Fuel 181 (2016) 916–925

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Autoignition of ternary blends for gasoline surrogate at wide

temperature ranges and at elevated pressure: Shock tube measurements
and detailed kinetic modeling
Hua Li, Liang Yu, Xingcai Lu ⇑, Linqi Ouyang, Shuzhou Sun, Zhen Huang
Key Laboratory for Power Machinery of M.O.E, Shanghai Jiao Tong University, Shanghai 200240, PR China

h i g h l i g h t s

Three ternary gasoline surrogates with the same RON95 were constructed.

Ignition delay of four toluene reference fuels was measured in shock tube.

A detailed chemical kinetic model for the ternary gasoline surrogates was established.

The ignition properties and reaction pathways of TRFs were analyzed.

A comparison of ignition characteristics of toluene reference fuels with same RON was made.

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

Article history: Considering the diverse compositions of commercial gasoline, the ignition delay time of toluene reference
Received 31 January 2016 fuels (TRF) composed of isooctane, n-heptane and toluene was studied in a shock tube under the condi-
Received in revised form 27 March 2016 tions of medium to high temperature ranges, different pressures (10–20 bar), and various equivalence
Accepted 4 May 2016
ratios (0.5, 1.0, 1.5 and 2) by reflected waves. To analyze the impacts of the component proportion on
Available online 12 May 2016
the gasoline surrogate combustion process, three different ternary blends, TRF2 (42.8% isooctane/13.7%
n-heptane/43.5% toluene), TRF3 (65% isooctane/10% n-heptane/25% toluene) and TRF4 (87.2% isooc-
Keywords:
tane/6.3% n-heptane/6.5% toluene), with the same Research Octane Number of 95 (RON = 95) were con-
Gasoline surrogate
TRF
structed; TRF1 was the same as Surrogate A in Gauthier et al. (2004). The experimental results showed
Shock tube that there was an obvious negative correlation between the ignition delay time of the toluene reference
Ignition delay fuels and the pressure, temperature and equivalence ratio; notably, the measured data showed a minimal
Detailed chemical kinetics discrepancy of TRF2, TRF3, and TRF4 at pressures of 10 and 20 bar in a stoichiometric ratio. Based on
Curran’s (2002) detailed kinetic model for PRF (primary reference fuel) and Yuan’s (2015) toluene pyrol-
ysis and oxidation model, which updated and integrated the thermodynamic parameter and reaction rate
for some key reactions, a detailed chemical mechanism consisting of 1251 species and 5705 reactions
was established to illustrate the surrogate combustion properties. The model captured the autoignition
behavior of all four ternary gasoline surrogates well in the shock tube experiments, especially at high
pressure and under rich fuel conditions. Furthermore, the sensitivities and a reaction pathway analysis
were also calculated for different blending ratios using CHEMKIN-PRO software, which exhibited and
analyzed partial ignition features with respect to the chemical reactions based on the model proposed
in this work.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction high-efficient and zero-emission internal combustion engine

For the past several decades, increasing numbers of new-type
combustion modes have been at the center of worldwide innova-
tion and research focus, with increasingly urgent requests for
⇑ Corresponding author.
E-mail addresses: lyuxc@sjtu.edu.cn (X. Lu), z-huang@sjtu.edu.cn (Z. Huang).
(ICE), more combustion modes [1–6] with highly efficient, ultra-
low NOx and soot emissions also are in the charge of chemical
kinetics are promoted. To this end, it’s imperative to exactly deter-
mine the detailed chemical mechanism of fuels in ICE operation
conditions.
However, study of the chemical kinetics of multicomponent fuel
is still immature and limited due to the computing resources. As a

