Professional Documents
Culture Documents
Investigation On Soot Emissions From Diesel-Cng Dual-Fuel: Sciencedirect
Investigation On Soot Emissions From Diesel-Cng Dual-Fuel: Sciencedirect
ScienceDirect
Article history: The burning of diesel and compressed natural gas (CNG) is attractive compared to diesel
Received 6 November 2018 fuel because of the reduction of CO2 emissions and particulate matter (PM) emissions.
Received in revised form While soot emissions from the diesel-CNG combustion can be tested in a real-world single-
31 January 2019 cylinder engine, the soot formation characteristics cannot be tested in the same way.
Accepted 4 February 2019 Therefore, to understand the mechanisms behind soot formation in diesel-CNG combus-
Available online 11 March 2019 tion, soot evolution must be investigated using a simulation model. In this study, the soot
evolution is investigated under different CNG substitution ratios with single and split fuel
Keywords: injection. An AVL 5402 single-cylinder diesel engine was modified to run diesel/CNG dual-
Dual-fuel combustion fuel to investigate the combustion and soot emissions. A new soot model using KIVA-3V R2
KIVA-3V R2 code code and integrated with a reduced heptane/methane PAH (polycyclic aromatic hydro-
Soot model carbons) mechanism was used to simulate soot behavior. For the combustion, the results
Single and split injection show that the ignition delay gets extended, the combustion duration gets shorter and the
peak pressure can be improved when CNG substitution ratio is increased both with single
and split injection. Additionally, a slight increase of pressure is observed when the split
injection is used. This is because the split injection is an effective strategy to change the
distribution and vaporization of fuel, which results in an incremental increase in com-
bustion efficiency and increase pressure. As the CNG substitution ratio is increased, soot
emissions get drastically reduced. The reason is the equivalence ratio distribution of air-
fuel becomes more homogenous and the local fuel-rich region shrinks with increasing of
CNG substitution ratios. Pyrene is an important intermediate specie to generate soot par-
ticles. The results show that pyrene distribution decreases, leading to a reduced generation
of soot precursors. As a result, the soot mass of CNG70 is less than the other two cases. The
basic reason is the prolonged ignition delay allowed for more time for fuelair mixing,
which reduces soot mass formation.
© 2019 Published by Elsevier Ltd on behalf of Hydrogen Energy Publications LLC.
* Corresponding author. Department of Mechanical Science and Engineering, University of Illinois at Urbana-Champaign, 61801, USA.
E-mail address: cflee@illinois.edu (C.-F.F. Lee).
https://doi.org/10.1016/j.ijhydene.2019.02.012
0360-3199/© 2019 Published by Elsevier Ltd on behalf of Hydrogen Energy Publications LLC.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 9 4 3 8 e9 4 4 9 9439
Since the experiment results were based on CNG with 99% (2) C2 H2 assisted surface growth
methane, methane was selected as the surrogate fuel. The
UBC-Mech 1.0 comprised of 38 species was applied as the base Research indicates that the surface growth contributes
mechanism for methane oxidation in this study [21]. considerably to soot particles mass growth. In this study, the
acetylene adsorption model which was proposed by Harris
(2) Lower carbon chemistry and Weiner [8] is used to describe this process. The process is
proportional to the C2 H2 concentration and the soot particles
A skeletal kinetic mechanism (Bin Pang's mechanism) surface area. TheC2 H2 assisted surface growth can be shown
comprised of 57 species and 176 reactions was applied for PRF as below
9
u_ 4 12100 pffiffiffi >
>
CðpreÞ þ C2 H2 !3CðsÞ þ H2 _u4 ¼ 9:0 104 exp ½C2 H2 S > >
>
RT >
>
>
=
2
S ¼ p,ds Ns (2)
>
>
1 >
>
6MWC ys 3 >
>
ds ¼ >
>
pNs rs ;
u_ 5
CðpreÞ þ A1 !7CðsÞ þ 3H2 u_ 5 (5) Soot oxidation by OH
6159 pffiffiffi
¼ 1:03 104 exp ½A1 S (3)
RT Soot oxidation by OH is also considered in this model [23].
u_ 7 1 3nOH 8RT
CðsÞ þ OH!CO þ H2 u_ 7 ¼ gOH S (6)
2 NA pMOH
(4) Surface oxidation
gOH is the OH collision efficiency (0.13 in this study), NA is
Avogadro constant, nOH is the OH number density, MOH is the
The NagleeStrickland and Constable model [22] is used to
mole fraction.
describe the surface oxidation process in this model.
