University of Southern California Dissertations and Theses

Studies on the flame dynamics and kinetics of alcohols and liquid hydrocarbon fuels

(USC Thesis Other)

Studies on the flame dynamics and kinetics of alcohols and liquid hydrocarbon fuels

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
STUDIES ON THE FLAME DYNAMICS AND KINETICS OF
ALCOHOLS AND LIQUID HYDROCARBON FUELS

by

Adam Takashi Holley

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements of the Degree
DOCTOR OF PHILOSOPHY
(MECHANICAL ENGINEERING)

December 2008

Copyright 2008 Adam Takashi Holley

ii
Dedication

This dissertation is dedicated to my family who has shown me love and patience
throughout my life.

iii
Acknowledgements

First and foremost I must thank my advisor Professor Fokion N. Egolfopoulos for the
tutelage and guidance he has given me over my many years at University of Southern
California. He is responsible for providing me with the foundation of my combustion
knowledge and the direction needed to help me grow. He has challenged me to become a
better speaker, writer, and scientist and been an example of hard work, dedication, and
genuine enthusiasm towards the advancement of combustion knowledge. I have no doubt
that without his assistance I would never have come as far as I have.
This work was only accomplished due to the assistance of my many colleagues in the
Combustion and Fuels Laboratory at the University of Southern California. I especially
wish to thank Dr. Yufei Dong and Dr. Mustafa Gurhan Andac. Yufei passed his
extensive experimental knowledge on to me, and was my partner in a large portion of the
experimental data that I have taken. Gurhan along with Dr. Egolfopoulos provided the
necessary helping hand while I was first learning about the numerical aspects of
combustion research.
I also wish to thank the Faculty of the Aerospace and Mechanical Engineering
Department of USC. My entire collegiate career has been at USC, and I have taken
classes from almost every faculty member in the department, and they have all
contributed to my understanding of the physical world. I would like to particularly thank
Dr. Hai Wang and his guidance over the past few years. He has enhanced greatly my

iv
knowledge of chemical kinetics mechanisms, their development, and their optimization
which has in turned opened my mind and caused me to grow as an experimentalist.
Finally, I wish to thank my family who has supported at every turn for all these years.
They have shown unyielding confidence in me throughout my years of study, and have
given me strength to go the distance. I feel fortunate to have a family that has the
patience and understanding to help me through this undertaking.

v
Table of Contents

Dedication ii

Acknowledgments iii

List of Tables viii

List of Figures ix

Abstract xiv

Chapter 1: Introduction and Significance 1
1.1 Liquid Fuels 1
1.2 Chapter 1 References 5

Chapter 2: Laminar Flames 6
2.1 Combustion Modes 7
2.2 Turbulent and Laminar Flames 8
2.3 Laminar Flame Speed 8
2.4 Flame Stretch 9
2.5 Flame Ignition 11
2.6 Flame Extinction 13
2.7 Chapter 2 References 16

Chapter 3: Practical Fuels 17
3.1 Distillate Fuels 17
3.2 Gasoline 18
3.3 Jet Fuels 19
3.4 Synthetic Fuels 21
3.5 Fuel Surrogates 22
3.6 Chapter 3 References 24

Chapter 4: Phase Change 25
4.1 Preferential Vaporization 26
4.2 Pyrolysis and Partial Oxidation 27
4.3 Condensation 27
4.4 Chapter 4 References 30

Chapter 5: Objectives 31
5.1 Experimental 31
5.2 Numerical 32

vi
Chapter 6: Experimental Approach 33
6.1 Experimental Apparatus 34
6.2 Verification of the Vaporization System 36
6.3 Fuels 42
6.4 Ignition Methodology 43
6.4.1 Premixed Configuration 44
6.4.2 Non-Premixed Configuration 45
6.5 Extinction Methodology 45
6.5.1 Premixed Configuration 46
6.5.2 Non-Premixed Configuration 46
6.6 Digital Particle Image Velocimetry 47
6.7 Chapter 6 References 48

Chapter 7: Numerical Approach 49
7.1 Numerical Code 49
7.2 Diffusion Sensitivity Modification 50
7.3 Extinction Approach 51
7.4 Chemical Kinetic Mechanisms 51
7.5 Chapter 7 References 53

Chapter 8: Extinction of Premixed Flames of Practical Single Component
Liquid Fuels 54
8.1 Introduction 54
8.2 Experimental Approach 56
8.3 Numerical Approach 56
8.4 Results and Discussion 58
8.4.1 Experimental Results on Extinction Strain Rates 58
8.4.2 Numerical Predictions of Extinction Strain Rates 60
8.4.3 Comparison of Phenomena of Laminar Flame
Propagation and Extinction 68
8.4.4 Diffusion Effects 80
8.5 Conclusions 83
8.6 Chapter 8 References 85

Chapter 9: Extinction of Benzene/Air and Toluene/Air Premixed Flames 87
9.1 Introduction 87
9.2 Experimental Approach 89
9.3 Numerical Approach 90
9.4 Results and Discussion 90
9.5 Conclusions 100
9.6 Chapter 9 References 102

vii
Chapter 10: An Experimental Study on the Ignition and Extinction of
Premixed Flames of Gasoline and its Surrogates 105
10.1 Introduction 105
10.2 Experimental Approach 107
10.2.1 Fuels Tested 107
10.2.2 Boundary Conditions 108
10.3 Results and Discussion 109
10.4 Conclusions 119
10.5 Chapter 10 References 121

Chapter 11: Ignition and Extinction of Non-Premixed Flames of
Single Component Liquid Hydrocarbons, Jet Fuels,
and their Surrogates 122
11.1 Introduction 122
11.2 Experimental Approach 124
11.2.1 Fuels Tested 124
11.2.2 Boundary Conditions 126
11.3 Results and Discussion 127
11.4 Conclusions 139
11.5 Chapter 11 References 141

Chapter 12: Sensitivity of Propagation and Extinction of Large
Hydrocarbon Flames to Fuel Diffusion 142
12.1 Introduction 142
12.2 Lennard-Jones Spherical Potential Parameters 145
12.3 Numerical Approach 146
12.4 Results and Discussion 147
12.4.1 Sensitivity of Flame Phenomena to Chemical
Kinetics and Molecular Transport 147
12.4.2 Species Flux Analysis 153
12.4.3 Effect of Transport Model 161
12.5 Conclusions 162
12.6 Chapter 12 References 164

Chapter 13: Concluding Remarks and Recommendations 166
13.1 Concluding Remarks 166
13.2 Future Work 170

Bibliography 174

viii
List of Tables

Table 8.1 Kinetics mechanisms used to simulate present and literature
experimental data. 57

Table 8.2 CH
2
CHO reactions involved in DL98 (revised) mechanism. 58

Table 8.3 List of reactions discussed in the text. 62

Table 8.4 List of reactions included in LHD03 that differ from HD98. 63

Table 8.5 Logarithmic sensitivity coefficients of S
u
o
and K
ext
to the rate of
the main branching reaction R1 and to the fuel diffusivity for
methanol/air, ethanol/air, n-heptane/air, and iso-octane/air
mixtures. 82

Table 10.1 Variation of T
ign
with fuel/air mass ratio for all fuels. The
uncertainty in fuel/air mass ratio is 2.5% and in T
ign
is ± 20
o
C. 109

Table 10.2 Variation of K
glb,ext
with fuel/air mass ratio for all fuels. The
uncertainty in fuel/air mass ratio is 2.5% and in K
glb,ext
is 3.5%. 115

Table 11.1 Test fuels with detailed composition data (ASTM D2425). 124

Table 11.2 Surrogate fuel composition. 125

Table 11.3 K
ext
vs Fuel/N
2
mass ratio for all fuels. The uncertainty in
fuel/N
2
mass ratio is 2.5% and in K
ext
is 3.5%. 127

Table 11.4 T
ign
vs Fuel/N
2
mass ratio for all fuels. The uncertainty in
fuel/N
2
mass ratio is 2.5% and in T
ign
is ± 20
o
C. 133

Table 12.1 Computational Test Cases. 146

ix
List of Figures

Figure 2.1 Axial velocity profile with the maximum velocity gradient
defined as K. 11

Figure 2.2 Ignition S-curve. 12

Figure 2.3 Extinction S-curve. 14

Figure 6.1 Schematic of the counterflow configuration. 34

Figure 6.2 Schematic of the experimental apparatus. 34

Figure 6.3 Schematic of the vaporization chamber. 35

Figure 6.4 Variation of the fuel/N
2
mass ratio at ignition as a function of
post chamber temperature for a constant ignition source. 37

Figure 6.5 Variation of the fuel/N
2
mass ratio at extinction as a function of
post chamber temperature at constant strain rate. 38

Figure 6.6 GC trace at an ambient post chamber temperature. 39

Figure 6.7 GC trace at a post chamber temperature of 250 °C. 40

Figure 6.8 GC trace at a post chamber temperature of 350 °C. 41

Figure 8.1 Variation of the experimentally determined K
ext
with φ, for
mixtures of air with all liquid fuels considered in this study. 59

Figure 8.2 Variation of experimental and computed K
ext
with φ for
methanol/air flames using the FDC00, HD98, and LHD03
mechanisms. 61

Figure 8.3 Variation of experimental and computed K
ext
with φ for
ethanol/air flames using the FDC00 and MRN99 mechanisms. 64

Figure 8.4 Variation of experimental and computed K
ext
with φ for n-
heptane/air flames using the DL98 and the DL98 (revised)
mechanisms. 65

x
Figure 8.5 Variation of experimental and computed K
ext
with φ for iso-
octane/air flames using the PPS96, DL98, and the DL98
(revised) mechanisms. 66

Figure 8.6 Variation of experimental and computed S
u
o
with φ for
methanol/air flames using the FDC00, HD98, and LHD03
mechanisms. The experimental data were taken from Ref. 1. 68

Figure 8.7 Integrated species consumption paths for a φ = 0.769 NEF and
FPF methanol/air flame, computed using the HD98 and LHD03
mechanisms. 70

Figure 8.8 Logarithmic sensitivity coefficients of S
u
o
and K
ext
to reaction
rate constants for a φ = 0.769 methanol/air flame computed
using LHD03. 71

Figure 8.9 Variation of experimental and computed S
u
o
with φ for
ethanol/air flames using the FDC00 and the MRN99
mechanisms. The experimental data were taken from Ref. 2. 73

Figure 8.10 Variation of experimental and computed S
u
o
with φ for n-
heptane/air flames using the DL98 and DL98 (revised)
mechanisms. The experimental data were taken from Ref. 3. 75

Figure 8.11 Variation of experimental and computed S
u
o
with φ for iso-
octane/air flames using the PPS96, DL98 and DL98 (revised)
mechanisms. The experimental data were taken from Ref. 3. 76

Figure 8.12 Integrated species consumption paths for a φ = 0.873 NEF and
FPF iso-octane/air flame, computed using the DL98 and DL98
(revised) mechanisms. 77

Figure 8.13 Logarithmic sensitivity coefficients of S
u
o
and K
ext
to reaction
rate constants for a φ = 0.873 iso-octane/air flame computed
using the DL98 mechanism. 79

Figure 9.1 Experimentally and numerically determined K
ext
of premixed
benzene/air flames as functions of φ. 91

Figure 9.2 Literature experimental data [27] and numerical predictions of
the S
u
o
of benzene/air flames as a function of φ. 92

xi
Figure 9.3 Fuel consumption reaction paths for a φ = 1.2 C
6
H
6
/air flame,
for both a FPF and a NEF. 93

Figure 9.4 Logarithmic reaction rate sensitivities for a φ = 1.2 benzene/air
flame, for both a FPF and NEF. 94

Figure 9.5 Logarithmic reaction rate sensitivities of the mass burning rate
of FPF for benzene/air mixtures at various φ. 95

Figure 9.6 Experimentally and numerically determined K
ext
for premixed
toluene/air flames as functions of φ. 96

Figure 9.7 Literature experimental data [27] and numerical predictions of
the S
u
o
of toluene/air flames as a function of φ. 98

Figure 9.8 Experimentally determined K
ext
for premixed benzene/air,
toluene/air, n-heptane/air, and iso-octane/air flames as functions
of φ. 99

Figure 10.1 Variation of T
ign
with the fuel/air mass fraction for the single
component hydrocarbons. 110

Figure 10.2 Variation of T
ign
with the fuel/air mass fraction for individual
samples of gasoline. 112

Figure 10.3 Variation of T
ign
with the fuel/air mass fraction for averaged
samples of gasoline and the gasoline surrogates. 113

Figure 10.4 Variation of K
ext
with the fuel/air mass fraction for the SCHs. 116

Figure 10.5 Variation of K
ext
with the fuel/air mass fraction for the
individual samples of gasoline. 117

Figure 10.6 Variation of K
ext
with the fuel/air mass fraction for the averaged
samples of gasoline and the gasoline surrogates. 118

Figure 11.1 Variation of K
ext
with the fuel/N
2
mass fraction for the SCHs. 129

Figure 11.2 Variation of K
ext
with the fuel/N
2
mass fraction for the 3 batches
of Jet-A. 130

xii
Figure 11.3 Variation of K
ext
with the fuel/N
2
mass fraction for the 2
synthetic fuels. 131

Figure 11.4 Variation of K
ext
with the fuel/N
2
mass fraction for the 2
surrogates, JP-8, and the averaged Jet-A-4658. 132

Figure 11.5 Variation of T
ign
with the fuel/N
2
mass fraction for the SCHs. 135

Figure 11.6 Variation of T
ign
with the fuel/N
2
mass fraction for the 3 batches
of Jet-A. 136

Figure 11.7 Variation of T
ign
with the fuel/N
2
mass fraction for the two
synthetic fuels. 137

Figure 11.8 Variation of T
ign
with the fuel/N
2
mass fraction for the two
surrogate fuels, JP-8, and the averaged batch of Jet-A-4658. 138

Figure 12.1 Counterflow non-premixed extinction strain rate K
ext
as a
function of number of carbon atoms in the nitrogen-diluted n-
alkane jet against an oxygen jet. Data (symbols) are taken from
[9]; lines: fits to data. 143

Figure 12.2 Logarithmic sensitivity coefficients with respect to reaction rate
and diffusion coefficients using mixture-averaged transport
formulation. Light bars: positive values; dark bars: negative
values. 148

Figure 12.3 Non-premixed flame extinction strain rate of normal alkanes as
a function of binary fuel-N
2
diffusion at 1200 K and 1 atm.
Symbols: experimental data for nitrogen diluted fuel jet against
an oxygen jet [9]; lines: fits to data. 150

Figure 12.4 Non-premixed flame extinction curves computed for n-heptane
and n-dodecane with fuel/nitrogen mass ratio of 0.08 against an
oxygen jet. The three n-C
12
H
26
cases are a) base case; b) the
binary diffusion coefficient of n-C
12
H
26
-N
2
replaced by that of
n-C
7
H
16
-N
2
; c) all binary diffusion coefficients of n-C
12
H
26

replaced by n-C
7
H
16
. 152

Figure 12.5 The convective and diffusive fuel flux of n-C
12
H
26
for a FPF. 154

Figure 12.6 The convective and diffusive fuel flux of n-C
12
H
26
for a NEF. 156

xiii
Figure 12.7 Convective and diffusive mass fluxes and species mass fraction
of n-C
12
H
26
as a functions of distance for a near extinction non-
premixed 1/30 fuel/N
2
molar ratio flame with air as oxidizer. 158

Figure 12.8 Convective and diffusive mass fluxes and species mass fraction
of H as functions of distance for a near extinction non-premixed
1/120 fuel/N
2
molar ratio flame with O
2
as oxidizer. 160

Figure 12.9 Logarithmic sensitivity coefficients with respect to reaction rate
and diffusion coefficients using the multi-component
formulation of transport. Light bars: positive values; dark bars:
negative values. 162

xiv
Abstract

An experimental and numerical study was conducted on the propagation and ignition
and extinction limits of alcohols and liquid hydrocarbon fuels. Experimentally, the
extinction and/or ignition limits were determined for a wide range of fuels including: C
1
-
C
2
alcohols, C
5
-C
14
n-alkanes, C
8
iso-alkanes, samples of gasoline, gasoline surrogates,
samples of jet fuels, jet fuel surrogates, and synthetic fuels. Experiments were conducted
for both premixed and non-premixed flames, and at ambient as well as elevated
temperature. Comparison between the ignition and extinction limits of flames involving
the various fuels provided insight into their burning characteristics, their relative
performance, and the effect of chemical classification. The insight that was gained will
aid in future surrogate development work and will provide direction for the desired
composition of practical fuels. Numerically, selected experiments were simulated
utilizing quasi-one-dimensional codes, which integrate the conservation equations of
mass, momentum, energy, and species with detailed descriptions of molecular transport
and chemical kinetics and by invoking a variety of chemical kinetic mechanisms.
Through comparison between the numerical simulations and experimental results the
adequacy and applicability of existing chemical kinetic mechanisms was assessed.
Insight was gained into the fundamental kinetic and transport mechanisms that control
flame phenomena of interest. Additionally, the inadequacies in current standard chemical
kinetic mechanisms optimization practices were identified, namely the exclusion of flame
phenomena other than propagation as a constraint, and the exclusion of transport
properties as parameters.

1

Chapter 1

Introduction and Significance

1.1 Liquid Fuels
Liquid fuels are among the primary sources of energy worldwide. They have been
one of the most important natural resources for over 100 years, and represent an economy
that is measured in the 100’s of billions of dollars each year in the United States alone.
Modern day society is dependent on their use, for both the distribution of goods and the
transportation of people. Typical practical liquid fuels contain large amount of chemical
energy stored in bonds between carbon, hydrogen, and oxygen with a wide range of
molecular weights and chemical classification, including alkanes, alkenes, aromatics, and
alcohols to name a few. A main draw back of the use of liquid fuels is the fact that their
combustion releases pollutants into the air such as particulate matter (PM), unburned
hydrocarbons (UHC), aldehydes, nitrogen oxides (NO
x
), carbon monoxide (CO), and

2
greenhouse gases, e.g. carbon dioxide (CO
2
), to name a few. Additionally, the oil prices
have increased to record levels, affecting all sectors of life. Based on those
considerations there is an urgent need on one hand to increase the efficiency of utilization
of conventional (liquid) fuels and on the other to explore the use of alternative non-
petroleum derived (liquid) fuels.
Liquid fuels, such as hydrocarbons, are widely distributed, have large energy density,
and they can be stored and transported rather conveniently. Thus, they are ideal for
transportation, cars and planes, due to the necessity to carry the fuel for the duration of
the trip. Due to their ideal nature for transportation, they will continue to be used despite
the growing economic and environmental concerns, as there are no viable alternatives
that can replace them at least in the foreseeable future. Their use, however, will most
likely decrease in stationary power generation, but it is doubtful that its use in
transportation will do anything but increase. In addition to crude oil, liquid fuels can be
produced from almost any carbon source, such as coal, oil sands, oil shale, biomass, and
waste. The energy content available in these other sources will last far into the next
century. Despite their vast importance, the fundamental burning processes of liquid fuels
are not well understood. This is true not only for new alternative fuels that are little
known, but even for conventional petroleum-derived fuels.
Significant research has been conducted assessing the performance of liquid fuels in
engines, e.g. engine “knock”, but significantly less fundamental research has been done.
Subsequently, the underlying physical and chemical processes controlling the oxidation
of such fuels are not well understood, especially when compared to lower molecular

3
weight gaseous fuels such as H
2
, CO, and C
1
-C
4
, collectively referred to as C
0
-C
4
fuels;
C
#
indicates the “carbon number” of a particular fuel. Generally, hydrocarbons with
C
#
> 4 are liquid at standard conditions. At the same time it has been recognized that the
C
0
-C
4
chemical kinetics constitute the foundation of the kinetics of the larger molecular
weight, practical, liquid fuels. Thus, it is necessary to understand the chemical kinetics of
small gaseous fuels first.
Combustion scientists have been doing research for over a century and have been
developing chemical kinetic models that simulate combustion chemistry for half a
century or so, focusing largely on gaseous fuels. Experimentally, it is much easier to
work with gaseous fuels as compared to liquid ones. It is a non-trivial task to vaporize
liquid fuels so that there is no chemical composition change, and subsequently maintain
them in the gaseous state. Additionally, chemical kinetic models for gaseous fuels can be
developed with greater degree of confidence compared to larger liquid fuel molecules due
to notably larger number of rate constants. Computationally, the power that is required to
simulate reacting flows even in simple one-dimensional steady laminar configurations
increases drastically as the molecular size grows. While ten years ago this would be an
impossible task, the current state of combustion chemistry knowledge, experimental
expertise, and computation power, renders feasible the development of fundamental
knowledge for large liquid fuels.
The development of a reliable chemical kinetic mechanism for any fuel, gaseous or
liquid, requires that experiments must be performed in geometries that can be modeled
directly and with very well defined boundary conditions. Homogeneous systems such as

4
flow reactors, shock tubes, and rapid compression machines, to name a few, provide
valuable experimental data that are typically used as the first step towards the
compilation of chemical kinetics mechanisms. However, these homogeneous systems
usually test a fairly narrow range of reactant concentrations and temperatures. The next
step is to conduct flame experiments in which the results are dependent on chemical
kinetics spanning a wide range of concentrations and temperatures typically from
ambient to nearly 2000 K. While the modeling of homogeneous systems is time-
dependent only, the modeling of flames involves one or more spatial dependencies. The
presence of molecular transport in flames makes the problem elliptic and the solution
algorithm is complicated notably. The prohibitive computational costs of simulating
multi-dimensional reacting flows and the complications associated with far-field
boundary conditions, underline the importance of modeling experiments that are either
one-dimensional or quasi-one-dimensional in space.
The opposed-jet stagnation flow has been shown to be a meritorious configuration for
fundamental flame studies, and can be modeled utilizing quasi-one-dimensional codes.
More specifically, the use of the stagnation flow configuration has been essential for the
determination of fundamental flame properties such as propagation speeds, ignition
limits, extinction limits, and thermal and concentration structures [e.g., 1-4]. The
accurate determination of such properties is very important, as they are essential towards
the development of reliable kinetic mechanisms. Such studies have contributed towards
the development of mechanisms that predict with increased degree of confidence the
oxidation characteristics of gaseous C
0
-C
4
fuels [e.g., 5-7].

5
1.2 Chapter 1 References
[1] C.K. Law, D.L. Zhu, G. Yu, Proc. Combust. Inst. 21 (1986) 1419-1426.
[2] C.G. Fotache, T.G Kreutz, D.L Zhu, C.K. Law, Combust. Sci. Technol. 109
(1995) 373–393.
[3] C.K. Law Proc. Combust. Inst. 22 (1988) 1381-1402.
[4] C.M. Vagelopoulos, F.N. Egolfopoulos, Proc. Combust. Inst. 27 (1998) 513-519.
[5] G.P. Smith, D.M. Golden, M. Frenklach, N.W. Moriarty, B. Eiteneer, M.
Goldenberg, C.T. Bowman, R.K. Hanson, S. Song, W.C. Gardiner, Jr., V.
Lissianski, Z.Qin, GRI-Mech 3.0, http://www.me.berkeley.edu/gri_mech/ (2000).
[6] C.J. Sun, C.J. Sung, H. Wang, C.K. Law Combust. Flame 107 (1996) 321-335.
[7] S.G. Davis, C.K. Law, H. Wang, Proc. Combust. Inst. 27 (1998) 305-312.

6

Chapter 2

Laminar Flames

A flame is characterized by a rapid conversion of reactants to products through a
highly exothermic process that takes place within a thin non-equilibrium zone separating
the unburned and burned equilibrium zones. Inherently, a flame exhibits high spatial
gradients in both species concentrations and temperature. Accounting for such spatial
variations of properties in experiments and modeling can be rather challenging. Testing
chemical kinetic models against flame data can provide validation over a wide range of
thermodynamic conditions. Additionally, the variations in the species concentrations add
another constraint to the model, that it to accurately simulate the molecular transport of
all species. Flames provide the most rigorous test of kinetic mechanisms, far more than
any homogeneous system. For any model to be used in complex reacting configurations,
it must be able first to predict fundamental flame data in well-controlled laminar flame
experiments that provide the linkage between what is learned in reactors and the realistic
conditions encountered in a combustor.

7

2.1 Combustion Modes
For combustion to take place, typically two reactants are required, a fuel and an
oxidizer. The fuel and the oxidizer must be mixed at the molecular level for combustion
to occur. Flames behave distinctly different if the fuel and oxidizer are mixed before they
burn or not. A premixed flame forms when the fuel and oxidizer are mixed before the
onset of combustion. Premixed flames are characterized by an equivalence ratio, φ,
which is defined as the ratio of fuel to oxidizer over the stoichiometric ratio of fuel to
oxidizer:

[]
[]
tric Stoichiome
Actual
Oxidizer Fuel
Oxidizer Fuel
= φ
2.1

The stoichiometric fuel/air ratio that is φ = 1, is defined as the ratio of fuel to oxidizer that
results in complete consumption of both initial reactants. Values of φ < 1 correspond to
lean flames that are fuel deficient, and φ > 1 correspond to rich flames that are oxidizer
deficient. Premixed flames also have a propagation speed. Non-premixed flames are
established at locations at which the fluxes of fuel and oxidizer are stoichiometrically
balanced. Since the reactants first mix and then burn, non-premixed flames are
commonly referred to as diffusion flames, due to their sensitivity to the transport

8
properties of the reactants. Unlike premixed flames, non-premixed flames do not have a
propagation speed.

2.2 Turbulent and Laminar Flames
Most practical combustion devices employ turbulent combustion, as turbulence
facilitates mixing and subsequently increases the overall mass-burning rate when
compared to laminar combustion. The high burning rate attainable utilizing turbulent
combustion allows for the design of smaller powerful combustors, which result in greater
overall efficiency. Turbulent flames are prohibitively difficult to study experimentally as
they are intrinsically unsteady, with the properties being time averaged.
Despite the inability to perform fundamental research on turbulent flames, in general,
turbulent combustion properties scale with laminar combustion properties. This is due to
the fact that an increase in molecular mixing does not change the underlying chemistry of
the system. Chemical kinetics, thermodynamic properties, and transport properties
developed and validated in laminar flames are still valid in turbulent combustion.
Additionally for certain regimes of turbulent combustion, the laminar flamelet model
[1,2] can be applied when the chemical time scales are sufficiently shorter compared to
the diffusion and convection time scales.

2.3 Laminar Flame Speed
The laminar flame speed, S
u
o
, is defined as the propagation speed of a steady, one-
dimensional, planar, unstrained, adiabatic, laminar premixed flame. It is a unique

9
property of a particular mixture since it is a measure of the mass consumption rate of an
ideal flame that is free from external influences. S
u
o
contains information regarding a
mixture’s reactivity, exothermicity, and diffusivity. Accurate knowledge of S
u
o
is
important towards the validation of chemical kinetic models and in modeling turbulent
combustion. It is related to other premixed flame phenomena such as flammability limits
and practical applications such as, for example, flame stabilization in a turbine.
No laboratory flame can ever fulfill the definition of the S
u
o
, given that an external
mechanism is required for flame stabilization in a laboratory frame of reference. The two
most common stabilization mechanisms are heat loss [3] and strain rate [4,5] to be
defined in the next section. To determine S
u
o
experimentally, a series of measurements is
required in which the magnitude of stabilization mechanism is varied, resulting in flame
speeds as function of this particular external (to the mixture) influence. Subsequently the
data can be extrapolated to the zero value of the external influence, resulting in S
u
o
.
Experimental measurements of the heat loss from a flame results in a total heat loss value
for the flame over the entire experimental domain, though the actual heat loss in any
given section of the experimental domain may not be known. Conversely, the strain rate
can be measured locally, allowing for a more precise measure of the external influence,
and this is the technique employed by the opposed jet stagnation flow.

2.4 Flame Stretch
Flame stretch, K, is defined as the rate of change of the area of a differential surface
element scaled by the value of the surface area [6]:

10
dt
dA
A
K
1
=
2.2
K can also be written in terms of flow velocity [7]:
( )( ) n n V V V K
t t t t
⋅ ∇ ⋅ + ⋅ ∇ = ⋅ ∇ =
2.3
where
t
∇ is the gradient operator along the surface of the flame, and V is the velocity
vector. Decomposing the velocity vector into its tangential and normal components
results in the two sources of flame stretch. The first term is called the strain rate, and is
the rate of change of the tangential velocity along the flame surface. This physically
represents the growth or decay of the size of the differential surface, as the velocities of
the two boundaries are different. The second term is the stretch caused by the movement
of a curved flame.
For the opposed jet stagnation configuration the velocity vector is:
0 ,
2
, r x V
α
α − =
2.4
where α is the strain rate of the flow imposed by the velocity gradient, x is the axial
coordinate, and r is the radial coordinate. Therefore for this configuration K = α/2.
Experimentally it is convenient to measure the axial velocity gradient, commonly known
as the imposed “strain rate”, as opposed to the radial velocity gradient in particular when
laser-based techniques are used and which measure velocity vectors only in one direction.
This approach has been the standard in counterflow experiments. Henceforth, K will be
referring to strain rate, defined as the maximum absolute value of the axial velocity
gradient in the hydrodynamic zone:

11
max
x
V
K
∂
∂
=
2.5
This is also shown graphically in Fig 2.1.

Figure 2.1 Axial velocity profile with the maximum velocity gradient defined as K.

2.5 Flame Ignition
The point at which the low intensity chemical activity transitions into high intensity
chemical activity sufficient to sustain a flame is defined as flame ignition. Ignition can
occur due to either thermal or radical runaway. Thermal runaway occurs when the heat
release due to chemical activity exceeds the diffusive/convective thermal losses out of the
ignition kernel, defined as the region within which the chemical activity is mostly
concentrated. On the other hand, radical runaway occurs when the radical production due
to chemical activity exceeds the diffusive/convective radical losses out of the ignition
kernel. Solving the conservation equations for the maximum concentration of, say, the H
Axial
Velocity
x
∂x
∂V
max
x
V
K
∂
∂
=

12
radicals as function of the boundary temperature that serves as the ignition source, results
in a S-shaped curve as shown in Fig. 2.2 which was first studied by Fendell [8] and Liñán
[9].

Figure 2.2 Ignition S-curve.

Typically, the H radical is chosen for such representations due its importance in the
main branching reaction, H + O
2
⇒ OH + O, but the same result would occur if any
other radical was chosen. Only the upper and lower branches of the S-shaped curve are
stable solutions. The transition of the solution from the bottom branch to top branch
corresponds to flame ignition. The temperature at which the transition occurs in defined
as the ignition temperature, T
ign
. T
ign
depends on the reactant type, the boundary
conditions, the initial thermodynamic state, as well as on the strain rate imposed as a
Temperature K
Ignition
Extinction
Stable Solution
Stable Solution
Unstable Solution
T
ign
Maximum
H Radical
Concentration

13
result of the convective forces. The thermodynamic state including reactant
concentrations, temperature, and pressure, directly affects the chemical kinetics and
subsequently the value of T
ign
. The strain rate is a measure of the rate of the convective
transport of species and thermal energy to and from the ignition kernel.

2.6 Flame Extinction
Flame extinction is the transition from a state of high intensity chemical activity to a
state of low intensity chemical activity. It can be caused by various mechanisms that
exert influence on the flame including heat loss, radical quenching, and flame stretch.
Heat loss reduces the temperature of the flame, which reduces the overall reaction rate.
Radical quenching leads to a reduction in the concentration of active radicals in the
flame, which also reduces the overall reaction rate. Flame stretch, K, can cause both heat
loss and radical quenching in flames. The value of K at the point of extinction is defined
as the extinction strain rate, K
ext
. Similar to ignition, the extinction phenomenon can also
be represented by a multi-branch curve [8,9], by solving for K as a function of maximum
H radical concentration resulting in an inverted S-curve as shown in Fig. 2.3.

14

Figure 2.3 Extinction S-curve.

