Fuel Chemistry Models for Simulating
Gasoline Kinetics in Internal Combustion
Engine Applications
31st International
Symposium on
Combustion
PrincetonUniversity
Zhenwei Zhao, Marcos Chaos, Andrei Kazakov,
Ponnuthurai Gokulakrishnan, Michele Angioletti, Frederick L. Dryer
August 6-11, 2006
Mechanical & Aerospace
Engineering Department
[email protected]
http://www.princeton.edu/~combust
University of Heidelberg,
Germany
Toluene selection and C/H ratio
Introduction
™ The overall C/H ratio of complex surrogates has typically not been
scrutinized so as to align with the C/H ratio of the gasoline of interest.
™Motivation: develop chemical kinetic sub-models that can properly
describe the important attributes of gasoline direct injection (GDI), and
™ Gasoline C/H ratios vary considerably by source, even though the
Research Octane Number (RON), Motored Octane Number (MON),
and Road Octane number, of the various gasoline compositions may be
nearly the same.
homogeneous charge compression ignition (HCCI) combustion
processes.
™ Practical fuels, such as gasoline, are mixtures of very large numbers of
different hydrocarbons ― extensive efforts have been focused on
™ The C/H mass ratio of PRF mixtures is functionally dependent on
octane number. Mass ratio values lie between 5.213 (PRF 0) and 5.296
(PRF 100). Those for gasoline samples are typically considerably
larger.
studying the kinetic behavior of single component or simple mixtures
of components that can represent gasoline.
™ PRF mixtures (primary reference fuels; n-heptane and iso-octane) are at
the foundations of rating gasoline octane number; however, PRF
™ The adiabatic equilibrium
flame temperature of a
2330
10
fuel/air mixture is not only a
2320
9
strong function of the
2310
8
mixture equivalence ratio,
2300
7
but also of the C/H mass
2290
6
ratio of the fuel (see Fig. 1).
2280
It is important, thus, to match
2270
5
0
20
40
60
80
100
this parameter, in addition to
% Toluene in Fuel (molar)
others
that
reflect
Figure 1. Calculated equilibrium reaction
temperature for a PRF 87 mixture with
autoignition properties by
added toluene to adjust the overall PRF+1
selecting a proper additive to
C/H mass ratio.
PRF components.
11
physical/chemical properties can behave significantly different in
homogenous charge, spark ignition engines than gasoline itself.
Tad (K)
™ More complex mixtures of individual species representing different
C/H mass ratio
2340
molecular groups have been suggested as surrogate fuels for gasoline.
As a result of the number of initial molecular types, the dimensional
character of surrogate kinetic models becomes large and the additional
mechanism parameters become quite uncertain.
™ The kinetic validation process itself becomes significantly more
complex as fewer experimental data are available for validating the
mechanism performance for the additional pure components as well as
the mixtures.
™ Toluene is a simple aromatic that appears in gasoline and that can be
used in test mixtures both to match C/H ratios and adjust over
autoignition properties. Of possible alternatives to add to PRF
components, it has also received the most detailed kinetic modeling
attention.
™ In this work, we have been driven by the above issues to consider
approaches that limit the number of species to three, including the
original PRF components. To this end we use PRF+toluene mixtures
(hereto forth named PRF+1) as surrogate fuels for gasoline combustion
research.
™ Toluene oxidation exhibits no negative temperature coefficient (twostage ignition) kinetic behavior. The octane behavior of a PRF+1
mixture that replicates C/H ratio can be independently adjusted by
changing the ratio of n-heptane and iso-octane without varying the
toluene molar fraction.
™ Experiments have been performed on PRF+1 mixtures with appropriate
C/H ratios to match real gasoline in order to achieve an optimal mixture
composition. Concurrently a minimized/optimized chemical kinetic
model has been developed to best match the behavior of PRF+1
This work has been supported by
HONDA R&D Co., Ltd., Japan
Model Formulation
™ Initial development started from a baseline iso-octane mechanism
(Curran et al. [1]) ― Minimized and optimized using methods similar
to those in [2] ― the model was shown to reproduce well autoignition
and oxidation characteristics of pure iso-octane
™ A suitable combination of minimized versions for iso-octane and nheptane (derived from [3]) was identified, and the resulting combined
mechanism was tested and optimized to reproduce pure component
(PRF=0 and PRF=100) combustion characteristics as well as those for a
PRF 90 (including laminar flame speeds).
