Sei sulla pagina 1di 8

Detailed Kinetic Modeling of Benzene and Toluene Combustion

Zoran M. Djurisic, Ameya V. Joshi, and Hai Wang


Department of Mechanical Engineering, University of Delaware, Newark, DE 19716
Introduction
The high-temperature reaction kinetics of
benzene and toluene oxidation is of fundamental
and practical importance in combustion. While
significant advances have been made in the
understanding of the oxidation kinetics of
aliphatic hydrocarbon combustion, comparatively
fewer studies have been conducted on aromatics
[1,2]. Such a disparity in emphasis is expected
because of the significantly more complex nature
of the aromatics' kinetics. It is also obvious that
since most practical fuel blends consist of large
amounts of aromatics, satisfactory modeling and
manipulation of the combustion processes would
not be possible without a quantitative description
of their oxidation kinetics. This concern is
further substantiated by recognizing the role of
aromatic kinetics in engine knock [3], soot
production [4-12], combustion emissions of
polycyclic aromatic carcinogen [13] and fullerene
synthesis [14].
Previously, Brezinsky and coworkers [2]
proposed a detailed kinetic model of benzene and
toluene oxidation. The model was developed on
the basis of thermochemical considerations and
flow-reactor experiments.
Several expanded
kinetic models have been proposed since the
work of Brezinsky et al. [2]. In general, these
models were developed for and tested against
data collected in a single type of experiment only.
Specifically, Lindstedt and Skevis [15], Zhang
and McKinnon [16], and Tan and Frank [17] have
independently published reaction models and
tested these models against the species profiles in
a burner-stabilized benzene flame of Bittner and
Howard [18]. There have also been reports of
simulation studies on the laminar flame speeds of
benzene and toluene using detailed reaction
kinetics and species transport [17, 19, 20]. We
have seen that these reaction models predicted the
individual experimental data reasonably well.
However, a comprehensive model capable of
predicting all experimental data is still not
available. Undoubtedly, this situation manifests
the significant uncertainties in the reaction
*

kinetics of aromatic compounds.


Our work has been largely motivated by the
need of a unified and physically justifiable
reaction model to describe one-ring aromatics'
combustion. Previously, we proposed [21,22] a
reaction model for benzene and toluene
combustion and tested this model, for the first
time, against selected experimental data obtained
under a variety of combustion conditions,
including the oxidation of benzene and toluene
in a flow reactor, a stirred reactor, premixed
flames and shock tubes. In the present work, we
compared the model predictions to a wider
variety of experimental data. Key uncertainties
of the model were systematically identified.
Methodologies
The current model of benzene and toluene
combustion consists of 65 species and 340
elementary reactions. The reaction kinetics of
C1 and C2 species is based on GRI-Mech 1.2
[23]. For larger species the reaction pathways
and rate coefficients were mostly obtained from
literature data and compilations [e.g., 24].
Selected rate parameters were analyzed using the
RRKM method [e.g., 25,26].
Calculations of the ignition delay were
carried out using the Sandia Chemkin-II [28-31]
codes.
The back rate coefficients were
computed via equilibrium constants.
The
thermochemical data of cyclopentadiene and its
derivatives was obtained from quantum chemical
calculations [27], other data was taken from
Refs. 32 and 33.
Results
We shall first present the modeling results
for global combustion properties namely the
laminar flame speed and shock-tube ignition
delay. This is followed by a comprehensive
comparison between model and experiment for
detailed species profiles during benzene and

Presented at the Second Joint Meeting of the U.S. Sections of the Combustion Institute, Oakland, CA, March 25-28, 2001.