The top branch of the curve corresponds to a vigorously burning flame, the lower
branch to a cold solution, and the middle branch to an unstable solution. Flame stretch
can always induce extinction for restrained flames. Flame stretch enhances both thermal
diffusion and mass diffusion in unrestrained flames, as well as reduces the allowable time
for reaction to occur in restrained flames. The effect of molecular transport on the flame
in the presence of stretch depends on the Lewis number, Le, defined as the ratio of the
thermal diffusivity of the mixture to the mass diffusivity of the deficient species:
D
Le
α
= 2.6
where α is the thermal diffusivity of the system, and D is the mass diffusivity of the
deficient species. When the Le = 1, the flame is unaffected by flame stretch until it is
restrained. For Le ≠ 1, the presence of the flame stretch will modify the energy and mass
Strain Rate s
-1
Ignition
Extinction
Stable Solution
Stable Solution
Unstable Solution
K
ext

Maximum
H Radical
Concentration

15
fluxes from and to the flame respectively, resulting thus in either hotter or cooler flames.
For example, in stagnation-type flames and for Le > 1 the thermal diffusivity is greater
than the mass diffusivity of the deficient species, resulting in more heat loss than mass
gain that cools the flame. All liquid fuels have Le ≠ 1, therefore the flame temperature
for all flames will depend on K.

16
2.7 Chapter 2 References

[1] N. Peters, Prog. Energy Comb. Sci. 10 (1984) 319-339.
[2] N. Peters, Proc. Combust. Inst. 21 (1986) 1231-1250.
[3] J.P. Botha, D.B. Spalding, Proc. Roy. Soc. Lond. A225 (1954) 71-96.
[4] C.K. Wu, C.K. Law, Proc. Combust. Inst. 20 (1984) 1941-1949.
[5] C.K. Law, Proc. Combust. Inst. 22 (1988) 1381-1402.
[6] F.A. Williams, AGARD Conference Proceedings 164 II 1-1 (1975).
[7] M. Matalon, Combust. Sci. Technol. 31 (1983) 169-181.
[8] F.E. Fendell, J. Fluid Mech. 21 (1965) 281-303.
[9] A. Liñán, Acta Astronautica 1 (1974) 1007-1039.

17

Chapter 3

Practical Fuels

3.1 Distillate Fuels
The dominant source of hydrocarbon fuels is crude oil. Crude oil’s composition
varies greatly depending on the source, but is always a mixture of a large number of
hydrocarbons spanning a wide range of chemical classes and molecule sizes. Distillate
fuels are liquid fuels that are generated by separating bulk crude oil into a smaller range
of products. This separation is done to select a smaller range of molecule sizes, and a
distinct ratio of chemical classifications to suit a particular application. All practical fuels
used today are distillate fuels, due to both economic and technological reasons [1,2].
With current technology the distillation process has an approximate energy efficiency of
90%, resulting in very high yields and at a relatively low cost. The resulting composition
of any fuel produced in this way has variability in it, due to both the process and the
initial crude oil composition.

18
3.2 Gasoline
Gasoline is a mixture of hydrocarbon fuels with C
5
≤ C
#
≤ C
9
, with the average C
#

being between C
6
and C
8
. It is the fuel mostly used in internal combustion engines,
which operate under premixed conditions. The small molecular weight of the fuel is
required to achieve complete evaporation in the small time interval between engine
strokes. The chemical composition of gasoline can vary significantly, but its
approximately 40% n/iso-paraffins, 20% n/iso-olefins, and 40% aromatics by mass
[1,3,4]. Additionally the C
#
varies across chemical classifications, predominately
paraffins are C
5
and C
6
while aromatics are C
7
and C
8
. Gasoline in the US also includes a
small fraction of oxygenates in its fuel to meet governmental guidelines.
This composition is a compromise between requirements related to performance and
pollutant emissions. Engine “knock” is one of the limiting factors on the allowable
compression ratio of an engine. “Knock” occurs when an undesired localized auto-
ignition takes place, which generates high-pressure waves within the cylinder causing a
knocking noise. This phenomenon reduces the efficiency of the engine and could
potentially damage the engine itself. The maximum allowable efficiency of an engine
scales with the compression ratio (r
v
) [5], which is defined as the ratio of the total
cylinder volume to the compressed gas volume. As the r
v
increases the temperature of
the gas mixture prior to the firing of the spark increases, which in turn can promote auto-
ignition. Octane number is a measure of a fuel’s resistance to auto-ignition. n-Alkanes
tend to have low octane numbers, while aromatics tend to have high octane numbers. To
increase the octane number of a fuel mixture, the percentage of aromatic compounds are

19
increased. Building engines with the highest possible efficiency may not be desirable
with existing technologies, as increasing the cylinder temperature and/or the amount of
aromatics in the fuel can increase the pollutant emissions notably.
The three main types of engine pollutants are NO
x
, UHC, and soot. NO
x
formation is
predominantly a function of flame temperature, with hotter flames producing more NO
x
.
There is a relatively small variation in the adiabatic flame temperature of hydrocarbons,
but it depends also on the initial gas temperature. UHC are caused by local extinction
events in the cylinder and the crevices that do not allow for complete oxidation. To
reduce the amounts of UHC that are produced, a fuel must be resistant to extinction, and
must have high burning rates. n-Alkanes are more resistant to extinction as compared to
iso-alkanes and aromatics, but as previously stated also prone to auto-ignition. Another
way to reduce the UHC that are produced is to add oxygenates, which are also resistant to
extinction, and this is why governments mandate their use. Sooting propensity depends
highly on chemical classification. Soot is made up of many joined aromatic rings. As
such, the precursors of soot include aromatics. Alkane flames have a very low sooting
tendency, and to minimize the soot produced in an engine, the amount of aromatics in the
fuel needs to be minimized. The current chemical composition of gasoline is constrained
by these requirements.

3.3 Jet Fuels
Many different jet fuels are in use today; each one is designed to perform under
certain operating conditions. This is in contrast to gasoline, which attempts to be as

20
universal as possible and only varies its octane number slightly depending on the grade.
The most commonly used jet fuels are Jet-A and JP-8. Jet-A is the standard commercial
airliner jet fuel used in USA and JP-8 is the military equivalent of Jet-A with extra
additives to prevent coking and increase its thermal resistance. Both Jet-A and JP-8 have
relatively loose requirements to minimize cost. There are also relatively specialized jet
fuels such as JP-7, which is a highly processed low-volatility fuel created for high
altitude supersonic speed flight used for the SR-71 Blackbird. Jet fuels are oxidized
primarily under non-premixed conditions, and have less strict performance criteria as
compared to gasoline since there is no “knock” equivalent problem in turbines. Jet fuels
have on the average C
7
≤ C
#
≤ C
16
with an average C
#
being between C
11
and C
12
. This
C
#
-range has more to do with the storage of the fuel than the operating conditions in the
engine. At high altitudes and correspondingly low pressures, the fuel tends to vaporize if
the attendant C
#
is too low. The approximate chemical composition of jet fuels is 60% n-
+i-paraffins, 20% cyclo-paraffins (naphthenes) and 20% aromatics by mass, though it has
been shown that the chemical composition of different batches of jet fuels can vary
significantly from each other (e.g., [2,6-8]).
The pollution concerns that were addressed for gasoline still apply to jet fuels, except
for the soot production. The sooting tendency of non-premixed systems is far greater
than that of premixed systems, but the relative propensity between the chemical
classifications still holds. Furthermore, sooting can affect engine performance and
lifetime due to the radiation heat transfer from the particles to the walls of the engine.

21
3.4 Synthetic Fuels
With the world’s supply of crude oil depleting at a substantial rate, the importance of
synthetic fuel research has grown. Various techniques exist to generate synthetic fuels,
but all follow the same approximate formulation. They all start with a carbon source (i.e.
biomass, coal, tar sands, oil shale) and is subsequently conversion into a liquid
hydrocarbon. The two most predominant techniques are the Fischer-Tropsch (FT)
process and Direct Coal Liquefaction (DCL). The FT process involves partially
oxidizing a carbon source at low temperatures to generate a H
2
/CO mixture. The H
2
/CO
mixture is then used to generate longer hydrocarbons including alkanes and olefins in the
presence of catalysts. A FT derived fuel tends to be made up of primarily n- and iso-
alkanes. This technology has been known for over 80 years, and is currently in mass
production use in parts of the world. DCL is performed by generating a pulverized coal
slurry, which under high pressures and temperatures is dissolved into a liquid fuel. The
final product of the DCL process is a fuel made up of cyclo-parrafins and aromatics. In
the coming decades the use of synthetic fuels will increase due to necessity, and
knowledge of their burning characteristics and the desired fuel chemical composition is
needed.
Among the synthetic fuels, alcohols are also considered as viable solutions to the
current energy crisis. Methanol and ethanol are the most relevant alcohols and are
currently being added into commercial gasoline. Alcohols can be produced by the
process of fermentation in addition to the other processes such as FT, with different
catalysts. Ethanol is primarily produced by fermenting food crops such as corn or

22
sugarcane, while methanol is produced from lower grade biomass such as wood chips.
Synthesizing alcohols to augment the domestically produced fuel supply can be
beneficial financially, environmentally, and geopolitically.

3.5 Fuel Surrogates
Due to the complexity of practical liquid fuels, it is not feasible to simulate their
combustion properties directly. Thus, in order to be able to perform large-scale
simulations of reacting flows, the development of mixtures of single-component
hydrocarbons (SCH) that mimic the combustion characteristics of the real fuels, known
as surrogates, is needed. Chemical kinetic mechanisms can be developed then for the
surrogates, which can be subsequently reduced so that they can be used in simulations of
practical combustors.
Fuel surrogates can be developed in multiple ways and take on multiple forms
depending on the final applications they are derived for. Some surrogates are derived
based on the physical properties of fuels, e.g. density or boiling temperature, and can be
used to model applications where these selected properties are of vital importance.
Surrogates are derived also based on chemical class distribution of a fuel, e.g. aromatic
content, and in general have a greater ability to reproduce the chemical kinetic behavior
of the fuel, though it does not guarantee accurate reproduction of fundamental
combustion properties. While those two surrogate development techniques rely solely on
knowledge of the real fuel, a third method to derive a surrogate involves matching

23
experimentally determined combustion properties. This is done by experimentally
matching such properties of real fuel and the surrogate.

24
3.6 Chapter 3 References

[1] T.J. Wallington, E.W. Kaiser, J.T. Farrell, Chem. Soc. Rev. 35 (2006) 335-347.
[2] L.Q. Maurice, H. Lander, T. Edwards, W.E. Harrison III, Fuel 80 (2001) 747-756.
[3] K. Brudzewski, A. Kesik, K. Ko łodziejczyk, U. Zborowska, J. Ulaczyk, Fuel 85
(2006) 553-558.
[4] J. Burri, R. Crockett, R. Hany, D. Rentsch, Fuel 83 (2004) 187-193.
[5] R.E. Sonntag, C. Borgnakke, G.J. Van Wylen, Fundamentals of Thermodynamics
5
th
ed., John Wiley & Sons, Inc., New York, 1998, p. 388.
[6] D.J. Cookson, J.L. Latten, I.M. Shaw, B.E. Smith, Fuel 64 (1985) 509-519.
[7] D.J. Cookson, B.E. Smith, I.M. Shaw, Fuel 66 (1987) 758-765.
[8] N. Kosal, A. Bhairi, M.A. Ali, Fuel 69 (1990) 1012-1019.

25

Chapter 4

Phase Change

Practical fuels exhibit additional experimental complexities compared to their gaseous
counterparts due the issues of phase change. Prevaporized flame experiments are
intricate enough without the added difficulties of phase change, so the liquid fuel is
vaporized and maintained in the gaseous phase. Practical fuels are mixtures of a large
number of components with a wide range of boiling points and vapor pressures. Thus,
they are susceptible to preferential vaporization. Additionally, their relatively weak
bonds can result in pyrolysis or partial oxidation at temperatures near to the boiling points
of the fuels. They are also very large molecules with low vapor pressures, which can lead
to condensation. Careful consideration is required to convert the liquid fuels accurately
into vapor and mix them with gaseous streams prior to entering the experimental domain
so that the boundary conditions of the experiment are will known. If done improperly,
this would affect the fidelity of the experimental data.

26
4.1 Preferential Vaporization
Due to the nature of practical fuels, preferential vaporization complicates greatly the
experimental procedure. It renders many experimental techniques useless, such as
bubblers, and necessitates the use of a vaporization system that is continuous. Namely a
system must have a constant supply of fresh liquid fuel and the fuel being introduced
must be vaporized immediately. Even with a continuous system, there will still be
variation in the vaporization rate of the components regardless of the prevailing
conditions, i.e. temperature and pressure. The only way to overcome this problem is to
ensure very rapid vaporization, since its rate cannot be modified but the total time it takes
to be completed can. The total vaporization time can be made to be so small, that any
slight variation in the gaseous fuel stream can be counteracted by molecular diffusion and
mixing in the gaseous phase prior to entering the experimental domain. This rapid
vaporization is achieved by ensuring that the vaporization takes place near the boiling
point of the heaviest component present in the fuel. Additionally, the vaporization time
can be decreased also by injecting fine droplets of fuel or preheating the liquid fuel as
shown by the d
2
-law [1].
r
s
2
= r
s,o
2
− K
v
t
(4.1)
where r
s
is the instantaneous droplet radius, r
s,o
is the initial droplet radius, K
v
is the
vaporization rate constant, and t is time.

27
4.2 Pyrolysis and Partial Oxidation
Even if a small fraction of the fuel undergoes some chemical activity prior to entering
the test section, the experimental results can be falsified notably. Any amount of
pyrolysis or partial oxidation would render the experimental boundary conditions
unknown. Both pyrolysis and oxidation reaction rates are highly sensitive to temperature,
but even at relatively low temperatures, compared to typical flame temperatures, the rates
are low but finite. Typical experimental mixing times within the vaporization chamber
and the heating path are of the order of seconds. Over these relatively long time scales,
by kinetic standards, notable composition changes could occur even at low temperatures.
To avoid altering the fuel stream’s composition, the temperatures along the heating path
must be kept as low as possible. The minimum temperature at the location of
vaporization is set by evaporation rate requirements, but the rest of the heating path
downstream must to be kept as low as possible to avoid composition changes.

4.3 Condensation
All substances have vapor pressures which are a function of temperature, defined by
the Clausius-Clapeyron relation [1]:
V T
L
T
P
Δ
=
∂
∂
4.1
where P is the vapor pressure, T the temperature, L the latent heat, and ∆V is the volume
change associated with the phase change. This relation dictates the vapor pressure as a

28
function of temperature for a substance, when combined with known critical points. In
practice, it is more convenient to use the Antoine equation [2]:
C T
B
A P
+
− = ) ( log
10
4.2
where P is the vapor pressure, T is the temperature, and A, B, and C, are experimentally
derived constants. Condensation will occur whenever the partial pressure of a substance
is greater than that of its vapor pressure at a given temperature. For gaseous fuels at
standard conditions, the vapor pressure is much greater than that of atmospheric, so a
pure fuel stream can be experimentally tested. In contrast, this is not true for liquid fuels,
as their vapor pressures are less than that of atmospheric pressure, and thus necessitate
dilution into other gasses to lower their partial pressure below the vapor pressure. This
adds further experimental constrictions as compared to liquid fuels.
For example, at atmospheric temperature and pressure, n-C
8
H
18
can be maintained in
the gaseous phase up to φ = 0.77 in air, while n-C
12
H
26
can only go up to φ = 0.01. The
vapor pressure of n-C
12
H
26
prohibits any experiments from being conducted at
atmospheric conditions, but there is a strong dependence of vapor pressure on
temperature. Increasing the experimental temperature by, say, 100 K, increases the
maximum φ to 4.5. The strong dependence of the maximum φ on C
#
is apparent, but
there can be just as strong of dependence due to molecular classification. The maximum
φ = 3.2, at standard conditions, for iso-C
8
H
18
, is over 4 times more than the normal
paraffin of equal C
#
, i.e. n-C
8
H
18
. Due to the strong dependence on both C
#
and chemical
classifications, the heaviest component with the most stringent vapor pressure

29
requirements must be considered when vaporizing practical fuels. These requirements
impose a minimum temperature throughout the heating path. This minimum temperature
must be maintained for the entire system downstream of the vaporization chamber in
order to avoid condensation.

30
4.4 Chapter 4 References

[1] C.K. Law, Combustion Physics, Cambridge University Press, New York, 2006, p.
19.
[2] N.A. Lange, Lange’s Handbook of Chemistry, 12
th
ed., McGraw-Hill, New York,
1979.

31

Chapter 5

Objectives

In light of the considerations given above, an experimental and numerical study of the
fundamental flame properties of practical liquid fuels, both single- and multi-component,
was conducted. The goal of this study was to further the understanding of the
combustion of liquid fuels, provide fundamental flame data on actual samples of
commercial fuels, and to understand the effects associated with the variability of distillate
fuels. The specific tasks are as follows:

5.1 Experimental
Experimental determination of ignition and extinction limits for a wide range of fuels
and reacting configurations. The fuels included single-component alcohols and
hydrocarbons, commercial gasoline, commercial and military jet fuels, and synthetic
fuels. Additionally, surrogates of both gasoline and jet fuel were tested. Fuels were

32
tested under premixed and non-premixed conditions depending on the practical relevance
of the fuels. Premixed combustion is most relevant to spark ignition engines, while non-
premixed is the predominant combustion mode of turbines. Due to the vapor pressure
requirements of the fuels, experiments were conducted at both ambient and elevated
temperatures.
To facilitate the testing of liquid fuels, a vaporization system was constructed with
the capability to handle all of the fuels previously mentioned under both premixed and
non-premixed conditions. A combination of heating and insulation was used to maintain
the piping and burners at elevated temperatures when necessary. Additionally straight
tube burners were constructed to allow for experiments to be conducted at temperatures
above those that the previously designed contoured burners could operate. A laser-based
diagnostics system was utilized to measure flow velocities.

5.2 Numerical
Selected experiments for single component fuels were numerically simulated to
assess both the current state of chemical kinetic and transport parameters of liquid fuels,
and to provide physical insight into the controlling combustion mechanisms. The
sensitivity of the fundamental flame properties on the chemical kinetic and transport
parameters was also determined to help provide direction of future study.
The existing codes where modified to improve their capability to determine the
sensitivity of the flame properties on diffusion parameters rigorously.

33

Chapter 6

Experimental Approach

The experiments were conducted in the opposed-jet counterflow configuration as
shown in Fig. 6.1, which has proven to be a meritorious configuration in past studies [e.g.
1-4]. To cover the wide range of experimental data, several sets of burners were used
with varying nozzle diameters, D, ranging from 7 mm to 22 mm. The opposed jets
impinge upon each other to create a stagnation-type flow. The distance between the two
nozzle exits is defined as the separation distance, L. The location at which the two jets
meet is referred to as the stagnation plane (SP). This technique allows for the
establishment of planar flames that are axisymmetric, which can be modeled using a
quasi-one-dimensional numerical approach.

34

Figure 6.1 Schematic of the counterflow configuration.

6.1 Experimental Apparatus

Figure 6.2 Schematic of the experimental apparatus.

Flow From
Gas Tanks
Flow From
Gas Tanks
Pressure
Gauges
Sonic
Nozzles Nebulizer
Laser Optics
Camera
Computer
Liquid Fuel
Pump
Vaporization
Chamber
Heating
Element
Insulation
Material
Burners
Laser Sheet
Post Chamber
Temperature T
pc

Nozzle
Diameter
Stagnation
Plane
Separation
Distance

35
A schematic of the experimental apparatus is shown in Fig. 6.2. The gaseous flow
rates were controlled using calibrated sonic nozzles and pressure gauges. The liquid flow
was metered using a high precision liquid pump. The gaseous flow was preheated prior
to entering the vaporization chamber. Liquid fuel was injected into the vaporization
chamber surrounded by a heated gaseous co-flow (air or nitrogen) to maximize the
mixing of the two phases as shown in Fig. 6.3.

Figure 6.3 Schematic of the vaporization chamber.

As previously discussed, introducing and maintaining a liquid fuel into the gaseous
phase requires notable care. Precision control of the temperature from the point of fuel
injection to the experimental domain is required. The vaporization chamber’s wall
temperature was held constant using digitally controlled thermocouples and was chosen
based on the fuel and the prevailing thermodynamic conditions. The walls of the
chamber were thick to ensure that the heat capacity of the chamber is orders of magnitude
greater than that of the fuel-containing stream, so that changes in the flow rate have a
negligible affect on the wall temperature. The post chamber mixture temperature was
monitored also, as the wall temperature is not necessarily equal to the mixture
Heating Element
To
Burners
Liquid Fuel
N
2

N
2

Vaporization
Chamber

36
temperature. The fuel-containing stream was maintained at high enough temperatures to
avoid condensation. For some fuels this could be accomplished at room temperature, but
for other fuels elevated temperatures were required. The elevated temperature was
maintained with both insulation and heating downstream of the chamber. Finally the
mixture temperature was monitored at the burner exit to determine the boundary
conditions of the experiment accurately, and to ensure that the mixture was hot enough to
avoid condensation. The desired post chamber mixture temperature and the burner exit
temperature depend on the fuel being tested. The contoured burner assembly is capable
of maintaining the fuel-containing jet at slightly elevated temperatures, but for
temperatures above 70°C straight tube burners were used. The thinner walls allow for
more direct heating of the test flow.
A nebulizer was used to seed the flow with silicon oil droplets that were between 0.5
and 0.3 μm in diameter [5]. Both the top and bottom burner can be seeded, depending on
the experimental configuration. The flow measurements were conducted using digital
particle image velocimetry (DPIV) [6].

6.2 Verification of the Vaporization System
To ensure that the liquid fuel was successfully converted into a gaseous stream
several tests were performed to verify it. A number of fuels were chosen to test the
system, including single components, mixtures of single components, and practical
distillate fuels. A series of ignition and extinction experiments were repeated at different
vaporization temperatures. The temperatures were varied by increasing both the

37
temperature of the preheated gas and the vaporization chamber. The experiments were
conducted in the non-premixed configuration with a heat fuel/N
2
jet counterflowing
against a pure O
2
jet. The details of both the ignition and extinction technique will be
discussed subsequently.

Figure 6.4 Variation of the fuel/N
2
mass ratio at ignition as a function of post
chamber temperature for a constant ignition source.

n-C
12
H
26
JP8
S6
S12
Post-Chamber Fuel/N
2
Temperature, T
pc
,

C
o
0.065 0.060 0.055 0.050 0.045 0.040
150
200
250
300
350
Fuel/N
2
Mass Ratio
n-C
5
H
12

38
Figure 6.4 depicts the fuel/N
2
mass ratio at ignition as a function of post chamber
temperature, T
pc
. As expected, the system more readily ignites as the initial thermal
energy of the fuel/N
2
jet is increased. This is true for both pure fuels and fuel mixtures,
as well as for large and small hydrocarbons. The thermal effect is approximately linear,
and there appears to be no added chemical affect, even though the fuels are expected to
undergo some pyrolysis.

Figure 6.5 Variation of the fuel/N
2
mass ratio at extinction as a function of post
chamber temperature at constant strain rate.
JP10
Post-Chamber Fuel/N
2
Temperature, T
pc
,

C
o
n-C
5
H
12
0.095 0.090 0.085 0.080 0.075 0.070
120
170
220
270
320
370
Fuel/N
2
Mass Ratio at Extinction
flame
Heated Fuel/N
2
O
2
n-C
12
H
26
JP8
S6
S12
RP
JP

39
The fuel/N
2
mass ratio at extinction as a function of T
pc
at constant strain rate for
selected fuels is shown in Fig. 6.5. Similarly with ignition, it is expected that as the
thermal energy increases the resistance to extinction also increases, resulting in a lower
fuel/N
2
mass ratio at extinction. At both low and high T
pc
this is true, but there is also a
negative dependence on temperature found at mid range of temperatures, T
pc
between
170 °C and 270 °C, resulting in an inverted S-curve. The temperature range at which the
negative temperature dependence was found corresponds to the expected onset of
pyrolysis. A series of samples were taken at 3 different T
pc
for n-C
12
H
26
to determine if it
was a chemical effect. The three samples where tested using a gas chromatograph (GC)
to determine if chemical conversion has begun. A sample was taken at T
pc
of ambient
temperature, 250 °C, and 350 °C, and the resulting GC analysis is shown in Figs. 6.6-6.8.

Figure 6.6 GC trace at an ambient post chamber temperature.

20 °C
Minute
0 5 10 15 20 25 30 35 40
0
500
1000
1500
2000
2500
3000
3500
N
2

O
2

Millivolts

40
Figure 6.6 shows that when the system is not heated no small hydrocarbons are
detected. At the T
pc
at which the negative temperature dependence for extinction starts
for n-C
12
H
26
notable amounts of small hydrocarbons show up on the GC trace as shown
in Fig. 6.7. This proves that the onset of pyrolysis causes the negative temperature
dependence. At even higher T
pc
as shown in Fig. 6.8 the amount of chemical conversion
grows substantially. The greater extent of chemical conversion results in again
increasing the system’s resistance to extinction. The additional conversion along with the
increased thermal energy causes a sharp increase in the resistance to extinction.

Figure 6.7 GC trace at a post chamber temperature of 250 °C.
250 °C
C
2
H
4
Minute
0 5 10 15 20 25 30 35 40
0
500
1000
1500
2000
2500
3000
3500
N
2
H
2

Millivolts

41

Figure 6.8 GC trace at a post chamber temperature of 350 °C.

All of the fuels show this general inverted S-curve behavior, but at slightly different
temperatures. The onset of the inverted temperature dependence is for values of T
pc

between 170 °C and 270 °C. This shows that for all of the fuels being tested as long as
the T
pc
is kept below 170 °C there will be no chemical conversion. For all experiments
the T
pc
was kept below this limit.
The reason for the S-curve behavior being found for extinction and not for ignition is
due to the nature of the flame phenomena. Extinction states depend on the overall heat
release of the flame, while ignition primarily depends on the external enthalpy supply.
When the system is heated to the point at which chemical activity occurs, it has been
shown also that coking occurs in the chamber. Coking is the accumulation of primarily
carbon deposits on the walls of the vaporization chamber, which means carbon is
Minutes
0. 2. 5. 7. 10. 12. 15. 17. 20. 22. 25. 27. 30. 32. 35. 37.
0
200
400
600
800
1000
1200
350 °C
CO
CH
4

N
2

H
2

C
2
H
6
C
2
H
4
C
3
H
8
Millivolts

42
removed from the fuel stream. The reduction in the total amount of carbon reduces the
heat release and flame temperature causing it to become less resistant to extinction, and
subsequently necessitating an increase of the initial fuel concentration, resulting thus in
the negative temperature dependence. With further increase in the temperature, the
extent of coking increases, but the increased thermal energy of the system combined with
the modification of the fuel’s composition to more reactive species, e.g. olefins,
overpower the coking effect. This coking effect does not hinder ignition through its
effect on the heat release. It has been shown that ignition state is independent on the fuel
concentration at sufficiently large values [e.g. 7].

6.3 Fuels
The study includes a large range of liquid fuels, ranging from pure hydrocarbons to
practical fuels. Pure hydrocarbon fuels considered were those with C
5
≤ C
#
≤ C
14
, and
chemical classifications including n/iso-paraffins and aromatics. In addition to the
hydrocarbons, methanol and ethanol were also tested due to their increased use in modern
day gasoline. Practical fuels include samples of gasoline and samples of various types of
jet fuel. Gasoline and jet fuel surrogates have also been tested. In light of the increased
importance of synthetic fuels in the future, samples of synthetic fuels produced by both
the Fischer-Tropsch (FT) process and Direct Coal Liquefaction (DCL) were tested.

43
6.4 Ignition Methodology
The counterflow ignition technique is a steady state measurement of an inherently
transient phenomenon. For both premixed and non-premixed configurations the fuel-
containing jet was impinged against a hot jet that was the ignition source [e.g. 4,7]. The
ignition source was established and slowly the fuel concentration was slowly increased
until ignition occurs. The fuel concentration was increased in small step increments, after
each increase the system is allowed to achieve steady state. The step increase that results
in a flame ignition is the ignition concentration. The ignition temperature, T
ign
, was
defined as the temperature of the hot source at which ignition occurs, for given reactant
compositions and strain rate.
An ultra-lean H
2
/CO premixed flame was utilized as the ignition source by
establishing it in one of the two burners. This approach results in hot vitiated oxidizer
that contains minor amounts of CO
2
and H
2
O and negligible amounts of radicals. In this
way, chemical energy was used to provide the required temperatures rather than a
complicated burner that includes electrical heaters. The location of the heat source in
relation to the fuel-containing jet affects the determined of T
ign
. If the heat source is too
close to the stagnation plane no clear ignition transition can be determined, but rather the
fuel-containing jet’s flame merely grows out of the ignition H
2
/CO flame. If the ignition
source is sufficiently far from the stagnation plane, a clear ignition event takes place, but
the resulting T
ign
depends on the location. Therefore, the location of the heat source must
be kept constant throughout the experiment to allow for comparison between data sets
taken at different temperatures.

44
The steady state location of a flame in a flow field is the location at which its
propagation speed equals the local fluid velocity. For a lean flame resulting from a single
fuel such as H
2
, an increase in φ will increase both the propagation speed and the flame
temperature. This would cause the ignition source to become hotter and also move
further away from the stagnation plane. H
2
flames have very high propagation speeds
compared to CO flames. Conversely, CO flames tend to be hotter than H
2
flames for the
same S
u
o
. Additionally, H
2
flames can also be sustained at very low φ and temperatures
while CO flames cannot. The combination of a H
2
/CO flame allows for the independent
adjustment of the flame’s temperature and propagation speed in the temperature range
required to ignite the fuels of interest. To increase the flame temperature while
maintaining the propagation speed, a small amount of H
2
can be replaced with CO
causing the flame to burn hotter without modifying the propagation speed.

6.4.1 Premixed Configuration
Premixed experiments were performed by counterflowing a fuel/oxidizer jet against
an ultra-lean H
2
/CO/oxidizer flame. The ultra-lean flame results in a hot vitiated stream,
which was used as the ignition source. For ignition of lean premixed flames, the oxygen
content in the heated stream has a negligible effect on the results. The excess O
2
can be
thought of as an inert, just like the N
2
, that is carrying the thermal energy to the fuel-
containing jet at the SP. The diffusion of thermal energy across the SP stimulates
chemical activity in the fuel-containing jet causing the mixture to ignite. Upon ignition
the premixed flame will propagate upstream, away from the SP, to a location at which its

45
propagation speed and the convective velocity are equal. For near-stoichiometric and
rich flames, the O
2
content of the hot stream does play a significant role on the ignition.
For these cases the flame used as the ignition source must be kept near φ = 1.0 to
consume all of the O
2
, but should be diluted with N
2
to maintain both slow flame speeds
and cool temperatures.

6.4.2 Non-Premixed Configuration
Non-premixed experiments were performed by counterflowing a fuel/N
2
jet against
an ultra-lean H
2
/CO/oxidizer flame, which was used as the ignition source. The excess
oxygen that survives the ignition source is critical for the non-premixed configuration,
since it is necessary for the ignition of the fuel/N
2
jet. The necessary concentration to
ignite and sustain the non-premixed flame depends on the fuel. The resulting location of
the non-premixed flame will occur naturally at the location at which the fuel and the
oxidizer fluxes balance stoichiometrically. Depending on the initial mixture
concentrations non-premixed flames could be established on either side of the SP.

6.5 Extinction Methodology
The procedure to determine the extinction limits in the counterflow configuration is
well established [e.g. 8-10]. A single flame was established for a given combination of
the jet exit velocities and species concentrations. To ensure that the SP resides in the
center between the two burners the momenta of the two jets were equal to each other.
Subsequently, the fuel flow rate was reduced, in the case of lean flames, or increased, in

46
the case of rich flames, until extinction occurs. A slightly stronger flame than that of the
extinction condition is established, seeding was added to the flow, and the local strain
rate, K, was obtained via DPIV. It has been shown both experimentally and numerically
that the strain rate of the slightly stronger flame is almost identical to that at the
extinction point. This technique allows for the direct measurement of the extinction
strain rate, K
ext
. The concentration of the fuel in the fuel-containing jet was very small,
less than 5% of the total flow, so small variations in the fuel flow rate result in negligible
changes in the total flow rate.

6.5.1 Premixed Configuration
The premixed single flame configuration involves a fuel/oxidizer jet counterflowing
against an inert N
2
jet. The K
ext
was measured on the fuel/oxidizer jet side of the SP on
which the (premixed) flame resides. The single-flame configuration was preferred over
the symmetric twin-flame one [10], because for the same mixture concentrations this
configuration results in lower K
ext
compared to the twin-flame one. This allows for more
accurate determination of the K and operation in a lower Reynolds number (Re) regime.
Note that high Re can result in flow instabilities that can affect the quality of the data.

