Chemical Kinetic Modeling of Biofuel Combustion by Subram Maniam Sarathy A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Chemical Engineering and Applied Chemistry University of Toronto Copyright c 2010 by Subram Maniam Sarathy
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Chemical Kinetic Modeling of Biofuel Combustion
by
Subram Maniam Sarathy
A thesis submitted in conformity with the requirementsfor the degree of Doctor of Philosophy
Graduate Department of Chemical Engineering and Applied ChemistryUniversity of Toronto
Copyright c⃝ 2010 by Subram Maniam Sarathy
Abstract
Chemical Kinetic Modeling of Biofuel Combustion
Subram Maniam Sarathy
Doctor of Philosophy
Graduate Department of Chemical Engineering and Applied Chemistry
University of Toronto
2010
Bioalcohols, such as bioethanol and biobutanol, are suitable replacements for gaso-
line, while biodiesel can replace petroleum diesel. Improving biofuel engine performance
requires understanding its fundamental combustion properties and the pathways of com-
bustion. This study’s contribution is experimentally validated chemical kinetic com-
bustion mechanisms for biobutanol and biodiesel. Fundamental combustion data and
chemical kinetic mechanisms are presented and discussed to improve our understanding
of biofuel combustion.
The net environmental impact of biobutanol (i.e., n-butanol) has not been studied
extensively, so this study first assesses the sustainability of n-butanol derived from corn.
The results indicate that technical advances in fuel production are required before com-
mercializing biobutanol. The primary contribution of this research is new experimental
data and a novel chemical kinetic mechanism for n-butanol combustion. The results
indicate that under the given experimental conditions, n-butanol is consumed primarily
via abstraction of hydrogen atoms to produce fuel radical molecules, which subsequently
decompose to smaller hydrocarbon and oxygenated species. The hydroxyl moiety in
n-butanol results in the direct production of the oxygenated species such as butanal,
acetaldehyde, and formaldehyde. The formation of these compounds sequesters carbon
from forming soot precursors, but they may introduce other adverse environmental and
health effects.
Biodiesel is a mixture of long chain fatty acid methyl esters derived from fats and
ii
oils. This research study presents high quality experimental data for one large fatty acid
methyl ester, methyl decanoate, and models its combustion using an improved skeletal
mechanism. The results indicate that methyl decanoate is consumed via abstraction
of hydrogen atoms to produce fuel radicals, which ultimately lead to the production of
alkenes. The ester moiety in methyl decanoate leads to the formation of low molecular
weight oxygenated compounds such as carbon monoxide, formaldehyde, and ketene.
The study concludes that the oxygenated molecules in biofuels follow similar combus-
tion pathways to the hydrocarbons in petroleum fuels. The oxygenated moiety’s ability to
sequester carbon from forming soot precursors is highlighted. However, the direct forma-
tion of oxygenated hydrocarbons warrants further investigation into the environmental
and health impacts of practical biofuel combustion systems.
iii
Dedication
Sabbo pajjalito loko, sabbo loko pakampito
The entire universe is nothing but combustion and vibration
To my forefathers and their wives
Sivarajan the Medical Doctor
Viswanathan the Industrialist
Subramaniam the Entrepreneur
Radhakrishnan the Doctor of Engineering
Gurumoorthy the Renunciate
To all beings
May you be happy and peaceful.
May you enjoy good health and harmony.
iv
Acknowledgements
I offer greatest thanks to my wife, Nimisha Rajawat. I would have never been able to
complete this work without her contributions of patience, compassion, encouragement,
love, and delicious food. Thanks for being their and bringing Panya into my life.
Endless gratitude to Professor Murray Thomson for his guidance and support in my
academic and professional endeavours. Thanks for being a wise guide, a passionate
leader, and a good friend.
Much appreciation to Professor Kirk and Professor Wallace for being members on my
PhD supervisory committee. I also thank Professor Mims and Professor Tran for serving
on my examination committee. It was an honour to have Professor Seshadri from UC
San Diego serving as my external examiner. His insights into my work were invaluable.
Thanks to Professor Phillipe Dagaut, Professor Heather MacLean, Dr. Yimin Zhang,
Dr. William Pitz, Professor Tianfeng Lu, and my other coauthors for their efforts.
Gratefulness to all my colleagues in the Combustion Research Group, specifically Dr.
Qingan Zhang, Dr. Seth Dworkin, Dr. Salvador Rego, Professor Zhenyu Wen, Dr.
Jerome Thiebaud, Richard Mills, Phil Geddis, Parham Zabeti, Meghdad Saffaripour,
Tim Chan, and Coleman Yeung. My contemporary, Tom Tzanetakis, deserves a special
recognition for sharing his knowledge of both practical engineering systems and funda-
mental scientific theory. He is the smartest and most humble engineer that I know.
Acknowledgement to NSERC and AUTO21 for funding my research and studies.
My father, Roger Sarathy, deserves special a acknowledgement for proofreading this
dissertation. Also, thanks for always encouraging me to ask questions and to search for
answers.
Besides my wife, two other women deserve special acknowledegment: my mother, Saraswathi
Sarathy, and her sister (i.e., my aunt), Durga Krishnan. Their regular phone calls and
barrage of internet messages always kept me on my toes. Thanks for being there and
v
making sure I never slacked off.
Much respect to Dr. Siva Sarathy, Professor Brinda Sarathy, Brihas Sarathy, Maggie
Decarie, Professor Karthick Ramakrishan, Chef Siddarth Deepak, Dr. David Miller, Wei
Lun Huang, and all my family and friends. Words cannot express my gratitude towards
you, so here is a loving smile :)
Thanks to my Dhamma family at the Ontario Vipassana Centre for supporting my
journey towards fulfilling the ten paramis (i.e., perfect qualities): generosity, moral-
GC/FID gas chromatograph flame ionization detector
GC Gas chromatography
GDP gross domestic product
GHG greenhouse gases
GREET Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation
xvii
GSV gas sampling valve
GUI graphical user interface
HACA hydrogen abstraction carbon addition
ICEs internal combustion engines
ID inner diameter
JSR jet stirred reactor
LCA life cycle assessment
LDV light duty vehicle
LHV lower heating value
LOD limit of detection
MB methyl butanoate
MC methyl trans-2-butenoate
MD methyl decanoate
NDIR non-dispersive infrared
NOx oxides of nitrogen
NTC negative temperature coefficient
OD outer diameter
Pc critical pressure
PM particulate matter
PTFE polytetrafluoroethylene
PTW pump-to-wheel
RME rape seed oil methyl ester
RVP reid vapour pressure equivalent
xviii
SI spark-ignition
Tb boiling point
Tc critical temperature
TCD thermal conductivity detector
THC total unburnt hydrocarbons
VOC volatile organic compounds
WTP well-to-pump
WTW well-to-wheel
xix
Statement of Co-Authorship and Copyright
The material presented in this thesis benefits from collaborations with a number of re-
searchers. In addition, much of the material has been published in academic journals.
This section details the author’s contribution to each section and the right to reproduce
published material without infringing on any copyrights.
All journal articles published from material in this thesis fall under Elsevier’s copyright
policy1. The policy states that the author has the right to include the article in full or
in part in a thesis without obtaining specific permission from Elsevier.
Butanol LCA Studies
The biobutanol life cycle assessment research presented in Chapter 6 was performed in
collaboration with Professor Murray Thomson, Professor Heather MacLean, Professor
Mike Griffin, Yimin Zhang, and Sylvia Sleep. S.M. Sarathy conceptualized the research
and was the primary contributer towards literature survey, developing modeling assump-
tions, life cycle assessment modeling, and writing the manuscripts. Yimin Zhang also
contributed significantly towards literature survey, developing modeling assumptions, life
cycle assessment modeling, and manuscript revisions. Professor Heather MacLean and
Professor Mike Griffin contributed towards developing modeling assumptions, research
supervision, and manuscript revisions. Sylvia Sleep contributed towards literature sur-
vey. Professor Murray Thomson provided financial support to S.M. Sarathy. The above
co-authors have approved the material published in this dissertation.
This study was presented at the following peer reviewed conference proceedings:
1. S.M. Sarathy, Y. Zhang, W.M. Griffin, M.J. Thomson, and H. Maclean, Life Cy-
cle Assessment of Biobutanol for use in Transportation Applications. 8th World
Congress of Chemical Engineering Conference, Montreal, Canada, 2009.
Butanol Combustion Studies
The butanol combustion research presented in Chapter 7 was performed in collaboration
with Professor Murray Thomson, Professor Philippe Dagaut, Dr. C. Togbe, Professor
Christine Rouselle, and Professor Fabien Halter. S.M. Sarathy’s contribution included
1Available online at http://www.elsevier.com/wps/find/authorsview.authors/copyright
xx
literature survey, developing chemical kinetic mechanisms, simulating the jet stirred re-
actor and opposed-flow diffusion flame, acquiring experimental data in the opposed-flow
diffusion flame, and preparing the manuscripts. Professor Murray Thomson contributed
towards financial support to S.M. Sarathy, research supervision, and manuscript revi-
sions. Professor Philippe Dagaut’s contribution was research supervision, developing
chemical kinetic mechanisms, manuscript revisions, and financial support to C. Togbe
who obtained the jet stirred reactor experimental data. Professor Christine Rouselle and
Professor Fabien Halter contributed the laminar flame speed experimental data and sim-
ulations. The above co-authors have approved the material published in this dissertation.
This study was published in the following journals:
1. S.M. Sarathy, M.J. Thomson, C. Togbe, P. Dagaut, F. Halter, C. Mounaim-
Rousselle. An experimental and kinetic modeling study of n-butanol combustion.
Combustion and Flame, 2009, Vol. 156, 852-864.
2. P. Dagaut, S.M. Sarathy, M.J. Thomson. A Chemical Kinetic Study of n-Butanol
Oxidation at Elevated Pressure in a Jet Stirred Reactor. Proceedings of the Com-
bustion Institute, 2009, Vol. 32, 229-237.
Biodiesel Combustion Studies
The biodiesel combustion research presented in Chapter 9 was performed in collabora-
tion with Professor Murray Thomson, Professor Tianfeng Lu, and Doctor William Pitz.
S.M. Sarathy’s contribution included literature survey, developing the modified detailed
chemical kinetic mechanism, performing computer simulations, acquiring experimental
data in the opposed-flow diffusion flame, and preparing the manuscripts. Professor Mur-
ray Thomson contributed towards financial support to S.M. Sarathy, research supervi-
sion, and manuscript revisions. Dr. William Pitz’s contribution was the original methyl
decanoate chemical kinetic mechanism and additional research supervision. Professor
Tianfeng Lu contributed the algorithm used to reduce the modified detailed chemical
kinetic mechanism developed by S.M. Sarathy. The above co-authors have approved the
material published in this dissertation.
This study has been submitted for publication in the following peer-reviewed confer-
ence proceedings:
xxi
1. S.M. Sarathy, M.J. Thomson, T. Lu, W.J. Pitz. An experimental and kinetic
modeling study of methyl decanoate combustion. Proceedings of the Thirty Third
International Combustion Symposium, 2010, Beijing, China.
2. S.M. Sarathy, M.J. Thomson. Chemical Kinetic Modeling of Biodiesel Combustion.
8th World Congress of Chemical Engineering Conference, Montreal, Canada, 2009.
xxii
Part I
Background and Methods
1
Chapter 1
Introduction
The National Resources Canada 2005 report on energy efficiency trends in Canada [1]
indicates that the transportation sector accounts for 30% of total energy use, second only
to the industrial sector. However, transportation produces the largest share of greenhouse
gases (GHG) because the fuels used in transportation are the most GHG intensive. In
addition, Canada spent nearly $61 billion on transportation fuels, the most of any sector.
The transportation sector includes road, air, rail, and marine vehicles; however, the main
source of energy use and GHG was road vehicles used for moving passengers and freight.
From 1990-2005, transportation energy use and GHG increased by more than 30 % due
to an increase in passenger kilometers driven, the consumer shift from cars to minivans
and light trucks, and the increased use of energy intensive modes for transportation.
Although these statistics are for the Canadian economy, the trends are consistent with
other high gross domestic product (GDP) economies.
Road vehicles are powered by internal combustion engines (ICEs) fueled by either
gasoline or diesel. These petroleum derivatives are inherently expensive, and upon
combustion they release large amounts of GHG and other pollutants. Alternatives to
petroleum derived transportation fuels are attractive due to the increasing demand and
limited supply of conventional fossil fuels. Liquid fuels derived from biomass feedstock
(i.e., biofuels) are attracting interest as transportation fuels because they are renewable,
can be locally produced, are more biodegradable, and may reduce net GHG [2]. The
primary driver for using biofuels in the transportation sector is to displace fossil fuel
use. Reducing harmful emissions to the atmosphere is also imperative for mitigating
global warming and sustaining healthy metropolitan areas for human inhabitation. In
addition, the ability for citizens to locally grow and produce their own fuel minimizes the
2
Chapter 1. Introduction 3
dependence on nonrenewable and foreign energy sources.
Currently, gasoline fuel is displaced with bioethanol, while diesel fuel is displaced
with biodiesel. Bioethanol is an alcohol conventionally produced via fermentation of
agriculturally derived starches and sugars. It is blended with gasoline for use in spark-
ignition (SI) engines with only minor modifications required to the engine and fueling
systems. Biodiesel is defined as a mixture of mono-alkyl esters of long chain fatty acids
derived from vegetable oils or animal fats [3]. Biodiesel can be used in its pure form
or it can be blended with petroleum diesel without major modifications to the existing
compression-ignition (CI) engine and fuel distribution infrastructure. In 2007, biofuels
accounted for over 1.5% of global transport fuels [4] with bioethanol and biodiesel con-
tributing an estimated 47 billion liters and 8 billion liters, respectively [5]. The global
demand for biofuels has tripled since 2000, and strong growth is expected in the near
future due to favourable policies from North American and European governments.
A recent review on biofuels provides a unique perspective on the environmental and
societal impacts of biofuels [6]. The rapid policy-driven growth of biofuel use has led to
serious environmental and food security concerns. Current biofuel technologies compete
with the food industry for feedstock, and the diversion of corn, rice, and oilseeds to biofuel
production is cited as the cause of rising food prices and global food shortages. In addi-
tion, large amounts of forest land are being destroyed for biofuel feedstock production,
leading to a loss in biodiversity and carbon-rich sinks. The competition for fresh water
resources presents an additional barrier towards widespread biofuel use. Despite these
challenges, advances in biofuel feedstock and production technologies can ameliorate the
negative impacts of biofuels. Along with environmental stewardship, energy conserva-
tion, efficiency improvements, and other renewable energy technologies (e.g., solar, wind,
geothermal, etc.), biofuels can safely be part of a diverse energy portfolio that reduces
fossil fuel consumption.
1.1 Research Motivation
The rapid increase in biofuel use has sparked an equally rapid growth in research on
biofuel sustainability assessment and combustion properties. Biofuels can either benefit
or harm the environment, so sustainability assessment research attempts to determine the
net environmental impacts associated with biofuel production and use. The most widely
used tool for sustainability assessment is the life cycle assessment (LCA) methodology,
Chapter 1. Introduction 4
which determines the biofuel’s environmental performance based on a set of user-defined
metrics, such as fossil energy input and GHG output. Biofuel combustion research studies
the fundamental combustion properties in order to improve vehicle performance and
minimize harmful emissions. One important combustion tool is the chemical kinetic
mechanism, which describes the molecular level transformation of reactants (i.e., fuel
and air) into products via a series of elementary steps. These mechanisms can be used to
predict ignition properties, heat release rates, amounts of emissions, and the types and
levels of intermediate species [7] in any combustion system.
The broad research questions that this dissertation addresses are:
∙ Do biofuels offer better environmental performance than fossil fuels?
∙ How does biofuel combustion differ from fossil fuel combustion?
LCAs and chemical kinetic mechanisms of bioethanol fuel are already under intense
research, so such work is beyond the scope of this dissertation. Recently, biobutanol has
attracted attention with British Petroleum and DuPont announcing they would begin
selling sugar beet derived butanol as a gasoline blending component in the United King-
dom [8]. This announcement, in combination with reported research and development
advances in biobutanol production, and cited fuel property advantages of biobutanol
compared to bioethanol, suggest that sustainability assessment and combustion research
on biobutanol is warranted.
Biodiesel sustainability assessment research is already well established, but funda-
mental combustion research is limited due to biodiesel’s complex composition. Biodiesel
is typically comprised of a mixture of saturated and unsaturated fatty acid alkyl esters
(i.e., fatty acid methyl esters (FAME)1) with chain lengths ranging from 12 to 18 carbon
atoms. Developing chemical kinetic models for biodiesel has been challenging due to the
large size of the fatty acid alkyl esters found in practical fuels. The added complexity
of varying chain length and degrees of unsaturation has led to the use of surrogate fuels
of well characterized composition for chemical kinetic modeling. Well validated chemical
kinetic mechanisms exist for short chain FAME surrogates; however, there are few vali-
dated chemical kinetic mechanisms for long chain FAME surrogates, so this dissertation
fills the void.
1Biodiesel can also be comprised of ethyl esters, but this study is only concerned with methyl esterssince these constitute the majority of biodiesel in production today.
Chapter 1. Introduction 5
1.2 Dissertation Objectives and Layout
The primary focus of this dissertation is to research the combustion kinetics of biobutanol
and FAME. The study proceeds by performing fundamental combustion experiments and
then using the experimental data to validate chemical kinetic mechanisms for the biofuels.
In addition, since biobutanol is a new biofuel that has not been critically assessed for
sustainability, this dissertation also performs an LCA of biobutanol.
This dissertation is divided in three parts. Part I includes this introduction, as well
as background material relevant to both biobutanol and biodiesel combustion chemistry.
Part II contains all research material related to biobutanol, which includes an LCA of
biobutanol and an experimentally validated chemical kinetic mechanism. Part III is
dedicated to research on biodiesel, which focuses on creating an experimentally validated
chemical kinetic mechanism for long chain FAME. Below is a specific list of objectives
for each part of this dissertation.
Part I Background and Methods
Chapter 1 Present the research motivation and objectives of this dissertation
Chapter 2 Provide a background on combustion chemistry in practical applications
Chapter 3 Discuss the modeling of combustion chemistry
Chapter 4 Describe the experimental methods used for validating chemical kinetic
mechanisms
Part II Biobutanol
Chapter 5 Provide background information related to biobutanol
Chapter 6 Present the LCA of biobutanol
Chapter 7 Present the validated chemical kinetic mechanism for biobutanol combustion
Part III Biodiesel
Chapter 8 Provide background information related biodiesel
Chapter 9 Present the validated chemical kinetic mechanism for biodiesel combustion
Chapter 1. Introduction 6
Part IV Closing
Chapter 10 Summarizes the contributions of this dissertation
Literature Cited
[1] NRCAN, “Energy efficiency trends in canada, 1990 to 2005,” Natural Resources
Canada Office of Energy Efficiency, Tech. Rep., 2009.
[2] A. Demirbas, “Importance of biodiesel as transportation fuel,” Energy Policy, vol. 35,
no. 9, pp. 4661–4670, September 2007.
[3] ASTM, “ASTM D 6751 — specification for biodiesel fuel blend stock (B100) for
middle distillate fuels,” in ASTM Book of Standards. ASTM, 2003.
[4] R. Sims, M. Taylor, J. Saddler, and W. Mabee, “From 1st to 2nd generation biofuel
technologies - an overview of current industry and RD&D activities,” International
Energy Agency and IEA Bioenergy, Tech. Rep., 2008.
[5] REN21, “Renewables 2007 global status report,” Renewable Energy Policy Network
for the 21st Century, Tech. Rep., 2008.
[6] L. P. Koh and J. Ghazoul, “Biofuels, biodiversity, and people: Understanding the
conflicts and finding opportunities,” Biological Conservation, vol. 141, no. 10, pp.
2450–2460, 2008.
[7] C. K. Westbrook, W. J. Pitz, P. R. Westmoreland, F. L. Dryer, M. Chaos, P. Osswald,
K. Kohse-Hoeinghaus, T. A. Cool, J. Wang, B. Yang, N. Hansen, and T. Kasper, “A
detailed chemical kinetic reaction mechanism for oxidation of four small alkyl esters
in laminar premixed flames,” Proceedings of the Combustion Institute, vol. 32, no.
Part 1, pp. 221–228, 2009.
[8] G. Hess, “BP and DuPont to make biobutanol,” Chemical & Engineering News,
vol. 84, pp. 9–10, 2006.
7
Chapter 2
Background
Since the late 1800s when Otto invented the SI engine and Diesel invented the CI engine,
ICEs have become a leading source of stationary and mobile power. ICEs convert the
fuel’s chemical energy into mechanical energy via oxidation (i.e., combustion) within the
engine. Therefore, the reactant fuel-air mixture and the combustion products are the
working fluids of the engine. The difference between the SI and CI engine lies in the
method by which the fuel-air mixture is introduced to the combustion chamber and ig-
nited. The method of ignition determines the key engine characteristics, including the fuel
requirements, operating temperatures and pressures, emission formation mechanisms,
and performance and efficiency [1]. Table 2.1 summarizes several unique characteristics
of SI and CI engines.
The vast majority of ICEs in mobile power applications burn petroleum derived liquid
hydrocarbon fuels. However, early pioneers in the auto industry envisioned the use of
liquid fuels derived from biomass; Henry Ford’s Model T was designed to run on ethanol
while Rudolf Diesel operated his CI engine on peanut oil. This chapter begins with a
brief background of SI and CI engines, fuel properties, and combustion emissions, so that
the reader understands the importance of studying the combustion chemistry of biofuels1.
The following chapters then focuses on combustion kinetics, including its importance, the
development of kinetic mechanisms, and the use of computer simulations as a modeling
tool.
1A thorough description of engine fundamentals is available in Heywood’s text “Internal CombustionEngine Fundamentals” [1].
8
Chapter 2. Background 9
Table 2.1: Selected characteristics of SI and CI engines
SI CI
Ignition Mode spark ignition compression ignition
Combustion Mode premixed nonpremixed
turbulent flame turbulent diffusion flame
Petroleum Fuel gasoline diesel
Biofuel alcohols fatty acid alkyl esters
Compression Ratio 8 to 12 12 to 24
2.1 Reciprocating ICEs
In reciprocating ICEs, the piston moves cyclically up and down in the cylinder chamber
to produce work. The ratio of the maximum cylinder volume (i.e., when the piston is
at the bottom of its stroke) to the minimum cylinder volume is called the compression
ratio. Typically, to generate one power stroke, the piston goes through a four-stroke cycle
which consists of the following [1]:
1. The intake stroke draws fresh mixture into the cylinder chamber by opening the
intake valve and moving the piston from the top of the cylinder to the bottom.
2. During the compression stroke the intake valve closes and the piston moves back
towards the top of the cylinder, compressing the cylinder mixture. Combustion
begins near the end of the compression stroke causing a rapid pressure rise.
3. The expansion stroke occurs as the rapid pressure rise in the cylinder forces the
piston downwards to the bottom of the cylinder chamber. The work generated dur-
ing expansion, which is five times greater than the work used during compression,
turns a crank shaft that delivers power to the vehicle’s wheels.
4. Finally, the exhaust stroke pushes the burned gases out of the cylinder by moving
the piston upwards and opening the exhaust valve. When the piston reaches the
top of the cylinder chamber, the exhaust valve closes and the intake valve opens,
and the four-stroke cycle is reciprocated.
Chapter 2. Background 10
2.1.1 SI Engines
SI engines are characterized by air and fuel being premixed prior to entering the cylinder
chamber through the intake valve. Air-to-fuel ratios in SI engines are typically near
stoichiometric. As the piston moves upwards and the cylinder volume decreases, the
premixed cylinder gas is compressed to 0.8-1.4 MPa (8-14 atm). A typical compression
ratio in SI engines is 8 to 12, which results in less work per stroke when compared to CI
engines [1]. Lower compression ratios are required in SI engines to minimize auto-ignition
(i.e., knocking) of the air-fuel mixture during the compression stroke. Near the top of the
compression stroke, a spark plug ignites the cylinder gases and propagates a turbulent
flame through the cylinder chamber. The rapid pressure and temperature rise forces the
piston downwards through the expansion stroke. Finally, the exhaust stroke forces the
burned gases out of the cylinder and the process is repeated.
2.1.2 CI Engines
In CI engines, only air enters the cylinder during the intake stroke. The cylinder air is
then compressed to approximately 4 MPa (40 atm) and 800 K. A typical compression
ratio in CI engines is 12 to 24, allowing for a greater amount of work to be done during
each cycle [1]. Near the end of the compression stroke, fuel is injected directly into the
cylinder chamber. The liquid fuel jet impinges upon the hot cylinder air and begins to
vaporize. Small pockets of premixed fuel and air then auto-ignite creating additional
heat and radicals sufficient to generate a diffusion flame which consumes the remaining
liquid fuel jet. The exhaust valve opens near the end of the expansion stroke and the
burned gases are exhausted as the piston moves back upwards to its starting position.
2.2 Fuel Properties
The nature of the ignition process in SI and CI engines determines each engine’s fuel
requirements. In SI engines, ignition is initiated by a spark and it is important for the
fuel-air mixture to avoid autoignition. However in CI engines, fuel autoignition is desired
in order to initiate the combustion process during the expansion stroke. ASTM standards
for petroleum derived gasoline [2] and diesel [3] provide specifications for fuel properties
such as density, viscosity, volatility, autoignition characteristics, composition, stability,
and seasonal performance. It is important for biofuels to have properties similar to their
Chapter 2. Background 11
hydrocarbon counterparts, so that major changes are not required to the fueling system
and engine.
