Journal of KONES Internal Combustion Engines 2003, vol. 10, No 1-2
SIMULATING COMBUSTION AND EXHAUST GAS EMISSIONS IN A DI
DIESEL ENGINE BY USING A CFD CODE COMBINED WITH
DETAILED CHEMISTRY
Jin Kusaka and Yasuhiro Daisho
Dep. of Mechanical Engineering, School of Science and Engineering, Waseda University
3-4-1 Okubo SHINJUKU, Tokyo, JAPAN, Zip:169-8555,
Tel & Fax : +81-3-5286-3917, e-mail : [email protected]
web : http://faculty.web.waseda.ac.jp/jkusaka/index.htm
1. Introduction
Complicated, fast and unsteady phenomena including the temporal changes of the
concentration of chemical species and temperature have been observed during diesel combustion
process. Thus, it is beneficial during the process of the engine development that the diesel
combustion will be reproduced or even predicted by means of CFD (Computed Fluid Dynamics)
codes such as KIVA-3[1] developed by Los Alamos national laboratory in the U.S. With the
extensively advanced sub-models used in these CFD codes concerned with atomization and
evaporation of fuel spray, current sub-models [i.e. Ref. 2] have consistency with the results of
visualization experiments to some extent. Nevertheless, chemical reactions relative to ignition
and combustion processes in the calculation codes are usually expressed by those with one or
several stages, and unfortunately these simplified reactions can’t properly describe the low
temperature oxidation process and the high temperature oxidation process that play important
roles in auto-ignition of fuel. Therefore, if the more detailed chemical reactions are considered,
we will be able to analyze ignition and combustion processes or NOx and PM formation
processes in detail in diesel combustion. Then beneficial information will be given in the engine’s
R&D. Of course, the current chemical kinetic scheme cannot describe the close phenomena
occurs in the cylinder of engines. Moreover, LES (Large Eddy Simulation) in which the variables
are filtered not in time scale but in physical space may be introduced in future. However, it is
assumed very productive in the current situation to simulate and observe the phenomena inside a
cylinder by combining the detailed chemistry with the current CFD code which utilizes the k-ε
type turbulence model.
From a point of view described above, in this study, we tried to conduct a calculation
analysis of diesel combustion by using the CFD code combined with the detailed elementary
reaction processes, and considered the calculation results about ignition and combustion and NOx
and soot formation processes calculation procedure
2. Calculation procedure
2.1 Test Engine
The calculation was conducted to simulate the turbo-charged DI diesel engine with
common-rail fuel injection system that has 7.79 l of total swept volume, inline 6 cylinders and 4
stroke-cycle, manufactured for the medium-duty trucks. The engine specification is shown in Table
Table 1 Engine specifications
4 cycle, OHC, DI
Engine type
Turbo charged,
Common-rail
Number of cylinders
In-line 6
Chamber shape
Reentrant
Bore × stroke mm
115 × 125
Swept volume L
7.790
1. PM was measured by using AVL SPC472 while exhaust gas emissions were measured by using HORIBA
MEXA9100DEGR.
2.2 Calculation Model
This calculation code based on KIVA-3 [1], which was developed in Los Alamos national
laboratory in the U.S, is directly linked with the originally modified subroutines in
CHEMICHIN-II [3], which was developed in Sandia national laboratory. As a model of a spray,
modified Wave Model proposed by Wakisaka et al [4] was employed. In the stratified combustion
field such as diesel combustion, its rate of fuel/air mixing is very slow compared to that of
chemical reactions. As a result, because the fuel oxidation is mainly governed by the mixing rate,
it was necessary to introduce a mixing process into calculation. In the conventional CFD code,
sub-models based on Magnussen model [5] are used. Magnussen model, in which a single-step
oxidation reaction resides, is usually used for the oxidation reaction rate that is controlled by the
mixing before fuel/air mixing reaches molecular level. However, when hundreds of elementary
reactions are considered, because of the existence of many reactions, Magnussen model can’t be
applied. From this point of view, as a model for the interaction between turbulent mixing
processes and gas phase chemical kinetics, the model based on PaSR (Partially stirred Reactor) [6]
concept by Valeri Golovichev et al was introduced into the calculation code.