6.5.2 Non-Premixed Configuration
The non-premixed flame configuration involves a fuel/N
2
jet counterflowing against
an oxidizer jet. Depending on the fuel being tested and its concentration in the fuel/N
2
jet
the oxidizer ranges from air to pure O
2
. The enriched oxidizer stream strengthens the

47
system’s resistance to extinction. The minimum K that can be measured reliably was
about 50 s
-1
. K
ext
is measured on the oxidizer side of the SP, because this is where the
main branching reaction H + O
2

⇒
OH + O takes place, and the fluid mechanics
influence on this reaction is critical to the extinction process.

6.6 Digital Particle Image Velocimetry
DPIV was used to determine the flow velocities [6]. The flow field was determined
and analyzed, and the maximum axial velocity gradient along the centerline in the
hydrodynamic zone was recorded as K. Thirty image pairs were taken for each data point
over the period of a few seconds, and the average local strain rate of all 30 pairs was
determined as the experimentally determined K.
Two different seeding methods were used for the experiments, silicon oil droplets
ranging from 0.1 to 0.3 microns in diameter [5] and 0.3 microns diameter Al
2
O
3
particles.
Experiments were conducted to ensure that results obtained from both seeding methods
agree. The DPIV system is composed of a 50 mJ double pulse Nd-Yag laser (532 nm
wavelength), and a Megaplus ES 4.0/E digital camera with a resolution of 2048x2048
pixels. The resulting camera accuracy was approximately 600 pixels per centimeter, and
an image pixel depth of 10 pixels. The camera and laser timing was controlled by a
digital pulse generator (Model# DG 535). The captured images were processed using the
CIV software [11].

48
6.7 Chapter 6 References

[1] C.K. Law, D.L. Zhu, G. Yu, Proc. Combust. Inst. 21 (1986) 1419-1426.
[2] C.G. Fotache, T.G Kreutz, D.L Zhu, C.K. Law, Combust. Sci. Technol. 109
(1995) 373–393.
[3] C.K. Law Proc. Combust. Inst. 22 (1988) 1381-1402.
[4] C.M. Vagelopoulos, F.N. Egolfopoulos, Proc. Combust. Inst. 27 (1998) 513-519.
[5] T. Hirasawa, C.J. Sung, A. Joshi, Z. Yang, H. Wang, C.K. Law, Proc. Combust.
Inst. 29 (2002) 1427-1434.
[6] Y. Dong, C.M. Vagelopoulos, G. Spedding, F.N. Egolfopoulos, Proc. Combust.
Inst. 29 (2002) 1419-1426.
[7] J.A Langille, Y. Dong, M.G. Andac, F.N. Egolfopoulos, T.T Tsotsis, Combust.
Sci. Technol. 178 (2006) 635-653.
[8] F.N. Egolfopoulos, P.E. Dimotakis, Combust. Sci. Technol. 162 (2001) 19-36.
[9] Y. Dong, A.T. Holley, M.G. Andac, F.N. Egolfopoulos, S.G. Davis, P. Middha,
H. Wang, Combust. Flame 142 4 (2005) 374-387.
[10] C.K. Law, D.L. Zhu, G. Yu, Proc. Combust. Inst. 21 (1986) 1419-1426.
[11] A.M. Fincham, G.R. Spedding, Experiments in Fluids 23 (1997) 449-462.

49

Chapter 7

Numerical Approach

7.1 Numerical Code
The experiments were numerically simulated, by solving the conservation equations
of mass, momentum, species concentrations, and energy along the stagnation streamline.
The ignition and extinction limits were computed using a stagnation-flow code that was
based on the original formulation of Kee and coworkers [1] for a twin-flame
configuration, and has been subsequently modified to allow for asymmetric flame
configurations by accounting for independent boundary conditions at the exits of the two
opposing nozzles [e.g. 2,3]. Laminar flame speeds were numerically determined by using
the Premix code [4] to simulate freely propagating flames (FPF). Both codes account for
the effect of thermal radiation from CH
4
, H
2
O, CO
2
, and CO at the optically thin limit [2].
The codes were integrated with the CHEMKIN [5] and Sandia Transport [6] subroutine
libraries. The codes have also been modified to increase their capabilities in the

50
determination of sensitivity coefficients, specifically that of binary diffusion coefficients,
BDC.

7.2 Diffusion Sensitivity Modification
The sensitivities on BDCs were performed for both codes using a similar approach to
that used for S
u
o
[7,8]. The sensitivities are determined by solving the linear system of
equations given by
0 =
∂
∂
+
∂
∂
∂
∂
α α
θ
θ
F F
7.1
where F is the residual vector, θ is the solution vector (velocity field, species
concentrations, temperature), and α is the parameter vector. ∂F/ ∂θ is the numerical
Jacobian that is known for any converged solution, and ∂F/ ∂α is the variation in the
residual vector due to the perturbation of a parameter that is solved numerically. The
change of any dependent variable in the solution vector due to the modification of the
parameter can then be determined. Subsequently the ∂θ/ ∂α value is properly scaled [4]
resulting in a “logarithmic” sensitivity coefficient (LSC), ∂lnθ/ ∂lnα, of a dependent
variable to an independent variable.
The sensitivities of S
u
o
were determined by invoking the computed mass flow rate.
The sensitivities of K
ext
are calculated by invoking the computed strain rate at the location
of maximum velocity gradient upstream of the flame, similar to the experiments.

51
7.3 Extinction Approach
To allow for an accurate determination of K
ext
a two-point continuation approach was
implemented [9,10] by imposing predetermined species mass fractions at two points in
the flow field, so that the strain rate is solved as a dependent variable. The H radical was
chosen as the species due to its importance to the main branching reaction. The
maximum gradient of the H radical concentration upstream and downstream of the peak
concentration were the locations at which the species mass fractions are imposed. The
concentrations are decreased by a small percentage of their previous values, and
subsequently the flame is solved so that the strain rate is computed as part of the solution.
This approach allows the code to capture the inverted S-curve which was previously
discussed.

7.4 Chemical Kinetics Mechanisms
The bulk of fundamental combustion research is devoted to the development of
chemical kinetic mechanisms along with appropriate thermodynamic and transport
databases. This process involves a large number of input parameters that describe the
thermodynamic, chemical kinetic, and transport behavior of the system. Invariably the
number of parameters far exceeds the number of constraints leading to a non-unique
solution. It is expected that continued research and development of mechanisms will lead
to the eventual convergence of the accepted values of parameters. The specific
parameters that are inputs into the codes are the thermodynamic properties in the form of
NASA polynomials, the elements, species, and reactions with their attending rate

52
parameters, and the Lennard-Jones transport parameters. All of these parameters affect
the final output of the simulations.

53
7.5 Chapter 7 References

[1] R.J. Kee, J.A. Miller, G.H. Evans, G. Dixon-Lewis, Proc. Combust. Inst. 22
(1988) 1479-1494.
[2] F.N. Egolfopoulos, Proc. Combust. Inst. 25 (1994) 1375-1381.
[3] F.N. Egolfopoulos, C.S. Campbell, J. Fluid Mech. 318 (1996) 1-29.
[4] R.J. Kee, J.F. Grcar, M.D. Smooke, J.A. Miller, PREMIX: A Fortran Program for
Modeling Steady Laminar One-Dimensional Premixed Flames, Report No.
SAND85-8240, Sandia National Laboratories, 1985.
[5] R.J. Kee, F.M. Rupley, J.A. Miller, Chemkin-II: A Fortran Chemical Kinetics
Package for the Analysis of Gas-Phase Chemical Kinetics, Report No. SAND89-
8009, Sandia National Laboratories, 1989.
[6] R.J. Kee, J. Warnatz, J.A. Miller, A FORTRAN Computer Code Package for the
Evaluation of Gas-Phase Viscosities, Conductivities, and Diffusion Coefficients,
Report No. SAND83-8209, Sandia National Laboratories, 1983.
[7] Z. Yang, B. Yang, H. Wang, paper 237, “Inference of H-Atom Diffusion
Coefficient on Laminar Flame Simulation,” Proceedings of the Second Joint
Meeting of the U.S. Sections of the Combustion Institute, Berkeley, CA, March
2001.
[8] P. Middha, B. Yang, H. Wang, Proc. Combust. Inst. 29 (2002) 1361-1369.
[9] F.N. Egolfopoulos, P.E. Dimotakis, Proc. Combust. Inst. 27 (1998) 641-648.
[10] M. Nishioka, C.K. Law, T. Takeno, Combust. Flame 104 (1996) 328-342.

54

Chapter 8

Extinction of Premixed Flames of Practical Single Component
Liquid Fuels

8.1 Introduction
Some of the most commonly studied liquid fuels are the C
1
-C
2
alcohols, CH
3
OH and
C
2
H
5
OH, and the primary reference fuels of gasoline, n-C
7
H
16
and iso-C
8
H
18
. Alcohols
have both fundamental and practical relevance due to the importance of understanding the
chemical kinetics of oxygenates, and the mandated use of alcohols in commercial
gasoline. The primary reference fuels are the foundation of the definition of octane
number, and are used in almost every gasoline surrogate. Significant work has been done
by many groups to develop chemical kinetic mechanisms for these fuels, primarily based
on experimental data obtained in homogeneous systems. Despite their importance
though, relatively limited flame studies have been conducted.
Using the counterflow configuration, S
u
o
of mixtures of CH
3
OH and C
2
H
5
OH with air
have been determined by Egolfopoulos, Law and coworkers in the early 90’s [1,2], while

55
S
u
o
of mixtures of C
7
-C
8
hydrocarbons with air have been recently determined by Davis
and Law [3]. Traditionally, S
u
o
has been the most commonly determined flame property
for a number of reasons. First, it is in its own right a very important property of a
combustible mixture and it is free of any influences caused by mechanisms that are
external to the mixture. Second, it is conveniently used in kinetics mechanisms
validation and optimization [e.g., 4]. Fundamental extinction limits for these fuels have
never been determined. It is expected that since both propagation and extinction are
high-temperature phenomena, they must be controlled by similar kinetics. This has been
shown to be the case by sensitivity analysis performed in a recent study of ethylene/air
flames [5]. Furthermore, it has been shown [6,7] that C
1
/C
2
kinetics mechanisms that
closely predict S
u
o
of methane/air flames, they also satisfactorily predict experimentally
determined K
ext
for such flames.
In view of these considerations the main goal of this investigation was to provide
archival experimental K
ext
data for premixed flames of practical alcohol and liquid
hydrocarbon fuels and to compare them with predictions derived from detailed numerical
simulations. Through these comparisons, various published chemical kinetics
mechanisms were tested against extinction data for the first time. Additionally, the thesis
that flame propagation and extinction are controlled by similar kinetics was further
assessed. This is an important point, as in large-scale simulations of practical combustors
the semi-detailed or reduced kinetics models that are used have been typically validated
against S
u
o
data only. However, if the extinction response of the flamelets, which are

56
typically induced by fluid mechanics and/or heat loss influences, is not predicted
accurately, the fidelity of such simulations is compromised.
8.2 Experimental Approach
Experiments were conducted at atmospheric pressure and ambient temperature of
293 K. n-C
7
H
16
and iso-C
8
H
18
are limited to a maximum φ of about 2.7, while CH
3
OH
and C
2
H
5
OH are limited to a maximum φ of about 0.95 at ambient conditions. K
ext
of
mixtures of all four fuels with air was determined for the entire achievable φ-range for
each fuel. S
u
o
values were taken from recent literature and were compared with
numerical predictions. More specifically, the S
u
o
for CH
3
OH/air and C
2
H
5
OH/air
mixtures were taken from Refs. 1 and 2, respectively, while those for n-C
7
H
16
/air and iso-
C
8
H
18
/air mixtures were taken from Ref. 3.

8.3 Numerical Approach
The experimental results were simulated using a number of kinetics mechanisms
summarized in Table 8.1. The mechanisms of Fischer et al. [8] (hereafter referred to as
“FDC00”), Held and Dryer [9] (hereafter referred to as “HD98”), and Li et al. [10,11]
(hereafter referred to as “LHD03”) were used to model CH
3
OH/air mixtures; LHD03 is a
modified version of the HD98 mechanism, optimized to better predict S
u
o
of CH
3
OH/air
flames. The Marinov [12] (hereafter referred to as “MRN99”) and the FDC00
mechanisms were used to model C
2
H
5
OH/air flames. To model n-C
7
H
16
/air and iso-
C
8
H
18
/air flames the Davis and Law [3] mechanism (hereafter referred to as “DL98”) and

57
its revised version, DL98 (revised), were used. The DL98 (revised) mechanism was
derived from the original DL98 with the addition of CH
2
CHO and its kinetics. The
added, to DL98, kinetics of CH
2
CHO were taken from Wang et al. [13,14] and are listed
in Table 8.2. iso-C
8
H
18
/air flames were also modeled using the Pitsch et al. [15]
(hereafter referred to as “PPS96”) mechanism. PPS96 is a greatly reduced mechanism
that was compiled to predict S
u
o
of iso-C
8
H
18
/air mixtures.

Mechanism Tested Fuels Species Reactions Validation Against Data Obtained in
FDC00
[8]

CH
3
OH
C
2
H
5
OH
81 359
Flow Reactors, Shock Tubes,
Stirred Reactors
HD98
[9]
CH
3
OH 21 93
Static Reactors, Flow Reactors,
Shock Tubes, Laminar Flame Speeds
LHD03
[10,11]
CH
3
OH 21 93 Laminar Flame Speeds
MRN99
[12]
C
2
H
5
OH 57 383
Ignition Delays, Flow Reactors,
Laminar Flame Speeds
DL98
[3]

n-C
7
H
16

iso-C
8
H
18

68 399
Flow Reactors, Laminar Flame
Speeds
DL98 (rev)
n-C
7
H
16

iso-C
8
H
18

69 421 N/A
PPS96
[15]
iso-C
8
H
18
26 40 Laminar Flame Speeds
Table 8.1 Kinetics mechanisms used to simulate present and literature experimental data.

58
Table 8.2 CH
2
CHO reactions involved in DL98 (revised) mechanism
A a E
Reaction cm
3(n-1)
mol
-(n-1)
s
-n
cal/mole
C
2
H
3
+ O
2
= CH
2
CHO + O 3.000E+11 0.290 11.00
C
2
H
3
+ HO
2
= CH
2
CHO + OH 1.000E+13 0.000 0.00
CH
2
CHO + H = CH
3
CO + H 5.000E+12 0.000 0.00
CH
2
CHO + H = CH
3
+ HCO 9.000E+13 0.000 0.00
CH
2
CHO + H = CH
2
CO + H
2
2.000E+13 0.000 4000.00
CH
2
CHO + O = CH
2
CO + OH 2.000E+13 0.000 4000.00
CH
2
CHO + OH = CH
2
CO + H
2
O 1.000E+13 0.000 2000.00
CH
2
CHO + O
2
= CH
2
CO + HO
2
1.400E+11 0.000 0.00
CH
2
CHO + O
2
= CH
2
O + CO + OH 1.800E+10 0.000 0.00
CH
2
CHO = CH
3
+ CO 7.800E+41 -9.147 46900.00
*
CH
2
CHO + H + M = CH
3
HCO + M 1.000E+14 0.000 0.00
Low /5.200E+39 -7.297 4700.00 /
Troe /0.55 8900. 4350. 7244. /
Enhanced third body efficiencies
H
2
/2 H
2
O/6 CH
4
/2 CO/1.5 CO
2
/2 C
2
H
6
/3 C
2
H
2
/3.00 C
2
H
4
/3.00
n is the reaction order
*
Pressure dependent reaction with multiple reaction rates for different pressure regimes, the rate constants for
atmospheric pressure are shown.

In order to compare the controlling mechanisms for Freely Propagating Flames (FPF)
and Near-Extinction Flames (NEF) integrated reaction path and sensitivity analyses were
performed. Assessing the ability of all mechanisms to predict K
ext
, after having been
optimized to closely predict experimental values of S
u
o
, was one of the main goals of this
investigation.

8.4 Results and Discussion
8.4.1 Experimental Results on Extinction Strain Rates
K
ext
of mixtures of CH
3
OH, C
2
H
5
OH, n-C
7
H
16
, and iso-C
8
H
18
with air is shown in
Fig. 8.1 as functions of φ. As mentioned earlier, at 293 K unburned mixture temperature,

59
only lean CH
3
OH/air and C
2
H
5
OH/air flames were established and measured due to their
low partial pressure, while for n-C
7
H
16
/air and iso-C
8
H
18
/air flames the measurements
were extended well into the fuel-rich domain. For all data reported in this investigation,
the experimental uncertainty was determined to be less than 2% in φ and less than 4% in
K
ext
. It should be also noted that the experiments were performed for φ that are neither
ultra-lean nor ultra-rich. Thus, the effect of radiation on the flame response is negligible,
as it has been shown to be the case for mixtures that are not near to their flammability
limits [16]. Additionally, soot was not formed in the reported experiments.

Figure 8.1 Variation of the experimentally determined K
ext
with φ, for mixtures of air
with all liquid fuels considered in this study.

0
100
200
300
400
500
600
700
0.60.7 0.80.9 1 1.1 1.21.3 1.41.5 1.6
Extinction Strain Rate, K
ext
, 1/s
Equivalence Ratio, φ
CH
3
OH
C
2
H
5
OH
n-C
7
H
16
iso-C
8
H
18

60
Comparing the K
ext
of all four fuels, the alcohols appear to be more resistant to
extinction than the hydrocarbons for the same φ. Also the CH
3
OH/air and the
C
2
H
5
OH/air mixtures exhibit rather similar extinction characteristics, for the range of φ
tested. The K
ext
values of n-C
7
H
16
/air mixtures exceed those of iso-C
8
H
18
/air mixtures for
all φ tested except for the more fuel lean concentrations. These findings have
implications on flame stability; K
ext
is a measure of flame stability considering that flames
that are more resistant to extinction result in more stable combustion given that local
extinction is less likely to occur. Thus, mixing alcohols with hydrocarbons to formulate
alternative fuels for Spark Ignition Engines (SIE) could improve the overall engine
performance.

8.4.2 Numerical Predictions of Extinction Strain Rates
Figure 8.2 depicts the experimental and predicted K
ext
for CH
3
OH/air mixtures.
FDC00 under-predicts the experimental K
ext
by a factor of 2.25! The kinetics described
by FDC00 become so slow at the state of extinction as φ decreases, that solutions for
φ < 0.7 were not possible. However, flames with φ < 0.63 were experimentally
established. The K
ext
data are slightly over-predicted by HD98 for φ range tested and the
discrepancy grows as φ increases, though the average discrepancy is only 10%. LHD03
under-predicts K
ext
by factors that are slightly smaller that those of FDC00. LHD03 is a
modified version of HD98 and has been optimized to predict S
u
o
of CH
3
OH/air mixtures.
The two mechanisms are, in general, very similar. Three reaction rates were modified
(R10, R15, R16), one reaction was removed (R18), one duplicate reaction was added

61
(R5), and one 3
rd
body collision efficiency was modified (R11); R index refers to the
reactions shown in Table 8.3; details of the modifications to HD98 that resulted in
LHD03 are shown in Table 8.4. The present results reveal that modifications in the
original mechanism (HD98) that could be casually considered as “minor,” caused a large
change in the predicted K
ext
.

Figure 8.2. Variation of experimental and computed K
ext
with φ for methanol/air
flames using the FDC00, HD98, and LHD03 mechanisms.
0
100
200
300
400
500
600
0.625 0.675 0.725 0.775 0.825 0.875
Extinction Strain Rate, K
ext
, 1/s
Equivalence Ratio, φ
CH
3
OH
HD98
LHD03
FDC00

62
Table 8.3 List of reactions discussed in the text
H
2
and CO reactions
Reaction # Reaction Mechanisms containing reaction
R1 H + O
2
= O + OH All
R2 O + H
2
= H + OH All
R3 H + OH + M = H
2
O + M All
R4 H
2
+ OH = H
2
O + H All
R5 HO
2
+ OH = H
2
O + O
2
All
R6 HO
2
+ H = OH + OH All
R7 HO
2
+ H = H
2
+ O
2
All but PPS96
R8 HO
2
+ HO
2
= H
2
O
2
+ O
2
All but PPS96
R9 CO + OH = CO
2
+ H All
R10 CO + O + M = CO
2
+ M All but PPS96
R11 HCO + M = H + CO + M All
R12 HCO + O
2
= CO + HO
2
All but PPS96
R13 HCO + H = CO + H
2
All
R14 CH
2
+ O
2
= CO + H
2
O DL98, DL98 (rev), MRN99, FDC00

Alcohol reactions
Reaction # Reaction Mechanisms containing reaction
R15 CH
2
O + OH = HCO + H
2
O All but PPS96
R16 CH
2
O + HO
2
= HCO + H
2
O
2
DL98, DL98 (rev), FDC00, HD98, LHD03
R17 CH
2
OH + O
2
= CH
2
O + HO
2
MRN99, FDC00, HD98, LHD03
R18 CH
3
O + H = CH
3
+ OH MRN99, HD98
R19 CH
3
O + M = CH
2
O + H + M All but PPS96
R20 CH
3
O + O
2
= CH
2
O + HO
2
All but PPS96
R21 CH
3
OH + H = CH
2
OH +H
2
MRN99, FDC00, HD98, LHD03
R22 CH
3
OH + OH = CH
3
O + H
2
O MRN99, HD98, LHD03
R23 CH
3
OH + OH = CH
2
OH + H
2
O MRN99, FDC00, HD98, LHD03

n-C
7
H
16
/iso-C
8
H
18
reactions
Reaction # Reaction Mechanisms containing reaction
R24 CH
2
CHO + H = CH
3
+ HCO DL98 (rev), MRN99
R25 CH
2
CHO = CH
3
+ CO DL98 (rev), MRN99
R26 C
2
H
3
+ O
2
= CH
2
CHO + O DL98 (rev), MRN99, FDC00
R27 C
2
H
3
+ M = C
2
H
2
+ H + M DL98, DL98 (rev), MRN99, FDC00
R28 C
3
H
6
+ OH = C
3
H
5
+ H
2
O DL98, DL98 (rev), MRN99
R29 iso-C
4
H
8
+ O = iso-C
4
H
7
+ OH DL98, DL98 (rev), PPS96

C
3
H
5
/C
3
H
4
loop reactions
Reaction # Reaction Mechanisms containing reaction
R30 C
3
H
5
+ H = C
3
H
4
+ H
2
DL98, DL98 (rev), PPS96, MRN99
R31 C
3
H
5
+ OH = C
3
H
4
+ H
2
O DL98, DL98 (rev), MRN99
R32 C
3
H
5
+ O = C
2
H
3
HCO

+ H DL98, DL98 (rev), MRN99
R33 C
3
H
5
+ O
2
= C
3
H
4
+ HO
2
DL98, DL98 (rev), PPS96, MRN99
R34 C
3
H
4
+ H = C
3
H
5
DL98, DL98 (rev), PPS96, MRN99
R35 C
3
H
4
+ O = CH
2
O + C
2
H
2
DL98, DL98 (rev)
R36 C
3
H
4
+ O = CO + C
2
H
4
DL98, DL98 (rev), MRN99
R37 C
3
H
4
+ OH = CHO + C
2
H
4
PPS96

63
Table 8.4 List of reactions included in LHD03 that differ from HD98
A a E
cm
3(n-1)
mol
-(n-1)
s
-n
cal/mole
Modified reaction rates
CH
2
O + OH = HCO + H
2
O 3.430E+09 1.18 -447.00
CH
2
O + HO
2
= HCO + H
2
O
2
2.000E+12 0.00 11660.00
CO + O + M = CO
2
+ M 1.800E+10 0.00 2384.00
Low /1.55E+24 -2.79 4191./
H
2
/2.5 H
2
O/12 CO/1.9 CO
2
/3.8
Modified third body efficiencies of H
2
O and CO
2

HCO + M = H + CO + M 0.186E+18 -1.00 17000.00
H
2
/2.5 H
2
O/3.0 CO/1.9 CO
2
/3.0
Additional reaction
HO
2
+ OH = H
2
O + O
2
5.000E+16 0.00 22000.00
*

Removed reaction
CH
3
O + H = CH
3
+ OH 3.200E+13 0.00 0.00
n is the reaction order
*
Duplicate reaction

64

Figure 8.3 Variation of experimental and computed K
ext
with φ for ethanol/air flames
using the FDC00 and MRN99 mechanisms.

Figure 8.3 depicts the experimental and predicted K
ext
for C
2
H
5
OH/air mixtures. Both
FDC00 and MRN99 under-predict the experimental K
ext
by approximately a factor of 1.2,
with the predictions of FDC00 being slightly higher than those of MRN99. The
numerical predictions deviate more at low φ and are closer at higher φ. This suggests a
steeper dependency of K
ext
as a function of φ. Both mechanisms satisfactorily predict the
extinction characteristics of ethanol for the φ range tested.

0
50
100
150
200
250
300
350
400
450
500
0.64 0.69 0.74 0.79 0.84
Extinction Strain Rate, K
ext
, 1/s
Equivalence Ratio, φ
C
2
H
5
OH
FDC00
MRN99

65

Figure 8.4 Variation of experimental and computed K
ext
with φ for n-heptane/air
flames using the DL98 and the DL98 (revised) mechanisms.

The experimentally determined and numerically predicted K
ext
for n-C
7
H
16
/air
mixtures are shown in Fig. 8.4. The DL98 and DL98 (revised) mechanisms were used in
the numerical simulations. For conditions that are fuel-lean and near stoichiometric,
DL98 under-predicts the experimental data by almost a factor of 1.5. The discrepancy
decreases to a just under a factor of 1.3 at φ = 1.2 but then increases to its maximum of
1.9 for ultra-rich conditions. DL98 (revised) at worst under-predicts the experiments by a
factor less than 1.6 and on average predicts closer the experiments by about 25%
0
100
200
300
400
500
600
700
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6
Extinction Strain Rate, K
ext
, 1/s
Equivalence Ratio, φ
n-C
7
H
16
DL98
DL98 (Revised)

66
compared to DL98. This improvement resulted from the addition of 1 species and 11
reactions, only 3 of which have a notable effect on the burning response of n-C
7
H
16
/air
mixtures, i.e. reactions R24, R25, and R26. Both DL98 and DL98 (revised) predict the
maximum K
ext
to be at φ = 1.22, while the experimental K
ext
peaks at about φ = 1.15, the
small shift explains the closest predictions at φ = 1.2.

Figure 8.5 Variation of experimental and computed K
ext
with φ for iso-octane/air
flames using the PPS96, DL98, and the DL98 (revised) mechanisms.

0
100
200
300
400
500
600
0.65 0.75 0.85 0.95 1.05 1.15 1.25 1.35 1.45 1.55
Extinction Strain Rate, K
ext
, 1/s
Equivalence Ratio, φ
iso-C
8
H
18
PPS96
DL98 (Revised)
DL98

67
Figure 8.5 depicts the experimentally determined K
ext
values for iso-C
8
H
18
/air
mixtures along with the numerical predictions. The PPS96, DL98, and DL98 (revised)
mechanisms were used. All three mechanisms fail to predict the experimental data,
especially for lean mixtures for which they under-predict the experimental data by a
factor of 2 or more, with the DL98 (revised) exhibiting a relatively better agreement with
the experiments. All mechanisms except the DL98 (revised) tested also predicted that
K
ext
peaks at a notably greater φ compared to the experiments. The experimental K
ext

peaks around φ = 1.2 while DL98 predict the peak at approximately φ = 1.25, and PPS96
at about φ = 1.30. The maximum φ shift results in the PPS96 mechanism to only slightly
under-predict the rich extinction, by approximately a factor of 1.15. The DL98 and DL98
(revised) mechanisms predict more accurately the φ dependence, but still under-predict
the experimental results by an average factor of 1.74 and 1.57 respectively. Overall
DL98 (revised) exhibits a 17% improvement of the over the range of φ tested compared
to DL98.
Of all seven mechanisms tested, only one of them, the HD98 mechanism, comes close
to reproducing the experimental K
ext
data. Even the HD98 mechanism over-predicted the
data by 10%, and the next two closest mechanisms barely came within 20% of the
experimental results. With few exceptions, all mechanisms were found to under-predict
the experimental K
ext
by factors as large as 2. Furthermore, they all predict excessively
weak burning under fuel-lean conditions, to the point that convergence was not possible
even though flames can be established experimentally. It is of interest to note that the
HD98 mechanism was subsequently optimized to predict S
u
o
of CH
3
OH/air flames, which

68
led to the LHD03 mechanism. However, this optimization to closer predict S
u
o
causes a
rather notable under-prediction of the experimental K
ext
by LHD03 as reported in Fig. 8.2.
Based on this observation and the fact that most of the mechanisms have been tested only
against S
u
o
, the ability of all mechanisms to simultaneously predict S
u
o
and K
ext
was
further assessed and is discussed next.

8.4.3 Comparisons of Phenomena of Laminar Flame Propagation and Extinction

Figure 8.6. Variation of experimental and computed S
u
o
with φ for methanol/air
flames using the FDC00, HD98, and LHD03 mechanisms. The experimental data
were taken from Ref. 1.
0
10
20
30
40
50
0.55 0.65 0.75 0.85 0.95 1.05
Laminar Flame Speed, S
u
o
, cm/s
Equivalence Ratio, φ
Egolfopoulos et al. [1]
HD98
FDC00
LHD03

69
Figure 8.6 depicts numerical predictions and literature [1] S
u
o
data for CH
3
OH/air
mixtures. While HD98 notably over-predicts the experimental data, LHD03 closely
predicts them except for fuel-lean conditions and FDC00 slightly under-predicts them.
Integrated reaction path analyses reveals that all three mechanisms have similar
combustion paths for CH
3
OH; the percentage consumption paths for the HD98 and
LHD03 mechanisms both NEFs and FPFs are shown in Fig. 8.7 for a φ = 0.769
CH
3
OH/air flame. The reaction path analysis reveals that the combustion pathways for
NEFs and FPFs are nearly identical and that each mechanism favored slightly different
paths. For example, the FDC00 mechanism (not shown in Fig. 8.7) was found to favor
the H
2
production directly from methanol via R21 compared to the other two
mechanisms. R21 accounts for over 50% of the CH
3
OH consumption for FDC00, but
under 10% for HD98 and LHD03.
Comparison of the reaction path analysis results obtained for LHD03 that closely
predicts S
u
o
, and for HD98 that closely predicts K
ext
, reveals that relatively “minor”
changes in a mechanism can have significant changes in the prediction of global flame
properties. The main difference between the two mechanisms is in the HO
2
chemistry.
The LHD03 mechanism places a stronger importance on the HO
2
chemistry by
decreasing the third body efficiencies of H
2
O and CO
2
for reaction R11. The reduction of
the R11 rate favors the consumption of HCO through R12, which produces more HO
2
.
The HO
2
chemistry has also been modified by adding reaction R5 that shifts almost 30%
of the consumption of HO
2
towards the production of H
2
O. This has a negative effect on
the overall reactivity of the flame as it slows down the rate of R6, which produces two

70
OH radicals that are important to the overall branching. Thus, LHD03 predicts weaker
burning characteristics compared to HD98. This weakening allows LHD03 to closer
predict S
u
o
compared to HD98. However, LHD03 fails to predict K
ext
. This indicates that
validations against S
u
o
are not sufficient for a mechanism and its reduced versions to be
used in large scale simulations as other high-temperature flame phenomena are not
necessarily predicted.

Figure 8.7 Integrated species consumption paths for a φ = 0.769 NEF and FPF
methanol/air flame, computed using the HD98 and LHD03 mechanisms.
CH
3
OH
HO
2
HCO
CO
CH
2
O
CO
2
H
2
O
H
2
CH
2
OH
25.0
24.9
22.4
22.0
89.3
89.0
90.0
89.1
39.3
38.9
32.9
31.7
97.6
97.6
98.1
98.3
CH
3
O
7.9
8.2
8.0
8.8
34.6
34.5
36.6
36.4
39.0
40.4
11.5
13.3
63.4
63.1
74.4
72.2
78.3
78.1
74.3
72.9
29.7
29.6
30.5
30.2
50.3
50.2
58.5
58.3
73.3
72.8
80.5
79.0
20.0
20.5
14.5
16.1
1 LHD03 NEF
2 LHD03 FPF
3 HD98 NEF
4 HD98 FPF
9.8
9.8
14.2
14.9
OH

38.0
37.7
55.0
57.9

71
The logarithmic sensitivity coefficients of S
u
o
and K
ext
to reaction rate constants for a
φ = 0.769 CH
3
OH/air flame computed using the LHD03 mechanism are shown in Fig.
8.8. (The term “logarithmic sensitivity coefficient” was chosen over the term
“normalized sensitivity coefficient” that is properly defined in Ref. 17. Similar
discussion can be also found in Ref. 18.)