A fuel’s ability to resist autoignition in SI engines is known as its antiknocking ten-
dency, and it is quantified by its octane number. The octane number is based on the
fuel’s knocking tendency relative to n-heptane (octane number 0) and iso-octane (2,2,4-
trimethylpentane, octane number 100) [1]. Higher octane numbers indicate a greater
resistance to knock, and are therefore desirable. The octane number of a hydrocarbon
fuel decreases with increasing chain length. Branched hydrocarbons have higher octane
numbers than straight hydrocarbons of the same carbon number because the length of
the basic chain is reduced. SI engine fuels (i.e., gasoline) have octane ratings ranging
from 87 to 105, and mainly consist of straight and branched hydrocarbons of less than
12 carbon atoms with a smaller amount of aromatic species. The addition of oxygenated
compounds, such as ethanol, to hydrocarbon fuel increases the octane number because
oxygenates have greater antiknocking tendencies.
A fuel’s ability to autoignite in CI engines is measured by its cetane number, and it
is inversely related to octane number. The cetane number is based on the fuel’s ability
to autoignite relative to n-hexadecane (cetane number 100) and 1-methylnaphthalene
(cetane number 0). For hydrocarbon fuels, the cetane number increases with increasing
chain length and decreases with branching and cyclication. CI engine fuels (i.e., diesel)
have cetane ratings ranging from 40 to 55 [1], and consist of straight chain hydrocarbons
between 10 and 20 carbon atoms in length with lesser amounts of branched and aromatic
hydrocarbons. Alcohols have low cetane numbers, thereby making them difficult to use
in CI engines. Fatty acid methyl ester species like those found in biodiesel have high
cetane numbers, making them suitable for CI engines.
Besides octane/cetane rating, another important fuel property in SI and CI engines
is the amount of energy per unit of volume. Fuels are sold on a volumetric basis, but
it is energy that powers a vehicle. The energy and volume of a fuel are related through
the density (i.e., kg/L) and the lower heating value (LHV) (i.e., MJ/kg). The LHV is
defined as the amount of energy (i.e., MJ) released during the combustion of a specified
mass of fuel (i.e., kg) at 25 ∘C and returning the combustion products to 25∘C, and then
subtracting the latent heat of vaporization of the water vapor formed during combustion.
Thus, the LHV assumes that water in the combustion products is in the vapor state,
and energy is not recovered by condensing it out of the combustion gas. The volumetric
heating value (i.e., MJ/L) is obtained by multiplying the fuel’s LHV by the fuel’s density,
Chapter 2. Background 12
and it is this value that is important for consumers. Fuels with a larger volumetric heating
value increase fuel economy (i.e., L/100 km) because the amount of energy per unit of
fuel volume is greater. Biofuels tend to have smaller volumetric heating values than their
hydrocarbon counterparts; therefore, it is expected that the fuel economy of a vehicle
powered by biofuel is lower than when powered by petroleum. It should be noted that
some studies have reported an improvement in engine efficiency when using biofuels,
which offsets the lower volumetric heating value, and results in no net change in fuel
economy [4].
2.3 Combustion Emissions
ICEs are a major source of air pollutants and GHG emissions. These emissions have
short-term and long-term health effects on humans, which include irritation of the eyes
and respiratory tract, severe respiratory illnesses, heart disorders, and cancer. Environ-
mental effects include global warming caused by GHG emissions, ozone layer depletion,
acidification, and urban smog formation [5].
Both SI and CI engines emit oxides of nitrogen (NOx), carbon monoxide (CO), par-
ticulate matter (PM), and total unburnt hydrocarbons (THC). These emissions are
regulated by government agencies, so vehicles must meet stringent emission standards.
Combustion in SI engines creates low levels of PM, high levels of CO, THC, and NOx,
but the use of three-way catalytic converters greatly reduces the tail-pipe emissions of
these compounds. CI engine combustion generally produces lower levels of THC and
CO, comparable levels of NOx, and higher levels of PM. Until recently, tail-pipe emis-
sions of CO and THC were reduced using two-way catalytic converters, and little exhaust
treatment was conducted to reduce NOx and PM emissions. However, new environmen-
tal regulations are introducing the use of NOx traps and diesel particulate filters for CI
engines.
Typically, biofuels produce similar amounts of regulated emissions as hydrocarbon
fuels, but there is one notable difference. The use of oxygenated fuels (e.g., biofuels)
has been shown to be an effective way of reducing soot emissions in diesel engines [5, 6].
Oxygenated fuels reduce soot formation by i. sequestering carbon atoms from forming
soot by creating carbon-oxygen bonds, and ii. reducing the aromatic content compared
Chapter 2. Background 13
to petroleum fuels2 . A recent study by Pepiots-Desjardins et al. [7] studied the sooting
tendency of various oxygenated (e.g., alcohols, esters, aldehydes, etc.) and hydrocarbon
compounds, and concluded that oxygenated fuels have soot reducing efficiencies that are
directly related to nature of the functional group.
In addition to the regulated emissions mentioned above, there are several unregulated
emissions that may be of concern. Carbon dioxide is a major cause of global warming
and its emission is currently not regulated by government agencies. Oxygenate emissions,
such as aldehydes and ketones, may become more important when using oxygenated fuels
because the fuel bound oxygen could lead to direct formation of oxygenated compounds.
A study by Jacobson [8] concluded that widespread ethanol use may increase the risk of
cancer and ozone-related illness due to higher aldehyde emissions and increased unburnt
ethanol emissions, which break down to acetaldehyde in the atmosphere. It is uncertain
whether or not biodiesel leads to higher oxygenate emissions since engine studies have
shown both increases and decreases when compared to diesel [4]. Combustion chem-
istry studies of biofuels can help determine the importance of oxygenated emissions by
elucidating the role of fuel bound oxygen during combustion.
The composition of the pollutant gases in the cylinder chamber at the end of the ex-
pansion process varies depending on the engine operating parameters. The concentration
of the pollutants can be calculated assuming chemical equilibrium, as described in the
next section, but these values tend to differ greatly than measured values. This discrep-
ancy is because the combustion products cool rapidly during the expansion process, and
the chemical reactions controlling pollutant formation become rate limited (i.e., they
cannot achieve equilibrium). The pollutant concentrations are essentially “frozen” at
their higher temperature values. Detailed chemical mechanisms and their corresponding
kinetic parameters are required for an accurate calculation of pollutant concentrations
[1]. Therefore, combustion chemistry studies, such as those presented in this thesis, are
required to design engines that curtail pollutant emissions.
2Aromatic hydrocarbons lead directly to soot formation.
Literature Cited
[1] J. Heywood, Internal Combustion Engine Fundamentals. McGraw-Hill Book Com-
pany, New York., 1988.
[2] ASTM, “ASTM D 4814 — standard specification for automotive spark-ignition engine
fuel,” in ASTM Book of Standards. ASTM, 2003.
[3] ASTM, “ASTM D 975 standard specification for diesel fuel oils,” in ASTM Book of
Standards. ASTM, 2003.
[4] M. Lapuerta, O. Armas, and J. Rodriguez-Fernandez, “Effect of biodiesel fuels on
diesel engine emissions,” Progress in Energy and Combustion Science, vol. 34, no. 2,
pp. 198–223, April 2008.
[5] A. Agarwal, “Biofuels (alcohols and biodiesel) applications as fuels for internal com-
bustion engines,” Progress in Energy and Combustion Science, vol. 33, no. 3, pp.
233–271, June 2007.
[6] J. Song, K. Cheenkachorn, J. Wang, J. Perez, A. L. Boehman, P. J. Young, and F. J.
Waller, “Effect of oxygenated fuel on combustion and emissions in a light-duty turbo
diesel engine,” Energy & Fuels, vol. 16, no. 2, pp. 294 – 301, 2002.
[7] P. Pepiot-Desjardins, H. Pitsch, R. Malhotra, S. Kirby, and A. Boehman, “Structural
group analysis for soot reduction tendency of oxygenated fuels,” Combustion and
Flame, vol. 154, pp. 191–208, 2008.
[8] M. Z. Jacobson, “Effects of ethanol E85 versus gasoline vehicles on cancer and mor-
tality in the united states,” Environmental Science and Technology, vol. 41, no. 11,
pp. 4150–4157, June 2007.
14
Chapter 3
Modeling Combustion Chemistry
Combustion in ICEs is a complex process involving fuel atomization, vaporization, fuel-air
mixing, ignition, and combustion. For example, in a CI engine liquid fuel is injected as a
high velocity spray into the combustion chamber, where it vaporizes upon impingement
with high-temperature high-pressure cylinder gases. Low temperature reactions then
spontaneously ignite portions of premixed fuel and air causing rapid heat release. The
remaining fuel spray is then consumed in a high temperature diffusion flame, and burned
gases are produced through the entire expansion process. This unsteady, heterogeneous,
3-dimensional process is challenging to model, and it is difficult to decouple mixing
processes from chemical kinetic processes [1].
Computer simulations based on the KIVA code [2] are capable of combining fluid
dynamics, spray dynamics, chemically reacting flows, and heat and mass transfer in
an engine cylinder to predict ignition behavior, pollutant formation, energy release, and
other features of engine operation. Such codes are widely used in the automobile industry
to increase fuel economy and reduce emissions. Typically, these engine simulations are
computationally expensive, so simplifications to fluid dynamics, spray dynamics, and
elementary chemical kinetics are required. However, reducing the chemical kinetic model
(i.e., mechanism) reduces chemical fidelity and limits our ability to fully understand
combustion chemistry. This chapter describes how detailed chemistry is modeled in
idealized combustion systems, so that the effects of molecular structure on combustion
and emissions can be understood.
15
Chapter 3. Modeling Combustion Chemistry 16
3.1 Chemical Kinetics
Many processes in the engine, including reactions in the flame zone which determine heat
release, reactions controlling ignition, and air pollutant formation mechanisms, occur
at times when temperature and pressure are changing rapidly. These nonequilibrium
processes depend on the rate of each individual chemical reaction (i.e., reaction kinetics),
which are governed by the temperature and the concentration of reactants.
The rates at which reactant species are consumed and product species are produced in
a kinetically controlled process is governed by the law of mass action. For the elementary
reaction in Equation 3.1, the law of mass action states that the rate at which reactants
are consumed is proportional to the product of concentration of each reactant raised to its
stoichiometric coefficient, as shown in Equation 3.2. The forward reaction rate constant
(kf ) shown in the equation follows the Arrhenius form and is further discussed in a later
section.
aA+ bB = cC + dD (3.1)
−dAdt
= kf [A]a[B]b (3.2)
A comprehensive list of chemical reactions and their rates (i.e., a chemical kinetic
mechanism) is required to accurately predict the rate of energy release, soot and pollu-
tant formation, ignition behaviour, knocking limits, and cool flame characteristics [3, 4, 5].
Rather than attempting to validate the detailed kinetic model in an engine, a better op-
tion is to study combustion chemistry and flame structure in idealized chemically react-
ing flow systems (e.g., an opposed-flow diffusion flame) [6]. The combustion phenomenon
observed in the laboratory experiment can then be used to validate a chemical kinetic
mechanism and understand combustion performance in an engine. Furthermore, com-
prehensive chemical kinetic mechanisms validated against a wide range of experimental
data provide the foundation for the reduced mechanisms used in engine simulations [7].
3.2 Computer Simulations for Mechanism Validation
Chemical kinetic mechanisms describe the molecular level transformation of reactants
(i.e., fuel and air) into products via a series of elementary steps. The mechanism valida-
tion first requires a model describing the geometry and operating regime of the specific
Chapter 3. Modeling Combustion Chemistry 17
combustion application. A large number of differential equations describing the mass,
momentum, energy, and species concentration are numerically integrated to generate
concentration profiles for reactants, intermediates, and products [8]. The computed pro-
files are then validated against experimental data from one or more well-characterized
combustion apparatuses.
A chemical kinetic mechanism is typically validated against an idealized chemically
reacting flow system. The experimental setups modeled in the present study included
laminar flame speed). This section describes the governing equations used for modeling
chemically reacting flow systems. Numerical modeling is not the focus of this dissertation
study, so a complete derivation of equations for the various combustion systems (e.g.,
opposed-flow diffusion flames, jet stirred reactors, and premixed laminar flames) and their
solution methodology is not presented herein. The reader is directed to the “CHEMKIN
Theory Manual” for further elaboration on the specific computer codes used for numerical
modeling [9].
3.2.1 Governing Equations for Chemically Reacting Flows
Chemically reacting flow problems are mathematically formulated using equations for
conservation of mass, momentum, energy, and concentration of chemical species, along
with thermodynamic relationships [5, 6, 10]. The chemical kinetic mechanism couples
chemical species concentrations with the energy equation via the enthalpy of reaction.
A set of ordinary differential equations for species and energy, with time as the indepen-
dent variable, make up the conservations equations for problems where spatial transport
is negligible (e.g., plug flow reactors, perfectly stirred reactors, etc.). When transport pro-
cesses are important (e.g., laminar flames), the conservation equations becomes a partial
differential equations, with time and space as the independent variables. The compu-
tational cost for kinetically controlled problems is small, but when transport processes
are included the computational load increases dramatically [5]. The following sections
discuss the governing equations for modeling chemically reacting flow systems comprised
of laminar gaseous flows.
Chapter 3. Modeling Combustion Chemistry 18
Conservation of Mass
Mass is always conserved in a system, and therefore, in a steady state process the rate at
which mass enters a differential element (i.e., a point in space) is equal to the rate at which
it leaves the element. In fluid mechanics, the conservation of mass is mathematically
formulated using the continuity equation shown in Equation 3.3.
∂�
∂t+∇ ⋅ (�v) = 0 (3.3)
where
� is the fluid density
t is the time
v is the fluid velocity vector
∇() is the divergence operator
The equation indicates that the rate of change in mass with respect to time in a
differential element, ∂�∂t
, plus the net mass flow into and out of that element, ∇(�V), is
zero.
Conservation of Momentum
The conservation of momentum (i.e., Navier-Stokes equations) together with the conti-
nuity equation for conservation of mass are the fundamental formulations in fluid me-
chanics. The general differential form of the Navier-Stokes equations where gravity is the
only acting force is shown in Equation 3.4.
�∂v∂t
+�v ⋅ ∇v = fsurface + fbody = ∇¯� −∇p+ �g (3.4)
where
� is the fluid density
t is the time
v is the fluid velocity vector
p is the pressure representing a surface force, fsurface
¯� is the viscous stress tensor representing a surface force, fsurface
g is the gravitational force constant representing a body force, fbody
Chapter 3. Modeling Combustion Chemistry 19
The equation states that the change in momentum with respect to time in a differential
element, �∂v∂t
, plus the contribution of convection on momentum, �V ⋅∇v, is equal to the
contribution of viscous stress on momentum, ∇¯� , minus the contribution of pressure on
momentum, ∇p, plus the contribution of gravitational forces on momentum, �g.
Conservation of Species
The continuity equation presented above defines mass conservation in a fluid flow, but it
does not provide any distinction on the chemical species present in the flow. However, the
mass conservation of individual species is important in chemically reacting flow systems
consisting of a multicomponent gaseous mixture. The mass fraction of an individual
species is shown in Equation 3.5.
Yk =�k�
(3.5)
where
Yk is the mass fraction of the ktℎ species, andK∑k=1
Yk = 1
� is the total fluid density
�k is mass density of the ktℎ species, andK∑k=1
�k = �
The chemical composition of a gaseous mixtures in a differential element can be
derived from species mass conservation equations. The mass conservation of a the ktℎ
species in an element is altered by homogeneous chemical reactions, molecular diffusion,
and convection, as shown in Equations 3.6
�∂Yk∂t
+ �v ⋅ ∇Yk = !kWk −∇Jk (3.6)
where
� is the fluid density
t is the time
v is the fluid velocity vector
Yk is the mass fraction of the ktℎ species
Jk is the diffusive mass flux vector of the ktℎ species
!k is the net molar production rate of the ktℎ species
Wk is the molecular weight of the ktℎ species
Chapter 3. Modeling Combustion Chemistry 20
The equation states that the change in concentration (i.e., mass fraction) of the ktℎ
species with respect to time in a differential element, �∂Yk∂t
, plus the contribution of
convection on concentration, �v ⋅∇Yk, is equal to the contributions of chemical reactions
on concentration, !Wk, minus the contribution of molecular diffusivity on concentration,
∇Jk.The diffusivity (i.e., diffusion mass flux) of a species, jk, can be described using Fick’s
law as shown in Equation 3.7. The theory states that the diffusivity depends linearly on
the negative concentration gradient multiplied by the binary diffusion coefficient, Dk.
Jk = −� YkXk
Dk∇Xk = −�Wk
WDk∇Xk (3.7)
where
Jk is the diffusive mass flux vector of the ktℎ species
� is the fluid density
Yk is the mass fraction of the ktℎ species
Xk is the mole fraction of the ktℎ species
Dk is the binary diffusion coefficient
Wk is the molecular weight of the ktℎ species
W is the mean molecular weight of the mixture
Conservation of Energy
Thermal energy is conserved in chemically reacting flow systems, and the energy equation
is used as the basis for such systems. The energy equation is used to describe the
temperature profile of a chemically reacting flow, which affects processes such as chemical
reaction, convection, and molecular diffusion. The thermal energy equation, shown in
Equation 3.8, stems from the first law of thermodynamics, and assumes ideal gases, low
Mach numbers. and Fourier’s law for heat conduction.
�cp∂T
∂t+ �cpvk ⋅ ∇T = ∇ ⋅ (�∇T )− �
K∑k=1
cp,kYkvk ⋅ ∇T −K∑k=1
ℎk!kWk + qrad (3.8)
where
cp is the constant pressure heat capacity of the ktℎ species
� is the fluid density
Chapter 3. Modeling Combustion Chemistry 21
t is time
T is the temperature
� is the thermal conductivity
Yk is the mass fraction of the ktℎ species
vk is the fluid velocity vector of the ktℎ species
!k is the net molar production rate of the ktℎ species
ℎk is the enthalpy of formation of the ktℎ species
Wk is the molecular weight of the ktℎ species
qrad is the radiative heat transfer
The equation states that the change in thermal energy with respect to time in a
differential element, �cp∂T∂t
, plus the thermal energy convected to the element by the
temperature gradient, �cpv ⋅ ∇T , is equal to the contribution of thermal heat conduc-
tion (i.e., Fourier’s law) on thermal energy, ∇(�∇T ), minus the contribution of thermal
diffusivity on thermal energy, �K∑k=1
cp,kYkvk ⋅ ∇T , minus the contribution of heat from
chemical reaction on thermal energy,K∑k=1
ℎk!kWk, plus the contribution of radiative heat
transfer on the the element, qrad.
3.3 Solving the Governing Equations
The governing equations can be solved using a numerical solver that evaluates the chemi-
cal kinetic, thermodynamic, and transport properties in each differential element as time
proceeds. This study models chemically reacting flow systems using the CHEMKIN
software package [11]. CHEMKIN provides modeling of a wide range of combustion ap-
paratuses, including shock tubes, premixed flames, diffusion flames, and partially and
perfectly stirred reactors. Chemical kinetic mechanisms are coupled with thermochem-
ical data for all the species in the mechanism to calculate forward and reverse reaction
rates. Transport properties for the species are also included when attempting to model a
combustion process in which transport processes are rate-controlling (e.g., opposed-flow
diffusion flames).
The combustion setups were modeled using the CHEMKIN 4.1 software package. The
first step used the CHEMKIN 4.1 graphical user interface (GUI) to set up a diagram of
the experimental apparatus, including all reactant and product streams. The next step
Chapter 3. Modeling Combustion Chemistry 22
was to generate the linking files for the numerical code. This required using the pre-
processors to access three important information files: i.) the chemical kinetic database;
ii.) the thermodynamic database; and iii.) the transport database. More information
on these files is given below. The “CHEMKIN Gas-Phase Interpreter” reads the first
two files and generates the “CHEMKIN Linking File”. The third file is used by the
“TRANSPORT Preprocessor” to generate the “Transport Linking File” when modeling
systems where transport processes are important.
Next, the characteristics of the chemically reacting flow system and inlet flows were
input. This includes the velocity of each inlet stream, initial concentrations, pressure,
physical configuration, temperature, and a number of solution method options. The
model was run, and the numerical simulation output a text file containing the solution.
The “Solution Export Utility” was used to convert this text file into a comma separated
values file format that was readable by Microsoft Excel.
This section discuss the the three input files required by CHEMKIN for solving the
chemically reaction flow system problem. The development of these input files is the
primary focus of this dissertation, so a thorough background is presented for the reader
to appreciate this study’s contributions.
3.3.1 Chemical Kinetic Database
The chemical kinetic database identifies all the gaseous species present, and it provides
a user-defined chemical kinetic mechanism for the production and consumption of these
species. The chemical kinetic mechanism details each reaction taking place and the ap-
propriate reaction rate parameters in the modified Arrhenius form, as shown in Equation
3.12. The gas phase kinetic file conforms to the CHEMKIN input format. Additional
information of the development of chemical kinetic mechanisms is provided in Section
3.4.
The chemical source term, !k, describes the net molar production rate of the ktℎ
species, and it appears in Equations 3.6 and 3.8. The chemical kinetic mechanism con-
tains the information needed to evaluate !k. The chemical system consisting of N species
and M reversible reactions can be expressed as
N∑k=1
� ′nk[Xk] ⇀↽N∑k=1
� ′′nk[Xk] (3.9)
where
Chapter 3. Modeling Combustion Chemistry 23
� ′nk is the stoichiometric coefficient of the ktℎ reactant species in the ntℎ
reaction
� ′′nk is the stoichiometric coefficient of the ktℎ product species in the ntℎ
reaction
[Xk] is the molar concentration of the ktℎ species
The molar production of the ktℎ species, !k, is expressed as
!k =M∑k=1
(� ′′nk − � ′nk)�n (3.10)
where
�n is the progress variable of the ntℎ reaction, given in Equation 3.11
�n = kfn
N∏k=1
[Xk]�′nk − krn
N∏k=1
[Xk]�′′nk (3.11)
Each ntℎ reversible reaction is characterized by a forward reaction rate, kfn following
the Arrhenius form shown in Equation 3.12. The reverse reaction rate constant, krn, is
calculated from thermochemistry.
kfn = A ⋅ (T )n exp−EaR ⋅ T
(3.12)
where
kfn is the reaction rate constant
A is the pre-exponential collision frequency factor in cm3
mol⋅s
T is temperature in kelvin
n is the temperature dependence factor
Ea is the activation energy in calmol
R is the ideal gas constant in calmol⋅K
The reaction rate constant, kfn, in Equation 3.12 depends on the temperature, activa-
tion energy, and the collision frequency factor. A temperature dependence term is incor-
porated into the equation (i.e., n ∕= 0) for reactions that exhibit non-Arrhenius behaviour
over the range of temperatures encountered in combustion. Typically, experiments are
conducted to determine the coefficients in the rate equation; however estimations and
calculations using ab initio quantum chemistry methods are also employed. This study
Chapter 3. Modeling Combustion Chemistry 24
uses, wherever possible, rate coefficients from published experimental and computational
studies available through the “NIST Chemical Kinetics Database” [12]. For reactions
that have not been studied previously, this study estimates rate coefficients heuristically.
3.3.2 Thermochemical Database
Thermochemical data for each species in the chemical kinetic mechanism are required for
CHEMKIN to calculate thermodynamic properties, thermal transport properties, and
reaction equilibrium constants. Contained within the thermochemical data file are the
species’ name, elemental composition, electronic charge, and phase. In addition, fourteen
polynomial fitting coefficients are provided to calculate the constant pressure molar heat
capacity (C0p), molar enthalpy of formation (ΔfH
0), and molar entropy of formation
(S0p) at any temperature. From these calculated properties, CHEMKIN can calculate
other important thermochemical properties, such as the constant volume heat capacity,
internal energy, Gibb’s free energy, and Hemholtz free energy. Mass based properties are
generated by dividing the property in molar units by the molecular weight [9].
Both computational and experimental methods can be used to determine thermo-
chemical data properties, and the “NIST Chemistry WebBook” has a good compilation
of previously published data [13]. For larger molecules which have not been experimen-
tally studied and for which quantum chemical calculations are computationally expensive,
thermochemical properties can be estimated based on group additivity methods. Ben-
son [14] has proposed a systematic way of estimating thermochemical properties for a
molecule from data on the bonded atomic groups which comprise it. The additivity
law determines the property X of a complex molecule by adding the tabulated bond
properties for simple molecules (e.g.,∑Xmolecule =
∑Xbonds). The THERGAS [15] and
THERM [16] softwares, which is based on Benson’s method, are used in the present study
to estimate thermochemical properties of new species. The user inputs the structure of
the molecule in a specified format, and the computer determines its thermochemical
properties by adding together tabulated bond properties.
3.3.3 Transport Database
Combustion is typically a combination of chemical kinetic driven processes (i.e., produc-
tion and destruction of species) and transport driven processes (i.e., convection, diffusion,
and conduction). In certain combustion applications, such as a perfectly stirred reactor or
Chapter 3. Modeling Combustion Chemistry 25
plug-flow reactor, the overall rate is assumed to be kinetically controlled since the trans-
port processes occur infinitely fast. However in other cases, such as laminar and diffusion
flames, the transport processes are rate-controlling. Therefore, the molecular transport
of species, momentum, and energy in the gas mixture must be evaluated from the dif-
fusion coefficients, viscosities, thermal conductivities, and thermal diffusion coefficients.