2.3 Chemical Reaction Scheme
Chemical reaction scheme for hydrocarbon fuel of which carbon number corresponds to that
of light gas oil has not been found yet. Thereupon, in the calculation, we assumed that the
hypothetical fuel had surface tension and vapor pressure equal to those of light gas oil, and its
kinetics adopted those of n-heptane that has the cetane index almost equal to that of light gas oil.
Thus the calculation results on penetration and evaporation characteristics were to become highly
close to those of diesel oil, in the mean time, for the thermal decomposition, oxidation reactions
were considered under control of the rates of reactions of n-heptane. The chemical reactions were
based on the reduced scheme for n-heptane [7], and incorporated N series reactions by Sano et al
[8]. N series reactions includes extended Zeldovich mechanism, Fenimore NO and NO formed
via N2O. Currently, several calculation models for each PAHs, soot formation reactions [i.e. Ref.
9] have been proposed. However, these models included too many elementary reactions to finish
the calculation in an allowable CPU time. Considering CPU time, the simple soot model, in
which olefin hydrocarbons grow into benzene before they finally produce solid soot, was
employed as shown in Fig.1. The total number of elementary reactions was 336, and that of
chemical species was 68.
2.4 Calculation Conditions
Calculation conditions are shown in Table 1. Calculations were conducted on the tenth
measuring point (revolution 60%, load 80%) in Japanese D-13 mode test.
3. Results and discussion
3.1 Pressure Diagram, Rate of Heat Release (RHR)
In this interaction model between turbulence and chemical kinetics, the turbulent characteristics
time-scale, τmix , is defined as follows
H
H
H
+C2H2
C
C C
œ
C
H C C H
C
C2H2
C
H
œ
H
H
C2H3
C
C
H
H
C
n-C4H5
n-C4H4
+C2H2
+C2H2
+H
œ
œ
H
+H
( C
|2H)
6C(S)
Table 1 Calculation conditions
Fuel
n-heptane
H
H
H
|
C
( H
| 2)
Number of revolutions rpm
1440
Injection timing CA deg.
ATDC
-1.1
Injection duration CA deg.
9.5
Dimension
2
Calculation range
deg. ATDC
-40<θ<60
Number of chemical species
68
Number of reactions
332
H
H
C
C
œ
C
3H2
C
Benzene
Phenyl
n-C4H3
H
Fig. 1 Schematic of soot formation model used in
Fig. 5.16
this Soot
workformation
ν
k
(1)
Here, ν and k represent kinetic viscosity and
turbulent energy, respectively. The notification ‘A’
is a model constant as well as a function for the
crank angle. ‘A’ was adjusted in order to agree the
predicted pressure diagram and RHR with the
actual measurement. Fig.2 shows the comparisons
of pressure diagram and RHR between actual
measurement and prediction. Since the calculation
result reproduced the similar tendency of actual
measurement, we will analyze this calculation
result in the next section.
Calculation
C
Experiment
8
6
200
Heat releases J/deg
τ mix = A
4
150
Pressure MPa
H
+H
2
100
Injection
period
0
50
3.2 Ignition and Combustion Process
0
Iso-contours of fuel vapor and HCHO
-40
-20
0
20
40
60
concentrations are shown in Fig.3. Iso-contours of
Crank Angle deg ATDC
CO
(carbon monoxide) and CO2 (carbon
Fig. 2 Comparisons of pressures and RHR
dioxides) concentration, OH concentration and
between experiment and prediction
isotherms are also shown in Fig. 4. To simplify the
explanation of the calculation results, the discussion
is firstly focused on the mixture that evaporates. As shown in Fig.3, fuel evaporated to form rich
mixture at the crank angleθ=4 deg. ATDC. This mixture was decomposed to HCHO during the
low-temperature reaction. Thus, HCHO was formed in the same region where the high
concentration fuel vapor was observed at the same crank angle. After that, this mixture penetrated
the surroundings accompanied by its progressing
chemical kinetics. Since CO was also formed
during low temperature reactions after the HCHO
formation, which was observed at θ=4 deg
ATDC in Fig.3, a little CO was also observed at
the same crank angle in Fig.4. Finally,
CO2 ,which was formed after being converted
from CO observed at θ=4 deg ATDC, appeared
at θ=8 deg ATDC. It is suggested from Fig.4
that the slight temperature increase was also
observed in the place where CO2 started being
formed during high temperature reaction. At this
time, OH was also formed. This sequential
phenomena of the fuel vapor and the HCHO and
CO formation at θ=4 deg ATDC followed by
the CO2 and OH formation and temperature
increase at θ=12 deg ATDC continued to occur
one after another. For example, the result in vapor
and HCHO and CO formation at 8 deg ATDC
and CO2, OH formation and high temperature
region at 12 deg ATDC was in the same sequence
as the above.