Figure 8.8 Logarithmic sensitivity coefficients of S
u
o
and K
ext
to reaction rate
constants for a φ = 0.769 methanol/air flame computed using LHD03.

HO
2
+HO
2
= H
2
O
2
+O
2
(R8)
H
2
+OH = H
2
O+H (R4)
O+H
2
= H+OH (R2)
CH
3
O+M = CH
2
O+H+M (R19)
HO
2
+H = OH+OH (R6)
CH
3
OH+OH = CH
3
O+H
2
O (R22)
H+O
2
= O+OH (R1)
HCO+M = H+CO+M (R11)
CO+OH = CO
2
+H (R9)
(R12) HCO+O
2
= CO+HO
2
(R5) HO
2
+OH = H
2
O+O
2
(R7) HO
2
+H = H
2
+O
2
(R23) CH
3
OH+OH = CH
2
OH+H
2
O
(R13) HCO+H = CO+H
2
(R17) CH
2
OH+O
2
= CH
2
O+HO
2
(R20) CH
3
O+O
2
= CH
2
O+HO
2
0.6 0.5 0.4 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.3 -0.4 -0.5
Logarithmic Sensitivity Coefficients
K
ext
S
u
o
CH
3
OH/air
φ = 0.769

72
It can be seen that while the reactions that largely affect S
u
o
and K
ext
are the same, the
sensitivities of K
ext
are approximately twice as much compared to S
u
o
. The latter
indicates the greater influence of kinetics on the phenomenon of extinction as compared
to propagation. It is apparent that the observed qualitative similarities of the sensitivities
of S
u
o
and K
ext
do not point into a rate or rates that could be responsible for the failure of
the kinetics to predict both phenomena. On the other hand, the quantitative differences of
the sensitivity coefficients that are uniformly described by a factor of 2 could be
considered as a potential source of the observed discrepancies between the two
phenomena. However, such improvements of rate constants could only be achieved via
rigorous global mechanism optimization [e.g., 4, 13], rather than individual rate constants
modifications.
The relative performance of the mechanisms between the two different flame
phenomena is summarized as follows: FDC00 under-predicts S
u
o
by about 25% and it
under-predicts K
ext
by almost 225% for the range tested. Note that the flame temperature
difference between NEF and FPF at φ = 0.769 is only 138 K. Comparing LHD03 and
HD98, it can be seen that modifying a mechanism to decrease S
u
o
by 30% reduces the
predicted K
ext
by 60%, i.e. by a factor of two that is consistent with the sensitivity results
shown in Fig. 8.8.
Figure 8.9 depicts numerical predictions using FDC00 and MRN99 and literature [2]
S
u
o
data for C
2
H
5
OH/air mixtures. It can be seen that both FDC00 and MRN99 predict
rather closely the experimental S
u
o
, and that there is on average less than 10% difference
between the predicted values and the experimental data for both mechanisms. Similarly

73
to CH
3
OH/air flames, integrated reaction path analysis reveals that there are only minor
differences between the pathways for NEFs and FPFs as predicted by the two
mechanisms. Similarly to CH
3
OH/air flames, the sensitivity analysis reveals that the two
phenomena share similar sensitivity values, with the sensitivities of K
ext
being greater
than those of S
u
o
by approximately a factor of two.

Figure 8.9 Variation of experimental and computed S
u
o
with φ for ethanol/air flames
using the FDC00 and the MRN99 mechanisms. The experimental data were
taken from Ref. 2.

10
15
20
25
30
35
40
45
50
0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.05
Laminar Flame Speed, S
u
o
, cm/s
Equivalence Ratio, φ
MRN99
FDC00
Egolfopoulos et al. [2]

74
From the analysis performed for CH
3
OH/air and C
2
H
5
OH/air FPFs and NEFs, it is
apparent that the kinetics mechanisms that were tested exhibit similar reaction paths and
sensitivities for both the phenomena of propagation and extinction, and that no reaction
was identified as responsible for the observed discrepancies.
There are several points to be made here. First, the analysis was performed based on
mechanisms that were developed based on S
u
o
predictions. Thus, if species are missing
and/or rate constants have not been optimized to account for phenomena other than
laminar flame propagation, it is possible that the reaction path and/or sensitivity analyses
performed were falsified and they cannot identify accurately the causes responsible for
the poor prediction of K
ext
. While no attempt is made to modify rate constants, the effect
of adding species to a mechanism that has been compiled based on S
u
o
predictions, was
assessed and discussed next.
Figure 8.10 depicts numerical predictions using DL98 and DL98 (revised) and
literature [3] S
u
o
data for n-C
7
H
16
/air mixtures. While both mechanisms accurately
predict S
u
o
for φ < 1.0, they slightly under-predict S
u
o
for φ ≥ 1.0. The DL98 (revised)
mechanism was an improvement for predicting S
u
o
of n-C
7
H
16
/air mixtures as well as K
ext

as shown earlier in Fig. 8.4. For example, the addition of CH
2
CHO and its kinetics
improved the predictions of S
u
o
by 9% for φ = 1.0 compared to DL98. However, Fig. 8.4
depicts that the prediction of K
ext
was improved by nearly 20% for φ = 1.0, i.e. a factor of
more than 2 improvement over that for S
u
o
. Reducing a mechanism to speed up the
processing time is necessary for these complex fuels. However, the reduction of a

75
mechanism could result in errors with respect to phenomena that have not been
investigated and which are of importance to the modeling of practical combustors. In
this particular case, the introduction of one species and 11 reactions was able to notably
improve the performance of the mechanism. Integrated reaction path analysis and
logarithmic sensitivity analyses were performed and will be discussed in reference to iso-
C
8
H
18
/air flames.

Figure 8.10 Variation of experimental and computed S
u
o
with φ for n-heptane/air
flames using the DL98 and DL98 (revised) mechanisms. The experimental data
were taken from Ref. 3.

10
15
20
25
30
35
40
45
0.55 0.65 0.75 0.85 0.95 1.05 1.15 1.25 1.35 1.45
Laminar Flame Speed, S
u
o
, cm/s
Equivalence Ratio, φ
Davis & Law [3]
DL98
DL98 (Revised)

76
Figure 8.11 depicts numerical predictions using PPS96, DL98, DL98 (revised) and
literature [3] S
u
o
data for iso-C
8
H
18
/air mixtures. While all three mechanisms closely
predict the experiments for φ < 1.0 they notably under-predict them for φ ≥ 1.0. The
DL98 (revised) mechanism is an improvement over DL98 but only by 4% on average.
The modification improves the prediction of K
ext
by 15% on average, as discussed earlier.

Figure 8.11 Variation of experimental and computed S
u
o
with φ for iso-octane/air
flames using the PPS96, DL98 and DL98 (revised) mechanisms. The
experimental data were taken from Ref. 3.

5
9
13
17
21
25
29
33
0.60.7 0.80.9 1 1.1 1.21.3 1.41.5 1.6
Laminar Flame Speed, S
u
o
, cm/s
Equivalence Ratio, φ
Davis & Law [3]
DL98
DL98 (Revised)
PPS96

77

Figure 8.12 Integrated species consumption paths for a φ = 0.873 NEF and FPF iso-
octane/air flame, computed using the DL98 and DL98 (revised) mechanisms.

Integrated reaction path analysis reveals, again, relatively minor differences between
NEFs and FPFs for iso-C
8
H
18
/air mixtures. This is shown in Fig. 8.12 in which only the
“lower section” of the hydrocarbon oxidation chain is shown, due to its considerable size
and the fact that all of the fuel consumption paths largely converge through these
molecules: C
2
H
4
, C
3
H
4
, and C
3
H
5
. The added CH
2
CHO species and chemistry is also
highlighted in Fig. 8.12. The extra C
2
H
3
consumption pathway (R26) leading to
CH
2
CHO and then to CH
3
via R24 and R25 strengthens the overall burning intensity.
C
2
H
4
C
3
H
4
C
3
H
5
C
2
H
3
C
2
H
3
HCO
C
2
H
2

CH
2

HCCO
HCO
CO CH
2
O
CH
3
CO
2
H
2
O H
2
CH
2
CHO
87.2
86.2
88.2
86.7
38.3
35.5
41.4
37.2
100.0
100.0
100.0
99.9
59.6
63.0
56.7
61.8
38.2
34.7
41.3
35.9
20.1
20.1
20.1
20.1
79.9
79.9
79.9
79.9
44.2
55.2
39.8
56.2
88.6
83.9
90.7
83.9
99.2
98.9
99.2
98.9
54.1
43.1
58.4
42.0
71.5
68.8
71.6
67.0
77.0
74.0
77.6
73.5
56.4
56.8
58.4
58.5
21.7
24.5
20.9
25.1
82.5
85.5
83.4
85.0
60.1
62.5
35.0
40.2
---
---
84.0
85.9
---
---
34.7
27.8
43.5
45.8
40.9
44.4
φ = 0.873
Iso-octane/air

Legend
DL98 NEF
DL98 FPF
Rev. DL98 NEF
Rev. DL98 FPF

Added paths:

78
The CH
3
path is favored over the main C
2
H
2
path (R27) since it adds more radicals to the
radical pool. The C
2
H
2
path (R27) is still important since it leads to the CO production
through HCCO and CH
2
, which in turn accelerates the heat release rate via the CO
oxidation. R26 constitutes between 25-35% of the consumption of C
2
H
3
and affects
NEFs more notably compared to FPFs.
Figure 8.13 depicts the logarithmic sensitivity coefficients of S
u
o
and K
ext
to reaction
rate constants for a φ = 0.873 iso-C
8
H
18
/air mixtures calculated using DL98. Similar to
the other fuels, the sensitivity coefficients of K
ext
are greater compared to S
u
o
. The
sensitivities are again qualitatively similar for the two phenomena with the exception of
reactions between C
3
H
4
+ O (i.e. R35 and R36) that do not result in C
3
H
5
(i.e. R34). It
was found that the sensitivity of

K
ext
to R35 and R36 for DL98 and R37 for PPS96 are
dominant. In Fig. 14 the sensitivities to R35 and R36 are shown and their effects are
apparent. On the other hand, the sensitivities of S
u
o
to R35, R36, and R37 were
determined to be minor. As expected, the main branching (R1) and the main CO-
oxidation (R9) reactions appear to affect notably K
ext
, but by far less than the C
3
H
4
+ O
channels.

79

Figure 8.13 Logarithmic sensitivity coefficients of S
u
o
and K
ext
to reaction rate
constants for a φ = 0.873 iso-octane/air flame computed using the DL98
mechanism.

The high sensitivities found for K
ext
as compared to S
u
o
, magnifies the significance of
a special set of reactions of C
3
H
5
and C
3
H
4
that are key to the oxidation of both n-
C
7
H
16
/air and iso-C
8
H
18
/air flames. A loop exists between C
3
H
5
and C
3
H
4
that is
basically described by reactions R30, R31, R33, R34, R35, R36, and R37 that cause the
1.1 0.9 0.7 0.5 0.3 0.1 -0.1
Logarithmic Sensitivity Coefficients
K
ext
S
u
o
iso-C
8
H
18
/air
φ = 0.873
(R1) H+O
2
= O+OH
(R11) HCO+M = H+CO+M
(R9) CO+OH = CO
2
+H
(R12) HCO+O
2
= CO+HO
2
(R30) C
3
H
5
+H = C
3
H
4
+H
2
(R3) H+OH+M = H
2
O+M
(R29) iso-C
4
H
8
+O = iso-C
4
H
7
+OH
(R31) C
3
H
5
+OH = C
3
H
4
+H
2
O
(R28) C
3
H
6
+OH = C
3
H
5
+H
2
O
(R32) C
3
H
5
+O = C
2
H
3
HCO+H
(R35) C
3
H
4
+O = CH
2
O+C
2
H
2
(R36) C
3
H
4
+O = CO+C
2
H
4

80
bulk of C
3
H
5
to be converted to C
3
H
4
and then nearly 90% of C
3
H
4
to form back C
3
H
5
.
This loop can be seen in the path diagram of Fig. 8.12. The analysis reveals that while
R35 and R36 are responsible for a rather small portion of the C
3
H
4
consumption, they
constitute “exit” channels from this loop that “traps” mass and slows down the overall
reaction intensity. Thus, the sensitivity of K
ext
to the response of this “loop” is rather
large. This point is further supported by the fact that R30 exhibits the largest negative
sensitivity, which is physically reasonable given that R30 constitutes the main loop-
initiating step. Additionally, R32 exhibits notable positive sensitivity, as it is a
consumption path of C
3
H
5
that bypasses the C
3
H
5
⇔ C
3
H
4
loop. The presence of this
loop slows down the rate of conversion of the main fuel to CO
2
and H
2
O, increasing thus
the time between fuel consumption and heat release. This effect is more profound for
NEFs for which residence time considerations at the state of extinction are crucial. This
is not the case for FPFs for which the flame is allowed to consume fully all intermediate
species.

8.4.4 Diffusion Effects
While typical analyses of computed flames are usually limited to chemical kinetics
effects, it is of interest to also assess diffusion effects that have been shown to be very
important for highly diffusive fuel molecules such as H
2
[18]. In the present
investigation, the fuel molecules are large and the effect of their diffusivities could be
important also. To test this thesis, the mass diffusivities of all fuels were perturbed in the
numerical simulations of both S
u
o
and K
ext
. This was done using a brute force approach,

81
where the perturbation was performed in the transport database prior to running the
simulation. Multiple runs with the exact same boundary conditions but with the
perturbations to the transport were performed and the resulting variation in the predicted
quantity was used to calculate the sensitivity, this is similar to the approach taken in Ref.
19. The logarithmic sensitivities of S
u
o
and K
ext
to the mass diffusivities were calculated
using the following equations in which the superscripts 1 and 2 correspond to the
unperturbed and perturbed values respectively for both the independent (i.e. fuel
diffusivities) and the dependent (i.e. S
u
o
and K
ext
) variables:
( )
()
D
D D
S
S S
o
u
o
u
o
u
1
1 2
1
1 2
−
−

( )
()
D
D D
K
K K
ext
ext ext
1
1 2
1
1 2
−
−
8.1, 8.2
The sensitivities of S
u
o
and K
ext
to the fuel diffusivities were then compared to those on
the main branching reaction R1 and the results are shown in Table 8.5.
For both lean iso-C
8
H
18
/air and n-C
7
H
16
/air flames, S
u
o
exhibits a small negative
sensitivity to diffusion with values that are an order of magnitude less compared to the
sensitivity to R1. On the other hand, the effect of diffusion on K
ext
appears to be notably
greater and of the same order compared to that of R1.
Similar results are shown for lean C
2
H
5
OH/air flames, with the exception that the
sensitivity of S
u
o
to diffusion is rather similar in magnitude to that to R1. For lean
CH
3
OH/air flames it can be seen that the magnitudes of the sensitivities of S
u
o
to
diffusion and R1 are very close, while that of K
ext
to R1 is an order of magnitude greater
compared to diffusion.

82

Fuel Mechanism φ
Sensitivity to
R1
Sensitivity
to Fuel
Diffusion
iso-octane DL98
S
u
o
0.873 0.320 -0.017
K
ext
0.873 0.409 0.186
n-heptane DL98
S
u
o
0.800 0.308 -0.056
K
ext
0.808 0.397 0.185
ethanol FDC00
S
u
o
0.682 0.455 -0.137
K
ext
0.682 0.794 0.422
methanol FDC00
S
u
o
0.769 0.347 -0.316
K
ext
0.769 0.581 0.032
Table 8.5 Logarithmic sensitivity coefficients of S
u
o
and K
ext
to the rate of the main
branching reaction R1 and to the fuel diffusivity for methanol/air, ethanol/air, n-
heptane/air, and iso-octane/air mixtures.

Negative sensitivities of S
u
o
on diffusion coefficients have been reported also in Ref.
17. In the absence of stretch, as is the case of FPFs, it can be shown [e.g., 18] that if the
fuel diffusivity increases by a certain amount, the diffusive layer of the fuel increases at a
greater rate reducing thus the fuel diffusive flux into the reaction zone. As a result the
burning intensity of these fuel-lean flames is reduced. This is not the case though for
stretched flames at their extinction states for which an increase of the diffusivity of the
heavier fuels effectively reduces the mixture’s Le number. This in turn augments the
gain of fuel flux relatively to the heat loss out of the flame zone. For the case CH
3
OH/air
flames whose Le number is close to one, this effect is less profound.

83
The results presented in Table 5 clearly suggest that the effect of uncertainties of
diffusion coefficients cannot be overlooked, as this could compromise the value of
kinetics rates that are validated against experimental data on flame propagation and
extinction.

8.5 Conclusions
A systematic study was conducted on the experimental and numerical determination
of K
ext
of mixtures of methanol, ethanol, n-heptane, and iso-octane with air under
atmospheric temperature and pressure. The experiments involved the use of the
counterflow configuration, a high-precision liquid fuel feeder, and a Digital Particle
Image Velocimetry system for the accurate measurement of flow velocities. The
numerical simulations of the experiments involved the use of seven recent kinetics
mechanisms that have been validated against data obtained in flow, static, and stirred
reactors, shock tubes, and laminar flame speeds.
Results show that for the same equivalence ratio, the alcohol flames are more
resistant to extinction compared to n-heptane and iso-octane under fuel lean conditions.
Comparisons between experimentally determined and computed K
ext
reveal that
mechanisms that closely predict S
u
o
fail to predict K
ext
by factors as large as 2 or more,
suggesting that while propagation and extinction are both high-temperature phenomena,
the controlling kinetics may notably differ. This finding also reveals that validating a
mechanism in flames against S
u
o
only is not sufficient in describing the global flame
response.

84
Through the aid of detailed sensitivity and reaction path analyses, further insight was
provided into the mechanisms controlling NEFs and FPFs. The reaction path analysis
reveal that the species consumption is rather similar for NEFs and FPFs. However, it was
found that the sensitivities to kinetics are higher notably for NEFs compared to FPFs.
Furthermore, for n-heptane/air and iso-octane/air flames a set of reactions was identified
having dominant sensitivities for NEFs but not for FPFs.
The effect of diffusion was assessed also and sensitivity coefficients of S
u
o
and K
ext
to
the fuel diffusivity were determined for all fuels and compared to the ones obtained for
the main branching reaction H + O
2

⇒
OH + O. Results show that under certain
conditions, the sensitivities on diffusion can be of the same order with those on kinetics.
Thus, validation of kinetics against fundamental flame phenomena without considering
the effects of the uncertainties of diffusion coefficients, can compromise the fidelity of
the rate constants that are derived from flame studies.

85
8.6 Chapter 8 References
[1] F.N. Egolfopoulos, D.X. Du, C.K. Law Combust. Sci. Technol. 83 (1992) 33-75.
[2] F.N. Egolfopoulos, D.X. Du, C.K. Law Proc. Combust. Inst. 24 (1992) 833-841.
[3] S.G. Davis, C.K. Law Proc. Combust. Inst. 27 (1998) 521-527.
[4] G.P. Smith, D.M. Golden, M. Frenklach, N.W. Moriarty, B. Eiteneer, M.
Goldenberg, C.T. Bowman, R.K. Hanson, S. Song, W.C. Gardiner, Jr., V.
Lissianski, Z.Qin, GRI-Mech 3.0, http://www.me.berkeley.edu/gri_mech/ (2000).
[5] F.N. Egolfopoulos, P.D. Dimotakis, Combust. Sci. Technol. 162 (2001) 19-36.5
[6] F.N. Egolfopoulos, Proc. Combust. Inst. 25 (1994) 1375-1381.
[7] F.N. Egolfopoulos, H. Zhang, Z. Zhang, Combust. Flame 109 (1997) 237-252.
[8] S.L. Fischer, F.L. Dryer, H.J. Curran, International Journal of Chemical Kinetics.
32 (12) (2000) 713-740.
[9] T.I. Held, F.L. Dryer, International Journal of Chemical Kinetics. 30 (11) (1998)
805-830.
[10] J. Li, Z. Zhao, A. Kazakov, F.L. Dryer, “An updated comprehensive kinetics
model of H
2
combustion,” in Chemical and Physical Processes in Combustion,
the 2003 Technical Meeting of the Eastern States Section of the Combustion
Institute, University Park, PA, October 2003, pp. 169-171.
[11] J. Li, F.L. Dryer, personal communications, 2004.
[12] N.M. Marinov, International Journal of Chemical Kinetics. 31 (3) (1999) 183-
220.
[13] H. Wang, A. Laskin, Z.M. Djurisic, C.K. Law, S.G. Davis, D.L. Zhu “A
comprehensive mechanism of C2Hx and C3Hx fuel combustion," in Chemical
and Physical Processes in Combustion, the 1999 Fall Technical Meeting of the
Eastern States Section of the Combustion Institute, Raleigh, NC, October, 1999,
pp. 129-132.
[14] H. Wang, personal communications, 2004.
[15] H. Pitsch, N. Peters, K. Seshadri, Proc. Combust. Inst. 26 (1996) 763-771.
[16] F.N. Egolfopoulos, Proc. Combust. Inst. 25 (1994) 1375-1381.

86
[17] R.J. Kee, J.F. Grcar, M.D. Smooke, J.A. Miller, PREMIX: A Fortran Program for
Modeling Steady Laminar One-Dimensional Premixed Flames, Report No.
SAND85-8240, Sandia National Laboratories, 1985.
[18] Y. Dong, A.T. Holley, M.G. Andac, F.N. Egolfopoulos, S.G. Davis, P. Middha,
H. Wang, Combust. Flame 142 4 (2005) 374-387.
[19] C.M. Vagelopoulos, F.N. Egolfopoulos, Proc. Combust. Inst. 27 (1998) 513-519.

87

Chapter 9

Extinction of Benzene/Air and Toluene/Air Premixed Flames

9.1 Introduction
In the previous chapter the K
ext
of the two representative alkanes and alcohols was
determined. However, practical fuels contain aromatic compounds. For some fuels the
presence of aromatics is essential for engine operation, while for other fuels it is due to
the distillation process. For example, aromatics are known to suppress auto-ignition in
gasoline engines [e.g., 1], and thus they are added to eliminate engine knock. Fuels have
evolved over time to meet new restrictions, and during that evolution aromatics have
become absolutely necessary [2]. Aromatic compounds are approximately 30% of
modern day gasoline [e.g., 3], and they exist also in variable amounts in jet and diesel
fuels [e.g., 4]. Due to their presence in practical fuels, aromatics are essential
components in the corresponding fuel surrogates. Experimental data of the pure
components contained in a surrogate fuel are critically important towards the validation
of the kinetic mechanism of each component. To this end, extensive data has been

88
obtained and models have been generated for selected aromatics. Benzene and toluene
are the most widely studied, with the bulk of the work being carried out in homogeneous
systems.
Benzene shock tube studies have included both pyrolysis studies [e.g., 5-13] to
measure species time evolutions and soot formation, and oxidation studies [e.g., 14-17] to
measure ignition delay times and species time evolutions. Turbulent flow reactors [e.g.,
18-21] and jet stirred reactors [e.g., 22,23] have been used also to study benzene.
However, there are limited studies in which fundamental flame properties have been
measured. Species profiles have been determined for burner-stabilized premixed flames
[24-26]. Utilizing the twin flame counterflow configuration and the spherical bomb
technique the propagation rate has been determined also [27-30]. Additionally, limited
lean extinction limits have been experimentally determined at elevated temperatures in
the counterflow configuration [30].
For toluene a similar set of experimental data exists in the literature. Ignition delay
times and species time evolutions have been measured in rapid compression machines
[e.g., 31-36]. Extensive shock tubes data on ignition delay times, pyrolysis, and species
time evolutions exist also in the literature [e.g. 37-45]. Turbulent flow reactors, jet stirred
reactors, and plug flow reactors have been employed also to gain further understanding of
toluene [e.g., 18-21,46-48]. Propagation rates have been determined as well in both the
counterflow configuration and the spherical bomb [e.g., 27-29,49,50]. Additionally,
species profiles and the non-premixed flame extinction limits have been determined in
the counterflow configuration over a condensed pool [51,52]. Limited studies of pre-

89
vaporized non-premixed flame ignition of toluene have been performed using the
counterflow configuration [53]. Despite all this work on both benzene and toluene, there
has never been a systematic study of the extinction criteria of aromatics.
Based on these considerations, the main goal of this work was to provide extensive
fundamental experimental data on the extinction of premixed benzene/air and toluene/air
flames, and compare them to numerical predictions obtained by invoking currently
published kinetic mechanisms. An additional goal was to compare the extinction limits
of aromatic compounds to other fuels which are commonly used in gasoline surrogates,
namely n-heptane and iso-octane. This work is intended to provide necessary data for the
improvement of kinetic mechanisms of aromatic compounds. It should also serve to aid
in the development of accurate gasoline and jet fuels surrogates.

9.2 Experimental Approach
Experiments were conducted at ambient temperature and pressure, which results in a
maximum φ attainable in air for benzene (C
6
H
6
) of over 3.0, but only 1.3 for toluene
(C
6
H
5
CH
3
). This is due to the attendant vapor pressures of the fuels at ambient
conditions. The experimental data range was 0.5 ≤ φ ≤ 1.7 for benzene, and 0.6 ≤ φ ≤ 1.2
for toluene with 60 s
-1
≤ K
ext
≤840 s
-1
. Due to the large strain rate range covered, 3
different burner nozzle diameters were utilized, to maintain laminar flow. For
K
ext
≤ 100 s
-1
, 22-mm diameter nozzles were used, for K
ext
between 100 s
-1
≤ K
ext
≤ 500 s
-
1
14-mm diameter nozzles were used, and for K
ext
> 500 s
-1
7-mm diameter were used.
All experiments were conducted with a ratio of nozzle separation distance to nozzle

90
diameter L/D = 1. The experimental uncertainty was determined to be less than 3% in φ,
and less than 5% in K
ext
.

9.3 Numerical Approach
The kinetic mechanism that was used to simulate benzene and toluene was a modified
version of the mechanism by Sivaramakrishnan et al. [41], hereafter referred to as the
modified “STB” mechanism. It is the product of many previous studies, with the most
recent improvements aiming towards better prediction of high-pressure data. The basis of
STB is the work of Emdee et al. [18], and was updated subsequently multiple times.
Modifications to improve the ability of the mechanism to predict S
u
o
were done by Davis
et al. [27]. Further improvements were done to the mechanism to allow for predictions of
data derived for toluene/n-butane blends [48]. The modified STB model that was used in
this study contains 97 species and 538 reactions.
To further assess the model, simulations of S
u
o
were performed also and compared to
the literature data of Davis et al. [27]. This was done to test, again, the relative ability of
a model to predict both phenomena.

9.4 Results and Discussion
The experimentally and numerically determined K
ext
of premixed benzene/air flames
are shown in Fig. 9.1 as functions of φ. The symbols represent the experimental data,
while the line the numerical predictions. As expected, K
ext
increases as φ approaches

91
unity, and subsequently decreases for rich flames. The peak K
ext
occurs at φ ≈ 1.25, with
a maximum value of K
ext
≈ 850 s
-1
. For the extreme values of φ that flames were
stabilized and extinguished, i.e. 0.56 and 1.67, K
ext
was found to be below 60 s
-1
.
Additionally, no soot was observed for all φ tested.

Figure 9.1 Experimentally and numerically determined K
ext
of premixed benzene/air
flames as functions of φ.

Comparing the experimental data and numerical predictions, it can be seen that the
experimental data are under-predicted on the lean side by factors as large as 2. On the
0
100
200
300
400
500
600
700
800
900
0.5 0.7 0.9 1.1 1.3 1.5 1.7
Extinction Strain Rate, K
ext
, 1/s
Equivalence Ratio, φ
Experiments
Simulations

92
rich side a closer agreement is observed. The computed peak K
ext
is only 660 s
-1
, which
is almost 30% below the maximum experimental value. The computed K
ext
peaks at
approximately the same φ as the experimental data. The numerical curve appears to be
slightly shifted to higher φ. One should note that the modified STB mechanism was not
optimized against flame extinction data, but instead against laminar flame speed data
[28].

Figure 9.2 Literature experimental data [27] and numerical predictions of the S
u
o
of
benzene/air flames as a function of φ.

20
25
30
35
40
45
0.75 0.85 0.95 1.05 1.15 1.25 1.35 1.45
Laminar Flame Speed, S
u
o
, cm/s
Equivalence Ratio, φ
Davis et al. [27]
Simulations

93
Figure 9.2 depicts literature experimental data [27] and current numerical predictions
of S
u
o
as a function of φ. Similar to K
ext
, S
u
o
is under predicted also on the lean side. The
experimental data and numerical predictions differ by as much as a factor of 1.5 on the
lean side, while close agreement is seen for rich flames. The same shift of the numerical
predictions to higher φ as compared to the experimental data was found for both S
u
o
and
K
ext
. This shows that the mechanism under predicts both high temperature flame
properties for lean flames, and exhibits incorrect dependency on φ.

Figure 9.3 Fuel consumption reaction paths for a φ = 1.2 C
6
H
6
/air flame, for both a
FPF and a NEF.

To assess the difference in the mechanism’s predictions of the two phenomena
reaction path analysis and reaction rate sensitivities were calculated. Figure 9.3
illustrates the initial fuel decomposition reactions for a φ = 1.2 benzene/air flame for both
C
6
H
6
C
6
H
5
C
6
H
5
O C
6
H
5
OH
C
6
H
4
O
2
C
5
H
5
C
5
H
6
C
5
H
4
O C
2
H
2
C
4
H
5
-N H
2
CCCH
42.5
44.4
61.0
60.1
12.6
14.5
35.6
31.7
17.4
20.5
16.3
14.1
17.8
15.9
100
100
70.6
77.5
14.5
10.3
14.6
12.1
66.3
65.4
86.4
84.6
99.0
99.2
39.1
34.8
C
6
H
6
C
6
H
5
C
6
H
5
O C
6
H
5
OH
C
6
H
4
O
2
C
5
H
5
C
5
H
6
C
5
H
4
O C
2
H
2
C
4
H
5
-N H
2
CCCH
42.5
44.4
61.0
60.1
12.6
14.5
35.6
31.7
17.4
20.5
16.3
14.1
17.8
15.9
100
100
70.6
77.5
14.5
10.3
14.6
12.1
66.3
65.4
86.4
84.6
99.0
99.2
39.1
34.8
FPF %
NEF %

94
a Freely Propagating Flame (FPF) and a Near Extinction Flame (NEF), where the
numbers are the % of the species consumed that the arrows originate from. The fuel is
consumed through the same pathways for both phenomena, with only minor differences
being seen, which are primarily due to the slightly lower flame temperature of a NEF as
compared to FPF. The reaction path diagram reveals also multiple carbon loops, and
recombination reactions. For example, a loop exists between C
6
H
5
OH and C
6
H
5
O, where
the only exit path from C
6
H
5
OH is to reform C
6
H
6
. The two main decomposition paths
of C
6
H
6
both have recombination paths, one directly and one indirectly.

Figure 9.4 Logarithmic reaction rate sensitivities for a φ = 1.2 benzene/air flame, for
both a FPF and NEF.
H+O
2
=O+OH
CO+OH=CO
2
+H
C
6
H
5
O=CO+C
5
H
5
C
6
H
5
O+O=C
6
H
4
O
2
+H
C
5
H
5
=C
2
H
2
+H
2
CCCH
C
5
H
5
+H=C
5
H
6
C
6
H
6
+O=C
6
H
5
O+H
C
6
H
5
+H=C
6
H
6
C
6
H
5
O+H=C
6
H
5
OH
-0.80 -0.40 0.00 0.40 0.80
Logarithmic Sensitivity
Propagation
Extinction
H+O
2
=O+OH
CO+OH=CO
2
+H
C
6
H
5
O=CO+C
5
H
5
C
6
H
5
O+O=C
6
H
4
O
2
+H
C
5
H
5
=C
2
H
2
+H
2
CCCH
C
5
H
5
+H=C
5
H
6
C
6
H
6
+O=C
6
H
5
O+H
C
6
H
5
+H=C
6
H
6
C
6
H
5
O+H=C
6
H
5
OH
-0.80 -0.40 0.00 0.40 0.80
Logarithmic Sensitivity
Propagation
Extinction

95
Reaction rate sensitivities were calculated for the same two flames and the results are
shown in Fig. 9.4. As expected negative sensitivities are seen for the loop initiation
reactions and recombination reactions, and positive sensitivities are seen for reactions
that bypass the loops and recombination steps. Both FPFs and NEFs are sensitive to the
same reactions and have similar relative sensitivities among those reactions. NEFs are
much more sensitive to all reactions than FPFs, by more than a factor of 2.