toluene oxidation in a flow reactor and in a


laminar premixed burner-stabilized flame.
Figure 1 shows the experimental and
computed laminar flame speeds of benzene and
toluene at the atmospheric pressure. It is seen
that the flame speed as predicted by the model is
in close agreement with the experimental data for
both benzene and toluene and over a wide range
of equivalence ratios.
In Figure 2, the computed ignition delay
times are compared to the experimental data of
Miyama [34], collected behind reflected shock
waves for 7 mixtures. The fuel-to-oxygen ratio
spans from 0.0256 to 0.379. It is seen that the
model predicts quite well the temperature
dependence of the ignition delay for all mixtures.
The model, however, consistently yielded
ignition delay times larger than those of the
experimental for mixtures with smaller fuel to
oxygen ratio. At the fuel-to-oxygen ratio of
0.0256, the model produced ignition delay times
larger than the experimental counterpart by a
factor of two. As this ratio is increased to 1/9 and
beyond, the experimental data are well
reproduced by the current reaction model.
Similar trend was found for the model
prediction of Burcat et al.s experimental data
[35], as seen in Figure 3. Specifically, our model
overpredicted the ignition delay times for
mixtures with the fuel-to-oxygen ratios smaller
than 0.066, but it reproduced the experimental
results for higher fuel-to-oxygen ratios. Figure 4
presents the comparison between model
predictions and the experimental data of
Thyagarajan and Bhaskaran [36]. It is seen that
the temperature dependence exhibited by the
model is more pronounced than the experiment.
It is most likely that the discrepancy is the result
of experimental inaccuracy because the model
predicted the temperature dependence for all
other mixtures as seen in Figures 2 and 3.
Figure 5 shows the ignition delay times of
four toluene-oxygen-argon mixtures, comparing
the present modeling results to the experimental
data [35]. In all cases, the agreement between
model and experiment is excellent.
Figure 6 shows selected species profiles
experimentally measured [2] and computed for
benzene oxidation in an atmospheric-pressure
flow reactor and under the fuel-lean condition
with the equivalence ratio of 0.65 and the
temperature of ~1108 K. The current model
reproduced the entire mole fraction profiles of
major species, including benzene, oxygen, and

carbon monoxide. A very good agreement was


also found for the concentration profile of
cyclopentadiene.
For species with mole
fractions at a few 100 ppm level, the model
predictions are accurate to within a factor of ~2.
In Figure 7, we present the similar comparison
for toluene oxidation at the equivalence ratio of
1.33. Again the mole fraction profiles of major
species are well reproduced by the model. For
minor species, however, the disagreement
between model and experiment is seen to be
significant. Notably the concentration of phenol
was severely underpredicted.
Figure 8 shows detailed mole fraction
profiles of selected species in a burnerstabilized, low-pressure benzene-oxygen-argon
flame. In general, the major species profiles
were extremely well reproduced by the current
reaction model. Similar agreement was found
for major radical species, H and OH. For a
variety of minor species, the model is capable of
reproducing their peak concentrations to within a
factor of 2 to 3. There are, however, a number
of species for which we found significant
disagreement between the model and
experiment.
These species include the
cyclopentadienyl radical, cyclopentadiene and
phenol. In all cases, the model severely over
predicts the experimental data. The modeling
results also show that for some minor species the
predicted mole fractions may be significantly
lower than measured values as in the case of
styrene, indicating that certain reaction channels
are missing in the model.
We verified the model against several other
sets of flame data, including the more recent
burner-stabilized flame data of Tregrossi et al.
[37]. These are fuel-rich, atmospheric-pressure
flames. The species concentration profiles were
determined by probe sampling and Gas
Chromatography. In general the model
reproduced quite well the measured species
profiles. The largest discrepancy is again seen in
the
concentrations
of
phenol
and
cyclopentadiene. In both cases, the model
predicts substantially larger mole fractions than
the experiment.
The model was also tested against the
benzene oxidation data collected in a stirred
reactor [38]. While the concentrations of the
major species were well reproduced, the model
over-predicts the phenol concentrations by as
much as a factor of 5.

Discussion

Summary

The modeling results presented in the


preceding section reveal that there are a number
of important issues concerning the current
reaction model. First, the reaction mechanism
seems to capture the main feature of the oxidation
processes for both benzene and toluene.
Specifically, the model reproduced quite well key
combustion properties including the laminar
flame speed and shock-tube ignition delay times.
The model is also capable of predicting the
concentration profiles of major species during
benzene and toluene oxidation under quite
different reaction conditions. There are, however,
several persistent problems. The first is that the
ignition delay times for mixtures with small fuelto-oxygen ratios are overpredicted. Secondly, the
model does not reproduce the concentration
profiles of phenol in all experiments. The degree
of disagreement appears to be dependent on the
reactor conditions, and more specifically on the
extent of back mixing, as seen in Figure 9. A
greater extent of back mixing causes the model to
more severely overpredict the experimental data.
We have made several attempts to isolate the
problem of ignition delay predictions from that of
the phenol problem. Specifically, it was thought
that the cause for overpredicting the ignition
delay times for mixtures of small fuel-to-oxygen
ratio is a missing radical-chain initiation reaction,
which involves a 1,2-H shift in benzene to form a
singlet carbene species followed by its reaction
with molecular oxygen.
Quantum chemical
calculations, however, suggest that this reaction
channel is not viable. Extensive sensitivity
analyses performed in the present study suggest
that it is very likely that the discrepancy in the
ignition delay times and the problem of phenol
predictions have the same cause, namely, the
kinetic uncertainty in the pathways of the reaction
between benzene and the oxygen atom. Our
analysis suggests that to reconcile the
discrepancy between model and experiment this
reaction cannot have a large rate producing
phenol. Regardless, a further development of the
current reaction model will have to rely on a
more accurate knowledge about the detailed
kinetics of this reaction.