CHEMKIN determines these temperature and pressure dependent flow properties of each
individual species using standard kinetic theory expressions, and then determines the gas
mixture properties using mixture averaging rules. Note, in some cases CHEMKIN sub-
stitutes the mixture-averaged approach with a multicomponent approach to determine
the transport properties of the gas mixture. The interested reader is referred to the
“CHEMKIN Theory Manual” for further elaboration on the methodology and the ex-
pressions used in determining the flow properties of individual species and gas mixtures
[9].
In this study, priority is placed on determining the molecular transport parameters for
each species in the gas mixture, such that the CHEMKIN can determine the flow prop-
erties using its standard kinetic theory expressions. The molecular transport parameters
are inputted into CHEMKIN via a specified data format, as follows, in order:
1. An index indicating the geometrical configuration of the molecule. If the index is
0, then the molecule is a monoatomic. If the index is 1, then the molecule is linear.
If the index is 2, then the molecule is nonlinear.
2. The Lennard-Jones potential well depth, �/kb, in Kelvin.
3. The Lennard-Jones collision diameter, �, in angstroms.
4. The dipole moment, �, in Debyes (10−18cm3/2ergs1/2).
5. The polarizability, �, in cubic angstroms
6. The rotational relaxation collision number, Zrot, at 298 K.
The molecular transport parameters can be obtained from a variety of sources, and
CHEMKIN itself contains a transport database of over 200 species. However, more
species are often required when dealing with new fuels. The modeler can then turn to
previous modeling work to search for transport parameters. Alternatively, much data can
be found in standard reference texts such as “Molecular Theory of Gases and Liquids”
Chapter 3. Modeling Combustion Chemistry 26
[17] or other chemistry handbooks. If the data is still not found, then estimation and
analogy with related molecules can be used.
For new molecules, the Lennard-Jones collision diameter and potential well depth can
be estimated via different methods. Svehla describes how these parameters are obtained
by computing “best fits” to experimental data of a macroscopic transport property (e.g.,
viscosity) [18]. When experimental data is not available, as is the case for most gases,
these parameters are estimated using a variety of techniques. Svehla describes how
they can be calculated using the physical-chemical properties (e.g., boiling point, molar
volume, etc.), several empirical or combining rules, or other theoretical relations [18].
This study uses the correlations developed by Tee, Gotoh, and Stewart [19] and described
in Wang and Frenklach [20]. The correlations allow for the calculation of the Lennard-
Jones collision diameter and potential well depth using the critical pressure (Pc) and
critical temperature (Tc) of the gas. The critical temperature of a substance is the
temperature at and above which separate gas and liquid phases do not exist, and only
the supercritical state exists. The critical pressure is the vapor pressure at the critical
point (refer to Figure 3.1).
Figure 3.1: A typical phase diagram showing critical point
The equations for calculating the Lennard-Jones collision diameter and potential well
depth are given below. The accentric factor ! in Equations 3.13 and 3.14 is evaluated
Chapter 3. Modeling Combustion Chemistry 27
using the Lee-Kesler vapor-pressure relations shown in Equation 3.15 [21]. The critical
temperature, critical pressure, and boiling point (Tb) used in the calculations are obtained
from the “NIST Chemistry WebBook” [13]. If Pc, Tc, and (Tb) are not readily available
for the species, they can be approximated from species with similar molecular structures.
�(PcTc
)1/3
= 2.3551− 0.0874! (3.13)
��
kbTc= 0.7915 + 0.1693! (3.14)
! =−ln(Pc)− 5.927 + 6.096Tb
Tc
−1+ 1.289lnTb
Tc− 0.169Tb
Tc
6
15.252− 15.688TbTc
−1 − 13.472lnTbTc
+ 0.436TbTc
6 (3.15)
The dipole moment (�) is a measure of the extent of polarity in covalent molecules.
It is dependent on the difference electronegativity of the bonding atoms, and is precisely
defined as the product of the magnitude of the charge and the distance between the
charges. Nonpolar compounds, such as fully saturated hydrocarbons have zero dipole
moments while oxygenated compounds display higher dipole moments. Many experi-
mentally measured dipole moments are available in McClellan’s “Tables of Experimental
Dipole Moments” [22]. If experimental data is not available then the the molecular dipole
moment, which is a vector property, can be calculated for using vector addition of known
bond moments [23]. Such a method requires detailed information of the geometry of
molecular bonds and their electronegativities.
The polarizability (�) of a molecule quantifies the tendency of a molecules charge
distribution (i.e., electron cloud) to be distorted from its normal shape by an external
electric field (e.g., a nearby dipole or ion). Experimentally measured polarizability values
in cubic Angstroms can be obtained from the “CRC Handbook of Chemistry and Physics”
[24]. Bosque and Sales [25] have presented an empirical additive formula that allows the
estimation of polarizability from the molecular formula (i.e., # of C, H, and O atoms),
as shown in Equation 3.16.
� = 0.32 + 1.51 ∗#C + 0.17 ∗#H + 0.51 ∗#O (3.16)
Chapter 3. Modeling Combustion Chemistry 28
3.4 Developing Chemical Kinetic Mechanisms
The process for developing and validating chemical kinetic mechanism was outlined by
Frenklach et al. [26] and summarized by Simmie [8]. Figure 3.2 is a flowchart of the
mechanism development process. This flowchart indicates that developing and validat-
ing a mechanism is a continuously evolving process wherein experiments and modeling
symbiotically achieve a satisfactory mechanism. The process can be summarized as fol-
lows:
1. Generate a list of elementary reactions.
2. Determine reaction rate constants for each reaction using literature sources or es-
timation, paying attention to temperature and pressure dependencies. Provide
thermochemical data to calculate equilibrium reverse rate constants.
3. Conduct controlled experiments that can be used to validate the reactions and rate
parameters given in the model.
4. Solve the reaction mechanism kinetics and transport equations using a computer
simulation of the experimental configuration. Conduct a sensitivity analysis to
determine the impact of specified rate constants on the final result.
5. Compare the experimental data to the model predicted values. Optimize reac-
tion rate parameters that have the greatest impact on fitting desired experimental
values.
The comprehensiveness of a mechanism is measured by its ability to describe com-
bustion phenomenon extensively. A mechanism is not considered comprehensive if it has
been tested against a single experiment because the role of each elementary reaction varies
with temperature, pressure, and composition. For example, reactions between hydrogen
atoms and fuel molecules are dominant in fuel-rich conditions, while reactions between
hydroxyl radicals and fuel molecules dominate in fuel-lean conditions. Many reactions
are important only at low temperatures, while others are dominant at high tempera-
tures. In an early treatise on chemical kinetic mechanisms for hydrocarbon combustion,
Westbrook [5] explains that a comprehensive mechanism must be validated against ex-
perimental data covering chemically reacting flows at various temperatures, pressures,
and reactant compositions.
Chapter 3. Modeling Combustion Chemistry 29
Determine thermo-
chemical & transport properties
Conduct controlled validation
experiments
Simulate experiments
using computer
model
Compare experiments
with computer simulation
Formulate elementary reaction set
and determine reaction rates
Figure 3.2: Flowchart for developing and validating chemical kinetic mechanisms
Autoignition characteristics are typically studied in shock tubes or rapid compression
machines, while reactions in a flameless premixed environment are studied in a jet stirred
reactor (i.e., perfectly stirred reactor). Laminar premixed and non-premixed flames are
often used to study combustion kinetics occurring at high temperatures. Each apparatus
can be operated at various temperature and pressure regimes to determine the effects of
these parameters on kinetic processes.
Since combustion of hydrocarbon fuels consists of sequential fragmentation of fuel
molecules into intermediates species, a comprehensive mechanism for any fuel must con-
tain detailed sub-mechanisms for the fuel’s intermediates. For example, since hydrogen
and carbon monoxide are products of hydrocarbon combustion, any hydrocarbon mecha-
nism must include reaction sub-mechanisms for hydrogen and carbon monoxide [7]. This
observation allows for the systematic and hierarchical development of kinetic mechanisms
by sequentially incorporating new species and reaction schemes in order of increasing
complexity [5].
Chapter 3. Modeling Combustion Chemistry 30
3.4.1 Mechanisms for Hydrocarbon Fuels
Chemical kinetic mechanisms for hydrocarbon fuels have been the focus of intense re-
search for several decades. Chemical kinetic mechanisms for biofuels, such as ethanol
and biodiesel, have only received attention recently. This section provides a background
on the combustion pathways of alkanes and alkenes because the same reaction types and
classes can be applied to biofuel combustion. In addition, detailed reaction rate stud-
ies for hydrocarbon fuels can be used to determine rate constants for similar reactions
for which no detailed reaction rate studies exist. Readers that are interested in a de-
tailed discussion of hydrocarbon combustion are directed towards recent review articles
by Battin-Leclerc [4] and Simmie [8].
A clear and simple explanation of the oxidation of fuels is given by Glassman [27].
Combustion reactions are driven by the formation of highly reactive radical, such as O,
OH, and H. During combustion, fuels are oxidized by a series of chain reactions which
can be categorized as one of following:
1. chain initiating,
2. chain propagating and chain branching,
3. chain terminating.
Chain initiating occurs when radical species are produced by dissociation of the re-
actants. The chain is propagated and branched as radicals react with stable compounds
to form additional radical species. Finally, the chain terminates when two radicals re-
combine to form stable species. The following subsections describe the major reaction
pathways for the combustion of alkanes and alkenes under low and high temperature
conditions. Comprehensive mechanisms would contain a number of minor reactions;
however, for the sake of simplicity, they have not been included here.
Combustion of Alkanes
Low Temperature Combustion of Alkanes
The combustion of hydrocarbons is different at low temperatures than high tempera-
tures. The general mechanism for the low temperature combustion of hydrocarbons was
developed by Semenov [28]. Benson [29] introduced the isomerization reaction of large
Chapter 3. Modeling Combustion Chemistry 31
hydrocarbons (Equation 3.20) to the Semenov mechanism. The following is a simplified
form of the Semenov mechanism including isomerization of large hydrocarbons:
RH +O2 → R ⋅+HO2⋅ (3.17)
R ⋅+O2 → alkene+HO2⋅ (3.18)
R ⋅+O2 → RO2⋅ (3.19)
RO2⋅ → ROOH (3.20)
ROOH +O2 → RO ⋅+OH⋅ (3.21)
HO2 ⋅+RH → H2O2 +R⋅ (3.22)
H2O2 +M → OH ⋅+OH ⋅+M (3.23)
The chain is initiated by low temperature combustion of the hydrocarbon (RH) to
form an alkyl radical (R⋅) and a hydroperoxy radical (HO2⋅), as shown in Equation 3.17.
Next, the chain is propagated by one of two parallel reactions between alkyl radicals
and oxygen to form an alkene, HO2⋅, and RO2⋅ (Equations 3.18 and 3.19). These reac-
tions compete with each other depending on the temperature. At temperatures above
500 K, Equation 3.18 predominates, wherein the oxygen abstracts a hydrogen from the
alkyl radical to form an alkene and a hydroperoxy radical. At temperatures below 500
K, Equation 3.19 is favored, wherein the oxygen adds to the alkyl radical to form an
alkylperoxy radical.
At low temperatures (below 500 K), propagation continues by the isomerization of
RO2⋅ to produce peroxide species (ROOH) (Equation 3.20). The radical pool then builds
up by degenerate branching of ROOH to form RO⋅ and OH⋅ radicals (refer to Equation
3.21). Further developments on this low temperature mechanism have been published by
Zhao et al. [30].
At intermediate temperatures (above 500 K), the HO2⋅ radical is more abundant,
so the reaction is propagated by hydrogen abstraction on the hydrocarbon by HO2⋅to form hydrogen peroxide (H2O2) and an alkyl radical (refer to Equation 3.22). As
the temperature increases, hydrogen peroxide decomposes to form two hydroxyl radicals
(refer to Equation 3.23). The fuel-air mixture explodes once the radical pool builds up,
and then high temperature combustion predominates.
Chapter 3. Modeling Combustion Chemistry 32
Intermediate and High Temperature Combustion of Alkanes
The intermediate and high temperature combustion (e.g., above 900 K) of alkanes larger
than methane proceeds via either unimolecular decomposition or H-atom abstraction.
Unimolecular decomposition involves the breaking of the fuel’s C-C and/or C-H bonds,
and such reactions are typically favored at very high temperatures (e.g., above 1300-1400
K) and fuel rich conditions. The breaking of C-C bonds is favored over the breaking of
C-H bonds because of the bond dissociation energy is lower, and the reaction proceeds
as shown in Equation 3.24.
RH + (M)→ R′ ⋅+R′′ ⋅+(M) (3.24)
where
RH is an alkane molecule
R′⋅ and R′′⋅ are alkyl radicals such as CH3, C2H5, etc.
(M) is a non-reacting collision partner
When highly reactive radicals, such as O, OH, and H, are present they can abstract
H atoms from the fuel, as shown in Equation 3.25. These reactions are favored at
Chapter 4. Experimental Apparatus and Analytical Methodology 47
chamber. The pumped fuel is delivered to a 30 kHz ultrasonic atomizer7 unit that breaks
the fuel into micro-droplets. The unit’s ultrasonic power supply converts 60 Hz energy to
a high frequency electrical energy at 30 kHz. Then, a piezoelectric transducer converts
the electrical energy to mechanical vibrations. The vibrations are intensified in a probe
and focused at its tip. The liquid fuel is dispensed in the probe, where it spreads out as
a thin film on the tip. The oscillations at the tip atomizes the liquid into micro-droplets
to form a gentle, low viscosity mist. The body of the atomizer was kept cool at 30 ∘C by
blowing compressed air over it.
A schematic of the fuel vaporization/mixing chamber is shown in Figure 4.3. The
atomized fuel is sprayed into the mixing chamber, where it mixes with nitrogen gas8.
The nitrogen was not preheated in this study because high temperatures near the top
of the mixing chamber can overheat the atomizer. The temperature in the lower half
of the chamber was maintained using heating tapes9 and a thermal blanket10. A high
temperature in only the lower portion of the mixing chamber served to vaporize the fuel
droplets while minimizing the heat transfer to the atomizer. The temperature inside the
chamber is measured using a stainless steel sheathed K-type thermocouple inserted at the
top of the chamber. Specific details on the temperature set point are provided in Table
4.2 in the next section. The mixing chamber is a custom-made stainless steel column11 30
cm in length and 10 cm in outer diameter (OD), with a 15 cm OD, 3/8′′ thick, stainless
steel flange bolted at the top to accommodate insertion of the atomizer, nitrogen, and
thermocouple. The column is packed with a 20 cm long bed of glass marbles. The
bed increases the path length traversed by gaseous fuel and nitrogen molecules; thereby,
increasing mixing of the two streams. The gaseous mixture of fuel and nitrogen then
flows to the bottom port of the burner apparatus via a 1/4′′ stainless steel transfer line
heated to 250 ∘C 12.
It was found that the atomizer did not operate properly when threaded onto a 3/8′′
thick stainless steel flange plate. The thick plate dampens the oscillations transmitted
to the atomizer tip and prevents the tip from vibrating. By working with the atomizer
manufacturer, it was determined that a portion of the flanged plate should be machined
7Sonaer custom built 30 kHz model8Linde Grade 4.89Omega SWH Ultra-High Temperature Heating Tapes
10Unifrax InsulfraxS blanket11Designed by Dr. John Z. Wen12Unique Heated Products Instrument Grade Heating Sample Line
Chapter 4. Experimental Apparatus and Analytical Methodology 48
to 1/32′′ thick, in order to minimize the dampening effects. The thin portion of the
machined flange plate allowed the atomizer to function properly while still meeting the
other design requirements of the flange. Refer to Figure 4.4 for a schematic of the
machined flange plate.
Chapter 4. Experimental Apparatus and Analytical Methodology 49
Figure 4.3: Schematic of the mixing chamber. Design by Dr. John Z. Wen
Figure 4.4: Schematic of the machined flange plate.
Chapter 4. Experimental Apparatus and Analytical Methodology 50
4.3 Supply of Fuel and Oxidizer Streams
The flow rates of fuel and oxidizer streams through the burner ports were key parameters
in these experiments. It is important that the momentums of the two streams be nearly
equal, so that a stagnation plane is created where the streams meet. The molar concen-
trations of fuel and oxidizer in each stream needs to be sufficient enough to light a flame;
however, high sooting flames are not desired. The flow rates of nitrogen, oxygen and air
are controlled using mass flow controllers13, while the flow rate of liquid fuel is controlled
by a peristaltic pump. The mass flow controllers are regularly recalibrated using a pos-
itive displacement gas flow meter 14. The inlet oxidizer and fuel stream concentrations
are selected based on the following criteria:
∙ a low Reynold’s Number to create a laminar flame
∙ a low sooting flame to prevent clogging of sampling probe
∙ avoid a very hot flame that will damage the probe
∙ a balanced momentum of the two streams to form a stagnation plane at their
intersection
∙ avoid excessive unburned fuel
∙ at the flame plane, an N2/O2 ratio near that of air to make the study relevant to
actual flames
The molar composition and measured temperature of the fuel and oxidizer streams
entering the burner for each experiment is indicated in Table 4.2. The temperature
of the oxidizer stream exiting the top burner was near 140 ∘C. The cause of this high
temperature was heat convected towards the top burner port from rising combustion
product gases.
It is important to ensure that the liquid fuel is sufficiently vaporized in the mixing
chamber and that it does not condense before exiting the burner, which has a maximum
operating temperature of 130 ∘C. To prevent condensation, the experiments conducted in
this study were performed at low partial pressures of fuel (i.e., 0.02-0.06 atm) wherein the
13Teledyne-Hastings Mass Flow Controller HFC20214BIOS Definer 220
Chapter 4. Experimental Apparatus and Analytical Methodology 51
vaporisation temperature is depressed. The vapour pressure of different FAME species
[3] was plotted to determine which fuels could be used under the temperature constraints
of the burner. Figure 4.5 indicates that FAME with 10 carbons or less can be sufficiently
vaporised at 130 ∘C.
Table 4.2: Experimental conditions
Mixing Chamber Fuel Stream Fuel Port Oxidizer Stream
T (∘C) Composition Exit T (∘C) Composition
n-butanol 110 5.9% Fuel 80 42.1% O2
94.1% N2 57.9% N2
n-butane N/A 5.9% Fuel 30 42.1% O2
94.1% N2 57.9% N2
methyl decanoate 150 1.8% Fuel 110 42.1% O2
98.2% N2 57.9% N2
1.0E-05
1.0E-04
1.0E-03
1.0E-02
1.0E-01
1.0E+00
1.0E+01
1.0E+02
0 50 100 150 200 250 300
Temperature (C)
Vapour
Pre
ssure
(a
tm)
Methyl Hexanoate
Methyl Octanoate
Methyl Decanoate
Methyl Dodecanoate
Methyl Oleate
Figure 4.5: Vapor pressure curves for fatty acid methyl esters
Chapter 4. Experimental Apparatus and Analytical Methodology 52
4.4 Reynold’s Number and Strain Rate Calculations
The Reynold’s Number (Re) is used to determined whether the flow exiting the burner
port is laminar (Re < 2300), transient (2300 < Re < 4000), or turbulent (Re > 4000).
The goal of these experiments is to generate a laminar diffusion flame, so the fluid velocity
must be set accordingly. Equation 4.1 is used to calculate the Reynold’s Number for flow
in the burner port assuming a plug flow behaviour.
Re =� ⋅ u ⋅ d�
(4.1)
where Re is the Reynold’s Number
� is the gas density in kgm3
u is the gas velocity in ms
d is the burner port diameter in m
� is the dynamic viscosity in N ⋅sm2
The strain rate is defined as the normal gradient of the normal component of the flow
velocity [4]. The strain is calculated using Equation 4.2.
a1 =2∣V1∣L
(1 +∣V2∣√�2
∣V1∣√�1
)(4.2)
where a1 is the strain rate on the fuel side in s−1
L is the distance between the two burner ports in cm
∣V1∣ is the absolute value of the fuel stream velocity at the fuel boundary
in cms
∣V2∣ is the absolute value of the oxidizer stream velocity at the oxidizer
boundary in cms
�1 is the fuel stream density in gcm3
�2 is the air stream density in gcm3
Chapter 4. Experimental Apparatus and Analytical Methodology 53
4.5 Gas Sampling System
The previous sections discussed the procedure for producing an opposed flow diffusion
flame. Once the flame was generated, a sampling system was used to obtain qualitative
and quantitative information about the flame’s characteristics. Specifically, the species
concentrations at various points between the two burner ports was measured to obtain
characteristic profiles. The following sections discuss the gas sampling system, as well as
the sampling procedure.
4.5.1 Sampling Apparatus
Microprobes, due to their small perturbation of flow fields, are commonly used in flame
studies to acquire the concentration of stable species [5, 6, 7, 8]. The gas sampling system
in these experiments consists of a quartz microprobe connected to a dual-stage pump15
via a heated 1/4′′ stainless steel line and a vacuum pressure gauge. The microprobe is
mounted on a sliding stage, allowing it to move into and out of the flame region easily.
The first stage (vacuum) of the pump creates a suction in the sampling line to withdraw
gas samples from the flame. An analytical instrument is connected downstream of the
second stage (compressor) of the pump to study the gases flowing through the line. The
compressor head on the pump pushes samples into the analytical instrument via a 1/4′′
stainless steel transfer line heated to 250 ∘C 16. The heated transfer lines are required to
prevent the condensation of high molecular weight species out of the sample gas.
Previous studies have identified precautions to take when using the microprobe sam-
pling technique. The primary objective is to eliminate chemical reactions within the
probe and sampling lines. A combination of rapidly reducing temperature and pressure
in the probe helps meet this objective.
Kassem and coworkers [9] studied the effect of microprobe cooling on fuel-rich, laminar
flat flames of chlorinated hydrocarbons. Their results indicate that cooled probes, as
opposed to uncooled probes, provided more accurate profiles of species concentration.
Schoenung and Hanson [5] showed that carbon monoxide (CO) measurements in the
post-flame region of a premixed methane/air flame were affected by the pressure within
the probe and sampling lines. Their results indicate that the CO concentration increases
as the pressure in the probe decreases, with the concentration reaching the actual value
15KNF oil-free dual-staged pump Model UN035.3 ST11 with heated heads16Unique Heated Products Instrument Grade Heating Sample Line
Chapter 4. Experimental Apparatus and Analytical Methodology 54
around 50 mm Hg. This finding suggests that CO is converted to CO2 in the probe region
unless the pressure is 7 kPA or below. Therefore, low temperatures and pressures in the
probe are required to quench the reactions.
Fristrom and coworkers [10] argue that it is not rapid temperature drop that is re-
quired for successful flame sampling. Instead, a combination of rapid pressure drop and
the destruction of radicals on the probe walls is responsible for quenching reactions at
the probe tip. They explain that 2nd order molecule-radical reaction rates vary with
the square of the gas density, which varies linearly with pressure; therefore, the reac-
tion rate decreases with a decrease in pressure. Furthermore, they argue that if a rapid
temperature drop is induced while keeping pressure constant, then reactions with acti-
vation energies of 20 kJ/mol or lower will increase in rate. These reaction rates vary
quadratically with density, which varies inversely with temperature at constant pressure;
therefore, decreasing the temperature alone would not quench all flame reactions.
The error introduced during microprobe sampling of laminar flame species profiles is
a concern. The insertion of a microprobe into a flame can disturb the fluid flow, provide a
surface for quenching reactions, offer poor spatial resolution, and absorb thermal energy.
These effects are certainly not reproducible in modeling simulations, so some researchers
make corrections to the experimental data to compensate for errors that microprobes
introduce (e.g., shifting species profiles, increasing measured concentrations, etc.). A
recent study by Struckmeier et al. [11] compared the quality of species profiles obtained
using intrusive microprobes and nonintrusive optical techniques. The results indicate
that data acquired using both techniques agreed well, and therefore suggests that probe
sampling techniques are an indispensable technique for reliable experimental data. In
addition, the authors state that only minor corrections (e.g., shifting profiles by 1 mm
and measured concentrations by 30%) should be made to compensate for errors in the
experimental data.
One set of experiments in this study used a quartz microprobe fabricated in the
Department of Chemistry Glass Blowing Shop. The inner diameter (ID) of the probe
tip could not be precisely controlled; therefore, a number of probes were made and
measurements were taken to select the best one. To avoid errors associated with probe
fabrication and the subsequent measurement procedures, a new sampling probe apparatus
was also designed as described below. A previous study by Syed [12] determined the
appropriate probe tip size. The study measured CO2 concentrations in a propane-air
flame using probe tips with various IDs. Syed suggests that a probe tip with an ID
Chapter 4. Experimental Apparatus and Analytical Methodology 55
of approximately 150-250�m is ideal. Probes with larger IDs do not successfully quench
reactions at the probe tip. Probes with smaller IDs restrict gas flow through the sampling
line, such that it is not possible to obtain gas samples from the flame.
In addition to having a small ID, the probe tip must also be long enough to allow
for a large pressure drop. As mentioned previously, the microprobe is connected to a
doubled-headed pump via 1/4′′ tubing. If the suction pump head creates an absolute
pressure of 4-6 kPa in the sampling line17, then this is sufficient to quench reactions. At
such low sampling pressures, the probe is not cooled since previous studies have shown
that a rapid pressure drop alone provides accurate sampling [10, 5]. This study found
that a probe tip ID of 250�m and a length of 3.5-4.0 cm achieves the aforementioned
pressure range in the sampling line.