Figure 9.5 Logarithmic reaction rate sensitivities on the mass burning rate of FPFs
for benzene/air mixtures at various φ.

Only minor differences were seen between the mechanisms ability to predict the K
ext

and the S
u
o
, but for both phenomena the mechanism fails to predict the correct
dependence on φ. Figure 9.5 exhibits the logarithmic reaction rate sensitivities on the S
u
o

-0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Logarithmic Sensitivity Coefficient
0.8
1.2
1.5
C
6
H
5
+H=C
6
H
6
C
6
H
5
O+H=C
6
H
5
OH
C
6
H
4
O
2
=C
5
H
4
O+CO
C
6
H
5
O+O=C
6
H
4
O2+H
CO+OH=CO
2
+H
HCO+M=H+CO+M
C
6
H
5
+O
2
=C
6
H
4
O
2
+H
C
6
H
5
O=CO+C
5
H
5
H+O
2
=OH+O
φ = 0.8
φ = 1.2
φ = 1.5

96
for lean, near-stoichiometric, and rich benzene/air flames. In general, as φ increases the
reaction rate sensitivity also increases. The opposite effect is seen for the
CO + OH ⇒ CO
2
+ H and C
6
H
4
O
2
⇒ C
5
H
4
O + CO reactions, which result in the two
largest sensitivities found for lean flames. Increasing either of these two reaction rates
would shift the numerical results to the left with respect to φ, but would increase
simultaneously the overall activity throughout the φ domain. Other reaction rates would
require further modification to continue to shift the results to lower φ and reduce the
overall reactivity of the system.

Figure 9.6 Experimentally and numerically determined K
ext
for premixed toluene/air
flames as functions of φ.
0
100
200
300
400
500
600
700
0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
Extinction Strain Rate, K
ext
, 1/s
Equivalence Ratio, φ
Experiments
Simulations

97

Figure 9.6 depicts the experimentally and numerically determined K
ext
as functions of
φ for premixed toluene/air flames. The symbols represent the experimental data, while
the line the numerical predictions. The φ-range that could be determined at ambient
temperature and pressure is limited by the toluene’s vapor pressure, and flames above
φ = 1.3 cannot be established. Due to added experimental difficulty at working with
liquid fuels near their vapor pressure limits the data reported in this paper are limited to
φ = 1.2, to ensure that all data taken is reliable. Comparing the experimental data and
numerical predictions, again it can be seen that experimental data were underpredicted by
factors as large as 2 for lean flames, similarly to what was found for benzene/air flames.
Due to the limited range of φ that could be tested it is not known how the mechanism
would perform for rich flames.
Figure 9.7 depicts the predicted S
u
o
of toluene/air flames as a function of φ along with
literature experimental data taken by Law and co-workers [27]. The results are similar to
those found for benzene/air flames, with underprediction on the lean side, slight
overprediction on the rich side, and the shift in peak S
u
o
to higher φ. Reaction path and
logarithmic sensitivity analyses of the toluene/air flames revealed similar results to those
of benzene/air flames.

98

Figure 9.7 Literature experimental data [27] and numerical predictions of the S
u
o
of
toluene/air flames as a function of φ.

The experimentally determined K
ext
for benzene/air and toluene/air flames, and
literature data of n-heptane/air and iso-octane/air flames [54] are compared in Fig. 9.8.
Among the four fuels, benzene/air flames are the most resistant to extinction, while iso-
octane/air flames are the least resistant ones. The peak K
ext
of benzene/air flames is
approximately 150% of that of iso-octane/air flames. Toluene/air and n-heptane/air
flames appear to have the same extinction resistance for the φ-range tested.
20
22
24
26
28
30
32
34
36
38
40
0.75 0.85 0.95 1.05 1.15 1.25 1.35 1.45
Laminar Flame Speed, S
u
o
, cm/s
Equivalence Ratio, φ
Davis et al. [27]
Simulations

99

Figure 9.8 Experimentally determined K
ext
for premixed benzene/air, toluene/air, n-
heptane/air, and iso-octane/air flames as functions of φ.

The results of Fig. 9.8 could have significant implications on the selections of
gasoline surrogates. Gasoline surrogates are made up of as many as four species
including, n-alkanes, iso-alkanes, cyclo-alkanes, and aromatics. Through the
combination of the different species the extinction response could be tuned, using the
species shown in Fig. 9.8 the extinction response of the surrogate can vary within the
0
100
200
300
400
500
600
700
800
900
0.5 0.7 0.9 1.1 1.3 1.5 1.7
Extinction Strain Rate, K
ext
, 1/s
Equivalence Ratio, φ
C
6
H
6
C
6
H
5
CH
3
iso-C
8
H
18
n-C
7
H
16

100
extreme ranges of benzene/air flames and iso-octane/air flames. Usually the aromatic
content of gasoline surrogates is vital to modify its ignition characteristics to match that
of the real gasoline. The addition of the aromatics, especially if it is benzene, could also
increase the surrogate’s resistance to extinction. It has been shown in previous work, that
the extinction resistance of real gasoline is similar to pure iso-octane [55], and therefore
toluene would be a better choice of aromatic compound compared to C
6
H
6
for gasoline
surrogates.

9.5 Conclusions
The extinction strain rates of premixed benzene/air and toluene/air flames were
experimentally determined over a wide range of equivalence ratios at ambient
temperature and pressure. The experiments were conducted in the counterflow
configuration and utilized DPIV to accurately determine the flow field. The experimental
data were simulated using a modified version of the STB mechanism, and compared with
the experimental results. It was found that the mechanism tends to under predict the
experimental results for fuel-lean conditions. The mechanism shows also a shift in the
equivalence ratio to higher values.
The experimental data were compared with those taken at the exact same
experimental configuration and conditions for n-heptane/air and iso-octane/air flames. It
was found that benzene/air flames are the most resistant to extinction, followed by
toluene/air and n-heptane/air flames, and finally iso-octane/air flames are the least
resistant to extinction. Due to previous work that shows that gasoline’s resistance to

101
extinction is close to that of iso-octane, toluene is recommended as a better choice than
benzene for the creation of gasoline surrogates.

102
9.6 Chapter 9 References

[1] W.G. Lovell, Ind. Chem. Engng. 40 (1948) 2388-2438.
[2] T.J. Wallington, E.W. Kaiser, J.T. Farrell, Chem. Soc. Rev. 35 (2006) 335-347.
[3] A.P. Singh, S. Mukherji, A.K. Tewari, W.R. Kalsi, A.S. Sarpal, Fuel 82 (2003)
23-33.
[4] Defense Energy Support Center, Petroleum Quality Information System, annual
fuel property surveys, http://www.desc.dla.mil/DCM/DCMPage.asp?pageid=99.,
1999-2004.
[5] R.D. Smith, A.L. Johnson, Combust. Flame 51 (1983) 1-22.
[6] D. S. Hsu, C.Y. Lin, M.C. Lin, Proc. Combust. Inst. 20 (1984) 623-630.
[7] R.D. Kern, C.H. Wu, G.B. Skinner, V.S. Rao, J.H. Kiefer, J.A. Towers, L.J.
Mizerka, Proc. Combust. Inst. 20 (1984) 789-797.
[8] J.H. Kiefer, L.J. Mizerka, M.R. Patel, J. Phys. Chem. 89 (1985) 2013-2019.
[9] H.J. Singh, R.D. Kern, Combust. Flame 54 (1983) 49-59.
[10] S. Hayashi, Y. Hisaeda, Y. Asakuma, H. Aoki, T. Miura, H. Yano, Y. Sawa,
Combust. Flame 117 (1999) 851-860.
[11] H. Bohm, H. Jander, D. Tanke, Proc. Combust. Inst. 27 (1998) 1605-1612.
[12] V.G. Knorre, D. Tanke, T.H. Thienel, H.G. Wagner, Proc. Combust. Inst. 26
(1996) 2303-2310.
[13] A. Laskin, A. Lifshitz, Proc. Combust. Inst. 26 (1996) 669-675.
[14] A.G. McLain, C.J. Jachimowski, C.H. Wilson, NASA Tech. Paper, (1979) 1472.
[15] B. Simmons, A. Williams, Combust. Flame 71 (1988) 219-232.
[16] A. Burcat, R.C. Farmer, R.L. Espinoza, R.A. Matula, Combust. Flame 36 (1979)
313-316.
[17] A. Burcat, C. Snyder, T. Brabbs, “Ignition Delay Times of Benzene and Toluene
with Oxygen in Argon Mixtures,” NASA TM 87312, NASA, 1986.
[18] J.L. Emdee, K. Brezinsky, I. Glassman, J. Phys. Chem. 96 (1992) 2151-2161.

103
[19] A. Lovell, K. Brezinsky, I. Glassman, Proc. Combust. Inst. 22 (1988) 1063-1074.
[20] C. Venkat, K. Brezinsky, I. Glassman, Proc. Combust. Inst. 19 (1982) 143-152.
[21] K. Brezinsky, Prog. Ener. Combust. Sci. 12 (1986) 1-24.
[22] R.A. Shandross, J.P. Longwell, J.B. Howard, Proc. Combust. Inst. 26 (1996) 711-
719.
[23] C.B. Vaughn, J.B. Howard, J.P. Longwell, Combust. Flame 87 (1991) 278-288.
[24] A. Tregrossi, A. Ciajolo, R. Barbella, Combust. Flame 117 (1999) 553-561.
[25] J.D. Bittner, J.B. Howard, Proc. Combust. Inst. 18 (1981) 1105-1116.
[26] F. Defoeux, V. Dias, C. Renard, P.J. Van Tiggelen, J. Vandooren, Proc. Combust.
Inst. 30 (2005) 1407-1415.
[27] S. Davis, H. Wang. K. Brezinsky, C.K. Law, Proc. Combust. Inst. 26 (1996)
1025-1033.
[28] S.G. Davis, C.K. Law, Combust. Sci. Technol. 140 (1998) 427-449.
[29] R.J. Johnston, J.T. Farrell, Proc. Combust. Inst. 30 (2005) 217-224.
[30] C.H. Wang, G.J. Ueng, M.S. Tsay, Combust. Flame 113 (1998) 242-248.
[31] M.P. Halstead, L.J. Kirsch, C.P. Quinn, Combust. Flame 30 (1977) 45-60.
[32] J.F. Griffiths, P.A. Halford-Maw, D.J. Rose, Combust. Flame 95 (1993) 291-306.
[33] R. Minetti, M. Ribaucour, M. Carlier, L.R. Sochet, Combust. Sci. Technol. 113-
114 (1996) 179-192.
[34] A. Roubaud, R. Minetti, L.R. Sochet, Combust. Flame 121 (2000) 535-541.
[35] G. Vanhove, M. Ribaucour, R. Minetti, Proc. Combust. Inst. 30 (2005) 1065-
1072.
[36] S. Tanaka, F. Ayala, J.C. Keck, J.B. Heywood, Combust. Flame 132 (2003) 219-
239.
[37] H. Miyama, J. Phys. Chem. 75 (1971) 1501-1504.
[38] M. Frenklach, S. Taki, R.A. Matula, Combust. Flame 49 (1983) 275-282.

104
[39] A. Alexiou, A. Williams, Combust. Flame 104 (1996) 51-65.
[40] R. Sivaramakrishnan, R.S. Tranter, K. Brezinsky, Combust. Flame 139 (2004)
340-350.
[41] R. Sivaramakrishnan, R.S. Tranter, K. Brezinsky, Proc. Combust. Inst. 30 (2005)
1165-1173.
[42] M.A. Oehlschlaeger, D.F. Davidson, R.K. Hanson, Combust. Flame 147 (2006)
195-208.
[43] D.F. Davidson, B.M. Gauthier, R.K. Hanson, Proc. Combust. Inst. 30 (2005)
1175-1182.
[44] R.A. Eng, C. Fittschen, A. Gebert, P. Hibomvschi, H. Hippler, A.N. Unterreiner,
Proc. Combust. Inst. 27 (1998) 211-218.
[45] V. Vasudevan, D.F. Davidson, R.K. Hanson, Proc. Combust. Inst. 30 (2005)
1155-1163.
[46] P. Dagaut, G. Pengloan, A. Restori, Phys. Chem. Chem. Phys. 4 (2002) 1846-
1854.
[47] E.B. Ledesma, M.J. Wornat, P.G. Felton, J.A. Sivo, Proc. Combust. Inst. 30
(2005) 1371-1379.
[48] S.D. Klotz, K. Brezinsky, I. Glassman, Proc. Combust. Inst. 27 (1998) 337-344.
[49] M.P. Halstead, D.B. Pye, C.P. Quinn, Combust. Flame 22 (1974) 89-97.
[50] T. Hirasawa, C.J. Sung, A. Joshi, Z. Yang, H. Wang, C.K. Law, Proc. Combust.
Inst. 29 (2002) 1427-1434.
[51] A. Hamins, K. Seshadri, Combust. Flame 64 (1986) 43-54.
[52] A. Hamins, K. Seshadri, Combust. Flame 68 (1987) 295-307.
[53] R. Seiser, K. Seshadri, E. Piskernik, A. Linan, Combust. Flame 122 (2000) 339-
349.
[54] A.T. Holley, Y. Dong, M.G. Andac, F.N. Egolfopoulos, Combust. Flame 144
(2006) 448-460.
[55] A.T. Holley, Y. Dong, F.N. Egolfopoulos, “Ignition and Extinction Studies of
Mixtures of Air with Single-Component Liquid Hydrocarbons, Gasoline, and
Gasoline Surrogates” Proc. Fall Meeting WSS/CI, Stanford CA, 05F-74 (2005).

105

Chapter 10

An Experimental Study on the Ignition and Extinction of
Premixed Flames of Gasoline and its Surrogates

10.1 Introduction
In chapters 8 and 9 the extinction criteria for potential gasoline surrogate components
were determined. Additionally the current state of numerical mechanisms of several
fuels was assessed, and it was found that currently published mechanisms fail to
completely capture a range of burning characteristics. Despite this, it was shown also
that the mechanisms did reproduce the qualitative behavior of all the fuels, and in some
cases come close in quantitative values as well. The evaluation of single components is
only the first step in surrogate development. Mixtures of single components need to be
tested against real gasoline to be able to determine an appropriate surrogate.
The simplest and oldest surrogate for gasoline is iso-octane, whose oxidation
characteristics have been experimentally determined in numerous reacting configurations.

106
Mixtures of iso-octane

and n-heptane, the two primary reference fuels (PRF), are
improved surrogates over iso-octane, and have also been widely studied in shock tubes
(e.g., [1-3]), rapid compression machines (e.g., [4,5]), jet-stirred reactors (e.g., [6]),
counterflow configurations (e.g., [7,8]), spherical bombs (e.g., [9]), and engines (e.g.,
[10,11]). Two-component surrogates of n-heptane and toluene have been studied also in
engines (e.g., [12]). Mixtures of n-heptane, iso-octane, and toluene appear to be the most
promising surrogates, as their composition could be adjusted to reproduce various
oxidation characteristics of gasoline. These three-component surrogates have been
developed in shock tubes (e.g., [13]), and rapid compression machines (e.g., [14]).
The development of an accurate surrogate would require testing against real gasoline
fuel in both homogeneous systems and flames. The bulk of the work on gasoline
surrogates has been conducted in homogeneous systems, such as shock tubes and rapid
compression machines. Few studies have been conducted on the determination of
fundamental flame properties.
Additionally, no systematic studies exist in which the effect of fuel variability has
been addressed. When considering a real fuel, such as gasoline, the fact that it has an
inconsistent composition must be taken into consideration. Gasoline is only defined by
more general requirements, such as density, octane number, and pollutant concentration,
e.g. sulfur and benzene. The question is how variable are the flame properties of real
samples of gasoline? Since a surrogate is a mixture of single components hydrocarbons
(SCH), it would have consistent flame properties, and the only real requirement would be
in the range of the various samples of gasoline.

107
In view of these considerations, the goals of the present study were to investigate the
flame ignition and extinction characteristics of a wide range of mixtures of air with
gasoline-related fuels in the counterflow configuration, and to assess the effects of the
variability of real gasoline on the burning characteristics. Recent studies have shown
[8,15,16] that both flame ignition and extinction limits are more sensitive to kinetics
compared to laminar flame speeds. Thus, such data will be important for developing
surrogates as well as compiling and optimizing the attendant kinetic mechanisms. The
present experimental investigation included seven (7) SCHs, four (4) samples of gasoline,
four (4) averaged mixtures of gasoline, and two (2) gasoline surrogates. Comparisons of
the experimental data will provide insight into the relative performance of the samples of
gasoline, its surrogates, and SCHs. This will allow for the assessment of current state of
surrogates, and will provide guidance for their further development.

10.2 Experimental approach
10.2.1 Fuels tested
SCHs, samples of gasoline, mixtures of gasoline, and gasoline surrogates were
considered. The SCHs tested were n-pentane (n-C
5
H
12
), n-hexane (n-C
6
H
14
), n-heptane
(n-C
7
H
16
), n-octane (n-C
8
H
18
), iso-octane (iso-C
8
H
18
), n-nonane (n-C
9
H
20
), and n-decane
(n-C
10
H
22
). The straight chain alkanes were chosen to span the C
#
range that is typically
found in conventional gasoline, and iso-octane due to its use as a PRF.
The samples of gasoline and averaged gasoline mixtures were obtained from local
vendors representing Arco, Shell, Chevron, and Mobil companies, all with an octane

108
number of 87. To assess the effect of the variability of gasoline, one sample from each
company was obtained and tested. Single samples can be subject to contamination, and
not necessarily represent the “average” fuel of a specific company. To moderate the
effect of variability of individual samples, four samples from four different gasoline
stations of the same company were obtained and mixed to create averaged samples of
gasoline. Gasoline surrogates need to be tested and compared to an averaged gasoline,
not any one sample.
Two published gasoline surrogates were also tested. Surrogate one (S1), with 63%
iso-octane, 20% n-heptane, and 17% toluene [13], and surrogate 2 (S2) with 55% iso-
octane, 35% n-heptane, and 10% toluene [17]; all percentages are per liquid volume.
Both S1 and S2 surrogates accurately reproduce the ignition delay characteristics of
gasoline in shock tubes.

10.2.2 Boundary Conditions
The lighter fuels tested were ranging from n-pentane to iso-octane, can be maintained
in the gaseous form at flammable concentrations in air at room temperature (e.g., [7,8]).
However, this is not the case for all fuels tested, including n-dodecane, the samples of
gasoline, and the gasoline surrogates. A slight increase in the fuel/air mixture
temperature does allow them to remain in the gaseous state, as the fuel vapor pressure
notably increases. Thus, for all cases, the fuel/air mixtures were maintained slightly
above 50
o
C to avoid condensation.

109
10.3 Results and discussion
The ranges of T
ign
and fuel/air mass ratio considered were approximately 875-1075
o
C
and 0.04-0.049 respectively. The ignition experiments were conducted with D = 14 mm
and L = 12 mm. The global strain rate, K
glb
, was kept constant for all ignition
experiments at K
glb
= 154 s
-1
; K
glb
≡ U
exit
/(L/2), where U
exit
is the mixture velocity at the
burner exit.
The ranges of global extinction strain rates, K
glb,ext
, and fuel/air mass ratio considered
were approximately 50-240 s
-1
and 0.046-0.08 respectively. Experiments were conducted
with D = 14 mm and 22 mm and L = 14 mm and 18 mm respectively; D = 14 mm was
used for the four highest K
glb,ext
reported in this investigation, while D = 22 mm was used
for the three lower K
glb,ext
.
Table 10.1 – Variation of T
ign
with fuel/air mass
ratio for all fuels. The uncertainty in fuel/air
mass ratio is 2.5% and in T
ign
is ± 20
o
C.
Temperature,
o
C 879 980 1076
Fuel Fuel/air mass ratio
n-C
5
H
12
0.0426 0.0414 0.0408
n-C
6
H
14
0.0429 0.0417 0.0410
n-C
7
H
16
0.0433 0.0420 0.0413
n-C
8
H
18
0.0434 0.0421 0.0414
iso-C
8
H
18
0.0476 0.0456 0.0449
n-C
9
H
20
0.0434 0.0421 0.0414
n-C
10
H
22
0.0451 0.0434 0.0427
Arco 0.0451 0.0431 0.0424
Arco Averaged 0.0488 0.0467 0.0457
Chevron 0.0477 0.0455 0.0448
Chevron Averaged 0.0485 0.0465 0.0454
Mobil 0.0479 0.0458 0.0451
Mobil Averaged 0.0479 0.0458 0.0451
Shell 0.0468 0.0451 0.0441
Shell Averaged 0.0479 0.0458 0.0451
S1 0.0466 0.0449 0.0438
S2 0.0472 0.0454 0.0443

110
The experimental results were presented in tables in their entirety to assure clarity, so
that they can be used for modeling purposes. Selected data were presented in figures to
illustrate graphically certain physical trends. Table 10.1 contains the ignition data. It
depicts T
ign
as function of fuel/air mass fraction for all fuels.

Figure 10.1 Variation of T
ign
with the fuel/air mass fraction for the single component
hydrocarbons.

As expected, T
ign
decreases as fuel/air mass fraction increases. Comparing the SCHs
in Fig. 10.1, the differences between the ignition characteristics for most of the n-alkanes
are within the experimental uncertainty. Under the conditions of the present
investigation, it appears that the process of ignition is moderately sensitive to fuel
825
875
925
975
1025
1075
0.040 0.041 0.042 0.043 0.044 0.045 0.046 0.047 0.048
Ignition Temperature, T
ign
, °C
Fuel/air Mass Ratio
n-C
7
H
16
n-C
5
H
12
n-C
6
H
14
n-C
8
H
18
n-C
9
H
20
n-C
10
H
22
iso-C
8
H
18

111
diffusion, as there is a small but finite variation of T
ign
with C
#.
Comparing the ignition
response of flames of the two C
8
species, n-octane and iso-octane, it can be seen that the
n-alkane flames are notably easier to ignite than iso-alkane flames, having a 10% lower
fuel/air mass fraction at the ignition state for the same T
ign
. This illustrates the effect of
the branched nature on the ignition propensity of the fuels. Additionally the ignition
response of both n-heptane and n-octane was found to be nearly identical.
If “octane” number is a measure of ignition that is valid for premixed flames, all
samples of gasoline should have ignition characteristics that are between those of n-
heptane and iso-octane. However, the results shown in Table 10.1 reveal that only a few
of the gasoline samples exhibit ignition behavior that is between the two PRFs, with the
rest having lower ignition propensity compared to iso-C
8
H
18
.
Figure 10.2 depicts the ignition characteristics of the individual samples of gasoline.
The different samples exhibit a significant variation in T
ign
. A difference of 6% in
fuel/air mass fraction can be seen between the two extreme samples. The Mobil sample
flames were the hardest to ignite, followed by Chevron, Shell, and Arco. It was found
that the densities of the four samples are different, and that their resistance to ignition
appears to scale with the fuel density. Chevrons’s sample’s density is the highest one
(0.743 gm/cm
3
), while Arco’s is the lowest one (0.703 gm/cm
3
). Given those density
differences, the liquid volumetric flow rates of the fuels at the state of ignition for all four
gasoline samples are much closer compared to the ones obtained based on mass flow rate
and reported in Table 10.1. Injection systems in cars operate on liquid volumetric flow

112
rate, not mass flow rate, therefore the variation in the ignition propensity would be less
notable during engine operation.

Figure 10.2 Variation of T
ign
with the fuel/air mass fraction for individual samples of
gasoline.

To further assess the effect of fuel variability that results in a wide range of T
ign
,
averaged samples obtained from the same companies were tested. The ignition results
obtained for those averaged samples appear to exhibit a much smaller variation as it can
be seen in Fig. 10.3. The difference between the two extreme averaged samples is only
3% in fuel/air mass fraction, which is half compared to what was found for the individual
825
875
925
975
1025
1075
0.042 0.043 0.044 0.045 0.046 0.047 0.048
Ignition Temperature, T
ign
, °C
Fuel/air Mass Ratio
Chevron
Arco
Mobil
Shell

113
samples. Flames of the Mobil averaged sample have the highest ignition propensity,
followed by Shell, Chevron, and Arco. The densities were also found to have a smaller
variation, with the averaged Arco having the highest density (0.739 gm/cm
3
) and the
averaged Chevron the lowest one (0.724 gm/cm
3
). The averaged samples also exhibit
rather different ignition characteristics compared to the corresponding individual
samples.

Figure 10.3 Variation of T
ign
with the fuel/air mass fraction for averaged samples of
gasoline and the gasoline surrogates.

825
875
925
975
1025
1075
0.0435 0.0445 0.0455 0.0465 0.0475 0.0485
Ignition Temperature, T
ign
, °C
Fuel/air Mass Ratio
S1
S2
Shell
(avg)
Chevron
(avg)
Arco
(avg)
Mobil
(avg)

114
Figure 10.3 also depicts the T
ign
for the two surrogates S1 and S2. Both surrogates
ignite more readily than the averaged samples of gasoline but the largest variation
between S1 and any averaged sample of gasoline is less than 5%. Flames of S1 ignite at
lower fuel/air mass fractions than those of S2, which is opposite to what is expected
based on their chemical composition, namely the higher concentration of aromatic
compounds. There is also a large density difference between the two surrogates, with the
density of S1 (0.726 gm/cm
3
) and S2 (0.752 gm/cm
3
). Considering this density variation
it was shown that the two surrogates have a very similar ignition performance based on a
liquid volumetric fuel flow rate basis, similar to what was found with the samples of
gasoline.
Table 10.2 contains the extinction data. It depicts K
glb,ext
as a function of fuel/air
mass fraction for all fuels. As expected, the fuel/air mass fraction at extinction increases
as the strain rate increases, for the same fuel. Notable differences in extinction behavior
can be seen between the different fuels, which are more profound for the higher K
glb,ext
.
The K
glb,ext
of SCHs are shown in Fig. 10.4. For the SCHs, there is no difference in their
resistance to extinction for the C
#
range that was tested, as K
glb,ext
for flames of the
straight-chain alkanes are within experimental uncertainty of each other. Comparison of
K
glb,ext
of iso-C
8
H
18
and n-C
8
H
18
flames reveals that iso-C
8
H
18
flames are far less resistant
to extinction compared to n-C
8
H
18
and all other straight-chain alkanes. The branched
nature of iso-C
8
H
18
causes the resulting flames to be less resistant to extinction than n-
alkanes.

115
Table 10.2 – Variation of K
glb,ext
with fuel/air mass ratio for all fuels. The uncertainty in fuel/air mass
ratio is 2.5% and in K
glb,ext
is 3.5%.
K
glb,ext
, s
-1
47 59 71 119 159 199 239
Fuel Fuel/air mass ratio
n-C
5
H
12
0.0469 0.0508 0.0523 0.0592 0.0651 0.0687 0.0745
n-C
6
H
14
0.0483 0.0508 0.0533 0.0608 0.0664 0.0692 0.0730
n-C
7
H
16
0.0492 0.0518 0.0540 0.0600 0.0655 0.0690 0.0740
n-C
8
H
18
0.0462 0.0500 0.0522 0.0594 0.0644 0.0683 0.0724
iso-C
8
H
18
0.0507 0.0543 0.0567 0.0623 0.0691 0.0750 0.0789
n-C
9
H
20
0.0467 0.0503 0.0526 0.0608 0.0653 0.0685 0.0714
n-C
10
H
22
0.0474 0.0510 0.0530 0.0601 0.0650 0.0689 0.0740
Arco 0.0492 0.0514 0.0537 0.0610 0.0667 0.0715 0.0755
Arco Averaged 0.0517 0.0541 0.0565 0.0649 0.0701 0.0757 0.0794
Chevron 0.0532 0.0553 0.0576 0.0653 0.0705 0.0761 0.0802
Chevron Averaged 0.0524 0.0544 0.0565 0.0636 0.0693 0.0746 0.0790
Mobil 0.0508 0.0533 0.0558 0.0630 0.0679 0.0729 0.0774
Mobil Averaged 0.0514 0.0538 0.0562 0.0638 0.0697 0.0753 0.0790
Shell 0.0503 0.0531 0.0550 0.0632 0.0688 0.0732 0.0773
Shell Averaged 0.0513 0.0535 0.0558 0.0637 0.0694 0.0742 0.0787
S1 0.0532 0.0550 0.0571 0.0646 0.0701 0.0758 0.0800
S2 0.0526 0.0545 0.0562 0.0636 0.0695 0.0750 0.0795

Similarly to ignition, there is a large difference between the extinction characteristics
of the two C
8
alkanes. Such difference between n-alkanes and iso-alkanes is expected. H
abstraction from n-alkanes results in primary and secondary alkyl radicals while H
abstraction from iso-alkanes form tert-alkyl radicals. The tert-alkyl radicals are more
stable as compared to primary and secondary alkyl radicals. The resulting peroxyl radical
(RO
2
·) formed from a tert-alkyl radical is unstable, resulting in dissociation back to the
original products. For primary and secondary alkyl radicals the resulting RO
2
· is more
stable, and subsequently undergoes rapid β-scission. This difference in the kinetics of n-
alkanes and iso-alkanes results in an overall weaker burning characteristics of iso-octane
as compared to n-octane, both in terms of ignition and extinction.

116

Figure 10.4 Variation of K
ext
with the fuel/air mass fraction for the SCHs.

Figure 10.5 depicts the relative extinction response of the individual samples of
gasoline. A large difference in fuel/air mass fractions at extinction was found for the
different samples, with a variation ranging from 6% to 8%. The most resistant sample to
extinction was Arco, followed by Mobil and Shell, which are approximately the same,
and Chevron that was the least resistant. This scaling is similar to the ignition
propensities of the samples, with Chevron being near to the least reactive and Arco being
the most reactive on mass basis.
0
50
100
150
200
250
0.045 0.050 0.055 0.060 0.065 0.070 0.075 0.080
Extinction Strain Rate, K
glb,ext
, 1/s
Fuel/air Mass Ratio
C5
C6
C7
C8
C9
C10
IsoC8
n-C
5
H
12
n-C
6
H
14
n-C
7
H
16
n-C
8
H
18
n-C
9
H
20
n-C
10
H
22
iso-C
8
H
18

117

Figure 10.5 Variation of K
ext
with the fuel/air mass fraction for the individual
samples of gasoline.

The extinction limits of averaged mixtures of gasoline and the two surrogates are
shown in Fig. 10.6. The averaged samples of gasoline exhibit very similar K
glb,ext

compared to each other, with the difference between the two extreme samples being 1%
at high and 2% at low strain rates. This is in agreement with what was earlier reported on
ignition. Additionally, it can be seen that the averaged samples of gasoline have very
0
50
100
150
200
250
0.045 0.055 0.065 0.075 0.085
Extinction Strain Rate, K
glb,ext
, 1/s
Fuel/air Mass Ratio
Arco
Mobil
Shell
Chevron

118
similar K
glb,ext
compared to iso-C
8
H
18
. The two surrogates exhibit very similar resistance
to extinction as the averaged samples of gasoline. The K
glb,ext
for the two surrogates were
close to the averaged samples of gasoline, with S1 being slightly less resistant to
extinction than S2. This is to be expected, as S1 has greater aromatic concentration and a
larger content of iso-C
8
H
18
compared to S2.

Figure 10.6 Variation of K
ext
with the fuel/air mass fraction for the averaged samples
of gasoline and the gasoline surrogates.

