1 s2.0 S0016236121002647 Main

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Fuel 293 (2021) 120388

Contents lists available at ScienceDirect

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

Full Length Article

Development of a reduced multi-component combustion mechanism for a


diesel/natural gas dual fuel engine by cross-reaction analysis
Zhenting Liu a, Liping Yang a, *, Enzhe Song a, Jiaqi Wang a, Ali Zare b, Timothy A. Bodisco b,
Richard J. Brown c
a
College of Power and Energy Engineering, Harbin Engineering University, Harbin, China
b
School of Engineering, Deakin University, VIC 3216, Australia
c
Biofuel Engine Research Facility, Queensland University of Technology, QLD 4000, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: In this paper, a four-component reduced mechanism (methane, n-dodecane, methylcyclohexane and toluene)
Natural gas with 150 species and 847 reactions was proposed for predicting the combustion characteristics and emissions
Dual fuel engine from natural gas-diesel dual fuel engines. Equivalence ratio (ϕ) from 0.5 to 2.0, pressure (P) from 30 to 90 bar,
Combustion
temperature (T) from 500 to 1700 K and ϕ from 0.5 to 2.0, P from 1 to 10 bar, T from 298 to 550 K were set as the
Reduced mechanism
Cross-reactions
reaction conditions for two reaction models respectively. The detailed mechanisms were reduced using the
directed relation graphs (DRG), directed relation graphs with error propagation (DRGEP) and full species
sensitivity analysis (FSSA) methods. The validation of the reduced mechanism was performed based on the
ignition delay and the laminar flame speed data available in the literature. Then the effects of cross-reactions on
the oxidation of diesel were further studied, associated with the reaction flux, concentration and sensitivity
analysis. Finally, the reduced mechanisms were verified at a reactivity controlled compression ignition (RCCI)
combustion mode at 25% and 75% loads, the maximum validation error is 3.3%. It was found that the effects of
cross-reactions on ignition were more pronounced in medium and low temperatures. Ignition was also enhanced
by an increase in the equivalence ratio, but was not found to be sensitive to pressure. Under lower temperatures,
adding cross-reactions can better reveal the formation of diesel intermediates. However, at higher temperatures,
the addition of cross-reactions did not significantly increase the reaction speeds of the intermediate products.

1. Introduction greenhouse gas emissions [8]. In addition, in order to achieve low


temperature combustion to simultaneously reduce NOx and PM emis­
With demand for vehicles ever increasing, there is a corresponding sions, reactivity controlled compression ignition (RCCI) combustion
increase in demand for oil based fossil fuels, which are known to cause mode has been widely studied, higher requirements are put forward for
severe environmental problems [1–2]. It is, therefore, necessary to the understanding of combustion process [9]. Therefore, the key to
explore alternative fuels. Natural gas (NG) has attracted attention owing being able to improve the efficiency, performance and emissions of dual
to its abundant reserves and reduced exhaust emissions [3–5]. NG-diesel fuel engines is a better understanding of the chemistry involved over a
dual fuel engines are employed in some heavy-duty vehicles and also in wide range of temperatures and pressures relevant to engine operation.
marine engines because of thermal efficiency and emissions advantages It is, therefore, important to study the reaction kinetics of combustion
[6]. chemistry to solve the main problems of engine combustion [10],
Dual fuel engines can produce a reliable output power and higher coupling the chemical kinetic mechanism with a computational fluid
efficiency under different engine loads with lower CO2 and particulate dynamics model has also become an important approach.
matter (PM), when compared to conventional diesel engines [7]. How­ Some researches have focused on the chemical mechanisms of in­
ever, the slower burning rate of natural gas needs more time for heat ternal combustion engines during compression and ignition. Using n-
transfer to the end gas, if the NG in the cylinder incompletely burns, the heptane and methane as the representative species of diesel and NG,
methane emission will increase, thus significantly increasing respectively, Aggarwal et al. [11], Maghbouli et al. [12], and Zhao et al.

* Corresponding author.
E-mail address: yangliping302@hrbeu.edu.cn (L. Yang).

https://doi.org/10.1016/j.fuel.2021.120388
Received 16 December 2020; Received in revised form 25 January 2021; Accepted 31 January 2021
Available online 4 March 2021
0016-2361/© 2021 Elsevier Ltd. All rights reserved.
Z. Liu et al. Fuel 293 (2021) 120388

[13] each developed reduced chemical mechanisms for dual fuel en­ accuracy of combustion. The present work provides foundations for
gines. However, all of these approaches used a single component as the incorporating realistic chemistry of hydrocarbons into large-scale tur­
substitute of diesel. Commercial diesel fuel is a mixture of more than 200 bulent or supersonic combustion modeling.
components [14]; as such, it is not truly representative to simulate the
combustion process of diesel fuel with a single component. Recognizing 2. Mechanism selection and methodology
the limitation of single-component diesel surrogate fuel models, it is
necessary to develop surrogate models with matching fuel compositions 2.1. The selection of detailed mechanisms of diesel and NG component
to actual diesel fuels. Some researchers have developed multi-
component chemical mechanism models to better represent the com­ The selection of mechanisms usually follows the following two
bustion process of commercial fuels [15–18]. However, the detailed principles:
chemical reaction mechanism of hydrocarbon fuel always contains a
large number of elemental reactions and components, if a detailed re­ (1) The scale of the mechanism should not be too large, an overly
action mechanism is directly coupled with a computational fluid dy­ complex characterization of the fuel can lead to an increase in the
namics (CFD) model, it will need to consume huge computer resources scale of the reaction mechanism and a decrease in its
[19–20]. Therefore the simplification of complex chemical mechanisms practicability.
is an important research direction in the field of fuel combustion. (2) As much as possible, representative fuels with similar physico­
Using reaction path, sensitivity analysis and slight modification of chemical properties as conventional fuels should be selected
thermodynamic data, Zhong et al. [15] developed a chemical kinetic [21].
mechanism consisting of 89 species and 355 elementary reactions which
were composed of five components, namely: iso-octane, n-heptane, The main components of diesel include alkanes (straight-chain and
ethanol, toluene and diisobutylene. Model validation results showed branched-chain alkanes), naphthenes, and aromatic hydrocarbons. Even
that using the reduced mechanism, the predicted ignition delay times though the composition of natural gas varies by regions, natural gas is
and laminar flame speeds were consistent with experimental results; not mainly made up of methane, ethane, propane and butane (methane
only for pure mono-component fuels, but also for mixed multi- accounts for more than 75%). Therefore, three different detailed
component gasoline surrogate fuels and commercial gasoline. Ra and mechanisms (n-dodecane, methylcyclohexane and toluene) and a
Reitz [16] developed a reduced chemical kinetic mechanism for the methane detailed mechanism were selected as the representative com­
oxidation of representative surrogate components of a typical multi- ponents of diesel and NG, respectively.
component automotive fuel and the final version of the mechanism
consisted of 113 species and 487 reactions. Poon et al. [17] combined
2.2. n-Dodecane mechanism
the reduced mechanisms including n-hexadecane (HXN), 2,2,4,4,6,8,8-
heptamethylnonane (HMN), cyclohexane (CHX) and toluene into two
A detailed n-dodecane mechanism was chosen to represent alkanes
different versions of multi-component diesel surrogate models. The in­
[22]. The detailed kinetic model consists 1306 elementary reactions and
tegrated mechanisms were then comprehensively validated in zero-
171 species. The base model is USC-Mech II [23] with 111 species and
dimensional chemical kinetic simulations under a wide range of condi­
784 reactions, which describes the oxidation of H2 and CO and the high-
tions. Chang et al. [18] have also formulated a diesel surrogate fuel
temperature chemistry of C1-C4 hydrocarbons. The model was validated
model with toluene, n-decane, iso-octane and methylcyclohexane to
against experimental data, including fuel pyrolysis in plug flow and jet-
represent diesel fuel. The integrated mechanism was comprehensively
stirred reactors, laminar flame speeds, and ignition delay times behind
validated against the experimental data for each fuel constituent and
reflected shock waves over a wide range of combustion conditions.
blend, as well as for commercial diesel fuel under a wide-range of
operating conditions.
In most of the multi-component chemical mechanism models used in 2.3. Toluene mechanism
the literature, a ‘reduced-then-combined’ model construction strategy
was employed. The detailed mechanisms of pure mono-component fuels Aromatics are non-negligible components in diesel fuel, which effect
were simplified separately after the scale and prediction accuracies were soot formation in the engine combustion process; therefore, toluene was
satisfied. These reduced mechanisms were subsequently combined to included to represent aromatic hydrocarbons in this study. Based on the
generate the final multi-component surrogate models. However, the work of Bounaceur et al. [24], the detailed toluene mechanism contains
interactions between the pure single-component fuels chosen, known as 329 species and 1888 reactions, which is generated by combining the
cross-reactions, in which free radicals produced by one fuel extract rate constants for benzyl radical decomposition from studies by Colket
hydrogen ions from another fuel during the oxidation process have been et al. [25] and Sakai et al. [26], and the addition reaction with a hy­
neglected, which may affect the prediction accuracy when coupling with droxyl radical taken from a study by Hippler et al. [27]. In addition, a
CFD models. Therefore, in order to improve the accuracy of dual fuel part of the H2/CO and C1 chemistry from a study by Li et al. [28] was
combustion CFD modeling, the effects of cross-reactions should be used as the C4 submechanism, C2-C4 chemistry was taken from the work
considered and analyzed in the mechanism development process. by Healy et al. [29] and the chemistry of some important unsaturated
Based on the above discussion, the main aim of this study is to species were taken from the work by Laskin et al. [30].
develop a reduced multi-component chemical kinetic mechanism to
simulate the combustion characteristics and emissions of dual fuel en­ 2.4. Methyl cyclohexane mechanism
gines. Firstly, the detailed mechanisms of the diesel and NG species have
been simplified using a directed relation graph (DRG), error propagation The methyl cyclohexane mechanism (MCH) employed in this study
extension to DRG (DRGEP) and full species sensitivity analysis (FSSA). [31] is based on the previous work of Pitz et al. [32]. Rate coefficients
After the mechanism size was satisfied, the reduced mechanisms were were changed based on recent recommendations from the literature to
examined and merged based on the comparison of ignition delay times get better agreements with more experimental data, an additional
and laminar flame speeds (determined with the detailed mechanism and concerted elimination pathway important at moderate to low tempera­
experimental data available in the literature over a range of equivalence tures was also introduced to the resulting chemical mechanism. Exten­
ratios, temperatures and pressures). In addition, the influence of cross- sive validation of the revised kinetic model was performed for flame
reactions between different diesel components was analyzed at speed and concentration of small hydrocarbon species using a wide
engine-relevant conditions to investigate their effect on the prediction range of experimental conditions and data sets [33].