3.3 Formation Processes of NO and NO2
Results of concentrations of NO and NO2 are
shown in Fig. 5. When the initial stage of
combustion was observed at crank angles like θ
=12, 16, deg. ATDC, NO was formed around the
same area as where high OH concentration was
observed in Fig.4. At θ=20 deg. ATDC, its band
shaped formation area with high concentration
more than 1000 ppm of NO appeared, and the
area of NO formation diffused in the cylinder. On
the other hand, NO2 appeared around the NO
formation area. NO2 was initially formed after
being converted from NO. Thus, formed NO was
diffused in the lean area and it was converted to
NO2 there. However, the concentration of NO2
itself was less compared to that of NO.
deg.
ATDC
0
4
8
12
16
20
24
28
32
0
Fuel vol%
2
0
HCHO ppm 3000
Fig.3 Iso-contours of Fuel vapor and HCHO
3.4 Evaluation of NO(Nitric Mono Oxide)
in the cylinder
Formation Paths
As described above, in diesel combustion, it has the wide window of excess air ratios
ranging from rich to lean exists, and variation becomes transitional at higher speed. Moreover,
temperature and amount of the chemical species also change. Therefore various NO formation
32
paths should be considered. In truth, 41 elementary reactions concerned with NO formation and
destruction were taken into account in our model. Thus, the simple technique as described below
was used accordingly to estimate the amount of NO formed through three different paths. Here,
NO formation path was classified into following three paths, thermal NO, Fenimore NO and NO
via NO2. In the analysis, it was assumed that these were independent of one another. Needless to
say, though each path isn’t actually independent, the rough dependence on each formation path
became known through this simple method. Calculations were conducted individually using the
reaction relative to NO formation only in each scheme, and the results for amount of NO were
compared to that of full scheme including all reactions relative to NO formation. The history of
NO formation and the frozen NO values for each path are shown in Figs.6 and 7 respectively. It is
clear from these figures that NO formation in diesel combustion is mainly governed by thermal
NO described by the extended Zeldovich mechanism. The NO that was generated through
thermal NO accounted for 88% of total NO. Fenimore NO and NO via N2O accounted for 7%
and 1.5%, respectively
3.5 Formation Process of Soot
As shown in Fig.1, there are mainly three paths for soot (Solid Carbon; C(S)) formation in the
0
CO ppm
deg.
ATDC
deg.
ATDC
0
0
4
4
8
8
12
12
16
16
20
20
24
24
28
28
32
32
40000 0
CO2 vol%
10
0
OH ppm
40
600
Fig.4 Iso-contours of CO, CO2, OH and isotherm in the cylinder
Temp. K
2200
deg.
ATDC
4
Thermal NO
NO mg/(cyc.cyl)
0
1.0
0.8
Full scheme
0.6
0.4
Fenimore NO
0.2
0
-40 -30 -20 -10
0
10
NO via N2O
20
30
Crank Angle deg. ATDC
8
12
40
Fig.6 Formation histories of NO in the
cylinder by using each scheme
NO via N2O (1.5%)
Others (3.5%)
Feniomore NO (7%)
16
20
24
Thermal NO (88%)
Fig.7 Fractions of frozen NO value in each
scheme reaction path
present chemical reaction scheme. Distributions
of the concentration for olefin hydrocarbon C2H2
28
(acetylene), C6H6 (benzene), soot, and O2 in the
cylinder are shown in Fig. 8. At crank angles 8
deg. ATDC, C2H2 was observed in the core
region of fuel spray where the oxygen
32
concentration became lower. After that, the
acetylene started to disappear after θ=12 deg.
ATDC. At crank angles from 8 to 16 deg. ATDC,
aromatic hydrocarbon benzene and soot were
0
NO ppm 1000 0
4
NO2 ppm
apparently observed in the core region of fuel
spay, which existed at the C2H2 formation region
Fig.5 Iso-contours of NO and NO2 formed
during combustion process
observed at 4 deg. ATDC. This can be explained
by the fact that C2H2 was polymerized to be
converted to soot. After θ=16 deg. ATDC, soot was decomposed, and the mass concentration
became low due to the increased cylinder volume. The histories of formations of soot and C2H2
as the olefin hydrocarbon are shown in Fig. 9. Amount of soot increased rapidly during the initial
combustion stage, and at 20 deg. ATDC the formation decreased, then it increased again. Around
60 deg. ATDC, the reactions in soot formation froze. However, the further extensive study to find
more accurate oxidation reactions in soot formation and decomposition are necessary to predict
soot formation since calculation result was ten times more than the actual measurement, which
was obtained by using the micro dilution PM sampling system.
4. Conclusions
deg.
ATDC
0
4
8
12
16
20
24
28
32
0
C2H2 ppm
700 0
C6H6 ppm
2×10-3
0
3
Soot g/cm
-5
3×10
3
O2 vol%
20
Fig.8 Iso-contours of C2H2, C6H6, Soot, and O2 in the cylinder
Simulation of the ignition and combustion processes of a DI diesel engine with a
supercharger and a common rail fuel injection system were tested under 60% of engine speed and
80% of load condition using the multi-dimensional model combined with detailed chemistry. The
conclusions were drawn as follows
Heat release J/deg
C2H2, soot mg/(cyc.cyl)
(1) The fuel directly injected into the cylinder is decomposed to HCHO and CO as it penetrates
inside the cylinder during the low temperature reaction. After that, the high temperature reaction
occurs when OH and CO2 are observed.
(2) NO is formed immediately around the fuel rich area after 16 deg. ATDC, and diffused in the
combustion chamber. This NO is partially converted into NO2 in the outer area of NO formation
region.
2.5
200
(3) 88% of NO formed in diesel combustion is
governed by the expanded Zeldovich mechanism.
2
7% and 1.5% NOs are formed through Fenimore
150
Soot
NO and NO via N2O, respectively.
1.5
(4) In the mechanism applied in this study, the
100
quantitative soot prediction was hardly achieved
Soot
1
(experiment)
since the calculation result was ten times more
RHR
50
than the actual result. More detailed oxidation
0.5
C2H2
reactions of soot are necessary. However, there is
a possibility to predict where and when soot is
0
0
-40
-20
0
20
40
60
formed, by considering a simplified soot
Crank angle deg ATDC
formation kinetic model.
Fig.9 Formation histories of Soot and C2H2
References
[1] Anthony A Amsden, KIVA-3:A KIVA Program with Block-Structured Mesh for Complex
Geometries, LA-12503-Ms, 1993 Anthony A Amsden, KIVA-3:A KIVA Program with
Block-Structured Mesh for Complex Geometries, LA-12503-Ms, 1993
[2] T. F. Su, M. A. Paterson, R. D. Reitz, P. V. Farrel, Experimental and Numerical Studies of
high Pressure Multiple Injection Sprays, SAE960861, 1996
[3] R. J. Kee, F. M. Rupley J. A. Miller : CHEMKIN-Ⅱ A Fortran Chemical Kinetics Package
For The Analysis of Gas-phase Chemical Kinetics, SAND89-8009, 19893
[4] Tomoyuki Wakisaka and Kunikazu ibaraki, An Improved Droplet Breakup Model for Three
Dimensional Diesel Spray Simulation, Proc. 3rd KSME-JSME thermal engineering
conference, II-167-172, KSME and JSME, 1996
[5] Magussen, B.E. and Hjertager, B.H., On mathematical modeling of turbulent combustion
with special emphasis on soot formation and combustion, Proc. 16th Symp.(inter.) on
combustion, 719-729, The combustion Institute, 1976
[6] Golovitchev,V.I., Tao, F., Chomial, L., 3-D Diesel spray Simulation Using a New Detailed
Chemistry Turblent COmbustion Using a New Detailed Chemistry Turbulent Combustion
Model’, SAE01-3552(1999)
[7] Valeri Golovichev, http://www.tfd.chalmers.se/~valeri/
[8] Tsutomu Kato, Taeko Sano, NOx formation in Methanol flame, Bulletin of Tokai University,
Vol. 32, No. l, 1992, pp. 223-234
[9] N. M. Marinov et al., “Modeling of Aromatic and Polycyclic Aromatic Hydrocarbon
Formation in Premixed Methane and Ethane Flames”, Combust. Sci. and Tech., Vol.116-117,
1996, 211-287