0
50
100
150
200
250
0.050 0.055 0.060 0.065 0.070 0.075 0.080 0.085
Extinction Strain Rate, K
glb,ext
, 1/s
Fuel/air Mass Ratio
Arco (avg)
Mobil (avg)
Shell (avg)
Chevron (avg)
S1
S2

119
10.4 Conclusions
Ignition temperatures and extinction strain rates were experimentally determined in
the counterflow configuration for mixtures of air with a wide range of gasoline-related
liquid fuels. Single component hydrocarbons, samples of gasoline, averaged mixtures of
gasoline, and gasoline surrogates were tested and the results were compared to assess
their relative performance.
The ignition limits of flames of straight chain alkanes between C
5
and C
9
were nearly
indistinguishable. As expected, iso-octane flames were found to be the hardest to ignite
among all single component hydrocarbons, demonstrating the substantial effect of the
branched nature on the ignition response. Vastly different ignition response was
determined for flames resulting from the various individual samples of gasoline, even
though all fuels are designated with an “octane” number of 87. However, averaged
samples of gasoline exhibit rather similar ignition response. The ignition propensity of
the surrogate flames was found to be slightly higher than that of the averaged samples of
gasoline.
The resistance to extinction for all n-alkane flames tested was determined to be very
similar, being within experimental uncertainty. Flames of n-octane were found to be
more resistant to extinction than iso-octane, revealing that the branched nature of the
molecule has a significant effect on the extinction characteristics, as it has on ignition.
The individual samples of gasoline tested exhibit a notable range of extinction conditions,
all again with the same “octane” number 87. Conversely, the averaged samples of
gasoline have virtually identical extinction characteristics. This demonstrates that any

120
given sample may deviate from the anticipated behavior of gasoline, but an averaged
sample of gasoline has very repeatable extinction conditions. The extinction
characteristics of flames of the two gasoline surrogates were found to be similar to each
other as well as to the averaged samples of gasoline.
The observed variation in ignition and extinction behavior between individual
samples of gasoline indicates that results obtained for any single sample of gasoline may
not represent accurately those of the “typical” gasoline fuel. Instead, reference samples
of gasoline should be used, such as a mixture of a large number of samples or a single
chosen sample of “typical” gasoline. This is essential in developing gasoline surrogates.

121
10.5 Chapter 10 References
[1] K. Fieweger, R. Blumenthal, G. Adomeit, Combust. Flame 109 (1997) 599–619.
[2] D.J. Vermeer, J.W. Meyer, A.K. Oppenheim, Combust. Flame 18 (1972) 327–
336.
[3] K. Fieweger, R. Blumenthal, G. Adomeit, Proc. Combust. Inst. 25: 1579–1585
(1994).
[4] S. Tanaka, F. Ayala, J.C. Keck, J.B. Heywood, Combust. Flame 132 (2003) 219-
239.
[5] J.F. Griffiths, P.A. Halford-Maw, C. Mohamed, Combust. Flame 111 (1997) 327-
337.
[6] A. Ciajolo, A, D'anna, Combust. Flame 112 (1998) 617-622.
[7] S.G. Davis, C.K. Law, H. Wang, Proc. Combust. Inst. 27 (1998) 305-312.
[8] A.T. Holley, Y. Dong, Y. Fan, M.G. Andac, F.N. Egolfopoulos, Combust. Flame
144 (2006) 448-460.
[9] D. Bradley, R.A. Hicks, M. Lawes, C.G.W. Sheppard, R. Woolley, Combust.
Flame 115 (1998) 26–144.
[10] X. Lu, W. Chen, Z. Huang, Fuel 84 (2005) 1074–1083.
[11] X. Lu, W. Chen, Z. Huang, Fuel 84 (2005) 1084–1092.
[12] J. Andrae, D. Johansson, P. Björnbom, P. Risberg, G. Kalghatgi, Combust. Flame
140 (2005) 267–286.
[13] B.M. Gauthier, D.F. Davidson, R.K. Hanson, Combust. Flame 139 (2004) 300–
311.
[14] M.P. Halstead, L.J. Kirsch, C.P. Quinn, Combust. Flame 30 (1997) 45-60.
[15] F.N. Egolfopoulos, P.D. Dimotakis, Combust. Sci. Technol. 162 (2001) 19-36.
[16] Y. Dong, A.T. Holley, M.G. Andac, F.N. Egolfopoulos, S.G. Davis, P. Middha,
H. Wang, Combust. Flame 142 (2005) 374-387.
[17] M. Bergman, V.I. Golovitchev, "Chemical Mechanisms for Modeling HCCI with
Gasoline and Diesel Oil Surrogates", Proceedings of the VII. Congress on Engine
Combustion Processes, pp. 323-334, Munchen, 15-16 March (2005).

122

Chapter 11

Ignition and Extinction of Non-Premixed Flames of Single
Component Liquid Hydrocarbons, Jet Fuels, and their
Surrogates

11.1 Introduction
The previous chapters focused on premixed flames. Premixed flames are of practical
importance to various engine types, such as spark-ignition engines, and are valuable
fundamental data towards the development of chemical kinetics models. Additionally the
previous chapters have focused predominately on the smallest of practical fuels, which
are found in gasoline, with the hydrocarbons varying in C
#
from C
5
to C
10
. With
advancements in computing power and combustion science, experimentalists and
modelers have begun to extend their range to fuels with higher C
#
, which are of
importance for aircraft turbines which run on jet fuels (JF).
JFs are mixtures of large numbers of hydrocarbons spanning a wide range of C
#
and
chemical classifications. Typically, the C
#
ranges from C
7
-C
16
(on the average C
11
-C
12
),

123
and the chemical classifications from n-paraffins to aromatics. The defining
characteristics of JFs are not necessarily their chemical compositions, but their physical
properties such as density and boiling range. It is also known that the chemical
composition of different batches of JFs can vary significantly from each other (e.g., [1]).
The large number of complex hydrocarbons present in real JFs renders the modeling
of their pyrolysis/oxidation characteristics a rather daunting task. Development of
reliable surrogate fuels is the only option to develop kinetics models for simulating real
combustors.
Surrogates have been developed for practical fuels based on chemical composition by
matching properties, such as volatility, chemical class distribution, average molecular
weight, H/C ratio, sooting tendency, heat release, flammability, and regression rate (e.g.
[2-6]). This matching, however, is based mostly on physical properties and much less on
the chemical kinetic and transport properties of the real fuels. Ideally, matching the
properties of JFs and surrogates should include kinetically controlled combustion
properties derived in both homogenous systems and flames. Experimental flame data
have the dual benefit of testing the chemical kinetics and transport properties for a very
wide range of temperatures and concentrations. Few past studies have considered flame
phenomena (e.g., [7]).
In the present investigation, the ignition and extinction limits of non-premixed flames
were determined experimentally for a wide range of liquid fuels in the well-controlled
counterflow configuration. It is known that both flame ignition and extinction are rather

124
sensitive to chemical kinetics (e.g., [8-10]). Thus, such data could be used to develop
appropriate surrogates as well as reliable kinetics mechanisms.

11.2 Experimental Approach
11.2.1 Fuels Tested
Table 11.1 – Test fuels with detailed composition data (ASTM D2425)
3327 4734 4572 4765 3773 4658 3602 3638 World
survey
average
JP-7 F-T RP-1 DCL JP-8 Jet A
blend
Jet A Jet A Jet-A
JP-8
JP-5
TS-1
Paraffins (n- + iso-) 67.9 99.7 57.6 0.6 57.2 55.2 49.4 64.5 58.8
Cycloparaffins 21.2 <0.2 24.8 46.4 17.4 17.2 15.8 13.2 10.9
Dicycloparaffins 9.4 0.3 12.4 47.0 6.1 7.8 10.8 7.1 9.3
Tricycloparaffins 0.6 <0.2 1.9 4.6 0.6 0.6 0.9 0.6 1.1
Alkylbenzenes 0.7 <0.2 2.1 0.3 13.5 12.7 14.0 10.8 13.4
Indans/Tetralins <0.2 <0.2 0.3 1.1 3.4 4.9 7.9 2.1 4.9
Indenes <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 <0.2
Naphthalene <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 0.4 0.13
Naphthalenes <0.2 <0.2 0.3 <0.2 1.7 1.3 1.2 1.3 1.55
Acenaphthenes <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 <0.2
Acenaphthylenes <0.2 <0.2 0.4 <0.2 <0.2 <0.2 <0.2 <0.2 <0.2
Tricyclic Aromatics <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 <0.2 <0.2

A wide range of liquid fuels was tested. Nine single-component hydrocarbons
(SCH), namely n-pentane (n-C
5
H
12
), n-hexane (n-C
6
H
14
), n-heptane (n-C
7
H
16
), n-octane
(n-C
8
H
18
), iso-octane (iso-C
8
H
18
), n-nonane (n-C
9
H
20
), n-decane (n-C
10
H
22
), n-dodecane
(n-C
12
H
26
), n-tetradecane (n-C
14
H
30
), and JP10 (exo-tetrahydrodicyclopentadiene C
10
H
16
)
were included. Ten practical JFs shown in Table 11.1, and two published JP-8
surrogates, namely a twelve-component (S12) [4] and a six-component (S6) [6] were also
tested; details of the surrogates composition are shown in Table 11.2. The practical

125
(petroleum distillate) fuels include JP-8 (standard military JF, MIL-DTL-831333E), three
Jet-A fuels (ASTM D1655, similar to JP-8 but for commercial use in U.S.), JP-7 fuel
(specialty JF, low volatility/highly processed), two RP-1 fuels (kerosene rocket
propellant, MIL-DTL-25576D), JP-10, a Fischer-Tropsch-derived JF, and a coal-derived
JF. The distillate fuels were supplied by Wright-Patterson Air Force Base and are
identified by a four-digit sample number and fuel type in the data that follows.
Table 11.2 – Surrogate fuel composition
Compound % Volume
S6
m-Xylene 15
iso-Octane 10
Methylcyclohexane 20
n-Dodecane 30
n-Tetradecane 20
Tetralin 5
Compound % Mass
S12
m-Xylene 5
iso-Octane 5
Methylcyclohexane 5
n-Dodecane 20
n-Tetradecane 15
Tetralin 5
cyclo-Octane 5
n-Decane 15
Butylbenzene 5
Tetramethylbenzene 5
Methylnapthalene 5
n-Hexadecane 10

JP-8-3773 is a typical batch of JP-8 fuel. Jet-A-3602 and -3638 were selected to have
levels of aromatics at either end of the distribution described above with 12% and 24%
respectively. Jet-A-4658 is an “average” Jet-A fuel composed by mixing in equal
proportions five samples from different U.S. manufacturers. JP-7 and RP-1 are fuels with
low aromatic content and more well defined specifications than other JFs, so it was
anticipated that these fuels would be consistent from batch-to-batch. As can be seen from
Table 11.1, the JP-8 and the “composite blend” Jet-A fuels have similar composition to

126
the average JF composition revealed in a recent world survey – roughly 60% n-+i-
paraffins, 20% cyclo-paraffins (naphthenes) and 20% aromatics. Separate analyses have
determined that typical JFs average approximately 20% n-paraffins. Two synthetic fuels
were tested, namely the Fischer-Tropsch jet fuel (4734) and one based on direct coal-
liquefaction (4765), and they are primarily composed of iso-paraffins and naphthenes,
respectively. Thus, the present experimental investigation includes the widest variation
in fuel composition of any similar flame study to date.

11.2.2 Boundary Conditions
The burners were not heated, resulting thus in temperatures at the burner exit in the
range of 110-130
o
C, which is the boundary condition for the heated fuel-jet. Under such
conditions, low fuel/N
2
mass fractions were considered to avoid condensation. For these
low fuel/N
2
mass fractions, flames could only be sustained in the presence of O
2
-enriched
oxidizer. Similarly, Cooke et al. [7] reported that for JP-8/N
2
mass fractions of about
0.02, the minimum O
2
mole fraction in the oxidizer stream was 60% to sustain flames. In
the present investigation pure O
2
was used in both the ignition and extinction studies.
For the extinction experiments the oxidizer stream was kept constant for all of the
fuels, resulting in a constant K
ext
between the fuels since the strain rate was measured on
the oxidizer side. The ranges of K
ext
and fuel/N
2
mass ratio considered were 40-360 s
-1

and 0.05-0.1, respectively.
For the ignition experiments the peak temperature of the H
2
/CO/O
2
flame was
measured with a thermocouple, and is reported as the ignition temperature, T
ign
. The
measured temperature was corrected for radiative heat losses. The ranges of T
ign
and

127
fuel/N
2
mass ratio considered in this study were approximately 1000-1100
o
C and 0.045-
0.065, respectively.

11.3 Results and discussion
Experimental results obtained for all fuels are presented in tables for completeness
and clarity. Graphical presentation will be used only for selected conditions.
Table 11.3 – K
ext
vs Fuel/N
2
mass ratio for all fuels. The uncertainty in fuel/N
2
mass ratio is 2.5% and
in K
ext
is 3.5%.
K
ext
, s
-1
60 78 97 155 223 290 357
Fuel Fuel/N
2
mass ratio
n-C
5
H
12
0.0571 0.0597 0.0613 0.0678 0.0734 0.0783 0.0829
n-C
6
H
14
0.0572 0.0599 0.0627 0.0698 0.0760 0.0811 0.0860
n-C
7
H
16
0.0572 0.0601 0.0625 0.0700 0.0770 0.0822 0.0870
n-C
8
H
18
0.0580 0.0609 0.0635 0.0701 0.0769 0.0837 0.0889
iso-C
8
H
18
0.0601 0.0625 0.0646 0.0707 0.0769 0.0838 0.0897
n-C
9
H
20
0.0575 0.0610 0.0635 0.0708 0.0779 0.0847 0.0908
n-C
10
H
22
0.0574 0.0609 0.0633 0.0704 0.0784 0.0851 0.0909
n-C
12
H
26
0.0625 0.0662 0.0696 0.0737 0.0815 0.0872 0.0945
n-C
14
H
30
0.0629 0.0661 0.0705 0.0750 0.0819 0.0877 0.0949
JP-7 0.0674 0.0706 0.0755 0.0817 0.0889 0.0925 0.0955
JP-8 0.0687 0.0712 0.0739 0.0802 0.0881 0.0916 0.0967
JP-10 0.0673 0.0704 0.0745 0.0820 0.0900 0.0943 0.0984
Jet-A-3602 0.0648 0.0686 0.0735 0.0791 0.0860 0.0932 0.0999
Jet-A-3638 0.0644 0.0672 0.0722 0.0788 0.0846 0.0916 0.0984
Jet-A-4658 0.0647 0.0680 0.0731 0.0797 0.0855 0.0921 0.0986
RP-1-3642 0.0654 0.0691 0.0732 0.0791 0.0853 0.0913 0.0976
RP-1-4572 0.0643 0.0682 0.0723 0.0777 0.0840 0.0902 0.0953
DCL-4765 0.0651 0.0680 0.0743 0.0787 0.0842 0.0904 0.0968
FT-4734 0.0628 0.0656 0.0701 0.0758 0.0820 0.0868 0.0921
S6 0.0620 0.0650 0.0684 0.0747 0.0817 0.0858 0.0886
S12 0.0631 0.0660 0.0685 0.0755 0.0833 0.0870 0.0897

Table 11.3 depicts K
ext
as a function of fuel/N
2
mass ratio for all fuels; for the
K
ext
= 60, 78, and 97 s
-1
cases, D = 22 mm was used, while for the K
ext
= 155, 223, 290,
and 357 s
-1
cases, D = 14 mm was used. Comparison of the SCHs is shown in Fig. 11.1,
it reveals that for the same K
ext
SCHs with smaller C
#
extinguish at lower fuel/N
2
mass

128
ratio, i.e., they have greater resistance to extinction. It can be seen that n-pentane is the
most resistant while n-tetradecane is the least resistant one. There is, however, less
separation between the various fuels extinction response at low strain rates than at higher
ones. Non-premixed burning is diffusion-controlled and given that the lighter molecules
are more diffusive, they result in more intense burning. The effect of strain rate can also
be explained based on a rather similar argument. At high strain rates, the fuel
concentration gradients just before the reaction zone are steeper compared to low strain
rates. Thus, the importance of mass diffusion is more apparent and there are notable
differences as the fuel molecular weight increases. At low strain rates these differences
become less distinct. Comparing n-octane and iso-octane flames, the former are more
resistant to extinction. This finding illustrates the importance of the branched nature of
the carbon chain on the flame response. This difference between the two C
8
species is
greater at low strain rates, where the kinetics plays a larger, while the diffusion plays a
larger role at high strain rates.

129

Figure 11.1 Variation of K
ext
with the fuel/N
2
mass fraction for the SCHs.

Comparing JFs to the SCHs, JF-flames are less resistant to extinction compared to
SCHs. The flame extinction characteristics of n-dodecane and n-tetradecane are close to
the JFs suggesting that they may be good candidates in surrogate fuel development. For
the fuel/N
2
mass ratio range considered, the flame extinction characteristics of the
different JFs are in general similar. While there were some systematic differences
between the JFs, no fuel was clearly identified as the least resistant to extinction.
0
50
100
150
200
250
300
350
400
0.050 0.055 0.060 0.065 0.070 0.075 0.080 0.085 0.090 0.095
Extinction Strain Rate, K
ext
, 1/s
Fuel/N
2
Mass Fraction
n-C5
n-C6
n-C7
n-C8
n-C9
n-C10
n-C12
n-C14
iso-C8
n-C
5
H
12
n-C
6
H
14
n-C
7
H
16
n-C
8
H
18
n-C
9
H
20
n-C
10
H
22
n-C
12
H
26
n-C
14
H
30
iso-C
8
H
18

130

Figure 11.2 Variation of K
ext
with the fuel/N
2
mass fraction for the 3 batches of Jet-
A.

Figure 11.2 depicts K
ext
for flames resulting from the three different batches of Jet-A.
Similar extinction behaviors were observed, with the higher aromatic-content Jet-A-3602
flames being slightly less resistant to extinction than the low aromatic-content Jet-A-
3638. The flames of the average Jet-A-4658 exhibit an extinction behavior between the
flames of the other two batches of Jet-A. This suggests that the greater the aromatic
content of the fuel is the lower is the resistance to extinction.
0
50
100
150
200
250
300
350
400
0.060 0.065 0.070 0.075 0.080 0.085 0.090 0.095 0.100
Extinction Strain Rate, K
ext
, 1/s
Fuel/N
2
Mass Fraction
JetA-3638
JetA-3602
JetA-4658

131
Figure 11.3 depicts K
ext
for the two synthetic fuels. FT-4734 flames are far more
resistant to extinction than the DCL-4765 flames. Comparing the purely n-+iso-paraffin
fuel to the purely cyclo-paraffin fuel reveals that the saturated alkanes have greater
resistance to extinction than the ring structured alkanes.

Figure 11.3 Variation of K
ext
with the fuel/N
2
mass fraction for the 2 synthetic fuels.

The results of Fig. 11.4 reveal that flames of the two JP-8 surrogates (S6 and S12) are
more resistant to extinction compared to JP-8 and Jet-A-4658. Though those surrogates
were compiled by matching the chemical class distribution of JP-8, they both contain a
large number of lighter compounds compared to JP-8. The high diffusivities of those
0
50
100
150
200
250
300
350
400
0.060 0.065 0.070 0.075 0.080 0.085 0.090 0.095 0.100
Extinction Strain Rate, K
ext
, 1/s
Fuel/N
2
Mass Fraction
FT-4734
DCL-4765

132
smaller molecules cause the flames of both surrogates to be more resistant to extinction
compared to JP-8 and Jet-A-4658. For S6, 50% of the mixture by volume has C
#
of 10 or
smaller, while only 20% has C
#
of 14; the approximate average C
#
of JP-8 is 12.
Comparing S6 and S12 flames, it can be seen that the extinction behavior is very similar.
Thus, the inclusion of a large amount of small hydrocarbons in a surrogate will result in a
fuel that may not mimic satisfactorily the flame behavior of the real fuel.

Figure 11.4 Variation of K
ext
with the fuel/N
2
mass fraction for the 2 surrogates, JP-8,
and the averaged Jet-A-4658.

0
50
100
150
200
250
300
350
400
0.060 0.065 0.070 0.075 0.080 0.085 0.090 0.095 0.100
Extinction Strain Rate, K
ext
, 1/s
Fuel/N
2
Mass Fraction
S12
JP-8
S6
JetA-4658

133
The difference in extinction characteristics cannot be attributed only to the chemical
composition. Since JP-7 contains nearly no aromatics and JP-8 about 20% aromatics, JP-
8 would be expected to be less resistant to extinction, which however is not the case. The
average molecular weight of the two fuels is also different. With the average molecular
weights of JP-7 and JP-8 being 170 and 152 respectively, the increased average
diffusivity of JP-8 counters the effect of chemical composition.
Table 11.4 – T
ign
vs Fuel/N
2
mass ratio for all
fuels. The uncertainty in fuel/N
2
mass ratio is
2.5% and in T
ign
is ± 20
o
C.
Temperature,
o
C 998 1050 1101
Fuel Fuel/N
2
mass ratio
n-C
5
H
12
0.0498 0.0482 0.0466
n-C
6
H
14
0.0501 0.0485 0.0468
n-C
7
H
16
0.0503 0.0486 0.0468
n-C
8
H
18
0.0511 0.0486 0.0479
iso-C
8
H
18
0.0556 0.0538 0.0518
n-C
9
H
20
0.0517 0.0499 0.0481
n-C
10
H
22
0.0537 0.0518 0.0500
n-C
12
H
26
0.0564 0.0545 0.0520
n-C
14
H
30
0.0580 0.0561 0.0535
JP-7 0.0582 0.0555 0.0528
JP-8 0.0607 0.0579 0.0551
JP-10 0.0556 0.0540 0.0517
Jet-A-3602 0.0603 0.0575 0.0548
Jet-A-3638 0.0580 0.0546 0.0526
Jet-A-4658 0.0600 0.0572 0.0545
RP-1-3642 0.0575 0.0555 0.0527
RP-1-4572 0.0585 0.0562 0.0538
DCL-4765 0.0567 0.0548 0.0530
FT-4734 0.0583 0.0557 0.0525
S6 0.0574 0.0547 0.0521
S12 0.0578 0.0552 0.0525

The T
ign
for all fuels are shown in Table 11.4 as a function of the fuel/N
2
mass ratio.
The data were obtained with D = 14 mm at a constant global strain rate of 194 s
-1
,
determined by dividing the burner exit velocity by L/2; strain rate effects will be assessed
in future investigations. For the 998
o
C flame the equivalence ratio, φ, was 0.0782 with a

134
H
2
/CO molar ratio of 3.548, for the 1050
o
C flame φ = 0.0825 with a H
2
/CO molar ratio
of 2.343, and for the 1101
o
C flame φ = 0.0859 with a H
2
/CO molar ratio of 1.609.
When considering uncertainty many fuels have very similar ignition characteristics,
but as all fuels were tested against the same ignition source their relative performance can
be assessed. Among the SCHs, the lighter the fuel is the easier it is for a flame to ignite,
as shown in Fig. 11.5. This is, again, reasonable given that the lighter molecules are
more diffusive and their transport into the ignition kernel (residing on the oxidizer side) is
facilitated. There is only a small difference between the ignition conditions of low C
#
fuels C
5
-C
7
but there is a notable difference between the larger fuels such as C
9
-C
14
.
Comparing fuels of different chemical class reveals that the iso-octane flames are much
harder to ignite than the n-octane flames, as expected. It is also seen that iso-octane
flames are only slightly easier to ignite compared to n-dodecane flames.
Comparing the JFs to the SCHs, all JF flames are harder to ignite than n-dodecane
except JP-10, which has a significantly smaller molecular mass than n-dodecane. JP-10
is lighter than n-decane but is still much harder to ignite, demonstrating that napthenic
compounds have reduced ignition propensity compared to n-paraffins. In general the JF
flames have similar ignition propensity to n-tetradecane flames.

135

Figure 11.5 Variation of T
ign
with the fuel/N
2
mass fraction for the SCHs.

There is a large range of T
ign
for the different JFs with JP-10 and JP-8 flames
exhibiting the greatest and lowest ignition propensity respectively. The DCL-4765
flames are just slightly harder to ignite than n-dodecane, and are the second easiest to
ignite compared to all JFs. It is of interest to note that while flames of DCL-4765 and JP-
10 are the easiest to ignite, they have reduced resistance to extinction as reported earlier
and that both fuels consist largely of cyclo-alkanes. Flames of low-aromatic fuels,
namely RP-1, JP-7, FT-4734, and Jet-A-3638, exhibit similar ignition characteristics. As
expected, flames of the three high-aromatic fuels, JP-8 and the two batches of Jet-A, are
970
990
1010
1030
1050
1070
1090
1110
0.045 0.047 0.049 0.051 0.053 0.055 0.057 0.059
Ignition Temperature, T
ign
, °C
Fuel/N
2
Mass Fraction
n-C
7
H
16
n-C
5
H
12
n-C
6
H
14
n-C
8
H
18
n-C
9
H
20
n-C
10
H
22
iso-C
8
H
18
n-C
12
H
26
n-C
14
H
30

136
the hardest to ignite. Figure 11.6 depicts the difference in ignition characteristics among
the three batches of Jet-A. As expected also, the fuels with higher aromatic content
exhibit reduced ignition propensity.

Figure 11.6 Variation of T
ign
with the fuel/N
2
mass fraction for the 3 batches of Jet-
A.

The ignition behavior of two synthetic fuels is shown in Fig. 11.7, and the difference
between a n-+iso-paraffin fuel and a cyclo-paraffin fuel is apparent. In contrast to
extinction characteristics, the cyclo-paraffin flames have greater ignition propensity
compared to n-+iso-paraffins. This suggests that there is a different order of reactivity
980
1000
1020
1040
1060
1080
1100
1120
0.051 0.053 0.055 0.057 0.059 0.061
Ignition Temperature, T
ign
, °C
Fuel/N
2
Mass Fraction
JetA-3638
JetA-3602
JetA-4658

137
from an ignition and extinction point of view. Flames of cyclo-paraffin compounds were
found to exhibit extinction behavior that is between that of n-+iso-paraffins and
aromatics, but it appears that flames of iso-paraffins are less likely to ignite than those of
cyclo-paraffins. The results for SCHs reveal that flames of n-paraffins are far easier to
ignite than iso-paraffins, and that n-paraffin flames are the easiest of all compounds to
ignite. This suggests that the iso-paraffin content of FT-4734 inhibits ignition compared
to the purely cyclo-paraffin fuel of DCL-4756.

Figure 11.7 Variation of T
ign
with the fuel/N
2
mass fraction for the two synthetic
fuels.

980
1000
1020
1040
1060
1080
1100
1120
0.052 0.053 0.054 0.055 0.056 0.057 0.058 0.059
Ignition Temperature, T
ign
, °C
Fuel/N
2
Mass Fraction
FT-4734
DCL-4765

138
Figure 11.8 depicts T
ign
of flames of JP-8, Jet-A-4658, and the S6 and S12 surrogates.
Flames of S6 and S12 exhibit, again, similar ignition responses (S6 flames ignite only
slightly easier than S12), which however, are different compared to the JP-8 flames that
are distinctly harder to ignite compared to both S6 and S12. The smaller average
molecular weight of the surrogates result in flames with stronger burning characteristics
compared to JP-8, as manifested by their greater resistance to extinction and their greater
ignition propensity.

Figure 11.8 Variation of T
ign
with the fuel/N
2
mass fraction for the two surrogate
fuels, JP-8, and the averaged batch of Jet-A-4658.

980
1000
1020
1040
1060
1080
1100
1120
0.051 0.053 0.055 0.057 0.059 0.061
Ignition Temperature, T
ign
, °C
Fuel/N
2
Mass Fraction
S12
JP-8
S6
JetA-4658

139
11.4 Conclusions
Extinction strain rates and ignition temperatures of a wide range of liquid fuels were
determined in the counterflow configuration under non-premixed conditions. Single
component hydrocarbons, practical jet fuels, and their surrogates were tested and the
results were compared to assess their relative performance. The reported experimental
flame data are the first ones to be reported for such a wide range of jet fuel and large
liquid hydrocarbons, and they are essential towards the development of reliable surrogate
fuels and the attendant kinetics models.
It was found that the fuels with lower carbon number result in flames that are more
resistant to extinction compared to the ones with higher carbon number. The difference
in the extinction characteristics of n-octane and iso-octane flames illustrates the
importance of branched nature of hydrocarbons on the flame response. Results show that
the jet fuel flames exhibit a rather wide range of resistance to extinction, which in general
was found to be close to n-tetradecane. Flames of the two JP-8 surrogates have similar
extinction characteristics, with both being more resistant to extinction compared to JP-8.
Flames of fuels with lower carbon number were also determined to ignite easier than
the ones with higher carbon number. Comparing n-octane and iso-octane flames revealed
that the branched nature of the fuel has a greater effect on ignition than extinction, with
the ignition characteristics of iso-octane flames being similar to n-dodecane flames. Jet
fuel flames exhibit a large range of ignition temperatures with JP-10 being the easiest and
JP-8 being the hardest to ignite. Flames of proposed JP-8 surrogates ignite more readily
compared to JP-8.

140
Comparing all data reveals certain hierarchy for both extinction and ignition
characteristics. In terms of extinction, flames of n-paraffins were found to be the most
resistant to extinction followed by iso-paraffins, then cyclo-paraffins, then aromatics. In
terms of ignition, flames of n-paraffins were found to be the easiest to ignite followed by
cyclo-paraffins, then iso-paraffins, then aromatics. The general flame response due to
chemical classification could be used in surrogate fuel development.
Surrogate fuels that have been generated by considering the physical properties and
chemical classification of the real fuels, do not necessarily match their transport
properties. Choosing lower molecular weight surrogate constituents with relatively
known kinetics may result in falsification of the real fuel’s transport properties, which
could notably affect the flame response.

141
11.5 Chapter 11 References
[1] Defense Energy Support Center, Petroleum Quality Information System, annual
fuel property surveys, http://www.desc.dla.mil/DCM/DCMPage.asp?pageid=99.,
1999-2004.
[2] J.T. Edwards, “USAF Supercritical Hydrocarbon Fuels Interests,” AIAA
Aerospace Sciences Meeting, Reno, Paper 93-0807, Jan. 1993.
[3] C.P. Wood, V.G. McDonell, R.A. Smith, G.S. Samuelson, J. Propulsion and
Power Vol. 5, No. 4, 1989, pp. 399-405.
[4] W.D. Schulz, J. Propulsion and Power Vol. 9, No. 1, 1993, pp. 5-9.
[5] R.C. Farmer, P.G. Anderson, G.C. Cheng, B.L. Myruski, R.W. Pike, “Propulsion
Chemistry for CFD Applications,” Final Report. On Contract NAS8-40574,
National Aeronautics and Space Administration, Marshall Space Flight Center,
AL, Sept. 1997.
[6] A. Violi, S. Fan, E.G. Eddings, A.F. Sarofim, S. Granata, T. Faravelli, E. Ranzi,
Comb. Sci. Technol. 174 (11-12) (2002) 399-417.
[7] J.A. Cooke, M. Bellucci, M.D. Smooke, A. Gomez, A. Violi, T. Faravelli, E.
Ranzi, Chemical and Physical Processes in Combustion 337-340 (2003).
[8] F.N. Egolfopoulos, P.D. Dimotakis, Combust. Sci. Technol. 162 (2001) 19-36.
[9] Y. Dong, A.T. Holley, M.G. Andac, F.N. Egolfopoulos, S.G. Davis, P. Middha,
H. Wang, Combust. Flame 142 (2005) 374-387.
[10] A.T. Holley, Y. Dong, Y. Fan, M.G. Andac, F.N. Egolfopoulos, Combust. Flame
144 (2006) 448-460.

142

Chapter 12

Sensitivity of Propagation and Extinction of Large
Hydrocarbon Flames to Fuel Diffusion

12.1 Introduction
The focus of the previous chapters was the determination of important fundamental
flame data for practical liquid hydrocarbons. This included fuels of a wide range of
molecular weights and chemical compositions. Generally global flame properties,
including the laminar flame speeds, ignition temperatures, and extinction strain rates, are
used for validating chemical kinetic models of hydrocarbon combustion [1]. The bulk of
the analysis of the numerical predictions in the previous chapters has focused on
chemical kinetics, though it is known that the transport properties do affect the results.
Recent flame studies have showed that for normal, higher alkanes the laminar flame
speeds are generally insensitive to fuel compound size [2-5]. Similar observations have
been made for the extinction strain rates of premixed n-decane and n-dodecane flames

143
over an extended range of equivalence ratios [4]. The observed insensitivity appears to
be related to the rapid decomposition of the fuel molecules, which renders the kinetic
subtlety of various normal alkanes unimportant [6]. This feature has been utilized for
developing simplified reaction models of normal alkane oxidation at high temperatures
(e.g., [6-8]). The role of molecular diffusion, however, remains unclear in the observed
flame phenomena, particularly considering that disparity in the mass diffusivity of the
fuel and oxidizer/diluent gas increases for higher alkanes.
0
100
200
300
400
500
600
700
4 6 8 10 12 14
Extinction Strain Rate, K
ext
(s
-1
)
Number of Carbon Atoms
Fuel/N
2
= 0.1 by mass
0.08
0.06

Figure 12.1 Counterflow non-premixed extinction strain rate K
ext
as a function of
number of carbon atoms in the nitrogen-diluted n-alkane jet against an oxygen jet.
Data (symbols) are taken from [9]; lines: fits to data.