2
Z. Liu et al. Fuel 293 (2021) 120388

must also be kept to avoid non-negligible error in the concentration of


species A. In addition, if the removal of species C will introduce a sig­
nificant error in species B, retaining species A will also require species C
to be retained in the skeletal mechanism, because of the indirect species
coupling between species A and C. On the other hand, if a species is not
required either directly or indirectly by any species to be retained in the
skeletal mechanism, this species can be safely removed because it will
not introduce any significant error.

2.6.2. DRGEP method


As mentioned above, the DRG can simplify the mechanism by
resolving component coupling without any knowledge of the reaction
system to identify the unimportant components in the reaction mecha­
nism. However, in the standard procedures of DRG, it is assumed that
every species selected to be kept in the mechanism is equally important,
and neglects the connection length between the coupling species. For
Fig. 1. Typical relations of the species in the DRG method.
example, in Fig. 1, if species A is selected as the starting important
species, species C is recognized as having the same importance as species
2.5. Methane mechanism B, which competently neglects the longer connection way to reach
species A.
The detailed methane (CH4) mechanism GRI Mech 3.0 [34] with 325 In order to overcome this deficiency, the DRGEP method was
reactions and 53 species, which is one of the most widely used CH4/O2 developed based on DRG simplification principle [37]. In the DRGEP
models to date, was selected in the study. Propane and C2 oxidation method, once component A is stored in the mechanism, all other com­
products, new formaldehyde and NO formation, and re-burn targets ponents that can be obtained from component A through direct and
have been added. GRI-Mech 3.0 was optimized and validated for indirect coupling will be detected using their ‘R value’, which is defined
methane and natural gas over a wide range of conditions (1000–2500 K, as:
10 Torr to 10 atm and ϕ = 0.1–5) based on flame speed measurements { }
and shock tube experiments [35]. RA (B) = max rij , (3)
s

where S is the set of all possible paths from component A to component


2.6. Simplification method
B, and rij is the chain product of weights along A given the reaction path.
For example, if A is connected to B by a reaction, and B is connected to C
The basic aim of the mechanism reduction process is to identify
by a reaction, there is a path from A to C by B, and the R value of that
unimportant species and reaction pathways in order to reduce the
path is rAB* rBC. According to this definition, if there is at least one
complexity of chemistry mechanisms, while still maintaining the
connection path from A to B and its r value is greater than the specified
important features over the desired range of operating conditions. Over
threshold, component B must be stored in the mechanism. In general,
the past 20 years, the reduction methods for chemical kinetic mecha­
the DRGEP method can generate skeleton a mechanism with about 10%
nisms have been continuously developed; considering the overall size of
less components than the DRG method, further reducing the mechanism
the detailed mechanism, DRG, DRGEP and FSSA methods were utilized
scale [38].
in this study.

2.6.3. FSSA method


2.6.1. DRG method
The more complex the kinetic model, the harder it is to identify the
The DRG method [36] identifies unimportant species in a reaction
relative importance of each species. To resolve this problem, sensitivity
mechanism by resolving species coupling without any prior knowledge
analysis is usually needed [39]. Sensitivity analysis can help us to better
of the system. Direct species coupling can be defined by the immediate
understand the importance of the chemical processes in the model. This
error to the production rate of a species A, introduced by the removal of
allows the identification of the pathways that determine the outputs or
another species B from the mechanism. Such immediate error, noted as
products of the model and how sensitive they are to change. If the results
rAB, can be expressed as
have relatively high sensitivity to a primitive reaction, the reaction will
∑ ⃒ ⃒

i=1, I vA,i ωi δBi
⃒ be retained. In this paper, full species sensitivity analysis (FSSA) was
rAB = ∑ ⃒ ⃒ ,
⃒vA,i ωi ⃒ (1) used as a last resort, when the mechanism could not be reduced further
i=1, I
by any other methods. The FSSA was performed by the following steps:
where
{ (1) Remove each species individually from the smallest skeletal
δBi =
1, if the ith reaction involes species B
(2) mechanism generated so far and calculate the induced error on
0, otherwise target parameters for this species.
A and B indicate the species, i indicates the ith reaction of the (2) Rank the species using their induced error in ascending order.
mechanism, νA,i indicates the stoichiometric coefficient of species A in (3) Remove candidate species from the skeletal mechanism accord­
the ith reaction, and ωi is the reaction rate of the ith reaction. The de­ ing to the ranked species and calculate the cumulative induced
nominator represents the total absolute contributions to the production error on target parameters after each species is removed. The
rate of species A by all reactions in the mechanism involving species A. process is stopped when the cumulative induced error is beyond
The numerator represents those contributions by reactions involving the set threshold.
species B as shown in Fig. 1.
When the immediate error, rAB, is larger than the specified error- 2.7. Theoretical discussion of cross-reactions
tolerance level, it means that the removal of species B from the mech­
anism will introduce an error to the production rate of species A. In the combustion process of multi-component fuels, whether there is
Therefore, if species A is to be kept in the skeletal mechanism, species B cross-reaction between hydrocarbon fuel molecules with different

3
Z. Liu et al. Fuel 293 (2021) 120388

carbon chain structures has always been controversial. In addition to the Table 1
discussion on the impact of cross-reactions on combustion, researchers Sample points of the closed homogeneous reactor (ϕ = 1).
also have different definitions of cross-reactions in the chemical kinetic Number Pressure Temperature Number Pressure Temperature
mechanism models. There are two kinds of cross-reactions, which can be (bar) (K) (bar) (K)
distinguished from the main components and types of cross-reactions. 1 30 700 13 70 700
The first kind of cross reaction can be defined as the elementary reac­ 2 30 900 14 70 900
tion between the small molecule free radicals (C1-C3 radical or C0 active 3 30 1100 15 70 1100
molecule) generated by oxidation cleavage reaction of a part of the 4 30 1300 16 70 1300
5 30 1500 17 70 1500
macromolecule hydrocarbon structure in multi-component fuel, and 6 30 1700 18 70 1700
between the C0 active molecule and the macromolecule. Because this 7 50 700 19 90 700
kind of reaction usually occurs in the middle and late stage of oxidation 8 50 900 20 90 900
reaction, which has a great influence on the consumption of macro­ 9 50 1100 21 90 1100
10 50 1300 22 90 1300
molecules and the overall combustion, so many scholars have conducted
11 50 1500 23 90 1500
in-depth studies on this reaction process [40,41]. 12 50 1700 24 90 1700
The other kind of cross-reaction is defined as the free radical ex­
change reaction between macromolecules generated by the dehydro­
genation of macromolecules of one component and molecules of another
Table 2
in the multi-component fuel system. Under this definition, typical cross-
Sample points of the laminar flame speed calculator (ϕ = 1).
reactions mainly includes the following three types of reactions:
Number Pressure Temperature Number Pressure Temperature
R1 + R2H ↔ R1H + R2(4) (bar) (K) (bar) (K)

