3829, KAY-ICHEC
that of propane, as ethylene and propane exhibit similar reactivi- via the project ngche079c, and to Shell Research Ltd. The au-
ties in the temperature range between 900 and 1100 K, as seen in thors from PCFC RWTH Aachen University would like to recog-
Fig. 4. nize the funding support from the German Research Foundation
It is interesting to note that, at high temperatures Eqs. (5) and (Deutsche Forschungsgemeinschaft, DFG) through the project num-
((6)), the magnitude of the oxidizer coefficients (–1.08 and –1.344 ber –322460823 (HE7599/2–1).
for C2 H4 /C3 H8 and C2 H6 /C3 H8 blends, respectively) are signifi-
cantly larger, by a factor of two or more, than for the correspond-
ing fuel coefficients, thus showing a higher sensitivity towards oxi- Supplementary materials
dizer concentrations at these conditions. As we approach the lower
temperature regimes Eqs. (9) and ((10)), the coefficients associated Supplementary material associated with this article can be
with each fuel become higher and are even larger than the cor- found, in the online version, at doi:10.1016/j.combustflame.2021.02.
responding oxidizer coefficients, thus representing the increasing 009.
importance of fuel-based kinetics. These characteristics observed in
the correlations corroborate the discussion of the underlying kinet-
ics in Section 4 above and enable the correlations to capture the References
IDT trends effectively.
[1] U.S.E.I. Administration, International energy outlook 2019 with projections to
2050, IEO2019, 2019.
5. Conclusions [2] S. Bilgen, Structure and environmental impact of global energy consumption,
Renew. Sustain. Energy Rev. 38 (2014) 890–902.
An experimental and kinetic modeling study of the IDT char- [3] 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.
acteristics of C2 – C3 binary blends of C2 H4 /C3 H8 and C2 H6 /C3 H8
J. Chem. Kinet. 45 (2013) 638–675.
mixtures over a wide range of experimental conditions, pressures [4] S.S. Nagaraja, J. Liang, S. Dong, S. Panigrahy, A. Sahu, G. Kukkadapu,
(1–135 atm), temperatures (~750–20 0 0 K), equivalence ratios (0.5 S.W. Wagnon, W.J. Pitz, H.J. Curran, A hierarchical single-pulse shock tube py-
≤ ϕ ≤ 2.0) and 75–90% of dilution percentage were presented. The rolysis study of C2–C6 1-alkenes, Combust. Flame 219 (2020) 456–466.
[5] M. Baigmohammadi, V. Patel, S. Martinez, S. Panigrahy, A. Ramalingam,
performance of NUIGMech1.1 and its corresponding derived corre- U. Burke, K.P. Somers, K.A. Heufer, A. Pekalski, H.J. Curran, A comprehen-
lations were evaluated against the experimental data collected. Re- sive experimental and simulation study of ignition delay time characteris-
sults show that NUIGMech1.1 is in good agreement within ~26.4% tics of single fuel C1−C2 hydrocarbons over a wide range of temperatures,
pressures, equivalence ratios, and dilutions, Energy Fuels 34 (3) (2020) 3755–
of model uncertainty to the measured IDTs over the studied con- 3771.
ditions, compared to ~35% for AramcoMech3.0. Moreover, correla- [6] M. Baigmohammadi, V. Patel, S. Nagaraja, A. Ramalingam, S. Martinez, S. Pani-
tions can predict the experimental IDTs appropriately under spe- grahy, A.A.E. Mohamed, K.P. Somers, U. Burke, K.A. Heufer, A. Pekalski, H.J. Cur-
ran, Comprehensive experimental and simulation study of the ignition delay
cific regimes, becoming a useful tool in predicting the behavior of time characteristics of binary blended methane, ethane, and ethylene over a
C2 – C3 binary blends at specific conditions. wide range of temperature, pressure, equivalence ratio, and dilution, Energy
Finally, the effects of blend composition, pressure and equiva- Fuels 34 (7) (2020) 8808–8823.