This study also designed a new type of microprobe apparatus that is low-cost, re-
producible, and easily reparable in case of probe tip breakage. This new probe also
eliminated flow field disturbances which were observed with the previous probe design
(refer to Chapters 7 and 9). A schematic of the microprobe setup is shown in Figure
4.6. The novelty of the design is the probe tip and the fittings used for connecting it to
the dual-stage pump via heated 1/4′′ stainless steel tubing. The probe tip is made of a
fused silica tubing typically found in gas chromatography applications. These commer-
cially available tubings are manufactured according to strict inner and outer diameter
specifications. This study used a fused silica tubing with an ID of 200�m and an OD of
360�m18. It should be noted that the fused silica tubing is manufactured with a poly-
imide resin outer coating which aids in sealing within couplings and provides flexibility to
the otherwise brittle silica material. When the probe tip is exposed to high temperatures
in a flame, this coating quickly burns off and does not contaminate the sampling system.
A stainless steel 1/4′′-to-1/16′′ coupling19 with a 1/16′′ graphite-reinforced composite re-
ducing ferrule20 connects the 360�m OD fused silica tubing to the 1/4′′ stainless steel
sampling line. The entire apparatus is capable of being heated to 350 ∘C.
Extra precautions were taken to eliminate leakage of gases from the surrounding
environment into the sampling line. Leaks into the sampling system dilute the sample
gas and lead to incorrect measurements. For this reason, Swagelok couplings are used for
17Gauge pressure measured by a vacuum pressure gauge18Agilent Deactivated Fused Silica Retention Gap19VICI Valco 1/4′′’ to 1/16′′ external-internal reducing20VICI Valco 1/16′′ one piece fused silica adapter, 0.35 mm tubing OD
Chapter 4. Experimental Apparatus and Analytical Methodology 56
Figure 4.6: Schematic of microprobe (not to scale)
all connections. Leaks into the sampling line were detected using a container filled with
dry ice (i.e solidified carbon dioxide). The container was placed near a suspected point
of leakage, and carbon dioxide gas fills the surrounding area. The NDIR analyzer (see
section 4.6.1) was connected to the sampling line and the dual-stage pump was turned
on. If a spike in CO2 concentration was observed on the analyzer, then there was a leak
present in the sampling line. The necessary steps were then carried out to eliminate the
leak (e.g., the couplings were changed, the tube was replaced, etc.).
4.5.2 Sampling Procedure
The sampling objective was to obtain samples at various points along the vertical axis
separating the two burner ports. Thus, the sliding stage, on which the microprobe is
mounted, was inserted between the two burner ports. A window was cut into the quartz
shroud enclosing the burner setup to permit insertion of the sampling probe. Figure 4.7 is
a schematic of the probe and burner setup. The tip of the probe was placed approximately
1.5 mm behind the central vertical axis separating the burner ports. This allows samples
to be withdrawn from the middle points of the flame region. After insertion, the probe
was held stationary, while the burner assembly was moved along the vertical axis with
the turn of a micrometer knob on the translation stage. One complete counterclockwise
rotation of the micrometer knob moves the burner assembly downwards by 0.5 mm.
A measuring system is required to define the exact position of the microprobe between
the two burner ports. For this purpose, the bottom (fuel) port was taken as the zero
Chapter 4. Experimental Apparatus and Analytical Methodology 57
Figure 4.7: Schematic of microprobe and burner setup (not to scale)
height, while the top (oxidizer) port was taken as the maximum height. The position
of the probe was zeroed by touching its tip to the bottom port. The reading on the
micrometer knob was noted as the zero distance. Each counterclockwise turn of the
micrometer knob moved the burner assembly down; thus, increasing the distance between
the probe and the fuel port (i.e., moving the probe upwards). The exact height at which
the probe withdraws samples is equal to the total distance plus the outer radius of the
probe.
Once the probe was positioned at the desired height, the dual-stage pump was turned
on and gases withdrawn from the flame fill the sampling line. The sampling lines were
purged before any analytical measurements were taken. The purge time varied depending
on the location of the probe in the flame. Sampling points away from the center of the
flame required short purge times (e.g., 15-20 minutes), since the low temperature, high
density gases permit high flowrates. In contrast, sampling points near the center of the
flame required longer purge times (e.g., 40-45 minutes), since gases in this high temper-
ature region have an extremely low density and permit low flowrates. The sampling line
filled with gases from the specified flame region were then ready for analysis.
Chapter 4. Experimental Apparatus and Analytical Methodology 58
4.6 Analytical Techniques
The hydrocarbon species and all oxygenated compounds were analyzed by a GC/FID.
Carbon dioxide and carbon monoxide were quantified by NDIR analysis. The flame
temperature was measured by an R-type thermocouple.
4.6.1 Non-Dispersive Infrared Analysis
NDIR analysis is a technique used to measure gas concentrations based on the energy
absorption characteristics of a gas in the infrared red region. The NDIR instrument
passes infrared light through two identical cells, in parallel, and then onto a detector.
The first cell is filled with nitrogen, which does not absorb the light and serves as a
reference cell. The second cell contains the sample gas, which absorbs infrared energy.
The detector measures the difference in energy between the two streams of light. This
difference is the absorption, which is proportional to the concentration of sample gas by
the Beer-Lambert Law in Equation 4.3:
A = � ⋅ b ⋅ c (4.3)
where
A is the gas absorbance.
� is the molar extinction coefficient (concentration−1 ⋅ length−1).
b is the path length that the beam travels in the sampling tube.
c is the gas concentration.
CO and CO2 Measurements
The NDIR instrument21 was used to quantify levels of CO and CO2 in the flame samples.
The instrument is capable of measuring concentrations from 0% to 40%. Initially, the
instrument is zeroed and calibrated. To zero the instrument, nitrogen was flowed through
the unit and the unit was zeroed. Next, the unit was calibrated with a gas mixture
containing 9.9% CO and 9.9% CO2. The gas mixture was passed through the detector,
and the span was set to match the calibration gas concentration.
Measurements were taken using the dual-stage pump to withdraw samples from the
flame and then pushing them towards the NDIR analyzer. Water, which is damaging
21NOVA NDIR Analyzer
Chapter 4. Experimental Apparatus and Analytical Methodology 59
to the analyzer, was removed from the sample gas by passing it through a cooling box
and a coalescing filter. The analyzer’s built-in pump was not used during sampling. The
concentrations were recorded once the displayed reading remains constant for 5 minutes.
As mentioned previously, samples withdrawn from points away from the flame center
have higher flowrates than samples from within the middle of the flame. Therefore, the
analyzer’s display reading stabilized much quicker for the former (e.g., 15-20 minutes)
than the latter (e.g., 40-45 minutes).
4.6.2 Gas Chromatography
Gas chromatography (GC) is a technique used for separating volatile organic compounds
based on their differences in partitioning between a flowing mobile phase and a stationary
phase. In this study, the GC method was used to measure C1 - C8 hydrocarbons and
C1 - C11 oxygenated species (e.g., esters, aldehydes, alcohols, etc.) A GC instrument
consists of a flowing mobile phase (carrier gas), a stationary phase (separation column),
an injection port, an oven, and a detector.
The carrier gas carries the sample gas through the separation column, and compounds
are separated due to partitioning between the two phases. Since partitioning behaviour is
a strong function of temperature, the separation column is placed inside a temperature-
controlled oven. Separation of compounds with a range of boiling points is achieved by
starting at low temperatures and then increasing the temperature until high boiling point
compounds are eluted. The injection port is always maintained at a temperature higher
than the boiling point of the least volatile compound in the mixture.
The amount of time a given component spends in the separation column is called the
retention time. The retention time of a given component remains the same provided the
mobile phase, stationary phase, temperature control, and gas flowrates remain constant.
As each component of the separated sample falls onto the detector, a quantitative re-
sponse in the form of a peak is generated. A series of peaks with the retention time on
the x-axis and the detector (e.g., voltage) on the y-axis is called a chromatogram. The
peak retention time is used to identify each compound, and the peak area is used to
determine the quantity of the compound.
The instrument used in these experiments is a Varian 3800 GC 22 with electronic
flow controllers, a 1079 injector, a methanizer, and two flame ionization detector (FID)s.
22Remotely controlled by a PC using STAR Chromatography Workstation 6.41
Chapter 4. Experimental Apparatus and Analytical Methodology 60
The FID consists of a hydrogen/air flame and a collector plate. As gases flow from the
separation column, the flame burns organic molecules to produce ions. The ions are
attracted to the collector plate, which generates a voltage depending on the quantity of
ions collected. Additional details of the GC setup are discussed in the follow subsections.
GC Carrier Gas
The carrier gas (mobile phase) used for the GC was 99.997 % helium. Hydrocarbon
and oxygen traps were placed before the column to filter out any contaminants from the
carrier gas. The flowrate of helium through the separation column was varied depending
on the compounds being studied. Details are available in subsection “GC Measurement
Procedures” below.
Injection System
The purpose of the injection system is to load the sample gas onto the separation column.
It consists of a sample loop, gas sampling valve (GSV), and an injector. Flame samples
were pushed into the GC sampling loop by the dual-stage pump. After the sample loop
was purged and the sample was ready for analysis, the GSV 23 rotated and delivered the
sample to the GC.
The GSV is a 10 port rotary valve which directs the sample and carrier gases into
the injector. The GSV has two positions; the fill position and the load position. In the
fill position, the sample gas flows through the 0.25 mL sample loop, while helium carrier
gas flows to the injector and separation column. As the GSV turns to the load position,
the sample trapped in the loop comes in-line with the carrier gas flow. The sample is
carried through into the injector and is directed into the column. The GSV’s duration
in the load position is a user controlled parameter, which is set to 2 minutes in these
experiments. As the GSV returns to the fill position, the normal flow pattern resumes.
The GC has a 1079 universal capillary injector, which can be run in several modes
based on the type of injector insert used. The injector temperature was set at 250 ∘C to
prevent condensation of sample components. An unpacked 3.4 mm ID insert for splitless
mode operation was used. The sample gas in the injector can be introduced to the column
via split mode or splitless mode. In splitless mode, the entire sample is loaded onto the
column. This mode is advantageous for detecting trace compounds in the sample gas.
23VICI Valco 10 port valve with air actuator
Chapter 4. Experimental Apparatus and Analytical Methodology 61
However, highly concentrated samples can damage the separation column, so split mode
can be used to load only a portion of the sample onto the column. In these experiments,
a dual column GC method was used, so splitless injection was employed to maximize the
amount of analyte reaching the columns.
GC Measurement Procedures
Two different GC measurement procedures have been developed to analyze the gaseous
samples. The first method was used to study a number of hydrocarbon compounds and
oxygenates in all the flames presented earlier. However, this method was not suitable for
analyzing fatty acid methyl ester compounds because the columns used are not capable
of separating them. Therefore, a second method was developed using a column suitable
for fatty acid methyl ester separation.
The first method used two separation columns, a methanizer, and two FIDs to study
a number of hydrocarbon and oxygenated compounds in a single run. A schematic of
this method is shown in Figure 4.8 and the specific operating parameters are in Table
4.3. As the sample passes through the injector, it enters a 0.5 meter fused-silica retention
gap24. The sample then passes through a y-splitter25 where it is split to two columns
for separation. C1-C8 hydrocarbons are separated on a non-polar phase HP-Al/S PLOT
capillary column26, and then detected on the front FID. Aldehydes, ketones, and alcohols
are separated on a polar phase Poraplot U capillary column27, and then passed through
a methanizer before detection.
The FID provides a detector response proportional to the analyte’s concentration and
number of carbon atoms. For example, 1 mole methane would produce exactly 1/2 the
detector signal of 1 mole ethane, and 1/3 the signal of 1 mol propane. Schofield [13] has
provided a table of molar response factors for a number of hydrocarbons and oxygenates.
Oxygenated species provide lower responses on an FID because carbon-oxygen bonds are
not broken in the flame. Therefore, species such as formaldehyde and carbon monoxide
provide zero response on an FID, while higher oxygenated hydrocarbons provide weaker
responses than their non-oxygenated counterparts. For example, 1 mole acetaldehyde
has two carbon atoms but produces the same detector signal of 1 mole of methane. In
24Varian 0.53mm ID retention gap methyl deactivated25Varian universal quick seal splitter26Agilent Technologies HP-Al/S 50 m x 0.53 mm (L x ID)27Varian Poraplot U 25m x 0.53mm (L x ID)
Chapter 4. Experimental Apparatus and Analytical Methodology 62
order to improve the FID’s response to oxygenated hydrocarbons, this study passed the
separated sample gas through a methanizer prior to the FID. The methanizer is a 1/16′′
stainless steel tube packed with a powdered nickel catalyst. The methanizer is heated to
380 ∘C and requires a constant flow of hydrogen gas. As the sample flows through the
methanizer, the nickel catalyst breaks the carbon-oxygen bonds and then saturates them
with hydrogen. Therefore, all carbon-oxygen bonds are effectively converted to carbon-
hydrogen bonds, and the analyte now appears on the FID as an alkane with the original
number of carbon atoms. For example, formaldehyde appears as methane, acetaldehyde
appears as ethane, etc.
FrontFIDGSV
Sample in
1079 injector
GC column oven
Retention gap
Plot column
Poraplot Ucolumn
Y-splitter
RearFID
methanizer
H2 Air He
Figure 4.8: Schematic of dual column GC Setup
The second GC method used in these experiments was designed to analyze fatty acid
methyl ester compounds in the methyl decanoate flame using a polar phase DB-WAX col-
umn28. This method used only one column and one FID. The GC parameters are provided
in Table 4.4. The method was capable of separating and identifying methyl propanoate,
Chapter 4. Experimental Apparatus and Analytical Methodology 68
Correction for Radiation Losses
The temperature measured by the thermocouple differs from the true flame temperature
due to aerodynamic, thermal, and/or chemical perturbations. The methods of minimizing
the effect of these perturbations are discussed in detail by Fristrom and Westenberg [6].
However, even if all disturbances are minimized, the thermocouple will register a different
temperature than the true stream temperature due to radiation losses. Correcting for
these losses is estimated by equating the heat transferred to the thermocouple from the
gas to the heat lost by radiation from the wires. The equation given for a spherical device
(i.e., Nusselt number of 2) is:
Tg − Tc =� ⋅ � ⋅ d ⋅ (T 4
c − T 4w)
2 ⋅ k(4.5)
where
Tg is the true gas temperature (K)
Tc is the measured gas temperature (K)
� is the emissivity of the thermocouple element (dimensionless)
� is the Stefan-Boltzmann Constant = 5.67x10−08 ( Wm2⋅K4 )
d is the wire diameter (m)
k is the thermal conductivity of gas ( Wm⋅K )
Tw is the wall (ambient) temp. to which heat is radiated = 300 (K)
The thermal conductivity, k, of the gases was estimated as the thermal conductivity of
air at the measured temperature. The thermal conductivity of air at various temperatures
was obtained from the CRC Handbook [3]. Linear interpolation and extrapolation was
used to determine thermal conductivity at temperatures not listed in the handbook.
The emissivity, �, of the thermocouple element was obtained from the study by
Bradley and Entwistle [15].
Literature Cited
[1] S. Sarathy, “Using an opposed flow diffusion flame to study the oxidation of C4 fatty
acid methyl esters,” Master’s thesis, University of Toronto, 2006.
[2] I. Glassman, Combustion, 3rd ed. San Diego, CA: Academic Press, 1996.
[3] D. R. Lide, Ed., CRC Handbook of Chemistry and Physics, 87th Edition. Boca Ra-
ton, FL: Taylor and Francis, 2007. [Online]. Available: http:/www.hbcpnetbase.com
[4] K. Seshadri, T. Lu, O. Herbinet, S. B. Humer, U. Niemann, W. J. Pitz, R. Seiser,
and C. K. Law, “Experimental and kinetic modeling study of extinction and ignition
of methyl decanoate in laminar non-premixed flows,” Proceedings of the Combustion
Institute, vol. 32, no. Part 1, pp. 1067–1074, 2009.
[5] S. M. Schoenung and R. K. Hanson, “CO and temperature measurements in a flat
flame by laser absorption spectroscopy and probe techniques.” Combustion Science
and Technology, vol. 24, no. 5-6, pp. 227 – 237, 1981.
[6] R. M. Fristrom, “Comments on quenching mechanisms in the microprobe sampling
of flames,” Combustion and Flame, vol. 50, pp. 239–242, 1983.
[7] A. M. Vincitore and S. M. Senkan, “Polycyclic aromatic hydrocarbon formation in
opposed flow diffusion flames of ethane,” Combustion and Flame, vol. 114, no. 1-2,
pp. 259–266, July 1998.
[8] A. Sinha and M. J. Thomson, “The chemical structures of opposed flow diffu-
sion flames of c3 oxygenated hydrocarbons (isopropanol, dimethoxy methane, and
dimethyl carbonate) and their mixtures,” Combustion and Flame, vol. 136, no. 4,
pp. 548–556, March 2004.
69
Literature Cited 70
[9] S. S. M. Kassem, M. Qun, “Chemical structure of fuel-rich 1,2-
C2H4Cl2/CH4/O2/Ar flames: Effects of micro-probe cooling on the sampling
of flames of chlorinated hydrocarbons,” Combustion Science and Technology,
vol. 67, pp. 147–157, 1989.
[10] R. M. Fristrom, Flame structure. New York: McGraw-Hill, 1965.
[11] U. Struckmeier, P. Osswald, T. Kasper, L. Boehling, M. Heusing, M. Koehler,
A. Brockhinke, and K. Kohse-Hoeinghaus, “Sampling Probe Influences on Temper-
ature and Species Concentrations in Molecular Beam Mass Spectroscopic Investiga-
tions of Flat Premixed Low-pressure Flames,” Zeitschrift Fur Physikalische Chemie-
International Journal of Research in Physical Chemistry & Chemical Physics, vol.
223, no. 4-5, pp. 503–537, 2009.
[12] S. A. Syed, “Oxidation studies of surrogate bio-diesel fuels in opposed flow diffusion
flames,” 2005.
[13] K. Schofield, “The enigmatic mechanism of the flame ionization detector: Its over-
looked implications for fossil fuel combustion modeling,” Progress in Energy and
Combustion Science, vol. 34, no. 3, pp. 330–350, June 2008.
[14] C. McEnally, U. Koylu, L. Pfefferle, and D. Rosner, “Soot volume fraction and
temperature measurements in laminar nonpremixed flames using thermocouples,”
Combustion and Flame, vol. 109, no. 4, pp. 701–720, June 1997.
[15] D. Bradley and A. Entwistle, “Deterimination of the emissivity, for total radiation,
of small diameter platinum-10% rhodium wires in the temperature range 600-450
degrees C,” British Journal of Applied Physics, vol. 12, pp. 708–711, December 1961.
Part II
Biobutanol
71
Chapter 5
Background
In 2006, British Petroleum (BP) and DuPont announced that they would start selling
biobutanol made from sugar beets, as a gasoline blending component in the United
Kingdom [1]. The biobutanol, would be produced using a fermentation process similar
to that of ethanol, and its feedstock include sugar beet, sugar cane, corn, wheat, and
potentially lignocellulosic biomass. Proponents [2, 3, 4] of biobutanol highlight many
characteristics that make it a superior biofuel, such as:
∙ the ability for blending with gasoline at higher concentrations than ethanol without
modifying current vehicle and engine technologies;
∙ a higher energy density than ethanol, which provides better fuel economy at higher
blend ratios;
∙ enhanced water tolerance compared to ethanol, allowing use of the existing fuel
distribution infrastructure;
∙ the use of existing feedstock and refineries (with minor modifications) used for
bioethanol production;
∙ the potential to produce from lignocellulosic and waste biomass feedstock;
∙ and a lower vapour pressure than ethanol and gasoline, thereby decreasing volatile
organic compounds (VOC) emissions.
It is apparent that biobutanol is being proposed as an eventual replacement for the
bioethanol currently used in SI engines. BP and DuPont stated that a complete environ-
mental LCA based on actual manufacturing design models is underway, but to date no
72
Chapter 5. Background 73
results have been published [5]. If biobutanol is to replace bioethanol, then its sustain-
ability needs to be critically assessed. In addition, the combustion kinetics of biobutanol
must be understood to aid in the design of combustion systems. This chapter provides
background information relevant to the development of an LCA and a chemical kinetic
mechanism for biobutanol.
5.1 Biobutanol History
The production of biobutanol from biomass feedstock is not novel. Jones and Woods [6]
have provided a detailed history of biobutanol production via fermentation. Figure 5.1
is a visual representation of this history and following is a summary.
Figure 5.1: Timeline of biobutanol history
The fermentation of sugars to biobutanol, specifically the isomer n-butanol 1, was first
documented by Louis Pasteur in 1861, although the yields he achieved were not sufficient
for commercialization. In the early 1900s, a German born chemist, Chaim Weizmann, was
attempting to produce butanol for synthetic rubber production, and discovered that the
Clostridium acetobutylicum organism was capable of converting large amounts of sugars
into a mixture of acetone-butanol-ethanol (ABE) in the molar ratio of 3:6:1. Weizmann
1Henceforth, the terms biobutanol, n-butanol, and butanol are used interchangeably
Chapter 5. Background 74
had performed his pioneering work while living in England, and at the onset of World
War 1, the English sought his help in producing bioacetone from biomass as a precursor
to cordite, a smokeless propellant used in munition. During the war, a large amount of
ABE was produced in England, Canada, and the U.S. from feedstock such as corn and
sugar beet molasses. During this period, the unwanted butanol was stored in large vats
waiting for some commercial use.
After World War 1, the automotive industry in the U.S. was rapidly growing and
the demand for paint lacquers increased dramatically. One company commercialized a
method of producing high quality lacquers using butanol and its ester, butyl acetate, as
solvents. Thus, in the 1920s and 1930s, there was a large industry built around convert-
ing biomass sugars into ABE for the automotive paints industry. In a 1927 Scientific
American article, Killeffer [7] states that the butanol production in the U.S. doubled to
60 tons per day in just 18 months.
In 1938, World War 2 began and all the U.S. fermentation facilities reverted to pro-
ducing acetone for cordite manufacturing. Japan, South Africa, India, Australia, China
and the U.S.S.R also opened a number of ABE fermentation facilities using wheat, rye,
and corn as feedstock. After World War 2, there was a rapid decline in ABE production
from biomass for two reasons. The cost of biomass feedstock rose sharply as corn, wheat,
and molasses became staples for animal feed. In addition, the petrochemical industry
boomed in the 1950s and 1960s, leading to superior solvents for use in automotive paint
lacquers. The demand for butanol decreased sharply, and by the 1980s virtually all the
ABE plants in the world had closed.
5.2 Biobutanol Production
2 Butanol can be produced from the same feedstock as ethanol (e.g., sugar beet, sugar
cane, corn, wheat, and lignocellulosic biomass) using a fermentation process. During
the first part of the last century, the production of butanol was a large-scale industrial
process. The conventional production process used C. acetobutylicum to produce acetone,
n-butanol, and ethanol (termed the ABE process). Almost two thirds of the total butanol
produced in the late 1940s was produced by fermentation. The traditional fermentation
process could not compete economically with the production of butanol from petroleum
2This sections includes background research conducted by S.M. Sarathy and Y. Zhang (PhD Candi-date, Dept. of Civil Engineering, University of Toronto).
Chapter 5. Background 75
due primarily to high feedstock (e.g., corn) costs and the large energy requirements for
butanol recovery [6].
The traditional process for production of butanol and ethanol from corn is as follows,
[8, 9]:
1. The corn crop is cultivated and milled.
2. The milled corn is slurried with water. For ethanol, the milled corn undergoes liq-
uefaction and sacharrification to convert starches to monosaccharides. This process
is not required for butanol because the fermentation organism is capable converting
starches to ABE.
3. The slurried mixture is then fermented in a batch reactor to convert sugars (and
starches) into a beer solution containing the desired product(s).
4. The desired solvent, or biofuel, is then recovered from the beer by a series of
separation processes.
The major consumer of energy in the production of butanol and ethanol are the
separation processes (i.e., distillation) required for product recovery. The production of
butanol via fermentation is severely handicapped due to the low concentrations of butanol
in the fermented beer. The C. acetobutylicum used to convert starches and sugars into
ABE cannot tolerate high solvent concentrations, so typical end-product concentration
is 20 g of ABE per liter of beer [8].
Several recent advances have occurred in butanol production, including the develop-
ment of strain, C. beijerinckii BA 101, with increased solvent tolerance to 33g/L total
solvents [10, 11], and the development of advanced fermentation techniques and down-
stream product recovery processes [12]. Environmental Energy Inc. (EEI), now Butyl-
Fuels LLC, has patented a novel dual-stage process and claimed the process significantly
improves butanol yield and minimizes undesired byproducts [3]. Green Biologics, a com-
pany based in the United Kingdom, has obtained significant funding to commercialize
the conventional ABE fermentation process by utilizing waste feedstock and advanced
fermentation and separation processes [4].
Historically, starch and sugar feedstock have been utilized for butanol production,
however, butanol could be produced from lignocellulosic biomass, a more abundant feed-
stock. During the last several decades, research has been conducted to convert lignocel-
lulosic biomass to butanol. Yu et al. [13] report that C. acetobutylicum can convert both
Chapter 5. Background 76
hexose and pentose sugars present in biomass hydrolyzates to butanol. Recently, Zverlov
et al. [14] documented the production of butanol from hydrolyzates of lignocellulosic
waste at a full-scale industrial plant in the 1980s in Russia.