1 u_ 6 12 kA PO2 (6) Soot particle coagulation
CðsÞ þ O2 !CO u_ 6 ¼ x
2 MWC 1 þ kZ PO2
The collision frequency in the present model is used by
þ kB PO2 ð1 xÞ S (4)
Pratsinis and Kazakov's expression [24].
kfm knc
u_ co ¼ (7)
kfm þ knc
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 9
Table 5 e The uncertainties of the measurements. 6kB Tds >
>
kfm ¼ 4a >
>
rs >
>
Measurement devices Uncertainty (U) >
=
A mass flow meter (CNG) ±0.032 mg/cyc 8kB T (8)
±0.027 mg/cyc
knc ¼ ð1 þ 1:257kn Þ >
>
>
>
An in-house fuel consumption device m >
>
Kistler 6125B pressure transducer ±0.03 bar >
;
kn ¼ 2l=ds
A flow meter (sampling flow rate) ±0.0041 L/cyc
Soot mass ±0.2 FSN a is Van der Waals factor (2.0 in this study), kn is the Knudsen
9442 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 9 4 3 8 e9 4 4 9
number, kB is the Boltzmann constant, l is mean free path for a where mn ði4Þ and Nn ði4Þ are the soot mass concentration and
gas molecule. soot molar concentration.
Considering the oxidation effect on the soot number, four
(7) Soot precursor oxidation situations were discussed as follow [27]:
The soot precursor's oxidation processes have been 1) RG O 0. It's indicated that surface growth is faster than
described as Tao and Bi's model [19,25]. surface oxidation. The soot mass will increase at this time
step. The effect of surface oxidation on soot number can be
u_ 8 1
CðpreÞ þ OH!CO þ H2 u_ 8 neglected in this situation. The soot molar concentration is
2
calculated using the same equation as that in Tao's model.
39740
¼ 1:0 10 exp
9
CðpreÞ ½OH (9)
RT
Nn ði4Þ ¼ Nn1 ði4Þ þ u_ pre u_ co Dt (14)
dmðsootÞ
¼ u_ pre MWInc þ u_ gr u_ ox MWc (10)
dt
The increase of soot mass between soot particle surface
growth and oxidation in one computational cell can be given
by Ref. [26]
RG O ¼ u_ gr u_ ox MWc (11)
13
=
6mn ði4Þ
dn ði4Þ ¼ (12)
pNn ði4Þrs
mn ði4Þ
MWn ði4Þ ¼ (13) Fig. 2 e Soot particles in the cell at current time step.
Nn ði4Þ
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 9 4 3 8 e9 4 4 9 9443
RG O
Nn ði4Þ ¼ 1þ u_ pre Dt
u_ pre Dt MWInc þ N ði4ÞMW ði4Þ
n1 n1
þ Nn1 ði4Þ u_ co Dt
(17)
Numerical implementation
In this study, a 72 sector mesh was developed using the KIVA-
3V Release 2 code. The inlet and exhaust process are not
considered in the calculation. The computation interval is
from the inlet valve close to the exhaust valve open
(226 CA~480 CA). Fig. 3 shows the grid relevant to the com-
bustion chamber at TDC. The mesh independent had been
Fig. 3 e Computational grid at TDC.
studied previously [28], and the mesh size used in this study
has little influence on the simulation results. For calculating
2) RG O < 0and dmði4Þ the fuel parcel injection dynamics, the ‘blob’ model is used.
dt < 0. It's indicated that surface oxidation is
faster than the combined effects of soot inception and For calculating the low Mach-number turbulent effect on
surface growth. In this case, it is assumed the incipient transportation and flow field, the RNG k-ε model is used. The
soot particles burn out first and the soot mass will KelvineHelmholtz model is used to calculate spray atomiza-
decrease. The soot molar concentration is calculated using tion and RayleigheTaylor model is used to calculate droplet
the equation below breakup process.
surface oxidation is faster than surface growth, but the Fig. 4 shows the comparisons of ignition delay timing at
soot mass will increase because of the incipient soot par- equivalence ratios equal to 0.5 and 1.0. It is shown that the
ticles. In this situation, the diameter of incipient soot par- simulation results and experiment data are in a good agree-
ticles is assumed to be 1.28 nm. If the mean diameter of ment, which means the present reaction mechanism can well
soot particles dn ði4Þ is larger than or equal to 1.28 nm, only predict the ignition delay of n-heptane oxidation.
the incipient soot molar concentration will decrease due to Fig. 5 shows the comparisons of mole fraction of important
oxidation effect. species. The experimental data were taken from Obwald[30]. It
is shown that the simulation results and experiment data are
in a good agreement, especially the aromatic species, which
RG O
Nn ði4Þ ¼ Nn1 ði4Þ þ u_ pre þ u_ pre u_ co Dt (16) are the important species in the soot formation.