A detailed reaction model is proposed in this


work with the goal to develop a comprehensive
model, which is capable of predicting a variety
of combustion data of benzene and toluene. The
model was validated against a large set of
experimental data. It was found that the reaction
mechanism captured the main features of the
oxidation processes for both benzene and
toluene. Notable discrepancies between model
and experiment were found to be the ignition
delay times of mixtures of small fuel-to-oxygen
ratio and the concentrations of phenol. These
discrepancies appear to originate from the same
kinetic uncertainty, namely the kinetics of the
reaction between benzene and the oxygen atom.
Acknowledgement. The authors wish to thank
Professor Lisa Pfefferle for helpful discussion. The
work was supported by the National Science
Foundation under Grant CTS9874768 and by the Air
Force Office of Scientific Research under Grant
F496200110144.

References
(1) Brezinsky, K. Prog. Energy Combust. Sci.
1986, 3, 1.
(2) Emdee, J. L.; Brezinsky, K.; Glassman, I. J.
Phys. Chem. 1992, 96, 2151.
(3) Sawyer, R. F. Proc. Combust. Inst. 1992,
24, 1423-1432.
(4) Haynes, B. S.; Wagner, H. Gg. Prog.
Energy Combust. Sci. 1980, 7, 229.
(5) Calcote, H. F. Combust. Flame 1981, 42,
215.
(6) Homann, K. H. Proc. Combust. Inst. 1984,
20, 857-870.
(7) Bittner, J. D.; Howard, J. B. in Soot in
Combustion Systems and its Toxic
Properties; Lahaye, J.; Prado, G. Eds.;
Plenum: New York, 1983.
(8) Bockhorn, H.; Fetting, F.; Wenz, H. W. Ber
Bunsenges. Phys. Chem. 1983, 87, 1067.
(9) Frenklach, M.; Warnatz, J. Combust. Sci.
Technol. 1987, 5, 265.
(10) Frenklach, M.; Wang, H. Proc. Combust.
Inst. 1991, 23, 1559-1566.
(11) Howard, J. B. Proc. Combust. Inst. 1991,
23, 1107-1127.
(12) McKinnon, J. T.; Howard, J. B. Proc.
Combust. Inst. 1992, 24, 965-971.
3

(13) Longwell, J. P. Proc. Combust. Inst. 1982,


19, 1339-1350.
(14) Howard, J. B., Proc. Combust. Inst. 1992,
24, 933-946.
(15) Lindstedt, R. P.; Skevis, G., Combust Flame
1994, 99, 551.
(16) Zhang, H.-Y.; McKinnon, J. T. Combust.
Sci. Technol. 1995, 107, 261.
(17) Tan, Y. W.; Frank, P. Proc. Combust. Inst.
1996, 26, 677-684.
(18) Bittner, J. D.; Howard, J. B. Proc. Combust.
Inst. 1980, 18, 1105-1116.
(19) Davis, S. G.; Wang, H.; Brezinsky, I.; Law,
C. K. Proc. Combust. Inst. 1996, 26, 10251033.
(20) Lindstedt, R. P.; Maurice, L. Q. Combust.
Sci. Technol. 1996, 120, 119.
(21) Wang, H.; Djurisic, Z. M., in Chemical and
Physical Processes of Combustion, the 1999
Fall Technical Meeting of the Eastern States
Section of the Combustion Institute,
Raleigh, NC, October, 1999, pp. 137-140.
(22) Djurisic, Z. M. MS. Thesis, University of
Delaware, Newark, DE, 1999.
(23) Frenklach, M.; Wang, H.; Goldenberg, M.;
Smith, G. P.; Golden, D. M.; Bowman, C.
T.; Hanson, R. K.; Gardiner, W. C.;
Lissianski, V. GRI-MechAn Optimized
Detailed Chemical Reaction Mechanism for
Methane Combustion; GRI Technical Report
No. GRI-95/0058, November 1, 1995.
(24) Baulch, D. L.; Cobos, C. J.; Cox, R. A.;
Frank, P.; Hayman, G.; Just, TH.; Kerr, J.
A.; Murrells, T.; Pilling, M. J.; Troe, J.;
Walker, R. W.; Warnatz, J. Combust. Flame,
1994, 98, 59.
(25) Wang, H.; Frenklach, M., J. Phys. Chem.
1994, 98, 11465.
(26) Wang, H.; Frenklach, M., Combust. Flame
1997, 110, 173.
(27) Wang, H; Brezinsky, K. J. Phys. Chem.
1998, 102, 1530.
(28) Kee, R. J.; Rupley, F. M.; Miller, J. A.,
Sandia Report SAND 89-8009B; Sandia
National Laboratories: Albuquerque, New
Mexico, 1989.
(29) Kee, R. J.; Grcar, J. F.; Smooke, M. D.;
Miller, J. A., Sandia Report SAND85-8240
UC4; Sandia National Laboratories:
Albuquerque, New Mexico, 1985.
(30) Kee, R. J.; Rupley, F. M.; Miller, J. A.,
Sandia Report SAND 89-8009B; Sandia
National Laboratories: Albuquerque, New
Mexico, 1989.