1 1 298 7 5 450
R1OO + R2H → R1OOH + R2(5) 2 1 350 8 5 550
3 1 450 9 10 298
R2OO + R1H → R2OOH + R1(6) 4 1 550 10 10 350
5 5 298 11 10 450
Theoretically, up to half of the fuel consumption rate cannot be 6 5 350 12 10 550
accounted for if such cross-reactions are neglected, so in many pure
component mechanisms, these kinds of cross-reactions are fully
considered. However, studies based on macromolecule active radical shown in Table 1.
cross reactions are very limited, and the influence of the cross reactions Similarly, ϕ from 0.5 to 2.0, P from 1 to 10 bar and T from 298 to
between macromolecules on the combustion process is not yet clear. 550 K were set as the reaction conditions of the laminar flame speed
Thus, the importance of cross-reactions, especially between different calculator. Selecting ϕ as 0.5, 1.0, 2.0, P as 1, 5, 10 bar and T as 298, 350,
diesel components, should be carefully investigated. In this paper, the 450, 550 K, the calculated sampling points (when ϕ = 1.0) are shown in
influence of cross-reactions between different diesel components (n- Table 2.
dodecane, methyl cyclohexane and toluene) is analyzed at engine-
relevant conditions and attempts to explain the influence of cross- 3.2. An introduction of a simplification strategy
reactions on the prediction accuracy of the combustion process and
mechanism model in the process of multi-fuel mixed combustion. In order to achieve better prediction accuracy, and study the inter­
action between the components of diesel during the oxidation combus­
3. Mechanism simplify strategy and development tion process, the DRGEP method was used to efficiently remove
redundant species of the single components, followed by the DRG
3.1. Computational reactor model and reaction conditions method to further remove insignificant reactions and species. The FSSA
method was then applied to delete the rest of the unimportant species.
Flame speed and ignition delay are two important parameters for the After the simplification works were accomplished, the reduced mecha­
evaluation of mechanism model, therefore in the reduction process, they nisms of single components were validated using laminar flame speed
were considered as two mechanism reduction targets, the premixed and ignition time data from the literature. Finally, the reduced mecha­
flame models and closed homogeneous reactor were employed respec­ nisms of single components were merged into the reduced dual fuel
tively in CHEMKIN Pro [42]. The closed homogeneous reactor was mechanism denoted as MODEL I; after the cross-reactions were added to
defined as a closed, adiabatic system ignoring wall effects. The premixed MODEL I, the mechanism was denoted as MODEL II. The concept flow
flame model solves the set of governing differential equations that block diagram of the simplification strategy is shown in Fig. 2.
describe the flame dynamics using implicit finite difference methods, as
well as a combination of time-dependent and steady-state methods,
3.3. Development of reduced multi-component natural gas-diesel
which simulate the propagating burning process.
surrogate fuel models
The target engine of this study is a 6-cylinder, direct-injection,
turbocharged dual fuel engine, with rated power of 309 kW and a
3.3.1. Development of reduced dual fuel mechanism
displacement of 12.2 L. The dual fuel engine adopts RCCI combustion
By analyzing the detailed mechanism, the fuel, oxidizer (O2) and the
mode and can achieve low temperature combustion by changing the
complete combustion products (H2O, CO2, N2) were used as target
activity distribution in the cylinder through two diesel injection pulse.
species to obtain results efficiently, some small molecular products that
Equivalence ratios (ϕ) from 0.5 to 2.0, pressures (P) from 30 to 90 bar
are important to the combustion process were also considered, such as
and combustion temperatures (T) from 500 to 1700 K were set as the
OH. As shown in the mechanism reduction strategy mentioned above,
reaction conditions for the closed homogeneous reactor. In order to
the most important species and reactions associated with the fuel
minimize the calculation time under all working conditions, T was
oxidation process were retained to derive a reduced mechanism. The
selected as an interval of 200 K, P was selected as an interval of 20 bar
comparison of the detailed and reduced mechanisms of the four com­
and ϕ was selected as 0.5, 1.0, 2.0, a total of 6*4*3 = 72 sampling points.
ponents is shown in Table 3:
As the number of selected points is not fully displayed, only sample
As shown in Table 3, we can see that the MCH reduced mechanism
points with an equivalent ratio of 1.0 are selected as an example, as
has the biggest scale, so it was employed as the base mechanism to

4
Z. Liu et al. Fuel 293 (2021) 120388

Fig. 2. Concept flow block diagram of simplify strategy.

Table 3 Table 4
A comparison of the detailed and reduced mechanisms. Comparison of physical and chemical parameters.

Program Species Reactions Program Commercial diesel Blended fuel

Methane Detailed mechanism 53 325 Cetane number 50.9 48


Reduced mechanism 23 82 Molar H/C ratio 1.85 1.89
Liquid density (g/cm3) 0.85 0.776
n-Dodecane Detailed mechanism 175 1318 Low heating value (MJ/kg) 43 43.45
Reduced mechanism 49 265

Methyl cyclohexane Detailed mechanism 369 1896


Reduced mechanism 84 402 776 reactions.
After the blending work was accomplished, it was also necessary to
Toluene Detailed mechanism 329 1888
determine the proportions of the three diesel ingredients to match the
Reduced mechanism 61 274
physical and chemical parameters of diesel. According to the physico­
chemical parameters of the market diesel fuel shown in Table 4 [43],
merge with the other three reduced mechanisms. The blending pro­ using the Surrogate Blend Optimizer (SBO) tool in the CHEMKIN Pro
cedure is time-consuming. If two chemical species have the same [42], which is designed to optimize the composition of a model fuel in
chemical formula, it is necessary to check their thermodynamic data­ order to match specified properties of a commercial diesel fuel, the
bases by comparing the enthalpy, heat capacity, and entropy at several ingredient proportions of the three fuels was determined as: 50% for n-
temperature values. If their thermodynamic properties return the same dodecane, 35% for MCH and 15% for toluene (volume fraction). The
(or similar) values, then both chemical species are considered to be comparison of the physical and chemical parameters between the
identical and the redundancy removed. The Mechanism Utilities tool of commercial diesel and the blended fuel from the three species is listed
CHENKIN Pro was employed to complete the mechanism merging pro­ below. As shown in Table 4, it can be seen that the blended fuel matches
cess, which allows detailed analysis of any overlap and differences be­ the cetane number and H/C ratio, which are the most important pa­
tween the mechanism data, to enable informed decisions to be made rameters as they have a significant influence on the flame temperature
during the merge process. After deleting the repeated components and and speed, heat of reaction, and emissions.
reactions, the MODEL I mechanism consisted of 150 components and

5
Z. Liu et al. Fuel 293 (2021) 120388

Fig. 3. Component consumption sensitivity and reaction path analysis of


n-dodecane. Fig. 5. Component consumption sensitivity and reaction path analysis of MCH.

were considered. It can be seen from the definition of cross-reaction that


the reactants of cross-reactions are mainly secondary products of fuel
components, in order to identify the species related to cross-reactions,
the main intermediate species were analyzed in the oxidation of n-
dodecane, MCH and toluene with the current MODEL I mechanism,
which are shown in Figs. 3–5.
It can be seen that for n-dodecane in Fig. 3, H atom abstraction
products like PXC12H25, SXC12H25 and S2XC12H25 are the most impor­
tant intermediates, some peroxidation products, such as PC12H25O2,
were also considered. For toluene in Fig. 4, benzyl (C6H5CH2), perox­
ybenzoyl (C6H5CHO), phenyl radical (C6H5), benzaldehyde (OC6H4CH3)
and benzene (C6H6) are dominant intermediates. As shown in Fig. 5, the
major hydrocarbon intermediates formed during MCH oxidation
(P = 30 bar, T = 1100 K, ϕ = 1) are four different carbon bond dehy­
drogenation products: MCHR1, MCHR2, MCHR3 and MCHR4, but ac­
cording to the main types of cross-reactions, MCH1OO and CYCHEXCH2
were also added as the intermediates of MCH. Therefore, reactions be­
tween these intermediates were assumed as important cross-reactions in
this study. The chemical structure of the intermediate components is
shown in Appendix file1. According to the previous research work of
Andrae et al. [44,45] and Ra and Reitz [16], a total of 71 cross-reactions
between three diesel components were added in the MODEL I, which is
denoted as MODEL II and consists of 150 species and 847 reactions. All
these added cross-reactions are presented in Appendix file2.
Fig. 4. Component consumption sensitivity and reaction path analysis
of toluene. 4. Verification and analysis of the reduced mechanism

3.3.2. Development of reduced dual fuel mechanism with cross-reactions Validation works were carried out by the CHEMKIN-PRO package for
In the present study, based on the MODEL I mechanism derived from all of the simulations presented in this section. Ignition delay and flame
Section 3.3.1, several cross-reactions between the surrogate components speed were selected as the verification objects. Numerous experimental
data from recent fundamental experiments on these two parameters

6
Z. Liu et al. Fuel 293 (2021) 120388

Fig. 6. Comparison of measured and calculated ignition delay of methane.