http://dx.doi.org/10.1016/j.fuel.2016.05.030
0016-2361/Ó 2016 Elsevier Ltd. All rights reserved.
 H. Li et al. / Fuel 181 (2016) 916–925
result, the construction of a detailed mechanism model for petro-
chemical fuel, which commonly contains thousands of hydrocar-
bons and varies with seasons and places of origin, is not feasible.
Therefore, establishing a gasoline surrogate mechanism model
involving limited components is turning into research emphasis,
in accordance with the carbon distribution and fundamental family
of the chemical constituents of the modeling goals. In addition, not
only physical properties (such as the density, the viscosity and the
evaporation characteristics) but also chemical properties (such as
the octane number/cetane number, the C/H, the flame speed and
the adiabatic flame temperature) should be matched by the surro-
gate. Hence, a series of foundation experimental facilities, such as
ST (shock tubes), RCM (rapid compression machines), JSR (jet-
stirred reactors), FL (flow reactors), combustion bombs and
single-cylinder engines, have been adopted to investigate the com-
bustion features of gasoline surrogates and to validate relevant
chemical kinetic models.
Gasoline is a complex mixture comprising branched paraffin,
olefins, cycloparaffins and aromatics [7,8], which contains 4–13
carbons. Due to its high octane number, isooctane was first
selected to represent real unleaded gasoline as a surrogate. Curran
et al. [9] at the LLNL (Lawrence Livermore National Laboratory)
constituted a detailed chemical kinetic model (858 species and
3606 reactions) for isooctane oxidation, followed by a series of
skeleton and reduced mechanism afterward [10–12]. Even though
isooctane can represent gasoline properties to some extent, it can-
not reproduce the combustion process and specific emission char-
acteristics of commercial gasoline. Increasing attention has been
focused on PRF (primary reference fuel), a binary gasoline surro-
gate that can achieve a variety of gasoline octane number flexibly
by mixing isooctane and n-heptane in a certain ratio. Curran et al.
[13] utilized an ST and a high pressure FL to measure the autoigni-
tion and combustion intermediate concentrations at 690–1220 K,
40 atm and 550–880 K (12.5 atm individually) and founded and
confirmed the detailed kinetics model for PRF (1034 species and
4238 reactions), and Chaos et al. [14] reduced it with only the high
temperature (>950 K) and low pressure condition (<15 atm) there-
with, while Andrae et al. [15] added 132 co-oxidation reactions of
two ingredients to it, updating the model to a new form that could
satisfy the PRF80 (80% isooctane) ignition delay in HCCI combus-
tion conditions. Moreover, Tanaka et al. [16], Ra and Reitz [17],
Tsurushima [18] and Neshat and Saray [19] also developed their
own reduced kinetic model of PRF.
The long-term study of PRFs with the same RON and
MON (Motor Octane Number) largely promoted the development
of the gasoline surrogate chemical kinetic model, in spite that it
could not match commercial gasoline sensitivity (S, S =
RON  MON), whose numerical value causes deviation in the
autoignition behavior. Because there is a relatively high propor-
tion of aromatics in commercial gasoline and their contribution
to soot precursors, an increasing number of researchers have
focused on the experimental study of toluene [20,21], especially
the autoignition behavior of it [22–25], meanwhile, they also
devoted themselves to developing the toluene chemical mecha-
nism [26–28], in light of the updated decomposition channels
and reaction rates [29–32]. However, it is obvious to see that most
of the toluene mechanism models only center on the oxidation
process of toluene. Recently, Yuan et al. [33,34] used SVUV-PIMS
to research the pyrolysis of toluene at pressures from 5 to 76 Torr
and temperatures from 1100 to 1730 K, and measured the concen-
tration profiles of the pyrolysis species and intermediates, while
the oxidation of toluene in a JSR was studied by GC combined with
flame ionization detector (FID), thermal conductivity detector
(TCD) and MS; he also proposed and corroborated a detailed
kinetic model of toluene pyrolysis and oxidation (272 species
and 1698 reactions), including the chemistry of aromatics growth
917
by miscellaneous methods, which is the most detailed and accu-
rate chemical model thus far.
On the basis of the development of the toluene chemical kinetic
model, TRF (toluene reference fuel) was kept drawing attention by
researchers. Gauthier et al. [8] found that the ignition delay of a
TRF named Surrogate A was consistent with the corresponding
commercial gasoline RD387 ((RON + MON)/2 = 87) and confirmed
that TRF is an efficient gasoline surrogate, and Andrae et al.
[35–37] built and reduced a detailed kinetic model of TRF by com-
bining PRF, toluene, benzene [38] sub-mechanism and cross-
reactions together, moreover, Saikai et al. [39], Niemeuer and Sung
[40] and Wang et al. [41] conducted a series of research on the
reduced chemical mechanism of TRF. Furthermore, more and more
scholars like Bruce et al. [42–44] are committed to applying surro-
gates contained TRF in experiments and simulations of engines
under various operation modes with the improvements of basic
experiments and the chemical kinetic model for gasoline
surrogates.
Therefore, the tendency of gasoline surrogates is to pursue high
reproduction in more respects of gasoline characteristic with lim-
ited components when the size of its matching chemical mecha-
nism becomes progressively smaller, which makes the ternary
blends for gasoline surrogates mixed with isooctane, n-heptane
and toluene increasingly vital. Considering the diverse components
in commercial gasoline that change with the seasons and origins,
the present study established three ternary blends for gasoline sur-
rogates, named TRF2/TRF3/TRF4, with the same compositions and
RON (RON = 95) by different blending ratios, aimed at investigating
the effects of proportion on the autoignition behavior of the surro-
gates in an ST. In addition, a detailed chemical kinetic model of TRF,
based on Curran’s [13] and Yuan’s [33,34] chemical kinetic model
of PRF and toluene, respectively, was proposed and validated in
this study, and the model was used to study the relationship
between the combustion characteristics of TRF and the pressure,
temperature and equivalence ratio.
2. Experimental approach
2.1. Shock tube apparatus and the definition of the ignition delay
The specific description and reliability analysis of all measure-
ments carried out in an ST are provided in Refs. [45,46], and a brief
introduction is given here. The entire ignition delay time experi-
ments in this study were performed in a 90-mm inner diameter,
unheated, high purity, helium-driven shock tube facility. Six
piezo-electric pressure transducers (PCB113B26) were positioned
axially along the driven section, which was used to record the arri-
val of the incident and reflected wave, whereas the OH⁄ emission
was detected by a photomultiplier (Hamamatsu, R928). The puri-
ties of all gases used in this study were 99.999%, toluene and isooc-
tane were 99.5%, and n-heptane was 99.9%. To ensure sufficient
blending, all fuels were mixed with synthetic air (N2:O2 = 3.76:1)
in a stainless-steel tank for at least 12 h at room temperature
before the experiments at every operation condition.
The definition of the ignition delay is different for kinds of
experimental facilities even for various STs used by each research
group. Commonly, the combustion pressure undergoes a sudden
rise when the fuels are ignited, yielding a tremendous amount
of OH⁄ and CH⁄ in the meantime. Thus the pressure history,
OH⁄ and CH⁄ are often regarded as the symbols of ignition, which
generally agree to within ±5% [25]. However, every research
group still has its precise definition of the ignition delay: Pitz
et al. [26] and Bounaceur et al. [28] defied the ignition delay as
the time interval between the pressure rise measured by the last
pressure transducer when the reflected shock wave arrived and
918 H. Li et al. / Fuel 181 (2016) 916–925
the growth of the OH⁄ signal detected by the photomultiplier up
to 10% of its maximum value; Vasudevan et al. [24] made the
definition of the ignition delay as the time that the OH⁄ concen-
tration reached 50% of the peak, with the initial time defined as
the arrival of the reflected shock front at the sidewall; Shen et al.
[23] and Daley et al. [47] defined the ignition delay as the time
between the shock arrival at the endwall, indicated by the inci-
dent shock velocity, and the extrapolation of the maximum slope
in OH⁄ emission to the baseline. Referred to the method of David-
son et al. [25,48,49], the ignition delay in this study is defined as
the time interval between the time that the reflected wave
arrives at the last pressure transducer, which is defined by the
step of the local pressure signal, and the intercept of the maxi-
mum slope of the OH⁄ emission profile back to the baseline,
which is showed in Fig. 1. And the combination of boundary layer
effects and the uncertainty of the temperature measurements
make the maximum overall uncertainty in ignition delay
approximately ±15% [23,50,51].
2.2. Methodology of ternary TRF blends as gasoline surrogates
It is meaningless to unwarrantedly define gasoline surrogate
components and their proportions. Both the physical and chemical
properties, especially the indexes such as RON and S, of commer-
cial gasoline should be matched by the corresponding surrogates.
By far, there are several ways to construct gasoline surrogates
according to special actual requirements: Nikolaou et al. [52] put
forward a nonlinear equation that can be easily integrated in an
online octane number GC analyzer to compute the RON of the sur-
rogates with numerous components, employing the blending
octane which was measured by blending 20 vol.% of the specific
hydrocarbon in 80 vol.% of a 60/40 iso-octane/n-heptane mixture
and comparing its calculated value with ASTM [53] measured
results. Ghosh et al. [54] presented a predictive model that could
calculate the RON and MON for a surrogate, considering the inter-
actions among components. Morgan et al. [55] also developed a
non-linear equation that is applicable to TRF only using the liquid
volume of each component. Pera and Knop [56] proposed a general
framework for defining the blending ratios of gasoline surrogate
compositions dedicated to simulating its autoignition process in
engines. Through the contradistinction of each method, the
methodology formulated and verified by Pera et al. [56–58] was
selected to establish TRF in this study, profiting from its simplicity
and comprehensiveness.
Fig. 1. Definition of ignition delay time.
X
n
f ðxi ; RONi Þ ¼ RON ð1Þ
i¼1
Xn
xi ¼ 1 ð2Þ
i¼1
It is distinctive that the percentage of every component in TRF
can be confirmed by at least three equations. The restricted condi-
tion requested by the target fuel is RON = 95; therefore, function
(1) in Pera’s [56] matrix was selected, where n is the number of
compounds in the surrogate, xi is the molar fraction of species i
and RONi represents the RON of each compound individually.
Admittedly, unity for the sum of the compound fractions is an ines-
capable constrict, which is expressed by function (2) in Pera’s [56]
matrix. By coupling the two functions presented above with the
properties of the components listed in Table 1, a series of blending
ratios could be calculated by assigning the proportion of n-heptane
ranges from 0 to 1 using a step of 0.001. Fig. 2 shows the formula-
tion of ternary blending of surrogates. Referencing the fifth stage of
the limits and the measurements methods for emissions from
light-duty vehicles of China (GB 18352.5-2013) [59], the volume
fraction of aromatics in gasoline cannot surpass 35%; thus, the
unshaded region in Fig. 2 is available. For the sake of investigating
the impact of the component proportion had on gasoline surrogate
combustion, three different ternary blends (TRF2, TRF3 and TRF4)
were uniformly defined, and TRF1, in reference to Surrogate A in
Gauthier et al. [8], was later chosen to validate the chemical kinetic
model. The specific composition of each surrogate is listed in
Table 2, and Table 3 shows the total blending ratios of the surro-
gate/synthetic air with all of the operation conditions in the ST.
3. Results and analysis
3.1. Chemical model
A new detailed kinetic model of gasoline surrogates with tern-
ary blends was constructed in this work, and it was based on the
methodology of blending the whole sub-mechanisms for different
compounds in the surrogate, which had been utilized and validated
by many predecessors [35,60,61]. This mechanism, including 1251
species and 5705 reactions, contained two main parts: the detailed
chemical model of PRF from Curran et al. [13], and the sub-
mechanism of toluene derived from Yuan et al. [33,34]. The dupli-
cated species and reactions were substituted and deleted in the
sub-mechanism of PRF after examining its feasibility with
CHEMKIN-PRO interpreter. And the pressure variations named
dp/dt were not taken into consideration in the simulation, due to
an average pressure rise less than 2% at the middle to high temper-
ature range (above 1000 K) in all experiment conditions [62].
Together with Andrae’s model of TRF [36], the chemical kinetic
mechanism proposed in this work (named the SJTU model below)
was validated against various experimental data in an ST and a JSR.
3.2. Model validation with neat toluene and TRF1
The chemical kinetic model of matched blending fuels may
simultaneously predict the combustion features of each
Table 1
Properties of compounds in the ternary blends of gasoline surrogates used for the
computations in this paper.
RON MON Molar weight (g/mol) Density at 298 K (kg/m3)
Isooctane 100 100 114.2 692
n-Heptane 0 0 100.2 684
Toluene 120 103.5 92.1 867
 H. Li et al. / Fuel 181 (2016) 916–925 919