MWInc
Fig. 4 e Ignition delay times of n-heptane oxidation between the present mechanism and shock tube test data.
9444 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 9 4 3 8 e9 4 4 9
Fig. 5 e Mole fraction of important species between the present mechanism and test data.
results are shown in Fig. 6. The simulation results and accumulated heat release is defined as the ignition delay. The
experiment data are in a good agreement under different CNG time interval between 10% and 90% mass-burned fraction is
substitution ratios. defined as combustion duration. The ignition delay and
Fig. 7 shows the ignition delay, combustion duration, and combustion duration have different trends. The ignition delay
peak pressure of single injection. The time 10% of the gets progressively larger with increasing CNG%, whereas the
Fig. 10 compares the evolution of in-cylinder soot precur- fuelair mixing, so soot precursor reduces with the increase
sor, as well as the A4 distribution at the equivalence ratio ¼ 2.0 of CNG%.
iso-surface. For the soot precursor, it increases and then de-
creases, reaching the highest at around 10 CAD of the three Validation and analysis of CNG rate on combustion process
cases. The decrease of soot precursors is because of particle with split injection
coagulation and soot precursor oxidation. For the peak soot
precursor, it has a significant downward trend with an in- The comparisons between the computational and experi-
crease of the CNG substitution rate, which is 2.58E-8 g/cyc at mental results are shown in Fig. 11. The simulation results
CNG70. Since the soot precursor is very sensitive to pyrene and experiment data are in a good agreement under different
(because it is important intermediate species related to soot CNG substitution ratios with split injection.
nucleation), the concentrations of pyrene at the equivalence The measured ignition delay, combustion duration, and
ratio ¼ 2.0 iso-surface is shown. By comparing the distribu- peak pressure under different substitution rates of split in-
tions of pyrene, it can be observed that the distribution of jection are shown in Fig. 12. The ignition delay and the com-
pyrene less with the increase of CNG%, leading to reduced bustion duration have the same trend with the single
soot precursor generators, so the soot mass of CNG70 is injection. The ignition delay increases, while the combustion
smaller than the other two cases. Thus, the decrease of the duration decreases. A difference of 0.7 and 6 of ignition delay
soot mass along with the increase of CNG% is because of the and combustion duration is observed between diesel and
reduction in soot precursor generators. The basic reason is CNG70. The longer ignition delay of CNG70 results in a higher
that the prolonged ignition delay allowed for more time for peak pressure, which is 71.69 bar. Comparing split and single
Conclusions
WLTP/GTR 15 test procedure[J]. SAE Int J Engines mechanism for practical engine simulation[J]. Energy Fuels
2015;8:1300e12 (2015-01-1061). 2013;27(3):1699e711.
[5] Papagiannakis RG, Krishnan SR, Rakopoulos DC, et al. A [17] Kazakov Andrei, Wang Hai, Frenklach Michael. Detailed
combined experimental and theoretical study of diesel fuel modeling of soot formation in laminar premixed ethylene
injection timing and gaseous fuel/diesel mass ratio effects flames at a pressure of 10 bar[J]. Combust Flame 1995;100(1-
on the performance and emissions of natural gas-diesel 2):111e20.
HDDI engine operating at various loads[J]. Fuel [18] Frenklach Michael, Wang Hai. Detailed modeling of soot
2017;202:675e87. particle nucleation and growth[M]Elsevier. 1991. p. 1559e66.