Fig. 7. Comparison of measured and calculated flame speed of CH4.

were employed to validate the present surrogate mechanism for each data well. Overall, good agreement is achieved between the methane
component. reduced mechanism and experimental data with respect to ignition
delay over the test conditions.
Fig. 7(a) shows laminar flame speeds of methane-air mixtures, the
4.1. Validation of the methane reduced mechanism
experimental data was taken from the research works of Park [50], Gu
[51] and Rozenchan [52]. It can be seen that the flame speed of the CH4
For validation of the methane reaction mechanism, several shock
simulated using the reduced mechanism is slightly lower than the
tube studies at engine-like conditions were used [46–49]. Fig. 6 com­
detailed model and measured results. In Fig. 7(b), with the increase of
pares the ignition delay of methane simulated with experimental data.
equivalence and pressure, the prediction accuracy of CH4 reduced
We can see in Fig. 6(a) that at a medium temperature, the predicting
mechanism reduced. Considering measurement errors, rate constant
error will slightly increase for the reduced mechanism; in Fig. 6(b) and
errors of the kinetic model and some unimportant elementary reactions,
(c), both the detailed and reduced mechanisms predict the experimental

Fig. 8. Comparison of measured and calculated ignition delay of n-dodecane.

7
Z. Liu et al. Fuel 293 (2021) 120388

Fig. 9. Comparison of measured and calculated flame speed of n-dodecane.

Fig. 10. Comparison of measured and calculated ignition delay of toluene.

which may affect the laminar flame generation, on the whole, the 4.2. Validation of the n-dodecane reduced mechanism
experimental values are in good agreement with current study. More­
over, because NG is predominately burned in the lean-fuel region in a The experimental data of n-dodecane about shock tubes previously
dual fuel engine, the flame speeds in this region can be well predicted [53–55] were used to evaluate the reduced model. Fig. 8 shows the
using the reduced CH4 mechanism. comparison between experimental and predicted ignition delay of n-
dodecane. It is seen that although the overall discrepancy is bigger with
an increase of initial pressure, the present mechanism agrees well with

Fig. 11. Comparison of measured and calculated flame speed of toluene.

8
Z. Liu et al. Fuel 293 (2021) 120388

Fig. 12. Comparison of measured and calculated ignition delay of MCH.

the experimental data. In addition, the mechanism also correctly cap­ was obtained from the shock tube test results of Andrae et al. [61]. The
tures the complicated negative temperature coefficient (NTC) behavior, prediction results fit well with the experimental data, especially in high-
which can be noticed in Fig. 8(b) when ϕ = 1 and P = 80 bar, at tem­ temperature regions. In addition, the NTC phenomenon of toluene in the
peratures below 950 K. medium temperature region can be also be well predicted. Thus, it can
Fig. 9 illustrates the variation of flame speed for n-dodecane/air, the be concluded that the reduced toluene mechanism has good agreement
experimental data was taken from [56–59]. In Fig. 10(a), the compu­ with the detailed toluene mechanism and measured results.
tational values are lower than Kumar’s [57] data at high equivalence Fig. 11 shows predictions of laminar flame speeds of toluene using
ratio conditions and the difference becomes larger with an increase in the detailed and reduced mechanisms compared to the experimental
the initial temperature. However, compared with another data from Ji results [62–68]. In Fig. 11(a), the predicted flame speeds are consistent
et al. [56,58] and Hui [59], at T = 400 K, P = 1 bar, it can be found that with the experimental data except at T = 298 K in the lean-fuel region
the prediction value is almost the same as the experimental data of Ji and T = 400 K in the rich-fuel region; when T = 470 K the predicted
et al. [58]. In Fig. 10(b), the predicted flame speeds are near consistent flame speeds are smaller than the experimental results. However, these
with the experimental data in the lean-fuel region, but in the rich-fuel results show an overall matching trend. In Fig. 11(b), it can be seen that
region, the predicted flame speeds are slightly smaller than the experi­ the reduced mechanism slightly over-predicts the flame speed in the
mental results. Overall, the model accurately follows the dependence of medium-fuel region when P = 2 bar, T = 353 K and P = 3 bar, T = 400 K,
laminar flame speed on pressure and the simulation results show respectively. When P = 3 bar, T = 450 K, although the simulation results
excellent agreement with Hui’s data [59]. Flame development in a dual are higher than Lipzig’s data [68] at low and medium equivalence ratio
fuel engine mainly depends on the turbulent intensity, where the conditions, compared with the data from Zhong et al. [67], it can be
laminar flame speed only has a slight effect [60]. However, the analysis found that the reduced reaction mechanism is well validated in this
demonstrates that the reduced model can provide reliable predictions condition.
for laminar flame speed at these conditions.
4.4. Validation of the MCH reduced mechanism
4.3. Validation of the toluene reduced mechanism
In order to validate the reduced MCH mechanism, the rapid
Fig. 10 presents a comparison of the measured ignition delay of compression machine (RCM) ignition delay times of a stoichiometric
toluene and the corresponding simulation data; the experimental data MCH + O2 + N2 + AR mixture, measured at two different initial

Fig. 13. Comparison of measured and calculated flame speed of MCH.

9
Z. Liu et al. Fuel 293 (2021) 120388

(a) =0.5 (b) =1 (c) =2


Fig. 14. The valves of δ under different conditions.

pressures by Bryan et al. [69], Gaurav et al. [70] were used in Fig. 12(a). as incident shock attenuation and boundary layer growth [76]. Overall,
It can be seen that the verification accuracy of the simplified mechanism all of the ignition behaviors for the pure components were well repro­
will increase slightly at low temperature, but the maximum error value duced by the model. The good agreement between the predictions and
is 3.1%, which is still within a reasonable range. The main source of the measured ignition delays and flame speeds suggests that the mech­
error is the deletion of some unnecessary low temperature reaction paths anisms can capture the ignition behavior for each component in the NG
during mechanism simplification, which leads to a slight decrease in the and diesel surrogate model.
prediction accuracy of the ignition delay at low temperature. Despite
certain deviations between the simulation results and the experimental 5. The influence of cross-reactions on the oxidation process
data at P = 50 bar, the NTC phenomenon of MCH in the low-temperature
range is well predicted. Shock tube test data by Subith et al. [71] and 5.1. Analysis of the influence of cross-reactions on the ignition delay
simulated results under 20 bar and 45 bar are illustrated in Fig. 12(b).
Strong agreement between literature data and simulated results can be To show the effects of cross-reaction on the ignition delay time, two
noticed both in Fig. 12(a) and (b) at different conditions. different mechanisms (MODEL I and MODEL II) were calculated and
To further test the performance of the present reduced MCH model, compared at different pressures, temperatures and equivalence ratios.
Fig. 13 shows the simulated results and literature data measured over a The proportion of diesel fuel mixture involved in the reaction is as fol­
wide temperature range from 298 to 450 K, under 1 and 3 bar. It is seen lows: 50% for n-dodecane, 35% for methylcyclohexane and 15% for
that all data follow the rule of temperature dependence that at a higher toluene (volume fraction). In order to evaluate the differences between
temperature the flame speed is faster, though slight differences are the two mechanism models with respect to ignition delay, the ignition
observed among the data at the same temperature measured by different delay relative deviation, δ, defined by Eq. (7) was used.
groups in Fig. 13. It can also be noticed that the present data reaches a
peak around the equivalence ratio of 1.1, which is consistent with the δ = (τno cross − τcross )/τno cross *100% (7)
literature [72–75] except for the results from Zhong et al. [67] which τno cross and τcross are the ignition delay times of the two models with
reported the peak value around an equivalence ratio of 1.05 both at 400 and without cross-reactions. Fig. 14 compares the values of δ under
and 450 K. different conditions.
From the above mechanism validations for pure components of NG It can be seen that with an increase in the initial temperature, the
and diesel, it can be seen that there are still some discrepancies between values of δ generally show a negative trend. The peak relative deviation
simulation and experimental results. The main reason for this phe­ occurs at medium and low-temperature conditions (700–900 K), this is
nomenon is that the simulation of constant energy and volume were not the most pronounced at 900 K. With an increase in pressure, the values
strictly valid for the experiments of the fuel/air mixtures with long of δ rise slightly; however, owing to the relatively small influence this
ignition delay times in shock tubes due to non-ideal facility effects, such has, the effect of initial pressure on the variation of δ during the ignition

Fig. 15. Reaction rate curves of diesel components.