[7] M.M. Kopp, N.S. Donato, E.L. Petersen, W.K. Metcalfe, S.M. Burke, H.J. Curran,
lence ratio on IDTs were investigated for various mixtures contain- Oxidation of ethylene–air mixtures at elevated pressures, part 1: experimental
ing C2 H4 , C2 H6, and C3 H8 as reactants. It was observed that for all results, J. Propuls. Power 30 (2014) 790–798.
mixtures, as the temperature and pressure increase, IDTs decrease. [8] M.M. Kopp, E.L. Petersen, W.K. Metcalfe, S.M. Burke, H.J. Curran, Oxidation
of ethylene-air mixtures at elevated pressures, part 2: chemical kinetics, J.
For the effect of equivalence ratio, an expected crossover point
Propuls. Power 30 (2014) 799–811.
was observed at TC ~ 1250 K, wherein the controlling chemistry [9] C.J. Aul, W.K. Metcalfe, S.M. Burke, H.J. Curran, E.L. Petersen, Ignition and ki-
switches such that the trends invert their behavior. The reactivity netic modeling of methane and ethane fuel blends with oxygen: a design of
experiments approach, Combust. Flame 160 (2013) 1153–1167.
of C2 H4 was found to be higher than for C2 H6 throughout the tem-
[10] S.M. Gallagher, H.J. Curran, W.K. Metcalfe, D. Healy, J.M. Simmie, G. Bourque, A
perature range examined in this study. At higher temperatures, vi- rapid compression machine study of the oxidation of propane in the negative
noxy radicals and oxygen atoms formed from the reaction of vinyl temperature coefficient regime, Combust. Flame 153 (2008) 316–333.
radicals with O2 , proceed via dissociation and bimolecular reac- [11] I.G. Zsély, T. Nagy, J.M. Simmie, H.J. Curran, Reduction of a detailed kinetic
model for the ignition of methane/propane mixtures at gas turbine conditions
tions with C2 H4 , to produce a substantial amount of Ḣ atoms re- using simulation error minimization methods, Combust. Flame 158 (2011)
sulting in faster ignition of C2 H4 /air mixtures. The concerted elimi- 1469–1479.
nation reaction between Ċ2 H5 and O2 is responsible for the reduc- [12] A. Burcat, R.W. Crossley, K. Scheller, Shock tube investigation of ignition in
ethane-oxygen-argon mixtures, Symp. (Int.) Combust. 18 (1972) 115–123.
tion in ethane reactivity. It was observed that C3 H8 blended fuels [13] H. Hashemi, J.G. Jacobsen, C.T. Rasmussen, J.M. Christensen, P. Glarborg,
were the fastest to ignite at lower temperatures (< 1250 K), how- S. Gersen, M. van Essen, H.B. Levinsky, S.J. Klippenstein, High-pressure oxida-
ever, the trend is reversed at higher temperatures, and C3 H8 exhib- tion of ethane, Combust. Flame 182 (2017) 150–166.
[14] S. Gersen, A.V. Mokhov, J.H. Darmeveil, H.B. Levinsky, P. Glarborg, Ignition-pro-
ited the slowest reactivity compared to both C2 H4 and C2 H6 at T > moting effect of NO2 on methane, ethane and methane/ethane mixtures in a
1250 K. In the case of C3 H8 , at low temperatures n-propyl radical rapid compression machine, Proc. Combust. Inst. 33 (2011) 433–440.
formation, followed the classical low-temperature chain branching [15] P. Dagaut, J.C.C. Boettner, M. Cathonnet, Ethylene pyrolysis and oxidation: a
kinetic modeling study, Int. J. Chem. Kinet. 22 (1990) 641–664.
pathways via its addition to O2 generate ȮH radicals promoting re-
[16] P. Dagaut, M. Cathonnet, J.C. Boettner, F. Gaillard, Kinetic modeling of propane
activity, while methyl radical recombination and reaction with HȮ2 oxidation, Combust. Sci. Technol. 56 (1987) 23–63.
leading to CH4 and O2 reduces reactivity at higher temperatures. [17] P. Dagaut, M. Cathonnet, J.c.c. Boettner, Kinetics of ethane oxidation, Int. J.
Chem. Kinet. 23 (1991) 437–455.
[18] W. Lowry, J. de Vries, M. Krejci, E. Petersen, Z. Serinyel, W. Metcalfe, H. Cur-
Declaration of Competing Interest ran, G. Bourque, Laminar flame speed measurements and modeling of pure
alkanes and alkane blends at elevated pressures, Proceedings of ASME Turbo
The authors declare that they have no known competing finan- Expo 2010: Power for Land, Sea and Air (2010).
[19] C.J. Brown, G.O. Thomas, Experimental studies of shock-induced ignition and
cial interests or personal relationships that could have appeared to transition to detonation in ethylene and propane mixtures, Symp. (Int.) Com-
influence the work reported in this paper. bust. 117 (1999) 861–870 1999.

412
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

[20] D. Davidson, W. Ren, R. Hanson, Experimental database for development of a [47] A. Ramalingam, S. Panigrahy, Y. Fenard, H. Curran, K.A. Heufer, A chemical ki-
HiFiRE JP-7 surrogate fuel mechanism, Proceedings of the 50th AIAA Aerospace netic perspective on the low-temperature oxidation of propane/propene mix-
Sciences Meeting including the New Horizons Forum and Aerospace Exposition tures through experiments and kinetic analyses, Combust. Flame 223 (2021)
(2012). 361–375.
[21] O.G. Penyazkov, K.L. Sevrouk, V. Tangirala, N. Joshi, High-pressure ethy- [48] Y. Sun, C.W. Zhou, K.P. Somers, H.J. Curran, An ab initio/transition state the-
lene oxidation behind reflected shock waves, Proc. Combust. Inst. 32 (2009) ory study of the reactions of C5 H9 species of relevance to 1,3-pentadiene, Part
2421–2428. I: potential energy surfaces, thermochemistry and high-pressure limiting rate
[22] S. Saxena, M.S.P. Kahandawala, S.S. Sidhu, A shock tube study of ignition delay constants, J. Phys. Chem. A 123 (2019) 9019–9052.
in the combustion of ethylene, Combust. Flame 158 (2011) 1019–1031. [49] Y. Sun, C.-.W. Zhou, K.P. Somers, H.J. Curran, An ab initio/transition state theory
[23] K.S. Alexander Burcat, A. Lifshitz, Shock-tube investigation of comparative ig- study of the reactions of C5 H9 species of relevance to 1,3-pentadiene, Part II:
nition delay times for C1-C5 alkanes, Symp. (Int.) Combust. 16 (1971) 29–33 pressure dependent rate constants and implications for combustion modelling,
1971. J. Phys. Chem. A 124 (2020) 4605–4631.
[24] J. de Vries, J.M. Hall, S.L. Simmons, M.J.A. Rickard, D.M. Kalitan, E.L. Petersen, [50] J. Power, K.P. Somers, C.-.W. Zhou, S. Peukert, H.J. Curran, A theoretical, exper-
Ethane ignition and oxidation behind reflected shock waves, Combust. Flame imental and modeling study of the reaction of hydrogen atoms with 1- and
150 (2007) 137–150. 2-Pentene, J. Phys. Chem. A 123 (2019) 8506–8526.
[25] Y. Hidaka, K. Sato, H. Hoshikawa, T. Nishimori, R. Takahashi, H. Tanaka, K. In- [51] N. Lokachari, S. Panigrahy, G. Kukkadapu, G. Kim, S.S. Vasu, W.J. Pitz, H.J. Cur-
ami, N. Ito, Shock-tube and modeling study of ethane pyrolysis and oxidation, ran, The influence of iso-butene kinetics on the reactivity of di-isobutylene and
Combust. Flame 120 (20 0 0) 245–264 20 0 0. iso-octane, Combust. Flame 222 (2020) 186–195.
[26] A.G. McLain, C.J. Jachimowski, Chemical kinetic modeling of propane oxidation [52] C.W. Zhou, Y. Li, U. Burke, C. Banyon, K.P. Somers, S. Ding, S. Khan, J.W. Hargis,
behind shock waves, National Aeronautics and Space Administration, 1977. T. Sikes, O. Mathieu, E.L. Petersen, M. AlAbbad, A. Farooq, Y. Pan, Y. Zhang,
[27] P. Cadman, G.O. Thomas, P. Butler, The auto-ignition of propane at interme- Z. Huang, J. Lopez, Z. Loparo, S.S. Vasu, H.J. Curran, An experimental and
diate temperatures and high pressures, Phys. Chem. Chem. Phys. 2 (20 0 0) chemical kinetic modeling study of 1,3-butadiene combustion: ignition de-
5411–5419. lay time and laminar flame speed measurements, Combust. Flame 197 (2018)
[28] Z. Qin, V.V. Lissianski, H. Yang, W.C. Gardiner, S.G. Davis, H. Wang, Combustion 423–438.
chemistry of propane: a case study of detailed reaction mechanism optimiza- [53] D.G. Goodwin, R.L. Speth, H.K. Moffat, B.W. Weber, Cantera: an object-oriented
tion, Proc. Combust. Inst. 28 (20 0 0) 1663–1669. software toolkit for chemical kinetics, thermodynamics, and transport pro-
[29] K.G. Yeong, S.G. Su, Shock tube and modeling study of the ignition of propane, cesses. https://www.cantera.org, doi:10.5281/zenodo.170284.
Bull. Korean Chem. Soc. 22 (2001) 2001vol. 2022, no. 2003. [54] Reaction-design, CHEMKIN-PRO 18.2, San Diego, 2013.
[30] J. Herzler, L. Jerig, P. Roth, Shock-tube study of the ignition of propane at in- [55] S. Chih-Jen, H.J. Curran, Using rapid compression machines for chemical kinet-
termediate temperatures and high pressures, Combust. Sci. Technol. 176 (2004) ics studies, Prog. Energy Combust. Sci. 44 (2014) 1–18.
1627–1637. [56] S.S. Goldsborough, S. Hochgreb, G. Vanhove, M.S. Wooldridge, H.J. Curran,
[31] E. Hu, Z. Zhang, L. Pan, J. Zhang, Z. Huang, Experimental and modeling study S. Chih-Jen, Advances in rapid compression machine studies of low- and in-
on ignition delay times of dimethyl ether/propane/oxygen/argon mixtures at termediate-temperature autoignition phenomena, Prog. Energy Combust. Sci.
20bar, Energy Fuels 27 (2013) 4007–4013. 63 (2017) 1–78.
[32] C. Lee, S. Vranckx, K.A. Heufer, S.V. Khomik, Y. Uygun, H. Olivier, On the chem- [57] V. Gururajan, F.N. Egolfopoulos, Direct sensitivity analysis for ignition delay
ical kinetics of ethanol oxidation: shock tube, rapid compression machine and times, Combust. Flame 209 (2019) 478–480.
detailed modeling study, Z. Phys. Chem. 226 (1) (2012) 1–28. [58] X. Li, A.W. Jasper, J. Zádor, J.A. Miller, S.J. Klippenstein, Theoretical kinetics of
[33] A. Ramalingam, K. Zhang, A. Dhongde, L. Virnich, H. Sankhla, H. Curran, O + C2 H4 , Proc. Combust. Inst. 36 (2017) 219–227.
A. Heufer, An RCM experimental and modeling study on CH4 and CH4 /C2 H6 [59] D.L. Baulch, C.T. Bowman, C.J. Cobos, R.A. Cox, Th. Just, J.A. Kerr, M.J. Pilling,
oxidation at pressures up to 160bar, Fuel 206 (2017) 325–333. D. Stocker, J. Troe, W. Tsang, R.W. Walker, J. Warnatz, Evaluated kinetic data
[34] L. Brett, J. Macnamara, P. Musch, J.M. Simmie, Brief Communication: simulation for combustion modeling: supplement II, J. Phys. Chem. 34 (2005).
of methane autoignition in a rapid compression machine with creviced pistons, [60] H.P. Upadhyaya, A. Kumar, P.D. Naik, A.V. Sapre, Discharge flow reaction ki-
Combust. Flame 124 (2001) 326–329. netic studies of O(3 P) with chloroethylenes CH2 CCl2 , CHClCCl2 , CCl2 CCl2 , Chem.
[35] E.L. Petersen, M. Röhrig, D.F. Davidson, R.K. Hanson, C.T. Bowman, High-pres- Phys. Lett. 321 (20 0 0) 411–418.
sure methane oxidation behind reflected shock waves, Symp. (Int.) Combust. [61] S.Y. Lee, H.S. Yoo, W.K. Kang, K.H. Jung, Reaction of O(3 P) atoms with
26 (1996) 799–806. CF2 = CXY (X, Y = H, F, C1, Br). Discharge flow-chemiluminescence imaging
[36] P.J. Ross, Taguchi techniques for quality engineering, McGraw-Hill Professional, technique, Chem. Phys. Lett. 257 (1996) 415–420.
New York, 1988. [62] S.E. Paulson, J.J. Orlando, G.S. Tyndall, J.G. Calvert, Rate coefficients for the re-
[37] S.M. Burke, U. Burke, R. Mc Donagh, O. Mathieu, I. Osorio, C. Keesee, A. Mo- actions of O(3 P) with selected biogenic hydrocarbons, Int. J. Chem. Kinet. 27
rones, E.L. Petersen, W. Wang, T.A. DeVerter, M.A. Oehlschlaeger, B. Rhodes, (1995) 997–1008.
R.K. Hanson, D.F. Davidson, B.W. Weber, C.J. Sung, J. Santner, Y. Ju, F.M. Haas, [63] V.D. Knyazev, V.S. Arutyunov, V.I. Vedeneev, The mechanism of O(3 P) atom
F.L. Dryer, E.N. Volkov, E.J.K. Nilsson, A.A. Konnov, M. Alrefae, F. Khaled, A. Fa- reaction with ethylene and other simple olefins, Int. J. Chem. Kinet. 24 (1992)
rooq, P. Dirrenberger, P.A. Glaude, F. Battin-Leclerc, H.J. Curran, An experimen- 545–561.
tal and modeling study of propene oxidation. Part 2: ignition delay time and [64] O. Horie, R. Taege, B. Reimann, N.L. Arthur, P. Potzinger, Kinetics and mecha-
flame speed measurements, Combust. Flame 162 (2015) 296–314. nism of the reactions of O(3 P) with SiH4 , CH3 SiH3 ,(CH3 )2 SiH2 , and (CH3 )3 SiH,
[38] S.M. Burke, W. Metcalfe, O. Herbinet, F. Battin-Leclerc, F.M. Haas, J. Santner, J. Phys. Chem. 95 (11) (1991) 4393–4400.
F.L. Dryer, H.J. Curran, An experimental and modeling study of propene oxi- [65] D.M.V. Fonderie, J. Peeters, The kinetic coefficient of the C2 H4 +O reaction over
dation. Part 1: speciation measurements in jet-stirred and flow reactors, Com- extended pressure and temperature ranges, Physico-chemical behaviour of at-
bust. Flame 161 (2014) 2765–2784. mospheric pollutants, Springer, Dordrecht (1984), pp. 274–282.
[39] U. Burke, K.P. Somers, P. O’Toole, C.M. Zinner, N. Marquet, G. Bourque, E.L. Pe- [66] R.B. Klemm, J.W. Sutherland, M.A. Wickramaaratchi, G. Yarwood, Flash photol-
tersen, W.K. Metcalfe, Z. Serinyel, H.J. Curran, An ignition delay and kinetic ysis-shock tube kinetic study of the reaction of atomic oxygen (3P) with ethy-
modeling study of methane, dimethyl ether, and their mixtures at high pres- lene: 1052K.ltoreq. T.ltoreq.2284K, J. Phys. Chem. 94 (8) (1990) 3354–3357.
sures, Combust. Flame 162 (2015) 315–330. [67] K. Mahmud, P. Marshall, A. Fontijn, A high-temperature photochemistry ki-
[40] D. Healy, H.J. Curran, S. Dooley, J.M. Simmie, D.M. Kalitan, E.L. Petersen, netics study of the reaction of oxygen (3P) atoms with ethylene from 290 to
G. Bourque, Methane/propane mixture oxidation at high pressures and at high, 1510K, J. Phys. Chem. 91 (6) (1987) 1568–1573.
intermediate and low temperatures, Combust. Flame 155 (2008) 451–461. [68] T.L. Nguyen, L. Vereecken, X.J. Hou, M.T. Nguyen, J. Peeters, Potential energy
[41] D. Healy, H.J. Curran, J.M. Simmie, D.M. Kalitan, C.M. Zinner, A.B. Barrett, surfaces, product distributions and thermal rate coefficients of the reaction of
E.L. Petersen, G. Bourque, Methane/ethane/propane mixture oxidation at high O(3 P) with C2 H4 (XAg): a comprehensive theoretical study, J. Phys. Chem. A
pressures and at high, intermediate and low temperatures, Combust. Flame 109 (33) (2005) 7489–7499.
155 (2008) 441–448. [69] L.N. Krasnoperov, J.V. Michael, Shock tube studies using a novel multipass ab-
[42] D. Healy, D.M. Kalitan, C.J. Aul, E.L. Petersen, G. Bourque, H.J. Curran, Oxidation sorption Cell: rate constant results for OH + H2 and OH + C2 H6 , J. Phys. Chem.
of C1−C5 alkane quinternary natural gas mixtures at high pressures, Energy A 108 (2004) 5643–5648.
Fuels 24 (2010) 1521–1528. [70] R. Sivaramakrishnan, J.V. Michael, B. Ruscic, High-temperature rate constants
[43] A. Mohamed, S. Panigrahy, A. Sahu, G. Bourque, H.J. Curran, An experimen- for H/D+ C2 H6 and C3 H8 , Int. J. Chem. Kinet. 44 (3) (2012) 194–205.
tal and modeling study of the auto-ignition of natural gas blends containing [71] R. Sivaramakrishnan, C.F. Goldsmith, S. Peukert, J.V. Michael, Direct measure-
C1–C7 n-alkanes, Proc. Combust. Inst. 38 (2021). ments of channel specific rate constants in OH + C3 H8 illuminates prompt dis-
[44] S. Nagaraja, J. Power, G. Kukkadapu, S. Dong, S..W. Wangon, W.J. Pitz, H.J. Cur- sociations of propyl radicals, Proc. Combust. Inst. 37 (2019) 231–238.
ran, A single pulse shock tube study of pentene isomer pyrolysis, Proc. Com- [72] A.T. Droege, F.P. Tully, Hydrogen-atom abstraction from alkanes by hydroxyl. 3.
bust. Inst. 38 (2021). Propane, J. Phys. Chem. 90 (1986) 1949–1954.
[45] S. Panigrahy, J. Liang, S. Nagaraja, Z. Zuo, G. Kim, T. MacDougall, S..S. Vasu, [73] A.W. Jasper, S.J. Klippenstein, L.B. Harding, Theoretical rate coefficients for the
H.J. Curran, A comprehensive experimental and improved kinetic modeling reaction of methyl radical with hydroperoxyl radical and for methylhydroper-
study on the pyrolysis and oxidation of propyne, Proc. Combust. Inst. 38 oxide decomposition, Proc. Combust. Inst. 32 (2009) 279–286.
(2021). [74] R. Zhu, C. Lin, The CH3 + HO2 Reaction: First-principles prediction of its
[46] S. Dong, K. Zhang, P.K. Senecal, G. Kukkadapu, S.W. Wagnon, S. Barrett, rate constant and product branching probabilities, J. Phys. Chem. A 105 (2001)
N. Lokachari, S. Panigrahy, W..J. Pitz, H.J. Curran, A comparative reactivity study 6243–6248.
of 1-alkene fuels from ethylene to 1-heptene, Proc. Combust. Inst. 38 (2021).

413
S. Martinez, M. Baigmohammadi, V. Patel et al. Combustion and Flame 228 (2021) 401–414

[75] Z. Hong, D.F. Davidson, K.-.Y. Lam, R.K. Hanson, A shock tube study of the rate [78] J. Zador, S.J. Klippenstein, J.A. Miller, Pressure-dependent OH yields in
constants of HO2 and CH3 reactions, Combust. Flame 159 (2012) 3007. alkene+HO2 reactions: a theoretical study, J. Phys. Chem. A 115 (2011)
[76] S.J. Klippenstein, From Theoretical reaction dynamics to chemical modeling of 10218–10225.
combustion, Proc. Combust. Inst. 36 (2017) 77–111. [79] M.S. Stark, D.J. Waddington, Oxidation of propene in the gas phase, Int. J.
[77] J. Zádor, R.X. Fernandes, Y. Georgievskii, G. Meloni, C.A. Taatjes, J.A. Miller, The Chem. Kinet. 27 (2) (1995) 123–151.
reaction of hydroxyethyl radicals with O2 : a theoretical analysis and experi-
mental product study, Proc. Combust. Inst. 32 (2009) 271–277.

414