Current metabolic engineering research is attempting to resolve the end-product in-
hibition associated with the conventional ABE process, so that higher yields of butanol
can be achieved. BP and DuPont disclosed that their partnership is patenting novel
biocatalysts for high yield production of n-butanol, as well as its higher octane isomers,
sec-butanol and iso-butanol [5]. Atsumi et al. are genetically modifying the metabolic
pathways of E. Coli to develop non-fermentative pathways for biofuel n-butanol, sec-
butanol, and iso-butanol production [15].
The aforementioned genetic engineering research may lead to novel biobutanol pro-
duction technologies in the long term (e.g., 15-20 years); however, in the short term (e.g.,
5-10 years), biobutanol could only be produced at industrial scales by reintroducing the
traditional fermentation technology, albeit with process improvements. Presently, there
is no commercial-scale production of biobutanol from starch, sugar or lignocellulosic feed-
stock and there is limited data on the processes. In spite of this, there is renewed interest
in butanol, particularly considering its apparent attractiveness as an automotive fuel.
5.3 Biobutanol Fuel Properties
3Although limited research has been conducted on the use of butanol in vehicles, re-
cent testing by BP found that butanol has several fuel property advantages compared
to ethanol [16]. Selected fuel properties are presented in Table 5.1. Butanol has a
higher energy density (i.e., LHV) than ethanol, which leads to better vehicle fuel econ-
omy. While ethanol has a higher octane rating than gasoline, butanol is octane neutral,
thus lessening requirements for additional fuel modification when blending with gasoline.
The lower oxygen content of butanol allows it to be blended at higher proportions than
ethanol without requiring engine modifications or exceeding current blending regulations.
When blended with gasoline, butanol contributes much less to the reid vapour pressure
equivalent (RVP) than ethanol, even having a negative impact on the RVP when bu-
tanol and ethanol are co-blended. Although ethanol/gasoline blends were found to lead
to distillation curve abnormalities, which may negatively impact driveability, this was
3This sections includes background research conducted by S.M. Sarathy and S. Sleep (UndergraduateResearch Assistant, Dept. of Civil Engineering, University of Toronto).
Chapter 5. Background 77
Table 5.1: Selected fuel properties of butanol, ethanol, and gasoline
n-Butanol Ethanol Gasoline
Chemical Formula C4H9OH C2H5OH C4 to C12 hydrocarbons
Energy Density (LHV)(MJ/L) 26.9 21.2 32.2-32.9
Fuel Density at 20 ∘C (kg/L) 0.81 0.79 0.72-075
Boiling Point (∘C) 117 78 <210
Octane Number (R+M/2)a 87 116 87
RVP at 10% v/v, (kPa) 34 130 <60/90b
Oxygen (%wt) 21.6 34.7 <2.7
a Octane number of alcohols is given for blends containing 5 vol% alcohol in
gasoline.b Summer/Winter specifications provided for gasoline DVPE.
not the case with butanol/gasoline blends, as no vapour pressure or distillation curve
abnormalities were noted. Butanol, unlike ethanol, is immiscible with water, so it can
be distributed in existing pipelines without the risk of water contamination; ethanol re-
quires an alternate distribution infrastructure. In addition, butanol is not corrosive to
engine components while ethanol corrodes copper and brass. Preliminary results of BP’s
vehicle testing of gasoline/butanol blends containing 5% and 10% butanol showed no
significant changes in carbon monoxide, hydrocarbon, and NOx emissions compared to
regular gasoline and gasoline/ethanol blends containing 5% ethanol. These combined fac-
tors indicates that butanol may be a superior gasoline blending component than ethanol.
However, further combustion testing and evaluation of the life cycle performance of bu-
tanol compared with ethanol and gasoline are required.
Literature Cited
[1] G. Hess, “BP and DuPont to make biobutanol,” Chemical & Engineering News,
vol. 84, pp. 9–10, 2006.
[2] BP-DuPont, “Bp and duPont biofuels fact sheet,” British Petroluem and DuPont,
Tech. Rep., June 2006.
[3] D. Ramey and S. Yang, “Production of butyric acid and butanol from biomass,”
US Department of Energy, Morgantown, Washington, Tech. Rep. DE-F-G02-
00ER86106, 2004.
[4] G. Clark, “Green biologics secures 1.58 million British pounds to develop next gen-
Ethanol from sugarcane and corn, and biodiesel from soybean and rapeseed are being
commercially produced today; however, the biofuels of choice for the future and their
methods of production are still uncertain [1, 2]. Initiatives to substantially increase bio-
fuel production and upcoming low carbon fuel standards motivate a critical examination
of the environmental implications of current and potential future fuel production and
end use pathways.
Over the past two decades, there have been many studies examining the environmen-
tal performance of corn-based ethanol. However, few studies have been conducted on
corn-based butanol. It is only recently that butanol production technology has become
competitive due to improved product yields. The following subsections include a review
of important studies on bioethanol LCA, as well as some recently published LCA research
on biobutanol.
6.1.1 Bioethanol LCA
The sustainability of bioethanol has been under intense scrutiny by the scientific com-
munity. Pimentel and Patzek [3, 4] concluded that the total energy required to produce
corn-based ethanol is greater than the energy content of the ethanol produced. However,
the majority of studies [5, 6, 7, 8, 9, 10] indicate that corn-based ethanol can displace
80
Chapter 6. LCA of Biobutanol for use in Transportation 81
fossil energy sources and reduce GHG emissions. A study by Farrell et al. developed
a meta-model to evaluate various LCA studies [11] and demystify the aforementioned
discrepancies. The results indicate that Pimentel and Patzek [3, 4] used obsolete data
from ethanol production technologies in the early 1980’s, and the authors did not cor-
rectly allocate credits for the ethanol coproducts. Farrell et al. also made the following
recommendations for future LCA evaluations:
∙ use the displacement method to credit coproducts,
∙ obtain accurate and current data,
∙ clearly define future scenarios, and
∙ define performance metrics such as GHG emissions, fossil energy use, soil erosion,
etc.
Kim and Dale [8, 9] used the system expansion approach to credit coproducts and also
considered the carbon sequestration effect from increasing the soil organic carbon level
during biomass production. The system boundaries include upstream processes (e.g.,
fertilizer production and transport), ethanol production and coproducts by dry milling,
urea production, and the corn farming system. The study concludes that using ethanol
can reduce GHG emissions and fossil energy use.
Studies at the Argonne National Laboratory (ANL) [10, 12] used the Greenhouse
Gases, Regulated Emissions, and Energy Use in Transportation (GREET) model to
conclude that ethanol could reduce GHG emissions and fossil energy use. The authors
used the displacement method to assign credits to the fuel ethanol coproducts. The
results by Wu et al. [10] project that corn ethanol produced by the dry milling process in
the year 2012 reduces fossil energy use by 38% and GHG emissions by 21%. The GREET
model version 1.8 by ANL is an LCA model with detailed descriptions of all the data sets
and procedures. Therefore, it will be used in this study to determine the environmental
performance of biobutanol and to compare it with bioethanol.
It is clear from the literature that conventional ethanol production from food crops,
most notably corn, has limited potential [13, 14]. Reasons for this include competition
from the food industry [15], limited agricultural land for crop growth [13], and high
energy input requirements for agricultural chemicals and harvesting [14]. For biofuels
to live up to their potential as attractive alternative fuels, lignocellulosic biomass must
Chapter 6. LCA of Biobutanol for use in Transportation 82
be considered. Consisting of energy crops, agricultural and forest residue as well as
other sources, lignocellulosic biomass offers advantages over food crop feedstock. These
advantages include: the ability to be cultivated on marginal agricultural land, lower
agricultural chemical requirements, lower energy requirements, and the potential of uti-
lizing the lignin portion of the biomass as an energy source for the production process.
For these reasons the net energy gains of lignocellulosic ethanol are much higher than
those of corn ethanol, and the resulting GHG emissions are much lower. Despite these
advantages, the conversion of lignocellulosic feedstock to fuels is not yet at commercial
scale, primarily because lignocellulosic biomass is more challenging to convert to alcohols
[13, 14, 15, 16].
Land Use Change and Food Security Issues
Recently, there have been a number of studies citing food security and land use change is-
sues associated with biofuel production. Prices of food, including corn and rice, increased
sharply in 2007 and 2008. Due to media sensationalisation, the rise in food prices was
largely attributed to the increasing diversion of food crops for biofuel use. In reality, the
rise in food prices can be attributed to a number of factors, including, ”increasing energy
costs, climate change, stagnation in crop productivity, and diversion of crops or crop-
lands to biofuel production” [17]. Furthermore, rising food prices may actually benefit
the largest population of undernourished individuals who are farmers [18]; however, the
urban poor would suffer.
As the value of biofuels and the price of their feedstock (i.e., corn) increase, farmers
worldwide are converting forests and grassland to cropland for additional grain produc-
tion. Searchinger et al. and Fargione et al. report [19, 20] that the conversion of carbon-
rich forest lands to cropland inevitably release CO2 emissions into the atmosphere. The
magnitude of the GHG released is much larger than the GHG reductions the biofuels offer
by displacing fossil fuels, suggesting that biofuels actually increase GHG emissions via
indirect land use change. Proponents of biofuels argue that the aforementioned studies
overestimate the carbon debt associated with biofuels. In any case, the use of biofuels
derived from traditional crops and croplands is likely unsustainable in the long term,
and therefore, it is vital that biofuels be produced from waste feedstock and/or feedstock
grown on degraded or abandoned land.
Chapter 6. LCA of Biobutanol for use in Transportation 83
6.1.2 Biobutanol LCA
For all the research on ethanol, very little has been published on the life cycle implications
of butanol. S&T Squared Consultants Inc., a consulting company performed an LCA of
biobutanol using the GHGenius model[21]. During the course of the present study, Wu
et al. published results using the GREET model [22]. A comparison of the two studies
shows some inconsistent results primarily due to Wu et al. utilizing recent data [23, 24]
while S&T Squared Consultants Inc. utilized data for the same process (ABE) but from
the mid-1980s. Although both studies found that the butanol production process was
more energy intensive than that of ethanol, they differed substantially in the energy
inputs estimated for the butanol production. S&T Squared Consultants Inc.Squared
Consultants Inc. found that butanol production consumed more fossil energy than the
energy the butanol contained. Wu et al. concluded that butanol resulted in a net
energy gain with GHG emissions slightly higher than ethanol, but still 20% to 60% lower
than gasoline. They also found that the butanol benefits were very dependent on the
treatment of the acetone coproduct due to the coproduct credit scheme. It should be
noted that neither study considered lignocellulosic biomass feedstock, or the possible
benefit of using existing gasoline pipeline infrastructure for butanol distribution, as both
assumed delivery by diesel truck.
The objective of this study is to examine the potential attractiveness of butanol as
a transportation fuel to replace gasoline use in the U.S. Selected life cycle environmen-
tal metrics associated with the production of butanol from corn grains are compared
with ethanol using the same feedstock. The goal is to determine the fossil energy use,
petroleum energy use, and GHG emissions associated with the production and use of
corn-based butanol. The results of this study are also compared with those of a previous
study [22].
6.2 Methodology
The LCA [25] method is utilized to evaluate the environmental performance of butanol
and ethanol derived from corn in the U.S. The environmental metrics examined are fossil
and petroleum energy use and GHG emissions. The GHGs considered are CO2, CH4,
and N2O which are aggregated as CO2 equivalent (CO2-eq.) based on the 100-year global
warming potentials recommended by [26].
Chapter 6. LCA of Biobutanol for use in Transportation 84
The butanol and ethanol LCA models consist of similar activities. The life cycle
begins with corn farming and the associated manufacture of upstream agronomic inputs
(i.e., fertilizers and herbicides). The corn is collected from the field and transported to a
conversion facility where it is converted to ethanol or butanol with associated coproducts.
The alcohol is then distributed to a facility for blending with gasoline. The life cycle of
the biofuel ends with the combustion of the alcohol/gasoline blend in an automobile.
Transportation between the various fuel production processes is also included in the
models. Figure 6.1 is a simplified diagram showing the life cycle of butanol, including its
use in a light duty vehicle (LDV).
This study divides the lifecycle into well-to-pump (WTP) and pump-to-wheel (PTW)
portions. The WTP portion includes the life cycle results solely associated with the
production and distribution of neat alcohol fuels, so as to distinguish from those when
gasoline is blended with them. The WTP results for neat alcohol are reported with
the functional unit (FU) being 1 MJ of alcohol available at the pump, where the pump
is defined as the gasoline blending facility. The PTW portion includes blending of the
biofuel with gasoline and subsequent processes, and this study reports the total well-to-
wheel (WTW) results for 85% alcohol and 15% gasoline blends. The FU for the WTW
analysis is one km driven by a midsize passenger vehicle.
6.2.1 Data sources and uncertainties
The often-cited, publicly available GREET 1.8b model contains corn ethanol and butanol
pathways [12]. The mix of energy sources (e.g., coal, natural gas, electricity, etc.) used
for biofuel production are considered the same for both ethanol and butanol, and the
values are the default presented in GREET 1.b. The comparison ethanol pathways are
obtained directly from this model and default values for the year 2010 are assumed. The
year 2010 was selected for the analysis as it is expected that butanol could be available
by that time based on the recent announcements. Models for butanol production are
developed in this paper based on production data available in the research literature, as
described below. The production data are based on experimental work which has not
been optimized from a process engineering perspective. There is considerable uncertainty
as to the feasibility and performance of the process at commercial scales.
Chapter 6. LCA of Biobutanol for use in Transportation 85
Figure 6.1: A simplified life cycle flowchart for corn-derived butanol and ethanol
6.2.2 Corn ethanol and butanol production
Corn production and its life cycle implications are identical whether the ultimate fuel
to be produced is ethanol or butanol. Activities included in corn production are: agri-
cultural chemical manufacture, including transport to the field and application; grain
harvest; and transport to the biofuel production facility. Corn ethanol plants can be
classified into dry milling and wet milling operations. The GREET 1.8b dry mill facility
pathway is utilized since all recent facilities constructed in the U.S. have been of this type.
The resulting ethanol yield is 405 L/Mg of corn with dried grains with solubles (DDGS)
produced as a co-product at a yield of 290 bone-dry kg/Mg of corn [12]. Prior to leaving
the facility, the ethanol is usually denatured with at least 1-5 vol% gasoline; however, in
this study the denaturant is excluded, in order to compare the life cycles of neat biofuels
leaving the ethanol and butanol facilities.
The production of ethanol and butanol from corn using the dry milling process is quite
similar, although as noted earlier ethanol production is a mature process while butanol is
not currently produced at commercial scale. For the butanol life cycle model, all stages
of the life cycle up to the corn delivery to the production facility are assumed identical
Chapter 6. LCA of Biobutanol for use in Transportation 86
to those modeled in the ethanol process. After delivery to the production facility, the
two models diverge as the saccharification, fermentation and separation processes differ.
The energy required for grain handling and saccharification in the butanol facility was
estimated using the USDA AspenPlus (Aspen Technologies, Inc., Cambridge, MA) model
for corn dry mill ethanol [27, 28]. The AspenPlus model was modified for the butanol
process to produce a fermentation broth containing 60 g/L of sugars and assuming 95%
of substrate utilization1. The conversion of glucose to acetone, butanol, and ethanol was
estimated based on lab-scale yields using the C. beijerinckii BA101 organism in a batch
reactor [29]. Estimates were based on batch conversion technology since it is widely used
in industrial fermentation technology, and other newer processes based on continuous
fermentation technologies, in-situ gas stripping, and fed-batch technologies [23, 24, 30]
are not.
The energy requirements for the downstream processes (i.e., separation of solvents
from broth) were estimated using an AspenPlus model consisting of a gas stripper and
two distillation columns based on previous work by Liu [31]. The model does not employ
advanced separation technologies, such as adsorption and pervaporation, which have
been proven to reduce energy consumption in recent laboratory experiments [32]. The
downstream process model was not optimized for energy efficiency, so it was estimated
that 40% of the energy prediction for the the separation processes could be saved using
heat-integrated distillation [33, 34]. It should be noted that since the completion of the
present study, an AspenPlus simulation of a complete process for producing butanol via
acetone, butanol, and ethanol corn fermentation has been published by Liu [35]. It is
expected that this new simulation could predict energy use better than the simulations
mentioned above since it includes all processes from grain processing through to product
purification in one model.
Beside DDGS, the butanol fermentation process also produces ethanol and acetone
as co-products. Data on butanol and co-product yields are presented in Table 6.12. Co-
product allocation methods could significantly impact the energy and emissions results
for corn ethanol and butanol. Various scenarios for butanol production were investigated
using different co-product allocation methods. The butanol scenarios developed in this
1These conversion estimates were determined by Dr. M. Griffin (Dept. of Civil and EnvironmentalEngineering, Carnegie Mellon University).
2The yield and energy use estimates were estimated by S.M. Sarathy and Y. Zhang (Dept. of CivilEngineering, University of Toronto).
Chapter 6. LCA of Biobutanol for use in Transportation 87
study are as follows3
∙ Butanol with no co-products
∙ Butanol with DDGS co-product credited by the displacement method (assumes
DDGS displaces corn and soybean meal as an animal feed)
∙ Butanol with DDGS, acetone, and ethanol co-products credited by the energy allo-
cation method (i.e., energy use and associated GHG emissions are allocated among
products based on their energy output shares)
In addition, our butanol scenarios are compared to those of Wu et al. [22] and ethanol
results in GREET 1.8b. The comparison scenarios are as follows:
∙ Wu et al. [22] results for butanol with DDGS and ethanol co-products credited by
the displacement method
∙ Wu et al. [22] results for butanol with DDGS, acetone, and ethanol co-products
credited by the energy allocation method
∙ Ethanol with no co-product (GREET 1.8b) [12]
∙ Ethanol with DDGS co-product credited by the displacement method (GREET
1.8b) [12]
6.2.3 Post Production Life Cycle Activities
Following the production of the biofuels they are transported from the plants to bulk
terminals, where they are blended with gasoline. Barge, rail and truck are assumed to
transport 40%, 40%, and 20% of the ethanol, respectively [12]. The distances for the
barge, rail and truck transport are 837 km, 1287 km, and 129 km, respectively. For
butanol, it is assumed that 100% pipeline transportation with an average distance of 966
km (i.e., the weighted average of the transportation distances for ethanol). 85% of each
biofuel is blended with 15% conventional gasoline (CG) (i.e., not reformulated) before
use in the vehicle to produce 85% ethanol-15% gasoline (E85) and 85% butanol-15%
gasoline (Bu85) blends. After blending, the fuel is distributed to refueling stations by
3The scenarios presented in this study were selected by Dr. H. MacLean (Dept. of Civil Engineering,University of Toronto.)
Chapter 6. LCA of Biobutanol for use in Transportation 88
Table 6.1: Data for estimating energy use and GHG emissions for corn butanol production
Assumptions Data Source
Corn Production 2010 Assumptions GREET 1.8b
in GREET 1.8b
Butanol production
Butanol yield (L/Mg of corn) 200 Estimated here
Acetone yield (L/Mg of corn) 100 Estimated here
Ethanol yield (L/Mg of corn) 34 Estimated here
DDGS yield (bone-dry kg/Mg of corn) 280 GREET 1.8b
Process fuel requirement (MJ/L of butanol)a 40 Estimated here
a Process fuel requirements are assumed to be met 80% by natural gas and 20% by
coal based on GREET 1.8b’s corn ethanol pathway [12].
diesel truck (48 km). The GREET 1.8b model is used to estimate energy use and GHG
emissions of the aforementioned distribution processes.
For the PTW analysis, the fuels are assumed to be utilized in a recent model year,
flexible fuel vehicle (FFV), the 2006 Chevrolet Impala. Flexible fuel vehicles can utilize
100% gasoline or an alcohol/gasoline blend with up to 85% alcohol. It is assumed that
there would be only minor modifications needed for the vehicle to utilize butanol blends
and that these would not impact vehicle performance with respect to energy use and GHG
emissions. The vehicle operation stage of the life cycle is based on fuel economy values
published by the U.S. Department of Energy [36]. The combined fuel consumption (fuel
economy) (55% city and 45% highway driving) for the Impala when fueled with gasoline
is 9.2 L/100km (25.5 mpg), whereas it is 12.3 L/100km (19.2 mpg) when fueled with E85.
Ethanol’s lower energy density accounts for the decrease in fuel economy; though, this is
very slightly offset by the greater thermal efficiency associated with ethanol combustion
[37]. Butanol’s higher energy density results in a smaller decrease in volumetric fuel econ-
omy than with ethanol. However, the combustion efficiency improvement possible with
ethanol blends does not occur for butanol [37]. Therefore, the calculated fuel economy
for Bu85 is 10.7 L/100km (21.8 mpg), assuming the vehicle efficiency is the same as when
fueled with gasoline. When inputted into the GREET model, the fuel economy values
are convereted to gasoline-equivalent units by multiplying the fuel economy by the ratio
of LHV gasoline/LHV alcohol-gasoline blend.
Chapter 6. LCA of Biobutanol for use in Transportation 89
6.3 Results and Discussion
In this section, the life cycle results for the butanol production scenarios are presented
and compared to the model and literature results for butanol and ethanol.
6.3.1 WTP fossil energy use
Figure 6.24 presents the WTP fossil energy use for the various scenarios. The results for
scenarios with no co-product(s) are broken down into, corn farming (consisting of the
activities associated with feedstock production), production (conversion of the feedstock
to alcohol) and transportation (distribution of the fuel to blending terminals). Butanol
has a high fossil energy input, as can be seen from the bar showing butanol with no co-
products. The inputs are more than double the inputs of ethanol under similar conditions.
The much higher fossil energy is attributed to the lower yields and higher downstream
product recovery energy requirements in butanol production. However, when the DDGS,
acetone, and ethanol co-products are accounted for using the energy allocation method,
the fossil energy used for butanol production is substantially lower. The results of this
study are higher than those of Wu et al. [22] when no co-products are considered due to
the more conservative butanol yields assumed in this study. Figure 6.2 also indicates that
fossil energy for biofuel production are highest followed by the fossil energy required for
feedstock production. Transportation energy contributes a small portion to the overall
fossil energy use.
6.3.2 WTP petroleum use
Figure 6.3 shows the WTP petroleum use for the scenarios listed above. In comparing
Figures 6.2 and 6.3, the petroleum contribution to fossil energy use is small indicating that
butanol (and ethanol) can contribute to reducing petroleum use. The WTP petroleum
input requirements for corn butanol and ethanol are comparable due to petroleum not
being used directly in the biofuel production processes5 but in the activities of corn
farming and transportation which are assumed to be very similar in the life cycles of
4Notes for WTP Figures: Wu (2008) refers to Wu et al. [22]. The scenarios are: ButOH w/DDGS= butanol with DDGS co-product credited by the displacement method; ButOH w/DDGS and EtOH= butanol with DDGS and ethanol credited by the displacement method; ButOH w/DDGS, Acetoneand EtOH = butanol with the three co-products credited by the energy allocation method. The EtOHresults are those for dry mill ethanol in GREET 1.8b
5Natural gas is the main fossil energy source used in the production of biofuels.
Chapter 6. LCA of Biobutanol for use in Transportation 90
0.0
0.5
1.0
1.5
2.0
2.5
ButOH no
credits
ButOH w/
DDGS
Wu (2008)
ButOH
w/DDGS
and EtOH
ButOH
w/DDGS,
Acetone,
and EtOH
Wu (2008)
w/DDGS,
Acetone,
and EtOH
EtOH no
credits
EtOH w/
DDGS
We
ll t
o P
um
p f
os
sil e
ne
rgy
us
e
(MJ
/MJ
of
fue
l p
rod
uc
ed
)
Transport
Production
Farming
WTP with credit
Figure 6.2: WTP fossil energy use (MJ of fossil fuel input/MJ of fuel produced)
the two fuels. Inclusion of co-product credits reduces the WTP petroleum use of corn
butanol by 50% to 75%.
6.3.3 WTP GHG emissions
Figure 6.4 presents the WTP GHG emissions for the various scenarios. A credit is
given for CO2 uptake during feedstock growth, but this is applied only when the biofuel
is combusted in the vehicle as per GREET convention [12]. Therefore, these WTP
results do not include a CO2 uptake credit. Corn butanol WTP results range from
60 to 180 g CO2-eq./MJ. The lower results are those which assume co-product credits.
Results for ethanol are comparable to those of butanol when co-products are considered.
It is interesting to note that fossil energy use in the most optimistic case for butanol
(considering all co-product credits) is greater than the fossil energy use for the ethanol
production with co-product credits, yet the GHG emissions are greater for ethanol than
butanol in the same scenarios. This result is because butanol consumes more fossil
energy in the form of less carbon intensive natural gas (i.e., during alcohol production)
while ethanol consumes more carbon intensive petroleum during transportation. Thus,
Chapter 6. LCA of Biobutanol for use in Transportation 91
0.0
0.1
0.1
0.2
0.2
0.3
ButOH no
credits
ButOH w/
DDGS
Wu (2008)
ButOH
w/DDGS
and EtOH
ButOH
w/DDGS,
Acetone,
and EtOH
Wu (2008)
w/DDGS,
Acetone,
and EtOH
EtOH no
credits
EtOH w/
DDGS
Well
to
Pu
mp
petr
ole
um
en
erg
y u
se
(MJ/M
J o
f fu
el
pro
du
ced
)
Transport
Production
Farming
WTP with credit
Figure 6.3: WTP petroleum energy use (MJ of petroleum input/MJ of fuel produced)
butanol’s ability to be transported by pipelines helps in reducing net GHG emissions.
6.3.4 WTW Fossil Energy Use and GHG Emissions
For the WTW results, only the most optimistic scenarios for butanol (i.e., DDGS, ace-
tone, and ethanol co-products credited by the energy allocation method) and ethanol
(i.e., DDGS co-product credited by the displacement method) are compared to that of
conventional gasoline for E85 and Bu85 blends. Figures 6.5 and 6.6 present the WTW
fossil energy use and GHG emissions per km driven by a passenger vehicle fueled by
conventional gasoline (CG), E85, and Bu85.
The trends for fossil energy use and GHG emissions closely match each other, as is
expected from the strong relation between fossil energy combustion and GHG emission.
E85 and Bu85 can offer reductions in fossil energy use by approximately 37-38% when
compared to CG. The fossil energy use and GHG emissions for E85 and Bu85 are almost
the same. Although not graphed, it is evident that when coproduct credits are not
considered for Bu85, then the fossil energy use and GHG emissions would largely exceed
those of CG.
Chapter 6. LCA of Biobutanol for use in Transportation 92
0
20
40
60
80
100
120
140
160
180
200
ButOH no
credits
ButOH w/
DDGS
Wu (2008)
ButOH
w/DDGS
and EtOH
ButOH
w/DDGS,
Acetone,
and EtOH
Wu (2008)
w/DDGS,
Acetone,
and EtOH
EtOH no
credits
EtOH w/
DDGS
Well
to
Pu
mp
GH
G e
mis
sio
ns
(g C
O2 e
q./
MJ o
f fu
el
pro
du
ced
)
WTP with credit
Transport
Production
Farming
Figure 6.4: WTP GHG emissions (CO2-eq/MJ of fuel produced)
6.4 Conclusions
The fossil and petroleum energy input as well as GHG emissions resulting from the
production and use of butanol from corn assuming three co-product allocation scenarios
were modeled. The co-product allocation method was found to have considerable impact
on the resulting environmental metrics. Uncertainty in the modeling was not examined
in this work but is a critical issue that should be examined in future research.
Butanol had a high fossil energy input, more than double the input to ethanol as
reported in GREET 1.8b if no co-product allocation was included. The much higher fossil
energy was attributed to the lower yields and higher product recovery energy requirements
for butanol production (i.e., primarily distillation energy). When energy was allocated
to the co-products, results were more in line with those of ethanol. Under all scenarios,
the petroleum contribution to fossil energy use was small, indicating that butanol can
contribute to reducing petroleum use in the transportation sector. The butanol scenarios
presented herein generally reported higher energy use and GHG emissions than those of
Wu et al. [22] due to our lower yield assumptions and higher energy requirements for
product recovery. On a WTW basis Bu85 and E85 can offer reductions in fossil energy
Chapter 6. LCA of Biobutanol for use in Transportation 93
0
1
2
3
4
5
CG
E85
Bu85
We
ll to
Wh
ee
l fo
ssil
en
erg
y u
se
(MJ/k
m)
Figure 6.5: WTW fossil energy use (MJ of fossil energy input/km driven)
use and GHG emissions when compared to conventional gasoline. Additional modeling of
butanol production based on state-of-the-art experimental and pilot-scale data is needed,
as is additional data from vehicle testing to facilitate the development of more detailed
WTW analyses.
6.4.1 Recommendations
The environmental performance of corn-butanol is poor because of low butanol yields and
high energy use for recovery (i.e., distillation). The use of lignocellulosic feedstock for
biobutanol production would greatly improve environmental performance, as has been
shown in bioethanol LCA studies [13, 16]. LCA need to be conducted on biobutanol
derived from lignocellulosic feedstocks in order to determine the fuel’s environmental
performance. Future LCA studies should also include additional functional units, such
as life cycle water use, emissions of volatile organic compounds, and the effects of land
use change.
From a biofuel production perspective, it would beneficial to use entire ABE mixture
as a biofuel rather than expending energy to separate the acetone, butanol, and ethanol
products from each other. ABE biofuel is likely to offer a better environmental perfor-
mance due to the higher yield of total solvents and lower energy required for solvent
Chapter 6. LCA of Biobutanol for use in Transportation 94
0
50
100
150
200
250
300
350
CG
E85
Bu85
We
ll to
Wh
ee
l G
HG
em
issio
ns
(g C
O2
eq
./km
)
Figure 6.6: WTW GHG emissions (CO2-eq/km driven)
recovery; however, this needs to be critically assessed using the LCA methodology. The
feasibility of ABE biofuel and its combustion performance in vehicles would first need to
be assessed, as is discussed in the following chapter.
A future research project would determine the net GHG emissions, fossil energy use,
and petroleum use associated with the production of ABE biofuel from corn and lignocel-
lulosic feedstock. An ASPEN model can be created to determine the energy requirements
for the ABE production process, and then the GREET software can be used to calculate
the net environmental impacts associated with feedstock (corn and corn stover) produc-
tion, through to conversion of the feedstock to fuel, and finally distribution of the fuel to
retail refueling stations.
The potential use of ABE biofuel would eliminate the need for sequential distilla-
tion to separate butanol from acetone and ethanol; however, water in the fermentation
broth would still require large amounts of energy to remove via distillation. Alternative
separation processes, such as pervaporation, gas stripping, liquid-liquid extraction, and
adsorption-desorption have been pursued [32]. Of these advanced separation processes,
adsorption-desorption requires the least amount of energy. Therefore, future research
should be directed towards developing a novel adsorbent for ABE removal from the fer-
mentation broth. The ideal adsorbent would have the following characteristics: quick
Chapter 6. LCA of Biobutanol for use in Transportation 95
adsorption kinetics, high capacity, low cost, quick desorption kinetics, and ease of re-
generation. Initially, this research should explore various materials (e.g. silicalite and
activated carbon) as adsorbents for ABE mixtures in water. Particular attention should
be paid to determining the effects of material composition, surface area, and particle size.
Literature Cited
[1] K. Sanderson, “US biofuels: A field in ferment,” Nature, vol. 444, pp. 673–676, 2006.
[2] A. Ragauskas, “The path forward for biofuels and biomaterials,” Science, vol. 311,
pp. 484–489, 2006.
[3] D. Pimentel and T. W. Patzek, “Ethanol production using corn, switchgrass, and
wood; biodiesel production using soybean and sunflower,” Natural Resource Re-
search, vol. 14, pp. 65–76, 2005.
[4] T. Patzek, “Thermodynamics of the corn-ethanol biofuel cycle,” Critical Reviews in
Plant Sciences, vol. 23, pp. 519–567, 2004.
[5] M. S. Graboski, “Fossil energy use in the manufacture of corn ethanol,” Prepared
for the National Corn Growers Association (2002)., Tech. Rep., 2002.
[6] H. Shapouri, J. A. Duffield, and M. Wang, “The energy balance of corn ethanol: An
update,” U.S. Department of Agriculture Agricultural Economic Report No. 814,
Tech. Rep., 2002.
[7] M. E. D. de Oliveira, B. E. Vaughan, and E. J. Rykiel, “Ethanol as fuel: Energy,
carbon dioxide balances, and ecological footprint,” Journal of BioScience, vol. 55,
p. 593, 2005.
[8] S. Kim and B. Dale, “Ethanol fuels: E10 or E85 - life cycle perspectives,” The
International Journal of Life Cycle Assessment, vol. 11, pp. 117–121, 2006.
[9] S. Kim and B. Dale, “Environmental aspects of ethanol derived from no-tilled corn
grain: fossil energy consumption and greenhouse gas emissions,” Biomass and Bioen-
ergy, vol. 28, pp. 475–489, 2005.
96
Literature Cited 97
[10] M. Wu, M. Wang, and H. Huo, “Fuel-cycle assessment of selected bioethanol pro-
duction pathways in the united states,” Centre for Transportation Research Energy
Systems Division,, Argonne National Laboratory, Tech. Rep., 2006.
[11] A. Farrell, R. Plevin, B. Turner, A. Jones, M. Oıhare, and D. Kammen, “Ethanol
can contribute to energy and environmental goals,” Science, vol. 311, pp. 506–508,
2006.
[12] Anonymous. Greenhouse gases, regulated emissions, and energy use in
transportation (greet) computer model. Center for Transportation Re-
search Argonne National Laboratory. Argonne, Illinois. [Online]. Available:
[36] Anonymous, “Model year 2006: fuel economy guide,” U.S. Department of Energy
Office of Energy Efficiency and Renewable Energy, Tech. Rep., 2006. [Online].
Available: http://www.fueleconomy.gov
[37] M. Gautam, D. Martin, and D. Carder, “Emissions characteristics of higher alco-
hol/gasoline blends,” Proceedings of the Institution for Mechanical Engineers, vol.
214, pp. 165–182, 2000.
Chapter 7
An Experimental and Kinetic
Modeling Study of Butanol
Combustion
7.1 Introduction
7.1.1 Engine Studies
There have been several engine studies using n-butanol as a fuel or as a blending agent
with gasoline [1, 2, 3, 4, 5, 6]. In the most notable, Yacoub et al. [6] used gasoline
blended with n-butanol) to fuel a single-cylinder SI engine. They found that the butanol
blends had less knock resistance than neat gasoline. The butanol blends also had reduced
CO and hydrocarbon emissions but increased NOx emissions. This may be due to the n-
butanol blends having a higher flame temperature and earlier spark timing. Of particular
interest to the present study is that the primary oxygenated hydrocarbon emissions were
butanol, formaldehyde and to a lesser extent, acetaldehyde. A study by Miller et al.
successfully operated unmodified gasoline and diesel engines on blends containing 0-20%
butanol in gasoline and 0-40% n-butanol in diesel fuel [7].
A recent study by Wallner et al. [8] studied 10%butanol-gasoline and 10%ethanol-
gasoline blends in a modern direct-injection four-cylinder SI engine at varying engine
speeds. The difference in heat release rate was found to be neglible between the biofuel
blends and neat gasoline. At high engine loads, the butanol blend had a reduced knock
resistance, in agreement with the aforementioned study [6]. Due to differences in energy
100
Chapter 7. Chemical Kinetic Modeling of Butanol Combustion 101
density, the brake specific volumetric consumption was the lowest for gasoline, second
lowest for the butanol blend, and highest for the ethanol blend. The regulated emissions
profiles for the ethanol and butanol blends were comparable, and the study concluded
that butanol use in existing engines is feasible.
7.1.2 Combustion Chemistry Studies
Predictive models provide a better understanding of the combustion performance and
emissions characteristics of biofuel compositions and why they differ from petroleum
derived materials. The development of a butanol model requires understanding of its
fundamental pyrolysis and oxidation kinetics. However, only a few studies have examined
the combustion chemistry of butanol.
The pyrolysis of n-butanol was studied by Barnard using a static reactor [9]. The
author suggests that pyrolysis is initiated by fission at the C3H7-CH2OH bond to pro-
duce the n-propyl radical and hydroxymethyl radical. The hydroxymethyl radical further
decomposes to formaldehyde and a hydrogen radical, while the n-propyl radical decom-
poses to ethylene and a methyl radical. In another study, Roberts measured the burning
velocities of n-butanol using shadowgraph images of the flame cone [10]. The results
indicate that the maximum burning velocity of n-butanol is similar to that of n-propanol
and isopentyl alcohol (i.e., approx. 46 cm/sec). A recent study by McEnally and Pfef-
ferle measured temperature and species in atmospheric-pressure coflowing laminar non-
premixed methane flames doped with one of four isomers of butanol (including n-butanol)
[11]. They concluded that unimolecular dissociation dominated over H-atom abstraction.
This consisted of C-C fission followed by beta-scission of the resulting radicals. Complex
fission involving four-center elimination of water was estimated to account for only 1% of
n-butanol decomposition. The most important measured species included propene and
ethylene. Yang and coworkers [12] studied laminar premixed flames fuelled by one of
four isomers of butanol (including n-butanol). Their results identified the combustion
intermediates in the butanol flames, but did not provide concentration profiles. The
qualitative data provided lends support to the aforementioned dissociation mechanism
[11].
Recently, Moss et al. published a chemical kinetic mechanism for four butanol iso-
mers (i.e., n-butanol, sec-butanol, iso-butanol, and tert-butanol) and validated it against
ignition delay times measured in a shock tube [13]. Under the given experimental con-
Chapter 7. Chemical Kinetic Modeling of Butanol Combustion 102
ditions, the study concluded that n-butanol is mainly consumed by H-atom abstractions
leading to the formation of C3CH7CHO, acetaldehyde, and ethylene, and, to a lesser
extent, propene and 1-butene. It should be noted that the chemical kinetic mechanism
was not validated against species profiles in the shock tube.
Recent work by Dagaut et al. studied the oxidation of 85%butanol-15%gasoline surro-
gate mixtures in a jet stirred reactor (JSR) at 10 atm, equivalence ratios spanning 0.3-2.0,
and temperatures ranging from 770-1270 K [14]. The surrogate mixture was comprised
of iso-octane, toluene, 1-hexene and n-butanol. A novel chemical kinetic mechanism was
derived using mechanisms for each pure component in the butanol-gasoline surrogate
mixture, and it was shown to provide good agreement with the experimental data. The
mechanism indicates that H-abstraction reactions are the main consumption pathway for
the butanol-gasoline surrogate fuel.
The goal of this study is to provide additional experimental data on pure n-butanol
oxidation in several well-defined environments. This study is the first to model detailed
chemistry in an n-butanol flame. Species and temperature profiles are provided for an
n-butanol opposed-flow diffusion flame. The paper also presents species profiles for an
n-butane opposed-flow diffusion flame to elucidate the differences in combustion between
an alkane and an alcohol. In addition, species profiles are presented for n-butanol in a
JSR at 101.325 kPA and 1013 kPa (i.e., 1 atm and 10 atm) with a range of equivalence
ratios (�=0.25-2.0) and temperatures (T)1. Burning velocity data is presented for an n-
butanol premixed laminar flame at various equivalence ratios2. This comprehensive data
set is used to validate an improved chemical kinetic model of n-butanol oxidation.
7.2 Experimental Methods
7.2.1 Opposed-flow Diffusion Flame
A detailed explanation of the experimental opposed-flow diffusion flame and correspond-
ing sampling setup was presented in Chapter 4. Briefly, the setup consists of two identical
flat flame burners with circular burner ports of diameter 25.4 mm, facing each other and
spaced 20 mm apart. A fuel mixture of 94.11% N2 and 5.89% fuel (99% pure n-butanol
or 99% pure n-butane) was fed through the bottom port at a mass flux rate of 0.0131
1The JSR experimental data was obtained by M.J. Thomson, P. Dagaut, and C. Togbe2The laminar flame speed data was obtained by F. Halter and C. Mounaim-Rousselle
Chapter 7. Chemical Kinetic Modeling of Butanol Combustion 103
g/cm2-sec, while an oxidizer mixture of 42.25% O2 and 57.75% N2 was fed through the
top port at a mass flux rate of 0.0126 g/cm2-sec. At these plug flow conditions, the
Reynold’s Number is in the laminar flow regime (i.e. Re < 400), the flame is on the fuel
side of the stagnation plane, and the fuel side strain rate is approximately 33 s−1. An
ultrasonic atomizer was used to spray the liquid fuel into a stream of N2 gas preheated
to 356 K. The gaseous fuel mixture was delivered to the burner via heated stainless steel
tubing. The temperatures of the gases exiting the top and bottom burner ports were
423 K and 356K , respectively. The gas sampling system in these experiments consists
of a quartz microprobe (250 m ID, 300 m OD) connected to a dual-stage pump with
heated heads (388 K) containing PTFE diaphragms. The suction side of the sampling
system consisted of 1/4′′ tubing and a vacuum pressure gauge connected to the quartz
microprobe. An absolute pressure of 4-6 kPa was measured downstream of the micro-
probe. This was sufficient to quench most reactions and ensure accurate data on flame
composition. The compression side of the pump delivered the samples to the analytical
instruments via 1/4′′ stainless steel tubing heated to 388 K.
Analytical techniques used to analyze the species in the sample included: NDIR for
CO and CO2; GC/FID with an HP-Al/S PLOT column for C1 to C5 hydrocarbons; and
GC/FID equipped with a methanizer (i.e., Ni catalyst) and Poraplot-U column for bu-
tanol, butanal, acetaldehyde, and formaldehyde. The precision of species measurements
is estimated to be ± 15%. Temperature measurements were obtained using a 254 �m
diameter wire R-type thermocouple (Pt-Pt/13% Rh) in an apparatus similar to that used
by McEnally et al. [15]. The measured temperatures were corrected for radiation losses.
7.2.2 Jet Stirred Reactor
The JSR experimental setup3. used in this study has been described earlier [16, 17]. The
reactor consists of a 4 cm diameter fused silica sphere equipped with four nozzles of 1
mm ID that promote rapid mixing of the gases as they enter the reactor. Prior to the
injectors, the reactants were diluted with high-purity nitrogen (<100 ppm H2O, <50 ppm
O2, <1000 ppm Ar, <5 ppm H2) and mixed. A high degree of dilution (0.1% mol. of fuel)
was used, reducing temperature gradients and heat release in the JSR. The reactants were
high-purity oxygen (99.995% pure) and high-purity n-butanol (99% pure butanol from
3The JSR setup is located at CNRS, Orleans, France. The experiments were performed by M.J.Thomson, C. Togbe, and P. Dagaut.
Chapter 7. Chemical Kinetic Modeling of Butanol Combustion 104
Aldrich), which was sonically degassed before use. The reactants were preheated before
injection to minimize temperature gradients inside the reactor. A Shimadzu LC10 AD
VP pump with an on-line degasser (Shimadzu DGU-20 A3) was used to deliver the fuel to
an atomizer-vaporizer assembly maintained at 473 K. Good thermal homogeneity along
the vertical axis of the reactor (gradients of approximately 1 K/cm) was observed for each
experiment by thermocouple (0.1 mm Pt-Pt/Rh (10%) located inside a thin-wall silica
tube) measurements. The reacting mixtures were sampled by a fused silica low pressure
sonic probe, and then analyzed online by fourier transform infrared spectroscopy (FTIR)
and off-line after collection and storage in 1 L Pyrex bulbs. Off-line analyses were done
using gas chromatographs equipped with capillary columns (DB-624 and Carboplot-P7),
a thermal conductivity detector (TCD), and a FID.
The experiments were performed at steady state under constant mean residence times
(�) of 0.07 seconds and 0.7 seconds corresponding to constant pressures (P) of 101.325
kPa (1 atm) and 1013 kPa (10 atm), respectively. The reactants were continually flowing
in the reactor, whereas the temperature of the gases inside the JSR was varied stepwise. A
good repeatability was observed in the experiments and reasonable good carbon balance
of 100 ± 15% was achieved.
7.2.3 Laminar Flame Speed Setup
The device to measure the laminar flame speed consists of a stainless steel cylindrical
combustion chamber with an inside volume of 2.4 L4. Two tungsten electrodes linked to a
conventional capacitive discharge ignition system, are used to form the spark gap (2.8mm)
at the center of this chamber. Four windows provide optical accesses into the chamber.
The air-fuel mixture was prepared directly in the chamber by adding the fuel and the
air at appropriate partial pressures to reach the total initial pressure. The pressure and
temperature conditions (i.e., 350 K and 90 kPa) were selected to optimize the saturated
vapour pressure of butanol. The chamber was warmed at the desired temperature and an
electric fan, located inside the chamber, mixed all the gases. Gas phase chromatography
analysis was performed to ensure adequate mixing. High-purity n-butanol was injected
by a gasoline injector. The spray of butanol was rapidly vaporised due to the relative
high temperature and the forced convection created by the fan. A delay before ignition
4The laminar flame speed setup is located at the University of Orleans, France. The experimentswere performed by F. Halter and C. Mounaim-Rousselle.
Chapter 7. Chemical Kinetic Modeling of Butanol Combustion 105
avoided any perturbation during the flame propagation. Measurements were limited to
flames having diameters less than 60 mm, implying that the total volume of burned gases
was less than 0.5%. In this initial part of the flame expansion, the total chamber pressure
can be considered constant. Observation times were lower than 15 ms for all the cases
investigated (depending on the equivalence ratio). A range of equivalence ratios, from
0.8 to 1.2, was tested. The error in the flame speed measurements is ± 2 cm/s.
The image of the flame was obtained by the classical shadowgraphy technique. The
parallel light from an Ar-Ion laser source was created by two plano-convex lenses (25 mm
and 1000 mm focal lengths respectively). The shadowgraphic images were recorded by
using a high speed video CMOS camera5 operating at 6000 frames per second with an
exposure time of 20 �s. The temporal evolution of the expanding spherical flame was
then analyzed to determine the laminar flame speed. A global schematic view of the
system is presented in Figure 7.1.
Figure 7.1: Schematic of the laminar flame speed measurement setup
7.3 Computational Methods
The kinetic modeling for n-butanol oxidation in the three experimental setups was per-
formed using the CHEMKIN modeling package [18]. The JSR was modeled using the
perfectly stirred reactor (PSR) code, the opposed-flow diffusion flame was modeled using
the OPPDIF code, and the laminar flame speed was modeled using the premix flame code
5Photron APX
Chapter 7. Chemical Kinetic Modeling of Butanol Combustion 106
(PREMIX). The inputs to each simulation include a detailed chemical kinetic reaction
mechanism, a dataset of thermochemical properties, and a dataset of transport proper-
ties. These input files are available as supplemental material to the journal publication
of this study [19].
7.3.1 Chemical Kinetic Mechanism
The chemical kinetic reaction mechanism developed here is based on a previously pro-
posed mechanism for 85%butanol-15%gasoline surrogate mixtures in a JSR [14]. The
combustion of n-butanol proceeds via unimolecular initiation and hydrogen abstraction
reactions. The fuel radical species formed are consumed via unimolecular decomposition
(beta-scission) and bimolecular reactions. Isomerization of radical species is also included
in the mechanism. Table 7.1 presents the structure of species produced during the oxi-
dation of n-butanol. Modifications have been made to the original mechanism to provide
better agreement with JSR data at 1 atm and 10 atm and opposed-flow diffusion flame
data. Below is a description of the proposed mechanism.
The Dagaut et al. n-butanol oxidation submechanism [14] was built upon a previ-
ously proposed C1-C4 hydrocarbon mechanism [20, 21, 22]. From the C1-C4 mechanism,
Dagaut et al. added 136 reactions to represent the oxidation of n-butanol and the vari-
ous species formed during its decomposition. Due to the absence of fundamental kinetic
studies on n-butanol unimolecular decomposition reactions and abstraction reactions, the
authors allocated reaction rate constants based on published rate data for structurally
similar hydrocarbons and oxygenates. The reaction rates were allocated to provide better
agreement with JSR data for 85%butanol-15%gasoline surrogate mixtures at 10 atm [14].
The present author (i.e., S.M. Sarathy) has modified Dagaut et al.’s mechanism to
better predict the experimentally measured opposed-flow diffusion flame species profiles.
The revised mechanism consists of 878 reactions involving 118 species. The following is
a description of specific changes made to the Dagaut et al. mechanism.
The simulations using the OPPDIF code were unable to converge upon a solution
using the aforementioned C1-C4 hydrocarbon mechanism. In order to obtain convergence,
the following modifications were made:
∙ removal of the reaction C3H6 + C2H⇀↽ BUTYNE + CH,
∙ replacement of the reaction nC3H7 + (M)⇀↽ C3H6 + H + (M) by the reaction
Chapter 7. Chemical Kinetic Modeling of Butanol Combustion 107
Table 7.1: Chemical structures of species during the oxidation of n-butanol
nC3H7 ⇀↽ C3H6 + H with reaction rate constants proposed by Curran [23].
This study has also made the following modifications to the former n-butanol sub-
mechanism, in order to provide better agreement with the experimental data presented
herein:
∙ The reaction rate constant6 for the reaction C4H9O ⇀↽ H + C3H7CHO has been
changed to
8.89x1010 ⋅ T 0.75 exp
(−21060 cal
mol
RT
)cm3
mol ⋅ s
based on the rate expression recommended by Curran for C3H7O ⇀↽ H + C2H5CHO
[23].
6Units are presented in cal, K, mol, cm3, and s according to the CHEMKIN convention
Chapter 7. Chemical Kinetic Modeling of Butanol Combustion 108
∙ The reaction rate constant for the reaction C4H9OH + O ⇀↽ OH + C4H9O has
been changed to
1.58x1010 ⋅ T 2.00 exp
(−448 cal
mol
RT
)cm3
mol ⋅ s
based on the rate expression recommended by Marinov for the ethanol reaction
C2H5OH +O ⇀↽ OH +C 2H5O [24].
∙ The reaction rate constant for the reaction C4H9OH + H ⇀↽ H2 + C4H9O has been
changed to
5.36x104 ⋅ T 2.53 exp
(−8754 cal
mol
RT
)cm3
mol ⋅ s
based on the rate expression recommended by Park et al. for the ethanol reaction
C2H5OH + H ⇀↽ H2 + C2H5O [25].
∙ The reaction rate constant for the reaction C4H9OH + OH ⇀↽ H2O + C4H9O has
been changed to
7.46x1011 ⋅ T 0.30 exp
(−1634 cal
mol
RT
)cm3
mol ⋅ s
based on the rate expression recommended by Park et al. for the ethanol reaction
C2H5OH + OH ⇀↽ H2O + C2H5O [25].
∙ The reaction rate constant for the reaction C4H9OH + H ⇀↽ H2 + aC4H8OH has
been changed to
2.58x107 ⋅ T 1.65 exp
(−2827 cal
mol
RT
)cm3
mol ⋅ s
based on the rate expression recommended by Marinov for the ethanol reaction
C2H5OH + H ⇀↽ OH +C2H4OH [24].
∙ The reaction rate constant for the reaction C4H9OH + OH ⇀↽ H2O + aC4H8OH
has been changed to
4.64x1011 ⋅ T 0.15 cm3
mol ⋅ s
based on the rate expression recommended by Marinov for the ethanol reaction
C2H5OH + OH ⇀↽ H2O +C2H4OH [24].
Chapter 7. Chemical Kinetic Modeling of Butanol Combustion 109
∙ The original mechanism [14] employed the same reaction rate constants for uni-
molecular dissociation at the C2H5-C2H4OH and C3H7-CH2OH bond sites. How-
ever, the reaction rate constant for the reaction C4H9OH ⇀↽ C2H5 + C2H4OH has
been changed to
5.0x1016 ⋅ T exp
(−86221 cal
mol
RT
)cm3
mol ⋅ s(7.1)
since the BDE of the C2H5-C2H4OH bond is lower than the BDE of the C3H7-
CH2OH due to the proximity of the OH group in the latter.
7.3.2 Thermochemical Data
The original thermochemical data for the butanol related species was calculated using
the software THERGAS [26], based on the group and bond additivity methods proposed
by Benson [27].
7.3.3 Transport Properties
In certain combustion applications, such as the JSR, the overall rate is assumed to
be kinetically controlled since the transport processes occur infinitely fast. Therefore,
the original Dagaut et al. mechanism [14] did not include transport properties of any
species. However, the transport processes are rate-controlling in laminar diffusion flames.
This study obtained the molecular transport parameters for species using a variety of
methods. The transport properties for the majority of compounds were already available
in the previously published C1-C4 hydrocarbon mechanism [20, 21, 22]. In addition, the
transport properties of several species were obtained from a study by Gail et al. on
the combustion of methyl butanoate [28]. The transport properties of species with no
previously published data were determined as follows. For stable species, this study used
the correlations developed by Tee, Gotoh, and Stewart [29], as described in Wang and
Frenklach [30], to calculate the Lennard-Jones collision diameter and potential well depth
using the critical pressure (Pc), critical temperature (Tc), and boiling point (Tb) of the
species. The Pc, Tc, and Tb for stable species were obtained from the NIST Chemistry
WebBook [31]. The polarizability in cubic Angstroms of stable species was obtained from
the CRC Handbook of Chemistry and Physics [32]. The dipole moment was obtained
from [33]. The index factor which describes the geometry of the molecule was determined
Chapter 7. Chemical Kinetic Modeling of Butanol Combustion 110
from the molecular structure. For new radical species, the aforementioned literature data
is not readily available, so the transport properties of their stable counterpart were used.
7.4 Results and Discussion
7.4.1 Jet Stirred Reactor
The JSR7 allows studying n-butanol oxidation in a flameless premixed environment. The
concentration of species at each equivalence ratio and temperature condition in the JSR
was measured by sonic probe sampling and GC and FTIR analyses. The measured
species included hydrogen (H2), oxygen (O2), water (H2O), carbon monoxide (CO), car-
these experiments. The probe 9 had a large inner diameter and short tip which caused
the flame to be drawn towards the fuel port. This probe effect was resolved by moving
to a new probe design 10.
The model performs well qualitatively, in that it well reproduces the shape of the
experimental profiles; however, there is a shift in the measured profiles towards from
the fuel port. In the following discussion, the model’s prediction is considered good if
predicted maximum mole fraction is within a factor 1.5 of the measured maximum mole
fraction. The model performs well at predicting the maximum concentrations of C2H4,
C2H6, 1-C4H8, and CH3CHO. The model moderately overpredicts (1.5-2.5 times) the
concentration of CH4, C2H2, pC3H4, C3H6, and CH2O. There are large overpredictions
(greater than 5 times) of 1,3-C4H6 and C3H7CHO, and large underpredictions of C3H8
(6 times). However, these compounds are only found in small quantities in both the
modeling and experimental results.
Another measure of the model’s qualitative performance is to compare the relative
concentration of species. Although the model poorly predicts the maximum concentra-
9This probe was made at the Department of Chemistry Glass Blowing Shop10The new probe design is described in Chapter 4 and improved results for methyl decanoate opposed-
flow diffusion flames are shown in Chapter 9
Chapter 7. Chemical Kinetic Modeling of Butanol Combustion 124
tions of several species, the relative concentration of species is reasonably well reproduced
by the model. For example, both the experimental data and model predictions show the
minor hydrocarbon species in order of decreasing maximum concentration are C2H4,
This section summarizes the important combustion related properties of several biodiesel
fuels, pure FAME, and standard low sulfur diesel2.
The LHV is a measure of the fuel’s energy density. Diesel engines are capable of
accepting a variation in heating values, so there is no specified LHV in the ASTM D 975
Standard. However, it is beneficial for biodiesel to have a volumetric LHV similar to that
of standard diesel so that differences in fuel economy (i.e., L/100 km) are not experienced
[19]. The volumetric LHV of biodiesel is slightly lower than that of standard diesel, so
fuel economy would be lower [8]. It should be noted that some studies have reported an
improvement in engine efficiency when using biofuels, which offsets the lower volumetric
heating value, and results in no net change in fuel economy [23]. For pure FAME, the
2Information on viscosity, oxidative stability, lubricity, and cold flow characteristics is available in theliterature [21, 16, 8, 22]
Chapter 8. Background 144
LHV increases with chain length [16].
The cetane number is a measure of the fuel’s ignition delay. Diesel combustion requires
the fuel to self-ignite as it is sprayed into the compressed cylinder gas. The self-ignition
leads to the characteristic diesel “knock”, wherein an explosion of premixed air and fuel
causes a rapid heat release and pressure rise. The magnitude of the explosion can be
decreased by shortening the ignition delay time. Higher cetane numbers result in shorter
ignition delay times, and therefore better engine operation. The ASTM D 975 minimum
acceptable cetane value is 40. Table 8.2 presents cetane numbers for common biodiesel
fuels and pure FAME. The iodine value is also shown for biodiesel fuels to indicate
the degree unsaturation. Biodiesel fuels have higher cetane numbers when compared to
standard diesel [8, 21, 22]. Similar to hydrocarbon compounds, the cetane number of
pure FAME decreases with increasing unsaturation and increases with increasing chain
length [16]. Soybean methyl ester and canola methyl ester have lower cetane numbers
than palm oil methyl ester because of the higher unsaturated fatty acid content in the
former [22]. Microalgal oils, which are rich in polyunsaturated fats with four or more
double bonds, have high iodine values and are likely to have depressed cetane numbers.
Current European biodiesel standards limit the iodine value and the concentration of
polyunsaturated FAME, so microalgal biodiesel must undergo hydrogenation3 to meet
current fuel standards [3].
8.3 Biodiesel Exhaust Emissions
There have been a number of studies on the use of biodiesel in CI engines. The goal of
these studies was to determine the effect of using biodiesel and diesel-biodiesel blends on
the emissions, fuel economy, and operation of the engine. Comprehensive review articles
on the use of biodiesel in CI engines and it effects on engine performance and exhaust
emissions have been published recently [8, 23].
The effect of biodiesel on chemical emissions depends on the specific pollutant of
concern, the type of engine used, the engine speed and load, and the biodiesel FAME
composition. The present study is mainly concerned with the role of FAME composition
because detailed chemical kinetic mechanisms can be used to elucidate any chemistry
related effects on combustion emissions. Therefore, this section briefly summarizes the
3Hydrogenations will saturate the double bonds and decrease the iodine value.
Chapter 8. Background 145
Table 8.2: Properties of common biodiesel fuels and pure FAMEa
Biodiesel feedstock or pure FAME Cetane number Iodine value
Beef Tallow 75 33-47
Palm Oil 61 49-55
Rapeseed/Canola Oil 55 110-126
Soybean Oil 49 118-139
Caprylic FAME (C8:0) 33.6 -
Lauric FAME (C12:0) 61.4 -
Myristic FAME (C14:0) 66.2 -
Palmitic FAME (C16:0) 74.5 -
Stearic FAME (C18:0) 86.9 -
Oleic FAME (C18:1) 55 -
Linoleic FAME (C18:2) 42.2 -
a Cetane numbers for biodiesel fuels are from [22, 21] and pure
FAME from [16]. Iodine values are from [24].
effects of biodiesel on diesel engine emissions with a particular focus on the role of FAME
composition4.
Table 8.3 presents the percentage of research publications that report increases, simi-
larities, or decreases in emissions when using biodiesel and diesel fuels [23]. The majority
of studies find that the use of biodiesel reduces the emissions of THC, CO, and PM, and
increases the emissions of NOx. The effect of biodiesel on the emissions of oxygenated
compounds is uncertain since research studies have shown both increases, decreases, and
insignificant differences compared to petroleum diesel.
8.3.1 CO, THC, and Oxygenate Emissions
The emissions of THC decreases when biodiesel is used [23]. Generally, the feedstock used
for biodiesel production does not affect THC emissions. However, research on pure FAME
indicates that THC emissions decrease with increasing fatty acid chain length [25, 26]
and decreasing unsaturation [25]. The most widely accepted reason for the decrease in
4The information presented is from a comprehensive review discussing the effect of biodiesel on dieselengine emissions by Lapuerta et al. [23]. The reader is directed to the original article for a more detaileddiscussion.
Chapter 8. Background 146
Table 8.3: Percent of publications that report changes in emissions for biodiesel [23].
Increase Same Decrease
NOx 85 10 5
PM 3 2 95
THC 1 4 95
CO 2 8 90
THC when compared to diesel fuel is that the oxygen content and higher cetane number
of biodiesel leads to a more complete and cleaner combustion. Rakapoulos has shown
that THC emissions decrease as the oxygen content in the cylinder increases, either by
enriching the oxygen content of the fuel or the air [27]. Increasing the cetane number
also helps reduce THC emissions, and this explains why longer chain and fully saturated
FAME have lower THC emissions.
CO emissions decrease when biodiesel is used [23]. For pure FAME, CO emissions
decrease with increasing chain length [16] and decreasing unsaturation [25, 28]. The de-
crease in emissions when compared to petroleum diesel is because of the higher oxygen
content and cetane number of biodiesel, which promote complete combustion. Cetane
numbers increase with increasing chain length and decreasing unsaturation, so this ex-
plains the decreases in CO emissions observed in longer chain and saturated FAME.
It is widely believed that biodiesel would lead to greater emissions of oxygenated
compounds, such as aldehydes and ketones, because of the fuel bound oxygen in FAME.
However, engine studies have shown varying results, and there is no conclusive evidence
on this matter [23]. One study that measured increased acrolein emissions in several
biodiesel fuels attributed this to the glycerol content of biodiesel [29]. The argument
appears valid since glycerol combustion is shown to produce high levels of acetaldehyde
and acrolein [30, 31].
8.3.2 PM and NOx emissions
It is nearly unanimous that PM emissions decrease when biodiesel is used [23]. The effects
of biodiesel feedstock and FAME molecular structure on PM emission is uncertain. The
EPA reported that PM emissions were lower for beef tallow methyl ester than soybean
methyl ester [28], which suggests that higher degrees of unsaturation may increase PM
emissions. However, studies on pure FAME have shown no correlation between FAME
Chapter 8. Background 147
molecular structure (i.e., chain length and unsaturation) and PM emissions [16, 25]. The
main factor affecting PM emissions is the oxygen content of the fuel, which is constant
for various biodiesel fuels and FAME [25]. There are a number of reasons explaining
the reductions in PM observed when using biodiesel. Biodiesel has no aromatic com-
pounds, so its use displaces the highly sooting aromatic compounds typically found in
petroleum diesel fuel. In additon, the oxygen atoms in the FAME bonds to carbon atoms,
and therefore prevents carbon atoms from participating in soot growth reactions. The
oxidation of FAME and other oxygenated intermediates also forms OH radicals, which
readily attack unsaturated hydrocarbons and prevent their participation in soot growth
reactions. The FAME combustion chemistry study presented in this dissertation offers
additional insights into the role of the ester moiety in reducing soot.
The use of biodiesel in diesel engines leads to a slight increase in NOx emissions
[23]. The EPA reported that NOx emissions were lower for beef tallow methyl ester
than soybean methyl ester [28], which suggests that higher degrees of unsaturation may
increase NOx emissions. For pure FAME, NOx emissions increase with decreasing chain
length and increasing unsaturation [16, 25].
Explanations for the observed increase in NOX in biodiesel are open for speculation.
The three prevailing mechanisms explaining NOx formation in combustion engines are: 1.
Figure 9.13 displays the measured and predicted species and temperature profiles ob-
tained in the opposed-flow diffusion flame. The experimental results (solid symbols)
show that the MD concentration begins decreasing quickly at a distance of 5 mm from
the fuel port. As the fuel is consumed, the CO and CO2 concentrations begin rising.
All of the MD is consumed at a distance of approximately 8.25 mm from the fuel port,
which corresponds closely the visually observed flame front. Both the temperature and
CO2 concentrations reach their maximum at approximately 9.5 mm from the fuel port.
Just before the flame front, at around 7.75 mm from the fuel port, the concentrations
of hydrocarbon species reach their maximum. Besides CO and CO2, the most abundant
measured species are C2H4, C2H2, CH4, and C3H6.
It should be noted that these results do not display the shift in the measured species
profiles towards from the fuel port, which was observed previously in n-butanol experi-
ments (refer to Chapter 7). This shift was caused by a poor probe design which induced
flow field disturbances. The results for MD used a new probe design with did not disturb
the flame 3.
3The different probe designs are described in Chapter 4
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 176
This probe effect was resolved by moving to a new probe design 4.
The model’s prediction of temperature species profiles in the opposed-flow diffusion
is also shown in Figure 9.13 5. The model reproduces the experimentally measured
temperature profile very well. The reactivity of MD is also well predicted by the model.
The maximum concentration of CO2 is underpredicted by approximately 0.3%, while the
maximum concentration of CO is underpredicted by approximately 0.1%.
The model performs well qualitatively, in that it well reproduces the shape of the
experimental profiles and the points of maximum measured concentrations. In the fol-
lowing discussion, the model’s prediction is considered good if predicted maximum mole
fraction is within a factor 1.5 of the measured maximum mole fraction. The model per-
forms well at predicting the maximum concentrations of CH4, C2H6, C3H4, C3H6, 1-C4H8,
C8H16, C5H10, C6H12, C7H14, and 1,3-C4H6. The model moderately underpredicts (i.e.,
1.5-2 times) the maximum concentration of C2H4 and overpredicts the concentration of
C2H2. Both model and the experimental data indicate that the concentration of 1-alkenes
decreases with increasing carbon number (e.g., [C2H4]>[C3H6]>[1-C4H8]>[C5H10] etc.).
A reaction path analysis was performed for MD at 1040 K, the temperature at which
approximately where 50% of fuel is consumed. Approximately 98% of the fuel is consumed
via H-atom abstraction by H atoms (57%), OH radicals (5%), and CH3 radicals (30%).
Abstraction is favoured for H atoms bonded to the methoxy carbon (20%) and the �
carbon (16%), and then secondary (8% each) and tertiary H atoms (3%).
Figure 9.14 shows the fate of the MDMJ radical which is formed via H-atom abstrac-
tion from the methoxy carbon. Interestingly, �-scission is not favored for this radical
because it requires the breaking of a strong C-O bond. Instead, the MDMJ radical un-
dergoes isomerization to form the MD2J radical (61%) and the MD3J radical (36%).
The fate of the MD2J radical is shown in Figure 9.15, while that of the MD3J radical is
discussed later.
As shown in Figure 9.15, H-atom abstraction from the � carbon leads to the forma-
tion of the MD2J radical, which undergoes �-scission (99%) to form a 1-heptyl radical
and methyl 2-propenoate. The 1-heptyl radical eventually leads to the formation of
ethylene, 1-pentene, 1-butene, and the radicals C3H7 and C2H5, while the fate of methyl
2-propenoate is discussed in the next section. The measureable levels of ethylene, 1-
4The new probe design is described in Chapter 4 and improved results for methyl decanoate opposed-flow diffusion flames are shown in Chapter 9
5Appendix C contains larger versions of the graphs shown here
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 177
0
200
400
600
800
1000
1200
1400
1600
1800
2000
0 2 4 6 8 10 12 14 16 18 20
DISTANCE FROM FUEL PORT (mm)
MO
LA
R C
ON
CE
NT
RA
TIO
N (
PP
M)
measured
corrected
predicted
0%
2%
4%
6%
8%
10%
0 2 4 6 8 10 12 14 16 18 20
DISTANCE FROM FUEL PORT (mm)
MO
LA
R C
ON
CE
NT
RA
TIO
N (
%) CO2
CO
MD
0
2000
4000
6000
8000
10000
12000
14000
0 2 4 6 8 10
DISTANCE FROM FUEL PORT (mm)
MO
LA
R C
ON
CE
NT
RA
TIO
N (
PP
M) C2H4
C2H2
CH4
0
200
400
600
800
1000
1200
1400
1600
1800
2000
0 2 4 6 8 10
DISTANCE FROM FUEL PORT (mm)
MO
LA
R C
ON
CE
NT
RA
TIO
N (
PP
M)
C2H6
1-C4H8
C3H6
C8H16
0
50
100
150
200
250
300
350
0 2 4 6 8 10
DISTANCE FROM FUEL PORT (mm)
MO
LA
R C
ON
CE
NT
RA
TIO
N (
PP
M) C3H4
C5H10
C6H12
C7H14
1,3-C4H6
0
100
200
300
400
500
600
700
800
900
1000
0 2 4 6 8 10
DISTANCE FROM FUEL PORT (mm)
MO
LA
R C
ON
CE
NT
RA
TIO
N (
PP
M) C2H4O
CH2O
CH2CO
Figure 9.13: Experimental and computed profiles obtained from the oxidation of MD in
an atmospheric opposed-flow flame (1.8% MD, 42% O2).
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 178
MDMJ
C8
C7
C6
C5
C4
C3
C2
1
OC
O
C9
C10
- 3%
+ C9H
19CO
OCH
2
MD2J
- 58%
- 39%
C8
C7
C6
C5
C4
C 3
C2
1
OC
O
C9
C10
MD3J
.....C
3
C
2
1
OC
O
Figure 9.14: Reaction pathway diagram for consumption of the MDMJ radical in the
opposed-flow diffusion flame at T=1040 K.
pentene, and 1-butene in the flame experiments agree with this reaction path analysis.
Approximately 56% of the fuel is consumed via abstraction of secondary H atoms from
the #4 to #9 carbon atoms. An example of the subsequent reaction pathways is shown
in Figure 9.16 for the MD4J radical. Approximately, 56% of the radical decomposes to
form 1-octene and the ME2J radical, while another 44% leads to the 1-pentyl radical
and methyl 4-pentenoate. The fate of the 1-pentyl radical is displayed in Figure 9.15,
which identifies ethylene as the final product species. The model indicates that methyl
4-pentenoate is consumed via unmimolecular decomposition to form an allyl radical (i.e.,
C3H5) and the ME2J radical. The ME2J either isomerizes (90%) to form the MEMJ
radical or decomposes (10%) to form ketene and a formaldehyde (i.e., via the methoxy
radical). The MEMJ radical goes to form formaldehyde and carbon monoxide (i.e., via
the acetyl radical). In this way, most of the MD4J radical goes to form formaldehyde,
carbon monoxide, ethylene, 1-octene, and ketene, all of which were measured in the
opposed-flow diffusion flame.
Other radicals formed via abstraction of secondary H atoms from other carbons in the
alkyl chain follow a similar path as MD4J. The radicals decompose via two routes, one
leading to an alkene and a methyl ester radical, and the other forming an unsaturated
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 179
MD2J
C8
C7
C6
C5
C4
C3
C
2
1
OC
O
C9
C10
- 99%
CC O
C
O
CC
CC
C
CC
+
MP2D
- 17% - 38%
CC
CC
C
CH2
CH2
CC
CC
CC
C
- 100% - C2H5
CC
CC C
- 77%
- C2H4
1-C5H
10
CC
C
CC
CC
- C3H7
- 44%
- 85%
CC
CC
1-C4H
8
- C3H7
C2H
4
Figure 9.15: Reaction pathway diagram for consumption of the MD2J radical in the
opposed-flow diffusion flame at T=1040 K.
methyl ester and an alkyl radical. The alkyl radicals eventually result in the formation of
1-alkenes, and as shown previously the concentration of 1-alkenes decreases with increas-
ing carbon number. The model predicts that the unsaturated methyl esters are consumed
mainly by unimolecular decomposition to form an allyl radical and a saturated methyl
ester radical that is three carbon atoms shorter (e.g., methyl 6-heptenoate decomposes
to the radical methyl 4-butanoate, methyl 7-octenoate decomposes to the radical methyl
4-pentanoate, etc.). These methyl ester radicals with the radical site on the terminal car-
bon undergo �-scission to form ethylene and smaller methyl ester radicals. The process
continues until the radical site nears the carbonyl group and the radical decomposes to
a low molecular weight oxygenated species.
9.7.3 Oxygenated Species
Table 9.2 presents the maximum predicted and measured mole fractions of several unsat-
urated methyl esters, aldehydes, enals, ketones, ketenes, and enols. The measured and
predicted concentrations of oxygenated product species can add insight into the role of
the ester moiety during combustion. The following is a discussion of several important
oxygenated species and their chemistry in the flame.
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 180
MD4J
C8
C7
C6
C5
C
4
C3
C2
1
OC
O
C9
C10
- 90%
C
OC
OME2J
- 44%
- C5H11
CC
CC O
C
O
MF4D
- 73%
- C3H5
- 56%
C OC
OMEMJ
- 10%
O
CH3
CH2
C
O
+
- 100%
OCH
2+
CC
O
- 98%
- 99%- H
- CH3
OC
CC
CC
CCC
C+
1-C8H
16
Figure 9.16: Reaction pathway diagram for consumption of the MD4J radical in the
opposed-flow diffusion flame at T=1040 K.
The model performs well at predicting the maximum concentrations of CH2CO, but
overpredicts the maximum concentrations of formaldehyde (CH2O) and underpredicts
acetaldehyde + ethenol (C2H4O). The measured concentration of CH2CO (i.e., ketene)
was 413 ppm, and the predicted concentration is 315 ppm. This is the first time ketene
concentrations have been measured in combustion studies of FAME. Figure 9.16 displays
the primary pathway which forms 83% of the ketene in the flame at 1040 K; the methyl
ester radical, ME2J, undergoes �-scission to form ketene and methoxy radical (CH3O).
Therefore, it is observed that the ester moiety contributes to the formation of ketene.
Approximately 86% of formaldehyde is formed via the decomposition of various
methyl ester radicals with a radical on the methoxy site, such as MDMJ, MEMJ, and
MP2DMJ. As shown in Figure 9.16 for MEMJ, these fuel radicals undergo �-scission to
form formaldehyde. Therefore, it is observed that the ester moiety contributes to the
formation of formaldehyde.
The maximum predicted concentration of formaldehyde is more than 2 times greater
than the measured concentration. This discrepancy can be attributed to either exper-
imental errors or modeling inaccuracies, so both are discussed here. The experimental
measurements for formaldehyde were performed using a GC/FID equipped with a meth-
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 181
Table 9.2: maximum Measured and Predicted Concentration of Oxygenated Species
Species Name Measured (ppm) Predicted (ppm)
in Mechanism
Formaldehyde CH2O 319 924
Ketene CH2CO 413 319
Ethanal+Ethenol CH3CHO+C2H3OH 100 15
Propanal C2H5CHO <LOD <1
2-propenal C2H3CHCO <LOD 44
2-propanone C3H6O <LOD 9
Methyl 2-propenoate MP2D <LOD 1149
Methyl 3-butenoate MB3D <LOD 120
Methyl 4-pentenoate MF4D <LOD 37
Methyl 5-hexenoate MH5D <LOD 29
anizer, which break the aldehyde’s C-O bond and replaces it with a C-H bond. This
method for detecting formaldehyde has yielded good results in JSR studies of FAME
[2, 22, 25, 51]. However, extractive sampling measurements in flames [2, 22, 52, 53] have
yielded similar discrepancies between measured and predicted formaldehyde, and one
group [53] suggests that formaldehyde may be lost due to polymerization in the sampling
lines.
Modeling overpredictions of formaldehyde concentrations have also been observed in
opposed-flow diffusion flames of MB [2, 4]. For MD at 1040 K, the present model predicts
that 86% of the formaldehyde is formed via decomposition of the metyl ester radical with
a radical on the methoxy site. A sensitivity analysis on formaldehyde at 1040 K revealed
that the predicted formaldehyde concentration is sensitive to the decomposition rates of
MDMJ, MEMJ, and MP2DMJ. The current rate estimate for the decomposition of these
radicals to formaldehyde are rough estimates, so detailed studies may reveal better rate
constants. Furthermore, the present model indicates that approximately 20% of CO in
the flame is formed directly from formaldehyde via the HCO radical. Since CO concen-
trations are underpredicted by approximately 1000 ppm, the 500 ppm overprediction of
formaldehyde by the model may be corrected by improving rate constants for reactions
linking CO and CH2O. Such modifications are beyond the scope of the present study,
and would require further investigation into fundamental experimental and theoretical
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 182
rate studies.