(31) Kee, R. J.; Grcar, J. F.; Smooke, M. D.;


Miller, J. A., Sandia Report SAND85-8240
UC4; Sandia National Laboratories:
Albuquerque, New Mexico, 1985.
(32) Wang, H.; Frenklach, M., J. Phys. Chem.
1993, 97, 3867.
(33) Burcat, A.; McBride, B. 1997 Ideal Gas
Thermodynamic Data for Combustion and
Air-Pollution Use; Technion Aerospace
Engineering (TAE) Report # 804, 1997.
(34) Miyama, H. J. Chem. Phys. 1970, 52, 38503851.
(35) Burcat, A.; Snyder, C.; Brabbs, T., NASA
Technical Memorandum 87312, 1986.
(36) Thyagarajan, K.; Bhaskaran, A. Int.
J.Energy Res. 1991, 15, 235-248.
(37) Tregrossi, A.; Ciajolo, A.; Barbella, R.
Combust. Flame 1999, 117, 553-561.
(38) Chai, Y.; Pfefferle, L. D. Fuel 1998, 77,
313.

Lam inar Flame Speed (cm/s)

45

40

Benzene

35

10

0.419% C H - 12.57% O - 87.011% Ar


6

Toluene

30

10

10

20

10
0.60

p = 2.0 atm
5

0.9

1.0

1.1

1.2

1.3

1.4

10

Ignition Delay Time ( s)

0.5% C H - 19.5% O - 80% Ar


6

10

p = 5.5-7.4 atm
5

10
0.65
4
10

0.70

0.75

0.80

0.85

0.90

1% C H - 19% O - 80% Ar
6

10

0.70

0.75

0.80

10

p = 2.27 atm
5

10
0.65
4
10

0.7 0

0.7 5

0.80

0.85

1.69% C H - 12.675% O - 85.635% Ar


6

10

10

p = 2.47 atm
5

10
0.60
4
10

0.6 5

0.7 0

0.75

0.80

1.35% C H - 5.09% O - 93.56% Ar


6

10

10

p = 2.52 atm
0.85

0.90

10
0.55
4
10

0.6 0

10

p = 5.5-7.0 atm

10

10

10

10
0.70
4
10
10

0.6 5

0.70

0.75

0.516% C H - 3.87% O - 95.614% Ar

0.75

0.80

0.85

p = 6.67 atm

0.90

10
0.60

1% C H - 9% O - 90% Ar
6

0.6 5

0.70

0.75

10 K/T
p = 5.9-6.6 atm
5

10
0.60
4
10

0.65

0.70

0.75

0.80

0.85

4% C H - 16% O - 80% Ar
6

Figure 3. Experimental (symbols) [35] and computed


(lines) ignition delay times for benzene-oxygen-argon
mixtures.