™ No suitable toluene submodel was identified that could properly couple
with the Min/Opt PRF model. Research conducted previously in our
laboratory has demonstrated the inability of several public-domain
toluene oxidation reaction models (i.e., [4-7]) to predict high-pressure
oxidation data taken in the VPFR.
™ We have developed an in-house sub-mechanism for toluene oxidation,
on current (and, still, incomplete) level of conceptual understanding of
the oxidation of aromatic single ring species. The redevelopment
process included several steps:
¾ Basic mechanism structure for main species (toluene, benzyl,
benzaldehyde, benzene, and their derivatives). Appropriate reaction
rate coefficients were assigned based on an extensive up-to-date
literature review.
¾ Thermochemistry was taken from the Burcat [8] and LLNL
databases, where possible.
¾ Cyclic C5 sub-mechanism added based on implementations by
Alzueta et al. [9] and Djurisic et al. [10].
Updated
cyclopentadienyl radical thermochemistry from Roy et al. [11] and
Kiefer et al. [12] were adopted.
™ The resulting set was then integrated with the PRF Min/Opt model
(linking it to C3, C2, and C1 intermediates and final products) producing
the PRF+1 mechanism reported here (adding an additional 22 species
and 102 reactions for a total of 469/1221 species/reactions).
mixtures at pure toluene, n-heptane, and iso-octane content, as well as
at the optimal PRF+1 composition. The present work provides
™ The presence of toluene in PRF mixtures, however, scavenges radical species that are produced in the reaction of alkanes at low and intermediate
temperatures, and therefore suppresses negative temperature coefficient behavior of a PRF+1 mixture relative to the PRF mixture without toluene.
Toluene has a higher hot ignition temperature than either PRF component, and thus addition of toluene to PRF mixtures will also affect hot ignition
characteristics.
additional contributions to improve the database on pure components,
mixtures, and gasoline autoignition and oxidation kinetic behavior for
optimizing and validating models by comparisons with predictions
Validation
Pure Toluene
Flow Reactor (12.5 atm, φ=1)
0.012
Mole Fraction
2
0.008
O2 x 0.5
CO
CO2
H2 O
0.004
™ Figure 2 shows model-predicted toluene/O2 reactivity against VPFR data. The model clearly captures the lack of NTC chemistry shown experimentally although it predicts a slightly
higher reactivity as temperature increases.
™ Figure 3 compares predictions of species time histories for the conditions shown in Fig. 2 at 920 K. Major species profiles are well reproduced. The agreement for intermediates
(benzene, formaldehyde, not shown) is somewhat less satisfactory and further work needs to be done to determine alternative channels for these species. It should be pointed out,
however, that the present agreement is a dramatic improvement over that found using the models of Refs. 4-6.
™ Agreement with the latest low– and high–pressure toluene/air ignition data [13, 14] is excellent (Fig. 4).
™ Flame speeds of toluene/air mixtures are faithfully reproduced by the model at all equivalence ratios (Fig. 5).
Binary Mixtures
0
650
750
850
950
T (K)
6
7
90/10 n-C7H16/C6H5CH3, 12.5 atm, φ=1, 670 ppm
90/10 i-C8H18/C6H5CH3, 12.5 atm, φ=1, 620 ppm
0.004
0.004
Mole Fraction
Mole Fraction
3
T = 920 K
O2 x 0.5
C6H5CH3 x 3
CO
CO2
H2O
0.008
Mole Fraction
0.004
0.012
O2 x 0.5
CO
CO2
H2O
0.002
0
0
550 600 650 700 750 800 850
T (K)
0
0.8
1.2
Time (s)
1.6
2
4
Ignition Delay (μs)
0.006
0.005
Mole Fraction
100
Data [13]
norm. to φ=1, 9% O2, 1 atm
Data [14]
norm. to φ=1, 50 atm
Model
O2 x 0.5
CO
CO2
H2O
CH2O x 20
0.003
0.002
0.6
0.7
0.8
0.9
1
550 600 650 700 750 800 850 900
T (K)
1000/T (1/K)
Laminar Flame Speed
τign (μs)
Flame Speed (cm/s)
1000
Davis et al. [15]
20
Data [16] norm. to 20 atm
PRF+1 Model (20 atm)
Data [16] norm. to 55 atm
PRF+1 Model (55 atm)
Model
15
0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
Equivalence Ratio, φ
0
8
SURROGATE B
φ = 1 in Air
30
25
0.002
0
100
0.85
0.5
1
1.5
Time (s)
9
2
Laminar Flame Speed
Shock Tube Ignition Delay
5
35
0.004
0.001
0
0.5
C6H5CH3 x 25
n-C7H16 x 25
i-C8H18 x 10
O2 x 0.5
H2O
CO
CO2 x 3
T = 820 K
0.004
10
Summary
™ A toluene oxidation submechanism has been produced that can properly
reproduce available experimental data.