10
Z. Liu et al. Fuel 293 (2021) 120388

Fig. 17. Reaction flux analysis at high temperature (T = 1500 K,


P = 30 bar, ϕ = 2).

Fig. 16. Reaction flux analysis at low temperature (T = 900 K, combustion process; the solid line represents the cross reaction involved,
P = 30 bar, ϕ = 2). and the dashed line does not. It can be seen that the reaction rate keeps
rising with the increase of the initial temperature, the value of the re­
delay period can be neglected. In addition, increasing the equivalence action rate is also improved at high equivalence ratio. In addition, the
ratio will also increase δ, especially at low-temperature conditions. reaction rate growth trend of diesel components will be different under
Based on the conclusion of Fig. 14, Fig. 15 shows the reaction rate the action of cross-reactions. At medium and low temperatures,
curves of diesel components under different temperatures and equiva­ compared to the other two diesel components, the reaction rate of n-
lence ratios. The value of the reaction rate is the maximum value of the dodecane increases significantly under the effect of cross-reactions, as

11
Z. Liu et al. Fuel 293 (2021) 120388

mechanism models, taking the maximum reaction rate of diesel com­


ponents, with and without cross-reactions, reaction flux analysis at the
beginning of the reaction was conducted on the two mechanism models
under different initial conditions.
Fig. 16 shows the results of the reaction flux analysis for NC12H26,
MCH and C6H5CH3 under calculation in the closed homogeneous reactor
at low temperatures made with two mechanisms, where A means no
cross-reactions and B means cross-reactions during the combustion
process. As shown in the reaction flux analysis at low temperatures, it
can be concluded that the cross-reactions have a great influence on the
oxidation of the three diesel components at the beginning of the reac­
tion, especially on the consumption of n-dodecane. It can be seen in
Fig. 16(a), for n-dodecane, the important element reactions are basically
the reaction of n-dodecane with OH, H, O and other active components,
the main products were secondary products of PXC12H25, SXC12H25 and
S2XC12H25. With the addition of a cross-reactions, it can be seen that
consumption proportion of PXC12H25, SXC12H25 and S2XC12H25 were
significantly increased, indicating that cross-reactions can accelerate the
component consumption of NC12H26 and was favorable for the further
formation of n-dodecane secondary products. For MCH, the overall ef­
fect of cross-reactions on its consumption is relatively weak, but under
the action of cross-reactions, the consumption ratio of MCHR1 is slightly
increased. Compared with the first two components, the reaction path of
toluene is more complex, as there are both consumption and formation
paths for toluene, the most important reaction path is the reaction of
toluene to C6H5CH2. Under the action of cross-reactions, the reaction
proportion of C6H5CH2 increased by 0.5%, thus the concentration of the
active molecules generated by R584 and R585 increased.
The reaction flux analysis for the three diesel components at high
temperature are shown in Fig. 17. As can be seen from Fig. 17, when the
initial temperature rises to 1500 K, with the action of the cross-reactions,
although the reaction flux values of the three diesel components have a
certain change, for some important secondary products, the increase of
its reaction flux proportion is limited. This shows that the cross-reactions
have a small influence on the consumption process of the three diesel
components under high temperature.
Under the action of cross-reactions, the reaction flux analysis of each
diesel component under the two models when ϕ = 1 is shown in Fig. 18.
Compared with the reaction flux data shown in Fig. 16, it can be seen
that with an increase in equivalence ratio, the difference in the reaction
flux values of the important secondary products will be increased with
and without adding cross-reactions, especially for secondary products
PXC12H25, SXC12H25 and S2XC12H25. The increase in the change value
for the reaction flux means that the role of cross-reactions in the reaction
process also improved with an increase in equivalence ratio, which has
an important influence on the subsequent reaction. Thus, it can be
concluded that with an increase in equivalence ratio, the influence of
cross-reactions on combustion can be promoted.

5.3. Analysis of key components of cross-reactions during combustion

5.3.1. Analysis of diesel intermediate component concentration


In order to further study the influence of cross-reactions, the key
components of cross-reactions on combustion was analyzed through
sensitivity analysis. The first step was to choose the representative in­
Fig. 18. Reaction flux analysis when ϕ = 1 (T = 900 K, P = 30 bar). termediates of the cross-reactions and small active molecules. As can be
seen from the above reaction flux analysis, S2XC12H25, MCHR2 and
the temperature further increases, the reaction rate difference under the C6H5CH2 all play important roles in reaction path and cross-reactions.
model with and without cross-reactions gradually decreases. It can be As such, they were selected as representative intermediate compo­
seen that the change rule of the reaction rate is consistent with the nents of the three diesel components. Fig. 19 shows the mass fraction
change rule of δ in Fig. 14. curves of the above components with and without cross reactions under
T = 900 K, ϕ = 2, where the solid line shows that there is cross reaction
and the dotted line shows no cross reaction.
5.2. Reaction flux analysis with and without cross-reactions As shown below, it can be seen that with the addition of cross re­
actions, the mass fraction curve of each component changes signifi­
In order to investigate the behavior of ignition under the reduced cantly. At first, the corresponding times of peak mass fraction of

12
Z. Liu et al. Fuel 293 (2021) 120388

Fig. 19. The concentration curves of diesel intermediates (T = 900 K, P = 30 bar, ϕ = 2).

intermediate components are greatly advanced, which is consistent with of the concentration curve of intermediate products, Fig. 20 shows the
the conclusion of Fig. 15. In addition to the increase in the formation sensitivity analysis of S2XC12H25, MCHR2 and C6H5CH2. As can be seen
rate of intermediate components, the addition of cross reaction also from Fig. 20, the reaction R18: H2O2 (+M) = 2OH is positively corre­
caused a certain change in the mass peak value of diesel intermediate lated with the formation of S2XC12H25, MCHR2 and C6H5CH2. As one of
component, indicating that cross reactions would promote the formation the reactions with the greatest influence on OH generation, the sensi­
of the intermediate product under low temperature. It can be concluded tivity coefficients of R18 of the three components are all at a high level,
from Fig. 16 that the formation process of the intermediate components indicating that the increase of concentration of S2XC12H25, MCHR2 and
is usually accompanied by the formation of certain active components, C6H5CH2 can effectively accelerate the reaction rate of R18, promote the
so the increase of such intermediate components can promote the generation of OH and thus reduce the ignition delay. In addition, it can
combustion of diesel. In addition, it can be seen that the concentration be seen that in the reaction process of diesel, the formation of inter­
curves of the four intermediate components of MCH change greatly with mediate components is also accompanied by interactions. In the
the addition of cross reaction. When there is no cross-reactions, MCHR1, S2XC12H25 and C6H5CH2 sensitivity analysis, the reaction R211: MCH-
MCHR2 and MCHR3 basically presents a gradually increasing tendency C7H14 + OH → MCHR3 + H2O is negatively correlated with these com­
with the reaction time. With the addition of cross reaction, the con­ ponents. According to the previous analysis, it can be seen that the
centration of MCHR1, MCHR2 and MCHR3 first increases rapidly and growth of these two intermediates has a certain blocking effect on this
reaches the first peak, then drops and comes to the second peak, and reaction, which can also explain the variation trend of MCHR3 con­
finally rapidly consumes. The difference in this trend was further centration curve rising first and then falling under the action of cross
explained below. On the whole, cross-reactions effectively improve the reaction in Fig. 19.
reaction rate and peak mass of diesel component intermediates, thus
promoting the improvement of the basic reaction rate and further fuel 5.3.2. Analysis of the concentration of active small molecule components
consumption. In view of the results of the above comparative analysis of concen­
In order to further understand the effect of cross reactions on the tration changes and sensitivity analysis of diesel component in­
subsequent combustion reaction and the reasons for the variation trend termediates under the action of cross reaction, the effects of cross-

13
Z. Liu et al. Fuel 293 (2021) 120388

Fig. 20. Sensitivity analysis of representative intermediates of diesel components.

Fig. 21. Sensitivity analysis of OH for ϕ = 2, pressure of 30 bar, and temperature of 900 K.

reactions on the concentration change of small active molecules will also analysis of OH, it can be concluded that CH2O, CH2CHO, H2O2, HO2 are
analyzed. Firstly, sensitivity analysis of OH was used to determine the sensitive to the change of OH concentration, so the concentration
small active molecules that need analyze. Fig. 21 shows the sensitivity changes of these four components under models with and without cross

14
Z. Liu et al. Fuel 293 (2021) 120388

most important reaction R18: H2O2 (+M) = 2OH (+M), the OH peak
concentration was also increased. Based on the above analysis, it can be
concluded that cross-reactions can not only promote the consumption of
diesel components at the initial stage of the combustion, but also pro­
mote the rapid formation of active small molecules at the later stage of
combustion, which can effectively accelerate the combustion and
shorten the ignition delay.
The small molecule concentration curves under the two models at
high temperature is shown in Fig. 24. Similar to the reaction flux anal­
ysis shown in Fig. 17, the variation trend of small molecule concentra­
tion is almost the same under the two mechanism models, indicating
that the cross-reactions has very limited influence on the combustion
process of diesel components at high temperature.
Fig. 25 shows the change of the concentration of active molecules
when ϕ = 1. Compared with the concentration curves in Fig. 21, the
relative difference of the peak time of the concentration of small

Fig. 22. Concentration curves of active components (T = 900 K, P = 30 bar,


ϕ = 2; black line: with cross reactions, red line: without cross reactions). (For
interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)

reactions will be compared.


The influence of cross-reactions on the change of the concentration
of active molecules under low temperature is shown in Fig. 22. As shown
in the figure, there is a big difference in the curves of active molecules
under the two mechanism models. Firstly, under the action of cross re­
actions, the time when the concentration of small molecules increases
significantly is greatly advanced; secondly, it can be found that the
addition of cross reactions can increase the peak concentrations of H2O2.
According to the H2O2 sensitivity analysis results shown in Fig. 23, the
reactions of SXC12H25 and S2XC12H25 have a positive effect on the for­
Fig. 24. Concentration curves of active components at high temperature
mation of H2O2. It is known by reaction flux analysis that cross-reactions (T = 1500 K, P = 30 bar, ϕ = 2; black line: with cross reactions, red line: without
can increase the reaction consumption proportion of these two compo­ cross reactions). (For interpretation of the references to colour in this figure
nents, so the concentration of H2O2 was increased. Then through the legend, the reader is referred to the web version of this article.)

Fig. 23. Sensitivity analysis of H2O2 for ϕ = 2, pressure of 30 bar, and temperature of 900 K.

15
Z. Liu et al. Fuel 293 (2021) 120388

Table 5
The initial conditions of dual fuel engine.
(a) 25% load

Parameter Value Parameter Value

Speed (rpm) 1134 Torque (N*m) 439


Gas flow (kg/h) 9.47 Inlet flow (kg/h) 414.7
Diesel flow (kg/h) 1.8 Diesel pre-injection ratio (%) 50
Diesel pre-injection timing 25 Diesel main injection timing 10
(◦ CA BTDC) (◦ CA BTDC)

(b) 75% load

Parameter Value Parameter Value

Speed (rpm) 1638 Torque (N*m) 1336


Gas flow per (kg/h) 48.21 Inlet flow (kg/h) 1414.13
Diesel flow (kg/h) 2.32 Diesel pre-injection ratio (%) 50
Diesel pre-injection timing 55 Diesel main injection timing 10
(◦ CA BTDC) (◦ CA BTDC)

Table 6
The temperature boundary conditions of dual fuel engine.
(a) 25% load

Parameter Value (K) Parameter Value (K)

Inlet 310 Cylinder head 400


Gas nozzle 300 Intake port 310
Intake valve 350 Intake valve seat 350
Exhaust valve 490 Combustion chamber 510

(b) 75% load

Parameter Value (K) Parameter Value (K)

Inlet 310 Cylinder head 590


Gas nozzle 300 Intake port 310
Intake valve 455 Intake valve seat 430
Exhaust valve 590 Combustion chamber 585
Fig. 25. Concentration curves of active components when ϕ = 1 (T = 900 K,
P = 30 bar; black line: with cross reactions, red line: without cross reactions).
(For interpretation of the references to colour in this figure legend, the reader is From the above analysis, it can be concluded that under low tem­
referred to the web version of this article.) perature when the overall reaction rate is relatively low, the addition of
cross-reactions better reveals the formation and consumption rate of
intermediate products. From the sensitivity analysis, it is clear that the
accelerated reaction rates of some intermediates promote the formation
and transformation of active molecules such as OH. By contrast, under
higher temperature the overall reaction rate is greatly increased. The
addition of cross-reactions did not significantly promote the consump­
tion rate of intermediates, therefore the ignition delay difference be­
tween the two mechanism models is limited.

5.4. Final comparative verification of MODEL I and MODEL II

In order to further validate the reliability of reduced dual-fuel


mechanism determine the final mechanism for the dual fuel engine,
two mechanism models were coupled into 3D CFD software AVL FIRE
for numerical simulation. Then, the modeled results were compared
with the experimental values. According to the structure of the dual fuel
engine (mentioned in Section 3.1), the CFD model used in this study is
shown in Fig. 26, the adaptive mesh refinement method was applied to
refine the meshes of high temperature and velocity gradient regions with
a minimum cell size of 0.125 mm.

Table 7
The calculation models.
Parameter Model
Fig. 26. Mesh model of dual fuel engine.
Turbulence model k-ζ-f
Diesel droplet decomposition model WAVE
molecules (such as OH) under the two models decreases from 49.75% to Diesel droplet evaporation model Dukowicz
38.74%. So it can be concluded that the effect of cross-reactions on Wall-impingement model Wall
combustion can be promoted with the increase of equivalent ratio. Combustion model General Gas Phase Reaction Module

16
Z. Liu et al. Fuel 293 (2021) 120388

engines was developed using a combination of DRG, DRGEP and FSSA


mechanism simplifying methods. Extensive validations were carried out
for the reduced multi-component fuel oxidation model against experi­
mental ignition delay and flame speed data. Based on the validation
works, the effects of cross-reactions among diesel components were
analyzed under varying pressures, temperatures, and equivalence ratios,
reaction flux and sensitivity analysis were also conducted to investigate
the auto-ignition behavior under reduced multi-component mechanism
models with and without cross-reactions. Finally, the reduced mecha­
nisms were further verified by comparing with the experimental data of
engine at low and high loads. Some conclusions were drawn, as follows:

1) Validation results indicated that for each surrogate fuel, the pre­
dicted ignition delay time and flame speed of the reduced mechanism
agreed well with the detailed reaction mechanism and experimental
results over a wide range of pressure, temperature, and equivalence
ratio conditions. Also, the ignition delay times of the surrogate
(a) 25% load
blends were well predicted with the present reaction mechanisms.
2) The effects of cross-reactions on ignition are more pronounced in
medium and low temperatures. The effect of cross-reactions on the
ignition process is not sensitive to changes in the initial pressure. In
addition, the influence of cross-reactions is enhanced with an in­
crease in equivalence ratio, especially at low temperature conditions.
3) At low temperatures, when the overall reactivity level is low, adding
cross-reactions better reveals the interactions between the diesel
components, promotes the formation of diesel component in­
termediates and small active molecule, which accelerate the re­
actions of the active components and decrease the ignition delay
time. However, at higher temperatures the overall reaction speed of
the elementary elements is greatly increased, the addition of cross-
reactions does not significantly increase the reaction speeds of in­
termediate products, so the effect of cross-reactions on ignition be­
comes limited.
4) After comparing in-cylinder HRR and pressure under the multi-
(b) 75% load
component mechanism models with and without cross-reactions
Fig. 27. Comparison of simulated in-cylinder HRR and pressure with experi­ with experimental data, a final reduced reaction mechanism for
mental data. dual fuel engines was determined, consisting of 150 species and 843
reactions.
In the simulation process, a mixture of diesel components (50% for n-
dodecane, 35% for methylcyclohexane and 15% for toluene, volume CRediT authorship contribution statement
fraction) was used as the representatives of diesel for modeling diesel
combustion, while methane was used as the representatives of NG for Zhenting Liu: Investigation, Writing - original draft. Liping Yang:
modeling NG combustion. Based on RCCI combustion mode, the initial Conceptualization, Methodology, Supervision, Funding acquisition,
conditions and temperature boundary conditions are shown in Tables 5 Writing - review & editing. Enzhe Song: Resources. Jiaqi Wang: Formal
and 6. The calculation models used in CFD is also listed in Table 7. analysis. Ali Zare: Writing - review & editing. Timothy A. Bodisco:
Fig. 27 shows the comparison of simulated in-cylinder HRR (heat Writing - review & editing. Richard J. Brown: Writing - review &
release rate) and pressure with experimental data with experimental editing.
data under two models at 25% and 75% loads. Overall, both models did
well in predicting the in-cylinder pressure and HRR, the maximum error Declaration of Competing Interest
value is 3.3%, which meets the need of precision, but the simulation
results of the mechanism with cross reactions can better fit the experi­ The authors declare that they have no known competing financial
mental data at 25% load. From the previous analysis, it can be concluded interests or personal relationships that could have appeared to influence
that cross reaction can promote the reaction rate of diesel and accelerate the work reported in this paper.
the combustion. Thus, the phenomenon of higher cylinder pressure rise
rate with MODEL II at 25% load can be explained. On the contrary, Acknowledgments
because of slower diesel combustion speed for Model I, its prediction of
in-cylinder pressure and HRR is affected. Therefore, the model with This work was supported by the Key Projects of the Natural Science
cross-reactions (MODEL II) was chosen as the final mechanism for Foundation of Heilongjiang Province of China (ZD2019E003) and the
modeling dual fuel engines. National Natural Science Foundation of China (51306041).

6. Conclusions Appendix A. Supplementary data

In this study, employing detailed mechanisms for methane, n- Supplementary data to this article can be found online at https://doi.
dodecane, methylcyclohexane and toluene, a reduced chemical reaction org/10.1016/j.fuel.2021.120388.
mechanism for modeling the oxidation of natural gas-diesel dual fuel

17
Z. Liu et al. Fuel 293 (2021) 120388

References [32] Pitz WJ, Naik CV, Mhaoldúin TN, Westbrook CK, Curran HJ, Orme JP, et al.
Modeling and experimental investigation of methylcyclohexane ignition in a rapid
compression machine. Proc Combust Inst 2007;31(1):267–75.
[1] Sangeeta, Moka S, Pande M, Rani M, Gakhar R, Sharma M, et al. Alternative fuels:
[33] Blanquart G, Pepiot-Desjardins P, Pitsch H. Chemical mechanism for high
an overview of current trends and scope for future. Renew Sustain Energy Rev
temperature combustion of engine relevant fuels with emphasis on soot precursors.
2014;32:697–712.
Combust Flame 2009;156(3):588–607.
[2] Yang LP, Ding SL, Litak G, Song EZ, Ma XZ. Identification and quantification
[34] Hu E, Li X, Meng X, Chen Y, Cheng Y, Xie Y, et al. Laminar flame speeds and
analysis of nonlinear dynamics properties of combustion instability in a diesel
ignition delay times of methane-air mixtures at elevated temperatures and
engine. Chaos 2015;25(1):301–2133.
pressures. Fuel 2015;158:1–10.
[3] Korakianitis T, Namasivayam AM, Crookes RJ. Natural-gas fueled spark-ignition
[35] Tingas EA, Manias DM, Sarathy SM, Goussis DA. CH4/air homogeneous
(SI) and compression-ignition (CI) engine performance and emissions. Prog Energy
autoignition: a comparison of two chemical kinetics mechanisms. Fuel 2018;223:
Combust Sci 2011;37(1):89–112.
74–85.
[4] Wang LY, Yang LP, Song EZ, et al. Effect of port gas injection on the combustion
[36] Lu T, Law CK. On the applicability of directed relation graphs to the reduction of
instabilities in a spark-ignition lean-burn natural gas engine. Chaos 2018;28(10):
reaction mechanisms. Combust Flame 2006;146(3):472–83.
1850124.
[37] Pepiotdesjardins P, Pitsch H. An efficient error-propagation-based reduction
[5] Yang LP, Song EZ, Ding SL, Brown RJ, Marwan N, Ma XZ. Analysis of the dynamic
method for large chemical kinetic mechanisms. Combust Flame 2008;154(1–2):
characteristics of combustion instabilities in a pre-mixed lean-burn natural gas
67–81.
engine. Appl Energy 2016;183:746–59.
[38] An J, Jiang Y. Differences between direct relation graph and error-propagation-
[6] Wang Z, Zhao Z, Wang D, Tan M, Han Y, Liu Z, et al. Impact of pilot diesel ignition
based reduction methods for large hydrocarbons. Procedia Eng 2013;62:342–9.
mode on combustion and emissions characteristics of a diesel/natural gas dual fuel
[39] Long L, Stevens JG, Farrell JT. A dynamic adaptive chemistry scheme for reactive
heavy-duty engine. Fuel 2016;167:248–56.
flow computations. Proc Combust Inst 2009;32(1):527–34.
[7] Yang LP, Bodisco TA, Zare A, Marwan N, Chu-Van T, Brown RJ. Analysis of the
[40] Dias V, Katshiatshia HM, Jeanmart H. The influence of ethanol addition on a rich
nonlinear dynamics of inter-cycle combustion variations in an ethanol fumigation-
premixed benzene flame at low pressure. Combust Flame 2014;161(9):2297–304.
diesel dual-fuel engine. Nonlinear Dyn 2019;95(3):2555–74.
[41] Yao C, Xu G. Chemical kinetic mechanism and a skeletal model for oxidation of n-
[8] Mousavi SM, Saray RK, Poorghasemi K, Maghbouli A. A numerical investigation on
heptane/methanol fuel blends. Fuel 2012;93:625–31.
combustion and emission characteristics of a dual fuel engine at part load
[42] CHEMKIN-PRO; Reaction Design: San Diego, CA, 2008.
condition. Fuel 2016;166:309–19.
[43] Szymkowicz PG, Benajes J. Single-cylinder engine evaluation of a multi-component
[9] Yang LP, Zare A, Bodisco TA, Nabi N, Liu ZT, Brown RJ. Analysis of cycle-to-cycle
Diesel surrogate fuel at partially-premixed and low-temperature combustion
variations in a common-rail compression ignition engine fueled with diesel and
modes. Fuel 2019;241:506–18.
biodiesel fuels. Fuel 2021;290.
[44] Andrae J, Johansson D, Bjoernbom P, Risberg P, Kalghatgi G. Co-oxidation in the
[10] Guo S, Wang J, Wei X, Yu S, Zhang M, Huang Z. Numerical simulation of premixed
auto-ignition of primary reference fuels and n-heptane/toluene blends. Combust
combustion using the modified dynamic thickened flame model coupled with
Flame 2005;140(4):267–86.
multi-step reaction mechanism. Fuel 2018;233:346–53.
[45] Andrae JCG, Bjoernbom P, Cracknell RF, Kalghatgi GT. Autoignition of toluene
[11] Aggarwal SK, Awomolo O, Akber K. Ignition characteristics of heptane–hydrogen
reference fuels at high pressures modeled with detailed chemical kinetics. Combust
and heptane–methane fuel blends at elevated pressures. Fuel Energy Abstracts
Flame 2007;149(1–2):2–24.
2011;36(23):15392–402.
[46] Huang J, Hill PG, Bushe WK, Munshi SR. Shock-tube study of methane ignition
[12] Maghbouli A, Saray RK, Shafee S, Ghafouri J. Numerical study of combustion and
under engine-relevant conditions: experiments and modeling. Combust Flame
emission characteristics of dual-fuel engines using 3D-CFD models coupled with
2004;136(1/2):25–42.
chemical kinetics. Fuel 2013;106:98–105.
[47] El Merhubi H, Kéromnès A, Catalano G, Lefort BT, Moyne LL. A high pressure
[13] Zhao W, Yang W, Fan L, Zhou D, Ma X. Development of a skeletal mechanism for
experimental and numerical study of methane ignition. Fuel 2016;177:164–72.
heavy-duty engines fuelled by diesel and natural gas. Appl Therm Eng 2017:
[48] Herzler J, Sakai Y, Fikri M, Schulz C. Shock-tube study of the ignition and product
1060–71.
formation of fuel-rich CH4/air and CH4/additive/air mixtures at high pressure.
[14] Anand K, Ra Y, Reitz RD, Bunting B. Surrogate model development for fuels for
Proc Combust Inst 2019;37(4):5705–13.
advanced combustion engines. Energy Fuels 2011;25(4):1474–84.
[49] Shao J, Choudhary R, Davidson DE, Hanson RK, Barak S, Vasu S. Ignition delay
[15] Zhong BJ, Zheng D. A chemical mechanism for ignition and oxidation of multi-
times of methane and hydrogen highly diluted in carbon dioxide at high pressures
component gasoline surrogate fuels. Fuel 2014;128:458–66.
up to 300 atm. Proc Combust Inst 2019;37:4555–62.
[16] Ra Y, Reitz RD. A combustion model for IC engine combustion simulations with
[50] Park O, Veloo PS, Liu N, Egolfopoulos FN. Combustion characteristics of alternative
multi-component fuels. Combust Flame 2011;158(1):69–90.
gaseous fuels. Proc Combust Inst 2011;33(1):887–94.
[17] Poon HM, Pang KM, Ng HK, Gan S, Schramm J. Development of multi-component
[51] Gu XJ, Hap MZ, Lawes M, Woolley R. Laminar burning velocity and Markstein
diesel surrogate fuel models - Part II: Validation of the integrated mechanisms in 0-
lengths of methane–air mixtures. Combust Flame 2000;121(1–2):41–58.
D kinetic and 2-D CFD spray combustion simulations. Fuel 2016;181:120–30.
[52] Rozenchan G, Zhu DL, Law CK, Tse SD. Outward propagation, burning velocities,
[18] Chang Y, Jia M, Li Y, Liu Y, Xie M, Wang H, et al. Development of a skeletal
and chemical effects of methane flames up to 60 ATM. Proc Combust Inst 2002;29
mechanism for diesel surrogate fuel by using a decoupling methodology. Combust
(2):1461–70.
Flame 2015;162(10):3785–802.
[53] Zeng M, Yuan W, Li W, Zhang Y, Wang Y. Investigation of n-dodecane pyrolysis at
[19] Nagy T, Turányi T. Reduction of very large reaction mechanisms using methods
various pressures and the development of a comprehensive combustion model.
based on simulation error minimization. Combust Flame 2009;156(2):417–28.
Energy 2018;155:152–61.
[20] Lu T, Law CK. A criterion based on computational singular perturbation for the
[54] Shao J, Choudhary R, Peng Y, Davidson DF, Hanson RK. A shock tube study of n-
identification of quasi steady state species: a reduced mechanism for methane
heptane, iso- octane, n-dodecane and iso- octane/n-dodecane blends oxidation at
oxidation with NO chemistry. Combust Flame 2008;154(4):761–74.
elevated pressures and intermediate temperatures. Fuel 2019;243:541–53.
[21] Wang Q, Chen CP. Simulated kinetics and chemical and physical properties of a
[55] Sun Xiuxiu, Liang Xingyu, Yu Hanzhengnan, Wang Yajun, Zhu Zhijie. Comparison
four-component diesel surrogate fuel. Energy Fuels 2017;31(12):13190–7.
the Performance of N-heptane, N-dodecane N-tetradecane and N-hexadecane.
[22] You X, Egolfopoulos FN, Hai W. Detailed and simplified kinetic models of n-
Energy Procedia 2017;105:1426–33.
dodecane oxidation: The role of fuel cracking in aliphatic hydrocarbon
[56] Ji C, You X, Holley AT, Wang YL, Wang H. Propagation and extinction of premixed
combustion. Proc Combust Inst 2009;32(1):403–10.
n-dodecane/air flames. 32nd International Symposium on Combustion 2007,
[23] http://ignis.usc.edu/USC_Mech_II.htm.
Montreal, Canada.
[24] Bounaceur R, Da Costa I, Fournet R, Billaud F, Battin-Leclerc F. Experimental and
[57] Kumar K, Sung CJ. Laminar flame speeds and extinction limits of preheated n-
modeling study of the oxidation of toluene. J Chem Kin 2005;37(1):25–49.
decane/O_2/N_2 and n-dodecane/O_2/N_2 mixtures. Combust Flame 2007;151:
[25] Colket MB, Seery DJ. Reaction mechanisms for toluene pyrolysis. Sympos Combust
209–24.
1994;25(1):883–91.
[58] Ji C, Dames E, Wang YL, Wang H, Egolfopoulos FN., Propagation and extinction of
[26] Sakai Y, Miyoshi A, Koshi M, Pitz WJ. A kinetic modeling study on the oxidation of
premixed C5–C12 n-alkane flames. Combust Flame 2010;157(2):277-287.
primary reference fuel–toluene mixtures including cross reactions between
[59] Hui X, Sung CJ. Laminar flame speeds of transportation-relevant hydrocarbons and
aromatics and aliphatics. Proc Combust Inst 2009;32(1):411–8.
jet fuels at elevated temperatures and pressures. Fuel 2013;109:191–200.
[27] Hippler H, Reihs C, Troe J. Shock tube UV absorption study of the oxidation of
[60] Haozhong H, Delin L, Jizhen Z, Zhaojun Z, Yingjie C, Yuping P, et al. Development
benzyl radicals. Symp Combust 1991;23(1):37–43.
of a new reduced diesel/natural gas mechanism for dual-fuel engine combustion
[28] Li J, Zhao Z, Kazakov A, Chaos M, Dryer FL, Scire JJ. A comprehensive kinetic
and emission prediction. Fuel 2019;236:30–42.
mechanism for CO, CH2O, and CH3OH combustion. J Chemical Kin 2010;39(3):
[61] Andrae JCG, Brinck T, Kalghatgi GT. HCCI experiments with toluene reference
109–36.
fuels modeled by a semidetailed chemical kinetic model. Combust Flame 2008;155
[29] Healy D, Curran HJ, Dooley S, Simmie JM, Kalitan DM, Petersen EL, et al.
(4):696–712.
Methane/ethane/propane mixture oxidation at high pressures and at high,
[62] Davis SG, Wang H, Breinsky K, Law CK. Laminar flame speeds and oxidation
intermediate and low temperatures. Combust Flame 2008;155(3):451–61.
kinetics of benene-air and toluene-air flames. Sympos Combust 1996;26(1):
[30] Laskin A, Hai W, Law CK. Detailed kinetic modeling of 1, 3-butadiene oxidation at
1025–33.
high temperatures. J Chemical Kin 2000;32(10):589–614.
[63] Davis SG, Law CK. Determination of and fuel structure effects on laminar flame
[31] Narayanaswamy K, Pitsch H, Pepiot P. A chemical mechanism for low to high
speeds of C1 to C8 hydrocarbons. Combust Sci Technol 1998;140(1):427–49.
temperature oxidation of n-dodecane as a component of transportation fuel
[64] Husson B, Bounaceur R, Tanaka K, Ferrari M, Herbinet O, Glaude PA, et al.
surrogates. Combust Flame 2015;162(4):1193–213.
Experimental and modeling study of the oxidation of n-butylbenzene. Combust
Flame 2012;159(4):1399–416.