Fig. 4 demonstrated the ignition delay measurements of neat

toluene at different pressures and equivalence ratios in an ST, com-
paring with the results simulated by the chemical kinetics of SJTU
and Andrae. It could be seen from Fig. 4(a) that neat toluene ignited
earlier as the pressure increased, with little distinction between
the activation energies at different pressures. Although the ignition
delay calculated by the SJTU model was approximately half of the
data measured in the ST at pressures of 10 and 15 bar, it represents
the tendency of the ignition delay with the variation of tempera-
ture and pressure, particularly at a pressure of 19 bar. Meanwhile,
the contrast of the experiments and calculated data with the equiv-
alence ratios of 0.5, 1 and 1.5 for neat toluene is shown in Fig. 4(b).
Both the SJTU and the Andrae models captured the tendency of
experiments at higher temperatures, whereas the bias predicted
by the Andrae model became increasingly larger below 1250 K.
In addition, the ignition features of TRF1, whose blending ratio
was the same as Surrogate A in Gauthier’s research [8], was also
Fig. 2. Formulation of the toluene reference fuels. studied in the ST at pressures of 10, 15 and 20 bar and equivalence
ratios of 0.5, 1 and 2. The experimental results illustrated a nega-
tive correlation between the ignition delay of TRF1 and the pres-
Table 2 sure, temperature and equivalence ratio, which can be observed
Composition of the ternary blends of gasoline surrogates in mole fraction.
in Fig. 5 apparently. And the prediction of the SJTU model agreed
RD387 RON95 well with the experimental data at all operation conditions, further
TRF1 (mol%) TRF2 (mol%) TRF3 (mol%) TRF4 (mol%) confirming its accuracy on the ternary blends of gasoline
Isooctane 56 42.8 65 87.2
surrogates.
n-Heptane 17 13.7 10 6.3
Toluene 27 43.5 25 6.5
3.3. Combustion characteristics of TRF2