™ The toluene model has been combined with a PRF model to generate a minimized
and optimized model (PRF+1) to simulate gasoline combustion kinetics.
™ The PRF+1 model performs well against experimental data from shock tubes,
laminar flames, and new flow reactor data collected as part of the present effort.
™ Comparison in all of the various experimental configurations are significant
improvements over other models presently in the literature, especially for
mixtures involving toluene.
Flow Reactor (12.5 atm, φ=1)
Shock Tube Ignition Delay
1000
Mole Fraction
0.4
550 600 650 700 750 800 850 900
T (K)
Ternary Mixtures
To = 400 K (19.68% EGR N2)
To = 400 K
To = 550 K
100
6.
7.
8.
9.
10.
80
60
40
20
0
0.9
0.95
1
1.05
1000/T (1/K)
1.1
1.15
1.
2.
3.
4.
5.
120
Flame Speed (cm/s)
0
O2 x 0.5
CO
CO2
H2O x 0.5
0.002
™ Figures 6 and 7 show PRF+1 model prediction against n-heptane/toluene and isooctane/toluene reactivity data collected in the VPFR. As with the ternary mixtures shown
below, very good agreement is observed for major species and the NTC region is
properly reproduced (Fig. 6). In addition, although not shown, characteristic reaction
time scales and dependences on equivalence ratio are properly captured by the model.
™ Figures 8-11 show comparison of model predictions against data from different
experimental systems for ternary mixtures (i.e. PRF+toluene). Figures 8, 9, and 11 show
data collected in our laboratory for a mixture emulating a gasoline with ON=84.
™ Of especial importance are the data shown in Fig. 10. These data appeared in the open
literature after the development of the present model and, although the model was not
optimized against this set, very good agreement can be observed.
10
0.6
0.8
1.0
1.2
Equivalence Ratio, φ
1.4
11
11.
12.
13.
14.
15.
16.
H.J. Curran; P. Gaffuri; W.J. Pitz; C.K. Westbrook, Combust. Flame 129 (2002) 253-280.
Z. Zhao; J.P. Conley; A. Kazakov; F.L. Dryer, SAE paper 2003-01-3265.
H.J. Curran; P. Gaffuri; W.J. Pitz; C.K. Westbrook, Combust. Flame 114 (1998) 149.177.
S.D. Klotz; K. Brezinsky; I. Glassman, Proc. Combust. Inst. 27 (1998) 337-344.
W.J. Pitz; R. Seiser; J.W. Bozzelli; I. Da Costa; R. Fournet, F. Billaud; F. Battin-Leclerc; K. Seshadri; C.K.
Westbrook, 2nd Joint Meeting, US Sections of the Combust. Inst., Paper 253 (2001).
P. Dagaut; G. Pengolan; A. Ristori, Phys. Chem. Chem. Phys 4 (2002) 1846-1854.
R. Bounaceur; I. Da Costa; R. Fournet; F. Billard; F. Battin-Leclerc, Int. J. Chem. Kinet. 37 (2005) 25-49.
A. Burcat, “Ideal Gas Thermochemical Database”, ftp.technion.ac.il.
M.U. Alzueta; P. Glarborg; K. Dam-Johnasen, Int. J. Chem. Kinet. 32 (2000) 498-522.
Z.M. Djurisic; A.V. Joshi; H. Wang, on”, 2nd Joint Meeting, US Sections of the Combust. Inst., Paper 238
(2001).
K. Roy; M. Braun-Unkhoff; P. Frank; Th. Just, Int. J. Chem. Kinet. 33 (2000) 821-833.
J.H. Kiefer; R.S. Tranter; H. Wang; A.F. Wagner, Int. J. Chem. Kinet. 33 (2000) 834-845.
V. Vasudevan; D.F. Davidson; R.K. Hanson, Proc. Combust. Inst. 30 (2005) 1155-1163.
D.F. Davidson; B.M. Gauthier;R.K. Hanson, Proc. Combust. Inst., 30 (2005) 1175-1182.
S.G. Davis; H. Wang; K. Brezinsky; C.K. Law, Proc. Combust. Inst. 26 (1996) 1025-1033.
B.M. Gauthier; D.F. Davidson; R.K. Hanson, Combust. Flame 139 (2004) 300-311.