Ignition Delay Time ( s)

Ignition Delay Time ( s)

0.80

2% C H - 18% O - 80% Ar
10

p = 5.4-7.4 atm
5

10
0.65
4
10

0.70

0.75

0.80

0.85

0.90

2% C H - 8% O - 90% Ar
6

10

10

p = 5.6-7.1 atm
5

10
0.60
3
10

0.65

0.70

0.75

0.80

0.85

0.90

2.75% C H - 7.25% O - 90% Ar


6

10

10

10

p = 5.9-7.0 atm
1

10

0.75

10
0.65
4
10

10

0.7 0

1.35% C H - 20.31% O - 78.34% Ar

Figure 1. Experimental (symbols) [19] and computed


(lines) laminar flame speed of benzene- and tolueneair mixtures at 1 atm. The experimental data were
obtained from nonlinear extrapolation to zero stretch
10

0.6 5

Equivalence Ratio

10

25

0.8

10

p = 5.4-6.7 atm
5

10
0.60

0.65

0.70

0.75

0.80

10
10
10

1.176% C H - 8.824% O - 90% Ar


6

p = 4 atm
5

10
0.45
4
10
10
10
10

0.50

0.55

0.60

0.65

0.70

0.75

0.80

0.85

2.353% C H - 17.647% O - 80% Ar


6

p = 4 atm
5

10
0.50

0.55

0.60

0.65

0.70

0.75

0.80

0.85

0.90

10 K/T

10 K/T
Figure 2. Experimental (symbols) [34] and computed
(lines) ignition delay times for benzene-oxygen-argon
mixtures. Ignition was defined by time lapse from the
arrival of a reflected shock wave to the time when fuel
concentration decreased by 3-5% (measured by
benzene light emission at 3.29 m).

10

Figure 4. Experimental (symbols) [36] and computed


(lines) ignition delay times for benzene-oxygen-argon
mixtures.

10

0.497% C H - 13.51% O - 85.993% Ar


7

Mole Fraction 10

10

p = 2.27 atm
5

10

0.65

0.70

0.75

0.497% C H - 4.48% O - 95.023% Ar


7

10

p = 2.28 atm
1

0.60

0.65

0.70

0.75

10

10

p = 2.82 atm
5

10
0.55

0.60

0.65

0.70

0.75

0.497% C H - 4.48% O - 95.023% Ar


7

10

CH

C H
2

C H
2

p = 6.81 atm
0.65

Mole Fraction 10

0.60

0.70

10 K/T
Figure 5. Experimental (symbols) [35] and computed
(lines) ignition delay times for toluene-oxygen-argon
mixtures.
4

C H OH
6

C H C HO
6

O /5

Mole Fraction 10

M ole Fraction 10

C H

10
0.55

CO
2

C H
6

C H 10
5

0
5

C H +C H

0
0

M ole Fraction 10

CO

1.495% C H - 13.45% O - 85.055% Ar

10

10

10
0.55
10

C H

Mole Fraction 10

Ignition Delay Time ( s)

10
0.60

20

40

60

80

100

120

Reaction Time (ms)

C H

C H

Figure 7. Species profiles experimentally determined


(symbols) [2] and computed (lines) for toluene
oxidation (0.16% C7H8-1.09% O2-98.75% N2, =
1.33) in a flow reactor at 1 atm and the temperature of
1190 K.

C H OH
6

0
0

20

40

60

80

100

120

Reaction Time (m s)

Figure 6. Species profiles experimentally determined


(symbols) [2] and computed (lines) for benzene
oxidation (0.14% C6H6-1.61% O2-98.24% N2, =
0.65) in a flow reactor at 1 atm and the temperature of
1108 K.

0.6

10

0.4

CO

Mole Fraction 10

Mole Fraction

0.5

H 3
2

0.3

C H

0.2

H O

0.1

CO

OH

0.0

0
0

12

10

Mole Fraction 10

Mole Fraction 10

H
6

CH O
2

HO

C H
2

C H 5

0
0

10

C H
3

6
3

Mole Fraction 10

Mole Fraction 10

C H
3

5
4

C H
4

3
2

C H
4

2
1
0

0
0

10

Mole Fraction 10

Mole Fraction 10

C H
5

C H
5

C H
6

C H
6

0
0

Mole Fraction 10

Mole Fraction 10

C H OH

C H
7

C H C H
6

0
0

Distance from Burner Surface (cm)

Distance from Burner Surface (cm)

Figure 8. Species profiles experimentally determined (symbols) [7] and computed (lines) for a laminar premixed
burner-stabilized benzene flame (13.5% C6H6-56.5% O2-30% Ar, = 1.8, cold gas velocity 50 cm/s, and p = 2.67
kPa).

[C6H5OH]model/[C6H5OH]exp

25

degree of backmixing

20
15
10
5
0
flow
reactor

burner-stabilized
flame

stirred
reactor

Figure 9. Influence of back mixing on the prediction of phenyl concentration. While back mixing in the flow
reactor is minimal, it is maximum in a stirred reactor. The back mixing in the laminar premixed burner-stabilized
flame is caused by diffusion.

Potrebbero piacerti anche