Both the model and experiments indicate that the C3 oxygenated species propanal,
2-propenal (i.e. acrolein), and 2-propanone (i.e. acetone) are not important product
species during the combustion of MD. However, a number of compounds are predicted
at appreciable levels in the flame, but were not detected in the experiments. It should
be noted that the flame sampling and analytical methods were capable of detecting
and quantifying these compounds. It is unlikely that these compounds were lost in the
sampling system since condensation and reactions in the sampling lines were minimized
by operating at low sampling pressures and using transfer lines heated to 200 ∘C.
The major discrepancy is for unsaturated methyl ester species. The experiments
did not measure detectable levels of any unsaturated methyl esters despite having the
appropriate sampling and analytical methods in place to measure them. Oxygenated
hydrocarbons typically create lower signal responses on an flame ionization detector, but
typically this affects detection levels of low molecular weight oxygenates (e.g., formalde-
hyde, acetaldehyde, etc.). Microliter injections of these unsaturated FAME verified that
the analytical instrument used in this study was suitable for their detection. Furthermore,
the accurate measurement of methyl decanoate in the flame samples indicates that the
sampling apparatus was adequate for high molecular weight FAME. It should be noted
that unsaturated methyl esters have been measured in experimental studies of methyl
hexanoate in a JSR at 10 atm[25], albeit at low concentrations.
Methyl 2-propenoate (i.e., MP2D) is the unsaturated FAME predicted in the high-
est concentration. Figure 9.17 displays the production and consumption pathways for
MP2D in the opposed-flow diffusion flame at 1040 K. 94% of the MP2D is formed via
�-scission of various methyl ester radicals with a radical site on the � carbon (e.g., MD2J,
etc.). The rest is formed via decomposition of the MP3J radical. Methyl 2-propenoate is
then consumed via H-atom abstraction reactions leading to the methyl ester radical with
a radical on the methoxy carbon (i.e., MP2DMJ). H abstraction is favoured from the
methoxy site because the H atoms bonded to the vinylic carbons have higher BDEs. The
MP2DMJ radical undergoes �-scission to ultimately form acetylene, carbon monoxide,
formaldehyde. High concentrations of methyl 2-propenoate are predicted because multi-
ple pathways lead to various methyl ester radicals with a radical site on the � carbon,
and these form methyl 2-propenoate faster than it can be consumed via H-atom abstrac-
tion reactions. The rate parameters for the methyl 2-propenoate consumption have been
determined based on analogies with saturated methyl ester molecules and unsaturated
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 183
hydrocarbons. Fundamental experimental and theoretical studies on H abstraction and
unimolecular decomposition reactions may improve the predicted concentration of methyl
2-propenoate.
Another explanation for the discrepancy between the model and predicted values is
possible decomposition of methyl 2-propenoate upon contact with hot surfaces in the
sampling line. Such a mechanism is plausible given the unstable nature of unsaturated
FAME, especially those of low molecular weight. It is possible that FAME are reacting
in the sampling line to form acetaldehyde+ethenol compounds which were measured
in appreciable quantities, but not predicted to be significant by the model. Further
investigation of the reactivity of methyl 2-propenoate at the conditions in the sampling
line (i.e., T=388 K, P=6-8 kPa) is required.
MP2D
CC O
C
O
.....C
3
C 2
1
OC
O
- CH2O
+ 94%
CC O
C
O
- H- 90 %
- 100 %
CC
C
O
MP2DMJ
- CxHy
- 100 %
- C2H3
OC
MX2J
Figure 9.17: Reaction pathways for the formation and consumption of methyl 2-
propenoate in the opposed-flow diffusion flame at T=1040 K.
The Fate of the Ester Moiety
Previous studies on methyl butanoate combustion discussed the fate of the ester moiety
at high temperatures. When applied to different experimental conditions (e.g., flames,
premixed reactors, etc.), the MB models predict that the methoxycarbonyl radical is an
important combustion intermediate. We ran the opposed-flow diffusion flame simulations
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 184
for MB using the experimental conditions of Gail et al. [2] and the mechanism of Dooley
et al. [4] to calculate how much of the fuel decomposes to form the methoxycarbonyl
radical at high temperatures. The results indicate that approximately 23% of the fuel
ends up in the methoxycarbonyl radical, as shown in Figure 9.18. In Figure 9.3, it
was shown that the methoxycarbonyl radical primarily decays to form a methyl radical
and CO2. From a soot reduction standpoint this decarboxylation of the ester is not an
efficient use of fuel-bound oxygen because two oxygen atoms are bonded to one carbon
atom. It would be more efficient if the ester moiety led to the production of CO since each
oxygen atom in the ester group would sequester one carbon atom from participating in
the production of soot. Many researchers have hypothesized that the long chain FAME in
biodiesel undergo similar reaction pathways as methyl butanoate, and therefore the ester
moiety in biodiesel is not as efficient at supressing soot compared to other oxygenated
moieties (e.g., aldehydes, ethers, alcohols, etc.). The present study on methyl decanoate
offers additional insights to the aforementioned hypothesis.
MB
C4
C3
C2
1
OC
O
- 27%
- H- 16%
- H
CC
C OC
OMB3J
- 84%
- C3H6
C
OC
O
CH3OCO
C4
C3
C
2
1
OC
O
CC O
C
O
MB2J
MP2D
- 84%
sum of all pathways
- 40%
- CH3
Figure 9.18: Reaction pathways leading to the formation of the methoxycarbonyl radical
in the opposed-flow diffusion flame at T=1030 K given the experimental and modeling
conditions of Gail et al. [2].
The above analysis on MB indicates that the radical sites on the � and � carbons can
lead to the methoxycarbonyl radical. At 1040 K, approximately 8% of MD is consumed
via H-atom abstraction from the � carbon leading to the MD3J radical. In addition, 8%
of MD leads to the MD3J radical via the MDMJ radical, as shown in Figure 9.14 . This
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 185
is 30% less than what was observed for methyl butanoate [2] under similar conditions, or
even for methyl hexanoate in a JSR at 950 K [25]. Similarly, 16% of MD is consumed via
H-atom abstraction from the � carbon whereas it is 27% for MB. These pathways in MD,
and for longer chain FAME too, becomes less important because the number of H-atom
abstraction sites increases with chain length. Furthermore, after absrtraction from the �
carbon, the resulting MD3J radical undergoes �-scission at equal rates to form either i.
1-nonene plus the methoxycarbonyl radical or ii. methyl 3-butenoate plus and a 1-hexyl
radical. However, the analogous radical in MB (i.e., MB3J) strongly favours the route
leading to the methoxycarbonyl radical, since �-scission forming methyl 3-butenoate is
thermodynamically unfavourable for the MB3J radical. The importance of fuel radical
isomerization reactions are become more important as the size of the FAME increases.
Similar to the previous studies on MB, the present MD mechanism predicts that the
methoxycarbonyl radical decomposes to form CO2. It is estimated that approximately
6% of the fuel ends up forming the CO2 via the methoxycarbonyl radical, which is much
lower than previous estimates based on MB (i.e., 23%). This number is expected to be
even lower in longer chain FAME because of the increased number of potential pathways
of fuel consumption.
It should be noted that the amount of fuel that ends up in the methoxycarbonyl
radical depends on the experimental conditions (i.e., temperature, pressure, mixture
fraction, etc.). The above analysis applies to high temperature oxidation where H-atom
abstraction is predominant. Simulations should be conducted for conditions in which uni-
molecular decomposition is predominant for additional insights. In any case, this study
suggests that decarboxylation of the ester group in long chain FAME is not significant.
In fact, much of the original fuel bound oxygen leads directly to oxygenated species such
as carbon monoxide, formaldehyde, and ketene, wherein one oxygen atom is bonded to
one carbon atom. Therefore, the soot reducing efficiency of long chain FAME may be
better than previously believed.
9.7.4 Jet Stirred Reactor
The proposed skeletal mechanism for MD was also validated against experimental JSR
data for RME at �=1.0, P=101.325 kPa, �=1.0 s [26]. Since MD is a smaller molecule
than RME, the inlet mole fraction of MD was proportionally increased to match the inlet
carbon flux of RME, as described by Herbinet et al. [27]. The comparison between the
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 186
model predictions and experimental data is shown in Figures 9.19 and 9.20.
The skeletal mechanism performs similarly to the detailed MD mechanism proposed
by Herbinet et al. [27]. The concentrations of CO2, CO, C2H4, O2, H2, CH4, and C3H6
are well predicted by the proposed skeletal mechanism. However, the concentrations of
1-C4H8, 1-C5H10, and 1-C6H12 are overpredicted by the model, which was also observed
by Herbinet et al. [27]. The prediction of alkenes is incorrect because RME consists of
longer chain FAME with various degrees of unsaturation, while MD is fully saturated
and has a small chain. Although there is no data available, RME is likely to have larger
carbon chains leading to the formation of larger 1-alkenes (e.g., >C8) than would MD;
therefore, MD over predicts the concentrations of the smaller 1-alkenes (e.g., C4-C6).
1.E-05
1.E-04
1.E-03
1.E-02
790 890 990 1090 1190
Temperature (K)
Mo
le F
ractio
n
O2 COCO2 H2CH4 C2H2co2 detailed o2 detailed
Figure 9.19: Comparison of proposed MD skeletal mechanism and experimental data for
RME in a JSR at �=1.0, P=101.325 kPa, �=1.0 s [26].
9.8 Conclusions and Recommendations
This study is the first to present experimental data for methyl decanoate combustion
that can be used for validating chemical kinetic mechanisms. The combustion of MD
in the opposed-flow diffusion flame generates typical hydrocarbon combustion products
(e.g., CO, CO2, CH4, C2H4, etc.). Of particular interest is the production of C5-C8 1-
alkenes which are formed after �-scission of fuel radicals. The production of low molecular
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 187
1.E-07
1.E-06
1.E-05
1.E-04
1.E-03
790 890 990 1090 1190
Temperature (K)
Mo
le F
ractio
n
C3H6
1-C4H8
1-C5H10
1-C6H12
Figure 9.20: Comparison of proposed MD skeletal mechanism (lines with symbols) and
experimental data (symbols) for RME in a JSR at �=1.0, P=101.325 kPa, �=1.0 s [26].
weight oxygenated compounds such as formaldehyde, ketene, and isomers of C2H4O is
also observed. The absence of acrolein and acetone in the measured data suggests that
FAME combustion does not produce these species.
The experimental data presented herein was used to validate an improved skeletal
mechanism for the high temperature oxidation of MD. Initially, modifications were made
to a previously proposed detailed mechanism for MD, and then an improved DRG al-
gorithm was performed to create a skeletal mechanism. This new skeletal mechanism
provides excellent qualitative prediction of experimentally measured species and temper-
ature profiles in the MD opposed-flow diffusion flame. This study highlights the effective-
ness of the DRG method in producing a mechanism that is computationally practical for
one-dimensional flame simulations yet also retains a high level of chemical fidelity. The
only major discrepancy between the proposed mechanism and the experimental data was
for methyl 2-propenoate, and it is suggested that the experimental sampling apparatus
be reassessed for accurate measurement of this compound.
The validated MD mechanism provides many new insights into the combustion chem-
istry of FAME. Firstly, the combustion of long chain FAME proceeds in much the same
way as a long chain alkane; however, the ester moiety does introduce unique combustion
pathways. It was shown that MD displays different combustion pathways than smaller
Chapter 9. Chemical Kinetic Modeling of Biodiesel Combustion 188
methyl esters, such as MB, due to the greater number of possible reaction pathways. This
study suggests that fuel bound oxygen in MD (i.e., the ester moiety) leads primarily to
the formation of carbon monoxide and low molecular weight oxygenated compounds, and
very little actually ends up in the form of CO2. Therefore, under the conditions of this
experimental and modeling study, the oxygen in the ester moiety does a good job of
sequestering carbon from participating in soot production. However, these findings need
to be verified across other fundamental combustion platforms with varying temperatures,
pressures, mixture fractions, and mixing conditions.
The proposed mechanism also indicates that unsaturated methyl esters are an im-
portant intermediate in the combustion of saturated FAME. The present mechanism
derives kinetic information for unsaturated methyl esters from analogous reaction path-
ways for alkenes. This approximation needs to be improved via fundamental studies on
the thermochemical properties of unsaturated FAME. Ab initio calculations for unsatu-
rated FAME will not only improve mechanisms for saturated FAME, but they will help
build comprehensive mechanisms for real biodiesel, which is a mixture of saturated and
unsaturated FAME.
Future experimental and modeling research should be directed towards unsaturated
FAME because current biodiesel fuels from soybean and canola oil, as well as next gen-
eration biodiesel from microalgal oil, have a high concentration of unsaturated FAME.
As shown in Figure 9.1 the unsaturated FAME typically found in biodiesel have double
bonds far away from the carbonyl group (e.g. methyl 9-octadecenoate, methyl 9,12-
octadecadienoate). However, as shown in this study, low molecular weight unsaturated
methyl esters with double bonds near the carbonyl group (i.e., methyl 2-propenoate,
methyl 3-butenoate, etc.) are important combustion intermediates. Therefore, this thesis
study recommends that future research be directed towards understanding the individual
effects of i. double bonds which are near the carbonyl group by using methyl 2-propenoate
and methyl 2-octenoate as surrogate compounds, and ii. double bonds which are far the
carbonyl group using methyl 9-nonenoate as as a surrogate compound6
6These surrogate compounds are selected because they represent the types of molecules encounteredin FAME combustion and they are readily available from chemical suppliers.
Literature Cited
[1] E. Fisher, W. Pitz, H. Curran, and C. Westbrook, “Detailed chemical kinetic mecha-
nisms for combustion of oxygenated fuels,” Proceedings of the Combustion Institute,
vol. 28, no. 2, pp. 1579 – 1586, 2000.
[2] S. Gail, M. Thomson, S. Sarathy, and S. Syed, “A wide-ranging kinetic modeling
study of methyl butanoate combustion,” Proceedings of the Combustion Institute,
vol. 31, pp. 305–311, 2007.
[3] W. Schwartz, C. McEnally, and L. Pfefferle, “Decomposition and hydrocarbon
growth processes for esters in non-premixed flames,” Journal of Physical Chemistry
A, vol. 110, no. 21, pp. 6643–6648, June 2006.
[4] S. Dooley, H. J. Curran, and J. M. Simmie, “Autoignition measurements and a
validated kinetic model for the biodiesel surrogate, methyl butanoate,” Combustion
and Flame, vol. 153, no. 1-2, pp. 2–32, April 2008.
[5] W. K. Metcalfe, S. Dooley, H. J. Curran, J. M. Simmie, A. M. El-Nahas, and M. V.
Navarro, “Experimental and modeling study of C5H10O2 ethyl and methyl esters,”
Journal of Physical Chemistry A, vol. 111, no. 19, pp. 4001–4014, May 2007.
[6] S. M. Walton, M. S. Wooldridge, and C. K. Westbrook, “An experimental investi-
gation of structural effects on the auto-ignition properties of two C-5 esters,” Pro-
ceedings of the Combustion Institute, vol. 32, no. Part 1, pp. 255–262, 2009.
[7] A. Farooq, D. F. Davidson, R. K. Hanson, L. K. Huynh, and A. Violi, “An ex-
perimental and computational study of methyl ester decomposition pathways using
shock tubes,” Proceedings of the Combustion Institute, vol. 32, no. Part 1, pp. 247–
253, 2009.
189
Literature Cited 190
[8] L. K. Huynh, K. C. Lin, and A. Violi, “Kinetic modeling of methyl butanoate in
shock tube,” Journal of Physical Chemistry A, vol. 112, no. 51, pp. 13 470–13 480,
December 2008.
[9] A. M. El-Nahas, M. V. Navarro, J. M. Simmie, J. W. Bozzelli, H. J. Curran, S. Doo-
ley, and W. Metcalfe, “Enthalpies of formation, bond dissociation energies and re-
action paths for the decomposition of model biofuels: Ethyl propanoate and methyl
butanoate,” Journal of Physical Chemistry A, vol. 111, no. 19, pp. 3727–3739, May
2007.
[10] A. Osmont, M. Yahyaoui, L. Catoire, I. Goekalp, and M. T. Swihart, “Thermochem-
istry of c-o, (co)-o, and (co)-c bond breaking in fatty acid methyl esters,” Combustion
and Flame, vol. 155, no. 1-2, pp. 334–342, October 2008.
[11] A. Osmont, L. Catoire, I. Goekalp, and M. T. Swihart, “Thermochemistry of c-c
and c-h bond breaking in fatty acid methyl esters,” Energy & Fuels, vol. 21, no. 4,
pp. 2027–2032, July-August 2007.
[12] A. Osmont, L. Catoire, and I. Goekalp, “Thermochemistry of methyl and ethyl esters
from vegetable oils,” International Journal of Chemical Kinetics, vol. 39, no. 9, pp.
481–491, September 2007.
[13] C. J. Hayes and D. R. Burgess, Jr., “Exploring the oxidative decompositions of
methyl esters: Methyl butanoate and methyl pentanoate as model compounds for
biodiesel,” Proceedings of the Combustion Institute, vol. 32, no. Part 1, pp. 263–270,
2009.
[14] L. K. Huynh and A. Violi, “Thermal decomposition of methyl butanoate: Ab initio
study of a biodiesel fuel surrogate,” Journal of Organic Chemistry, vol. 73, no. 1,
pp. 94–101, January 2008.
[15] T. Vaughn, Hammill, H. M., M., and A. Marchese, “Ignition delay of bio-ester fuel
droplets,” SAE Technical Paper Series, vol. 2006-01-3302, 2006.
[16] K. HadjAli, M. Crochet, G. Vanhove, M. Ribaucour, and R. Minetti, “A study of
the low temperature autoignition of methyl esters,” Proceedings of the Combustion
Institute, vol. 32, no. Part 1, pp. 239–246, 2009.
Literature Cited 191
[17] C. Westbrook, W. Pitz, and H. Curran, “Chemical kinetic modeling study of the
effects of oxygenated hydrocarbons on soot emissions from diesel engines,” Journal
of Physical Chemistry A, vol. 110, no. 21, pp. 6912–6922, June 2006.
[18] P. A. Glaude, W. J. Pitz, M. J. Thomson, and R. Pitz, “Chemical kinetic model-
ing of dimethyl carbonate in an opposed-flow diffusion flame,” Proceedings of the
Combustion Institute, vol. 30, no. 1, pp. 1111 – 1118, 2005.
[19] J. P. Szybist, A. L. Boehman, D. C. Haworth, and H. Koga, “Premixed ignition
behavior of alternative diesel fuel-relevant compounds in a motored engine experi-
ment,” Combustion and Flame, vol. 149, no. 1-2, pp. 112–128, April 2007.
[20] P. Pepiot-Desjardins, H. Pitsch, R. Malhotra, S. Kirby, and A. Boehman, “Structural
group analysis for soot reduction tendency of oxygenated fuels,” Combustion and
Flame, vol. 154, pp. 191–208, 2008.
[21] S. M. Sarathy, S. Gaeil, S. A. Syed, M. J. Thomson, and P. Dagaut, “A comparison
of saturated and unsaturated C4 fatty acid methyl esters in an opposed flow diffusion
flame and a jet stirred reactor,” Proceedings of the Combustion Institute, vol. 31, no.
Part 1, pp. 1015–1022, 2007.
[22] S. Gail, S. M. Sarathy, M. J. Thomson, P. Dievart, and P. Dagaut, “Experimental
and chemical kinetic modeling study of small methyl esters oxidation: Methyl (e)-
2-butenoate and methyl butanoate,” Combustion and Flame, vol. 155, no. 4, pp.
635–650, December 2008.
[23] Anonymous, “Draft Technical Report EPA420-P-02-001,” United States Environ-
mental Protection Agency, Tech. Rep., 2002.
[24] S. Garner, R. Sivaramakrishnan, and K. Brezinsky, “The high-pressure pyrolysis of
saturated and unsaturated c7 hydrocarbons,” Proceedings of the Combustion Insti-
tute, vol. 32, pp. 461–467, 2009.
[25] G. Dayma, S. Gail, and P. Dagaut, “Experimental and kinetic modeling study of
the oxidation of methyl hexanoate,” Energy & Fuels, vol. 22, no. 3, pp. 1469–1479,
May-June 2008.
[26] P. Dagaut, S. Gail, and M. Sahasrabudhe, “Rapeseed oil methyl ester oxidation over
extended ranges of pressure, temperature, and equivalence ratio: Experimental and
Literature Cited 192
modeling kinetic study,” Proceedings of the Combustion Institute, vol. 31, no. Part
2, pp. 2955–2961, 2007.
[27] O. Herbinet, W. J. Pitz, and C. K. Westbrook, “Detailed chemical kinetic oxidation
mechanism for a biodiesel surrogate,” Combustion and Flame, vol. 154, no. 3, pp.
507–528, August 2008.
[28] T. Lu and C. K. Law, “Toward accommodating realistic fuel chemistry in large-
scale computations,” Progress in Energy and Combustion Science, vol. 35, no. 2, pp.
192–215, April 2009.
[29] K. Seshadri, T. Lu, O. Herbinet, S. B. Humer, U. Niemann, W. J. Pitz, R. Seiser,
and C. K. Law, “Experimental and kinetic modeling study of extinction and ignition
of methyl decanoate in laminar non-premixed flows,” Proceedings of the Combustion
Institute, vol. 32, no. Part 1, pp. 1067–1074, 2009.
[30] C. McEnally, U. Koylu, L. Pfefferle, and D. Rosner, “Soot volume fraction and
temperature measurements in laminar nonpremixed flames using thermocouples,”
Combustion and Flame, vol. 109, no. 4, pp. 701–720, June 1997.
[31] R. J. Kee, F. M. Rupley, J. A. Miller, M. E. Coltrin, J. F. Grcar, E. Meeks, H. K.
Moffat, A. E. Lutz, G. Dixon-Lewis, M. D. Smooke, J. Warnatz, G. H. Evans,
R. S. Larson, R. E. Mitchell, L. R. Petzold, W. C. Reynolds, M. Caracotsios, W. E.
Stewart, P. Glarborg, C. Wang, C. L. McLellan, O. Adigun, W. G. Houf, C. P. Chou,
S. F. Miller, P. Ho, P. D. Young, D. J. Young, D. W. Hodgson, M. V. Petrova, and
K. V. Puduppakkam. (2006) CHEMKIN release 4.1. San Diego, California.
[32] H. Curran, “Rate constant estimation for C1 to C4 alkyl and alkoxyl radical decom-
position,” International Journal of Chemical Kinetics, vol. 38, no. 4, pp. 250–275,
April 2006.
[33] D. Healy, H. J. Curran, J. M. Simmie, D. M. Kalitan, C. M. Zinner, A. B. Barrett,
E. L. Petersen, and G. Bourque, “Methane/ethane/propane mixture oxidation at
high pressures and at high, intermediate and low temperatures,” Combustion and
Flame, vol. 155, pp. 441–448, 2008.
Literature Cited 193
[34] W. Tsang and R. Hampson, “Chemical kinetic databse for combustion chemistry:
Part I. Methane and related compounds,” Journal of Physical Chemical Reference
Data, vol. 15, 1986.
[35] T. Lu and C. Law, “A directed relation graph method for mechanism reduction,”
Proceedings of the Combustion Institute, vol. 30, pp. 1333–1341, 2005.
[36] ——, “On the applicability of directed relation graph to the reduction of reaction
mechanisms,” Combustion and Flame, vol. 146, pp. 472–483, 2006., 2006.
[37] C. K. Westbrook, W. J. Pitz, O. Herbinet, H. J. Curran, and E. J. Silke, “A com-
prehensive detailed chemical kinetic reaction mechanism for combustion of n-alkane
hydrocarbons from n-octane to n-hexadecane,” Combustion and Flame, vol. 156,
no. 1, pp. 181–199, January 2009.
[38] P. Dagaut, M. Reuillon, M. Cathonnet, and D. Voisin, “High-pressure oxida-
tion of normal-decemberane and kerosene in dilute conditions from low to high-
temperature,” Journal de Chimie Physique et de Physico-Chimie Biologique, vol. 92,
no. 1, pp. 47–76, January 1995.
[39] E. R. Ritter and J. W. Bozzelli, “THERM: Thermodynamic property estimation
for gas phase radicals and molecule,” International Journal of Chemical Kinetics,
vol. 23, pp. 767–778, 1991.
[40] S. Benson, Thermochemical Kinetics, 2nd edition. Wiley, New York, 1976.
[41] R. Sumathi and W. H. G. Jr., “Oxygenate, oxyalkyl and alkoxycarbonyl thermo-
chemistry and rates for hydrogen abstraction from oxygenates,” Physical Chemistry
Chemical Physics, vol. 5, pp. 3402–3417, 2003.
[42] L. Tee, S. Gotoh, and W. Stewart, “Molecular parameters for normal fluids - lennard-
jones 12-6 potential,” Industrial & Engineering Chemistry Fundamentals, vol. 5,
no. 3, p. 356, 1966.
[43] H. Wang and M. Frenklach, “Transport properties of polycyclic aromatic hydrocar-
bons for flame modeling,” Combustion and Flame, vol. 96, no. 1-2, pp. 163–170,
January 1994.
Literature Cited 194
[44] P. Linstrom and W. Mallard. (2005) NIST chemistry webbook, NIST standard
reference database number 69. National Institute of Standards and Technology,.