18
Z. Liu et al. Fuel 293 (2021) 120388

[65] Kumar K, Sung CJ. Flame propagation and extinction characteristics of neat [71] Vasu SS, Davidson DF, Hong Z, Hanson RK. Shock tube study of methylcyclohexane
surrogate fuel components. Energy Fuels 2010;24:3840–9. ignition over a wide range of pressure and temperature. Energy Fuels 2009;23(1):
[66] Han D, Deng S, Liang W, Zhao P, Wu F, Huang Z, et al. Laminar flame propagation 175–85.
and nonpremixed stagnation ignition of toluene and xylenes. Proc Combust Inst [72] Alekseev VA, Matveev SS, Chechet IV, Matveev SG, Konnov AA. Laminar burning
2017;36(1):479–89. velocities of methylcyclohexane + air flames at room and elevated temperatures: a
[67] Zhong BJ, Peng HS, Zheng D. The effect of different class of hydrocarbons on comparative study. Combust Flame 2018;196:99–107.
laminar flame speeds of three C_7 fuels. Fuel 2018;225(AUG.1):225-229. [73] Lam KY, Hong Z, Davidson DF, Hanson RK. Shock tube ignition delay time
[68] van Lipzig JPJ, Nilsson EJK, de Goey LPH, Konnov AA. Laminar burning velocities measurements in propane/O2/argon mixtures at near-constant-volume conditions.
of n-heptane, iso-octane, ethanol and their binary and tertiary mixtures. Fuel 2011; Proc Combust Inst 2011;33(1):251–8.
90(8):2773–81. [74] Ji C, Dames E, Sirjean B, Hai W, Egolfopoulos FN. An experimental and modeling
[69] Weber BW, Pitz WJ, Mehl M, Silke EJ, Davis AC, Sung C-J. Experiments and study of the propagation of cyclohexane and mono-alkylated cyclohexane flames.
modeling of the autoignition of methylcyclohexane at high pressure. Combust Proc Combust Inst 2011;33(1):971–8.
Flame 2014;161(8):1972–83. [75] Wu F, Kelley AP, Law CK. Laminar flame speeds of cyclohexane and mono-
[70] Mittal G, Sung CJ. Autoignition of methylcyclohexane at elevated pressures. alkylated cyclohexanes at elevated pressures. Combust Flame 2012;159(4):
Combust Flame 2009;156(9):1852–5. 1417–25.
[76] Chaos M, Dryer FL. Chemical-kinetic modeling of ignition delay: Considerations in
interpreting shock tube data. J Chemical Kin 2010;42(3):143–50.

19

You might also like