[6] Wei Lijiang, Geng Peng. A review on natural gas/diesel dual [19] Tao Feng, Reitz Rolf D, Foster David E, et al. Nine-step
fuel combustion, emissions and performance[J]. Fuel Process phenomenological diesel soot model validated over a wide
Technol 2016;142:264e78. range of engine conditions[J]. Int J Thermal Sci
[7] Wang Zhongshu, Zhao Zhongxiang, Wang Dan, et al. Impact 2009;48(6):1223e34.
of pilot diesel ignition mode on combustion and emissions [20] Nithyanandan Karthik, Zhang Jiaxiang, Li Yuqiang, et al.
characteristics of a diesel/natural gas dual fuel heavy-duty Diesel-like efficiency using compressed natural gas/diesel
engine[J]. Fuel 2016;167:248e56. dual-fuel combustion[J]. Journal of Energy Resources
[8] Harris Stephen J, Weiner Anita M. Determination of the rate Technology 2016;138(5):52201.
constant for soot surface growth[J]. Combust Sci Technol [21] Huang J, Hill PG, Bushe WK, et al. Shock-tube study of
1983;32(5-6):267e75. methane ignition under engine-relevant conditions:
[9] Karavalakis Georgios, Hajbabaei Maryam, Yu Jiang, et al. experiments and modeling[J]. Combust Flame 2004;136(1-
Regulated, greenhouse gas, and particulate emissions from 2):25e42.
lean-burn and stoichiometric natural gas heavy-duty [22] Gokul Vishwanathan, Reitz Rolf D. Development of a
vehicles on different fuel compositions[J]. Fuel practical soot modeling approach and its application to low-
2016;175:146e56. temperature diesel combustion[J]. Combust Sci Technol
[10] Yoshimoto Yasufumi. Combustion characteristics of a dual 2010;182(8):1050e82.
fuel diesel engine with natural gas (study with fatty acid [23] Fenimore Charles P, Jones George Wallace. Oxidation of soot
methyl esters used as ignition fuels)[R]: SAE Technical Paper. by hydroxyl radicals[J]. J Phys Chem 1967;71(3):593e7.
2010. [24] Kazakov Andrei, Frenklach Michael. Dynamic modeling of
[11] Kalsi Sunmeet Singh, Subramanian KA. Experimental soot particle coagulation and aggregation: implementation
investigations of effects of hydrogen blended CNG on with the method of moments and application to high-
performance, combustion and emissions characteristics of a pressure laminar premixed flames[J]. Combust Flame
biodiesel fueled reactivity controlled compression ignition 1998;114(3-4):484e501.
engine (RCCI)[J]. Int J Hydrog Energy 2017;42(7):4548e60. [25] Bi Xiaojie, Liu Haifeng, Ming Huo, et al. Experimental and
[12] Liu Jie, Yang Fuyuan, Wang Hewu, et al. Effects of pilot fuel numerical study on soot formation and oxidation by using
quantity on the emissions characteristics of a CNG/diesel diesel fuel in constant volume chamber with various
dual fuel engine with optimized pilot injection timing[J]. Appl ambient oxygen concentrations[J]. Energy Conversion Manag
Energy 2013;110:201e6. 2014;84:152e63.
[13] Imran S, Emberson DR, Ihracska B, et al. Effect of pilot fuel [26] Zhao Zhichao, Wu Han, Wang Mianzhi, et al. Computational
quantity and type on performance and emissions of natural investigation of oxygen concentration effects on a soot
gas and hydrogen based combustion in a compression mechanism with a phenomenological soot model of
ignition engine[J]. Int J Hydrog Energy 2014;39(10):5163e75. acetone–butanol–ethanol (ABE)[J]. Energy Fuels
[14] Gokul Vishwanathan, Reitz Rolf D. Application of a semi- 2015;29(3):1710e21.
detailed soot modeling approach for conventional and low [27] Zhao Zhichao, Xu Zhengxin, Liu Jingping, et al. Experimental
temperature diesel combustion–Part I: model performance[J]. and numerical investigation of soot mechanism of acetone-
Fuel 2015;139:757e70. butanol-ethanol (ABE) with various oxygen concentrations
[15] Hou Jie, Yan Fuwu, Lee Timothy H, et al. Computational [R]: SAE Technical Paper. 2015.
investigation on soot mechanism of diesel and diesel/n- [28] Li Menghan, Meng Xiangyu, Jie Hou, et al. A numerical study
butanol blend in constant volume chamber with various on the effects of hot EGR on the operation of natural gas
ambient temperatures[J]. Energy Fuels 2016;31(1):916e31. engine ignited by diesel-butanol blends[R]: SAE technical
[16] Pang Bin, Xie Mao-Zhao, Ming Jia, et al. Development of a paper. 2017.
phenomenological soot model coupled with a skeletal PAH