144
Non-premixed strained flames, however, present a different picture. In a recent study
[9], extinction strain rates of non-premixed flames of n-alkanes, ranging from n-pentane
to n-tetradecane, were determined in the counterflow configuration; the fuel stream was
nitrogen-diluted and while the oxidizer stream was pure oxygen. The results revealed
that the extinction strain rate was rather sensitive to the fuel size, as seen in Fig. 12.1. It
can be seen that for the same fuel/air mass ratio, the heat release characteristics of each
curve are similar, but the variation of the extinction strain rate is evident and systematic.
Namely, a decrease in the molecular size/weight leads to an increased resistance to
extinction. The observed variation may be attributed to the coupling between reaction
kinetics and molecular transport. Yet, the importance of the two effects and more
specifically the role of molecular diffusion remain unclear. This question was addressed
in the current study.
The importance of transport processes in flames has been recognized some time ago
(e.g., [10]). Studies have addressed the effect of molecular transport on flame structure,
ignition, propagation, and extinction (e.g., [3,11-26]). The sensitivity to mass diffusion
has been quantified for laminar flame speeds of hydrogen and small hydrocarbons [3, 19-
23], the structure of burner-stabilized hydrogen/carbon monoxide flames [24], the
ignition states of hydrogen, n-heptane, and n-octane flames [25], and for flame speeds
and extinction strain rates of methanol, ethanol, n-heptane and n-octane flames [26].
Emerging from these studies is the notion that molecular diffusion can exert greater
influence on a flame phenomenon than chemical kinetics [21-23,25,26]; note that the
sensitivity to diffusion is expected to grow as the disparity in mass diffusivity of major

145
species grows, as in the case of hydrogen or large hydrocarbon flames. Yet, in recent
efforts of developing kinetic models for real fuel surrogates, transport properties for
higher hydrocarbon fuels have not received adequate attention.
The objective was to determine computationally and by invoking rigorous sensitivity
analyses, the relative importance of molecular diffusion and chemical kinetics for laminar
flame speeds and extinction strain rates, using n-dodecane as the primary test cases and
n-heptane as a supplement. This was done to assess the possible errors induced by
overlooking the transport properties in mechanism development.

12.2 Lennard-Jones Spherical Potential Parameters
To date, the only set of Lennard-Jones (LJ) 12-6 potential parameters available for
large normal alkanes is that of Mansoori et al. [27], who employed the method of
corresponding states and liquid properties, including molar volume, viscosity, and
thermal conductivity, to estimate the potential parameters. There exists no directly
experimentally determined LJ parameter for large hydrocarbons, such as n-dodecane.
The method of corresponding states relies on the extrapolation of a small set of data to
other species which haven’t been tested. Considering the non-spherical nature of the
molecules and that the molecular configurations and interactions are not identical
between the liquid and vapor states, the validity of these LJ parameters for calculating
gaseous diffusion coefficients remains questionable. The most up to date LJ parameters
[28] were used in this study which were determined with the correlations of
corresponding states of Tee et al. [29], as used in an earlier study on the transport

146
properties of aromatics [30]. Despite using the most advanced estimates of the transport
properties, there still exist large uncertainties.

12.3 Numerical Approach
Three types of flame phenomena were studied, as shown in Table 12.1. They include
laminar flame speeds, S
u
o
, and extinction strain rates, K
ext
, of premixed flames established
by counterflowing an n-dodecane/air jet at 403 K against a nitrogen jet at 298 K.
Extinction strain rates were also studied for non-premixed flames established by
counterflowing an n-dodecane/nitrogen jet at 403 K against an oxygen or air jet at 298 K.
For the case where molecular oxygen was used as the oxidizer, the fuel-to-nitrogen
dilution ratios span the range as employed in the experiment [9] (fuel/N
2
mass ratio =
0.05 to 0.1).
Table 12.1 Computational Test Cases
a

Flame Description
Laminar Flame Speed (T
u
= 403 K)
u
S
o
1.4 φ=1.4
u
S
o
1.0 φ=1.0
u
S
o
0.7 φ=0.7
Extinction strain rate of premixed flame
K
ext
1.4 Fuel/air (φ=1.4) vs. N
2

K
ext
1.0 Fuel/air (φ=1.0) vs. N
2

K
ext
0.7 Fuel/air (φ=0.7) vs. N
2

Extinction strain rate of non-premixed flame
K
ext
1-30 (air) Fuel/ N
2
=1/30
b
vs. air
K
ext
1-15 (air) Fuel/ N
2
=1/15
b
vs. air
K
ext
1-120 (O
2
) Fuel/ N
2
=1/120
b
vs. O
2

K
ext
1-60 (O
2
) Fuel/ N
2
=1/60
b
vs. O
2

a
All test cases are n-dodecane flames. For opposed
jet flames, the temperature of the fuel jet is 403 K;
the oxidizer jet temperature is 298 K.
b
molar ratio.

147
The computations were performed using a reduced n-dodecane kinetic model with 56
species and 289 reactions, based on a detailed model of 171 species and 1306 reactions
recently developed for high-temperature oxidation of n-paraffin compounds up to n-
dodecane [6, and http://ignis.usc.edu/Jet_Fuel_MechI.html]. The use of a reduced model
facilitates greatly the CPU time of the simulations, while the conclusions are not affected
by this simplification. The mechanism reduction was performed in two separate steps.
First, the base H
2
/CO/C
1
-C
4
model was trimmed from 111 species and 784 reactions into
a skeletal model of 52 species and 257 reactions using the method of Ref. 31. Second,
the chemistry of n-dodecane was reduced to a lumped reaction model, consisting of 4
species and 32 reaction steps [6], using a steady-state analysis of premixed n-dodecane
flames over a range of pressures and equivalence ratios, φ. The reduced model results in
laminar flame speeds that are within 1 cm/s of the detailed model for lean and
stoichiometric mixtures and 2 to 3 cm/s higher at φ = 1.4. An n-heptane sub-model of 12
species and 54 reactions was developed similarly for simulating non-premixed n-heptane
flame extinction.

12.4 Results and Discussion
12.4.1 Sensitivity of Flame Phenomena to Chemical Kinetics and Molecular Transport
We now examine the responses of various laminar flame phenomena to reaction
kinetics and diffusion process. Since the sensitivity computations were obtained by using
the mixture-averaged estimation, additional simulations were run to assess the impact of
this approximation as compared to the more rigorous multicomponent formulation. It

148
was found that overall, the mixture-averaged and multicomponent formulations result in
very similar sensitivity coefficients, and the details of the differences are discussed in a
subsequent section.

Figure 12.2. Logarithmic sensitivity coefficients with respect to reaction rate and
diffusion coefficients using mixture-averaged transport formulation. Light bars:
positive values; dark bars: negative values.

Figure 12.2 illustrates a summary of sensitivity coefficients of the model responses of
n-dodecane flames to key reaction rate parameters and binary diffusion coefficients.

149
Laminar flame speeds ( S
u
o
-φ) and extinction strain rates for premixed flames (K
ext
-φ) at
several representative φ, and extinction strain rates for non-premixed flames [K
ext
-fuel-
nitrogen (oxidizer)] were considered as specified in Table 12.1.
Several observations can be made from the computational results. The overall
influences of reaction rates and diffusion fluxes are about the same for the premixed
flame responses. The sensitivity to diffusion originates primarily from the O
2
-N
2
binary
diffusivity. Kinetically, the propagation rates of the premixed flames are found to be
insensitive to the chemistry specific to n-dodecane as shown here, since its
decomposition in the preheat zone of the flame is too rapid to be rate-limiting [6].
Rather, the sensitivity spectrum includes only the rate coefficients found typically in
H
2
/CO and small hydrocarbon combustion. As shown in Fig. 12.2, these are mainly
H + O
2
→ O + OH, CO + OH → CO
2
+ H, and H + O
2
(+M) → HO
2
(+M). The fact that
the computed responses of premixed flames are insensitive to kinetic or diffusion
parameters specific to n-dodecane suggests that the measured S
u
o
and K
ext
of premixed
flames should indeed be independent of the nature and size of the alkane fuel molecule.
Note that under similar conditions, the decomposition kinetics of higher n-alkanes is
similar to those of n-dodecane. Additionally, for smaller n-alkanes the larger diffusivities
would cause the flame propagation rate to depend even less on the fuel diffusion rate.
For non-premixed flame extinction though, the results are qualitatively different.
Figure 12.2 depicts that the computed strain rates are notably sensitive to the n-dodecane-
N
2
binary diffusion coefficient in addition to O
2
-N
2
. The large sensitivity to diffusivity
exists because the amount of fuel accessible to non-premixed counterflow flames is

150
diffusion controlled. The effect is further enhanced due to the proximity of the flames to
the stagnation plane, where the convective velocity approaches zero and diffusion is the
dominant mode of mass transport. While the extinction strain rates of non-premixed
flames are as sensitive to the reaction rate coefficients as to diffusion, the dominant
influence again comes from the reactions unrelated to the exact nature and size of the
fuel.
10
2
10
3
0.6 0.7 0.8 0.9 1.0
D
fuel-N2
(cm
2
/s)
Extinction Strain Rate, K
ext
(s
-1
)
Fuel/N
2
= 0.1 by mass
(dlnK
ext
/dlnD
fuel-N2
= 1.0 +/-0.2)
0.06 (dlnK
ext
/dlnD
fuel-N2
= 0.7 +/-0.3)
Number of Carbon Atom
5 6 7 8 9 10 12
0.08 (dlnK
ext
/dlnD
fuel-N2
= 0.8 +/-0.1)

Figure 12.3. Non-premixed flame extinction strain rate of normal alkanes as a
function of binary fuel-N
2
diffusion at 1200 K and 1 atm. Symbols: experimental
data for nitrogen diluted fuel jet against an oxygen jet [9]; lines: fits to data.

151
The fact that the response of nonpremixed flames is sensitive to the fuel diffusivity
indicates that K
ext
would be dependent on the molecular size of the fuel, since the fuel
diffusivity differs. To further examine this issue, the extinction strain rates of Fig. 12.1
are re-plotted as a function of the fuel-N
2
binary diffusion coefficient, as shown in Fig.
12.3. As expected, an increase in the diffusion coefficient increases the extinction strain
rate. As seen, lnK
ext
varies linearly with lnD
fuel-N2
, where the slope indicates the
experimental sensitivity of K
ext
to the binary diffusion coefficient,
2
ext
fuel
ln
ln
−
=
N
dK
s
dD
, (12.1)

Over the fuel-to-nitrogen mass ratios considered (0.06 to 0.1), s ranges from 0.7 to 1.0.
These values are consistent with the values of logarithmic sensitivity coefficients on the
n-dodecane-N
2
binary diffusion coefficient computed for K
ext
1-120(O
2
) and 1-60(O
2
),
which are on the order of 0.5.
Additional physical insight was gained by computing the extinction response of non-
premixed n-dodecane and n-heptane flames as a function of strain rate. The results
shown in Fig. 12.4 reveal K
ext
of n-heptane flames is greater compared to n-dodecane, in
agreement with experiments. Additionally, the n-dodecane-N
2
binary diffusion
coefficient was artificially replaced with that of n-heptane-N
2
. This led to a shift of the
turning point to larger strain rates (curve b). With all binary diffusion coefficients of n-
dodecane replaced by those of n-heptane (curve c), the strain rates at the turning point are
within 5% of the two fuels. The remaining difference is caused by small kinetic
differences of the two fuels. Under equal mass ratio, n-heptane results in slightly greater

152
heat release than n-dodecane due to the small difference in the carbon-to-hydrogen ratio.
Regardless, the experimental variation of nonpremixed K
ext
may be firmly attributed to
variations in the fuel diffusivity.

1450
1500
1550
1600
1650
1700
1750
300 400 500 600 700
Maximum Temperature (K)
Strain Rate, K (s
-1
)
n-C
7
H
16
n-C
12
H
26
a b c

Figure 12.4 Non-premixed flame extinction curves computed for n-heptane and n-
dodecane with fuel/nitrogen mass ratio of 0.08 against an oxygen jet. The three
n-C
12
H
26
cases are a) base case; b) the binary diffusion coefficient of n-C
12
H
26
-N
2

replaced by that of n-C
7
H
16
-N
2
; c) all binary diffusion coefficients of n-C
12
H
26

replaced by n-C
7
H
16
.

153
12.4.2 Species Flux Analysis
In general notable LSCs to the BDCs were found for reactants, active radicals, and
products. Depending on the configuration and stoichiometry, the magnitude and sign of
the LSCs can change. It is important to note that the total number of BDCs that exhibit
notable sensitivities, i.e. greater than 10
-2
, is much smaller than the total number of
reactions that do. Approximately 25% of the reaction rates and only 1-2% of the BDCs
exhibit sensitivities of the order of 10
-2
or greater. This is due to the fact that chemical
reactions have a first order effect on the profiles of most intermediate species, while
transport plays a secondary role. This is not the case for reactants and products, for
which transport can dictate their mass flux into or out of the flame, as well as for highly
active radicals such as H, O, and OH for which even a minor modification of their
concentration profiles due to transport can have a large effect on the flame response.
The LSC of S
u
o
on the n-C
12
H
26
⎯ N
2
BDC appears to be small, though it is still of
the order of 10
-2
. The LSC is negative for lean flames and positive for rich flames. In the
absence of stretch, it has been shown that as the fuel diffusivity increases, the diffusive
layer of the fuel increases at a greater rate for small molecular weight fuels [e.g., 11] thus
reducing the fuel diffusive flux into the reaction zone. This is not the case for very large
fuels, which have very low diffusivities. Figure 12.5 depicts the species profiles and the
convective and diffusive mass fluxes of n-C
12
H
26
for two stoichiometric freely
propagating flames (FPF). Simulations of the exact same conditions were performed, one
with original diffusivity of the n-C
12
H
26
⎯ N
2
BDC and one with the original BDC value
multiplied by a factor of 2.

154

Figure 12.5 The convective and diffusive fuel flux of n-C
12
H
26
for a FPFs.

As expected, the thickness of diffusive layer increases, but the growth is comparable
to that of the increased diffusivity. The extension of the diffusive layer does result in
lower fuel concentrations upstream of the flame, which results in reduced convective flux
into the flame. The ratio of the diffusive flux over the convective flux is less than unity
until approximately only 25% of the fuel remains for the original case, thus the
convective flux dominates the transport until the fuel reaches the flame. For the case of
increased diffusivity this is modified only slightly, and approximately 33% of the fuel
still remains when the two mass flux terms become equal. Increasing of the diffusivity
0.E+00
1.E-02
2.E-02
3.E-02
4.E-02
5.E-02
6.E-02
0.0E+00
5.0E-04
1.0E-03
1.5E-03
2.0E-03
2.5E-03
3.0E-03
3.5E-03
4.0E-03
4.5E-03
0.025 0.03 0.035 0.04 0.045 0.05 0.055 0.06
n-C
12
H
26
Mass Fraction
Mass Flux, g/cm
2
-s
Distance cm
conv orig
diff orig
conv-2
diff-2
c12 orig
c12-2
Convective Mass Flux
Diffusive Mass Flux
Perturbed Convective Mass Flux
Perturbed Diffusive Mass Flux
n-C
12
H
26
Mass Fraction
Perturbed n-C
12
H
26
Mass Fraction

155
does result in a shift in the relative magnitude of the two mass flux terms, but the total
mass flux experiences only a minor perturbation. In the convection-dominated region,
there is only a 2-3% decrease in the total mass flux, while in the diffusion dominated
region the increase between the two cases is less than 25% when there is only 5% of the
fuel remaining. The species profiles of n-C
12
H
26
shown in Fig 12.5, illustrate a decrease
in the concentration of the fuel in the convection dominated-region, but the two profiles
merge in the diffusion-dominated region. Therefore a forced perturbation of the n-
C
12
H
26
⎯ N
2
BDC by a factor of 2 results in almost negligible modifications to the total
mass flux for FPFs, but does result in an overall decrease in the n-C
12
H
26
concentration
through the flame. As a result the burning intensity of fuel-lean flames is reduced. Thus,
increasing the n-C
12
H
26
⎯ N
2
BDC makes lean flames more lean and rich flames less
rich locally that burn less and more intensely respectively. This effect has been shown to
result in large sensitivities for small fuels, such as H
2
[11,13], since a significant portion
of the fuel flux is due to diffusion. However, this is not the case for n-C
12
H
26
due to the
fact that convection is the main mode of transport for n-C
12
H
26
due to its low diffusivity.

156

Figure 12.6 The convective and diffusive fuel flux of n-C
12
H
26
for a NEF.

Figure 12.6 depicts the results of similar analysis performed for a stoichiometric
premixed n-C
12
H
26
near extinction flame (NEF) and there are differences compared to
FPF. The axial convective mass fluxes are pointing always towards the stagnation plane
and they are treated as positive quantities. The sign of the diffusive mass fluxes is either
positive or negative for fluxes when pointing towards the right and left boundary
respectively. A NEF is experiencing reactant leakage and incomplete combustion, which
results in lower flame temperatures, approximately 200 to 300 K for alkanes. This leads
-1.E-02
0.E+00
1.E-02
2.E-02
3.E-02
4.E-02
5.E-02
6.E-02
7.E-02
-1.5E-03
-5.0E-04
5.0E-04
1.5E-03
2.5E-03
3.5E-03
0.810 0.815 0.820 0.825 0.830 0.835 0.840
n-C
12
H
26
Mass Fraction
Mass Flux, g/cm
2
-s
Distance cm
Conv
Diff
Conv-2
Diff-2
C12
C12-2
Convective Mass Flux
Diffusive Mass Flux
Perturbed Convective Mass Flux
Perturbed Diffusive Mass Flux
n-C
12
H
26
Mass Fraction
Perturbed n-C
12
H
26
Mass Fraction

157
to reduced reaction rates and radical pools. Thus, the consumption rate of n-C
12
H
26
is
reduced and thus it can leak farther into the flame as compared to FPFs. Consequently,
the diffusive transport becomes more important, and the transition from the convection-
dominated to the diffusion-dominated region occurs while more of the fuel remains
unreacted. This results in magnitudes of the convective and diffusive transport rates that
are equal when there is still almost 40% of the fuel remaining, as opposed to only 25%
for FPF. Comparing the two species profiles in Fig. 12.6 it is clear that the mass fraction
of n-C
12
H
26
is higher through the flame for the case with the higher diffusivity. Thus,
increasing the fuel diffusivity for NEFs increases the local stoichiometry of the system,
which is not the case for FPFs.
The LSCs of K
ext
for non-premixed flames using O
2
as the oxidizer to the n-
C
12
H
26
⎯ N
2
BDC, are notably larger compared to premixed flames. This is because
non-premixed flames are inherently diffusion-controlled. Additionally, for large fuel
molecules with low diffusivity, a non-premixed flame is located close to the SP that is a
region of low convective velocities. Thus, the diffusive flux of fuel dominates the
convective one at the flame vicinity.
The LSCs of K
ext
for non-premixed flames with air as the oxidizer to the BDCs are
larger than the case of O
2
being the oxidizer. For this configuration the flame resides on
the oxidizer side, though the temperature is still 1200 K at the stagnation plane. This
results in approximately 3% unreacted n-C
12
H
26
at the stagnation plane. Due to the very
low velocities near the stagnation plane, the ratio of diffusive flux of fuel over its
convective counterpart is as much as 10 times greater than the case of O
2
being the

158
oxidizer. The mass fluxes of n-C
12
H
26
for the 1/30 fuel/N
2
molar ratio case are shown in
Fig 12.7. The diffusive transport dominates the convective transport for most of the
domain, and is the only mode of transport near the SP. These results suggest that a 10%
uncertainty in the n-C
12
H
26
⎯ N
2
BDC results in a 10% uncertainty in the K
ext
for the
1/30 fuel/N
2
dilution ratio.

Figure 12.7 Convective and diffusive mass fluxes and species mass fraction of n-
C
12
H
26
as a functions of distance for a near extinction non-premixed 1/30 fuel/N
2

molar ratio flame with air as oxidizer.

The other reactant BDC that exhibits large LSCs is the O
2
⎯ N
2
BDC, with values
comparable or greater than the LSCs to the rate constants of the main branching reaction
0.0E+00
2.0E-02
4.0E-02
6.0E-02
8.0E-02
1.0E-01
1.2E-01
1.4E-01
1.6E-01
1.8E-01
2.0E-01
-4.0E-04
-2.0E-04
0.0E+00
2.0E-04
4.0E-04
6.0E-04
8.0E-04
1.0E-03
0.60 0.62 0.64 0.66 0.68 0.70 0.72 0.74 0.76 0.78 0.80
n-C
12
H
26
Mass Fraction
Mass Flux, g/cm
2
-s
X cm
Convective
Diffusive
n-C12H26
Convective Mass Flux
Diffusive Mass Flux
n-C
12
H
26
Mass Fraction

159
(BR) H+O
2
⇔ OH+O. This is physically sound, as the diffusive transport of O
2
directly
affects the forward progress of BR. However, the O
2
⎯ N
2
BDC is a well-known value,
and easily measured unlike the BDCs of large alkanes. For FPFs an increase of the O
2

diffusivity results in lower O
2
flux and local concentration through the flame and as a
result in lower BR rate. Thus, for all φ the sensitivity is negative with notably large
absolute value for rich flames for which O
2
is the deficient reactant. For NEFs the
opposite effect is seen as compared to n-C
12
H
26
, and the increased diffusivity of O
2
results
in a slightly higher O
2
mass flux. On the other hand, the O
2
concentration values
decrease close to the reaction zone, as the thickness of the O
2
diffusive layer increases.
Thus, increasing the O
2
diffusivity has a negative affect on the forward progress of BR.
Notable LSCs on the BDCs of the active H, OH, and O radicals with N
2
, i.e. H ⎯ N
2
,
O ⎯ N
2
, and OH ⎯ N
2
were determined. These radicals are produced within the
reaction zone whose thickness is controlled by the activation energy. Figure 12.8 depicts
the species profile and mass flux of H radical in a non-premixed NEF with 1/120 fuel/N
2

molar ratio flame with O
2
as the oxidizer. It is clear that the H radical is transported
largely by diffusion. In Fig. 12.8 the right boundary corresponds to the fuel jet, while the
left boundary to the oxidizer jet. For all cases studies, the radical concentration and
temperature gradients are higher on the fuel side compared to the oxidizer side. Thus
increasing the H ⎯ N
2
BDC preferentially increases the diffusive flux of H into the fuel,
having thus a negative effect on rate of BR that peaks on the oxidizer side. For premixed
flames on the other hand, fuel and O
2
are mixed so that increasing the H ⎯ N
2
BDC
increases the diffusive flux of H into the oxidizer.

160

Figure 12.8. Convective and diffusive mass fluxes and species mass fraction of H as
functions of distance for a near extinction non-premixed 1/120 fuel/N
2
molar ratio
flame with O
2
as oxidizer.

LSCs on the BDCs of CO, CO
2
, and H
2
O with both N
2
and O
2
, are also non-
negligible and in general greater than that of the active radicals. For premixed flames in
general, the LSCs are all positive with respect to N
2
and all negative with respect to O
2
.
Similarly to the radicals, CO, CO
2
, and H
2
O are produced within the reaction zone and
increasing their diffusivity into N
2
increases their diffusive fluxes into the reactants. The
increased diffusion of those products into the reactants modifies the species profiles only
slightly, due to counter-balancing convective flux. The positive sensitivity is due to a
0.0E+00
2.0E-06
4.0E-06
6.0E-06
8.0E-06
1.0E-05
1.2E-05
1.4E-05
1.6E-05
1.8E-05
-1.5E-06
-1.0E-06
-5.0E-07
0.0E+00
5.0E-07
1.0E-06
1.5E-06
2.0E-06
2.5E-06
3.0E-06
1.15 1.20 1.25 1.30 1.35
H Mass Fraction
Mass Flux, g/cm
2
-s
Distance cm
conv
diff
h
Convective Mass Flux
Diffisive Mass Flux
H Mass Fraction

161
thermal effect, since the hot products are diffusing into the colder reactants, which results
in a slight increase in the temperature profile through the reaction zone. Diffusion into
the O
2
similarly would increase the temperature, though to a lesser degree, but would also
serve to increase the diffusivity of O
2
, which would negatively affect premixed flames as
previously discussed.

12.4.3 Effect of Transport Model
To assess the choice of the mixture-average formulation of transport, simulations
were repeated using full multi-component formulation the resulting LSCs are shown in
Fig. 12.9. To the first order there is very little change in calculated sensitivities compared
to those reported in Fig. 12.2. The reaction rate sensitivities do not change, and neither
did most of the BDC sensitivities. The largest effect of the formulation is to actually
reduce the diffusivity of the fuel, and resulting in making the flames more susceptible to
extinction. This change has a minor affect on premixed flames, but results in a large
increase in the LSCs to the BDCs of the reactants for non-premixed flames. The decrease
in the K
ext
, and subsequently the convective velocities, further enhances the importance of
diffusion. LSCs as large as 3 are found for the weakest flames! Qualitatively speaking,
the relative importance of molecular transport and chemical kinetics doesn’t really
change for premixed flames, but does increase for non-premixed flames. Additionally
the flame configurations that result in notable LSCs on BDC does not change due to the
choice of transport formulation.

162

Figure 12.9. Logarithmic sensitivity coefficients with respect to reaction rate and
diffusion coefficients using the multi-component formulation of transport. Light
bars: positive values; dark bars: negative values.

12.5 Conclusions
The logarithmic sensitivity coefficients on both reaction rates and binary diffusion
coefficients were calculated for laminar flame speeds as well as extinction strain rates for
premixed and non-premixed flames. Results reveal that the premixed flame response is

163
generally insensitive to the fuel diffusivity. Coupled with a lack of sensitivity to the
reaction kinetics specific to the fuel molecule, the computed results satisfactorily explain
the insensitivity of the phenomena of flame propagation and extinction to the variation of
the size of normal alkane fuels. In contrast, extinction strain rates of non-premixed
flames were found to be highly sensitive to the fuel diffusivity, in that larger fuel
diffusivities render the flames more resistant to extinction. Hence the variation of the
extinction strain rate of non-premixed flames with the normal alkane size, as observed in
Ref. 9, is satisfactorily explained by the variation of the mass diffusivity of the fuel.
Physical insight into the variation in flame structure due to changes in mass diffusivity
was provided. Additionally, it was shown that there is very little change to the qualitative
results due to choice of transport formulation, but some quantitative dependencies do
exist.

164
12.6 Chapter 12 References
[1] C. K. Law, C. J. Sung, H. Wang, T. F. Lu, AIAA J. 41 (2003) 1629-1646.
[2] S. G. Davis, C. K. Law, Combust. Sci. Technol. 140 (1998) 427-449.
[3] Z. Zhao, J. Li, A. Kazakov, F.L. Dryer, Combust. Sci. Technol. 177 (2005) 89-
106.
[4] K. Kumar, C. J. Sung, Combust. Flame 151 (2007) 209-224.
[5] C. Ji, X.Q. You, A.T. Holley, Y.L. Wang, F.N. Egolfopoulos, H. Wang,
“Propagation and Extinction of Premixed n-Dodecane/Air Flames,” submitted for
consideration at the 32nd International Symposium on Combustion, 2007.
[6] X.Q. You, F.N. Egolfopoulos, H. Wang, “Detailed and Simplified Kinetic Models
of n-Dodecane Oxidation: The Role of Fuel Cracking in Aliphatic Hydrocarbon
Combustion,” submitted for consideration at the 32nd International Symposium
on Combustion, 2007.
[7] T.J. Held, A.J. Marchese, F.L. Dryer, Combust. Sci. Technol. 123 (1997) 107-146.
[8] S.P. Zeppieri, S.D. Klotz, F.L. Dryer, Proc. Combust. Inst. 28 (2000) 1587-1595.
[9] A.T. Holley, Y. Dong, M.G. Andac, F.N. Egolfopoulos, T. Edwards, Proc.
Combust. Inst. 31 (2007) 1205-1213.
[10] G. Dixon-Lewis, in: J. Warnatz, W. Jäger (Eds.), “Complex Chemical Reaction
Systems, Mathematical Modelling and Simulations,” Springer, Berlin, 1984, p.
265.
[11] B.A. Williams, Combust. Flame 124 (2001) 330-333.
[12] P.A. Libby, K.N.C. Bray, J.B. Moss, Combust. Flame 34 (1979) 285-301.
[13] A. Ern, V. Giovangigli, Combust. Theor. Model. 2 (1998) 349- 372.
[14] A. Ern, V. Giovangigli, Combust. Sci. Technol. 149 (1999) 157-181.
[15] R. Knikker, A. Dauptain, B. Cuenot, T. Poinsot, Combust. Sci. Technol. 175
(2003) 1783-1806.
[16] R. Seiser, H. Pitsch, K. Seshadri, W. Pitz, H .J. Curran, Proc. Combust. Inst. 28
(2000) 2029-2037.
[17] V. Gopalakrishan, J. Abraham, Combust. Flame 136 (4) (2004) 557-566.

165
[18] P. H. Paul, J. Warnatz, Proc. Combust. Inst. 27 (1998) 495-504.
[19] H. Wang, Chem. Phys. Lett. 325 (2000) 661-667.
[20] Z. Yang, B. Yang, H. Wang, paper 237, “Inference of H-Atom Diffusion
Coefficient on Laminar Flame Simulation,” Proceedings of the Second Joint
Meeting of the U.S. Sections of the Combustion Institute, Berkeley, CA, March
2001.
[21] P. Middha, B. Yang, H. Wang, Proc. Combust. Inst. 29 (2002) 1361-1369.
[22] Y. Dong, A.T. Holley, M.G. Andac, F.N. Egolfopoulos, S.G. Davis, P. Middha,
H. Wang, Combust. Flame 142 (2005) 374-387.
[23] P. Middha, H. Wang, Combust. Theory Model. 9 (2005) 353-363.
[24] M.K. Mishra, R.A. Yetter, Y. Reuven, H. Rabitz, M.D. Smooke, Int. J. Chem.
Kinet. 26 (1994) 437-453.
[25] M.G. Andac, F.N. Egolfopoulos, Proc. Combust. Inst. 31 (2007) 1165-1172.
[26] A.T. Holley, Y. Dong, M.G. Andac, F.N. Egolfopoulos, Combust. Flame 144
(2006) 448-460.
[27] G.A. Mansoori, V.K. Patel, M. Edalat, Int. J. Thermophys. 1 (1980) 285-298.
[28] A.T. Holley, X.Q. You, E. Dames, H. Wang, F.N. Egolfopoulos, “Sensitivity of
Propagation and Extinction of Large Hydrocarbon Flames to Fuel Diffusion,”
accepted for publication at the 32nd International Symposium on Combustion,
2007.
[29] L.S. Tee, S. Gotoh, W.E. Stewart, I&EC Fundam. 5 (1966) 356-363.
[30] H. Wang, M. Frenklach, Combust. Flame 96 (1994) 163-170.
[31] H. Wang, M. Frenklach, Combust. Flame 87 (1991) 365-370.