constituent. In this study, the Perfectly Stirred Reactor in
CHEMKIN-PRO was adopted to calculate the concentration profiles
of A1CH3 (toluene), A1 (benzene), CO and CO2 as functions of the
heating temperature when neat toluene was oxidized in a JSR at
a pressure of 10 atm, a residence time of 1 s and a temperatures
ranging from 550 K to 1180 K [63]. Fig. 3 compares the experimen-
tal data and the calculated results and shows that the SJTU model
captured the decomposition of A1CH3 and predicted the formation
of CO and CO2 well within the experimental uncertainties, except
for under-predicting the variety of A1.
In terms of the verified chemical kinetic model of SJTU, TRF2,
which has the highest percentage of toluene of the three gasoline
surrogates with RON = 95, was selected to explore the combustion
characteristics by means of calculating the reaction path and the
sensitivity cooperated with its ignition delay measurements.
Fig. 6(a) and (b) indicates the ignition delay of TRF2 with the
variance of the pressure (10, 15 and 20 bar) and the equivalence
ratio (0.5, 1 and 2) individually, and the corresponding simulation
results are also provided. Similar to TRF1, the measurements of the
ignition delay showed a gradual decline with the rise of pressure,

Table 3
The composition of the mixtures in this study at different experimental conditions (mol%).

Mixtures U Isooctane (%) n-Heptane (%) Toluene (%) O2 (%) N2 (%) P (bar)
Toluene 1 0 0 2.2810 20.5292 77.1898 10
1 0 0 2.2810 20.5292 77.1898 15
1 0 0 2.2810 20.5292 77.1898 19
0.5 0 0 1.1537 20.7660 78.0803 19
1.5 0 0 3.3829 20.2977 76.3194 19
TRF1 1 1.0221 0.3103 0.4928 20.6250 77.5498 10
1 1.0221 0.3103 0.4928 20.6250 77.5498 15
1 1.0221 0.3103 0.4928 20.6250 77.5498 20
0.5 0.5158 0.1566 0.2487 20.8149 78.2640 20
2 2.0076 0.6095 0.9680 20.2553 76.1596 20
TRF2 1 0.8188 0.2621 0.8321 20.6065 77.4805 10
1 0.8188 0.2621 0.8321 20.6065 77.4805 15
1 0.8188 0.2621 0.8321 20.6065 77.4805 20
0.5 0.4133 0.1323 0.4201 20.8055 78.2288 20
2 1.6068 0.5143 1.6330 20.2197 76.0262 20
TRF3 1 1.1686 0.1798 0.4495 20.6307 77.5714 10
1 1.1686 0.1798 0.4495 20.6307 77.5714 15
1 1.1686 0.1798 0.4495 20.6307 77.5714 20
0.5 0.5896 0.0907 0.2268 20.8178 78.2751 20
2 2.2960 0.3532 0.8831 20.2663 76.2014 20
TRF4 1 1.4788 0.1068 0.1102 20.6521 77.6521 10
1 1.4788 0.1068 0.1102 20.6521 77.6521 15
1 1.4788 0.1068 0.1102 20.6521 77.6521 20
0.5 0.7457 0.0539 0.0556 20.8287 78.3161 20
2 2.9083 0.2101 0.2168 20.3077 76.3571 20
920 H. Li et al. / Fuel 181 (2016) 916–925

Fig. 3. Comparison of the experimental and analyzed data of the species concentrations for neat toluene oxidation in a JSR.

(a) different pressures (b) different equivalence ratios

Fig. 4. Comparison of the experimental and analyzed data of the ignition delay for neat toluene.

(a) different pressures (b) different equivalence ratios

Fig. 5. Comparison of the experimental and analyzed data of the ignition delay for TRF1.
 H. Li et al. / Fuel 181 (2016) 916–925
(a) different pressures
Fig. 6. Comparison of the experimental and analyzed data of the ignition delay for TRF2.
in which the overall activation energy was not sensitive to
pressure. What’s more, neither the Andrae model nor the SJTU
model could reproduce the ignition delay value of TRF2 identically:
the SJTU model slightly under-predicted the value when the pres-
sure was 15 bar and 20 bar for stoichiometry and rich ratio while
the Andrae model showed a big discrepancy nearly a factor of three
to four smaller. Additionally, there was a moderate disparity in the
activation energy when the equivalence ratio was 0.5, 1 and 2,
whereas the measurement of rich fuel mixtures was close to the
stoichiometric value, which was predicted more accurately with
the SJTU model in the temperature range of 1050–1300 K.
Generally, the dramatic production of OH may promote the
decomposition and isomerization of fuels. For the purpose of ana-
lyzing the effects that the main elementary reactions of the compo-
nents had on the combustion procedure of the surrogate, the first-
order sensitivity coefficients of the OH radical with respect to the
pre-exponentials for TRF2 were calculated based on the SJTU
model using CHEMKIN-PRO software at 1250 K, with pressures of
10 bar, 15 bar, and 20 bar and equivalent ratios of 0.5, 1 and 2.
The columns in Fig. 7(a) and (b) summarize the normalized sensi-
tivity coefficients at different pressures and equivalence ratios in
descending order, respectively, with negative values indicating
promotion and positive values indicating inhibition.
It could be seen from Fig. 7(a) that R1689: H + O2 = O + OH, the
typical chain-branching reaction at high temperatures, played the
most important role in accelerating the autoignition of TRF2 at
all three pressures due to its contribution to OH. Likewise, the
(a) various pressure at = 1.0
Fig. 7. Normalized first-order sensitivity coefficients of the OH radical with respect to preexponentials for TRF2 at various pressures when T = 1250 K.
921
(b) different equivalence ratios
promoting effects of R74: A1CH2 + HO2 = A1CH2O + OH, R1136:
IC4H7 + HO2 = IC4H7O + OH, R1705: C2H2 + O2 = HCCO + OH, and
R1343: AC3H5 + HO2 = C3H5O + OH were enhanced as the pressure
increased. On the contrary, R12: A1CH3 + OH = A1CH2 + H2O
reduced the overall reaction rate because it was prone to consume
abundant OH in competition with other promoting reactions, such
as R1447, R13 and R1616.
R12 was superior in decelerating the ignition of TRF2 with the
variation of equivalence ratios, as shown in Fig. 7(b). A special phe-
nomenon was observed for rich fuel (U = 2), where the elementary
reactions, in which the primary and secondary products produced
by toluene were involved, had a decisive effect on the autoignition.
For further investigation of how TRF2 was ignited, a sequence of
reaction pathways was established at the same conditions used in
the sensitivity analysis, when the consumptions of toluene were
20%. Fig. 8(a) and (b) represented the reaction flows of toluene in
TRF2 with the changes of pressures and equivalence ratios in the
structure formula, where the numbers from top to bottom indi-
cated the proportion of each path in the ascending order of the
pressure and the equivalence ratio.
The main decomposition channels of toluene blended in TRF2
were reactions to yield benzyl (A1CH2) and methylphenyl
(C6H4CH3) at pressures of 10, 15 and 20 bar. More than 64% of
the benzyl was converted to ethylbenzene (A1C2H5), whereas the
main oxidation products of methylphenyl were ortho-
benzoquinone (o-C6H4O2) and methylphenoxyl radical (OC6H4CH3).
When the pressure changed from low to high, there was no new
(b) various equivalence ratios at P = 20 bar
922 H. Li et al. / Fuel 181 (2016) 916–925
(a) different pressures at = 1.0
Fig. 8. Reaction pathways for toluene in TRF2 at T = 1250 K and 20% toluene consumption.
path of toluene produced, and only the extent of some reactions
was altered, such as the decrease of benzene (A1) and benzyl and
the increase of methylphenyl produced by toluene. Particularly,
the ethylbenzene produced by benzyl had a decline of 4%, and
the bibenzyl (C14H14) reduced to 2/3 of its original value due to
the increase of pressure from 10 to 20 bar. Similar to the pressure,
from the lean mixture to rich mixture, the proportion of benzene
was enhanced to 1.5 times, with the increase of bibenzyl and
indane (C9H10) to some extent.
3.4. Comparison of TRF2, TRF3, and TRF4 against combustion
properties
To further analyze the difference in the autoignition procedure
of ternary blends for surrogate gasoline with different blending
ratios and the same RON, TRF2, TRF3 and TRF4 were constructed.
Fig. 9(a) and (b) indicated the ignition delays in stoichiometric
ratio at pressures of 10 and 20 bar individually, where the points
correspond to the measurements and the lines stand for simulated
(a) P = 10 bar, = 1.0
Fig. 9. Comparison of the experimental and analyzed data of the ignition delay for TRF2/TRF3/TRF4.
(b) different equivalence ratios at P = 20 bar
data computed by the SJTU and Andrae models respectively. It can
be seen that the SJTU model could almost capture the tendency of
all three TRFs on these conditions when Andrae model predicted
too much smaller than experiment results. In the meantime, the
results obtained by the SJTU model for TRF2, TRF3 and TRF4 were
opposed to that of Andrae model by an order of magnitude at dif-
ferent temperature ranges, especially at low temperatures. For
example, as is shown in the partially enlarged drawing in Fig. 9
(a): when the temperature ranged from 1176 K to 1428 K, the sim-
ulated ignition delay obtained by the SJTU model elucidated that
TRF2 ignited the earliest, followed by TRF3 and TRF4, in contrast
to the results of the Andrae model.
Fig. 10 illustrates the purely experimental comparison of TRF2
to TRF4 when the pressure was 10 bar and 20 bar, equivalence
ratios range from 0.5 to 2.0. It cannot conclude a regular ignition
property of three TRFs with same RON on these conditions, except
that there existed no big disparity of ignition delay in both value
and tendency. So on the basis of this phenomenon, it can be
inferred that RON may play a moderated effect on the ignition
(b) P = 20 bar, = 1.0
 H. Li et al. / Fuel 181 (2016) 916–925 923

Fig. 10. Comparison of the experimental data of the ignition delay for TRF2/TRF3/TRF4.

behavior of testing fuels. In the meantime, a regression analysis For TRF2; / ¼ 1:0 : s ¼ 4:079  103 P1:207 expð35:87=RTÞ ð3Þ
applied to the measured data of TRF2–TRF4 in this study, fitted
For TRF3; / ¼ 1:0 : s ¼ 5:533  103 P0:9578 expð33:34=RTÞ ð4Þ
by three Arrhenius formulas separately, like the form:
s ¼ APn expðEa =RTÞ, where R represents the universal gas constant For TRF4; / ¼ 1:0 : s ¼ 4:748  103 P0:6625 expð31:38=RTÞ ð5Þ
of 1.986  103 kcal mol1 K1, Ea is the activation energy in kilo- In addition, the reaction pathways of toluene in TRF2, TRF3 and
calories, T is the temperature in kelvins, P denotes the reflected TRF4 were also computed by CHEMKIN-PRO with the SJTU model
shock pressure in bar, A is the pre-exponential factor and s is the at the temperatures of 1000 K and 1500 K, pressure of 10 bar and
ignition delay time in microseconds. It could be seen that the stoichiometric ratio, which are presented in Fig. 11. The major
pressure-scaling parameter and the activation energy declined primary decomposed product of toluene was benzyl and methyl-
from TRF2 to TRF4, consistent with the fraction of toluene in the phenyl in different blending ratios. From TRF2 to TRF4, 4% more
blending surrogate. toluene was converted to benzyl at 1500 K, with minimal variation

(a) T = 1000K (b) T = 1500K

Fig. 11. Reaction pathways for toluene in TRF2/TRF3/TRF4 at P = 10 bar, U = 1.0, and 20% toluene consumption.
924 H. Li et al. / Fuel 181 (2016) 916–925
(a) T = 1000 K
Fig. 12. Normalized first-order sensitivity coefficients of the OH radical with respect to pre-exponentials for TRF2, TRF3, and TRF4 at P = 10 bar and U = 1.0.
at 1000 K. From the perspective of the decomposition of benzyl, the
products indane and benzaldehyde (A1CHO) at 1000 K were
replaced by phenyl (C6H5) and ethylbenzene radical (A1CHCH3)
at 1500 K, and the production of ethylbenzene raised from TRF2
to TRF4 at both temperatures, contrary to bibenzyl. For the reason
that C14H14 ? C14H13 ? C14H12 ? C14H11 ? A3 was a considerable
approach to produce A3 (phenanthrene) [64–66], confining the
proportion of toluene in the gasoline surrogate may control the
generation of PAH (polycyclic aromatic hydrocarbon) to some
degree.
Furthermore, the normalized first-order sensitivity coefficients
of the OH radical with respect to pre-exponentials for the three
ternary blends of TRF2, TRF3 and TRF4 are also shown in Fig. 12
at a pressure of 10 bar and temperatures of 1000 K and 1500 K in
a stoichiometric ratio. It is obvious that R1667: H2O2(+M) = OH
+ OH(+M) superseded the lead position of R1689 at 1000 K, as
the main inhibiting effect of R12 was substituted by R1595: CH3 +
HO2 = CH4 + O2 and R1396: CH3 + CH3(+M) = C2H6(+M) from TRF2
to TRF4. Moreover, R74 affected the ignition process of TRF2 at
1000 K and 1500 K but only influenced TRF3 at 1000 K and was
not even at the top for TRF4. It is well-known that fuel specific
reactions that influence the formation of the radical pool in the ini-
tial stages may affect its autoignition chemistry [67], TRF2, which
has the most toluene, was dominated by the reactions of toluene
in addition to the reactions of small molecules at both tempera-
tures, and the decomposition reactions of isooctane were added
to affect the TRF3 ignition at 1000 K, whereas only the reactions
of small molecules decided the ignition of TRF4 at both tempera-
tures. Therefore, it may imply that there was no direct relationship
between the most important elementary reactions and the corre-
sponding component with the largest proportion in the blending
gasoline surrogates at temperatures of 1000 K and 1500 K, a pres-
sure of 10 bar and a stoichiometric ratio.
4. Conclusion
The ignition delay of neat toluene and four ternary gasoline
surrogates composed of different ratios of isooctane, n-heptane
and toluene was measured in an ST behind the reflected shock at
temperatures of 1052–1613 K; pressures of 10 bar, 15 bar, and
20 bar (19 bar for neat toluene); and equivalence ratios of lean
(U = 0.5), stoichiometric (U = 1) and rich (U = 2/U = 1.5 for neat
toluene). A kinetic model containing 1251 species and 5705 reac-
tions (SJTU model) was also established to compute the analyzed
data, which agreed well with the experimental results in all
conditions.
(b) T = 1500 K

The ignition delay decreased with increasing temperature, pres-
sure and equivalence ratio, not only for neat toluene, but also
for TRF1–TRF4, and there was no significant difference in the
activation energy with the variation of the equivalence ratio
in these cases.

The ignition delay of TRF2, TRF3 and TRF4 with the same RON
showed little disparity in the range of these temperatures and
pressures in measurements results, which proves the lead posi-
tion of RON rather than other characteristics when constructing
gasoline surrogates.

The major reaction pathways for toluene in TRF2 may not disap-
pear or be replaced with the change of pressure and equivalence
ratio at the same temperature and the same molar consump-
tions of toluene, and the reactions involving the primary and
secondary products yielded by toluene had a decisive position
in influencing the ignition of rich TRF2/synthetic air mixtures.

Although there was no considerable gap in the ignition delay in
the ST among TRF2, TRF3 and TRF4, the simulated ignition delay
computed by the SJTU model and the Andrae model illustrated
different orderings of TRF2, TRF3 and TRF4 at different temper-
ature ranges. Due to the impact that specific fuel reactions influ-
encing the formation of the radical pool in the initial stages had
on the autoignition chemistry, further study on the rates of
these reactions is needed by means of experiments and quan-
tum chemistry calculations.

The sensitivity analysis of TRF2, TRF3 and TRF4 may have an
implication that there was no direct relationship between the
important elementary reactions and the corresponding compo-
nent with the largest proportion in the blending gasoline
surrogate.
Acknowledgements
This work was supported by the National Science Fund for
Distinguished Young Scholars (51425602) and the National 973
Major Basic Project (2013CB228405).
References
[1] Yousefi A, Gharehghani A, Birouk M. Comparison study on combustion
characteristics and emissions of a homogeneous charge compression ignition
(HCCI) engine with and without pre-combustion chamber. Energy Convers
Manage 2015;100:232–41.
[2] Cinar C, Uyumaz A, Solmaz H, Sahin F, Polat S, Yilmaz E. Effects of intake air
temperature on combustion, performance and emission characteristics of a
HCCI engine fueled with the blends of 20% n-heptane and 80% isooctane fuels.
Fuel Process Technol 2015;130:275–81.
 H. Li et al. / Fuel 181 (2016) 916–925
[3] Jia M, Xie M, Wang T, Peng Z. The effect of injection timing and intake valve
close timing on performance and emissions of diesel PCCI engine with a full
engine cycle CFD simulation. Appl Energy 2011;88:2967–75.
[4] Jia M, Li Y, Xie M, Wang T. Numerical evaluation of the potential of late intake
valve closing strategy for diesel PCCI (premixed charge compression ignition)
engine in a wide speed and load range. Energy 2013;51:203–15.
[5] Li Y, Jia M, Liu Y, Xie M. Numerical study on the combustion and emission
characteristics of a methanol/diesel reactivity controlled compression ignition
(RCCI) engine. Appl Energy 2013;106:184–97.
[6] Zhou DZ, Yang WM, An H, Li J. Application of CFD-chemical kinetics approach
in detecting RCCI engine knocking fuelled with biodiesel/methanol. Appl
Energy 2015;145:255–64.
[7] Kalghatgi GT. Auto-ignition quality of practical fuels and implications for fuel
requirements of future SI and HCCI engines; 2005.
[8] Gauthier BM, Davidson DF, Hanson RK. Shock tube determination of ignition
delay times in full-blend and surrogate fuel mixtures. Combust Flame
2004;139:300–11.
[9] Curran HJ, Gaffuri P, Pitz WJ, Westbrook CK. A comprehensive modeling study
of iso-octane oxidation. Combust Flame 2002;129:253–80.
[10] Lu T, Law CK. Linear time reduction of large kinetic mechanisms with directed
relation graph: n-heptane and iso-octane. Combust Flame 2006;144:24–36.
[11] Glaude PA, Fournet R, Warth V, Battin-Leclerc F, Côme GM, Sacchi G; 2002.
http://www.ensic.u-nancy.fr/DCPR/Anglais/GCR/generatedmecanisms/
isooctane.
[12] Golovitchev V; 2008. http://www.tfd.chalmers.se/~valeri/.
[13] Curran HJ, Pitz WJ, Westbrook CK, Callahan CV, Dryer FL. Oxidation of
automotive primary reference fuels at elevated pressures. Proc Combust Inst
Int Symp Combust 1998;27:379–87.
[14] Chaos M, Kazakov A, Zhao Z, Dryer FL. A high-temperature chemical kinetic
model for primary reference fuels. Int J Chem Kinet 2007;39:399–414.
[15] Andrae J, Johansson D, Björnbom P, Risberg P, Kalghatgi G. Co-oxidation in the
auto-ignition of primary reference fuels and n-heptane/toluene blends.
Combust Flame 2005;140:267–86.
[16] Tanaka S, Ayala F, Keck JC, Heywood JB. Two-stage ignition in HCCI combustion
and HCCI control by fuels and additives. Combust Flame 2003;132:219–39.
[17] Ra Y, Reitz RD. A reduced chemical kinetic model for IC engine combustion
simulations with primary reference fuels. Combust Flame 2008;155:713–38.
[18] Tsurushima T. A new skeletal PRF kinetic model for HCCI combustion. Proc
Combust Inst 2009;32:2835–41.
[19] Neshat E, Saray RK. An optimized chemical kinetic mechanism for HCCI
combustion of PRFs using multi-zone model and genetic algorithm. Energy
Convers Manage 2015;92:172–83.
[20] Burcat A, Farmer RC, Espinoza RL, Matula RA. Comparative ignition delay times
for selected ring-structured hydrocarbons. Combust Flame 1979;36:313–6.
[21] Ribaucour M, Lemaire O, Minetti R. Low-temperature oxidation and
autoignition of cyclohexene: a modeling study. Proc Combust Inst Int Symp
Combust 2002;29:1303–10.
[22] Mittal G, Sung C-J. Autoignition of toluene and benzene at elevated pressures
in a rapid compression machine. Combust Flame 2007;150:355–68.
[23] Shen H-PS, Vanderover J, Oehlschlaeger MA. A shock tube study of the auto-
ignition of toluene/air mixtures at high pressures. Proc Combust Inst
2009;32:165–72.
[24] Vasudevan V, Davidson DF, Hanson RK. Shock tube measurements of toluene
ignition times and OH concentration time histories. Proc Combust Inst
2005;30:1155–63.
[25] Davidson DF, Gauthier BM, Hanson RK. Shock tube ignition measurements of
iso-octane/air and toluene/air at high pressures. Proc Combust Inst
2005;30:1175–82.
[26] Pitz WJ, Seiser R, Bozzelli JW. Chemical kinetic study of toluene oxidation
under premixed and nonpremixed conditions. In: Thirtieth (international)
symposium on combustion, Chicago, IL, United States.
[27] Metcalfe WK, Dooley S, Dryer FL. Comprehensive detailed chemical kinetic
modeling study of toluene oxidation. Energy Fuels 2011;25:4915–36.
[28] Bounaceur R, Da Costa I, Fournet R, Billaud F, Battin-Leclerc F. Experimental and
modeling study of the oxidation of toluene. Int J Chem Kinet 2005;37:25–49.
[29] Sivaramakrishnan R, Tranter RS, Brezinsky K. High-pressure, high-temperature
oxidation of toluene. Combust Flame 2004;139:340–50.
[30] Sivaramakrishnan R, Tranter RS, Brezinsky K. A high pressure model for the
oxidation of toluene. Proc Combust Inst 2005;30:1165–73.
[31] Oehlschlaeger MA, Davidson DF, Hanson RK. Investigation of the reaction of
toluene with molecular oxygen in shock-heated gases. Combust Flame
2006;147:195–208.
[32] Oehlschlaeger MA, Davidson DF, Hanson RK. Thermal decomposition of
toluene: overall rate and branching ratio. Proc Combust Inst 2007;31:211–9.
[33] Yuan W, Li Y, Dagaut P, Yang J, Qi F. Investigation on the pyrolysis and
oxidation of toluene over a wide range conditions. I. Flow reactor pyrolysis and
jet stirred reactor oxidation. Combust Flame 2015;162:3–21.
[34] Yuan W, Li Y, Dagaut P, Yang J, Qi F. Investigation on the pyrolysis and
oxidation of toluene over a wide range conditions. II. A comprehensive kinetic
modeling study. Combust Flame 2015;162:22–40.
[35] Andrae JCG, Björnbom P, Cracknell RF, Kalghatgi GT. Autoignition of toluene
reference fuels at high pressures modeled with detailed chemical kinetics.
Combust Flame 2007;149:2–24.
[36] Andrae JCG, Brinck T, Kalghatgi GT. HCCI experiments with toluene reference
fuels modeled by a semidetailed chemical kinetic model. Combust Flame
2008;155:696–712.
925
[37] Andrae JCG. Comprehensive chemical kinetic modeling of toluene reference
fuels oxidation. Fuel 2013;107:740–8.
[38] Alzueta MU, Glarborg P, Dam-Johansen K. Experimental and kinetic modeling
study of the oxidation of benzene. Int J Chem Kinet 2000;32:498–522.
[39] Sakai Y, Miyoshi A, Koshi M, Pitz WJ. A kinetic modeling study on the oxidation
of primary reference fuel–toluene mixtures including cross reactions between
aromatics and aliphatics. Proc Combust Inst 2009;32:411–8.
[40] Niemeyer KE, Sung C-J. Mechanism reduction for multicomponent surrogates:
a case study using toluene reference fuels. Combust Flame 2014;161:2752–64.
[41] Wang H, Yao M, Yue Z, Jia M, Reitz RD. A reduced toluene reference fuel
chemical kinetic mechanism for combustion and polycyclic-aromatic
hydrocarbon predictions. Combust Flame 2015;162:2390–404.
[42] Bunting BG, Eaton S, Naik CV, Puduppakkam KV, Chou C-P, Meeks E. A
comparison of HCCI ignition characteristics of gasoline fuels using a single-
zone kinetic model with a five component surrogate fuel. SAE International;
2008.
[43] Kalghatgi GT, Hildingsson L, Harrison AJ, Johansson B. Surrogate fuels for
premixed combustion in compression ignition engines. Int J Engine Res
2011;12:452–65.
[44] Solaka H, Tuner M, Johansson B, Cannella W. Gasoline surrogate fuels for
partially premixed combustion, of toluene ethanol reference fuels; 2013 [1].
[45] Geng Z, Xu L, Li H, Wang J, Huang Z, Lu X. Shock tube measurements and
modeling study on the ignition delay times of n-butanol/dimethyl ether
mixtures. Energy Fuels 2014;28:4206–15.
[46] Xu L, Ouyang L, Geng Z, Li H, Huang Z, Lu X. Experimental and kinetic study on
ignition delay times of liquified petroleum gas/dimethyl ether blends in a
shock tube. Energy Fuels 2014;28:7168–77.
[47] Daley SM, Berkowitz AM, Oehlschlaeger MA. A shock tube study of
cyclopentane and cyclohexane ignition at elevated pressures. Int J Chem
Kinet 2008;40:624–34.
[48] Hong Z, Lam K-Y, Davidson DF, Hanson RK. A comparative study of the
oxidation characteristics of cyclohexane, methylcyclohexane, and n-
butylcyclohexane at high temperatures. Combust Flame 2011;158:1456–68.
[49] Metcalfe WK, Pitz WJ, Curran HJ, Simmie JM, Westbrook CK. The development
of a detailed chemical kinetic mechanism for diisobutylene and comparison to
shock tube ignition times. Proc Combust Inst 2007;31:377–84.
[50] Tian Z, Zhang Y, Yang F, Pan L, Jiang X, Huang Z. Comparative study of
experimental and modeling autoignition of cyclohexane, ethylcyclohexane,
and n-propylcyclohexane. Energy Fuels 2014;28:7159–67.
[51] Darcy D, Tobin CJ, Yasunaga K, Simmie JM, Würmel J, Metcalfe WK, et al. A high
pressure shock tube study of n-propylbenzene oxidation and its comparison
with n-butylbenzene. Combust Flame 2012;159:2219–32.
[52] Nikolaou N, Papadopoulos CE, Gaglias IA, Pitarakis KG. A new non-linear
calculation method of isomerisation gasoline research octane number based
on gas chromatographic data. Fuel 2004;83:517–23.
[53] ASTM D 2699-92. IP237/69 Standard test method for knock characteristics of
motor fuels by the research method. Annual book of ASTM standards v.05.04;
1994.
[54] Ghosh P, Hickey KJ, Jaffe SB. Development of a detailed gasoline composition-
based octane model. Ind Eng Chem Res 2006;45:337–45.
[55] Morgan N, Smallbone A, Bhave A, Kraft M, Cracknell R, Kalghatgi G. Mapping
surrogate gasoline compositions into RON/MON space. Combust Flame
2010;157:1122–31.
[56] Pera C, Knop V. Methodology to define gasoline surrogates dedicated to auto-
ignition in engines. Fuel 2012;96:59–69.
[57] Knop V, Pera C, Duffour F. Validation of a ternary gasoline surrogate in a CAI
engine. Combust Flame 2013;160:2067–82.
[58] Knop V, Loos M, Pera C, Jeuland N. A linear-by-mole blending rule for octane
numbers of n-heptane/iso-octane/toluene mixtures. Fuel 2014;115:666–73.
[59] Limits and measurements methods for emissions from light-duty vehicles
(CHINA 5); 2013.
[60] Yahyaoui M, Djebaïli-Chaumeix N, Dagaut P, Paillard CE, Gail S. Experimental
and modelling study of gasoline surrogate mixtures oxidation in jet stirred
reactor and shock tube. Proc Combust Inst 2007;31:385–91.
[61] Cancino LR, Fikri M, Oliveira AAM, Schulz C. Autoignition of gasoline surrogate
mixtures at intermediate temperatures and high pressures: experimental and
numerical approaches. Proc Combust Inst 2009;32:501–8.
[62] Chaos M, Dryer FL. Chemical-kinetic modeling of ignition delay:
considerations in interpreting shock tube data. Int J Chem Kinet
2010;42:143–50.
[63] Moréac G, Dagaut P, Roesler JF, Cathonnet M. Nitric oxide interactions with
hydrocarbon oxidation in a jet-stirred reactor at 10 atm. Combust Flame
2006;145:512–20.
[64] Zhang L, Cai J, Zhang T, Qi F. Kinetic modeling study of toluene pyrolysis at low
pressure. Combust Flame 2010;157:1686–97.
[65] Li Y, Cai J, Zhang L, Yuan T, Zhang K, Qi F. Investigation on chemical structures
of premixed toluene flames at low pressure. Proc Combust Inst
2011;33:593–600.
[66] Matsugi A, Miyoshi A. Modeling of two- and three-ring aromatics formation in
the pyrolysis of toluene. Proc Combust Inst 2013;34:269–77.
[67] Mehl M, Chen JY, Pitz WJ, Sarathy SM, Westbrook CK. An approach for
formulating surrogates for gasoline with application toward a reduced
surrogate mechanism for CFD engine modeling. Energy Fuels
2011;25:5215–23.