166

Chapter 13

Concluding Remarks and Recommendations

13.1 Concluding Remarks
Significant contributions towards the understanding of large liquid hydrocarbon
combustion were made in the present dissertation. Ignition and extinction states of a
wide range of single component hydrocarbons and practical liquid fuels were
experimentally determined in the counterflow configuration. Both premixed and non-
premixed flames were studied. Detailed numerical simulations of the experiments were
conducted using a wide range of chemical kinetic mechanisms. Additionally, rigorous
sensitivity analyses on both chemical kinetics and molecular transport as well as reaction
path analysis were performed to gain insight into the controlling mechanisms of the
various combustion phenomena considered.
The main motivation for this work was the scarcity or lack of high quality
fundamental flame data for liquid fuels that are of relevance to transportation and air-
breathing propulsion. Such data are critically needed for two reasons. The first is to

167
evaluate the performance of various fuels importance so that appropriate surrogates of
practical fuels can be formulated. The second is the development and validation of
chemical kinetics and molecular transport models that are needed in large-scale
simulations. To that end, the counterflow configuration was utilized to experimentally
determine ignition and extinction limits for premixed and non-premixed flames of n-
alkanes with a carbon number ranging from C
5
to C
14
, small alcohols and simple aromatic
compounds, samples of gasoline, gasoline surrogates, as well as a range of practical jet
fuels their surrogates. Detailed numerical simulations were performed for all cases for
which chemical kinetic mechanisms were available.
The extinction studies of premixed flames of alcohols and primary reference fuels of
gasoline revealed that mechanisms developed to predict laminar flame speeds, do not
ensure accurate prediction of the extinction limits. This is despite the fact that the two
phenomena occur at high temperatures. Sensitivity on kinetics and reaction path analyses
showed that both flame propagation and extinction exhibit in general, similar qualitative
dependence on kinetics as well as similar species consumption pathways. Compared to
propagation, however, the sensitivity of extinction on kinetics was determined to be
notably greater. Compared to propagation, extinction was found, under certain
conditions, to be sensitive to some additional reactions. This is because of the incomplete
combustion that occurs for near-extinction flames, which has a notable effect on the
flame structure compared to freely propagating flames. Additionally, it was shown that
extinction exhibits a larger sensitivity to molecular transport compared to propagation.

168
These findings need to be considered during the development of detailed chemical
kinetics mechanisms. Namely, if laminar flame speed is the only flame property used for
validation/optimization mechanism this may not be sufficient for establishing reliable
high temperature (flame) kinetics. When the phenomenon of extinction is used also as
part of the validation/optimization additional information and constraints are imposed.
More specifically, in the case of extinction all sensitivities are increased and in some
cases additional reactions are sensitized. Given the complex interactions between fluid
mechanics and kinetics in large-scale simulations of practical combustors, it is essential
that kinetics models describe accurately flame propagation as well as extinction. Finally,
the effect of molecular transport needs to be quantified accurately as it plays a major role
on the response of laminar flames, but it can be also important even in turbulent
combustion under certain conditions.
The studies on aromatic compounds reinforced the previously stated conclusions
regarding the ability of a mechanism to predict both propagation and extinction
phenomena. Comparison between the experimental data showed that aromatic
compounds have similar extinction states to that of the gasoline primary reference fuels,
which has significant implications on surrogate development. Namely that the extinction
state will be affected only minimally by the aromatic content of a surrogate, and that
toluene appears to be a better compound choice for gasoline surrogates.
The studies on real gasoline yielded a very important result for the development of
gasoline surrogates, that the flame properties of any random sample of gasoline could
vary greatly. It was shown experimentally that both the ignition and extinction states of

169
the samples of gasoline varied greatly, but when averaged samples were tested their
properties begun to converge. This proves that either an averaged sample of gasoline, or
even better a single averaged batch distributed by the government agencies to ensure
consistency between research groups, need to be used when developing gasoline
surrogates.
Studies on non-premixed flames illustrated the dependency of both ignition and
extinction states on the fuel size. This is different from the premixed cases, where only a
minor variation due to fuel size was found. This suggests that the fuel surrogate must
come close to approximating the overall fuel size of the parent fuel, when fuel oxidation
takes place in the non-premixed mode. Additionally, considering the relative
performance of the JP-8 and its surrogates, both of which approximately match the
chemical classifications and hydrogen to carbon ratio, the smaller species present in the
surrogates are not representative of the real JP-8. Both surrogates exhibited stronger
burning intensity, being easier to ignite and harder to extinguish, and this is due to the
larger diffusivities that smaller molecules have. Comparing all data reveals certain
hierarchy for both extinction and ignition. The general flame response due to chemical
classification could be used in surrogate fuel development.
To further explore the influence of fuel size on various experimental data, a
systematic study was performed on the relative effects of reaction rates and binary
diffusion coefficients. The simulations revealed that in general premixed flame
phenomena exhibit only minor sensitivities to the diffusion coefficients, while non-
premixed flames can have very large sensitivities. This is also in agreement with well-

170
established flame theories. The same numerical experiment showed that the sensitivities
to fuel specific reactions for all cases are minor. This explains why in past studies it has
been shown that the laminar flame speeds of various large hydrocarbons have very
similar values, since their kinetic differences are inconsequential and their diffusivities
don’t affect the results.

13.2 Future work
The present work was the beginning of a major effort to understand the combustion
processes of practical liquid fuels. The body of work that still needs to be done dwarfs
current literature. The experimental work will include premixed and non-premixed
flames, studies at both reduced and elevated pressures, burning rates, ignition and
extinction states, species and thermal profiles through flames, and a large amount of work
in homogeneous systems. Equally large, is the task yet to be done in the area of chemical
kinetics. Specific issues that need to be addressed are indicated in the following.
1. While global flame properties can be determined more readily it is difficult to
extract chemical kinetic information. They provide a rigorous test of chemical
kinetic mechanism, requiring accurate modeling of both the chemistry and
transport, and thus can be used in the validation process. Though sensitivity
analysis information about the controlling processes can be derived, it is
insufficient to deduce the source of the deficiency of a mechanism. Local
measurements of temperature and species through the flame are needed, so that
they can be compared with numerical predictions. Ideally laser diagnostics

171
should be introduced to measure the profiles of either stable or radical
intermediates through the flame. This would provide more confidence in the
kinetics allowing thus for the validation of transport coefficients.
2. The experimental parameter space must be increased, primarily the initial
thermodynamic state, i.e. initial temperature and pressure. Pressure is especially
important for jet turbine operation, where the inlet pressure can vary from 0.5 to
50 atmospheres. High altitude flight occurs with low engine pressure, but takeoff
occurs at very high pressures, and as turbine technology improves even higher
pressures will be experienced as the compression ratio increases. This pressure
range requires chemical kinetic mechanisms developed for jet fuels to reproduce
accurately the pressure dependencies of rate constants to allow for high fidelity
engine simulation. In addition to extending the range of the initial
thermodynamic state, further experiments in premixed, non-premixed, and
partially premixed configurations are needed. In terms of global flame properties,
experiments must include the determination of the laminar flame speeds, in
addition to ignition and extinction limits.
3. The study needs to extend to other chemical classifications. The bulk of the
single component hydrocarbons that were tested in this dissertation are n-alkanes,
though straight-chain paraffins make up only a small portion of most practical
fuels. Experiments need to extend to cyclo-alkanes, aromatics, naphthalenes, and
iso-alkanes with variation in the branched nature. The fuels must also range from

172
C
5
to C
16
that will aid the development of reliable chemical kinetic mechanisms
and jet fuel surrogates.
4. The importance of diffusion coefficients was shown to be significant, but still
very little work has been done at deriving diffusion coefficients especially for
large linear fuel molecules. This is partially due to the difficulty of the
experiments to derive the temperature dependency for large fuels. Classical
techniques should be employed to derive diffusion coefficients for large
hydrocarbons. In addition to classical techniques, it is clear that a series of non-
premixed extinction experiments could be conducted which systematically
increase the sensitivity of the measured flame response to diffusion coefficients.
Locking the chemistry and comparing the simulated extinction results against the
experimental data could optimize the transport parameters. Eventually,
correlations could be derived and used to determine the transport properties of a
large number of fuels.
5. Currently, kinetics mechanism optimization does not include limit flame
phenomena, namely ignition and extinction limits. This is due to both the
difficulty in opposed flow simulations, and the fact that automated optimization
can only be performed on dependent values. The standard opposed flow code
does not have the capability to invert the problem, but modifications to the code
have been performed to allow this. This needs to be taken a step farther and
integrated into the automatic optimization subroutines utilized for other
experimental data. This would require writing appropriate codes that can both

173
identify the turning point behavior and then perform the necessary sensitivity
calculations at that point.
In closing it is important to stress the necessity of continuing this work in other
experimental configurations. Experiments in shock tubes, flow reactors, rapid
compression machines, and burner stabilized flames are traditionally utilized to generate
fundamental reaction rates, and are necessary in the development of the chemical kinetic
mechanism of large molecular weight liquid fuels. Additionally continuous collaboration
between experimentalists and chemical kineticists is necessary for the advancement of
the understanding of the combustion of practical fuels.

174
Bibliography
A. Alexiou, A. Williams, Combust. Flame 104 (1996) 51-65.
M.G. Andac, F.N. Egolfopoulos, Proc. Combust. Inst. 31 (2007) 1165-1172.
J. Andrae, D. Johansson, P. Björnbom, P. Risberg, G. Kalghatgi, Combust. Flame 140
(2005) 267–286.
M. Bergman, V.I. Golovitchev, "Chemical Mechanisms for Modeling HCCI with
Gasoline and Diesel Oil Surrogates", Proceedings of the VII. Congress on Engine
Combustion Processes, pp. 323-334, Munchen, 15-16 March (2005).
J.D. Bittner, J.B. Howard, Proc. Combust. Inst. 18 (1981) 1105-1116.
H. Bohm, H. Jander, D. Tanke, Proc. Combust. Inst. 27 (1998) 1605-1612.
J.P. Botha, D.B. Spalding, Proc. Roy. Soc. Lond. A225 (1954) 71-96.
D. Bradley, R.A. Hicks, M. Lawes, C.G.W. Sheppard, R. Woolley, Combust. Flame 115
(1998) 26–144.
K. Brezinsky, Prog. Ener. Combust. Sci. 12 (1986) 1-24.
K. Brudzewski, A. Kesik, K. Ko łodziejczyk, U. Zborowska, J. Ulaczyk, Fuel 85 (2006)
553-558.
A. Burcat, R.C. Farmer, R.L. Espinoza, R.A. Matula, Combust. Flame 36 (1979) 313-
316.
A. Burcat, C. Snyder, T. Brabbs, “Ignition Delay Times of Benzene and Toluene with
Oxygen in Argon Mixtures,” NASA TM 87312, NASA, 1986.
J. Burri, R. Crockett, R. Hany, D. Rentsch, Fuel 83 (2004) 187-193.
A. Ciajolo, A, D'anna, Combust. Flame 112 (1998) 617-622.
J.A. Cooke, M. Bellucci, M.D. Smooke, A. Gomez, A. Violi, T. Faravelli, E. Ranzi,
Chemical and Physical Processes in Combustion 337-340 (2003).
D.J. Cookson, J.L. Latten, I.M. Shaw, B.E. Smith, Fuel 64 (1985) 509-519.

175
D.J. Cookson, B.E. Smith, I.M. Shaw, Fuel 66 (1987) 758-765.
P. Dagaut, G. Pengloan, A. Restori, Phys. Chem. Chem. Phys. 4 (2002) 1846-1854.
D.F. Davidson, B.M. Gauthier, R.K. Hanson, Proc. Combust. Inst. 30 (2005) 1175-1182.
S.G. Davis, C.K. Law, H. Wang, Proc. Combust. Inst. 27 (1998) 305-312.
S.G. Davis, C.K. Law, Combust. Sci. Technol. 140 (1998) 427-449.
S.G. Davis, C.K. Law Proc. Combust. Inst. 27 (1998) 521-527.
S. Davis, H. Wang. K. Brezinsky, C.K. Law, Proc. Combust. Inst. 26 (1996) 1025-1033.
Defense Energy Support Center, Petroleum Quality Information System, annual fuel
property surveys, http://www.desc.dla.mil/DCM/DCMPage.asp?pageid=99., 1999-
2004.
F. Defoeux, V. Dias, C. Renard, P.J. Van Tiggelen, J. Vandooren, Proc. Combust. Inst.
30 (2005) 1407-1415.
G. Dixon-Lewis, in: J. Warnatz, W. Jäger (Eds.), “Complex Chemical Reaction Systems,
Mathematical Modelling and Simulations,” Springer, Berlin, 1984, p. 265.
Y. Dong, C.M. Vagelopoulos, G. Spedding, F.N. Egolfopoulos, Proc. Combust. Inst. 29
(2002) 1419-1426.
Y. Dong, A.T. Holley, M.G. Andac, F.N. Egolfopoulos, S.G. Davis, P. Middha,
H. Wang, Combust. Flame 142 4 (2005) 374-387.
J.T. Edwards, “USAF Supercritical Hydrocarbon Fuels Interests,” AIAA Aerospace
Sciences Meeting, Reno, Paper 93-0807, Jan. 1993.
F.N. Egolfopoulos, D.X. Du, C.K. Law Proc. Combust. Inst. 24 (1992) 833-841.
F.N. Egolfopoulos, D.X. Du, C.K. Law Combust. Sci. Technol. 83 (1992) 33-75.
F.N. Egolfopoulos, Proc. Combust. Inst. 25 (1994) 1375-1381.
F.N. Egolfopoulos, C.S. Campbell, J. Fluid Mech. 318 (1996) 1-29.
F.N. Egolfopoulos, H. Zhang, Z. Zhang, Combust. Flame 109 (1997) 237-252.
F.N. Egolfopoulos, P.E. Dimotakis, Proc. Combust. Inst. 27 (1998) 641-648.

176
F.N. Egolfopoulos, P.E. Dimotakis, Combust. Sci. Technol. 162 (2001) 19-36.
J.L. Emdee, K. Brezinsky, I. Glassman, J. Phys. Chem. 96 (1992) 2151-2161.
R.A. Eng, C. Fittschen, A. Gebert, P. Hibomvschi, H. Hippler, A.N. Unterreiner, Proc.
Combust. Inst. 27 (1998) 211-218.
A. Ern, V. Giovangigli, Combust. Theor. Model. 2 (1998) 349- 372.
A. Ern, V. Giovangigli, Combust. Sci. Technol. 149 (1999) 157-181.
R.C. Farmer, P.G. Anderson, G.C. Cheng, B.L. Myruski, R.W. Pike, “Propulsion
Chemistry for CFD Applications,” Final Report. On Contract NAS8-40574, National
Aeronautics and Space Administration, Marshall Space Flight Center, AL, Sept.
1997.
F.E. Fendell, J. Fluid Mech. 21 (1965) 281-303.
K. Fieweger, R. Blumenthal, G. Adomeit, Proc. Combust. Inst. 25: 1579–1585 (1994).
K. Fieweger, R. Blumenthal, G. Adomeit, Combust. Flame 109 (1997) 599–619.
A.M. Fincham, G.R. Spedding, Experiments in Fluids 23 (1997) 449-462.
S.L. Fischer, F.L. Dryer, H.J. Curran, International Journal of Chemical Kinetics. 32 (12)
(2000) 713-740.
C.G. Fotache, T.G Kreutz, D.L Zhu, C.K. Law, Combust. Sci. Technol. 109 (1995) 373–
393.
M. Frenklach, S. Taki, R.A. Matula, Combust. Flame 49 (1983) 275-282.
B.M. Gauthier, D.F. Davidson, R.K. Hanson, Combust. Flame 139 (2004) 300–311.
V. Gopalakrishan, J. Abraham, Combust. Flame 136 (4) (2004) 557-566.
J.F. Griffiths, P.A. Halford-Maw, D.J. Rose, Combust. Flame 95 (1993) 291-306.
J.F. Griffiths, P.A. Halford-Maw, C. Mohamed, Combust. Flame 111 (1997) 327-337.
M.P. Halstead, L.J. Kirsch, C.P. Quinn, Combust. Flame 30 (1977) 45-60.
M.P. Halstead, D.B. Pye, C.P. Quinn, Combust. Flame 22 (1974) 89-97.

177
A. Hamins, K. Seshadri, Combust. Flame 64 (1986) 43-54.
A. Hamins, K. Seshadri, Combust. Flame 68 (1987) 295-307.
S. Hayashi, Y. Hisaeda, Y. Asakuma, H. Aoki, T. Miura, H. Yano, Y. Sawa, Combust.
Flame 117 (1999) 851-860.
T.I. Held, F.L. Dryer, International Journal of Chemical Kinetics. 30 (11) (1998) 805-
830.
T.J. Held, A.J. Marchese, F.L. Dryer, Combust. Sci. Technol. 123 (1997) 107-146.
T. Hirasawa, C.J. Sung, A. Joshi, Z. Yang, H. Wang, C.K. Law, Proc. Combust. Inst. 29
(2002) 1427-1434.
A.T. Holley, Y. Dong, F.N. Egolfopoulos, “Ignition and Extinction Studies of Mixtures
of Air with Single-Component Liquid Hydrocarbons, Gasoline, and Gasoline
Surrogates” Proc. Fall Meeting WSS/CI, Stanford CA, 05F-74 (2005).
A.T. Holley, Y. Dong, M.G. Andac, F.N. Egolfopoulos, Combust. Flame 144 (2006) 448-
460.
A.T. Holley, X.Q. You, E. Dames, H. Wang, F.N. Egolfopoulos, “Sensitivity of
Propagation and Extinction of Large Hydrocarbon Flames to Fuel Diffusion,”
accepted for publication at the 32nd International Symposium on Combustion, 2007.
A.T. Holley, Y. Dong, M.G. Andac, F.N. Egolfopoulos, T. Edwards, Proc. Combust. Inst.
31 (2007) 1205-1213.
D. S. Hsu, C.Y. Lin, M.C. Lin, Proc. Combust. Inst. 20 (1984) 623-630.
C. Ji, X.Q. You, A.T. Holley, Y.L. Wang, F.N. Egolfopoulos, H. Wang, “Propagation
and Extinction of Premixed n-Dodecane/Air Flames,” submitted for consideration at
the 32nd International Symposium on Combustion, 2007.
R.J. Johnston, J.T. Farrell, Proc. Combust. Inst. 30 (2005) 217-224.
R.J. Kee, J.A. Miller, G.H. Evans, G. Dixon-Lewis, Proc. Combust. Inst. 22 (1988) 1479-
1494.
R.J. Kee, J.F. Grcar, M.D. Smooke, J.A. Miller, PREMIX: A Fortran Program for
Modeling Steady Laminar One-Dimensional Premixed Flames, Report No. SAND85-
8240, Sandia National Laboratories, 1985.

178
R.J. Kee, F.M. Rupley, J.A. Miller, Chemkin-II: A Fortran Chemical Kinetics Package
for the Analysis of Gas-Phase Chemical Kinetics, Report No. SAND89-8009, Sandia
National Laboratories, 1989.
R.J. Kee, J. Warnatz, J.A. Miller, A FORTRAN Computer Code Package for the
Evaluation of Gas-Phase Viscosities, Conductivities, and Diffusion Coefficients,
Report No. SAND83-8209, Sandia National Laboratories, 1983.
R.D. Kern, C.H. Wu, G.B. Skinner, V.S. Rao, J.H. Kiefer, J.A. Towers, L.J. Mizerka,
Proc. Combust. Inst. 20 (1984) 789-797.
J.H. Kiefer, L.J. Mizerka, M.R. Patel, J. Phys. Chem. 89 (1985) 2013-2019.
S.D. Klotz, K. Brezinsky, I. Glassman, Proc. Combust. Inst. 27 (1998) 337-344.
R. Knikker, A. Dauptain, B. Cuenot, T. Poinsot, Combust. Sci. Technol. 175 (2003) 1783-
1806.
V.G. Knorre, D. Tanke, T.H. Thienel, H.G. Wagner, Proc. Combust. Inst. 26 (1996)
2303-2310.
N. Kosal, A. Bhairi, M.A. Ali, Fuel 69 (1990) 1012-1019.
K. Kumar, C. J. Sung, Combust. Flame 151 (2007) 209-224.
N.A. Lange, Lange’s Handbook of Chemistry, 12
th
ed., McGraw-Hill, New York, 1979.
J.A Langille, Y. Dong, M.G. Andac, F.N. Egolfopoulos, T.T Tsotsis, Combust.
Sci. Technol. 178 (2006) 635-653.
A. Laskin, A. Lifshitz, Proc. Combust. Inst. 26 (1996) 669-675.
C.K. Law, D.L. Zhu, G. Yu, Proc. Combust. Inst. 21 (1986) 1419-1426.
C.K. Law, Proc. Combust. Inst. 22 (1988) 1381-1402.
C. K. Law, C. J. Sung, H. Wang, T. F. Lu, AIAA J. 41 (2003) 1629-1646.
C.K. Law, Combustion Physics, Cambridge University Press, New York, 2006, p. 19.
E.B. Ledesma, M.J. Wornat, P.G. Felton, J.A. Sivo, Proc. Combust. Inst. 30 (2005) 1371-
1379.

179
J. Li, Z. Zhao, A. Kazakov, F.L. Dryer, “An updated comprehensive kinetics model of H
2

combustion,” in Chemical and Physical Processes in Combustion, the 2003 Technical
Meeting of the Eastern States Section of the Combustion Institute, University Park,
PA, October 2003, pp. 169-171.
J. Li, F.L. Dryer, personal communications, 2004.
P.A. Libby, K.N.C. Bray, J.B. Moss, Combust. Flame 34 (1979) 285-301.
A. Liñán, Acta Astronautica, 1 (1974) 1007-1039.
A. Lovell, K. Brezinsky, I. Glassman, Proc. Combust. Inst. 22 (1988) 1063-1074.
W.G. Lovell, Ind. Chem. Engng. 40 (1948) 2388-2438.
X. Lu, W. Chen, Z. Huang, Fuel 84 (2005) 1074–1083.
X. Lu, W. Chen, Z. Huang, Fuel 84 (2005) 1084–1092.
G.A. Mansoori, V.K. Patel, M. Edalat, Int. J. Thermophys. 1 (1980) 285-298.
N.M. Marinov, International Journal of Chemical Kinetics. 31 (3) (1999) 183-220.
M. Matalon, Combust. Sci. Technol. 31 (1983) 169-181.
L.Q. Maurice, H. Lander, T. Edwards, W.E. Harrison III, Fuel 80 (2001) 747-756.
A.G. McLain, C.J. Jachimowski, C.H. Wilson, NASA Tech. Paper, (1979) 1472.
P. Middha, B. Yang, H. Wang, Proc. Combust. Inst. 29 (2002) 1361-1369.
P. Middha, H. Wang, Combust. Theory Model 9 (2005) 353-363.
R. Minetti, M. Ribaucour, M. Carlier, L.R. Sochet, Combust. Sci. Technol. 113-114
(1996) 179-192.
M.K. Mishra, R.A. Yetter, Y. Reuven, H. Rabitz, M.D. Smooke, Int. J. Chem. Kinet. 26
(1994) 437-453.
H. Miyama, J. Phys. Chem. 75 (1971) 1501-1504.
M. Nishioka, C.K. Law, T. Takeno, Combust. Flame 104 (1996) 328-342.
M.A. Oehlschlaeger, D.F. Davidson, R.K. Hanson, Combust. Flame 147 (2006) 195-208.

180
P. H. Paul, J. Warnatz, Proc. Combust. Inst. 27 (1998) 495-504.
N. Peters, Prog. Energy Comb. Sci. 10 (1984) 319-339.
N. Peters, Proc. Combust. Inst. 21 (1986) 1231-1250.
H. Pitsch, N. Peters, K. Seshadri, Proc. Combust. Inst. 26 (1996) 763-771.
A. Roubaud, R. Minetti, L.R. Sochet, Combust. Flame 121 (2000) 535-541.
W.D. Schulz, J. Propulsion and Power Vol. 9, No. 1, 1993, pp. 5-9.
R. Seiser, K. Seshadri, E. Piskernik, A. Linan, Combust. Flame 122 (2000) 339-349.
R. Seiser, H. Pitsch, K. Seshadri, W. Pitz, H .J. Curran, Proc. Combust. Inst. 28 (2000)
2029-2037.
R.A. Shandross, J.P. Longwell, J.B. Howard, Proc. Combust. Inst. 26 (1996) 711-719.
B. Simmons, A. Williams, Combust. Flame 71 (1988) 219-232.
A.P. Singh, S. Mukherji, A.K. Tewari, W.R. Kalsi, A.S. Sarpal, Fuel 82 (2003) 23-33.
H.J. Singh, R.D. Kern, Combust. Flame 54 (1983) 49-59.
R. Sivaramakrishnan, R.S. Tranter, K. Brezinsky, Combust. Flame 139 (2004) 340-350.
R. Sivaramakrishnan, R.S. Tranter, K. Brezinsky, Proc. Combust. Inst. 30 (2005) 1165-
1173.
G.P. Smith, D.M. Golden, M. Frenklach, N.W. Moriarty, B. Eiteneer, M. Goldenberg,
C.T. Bowman, R.K. Hanson, S. Song, W.C. Gardiner, Jr., V. Lissianski, Z.Qin, GRI-
Mech 3.0, http://www.me.berkeley.edu/gri_mech/ (2000).
R.D. Smith, A.L. Johnson, Combust. Flame 51 (1983) 1-22.
R.E. Sonntag, C. Borgnakke, G.J. Van Wylen, Fundamentals of Thermodynamics 5
th
ed.,
John Wiley & Sons, Inc., New York, 1998, p. 388.
C.J. Sun, C.J. Sung, H. Wang, C.K. Law Combust. Flame 107 (1996) 321-335.
S. Tanaka, F. Ayala, J.C. Keck, J.B. Heywood, Combust. Flame 132 (2003) 219-239.
L.S. Tee, S. Gotoh, W.E. Stewart, I&EC Fundam. 5 (1966) 356-363.

181
A. Tregrossi, A. Ciajolo, R. Barbella, Combust. Flame 117 (1999) 553-561.
C.M. Vagelopoulos, F.N. Egolfopoulos, Proc. Combust. Inst. 27 (1998) 513-519.
G. Vanhove, M. Ribaucour, R. Minetti, Proc. Combust. Inst. 30 (2005) 1065-1072.
V. Vasudevan, D.F. Davidson, R.K. Hanson, Proc. Combust. Inst. 30 (2005) 1155-1163.
C.B. Vaughn, J.B. Howard, J.P. Longwell, Combust. Flame 87 (1991) 278-288.
C. Venkat, K. Brezinsky, I. Glassman, Proc. Combust. Inst. 19 (1982) 143-152.
D.J. Vermeer, J.W. Meyer, A.K. Oppenheim, Combust. Flame 18 (1972) 327–336.
A. Violi, S. Fan, E.G. Eddings, A.F. Sarofim, S. Granata, T. Faravelli, E. Ranzi, Comb.
Sci. Technol. 174 (11-12) (2002) 399-417.
T.J. Wallington, E.W. Kaiser, J.T. Farrell, Chem. Soc. Rev. 35 (2006) 335-347.
C.H. Wang, G.J. Ueng, M.S. Tsay, Combust. Flame, 113 (1998) 242-248.
H. Wang, M. Frenklach, Combust. Flame 87 (1991) 365-370.
H. Wang, M. Frenklach, Combust. Flame 96 (1994) 163-170.
H. Wang, A. Laskin, Z.M. Djurisic, C.K. Law, S.G. Davis, D.L. Zhu “A comprehensive
mechanism of C2Hx and C3Hx fuel combustion," in Chemical and Physical
Processes in Combustion, the 1999 Fall Technical Meeting of the Eastern States
Section of the Combustion Institute, Raleigh, NC, October, 1999, pp. 129-132.
H. Wang, Chem. Phys. Lett. 325 (2000) 661-667.
H. Wang, personal communications, 2004.
F.A. Williams, AGARD Conference Proceedings, 164 II 1-1 (1975).
B.A. Williams, Combust. Flame 124 (2001) 330-333.
C.P. Wood, V.G. McDonell, R.A. Smith, G.S. Samuelson, J. Propulsion and Power, Vol.
5, No. 4, 1989, pp. 399-405.
C.K. Wu, C.K. Law, Proc. Combust. Inst. 20 (1984) 1941-1949.

182
Z. Yang, B. Yang, H. Wang, paper 237, “Inference of H-Atom Diffusion Coefficient on
Laminar Flame Simulation,” Proceedings of the Second Joint Meeting of the U.S.
Sections of the Combustion Institute, Berkeley, CA, March 2001.
X.Q. You, F.N. Egolfopoulos, H. Wang, “Detailed and Simplified Kinetic Models of n-
Dodecane Oxidation: The Role of Fuel Cracking in Aliphatic Hydrocarbon
Combustion,” submitted for consideration at the 32nd International Symposium on
Combustion, 2007.
S.P. Zeppieri, S.D. Klotz, F.L. Dryer, Proc. Combust. Inst. 28 (2000) 1587-1595.
Z. Zhao, J. Li, A. Kazakov, F.L. Dryer, Combust. Sci. Technol. 177 (2005) 89-106.

Abstract (if available)

Abstract An experimental and numerical study was conducted on the propagation and ignition and extinction limits of alcohols and liquid hydrocarbon fuels. Experimentally, the extinction and/or ignition limits were determined for a wide range of fuels including: C1-C2 alcohols, C5-C14 n-alkanes, C8 iso-alkanes, samples of gasoline, gasoline surrogates, samples of jet fuels, jet fuel surrogates, and synthetic fuels. Experiments were conducted for both premixed and non-premixed flames, and at ambient as well as elevated temperature. Comparison between the ignition and extinction limits of flames involving the various fuels provided insight into their burning characteristics, their relative performance, and the effect of chemical classification. The insight that was gained will aid in future surrogate development work and will provide direction for the desired composition of practical fuels. Numerically, selected experiments were simulated utilizing quasi-one-dimensional codes, which integrate the conservation equations of mass, momentum, energy, and species with detailed descriptions of molecular transport and chemical kinetics and by invoking a variety of chemical kinetic mechanisms. Through comparison between the numerical simulations and experimental results the adequacy and applicability of existing chemical kinetic mechanisms was assessed. Insight was gained into the fundamental kinetic and transport mechanisms that control flame phenomena of interest. Additionally, the inadequacies in current standard chemical kinetic mechanisms optimization practices were identified, namely the exclusion of flame phenomena other than propagation as a constraint, and the exclusion of transport properties as parameters.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Flame ignition studies of conventional and alternative jet fuels and surrogate components

PDF

Investigations of fuel effects on turbulent premixed jet flames

PDF

Experimental and kinetic modeling studies of flames of H₂, CO, and C₁-C₄ hydrocarbons

PDF

Pressure effects on C₁-C₂ hydrocarbon laminar flames

PDF

Studies of combustion characteristics of heavy hydrocarbons in simple and complex flows

PDF

Studies of methane counterflow flames at low pressures

PDF

Experimental investigation of the propagation and extinction of edge-flames

PDF

Experimental studies of high pressure combustion using spherically expanding flames

PDF

A theoretical study of normal alkane combustion

PDF

CFD design of jet-stirred chambers for turbulent flame and chemical kinetics experiments

PDF

Investigations of fuel and hydrodynamic effects in highly turbulent premixed jet flames

PDF

Modeling investigations of fundamental combustion phenomena in low-dimensional configurations

PDF

End-gas autoignition investigations using confined spherically expanding flames

PDF

Studies of siloxane decomposition in biomethane combustion

PDF

Flame characteristics in quasi-2D channels: stability, rates and scaling

PDF

Dynamics of direct hydrocarbon polymer electrolyte membrane fuel cells

PDF

Kinetic modeling of high-temperature oxidation and pyrolysis of one-ringed aromatic and alkane compounds

PDF

Design, dynamics, and control of miniature catalytic combustion engines and direct propane PEM fuel cells

PDF

Accuracy and feasibility of combustion studies under engine relevant conditions

PDF

Determination of laminar flame speeds under engine relevant conditions

Asset Metadata

Creator Holley, Adam Takashi (author)

Core Title Studies on the flame dynamics and kinetics of alcohols and liquid hydrocarbon fuels

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Mechanical Engineering

Degree Conferral Date 2008-12

Publication Date 12/11/2008

Defense Date 10/14/2008

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag flame extinction,flame ignition,Gasoline,jet fuel,liquid hydrocarbons,molecular diffusion,OAI-PMH Harvest

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Egolfopoulos, Fokion N. (committee chair), Campbell, Charles (committee member), Sadhal, Satwindar S. (committee member), Shing, Katherine S. (committee member), Wang, Hai (committee member)

Creator Email aholley@usc.edu,easytarget239@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m1916

Unique identifier UC197793

Identifier etd-Holley-2516 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-151429 (legacy record id),usctheses-m1916 (legacy record id)

Legacy Identifier etd-Holley-2516.pdf

Dmrecord 151429

Document Type Dissertation

Rights Holley, Adam Takashi

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu