University of Southern California Dissertations and Theses

End-gas autoignition investigations using confined spherically expanding flames

(USC Thesis Other)

End-gas autoignition investigations using confined spherically expanding flames

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content End-gas autoignition investigations using confined
spherically expanding flames.

BY

ROBERT LAWSON

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

AUGUST 2021

ii

Acknowledgements
It has been a tortuous path full of ups and downs and I am sincerely grateful to everyone
who has immensely contributed along this journey. Much gratitude goes to Professor Fokion N.
Egolfopoulos, for offering an opportunity to conduct research and pursue my PhD under his
tutelage. His guidance, encouragement and support will always be appreciated and respected. I
would especially like to thank Professors Ronney, Bermejo-Moreno and Jessen for fruitful
discussions, ideas and serving on my committee.
Genuine gratitude goes to all colleagues, namely, Drs. Hugo Burbano, Christodoulos
Xiouris, Jagannath Jayachandran, Vyaas Gururagan, Abtin Ansari, for their mentorship. I would
particularly like to thank Dr. Vyaas Gururagan for the collaborative work on end-gas autoignition
and teaching me the basics of numerical work. Special thanks and appreciation go to Ashkan
Movaghar, for the many days of frustration, failure, help, fruitful discussions, and research on
autoignition and laminar flame speeds of C1-C4, C5-C10 hydrocarbons; Gracias. Additional thanks
to Anguo, Steven and Hiba. Additional thanks to Professor Alon Grinberg Dana of Technion –
Israel Institute of Technology for the collaborative work on RMG.
I would like to acknowledge Drs. Allen Aradi and Tushar Bera of Shell Global Solutions
(US) for sponsoring this project during the past four years and asking tough questions to push the
scope of the project.
To my parents and siblings who have loved, sacrificed and ingrained honesty, dedication
and hard work, a heartfelt appreciation and thank you go to them. Kofi Gyan, Professor Tim
Johnson, of the University of Michigan, Aunt Gladys, Dr Amanfo, Korle-Bu Community Chapel,
Ghana, Qodesh Family Church, Los Angeles, and Yayra, who have contributed their quota by way
of support are also appreciated.
Last but most important, I am forever indebted to my Heavenly Father and Jesus, my
Savior.

In everything give thanks: for this is the will of God in Christ Jesus concerning you.
1 Thessalonians 5:18

iii

Table of Contents
Acknowledgements ............................................................................................................................. ii
List of Tables....................................................................................................................................... vi
List of Figures .................................................................................................................................... vii
Abstract ............................................................................................................................................... x
Chapter 1: Introduction ........................................................................................................................ 1
1.1 Motivation and significance ........................................................................................................... 1
1.2 Literature Review ........................................................................................................................... 3
Experimental configurations for autoignition phenomena ....................................................... 3
Computational Efforts to study autoignition ............................................................................ 4
Confined spherically expanding flame configuration................................................................ 5
Gasoline components/additive effects on autoignition ............................................................ 9
1.3 Objectives and overview .............................................................................................................. 10
Chapter 2: Background ....................................................................................................................... 12
2.1 Laminar Flame Theory.................................................................................................................. 12
Premixed Flames ................................................................................................................... 12
Laminar Flame Speed ............................................................................................................ 13
2.2 Hydrocarbon Combustion ............................................................................................................ 15
Fuel Oxidation ....................................................................................................................... 16
Low Temperature Chemistry and NTC ................................................................................... 17
2.3 Ignition Delay Time (IDT) .............................................................................................................. 20
Chapter 3: Experimental Approach ................................................................................................... 21
3.1 Experimental Setup ...................................................................................................................... 21
3.2 Confined spherical flame methodology ........................................................................................ 23
3.3 Fuel Handling and Vaporization System ....................................................................................... 25
3.4 Hot-Spot Ignition ......................................................................................................................... 27
Chapter 4: Numerical Methodology ................................................................................................... 28
4.1 Freely propagating flames ............................................................................................................ 28
4.2 End-gas autoignition and sensitivity analysis ................................................................................ 29
4.3 Spherically Expanding Flames (Direct Numerical Simulations) ...................................................... 30
4.4 Automatic Mechanism Generator ................................................................................................ 31
4.5 Kinetic Models ............................................................................................................................. 33
Chapter 5: Autoignition of reacting mixtures at engine-relevant conditions using confined
spherically expanding flames .............................................................................................................. 34

iv

5.1 Introduction ................................................................................................................................. 34
5.2 Experimental approach ................................................................................................................ 35
5.3 Numerical Approach .................................................................................................................... 36
5.4 Characterization of end-gas autoignition ...................................................................................... 36
5.5 Characteristic ignition delay time and sensitivity analysis ............................................................. 43
5.6 Unsteady effects of pressure on kinetics ...................................................................................... 48
5.7 Conclusion ................................................................................................................................... 51
Chapter 6: Assessment of observables from end-gas autoignition spherically expanding flame
experiments .......................................................................................................................................... 53
6.1 Introduction ................................................................................................................................. 53
6.2 Experimental parameter space and numerical approach .............................................................. 54
6.3 Characterization of end-gas autoignition observables .................................................................. 54
1
st
stage ignition point ........................................................................................................... 54
Maximum heat release during 1
st
stage ignition .................................................................... 56
Measure of 1
st
stage ignition heat release ............................................................................. 59
6.4 Concluding remarks ..................................................................................................................... 60
Chapter 7: Preliminary investigations of the pressure rise rate influence on end-gas reactivity ...... 62
7.1 Introduction ................................................................................................................................. 62
7.2 Experimental approach and parameter space .............................................................................. 62
7.3 Numerical approach..................................................................................................................... 63
7.4 Thermal characteristics of the end-gas ......................................................................................... 63
7.5 Effect of flame burning rate on end-gas reactivity ........................................................................ 66
7.6 Concluding remarks ..................................................................................................................... 69
Chapter 8: Fuel additive effects on end-gas autoignition in spherically expanding gasoline flame
experiments .......................................................................................................................................... 71
8.1 Introduction ................................................................................................................................. 71
8.2 Experimental approach ................................................................................................................ 75
8.3 Parameter space .......................................................................................................................... 75
8.4 Modeling approach ...................................................................................................................... 76
Gasoline fuel composition and its properties ......................................................................... 76
Ignition delay time ................................................................................................................ 78
Kinetic model generation and refinement ............................................................................. 78
8.5 Fuel additive effects on experimental observables ....................................................................... 80
8.6 Ignition delay time predictions ..................................................................................................... 91

v

8.7 Concluding remarks ..................................................................................................................... 93
Chapter 9: Conclusions and Recommendations ................................................................................. 95
9.1 Concluding Remarks .................................................................................................................... 95
9.2 Recommendations for Future Work ............................................................................................. 97
References …... …… …… … … … … … … … … … … … … … … … … … … … …... … … … … … … … … … … … … … … … … … … … … … … . .…… …99

vi

List of Tables
Table 1.1 Summary of features and conditions for autoignition phenomena studies ([10]) ..........4
Table 3.1 Summary of thermodynamic conditions investigated................................................. 24
Table 5.1 Parameter space considered in the present investigation ............................................ 35
Table 5.2 Parameter space considered for present investigation with corresponding 1-D laminar
flame speeds simulations ........................................................................................................... 37
Table 5.3 Characteristic values, C, at 1
st
and 2
nd
stage ignition .................................................. 41
Table 5.4 IDT computation for dynamic and stagnation thermodynamic values ........................ 42
Table 5.5 Computed and measured characteristic ignition delay times from sensBVP and
experiment ................................................................................................................................ 45
Table 6.1 Experimental thermodynamic condition and mixture composition used for current
investigation .............................................................................................................................. 54
Table 7.1 Thermodynamic conditions considered in the present investigation ........................... 63
Table 8.1 Thermodynamic conditions and mixture summary considered in the present
investigation. ............................................................................................................................. 76
Table 8.2 Key properties of gasoline base-fuel. ......................................................................... 76
Table 8.3 RMG Model nomenclature and size. ......................................................................... 80
Table 8.4. Parameter space considered in the present investigation. ......................................... 90

vii

List of Figures
Figure 1.1 Future energy consumption predictions......................................................................1
Figure 1.2 Compression of the unburned gas induced by CSEF propagation. Color map indicates
the temperatures of both burned and unburned gas ......................................................................7
Figure 1.3 Schematic diagram of constant volume spherical combustion bomb used to determine
explosion limits ...........................................................................................................................7
Figure 1.4 Pressure-time history in the presence of autoignition .................................................8
Figure 2.1 Structure of a premixed flame .................................................................................. 13
Figure 2.2 Structure of the ideal flame in the laminar flame speed definition ............................ 14
Figure 2.3 Temperature dependent process of fuel oxidation [Courtesy: Prof. H.J. Curran]...... 17
Figure 2.4 Key pathways in low-temperature hydrocarbon ignition [Courtesy: Prof. H.J. Curran]
................................................................................................................................................. 18
Figure 2.5 Calculated and measured ignition delay times in hydrocarbon-air mixtures .............. 20
Figure 3.1 Schematic of the spherical chamber configuration ................................................... 21
Figure 3.2 Schematic of the spherical chamber configuration ................................................... 22
Figure 3.3 Experimental pressure-time (P-t) history and (b) dP/dt as a function of P without
autoignition ............................................................................................................................... 25
Figure 3.4 Volume of injected fuel as a function its pressure .................................................... 26
Figure 5.1 Experimental results for Mixture 1: (a) temporal evolution of P and (b) d(lnP)/dt as a
function of P/Po for Po = 3 atm ( ─), Po = 4 atm ( ─), Po = 5 atm ( ─), Po = 6 atm ( ─) .................. 38
Figure 5.2 Experimental results for Mixture 2: (a) temporal evolution of P and (b) d(lnP)/dt as a
function of P/Po for Po = 3 atm ( ─), Po = 4 atm ( ─), Po = 5 atm ( ─), Po = 6 atm ( ─) .................. 39
Figure 5.3 Computed radial variation of T for Mixture 1 at Po = 6 atm with corresponding time
sequence (a) a - 36.12 ms, b - 47.84 ms, c - 54.31 ms, d – 58.60 ms, e – 61.80 ms, (b) f – 53.72
ms, g – 57.20 ms, h - 59.94 ms, i – 61.58 ms, j -61.96 ms, k - 62.2040 m .................................. 40
Figure 5.4 Comparison of experimental ( …) and computed ( ─) dP/dt as a function of P for
Mixture 1 at Po = 6 atm ............................................................................................................. 42
Figure 5.5 Computed P-Tu functions for an unreactive ( ─) and reactive ( ─) end-gas for Mixture
1 at Po = 6 atm .......................................................................................................................... 44
Figure 5.6 Computed temperature profile for Mixture 1 at Po = 6 atm with 𝜏𝐶𝐼𝐷𝑇 . Insert shows
a section of the zoomed temperature profile starting before 1
st
stage ignition to after 2
nd
stage
ignition ..................................................................................................................................... 45

viii

Figure 5.7 Computed temperature profile, T ( ─) and DME ( ─), OH ( ─) and CH2O ( ─) mass
fractions in the unburned mixture as a function of the normalized time for Mixture 1 at Po = 6
atm ............................................................................................................................................ 46
Figure 5.8 Ranked LSC of 𝜏𝐶𝐼𝐷𝑇 to kinetics for Mixture 1 at Po = 3 (blue) and 6 atm (orange) 47
Figure 5.9 Ranked logarithmic sensitivity coefficient of 𝜏𝐶𝐼𝐷𝑇 (blue) and 𝜏𝐶𝐼𝐷𝑇 1 (red) to
kinetics for Mixture 1 at Po = 6 atm ........................................................................................... 48
Figure 5.10 (a) P-t history, (b) Tu-t variation and (c) Net rate of progress of CH3OCH2O2
production from sensBVP for Mixture 1 at Po = 25 atm, Tu,o = 650 K for dP/dt = 7.33 atm/ms
( ─), dP/dt = 14.67 atm/ms ( ─), dP/dt = 36.67 atm/ms ( ─), dP/dt = 58.67 atm/ms ( ─),
dP/dt = 73.33 atm/ms ( ─). ......................................................................................................... 50
Figure 6.1 Experimental pressure gradient for Mixture 1 in Table 6.1 during the compression
stage of CSEF propagation. Point ‘a’ – 1
st
stage ignition point (Pign_1); Point ‘b’ – dP/dtmax;
Hatched Area ‘c’ – Measure of 1
st
stage exothermicity. ............................................................. 55
Figure 6.2 Computed temperature profile T and species mass fraction in the unburned mixture as
a function of the normalized time. ............................................................................................. 56
Figure 6.3 (a) Computed thermodynamic pressure and normalized heat release rate as a function
of unburned temperature for Mixture 1 (b). Section of computed thermodynamic pressure and
normalized heat release rate as a function of unburned temperature for Mixture 1 at 1
st
stage
ignition...................................................................................................................................... 57
Figure 6.4 Reaction path analysis at T = 760 K, T = 854 K and T = 880 K for Mixture 1. ......... 59
Figure 7.1 Experimental rate of change of pressure as a function of P/Po for Mixtures 1 - 4. ..... 64
Figure 7.2 Computed P-Tu functions for an frozen unburned gas for Mixtures 1 – 4 in Table 7.1.
................................................................................................................................................. 65
Figure 7.3 Comparison of computed Su vs Tu for flames of Mixtures 1 - 4 in Table 7.1 for a
chemically frozen unburned gas ................................................................................................ 66
Figure 7.4 Computed P-Tu functions for Mixtures 1 – 4 in Table 7.1 ........................................ 67
Figure 7.5 Computed temperature, OH and HO2 mass fractions profiles in the unburned mixture
as a function of the normalized pressure for 100% He (0%Ar) [ ─] and 70% He (30% Ar) [- -]. . 68
Figure 7.6 Ranked logarithmic sensitivity coefficient for Mixtures 1-4 in Table 7.1 ................. 69
Figure 8.1 Mass percentages of the fuel components. ............................................................... 77
Figure 8.2 Ignition characteristics for Mixture 4 for Po = 9 atm and To = 493 K during the
compression stage of CSEF propagation. Point ‘a’ – 1
st
stage ignition point (Pign_1); Point ‘b’ –
1
st
stage pressure rise rate; Point ‘c’ – dP/dtmax; Point ‘d’- 2
nd
stage ignition point (Pign_2); Point
‘e’ – 2
nd
stage pressure rise rate; Hatched Area ‘f’ – Measure of 1
st
stage exothermicity; τ1 –
Characteristic 1
st
stage ignition delay time. ................................................................................ 82

ix

Figure 8.3 Scaled 1
st
and 2
nd
stage ignition point for all fuels ignition at (a) To = 468 K, ϕ = 1.0
and (b) To = 493 K, ϕ = 0.8. ....................................................................................................... 83
Figure 8.4 To = 493 K, ϕ = 0.8 (a) Section of experimental pressure rise rates up to 1
st
stage
ignition (b) Scaled characteristic time τc during 1
st
stage pressure rise. (c) Section of
experimental pressure rise rates during both ignition stages (d) Scaled τc during 2
nd
stage pressure
rise. ........................................................................................................................................... 84
Figure 8.5 To = 468 K, ϕ = 1.0 (a) Section of experimental pressure rise rates up to 1
st
stage
ignition (b) Scaled characteristic time τc during 1
st
stage pressure rise. (c) Section of
experimental pressure rise rates during both ignition stages (d) Scaled τc during 2
nd
stage pressure
rise. ........................................................................................................................................... 85
Figure 8.6 Scaled maximum dP/dt at 1
st
stage ignition (a) To = 468 K, ϕ = 1.0 (b) To = 493 K,
ϕ = 0.8. ...................................................................................................................................... 86
Figure 8.7 Scaled measure of heat release as a function of 1
st
stage ignition delay time (a)
To = 468 K, ϕ = 1.0 (b) To = 493 K, ϕ = 0.8. .............................................................................. 87
Figure 8.8 Scaled measure of heat release and 1
st
stage ignition delay time as a function of the
estimated RON at To = 493 K, ϕ = 0.8. ...................................................................................... 89
Figure 8.9. Experimental comparison of the Basefuel and 2-EHN for (a) Test Mixtures 1 and 2
at To = 468 K (b) Test Mixtures 3 and 4 at To = 493 K. .............................................................. 90
Figure 8.10. Predicted ignition delay time modeling for mixtures in air for gasoline surrogate
/additive blend ϕ = 1.0, P = 30 atm. ........................................................................................... 92

x

Abstract
Propagation of a confined spherically expanding flame induces isentropic compression that
can culminate in autoignition and/or detonation under conducive thermodynamic conditions. This
relatively simple technique measures a distinct ‘characteristic ignition delay time’ to identify the
onset of autoignition and complements other established legacy approaches. The present study
details this methodology by examining the experimental autoignition characteristics of reactive
dimethyl ether (DME)-oxygen-inert-mixtures. Experimental results displayed the classic two-
stage ignition typical of dimethyl-ether oxidation at low temperatures with 1
st
stage ignition
occurring approximately 3.6 times the initial pressure.
Direct numerical simulation tools were used to model with results adequately capturing the
appropriate physics of unsteady flame propagation, end-gas reactivity upon compression, end-gas
autoignition and heat loss to the walls, all in a bid to aid in the interpretation of the experimental
results. Furthermore, 0-D thermochemical evolution simulations, sensitivity analysis and heat
release calculations using a detailed DME kinetic model were performed to assess the controlling
reactions of end-gas autoignition and provide substantial insight into the unsteady flame
propagation effects on end-gas kinetics.
Upon demonstration of this approach, a methodology was developed to screen and predict
the mechanistic effects of fuel additives/components on the ignition propensity of reactive
gasoline-additive (2-ethylhexyl nitrate, aniline, decalin, 1,3,5-trimethylbenzene) mixtures using
this technique. Distinctive features from the two-stage ignition process end-gas autoignition for
these reactive gasoline-additive mixtures were characterized in order to gain relevant information,
applicable to engine testing for screening purposes.

xi

The role of additives on end-gas autoignition at low-temperatures was probed by
developing detailed chemical kinetic models of gasoline surrogate - fuel additives using an open-
source automatic reaction mechanism generator software package for qualitative predictions
through ignition delay time simulations. Specifically, aniline, a known anti-knock gasoline
additive was found to effectively suppress the 2
nd
stage ignition by a factor of 3. Simulation
predictions closely followed experimental trends and highlighted the versatility of using this
technique in gleaning information towards engine applications while demonstrating the immense
capabilities of Reaction Mechanism Generator to explore the qualitative performance and chemical
kinetics of fuel additives on autoignition.

1

Chapter 1: Introduction
1.1 Motivation and significance
Global energy demands based on current and projected energy consumption forecast a
50% increase by 2050 [1] with fossil fuels providing 77% of the energy demands as seen in
Figure 1.1. Despite nuclear and renewable energy sources gaining traction in recent years, due to
their environmentally benign nature, fossil fuels still hold a firm advantage due to their high
energy density. With wide-ranging applications of the latter alongside its associated challenges,
chief amongst them being pollution, the need to gain insights into the fundamental aspects of
combustion science and engine technology is paramount.

Figure 1.1 Future energy consumption predictions.

Practical combustion systems such as internal combustion engines (ICE) and gas turbines typically
operate at high pressures, temperatures and under turbulent conditions to achieve power and
efficiency requirements. Scientific inquiry into fundamental combustion processes such as
turbulent flame propagation, engine knock and pollutant-formation in practical engine combustor
systems to develop computational predictive tools can prove a daunting task due to the complex

2

nonlinear coupling of thermodynamics, chemical kinetics, fluid mechanics and transport
phenomena [2] as well as the resolution of various length and time scales. Thus, high fidelity
kinetic data from laboratory scale canonical experimental configurations offer an alternative to
reduce the complexity and serve as a basis for developmental work.
Meritorious low-dimensional laboratory scale experimental configurations such as the
shock tube [3], Rapid Compression Machines (RCMs) [4] and flow reactors [5] have principally
been used for chemical kinetics and ignition studies to develop chemical models, whiles stagnation
flow flames, burner stabilized flames and spherically expanding flame have found their usefulness
in laminar flame speed measurement [6] for testing and/or validation purposes. These established
approaches which permit the isolation of relevant physics for in-depth parametric studies, have
produced kinetic data for engine developmental work.
Furthermore, their well-characterized initial and boundary conditions can be effectively
modeled with computational codes utilizing chemical models that have been validated against
these kinetic data. Chemical models ideally incorporate detailed descriptions of chemistry,
transport and thermodynamic properties which serve as input into three-dimensional direct
numerical simulations of combustion.
From a practical stance, spark ignition (SI) engines traditionally operate by spark-igniting
a piston-compressed gasoline fuel and air mixture at the top of the cylinder. Combustion and heat
release take place by a turbulent flame consuming the unburned mixture with the simultaneous
burned gases expanding in the process. However, abnormal combustion does occur in some cases
as the piston nears the end of compression resulting in autoignition and eventually knocking
combustion [7]. This spontaneous ignition phenomenon is strictly governed by the reactive
mixtures’ thermo-chemical history and typically characterized by the fuel’s Research Octane

3

Number (RON). This phenomenon has led to an aggressive drive to develop antiknock additives
blended to gasoline fuels to increase their ‘autoignition resistance’ and thereby prolong normal
combustion.
1.2 Literature Review
Experimental configurations for autoignition phenomena
SI engines in a bid to attain efficiency goals and conform to stringent societal regulations
have gravitated towards direct injection, low-temperature combustion, lean burning strategies,
engine downsizing and boosting [8]. Low-temperatures are particularly attractive due to the
reduction in pollutant formation and the increase in engine thermal efficiency. However, a
drawback associated with low-temperatures is the possibility of knocking combustion. The need
to gain insights into its controlling phenomena i.e. thermodynamic conditions, equivalence ratios
( ϕ), and low-temperature chemistry (LTC) at engine-relevant conditions has long been recognized
and has spurred scientific inquiry by extensive experimental, theoretical and computational efforts
to examine this phenomenon.
To date, experimental investigations of autoignition phenomena have chiefly been carried
out in “legacy” combustion facilities such as shock tubes, RCMs and motored engines. Shock tubes
and RCMs make use of good measurement (pressure, temperature, species) access to interrogate
the region responsible for many of the fuel-specific effects during low-temperature combustion in
engines including the negative temperature coefficient (NTC) region. The motored engine, which
essentially is a single cylinder engine but with the capability of varying the compression ratio, is
mostly used to study autoignition chemistry, chemical kinetics and ignition delay time (IDT) [9].
Table 1.1 summarizes the time scales, thermodynamic conditions characteristic of the above
experimental approaches.

4

Shock Tube RCM Motored Engine
Time scales (ms) 0.01 – 2 2-150 1-10
Pressure (bar) 2 – 80 5-80 5 – 40
Temperature (K) 800 – 2500 400 – 1200 400 – 900
Operation Single shot, unsteady Single shot, unsteady Multi-shot, unsteady
Flow Conditions Laminar (transition to
turbulent)
Laminar, turbulent Turbulent
Advantages High pressures/
temperatures,
instantaneous
compression
Inexpensive, high
pressure/
temperatures,
pressure history
similar to engines
Inexpensive,
moderate pressures
and temperatures
Disadvantages Single shot, boundary
layers, test times,
expensive to build and
run, inhomogeneous
ignition, fluid mechanics
interferences
Single shot, heat
transfer, expensive to
build and maintain,
fluid mechanics
interferences,
inhomogeneous
ignition
Limited range,
turbulent, high
residual
concentrations,
cyclical variation.
Table 1.1 Summary of features and conditions for autoignition phenomena studies ([10])

Despite their well-documented drawbacks such as boundary layer growth [11], shock
bifurcation [12], shock boundary layer interaction [13][14], inhomogeneous ignition [15],
detonation transition [16][17][18], heat loss [19], residual gas and cycle-to-cycle variation [20],
which compromise the fidelity of experimental data, remarkable improvements have been made
to quantify and/or reduce these non-idealities to large extents to produce scientific-worthy data
[3][20]. Furthermore, these experimental tools with their characteristic measuring strategies and
observations have provided key fundamental insights into the processes of autoignition.

Computational Efforts to study autoignition
Much computational effort has seen substantial progress devoted to researching key
physical mechanisms of autoignition, its prediction, influence of hot spots, detonation
development and pressure wave-chemistry interactions. The effects of the unburnt gas

5

characteristics on the reactive front propagation have been investigated in the following studies:
[21][22][23][24][25][26]. In a noteworthy study, Yu and Chen [27] numerically studied the end-
gas combustion mode development based on the chamber size and prevailing thermodynamic
conditions using A-SURF [28][29][30], a 1-D compressible flow code for flame propagation and
ignition with compression, shock and detonation wave capturing capabilities. The authors
identified three modes of end-gas combustion that depended on chamber lengths and initial
conditions. Furthermore, the ensuing pressure oscillations were also dependent on the mode of
autoignition. In a follow-up study, Chen et al. [31] numerically investigated the effect of the
combustion speed on end-gas autoignition and knock with results indicating the avoidance of
autoignition or knock with increased flame propagation speed or smaller chamber sizes. These
studies have provided fundamental insights into the physics of end-gas autoignition.
Confined spherically expanding flame configuration
Combustion at low temperatures is commonly associated with relatively long timescales
for chemical kinetics due to the slow peroxy kinetics. In some of these ‘legacy’ configurations, the
timescales required for autoignition to occur at low-temperatures may compromise the measuring
area due to boundary layer growth and heat loss effects which inadvertently affect the assumed
adiabatic core homogeneity [32][33][34].
While these phenomena are undesirable, the boundary layer effects have far-reaching
consequences as ignition has been observed to originate close to the wall, i.e. the boundary layer
rather than the adiabatic core. Griffiths and co-workers [35] identified that ignition could first
occur at a location that is at a lower temperature relative to its surroundings for fuel-air mixtures
that displaying NTC. Numerical simulations showed that reactivity commenced near the wall with
a lower temperature compared to the adiabatic core. Exothermic centers have also been known to

6

develop close to the wall in RCMs and shock tubes which coincidentally, are key experimental
facilities to obtain data for kinetic model development and validation, as well as provide insight
into the physics of knock. In recent times, the authors in [36] indicated that under certain conditions
encountered in kinetic experimental facilities, the location of ignition is contingent on the
prevailing thermodynamic conditions and the type of fuel. Based on these findings, it is evident
that, an approach is needed to perform homogeneous ignition studies that are devoid of the above-
mentioned fluid mechanics interferences (boundary layer effects).
The confined spherically expanding flame (CSEF) approach has gained relevance in the
measurement of laminar flame speed (𝑆 u
o
) [37] at engine-relevant conditions, with significant
progress being made in recent times [38]. Review articles of the CSEF method of measuring 𝑆 u
o

can be consulted in [6][39] where the challenges, potential sources of uncertainties and
recommendations for future studies are prescribed. Distinction is hereby made with another
spherically expanding flame configuration where the flame radius trajectory is obtained using high
speed photography during initial propagation usually under isobaric conditions. At the later stages
of flame propagation under isochoric conditions, this method utilizes only the monitored pressure
rise in the vessel without any optical access [38][40]. Also resulting during isochoric conditions is
a flame-induced isentropic compression of the unburnt gas mixture, which raises its pressure and
temperature as shown by the numerical results of Figure 1.2. All results and discussions presented
herein will focus on the latter.

7

Figure 1.2 Compression of the unburned gas induced by CSEF propagation. Color map indicates
the temperatures of both burned and unburned gas

Based on this isochoric process, Hu and Keck [41] advanced this technique by studying
the end-gas autoignition limits of reacting mixtures. In this simple configuration as shown in
Figure 1.3 below, autoignition of the unburnt gas occurs, provided the timescales of flame
propagation and chemical reaction are comparable.

Figure 1.3 Schematic diagram of constant volume spherical combustion bomb used to determine
explosion limits

Unburned
Gas
Burned
Gas

8

Their experimental investigations of n-butane to n-octane reactive mixtures demonstrated
that the CSEF acted as a piston to induce ignition of the unburned mixture being accompanied by
oscillations following an abrupt pressure rise as depicted in Figure 1.4. In the absence of
autoignition, however, smooth pressure traces are obtained.

Figure 1.4 Pressure-time history in the presence of autoignition

Using the CSEF technique in a completely spherical chamber for autoignition investigation
is generally, a complementary low-cost technique which is simpler to construct, robust, easier to
operate, offers high experimental repeatable and most importantly, the ignition process can be
evaluated in the presence of high rates of change of pressure typified by ICE. In addition, the
absence of any movable parts reduces the maintenance and operating cost. The available
experimental test times for compression range from ~25-50 ms and compression ratios ~5-8,
sufficient to promote chemical induction and radical initiation processes at low-temperatures.
Achieving this requires tailoring 𝑆 u
o
through mixture conditions such that the timescales of flame
propagation become comparable to unburnt gas chemical reaction. Other advantages stem from
the fact that, the process is to a large extent one-dimensional (1-D) [27]. Thus, the CSEF offers a
unique methodology to probe the fuel-specific reactions that promote low-temperature ignition
No autoignition
Autoignition

9

including, NTC behavior and the experimental conditions leading to autoignition at engine-
relevant conditions when combined with numerical tools, while concurrently, providing
complementary ignition data. In contrast, should one consider the benefits and drawbacks of other
spatially homogeneous legacy configurations, this poses a great challenge to perform high fidelity
direct numerical simulations (DNS) considering first and second order effects such as boundary
layer growth, recirculation and bifurcation, among other known phenomena etc. However, this
approach to study ignition, has unfortunately, not been afforded the necessary attention to advance
this methodology due to the unavailability of promising DNS tools.
Gasoline components/additive effects on autoignition
The study of autoignition chemistry of neat fuels and multicomponent fuels has been
probed to a large extent in experimental facilities to understand its oxidation at both low and high
temperatures. In addition to the detailed understanding of low-temperature kinetics at engine-
relevant conditions to understand the physico-chemical processes leading to knock, practical fuel
formulation with enhanced knock resistance is equally important for improved engine
performance. Real fuels, for example, gasoline, are a blend of numerous hydrocarbon compounds
derived from the distillation of crude oil and composed of a range of functional groups such as
paraffins, olefins, naphthenes and aromatics to meet various specifications of volatility, emissions,
octane rating, etc [43] in order to attain certain physical and chemical properties due to operating
conditions. Fuels are typically enhanced with components and/or additives to realize these gains
of increased knock resistance. Few experimental investigations into the effects of fuel additives on
autoignition have been performed, and even more so at constant pressure conditions. However, in
a real engine, steep gradients of pressure exist which cannot be realized in typical legacy
experiments. Furthermore, very little progress has been made to understand the mechanistic and

10

synergistic role of fuel additives and components in altering the kinetics of gasoline fuels to
suppress knock except a recent study by Zhang et al. [44]. This has been partly due to the
unavailability of detailed chemical kinetic models of these additives. Kinetic models of additives
will give an indication and qualitative understanding of their oxidation, inhibition effects on the
main fuel breakdown, and the generation of intermediate species that can possibly affect the
autoignition behavior.
1.3 Objectives and overview
In view of the above considerations, the overall goal of this research is three-fold. First,
autoignition is revisited using an alternative approach by advancing the initial proposition by Hu
and Keck [41][42]. To achieve this, flames of CH3OCH3 (DME)/O2/N2/He reactive mixtures will
be used to fully-characterize end-gas autoignition experimentally using a confined spherically
expanding flame leveraging 1-D DNS modeling. The salient observables will be highlighted at
low-temperatures where fuel-specific reactions are prominent and provide new information about
the NTC region under a pressure gradient.
Secondly, the experimental observables will be assessed through heat release computations
with a goal of providing benchmarks for gasoline additive comparison. Zero-dimensional (0-D)
numerical simulations will be performed using end-gas autoignition numerical results of DME to
gain insights into the low-temperature chemistry and its effects on heat release. Additionally, the
importance of the burning rate and unsteady rate of pressure rise on autoignition will be
investigated to understand the time-history on kinetics.
Finally, the autoignition limits of gasoline reactive mixtures and its additives will be
investigated at engine-relevant conditions by examining the effects of gasoline
additive/component structures in promoting or suppressing autoignition under a rapidly varying

11

pressure environment, thus, providing fundamental background to screen fuel additives. This will
be achieved alongside developing chemical kinetic models for these additives to provide
qualitative understanding of the additive mechanistic pathways through sensitivity and reaction
path analysis. In the process, guidance will be provided towards the development of new additives.
Chapter 2 establishes the basics of laminar premixed flames and hydrocarbon oxidation to
provide context in terms of premixed flame theory. Subsequently, topics relating to the generation
of active radicals, regimes of oxidation chemistry, two-stage ignition, including NTC, will briefly
be discussed.
The experimental methodology used to first establish propagation of laminar flames, and
realize end-gas autoignition is detailed in Chapter 3. A brief description is given of the
experimental procedure in addition to post-processing of the experimental results.
Key aspects of experimental end-gas autoignition results in this dissertation are buttressed
with numerical simulations. Numerical modeling proceedures of the experimental data are
presentated in Chapter 4, accompanying codes used to gain fundamental understanding into the
process of autoignition.
Chapters 5 through 8 detail the pertinent results through the course of the project. A
comprehensive discourse of the experimental and numerical results is addressed. Detailed analysis
is provided for the DME experimental and computational results, while gasoline end-gas
autoignition results are discussed for screening purposes. Reaction path analysis provides details
as to the effects of additives on autoignition. In addition, RMG modeling to develop gasoline
additives is elucidated towards autoignition suppresion.
Chapter 9 presents future directions for the project and summarizes the impacts to the
combustion community of the completed studies.

12

Chapter 2: Background
2.1 Laminar Flame Theory
A major objective of this work is employing the use of CSEF to achieve end-gas
autoignition of reactive mixtures under various thermodynamic conditions. Therefore, an accurate
description of fundamental concepts of flame theory is imperative.
Premixed Flames
A flame is a thin region of non-equilibrium space where reactants typically a fuel and an
oxidizer are depleted to products resulting in the conversion of chemical energy stored in the bonds
of fuel molecules to thermal energy. Large gradients of temperature and species concentrations
also occur in this thin region. Depending on the mixing process of the reactants prior to depletion,
two main flames are possible in combustion which are the premixed flame and non-premixed
flame. The focus of this discussion, premixed flames, occurs when the fuel and oxidizer are
homogeneously mixed before consumption. The comparative proportion of fuel-air mixing is
dictated by their equivalence ratio, 𝜙 defined as:
𝜙 =
(
𝐹 𝐴 ⁄ )
𝑎𝑐𝑡𝑢𝑎𝑙 (
𝐹 𝐴 ⁄ )
𝑠𝑡𝑜𝑖𝑐 ℎ𝑖𝑜𝑚𝑒𝑡𝑟𝑖𝑐 (2.1)
where F and A are the molar/mass quantities of the fuel and air, respectively. Based on this
definition of 𝜙 , three possibilities exist: 𝜙 < 1, 𝜙 = 1 and 𝜙 > 1 which represent fuel-lean,
stoichiometric and fuel-rich mixture, respectively. The equivalence ratio plays a major role in
terms of the global response of the combustion process as well as kinetics and pollutant formation.
At the molecular level, the structure of the premixed flame, diagrammatically displayed in Figure
2.1, can be divided into distinct regions. The preheat zone, lD, is where reactants are preheated due
to a balance between convection and diffusion. Reactants are preheated up by virtue of heat

13

diffusion to a temperature at which reaction rates become significant. The reaction zone, lR, where
vigorous chemical reaction occurs. The downstream equilibrium zone contains the high
temperatures and concentrations of combustion products.

Figure 2.1 Structure of a premixed flame

Laminar Flame Speed
A consequential feature of the premixed flame structure is its ability to propagate into an
unburnt mixture with a characteristic propagation speed. This important global phenomenon, the
flame burning speed, is a direct indicator of its exothermicity, reactivity and diffusivity of the
combustible mixture [2]. The laminar flame speed, 𝑆 u
o
, is defined as the propagation speed of a
steady, laminar, 1-D, planar, stretch-free and adiabatic premixed flame as shown schematically in
Figure 2.2.

14

Figure 2.2 Structure of the ideal flame in the laminar flame speed definition

Fundamentally, 𝑆 u
o
reflects the mass burning rate per unit flame front area, 𝑓 o
, however, it
is often easier to measure velocity and thus the former is more commonly reported. 𝑆 u
o
is dependent
on some characteristic parameters highlighted in the Equation 2.2 as:
𝑆 𝑢 𝑜 ~𝑃 (
𝑛 2
−1)
√(
𝜆 𝐶 𝑝 ) 𝑒 (−
𝐸 𝑎 2𝑅 𝑇 𝑏 )
(2.2)
From Equation 2.2, P is the mixture pressure, n is the overall reaction order, λ is the mixture
thermal conductivity, Cp is the mixture heat capacity at constant pressure, Ea is the overall
activation energy, R is the universal gas constant and Tb is the adiabatic flame temperature. The
term under the square root represents the mixture diffusivity while the exponential term is the
overall reactivity. Thus, 𝑆 u
o
is dependent on transport properties and kinetics.
The thickness of a laminar flame can be approximated by the balance of diffusive and
convective fluxes in the preheat zone, lD, which can be expressed as:
𝑙 𝐷 =
𝜆 /𝜌𝐶
p
𝑆 u
o

(2.3)

15

The flame thickness is known to decrease at high pressure and becomes susceptible to flow
instabilities.

2.2 Hydrocarbon Combustion
The hydrocarbon combustion process predominantly involves the detailed interrelations
between chemical reactions resulting in significant amounts of heat release coupled with large
gradients of species and temperature all occurring in a combustion device. This high activation
energy process is responsible for converting the energy between chemical bonds to thermal energy
through a series of elementary reactions of carbon and hydrogen containing species. The reactions
can also occur with or without the presence of oxygen in an oxidative or pyrolytic manner,
respectively, forming stable and radical species in the process.
While the application and combustion device does usually dictate the fuel type, most
hydrocarbon fuels undergo a similar multistep reaction process with a wide range of intermediates
and radicals formed based on the molecular structure. As most of the unstable intermediates and
radicals are short-lived, a complete understanding of their formation and destruction helps gain
control of the entire combustion since the explosion and detonation of reactive hydrocarbon
mixtures can be detrimental to devices and most importantly, lives.
Despite the large number of intermediates formed in the combustion process, only a few
key reactions in the presence of active radicals (O, OH, H, HO2) at particular temperatures, tend
to tremendously affect the global system reactivity. Thus, first identifying these important
intermediates and focusing on their associated controlling reactions provide a possibility to
understanding the combustion process. It is noted though that hydrocarbon combustion follows a
hierarchical nature where H2/CO/C1-C4 chemistry is effectively rate controlling [45].

16

Also important in hydrocarbon combustion is the system temperature. Since the activation
energy of the reaction rate is exponentially dependent on temperature, various kinetic pathways
are activated at different temperatures and compete with existing pathways, inadvertently,
affecting the global system reactivity. Relatively, the system pressure, though important, affects
the global reactivity to a lesser extent compared to the temperature. Experimental investigations
into these phenomena have been carried out in low dimensional configurations to understand these
sequences of stable chemical intermediates and combustion processes.

Fuel Oxidation
The rate of fuel oxidation is primarily determined by the adequate supply of free radicals,
temporally generated from the reactants, that partake in chemical reactions for a reacting system,
such as autoignition. Such kinetically controlled phenomenon is contingent on the fuel molecular
structure.
For a controlled or exponential increase in the global reaction rate leading to an ignition
event, there must be a chain-branching process in the elementary chemical reactions of the active
radicals. Due to the exothermic nature of oxidation, this rate invariably increases with temperature.
As such, the increase in temperature feeds the chemical reactions and generation of chain
branching process in the thermo-kinetic feedback loop [46].
At low temperatures ~ 600 K where combustion is relatively slow, oxidation precedes
pyrolysis where fuel species initially undergo hydrogen abstraction reactions by OH producing a
fuel radical. Subsequently, molecular oxygen adds to this fuel radical in an oxidative process,
before an internal isomerization process and eventually forms ‘keto-hydroperoxide’ which later
decomposes to give smaller fuel fragments as shown in Figure 2.3.

17

Figure 2.3 Temperature dependent process of fuel oxidation [Courtesy: Prof. H.J. Curran]

At higher temperatures (T > 1200 K) the converse is true. The fuel decomposes into
smaller radicals by pyrolysis followed by oxidation of these fragments into CO2 and H2O. The
pyrolytic fragments distribution has a controlling influence on the radical buildup and ensuing
heat release. Thus, at high temperatures alkyl radicals are favored over the peroxy radicals or
pyrolysis is favored over oxidation as seen in Figure 2.4.

Low Temperature Chemistry and NTC
For propane and larger alkanes, low-temperature oxidation follows the general scheme
shown in Figure 2.4 at temperatures between 500 and 1000 K depending on the pressure.

18

Figure 2.4 Key pathways in low-temperature hydrocarbon ignition [Courtesy: Prof. H.J. Curran]

The initial breakdown of nC7H16, a prototype n-alkane, starts with hydrogen atom
abstraction preferentially with OH radical to form an alkyl (nC7H 15) radical. At high temperatures
(T > 1000 K) this alkyl can readily fragment into smaller radicals (i.e. pentene and ethyl) or first
undergo isomerization before further decomposition into smaller radicals such as smaller olefins
and other radicals. At lower temperatures, however, nC7H15 reacts with O2 across the double bond
to form the alkyl peroxy (R7OO) radical; subsequently undergoing an internal hydrogen atom
isomerization thereby moving the hydrogen atom from one radical site to another through the
transition state to produce the hydroperoxyalkyl (Q7OOH) radical [47].
At the radical site, Q7OOH can once again add to O2 to produce peroxyhydroperoxyalkyl
(O2Q7OOH), which subsequently, can form a stable intermediate, ketohydroperoxide (KHP) +
OH. The O-O bond of KHP is broken to obtain two hydroxyl (OH) radicals and a carbonyl alcoxy
radical with the process eventually concluding in heat release. Thus, from a single radical at low-

19

temperature, the chain branching process eventually produces three reactive radicals, once the
KHP’s O-O bond is broken.
Bimolecular reactions preceding chain branching at low temperatures, approximately have
associated energy barrier to overcome, i.e. Ea ≈ 0 kJ. The gradual rise in exothermicity triggers
intermediate temperature unimolecular reactions predominantly associated with the Q 7OOH
radical due to their related energy barrier being overcome. Despite some intermediate temperature
unimolecular reactions producing OH radical, which are progressively favored at higher
temperatures, they are inherently chain propagating steps, thus competing with chain-branching
processes.
In effect, the NTC behavior stems from eventual fate of the Q 7OOH radical since for all
temperatures, it is produced, but depending on the system temperature and associated activation
energy barrier, the processes leading to the intermediate and high temperatures product species
cannot proceed, leaving only the second molecular oxygen addition route (low-temperature
processes) and ensuing chain branching mechanism to proceed. At higher temperatures (750 K <
T < 900 K), however, chain propagating channels produce radicals and stable species which reduce
the global reactivity of the system. Increased exothermicity and temperatures gradually shuts off
the low-temperature branching pathways and processes lead only to intermediate temperature
reactions with a resulting buildup of hydroperoxyl, HO2 radicals and afterwards, forms hydrogen
peroxide, H2O2. H2O2 then decomposes at temperatures above a 1000 K to release two OH radicals
to commence a second chain branching sequence, albeit at higher temperatures. The latter chain
branching sequence is responsible for the 2
nd
stage ignition. The entire pattern is exhibited with
varying degrees by hydrocarbons that exhibit NTC behavior.

20

2.3 Ignition Delay Time (IDT)
Ignition delay time 𝜏 𝑖𝑔𝑛 in chemically reacting systems is defined as the induction period
between combustion commencement to the ignition of the reactive mixture, usually involving
some form of exothermicity or flame structure typified by steep gradients of pressure and
temperature or maximum light output from some excited species. Experimental facilities used in
the measurement of ignition delay include RCMs, motored engines, constant volume chambers
and shock tubes [46]. For ignition to occur, the rate of heat loss ought to be greater than the heat
release rate. The latter being generated by the exponential increase in the radical pool at favoring
thermodynamic conditions. Due to the temperature dependence of the underlying elementary
reactions, IDT predominantly scales inversely on temperature as seen in the Figure 2.5 and
expressed by the functional relationship and linear variation, respectively, below

𝜏 ign
= 𝐴𝑒
(−
𝐸 𝑎 𝑅𝑇
)

(2.4)

ln( 𝜏 ign
) ~
1
𝑇
(2.5)

Figure 2.5 Calculated and measured ignition delay times in hydrocarbon-air mixtures

21

Chapter 3: Experimental Approach
3.1 Experimental Setup
End-gas autoignition investigations are performed in an existing spherical chamber
configuration depicted in Figure 3.1 which has been used in previous studies [38][48] and a
schematic is depicted in Figure 3.2.

Figure 3.1 Schematic of the spherical chamber configuration

22

Figure 3.2 Schematic of the spherical chamber configuration

The spherical chamber is made up of 316 stainless steel and completely capable of
withstanding post-combustion pressures up to 200 atm. It has an internal diameter of 203.2 mm
and a wall thickness of 30 mm. A specially designed 16.4 kW oven capable of ensuring a uniform
initial temperature (Tu,o) distribution up to 500 K houses the spherical chamber. Prior to the filling
process accomplished by the partial pressure method, the spherical chamber is evacuated to ~ 100
millitorr before supplying all the gaseous components using various high accuracy Omega PX-
409-A5V pressure transducers. All the pressure transducers are used in the relevant pressure ranges
to reduce the uncertainty arising from mixture preparation [38]. If liquid fuels are being used, a
Kulite XTEL-190 pressure transducer, calibrated at the experimental initial temperature, is used
to measure the completely vaporized fuel partial pressure. To remove any residual gas after each
experiment, the spherical chamber and supply lines are completely purged multiple times with
nitrogen before evacuation.

23

Introducing any optical access for diagnostics may interfere with the flame’s sphericity,
thus the dynamic pressure remains as the only experimental observable. A Kistler 601B1
dynamic piezoelectric pressure transducer is used to measure experimental pressure-time (P-t)
history (trace) and it is connected to a Kistler 5010B charge amplifier to record the dynamic
pressure-rise signal. This signal is then recorded by a National Instruments 6259 DAQ analog
input at a sampling rate of 100 kHz. The P-t trace is afterwards used for analysis, described in
the next section. Given the importance of this signal, the dynamic transducer is calibrated in the
pressure range of relevance to the experiments at the required initial temperature to reduce the
uncertainty of the pressure measurement (±0.7 psi). The initial temperature is read with an
Omega controller (CN9110A) which has an uncertainty of ±3 K. All gaseous components have
purities of more than 99%. Experimental uncertainties with respect to pressure and the rate of
pressure are quantified based on the method in [38].

3.2 Confined spherical flame methodology
To properly perform an experiment, the following procedure is followed to ensure
measurement accuracy and repeatability and most importantly, safety of the operator.
1. Seal the chamber tightly.
2. Pressurize the chamber to 150 psi and check for any possible leaks
3. Evacuate the chamber using the vacuum pump to 100 mTorr and check for leaks to the
chamber. This is done to ensure leakage of air into the chamber is at its minimum.
4. Enclose the chamber in the oven and heat it up to the desired temperature.
5. If a liquid fuel experiment is being performed, heat up its line to the desired temperature.

24

6. Supply the first gaseous component using the method of partial pressure while reading its
pressure with the static pressure transducer.
7. Close the chamber inlet shut with the high pressure and temperature valve.
8. Vacuum all supply lines to 100 mTorr.
9. Supply the next component while reading its corresponding pressure using the static
pressure transducer.
10. Repeat steps 7 to 9 again until the chamber has been filled with all gases.
11. Finally close the chamber and exhaust all gases in the supply lines.
12. Wait for 7-10 minutes for the gases to reach a quiescent state.
13. Ignite the mixture using a spark from the ignition system and acquire the data using the
dynamic pressure transducer in conjunction with the data acquisition unit.
14. Exhaust the burnt gases and purge the chamber with nitrogen.
15. Vacuum the chamber once again for the next experiment.
For a case of flame propagation in the absence of autoignition, a mixture and thermodynamic
condition was taken from [38] as shown in Table 3.1.
ϕ P o (atm) T o (K) X CH4 X O2 X N2 X HE
0.9 3 298 0.0664 0.1659 0.2303 0.5374
Table 3.1 Summary of thermodynamic conditions investigated

The experimental pressure-time history (Figure 3.3a) depicts a nearly constant pressure
initially up to about ~40 ms, followed by a steep rise due to isentropic compression by virtue of
flame propagation until the flame encounters the chamber wall where the pressure is maximum
and then decays slowly due to heat loss effects. Its corresponding pressure gradient (dP/dt) plotted
against pressure during the isentropic region in Figure 3.3b shows a positive linear range. This
contrasts the non-monotonicity exhibited when autoignition occurs as discussed in Chapter 5.

25

Figure 3.3 Experimental pressure-time (P-t) history and (b) dP/dt as a function of P without
autoignition

3.3 Fuel Handling and Vaporization System
As explained in Figure 3.2, liquid fuel is first vaporized in the heated delivery at the initial
temperature of the experiment and its measured partial pressure supplied to the chamber by virtue
of a vacuum created. In an instance where cold spots exist along the delivery line, the appropriate
partial pressure is not realized as the fuel condensation may occur at these locations.
Cold spots along the supply lines, may cause erroneous pressure measurement in the case
of real fuels which are multicomponent in nature e.g. gasoline fuels, jet fuels, diesel. Preferential
vaporization of the light components can potentially occur while the heavier components might
remain condensed. Appropriate heating and adequate insulation along the fuel delivery lines are
of utmost importance to obviate this experimental challenge.
To this end, different volumes of gasoline fuel are injected into the delivery line at 458 K
via a Hamilton 1750 syringe and its partial pressure measured by the Kulite XTEL-190 pressure
transducer. This test is performed to investigate the significance of cold spots, if any, in the heated

26

section of the delivery lines, on the measured partial pressure and effectively, equivalence ratio.
Based on the volume of fuel injected, its equivalent pressure is computed to serve as a reference
against the measure pressure by the transducer. The results in Figure 3.4 for five injected volumes
100μl – 500μl indicate the measured partial pressure corresponds to computed pressure within
measurement uncertainty. Thus, fuel is not lost by virtue of condensation along any cold regions.

Figure 3.4 Volume of injected fuel as a function its pressure

Gasoline is notoriously susceptible to mishandling whereby, low-molecular weight
components tend to vaporize quickly upon exposure to ambient conditions during handling and
storage, resulting in the potential modification of its composition due to preferential vaporization.
Opening and closing a container of gasoline risks the issue of the loss of light hydrocarbons. As
such, gasoline fuel samples must be refrigerated before opening to minimize the light-end loss.

0.000
0.050
0.100
0.150
0.200
0.250
0.300
0.350
0.400
0.450
0.500
0 100 200 300 400 500 600
Pressure (PSI)
Fuel Volume ( μL)
Injected Fuel
Calculated

27

3.4 Hot-Spot Ignition
During spherical expanding flame propagation, the unburned mixture undergoes isentropic
compression resulting in temporal rise in its pressure and temperature. For chemical reaction
timescales significantly less than that of flame propagation, autoignition of the unburned gases
occurs. Abnormal ignition occasionally transpires due to the presence of hot spots on surfaces
surrounded by unburned gas. In an instance of hot spot ignition prior to volumetric autoignition of
the unburned mixture, two flame fronts may interact producing intense high-frequency pressure
oscillations accompanied by a metallic ping sound [49].
A buildup of carbon deposits on the chamber surface increases the risks of hot spots
forming, potentially promoting preignition. Carbon deposition accumulation on a recessed interior
surface after numerous combustion experimental trials can glow at very high temperatures
fostering conducive preignition conditions. Other likely candidates assisting carbon deposition and
its effects are particles of teflon tape sealing and/ or rust buildup on extended electrodes, pieces of
room-temperature-vulcanizing (RTV) silicon sealant coating the dynamic pressure transducer and
particles from o-ring seal failure at high temperature.
Other than the unusually load metallic ping sound, pre-ignition can, however, be noticed
during post-processing of the experimental pressure trace due its erratic nature and non-repeatable
results. Thus, it is imperative to always repeat experiments.
To prevent preignition, the combustion chamber has to be purged with nitrogen several
times after each experimental run. Additionally, the chamber needs to be disassembled periodically
and all exposed surfaces thoroughly cleaned with ethanol. Finally, teflon tape on the extended
electrodes threads must not be exposed to the chamber interior.

28

Chapter 4: Numerical Methodology
4.1 Freely propagating flames
𝑆 𝑢 𝑜 was computed using the PREMIX code which models 1-D freely propagating steady,
laminar, planar, isobaric, adiabatic premixed flames. PREMIX, developed by Kee et al. [50], is
integrated with CHEMKIN [51] and the Sandia transport subroutine libraries [52]. Thermal
radiation at the optically thin limit has been considered for the species CO, CO2, CH4 and H2O
[53]. Additionally, binary diffusion coefficients of H and H2 with particular species are based on
a recently updated set of Lennard-Jones parameters [54]. All simulations herein are performed
using the full multicomponent transport formulation with thermal diffusion from Soret effect. The
PREMIX code predicts the temperature and species distribution by solving the following
governing continuity, species and energy conservation equations:

𝑑 𝑚 ̇ 𝑑𝑡
= 0

(4.1)

𝑚 ̇ 𝑑 𝑌 𝑘 𝑑𝑡 +
𝑑 ( 𝜌𝐴 𝑉 𝑘 𝑌 𝑘 )
𝑑𝑥 − 𝐴 𝜔 ̇ 𝑘 𝑊 𝑘 = 0

(4.2)

𝑚 ̇ 𝑑𝑇 𝑑𝑥 +
𝐴 𝐶 𝑝 ∑ 𝜌 𝑉 𝑘 𝑌 𝑘 𝐶 𝑝 ,𝑘 𝑑𝑇 𝑑𝑥 −
1
𝐶 𝑝 𝑑 𝑑𝑥 (𝜆𝐴
𝑑𝑇 𝑑𝑥 ) +
𝐴 𝐶 𝑝 ∑ 𝜔 ̇ 𝑘 ℎ
𝑘 𝑊 𝑘 𝐾 𝑘 =1
𝐾 𝑘 =1
= 0

(4.3)
In these equations, ṁ denotes the mass flux eigenvalue, Yk, the mass fraction of the k
th
species,
Vk, the diffusion velocity of the k
th
species, Wk, the molecular weight of the k
th
species, Cp, the
constant pressure specific heat of the mixture, λ, the thermal conductivity, ὡk, the molar rate of
production by chemical reaction of k
th
species and A, the cross-sectional area. A Newton algorithm
is used to solve the above equations after discretization iteratively with the help of a time-stepping
algorithm.

29

4.2 End-gas autoignition and sensitivity analysis
The time evolution of the unburned gas during isentropic compression has been modeled
using the 0-D adiabatic batch reactor model code (sensBVP) [55]. It has been integrated with
Cantera subroutines [56] and will employ the use of the experimental P-t history as input to
monitor the time evolution of species and temperature in the homogeneous unburnt mixture during
isentropic compression. Supplying sensBVP with an experimental pressure trace constrains the
chemical evolution of the unburned gas as it undergoes flame-induced compression. Additionally,
sensBVP computes a characteristic ignition delay time (𝜏 𝐶𝐼𝐷𝑇 ) by solving the energy and species
equations as specified below (all variables defined in the next section):

𝜕𝑇
𝜕𝑡
=
1
𝜌𝐶
𝑝 𝑑𝑃 𝑑𝑡 − ∑ 𝜔 𝑖 ℎ
𝑖 𝑖 1

(4.4)
𝜕 𝑌 𝑖 𝜕𝑡
=
𝜔 𝑖 𝑊 𝑖 𝜌
(4.5)

sensBVP calculates sensitivity of the 𝜏 𝐶𝐼𝐷𝑇 to reaction rates coefficients by formulating the initial
value problem (IVP) as a boundary value problem (BVP), thus obviating the traditional brute force
technique. sensBVP utilizes the SUNDIALS [57] suite of solvers including CVODE, IDA and
KINSOL numerical integrators to solve the initial value problem for ordinary differential equation
systems, differential-algebraic equation and the nonlinear algebraic system for the boundary value
problem and sensitivity analysis, respectively.
While the use of this code is fairly inexpensive since tractable modeling is afforded for large
reactions mechanisms, it does operate under the assumptions:
1. Spatial changes in pressure neglected (Low Mach number)
2. Flame evolution is ignored
3. dP/dt must be known and specified as input

30

4. Homogeneous unburned gas (cold chamber wall boundary layer effects being ignored).

4.3 Spherically Expanding Flames (Direct Numerical Simulations)
Flame propagation and end-gas autoignition was directly simulated using the Lagrangian
Transient One-Dimensional Reacting Flow Code (L-TORC) [58] developed to solve the
propagation and end-gas autoignition of the 1-D laminar flame under confinement under the low
Mach number regime in spherical coordinates. The Lagrangian method, applied in other
combustion related studies (e.g., [[59][60]]), has been utilized in the formulation of the governing
equations shown below to eliminate the non-linear advective term in the transport equations for
energy and species concentrations in comparison to the Eulerian formulation [38][61]. Other
studies utilizing this include numerical works of laminar flame speeds by Spalding [62] and Dixon
Lewis [63], while Stauch [64] and Cho [65] studied autoignition and burning of droplets,
respectively. An exposition on the derivation and numerical application of the Lagrangian
formulation can be found in Richmyer [66] and a summary shown in the supplementary material
of [67].

𝜕𝑟
𝜕𝜓
−
1
𝜌 𝑟 2
= 0

(4.6)
𝜕𝑃
𝜕𝜓
= 0

(4.7)

𝜌 𝐶 𝑝 𝜕𝑇
𝜕𝑡
+ 𝑚 ̇ 𝑜 𝜕𝑇
𝜕𝜓
− 𝜌 𝜕 𝜕𝜓
(𝑟 4
𝜌𝜆
𝜕𝑇
𝜕𝜓
) −
𝜕𝑃
𝜕𝑡
+ ∑ 𝜔 ̇ 𝑖 𝑁 𝑖 =1
∆ℎ
𝑓 ,𝑖 °
+ 𝜌 2
𝑟 2
𝜕𝑇
𝜕𝜓
∑ 𝐶 𝑝 ,𝑖 𝑁 𝑖 =1
𝑌 𝑖 𝑉 𝑖 = 0

(4.8)

𝜌 𝜕 𝑌 𝑖 𝜕𝑡
+ 𝑚 ̇ 𝑜 𝜕 𝑌 𝑖 𝜕𝜓
+ 𝜌 𝜕 𝜕𝜓
( 𝑟 2
𝜌 𝑌 𝑖 𝑉 𝑖 )− ( 𝑊 𝑖 𝜔 ̇ 𝑖 ) = 0
(4.9)

(
𝜕𝑇
𝜕𝜓
)
𝑟 =0
= 0, (
𝜕 𝑌 𝑖 𝜕𝜓
)
𝑟 =0
= 0
𝐴𝑑𝑖𝑎𝑏𝑎𝑡𝑖𝑐 : (
𝜕𝑇
𝜕𝜓
)
𝑟 =𝑅 = 0,
Non-adiabatic: ( 𝑇 )
𝑟 =𝑅 = Twall, Vi = 0

31

with:
r - spatial location
t - time
hi – specific enthalpy of species
R - domain length
𝜌 - mixture density
P - thermodynamic pressure
𝐶 𝑝 - bulk specific heat capacity at constant pressure
T - temperature
λ - thermal conductivity
Yi - i
th
species’ mass fraction
∆ℎ
𝑓 ,𝑖 °
- standard enthalpy of formation
Vi - diffusion velocity (computed using a mixture-average formulation)
Wi - molecular mass
𝜔 ̇ 𝑖 - molar production rate
N - total number of species
𝜓 - Lagrangian variable

The method of lines approach is used to isolate temporal and spatial derivatives, the latter
of which is discretized using finite differences. The system of equations is differential-algebraic
and is solved using IDA [57]. The fluid is treated as an ideal gas and the necessary thermodynamic,
kinetic and transport calculations are carried out using Cantera. To resolve the flame as it moves,
a moving point density function constructed from hyperbolic tangent functions with high density
at the flame at all times was used. Additional refinement was supplied at the wall (due to the
thermal boundary layer there) and the origin (due to the occurrence of terms proportional to the
square of the radius in the governing equations). Grid refinement is accompanied by third order
monotone upwind interpolation from variables on the old grid to the new one.

4.4 Automatic Mechanism Generator
Reaction Mechanism Generator (RMG) [68][69] software package is a predictive tool used
for generating chemical kinetic models for a wide range of combustion systems of interest by

32

performing extensive computations and estimations for all species, significant reactions and rate
coefficients.
RMG utilizes a rate-based algorithm to automatically discover new species and associated
elementary reactions in order to build models. The respective thermodynamic parameters and rate
constants for generated species and reactions are obtained from a database and estimated by rate
rules. Species and elementary reactions incorporated into the final model (core) are selected once
a user-specified flux is attained.
The simulation is commenced by specifying initial conditions (Po, To, Xo) and termination
criteria after which the rate-based algorithm is used for reaction generation. All initial species are
reacted together in an isothermal isobaric batch reactor using know reaction family templates and
a time-integration of the kinetic equations is performed. Afterwards, the rates of new chemical
species produced from the previous step are evaluated.
When a specified flux is surpassed, the new species with its corresponding reactions is
migrated to the core. Subsequently, the new core species undergoes a process of reacting with each
other to generate new edge species and associated reactions followed by the time integration using
the updated core. The chemical model growth follows this cycle until user-specified end-
conditions are met.
All models presented were generated using RMG-Py version 2.4.1 and 3.0.0.
Thermochemistry was estimated using Benson’s group additivity [70] for all species.

33

4.5 Kinetic Models
For 1-D freely propagating flames and spherically expanding flames and IDT numerical
simulations, the Zhao et al. [71] kinetic model with detailed molecular transport made up of a 55
species and 290 reactions including both low and high temperature chemistry was used.
The gasoline kinetic model by Wu et al. [72] containing 165 species and 1403 reactions including
both low and high temperature chemistry was used in conjunction with RMG. Both models have
been validated for conditions of relevance to this study.

34

Chapter 5: Autoignition of reacting mixtures at engine-relevant
conditions using confined spherically expanding flames
5.1 Introduction
Knock has been one among the major limitations in gasoline engine technology towards
achieving higher efficiencies and has been the motivation of numerous studies. Experimental
investigations have been performed using shock tubes (e.g.,[3]), RCMs (e.g.,[10]), and motored
engines. The strength of STs and RCMs lies in their capacity to attain high thermodynamic
conditions (400-3000 K, 2-100 bar) coupled with laser and optical diagnostics for IDT
measurement, speciation time-history, reaction rate measurements, and knock studies. As
described by Donovan et al. [73], the key desirable factors worth considering in an experimental
study of autoignition phenomena are ample test times, high pressures and temperatures, good flow
field and measurement access, for which the aforementioned duly satisfy.
CSEFs employed for the measurement of 𝑆 𝑢 𝑜 at high pressures and temperatures was
pioneered by Lewis and von Elbe [37] and also offer complementary low-cost approach to yield
insight into knocking phenomena as first suggested by Hu and Keck [41][42]. Their experimental
investigations of C4-C8 alkanes demonstrated that a CSEF acts as a “piston” through the thermal
expansion of the burned gas that isentropically compressed the end-gas to thermodynamic
conditions favoring its homogeneous ignition or in some cases detonation being accompanied by
oscillations following an abrupt pressure rise. Thus, the isentropic compression of the end-gas for
particular initial conditions, could bring about autoignition under variable pressure conditions that
closely mimic SI engine behavior [41][42].
There are only two experimental studies that used CSEFs to investigate autoignition
systematically [74][75]. The limited use of the initial proposition by Hu and Keck [41][42] is

35

mainly due to the lack of high-fidelity tools that could simulate simultaneously the phenomena of
flame propagation and autoignition, which is a stringent requirement for chemical kinetic model
development. In other studies such as [76][77][78][79], the autoignition via flame-induced
compression has also been investigated.
The goal of the present investigation is to revisit the original proposition of Hu and Keck
[41][42] by experimentally characterizing end-gas autoignition and interpreting these
observables using DNS. The canonical nature of the experiment suits the use of a 1-D low Mach
number reacting flow code to accurately model all the relevant physics: flame propagation, heat
loss at the walls, and end-gas autoignition. The method is demonstrated by using reactive DME-
oxygen-inert-mixtures, known to exhibit NTC behavior.

5.2 Experimental approach
All CSEF experiments and pertinent procedures were performed in a heated spherical
chamber with all details described in Chapter 3. End-gas autoignition was investigated using DME
flames with initial thermodynamic condition, mole fraction (Xi) compositions and initial adiabatic
flame temperatures (Tad) as listed in Table 5.1. The mixture dilution of He/N2 = 70/30 on a molar
basis was used with experiments conducted at an initial temperature To = 468 K and various initial
pressures (Po).
Mixture ϕ Tad (K) Po (atm) XCH3OCH3 XO2 XN2 XHE
1 0.9 1950 3,4,5,6 0.032967 0.109890 0.257143 0.60000
2 0.6 1925 3,4,5,6 0.033009 0.165047 0.240583 0.561361
Table 5.1 Parameter space considered in the present investigation

36

5.3 Numerical Approach
End-gas reactivity was modeled using the 0-D adiabatic batch reactor model code sensBVP
which accounts for the temporal isentropic compression of the unburned mixture.
Flame propagation and end-gas autoignition were simulated using L-TORC for initial mixture
composition, Po and To are as specified in Table 5.1. The premixed flame front propagates from
the left to the right wall, compressing the end-gas in the process resulting in autoignition. A
computational spherical domain of radius 10.16 cm is used with a non-adiabatic wall. An ignition
kernel with a chemically equilibrated composition was used to initiate the flame. Details of both
codes are provided and discussed in Sections 4.2 and 4.3.

5.4 Characterization of end-gas autoignition
With the pressure trace being the only experimental observable, attention was devoted to
any phenomena that affect its measurement, such as flame-front instability and buoyancy.
For weakly burning CSEFs, the burned volume of gas is likely to lose its sphericity by becoming
buoyant, a tendency promoted at higher pressures. To preclude this, the reactive mixtures were
formulated by partially replacing N2 with He, decreasing the molar specific heat of the diluent and
raising T ad resulting in increasing 𝑆 u
o
. As such, 𝑆 u
o
s for freely propagating flames using the
PREMIX code [50] were computed at the initial conditions in Table 5.1 to ensure buoyant-free
flames. The computed 𝑆 u
o
results in Table 5.2 were all greater than 35 cm/s guaranteeing a flame
propagation unaffected by buoyancy.

37

Mixture ϕ X CH3OCH3 X O2 X N2 X HE T o (K) P o (atm) 𝑆 𝑢 𝑜 (cm/s)

1

0.9

0.032967

0.109890

0.257143

0.60000

468.15
3 43.84
4 40.13
5 37.51
6 35.33

2

0.6

0.033009

0.165047

0.240583

0.561361

468.15
3 54.04
4 49.99
5 46.73
6 44.11
Table 5.2 Parameter space considered for present investigation with corresponding 1-D laminar
flame speeds simulations

The potential development of flame-front instabilities during propagation exists: thermo-
diffusive and hydrodynamic [80] modes augment the mass burning rate by creating flame surface
area. Helium was used for this reason as well as a diluent by increasing the mixture’s Lewis
number beyond unity [81], thus ensuring thermo-diffusional flame stability; high pressure 𝑆 u
o

experiments (e.g., [38][48][82]) have employed this approach to achieve stable flames. It is worth
noting also that the use of He, due to its higher heat capacity ratio, helps promote end-gas
autoignition by attaining higher unburned gas mixture temperatures.
A typical experimental P-t trace and its gradient (dP/dt) obtained in a previous study in the
absence of autoignition, is shown in Figure 3.3, Section 3.2. In the presence of end-gas
autoignition, (for initial conditions in Table 5.1) the P-t plot (Figure 5.1a) exhibits pressure
oscillations similar to previous studies [41][42][74][75]. The fractional rate of change of pressure
(d(lnP)/dt) is plotted against P/Po in Figure 5.1b and the non-monotonic behavior is due to low-
temperature kinetics similarly to ignition caused by thermal boundary layers (TBL) [36].

38

Figure 5.1 Experimental results for Mixture 1: (a) temporal evolution of P and (b) d(lnP)/dt as a
function of P/Po for Po = 3 atm ( ─), Po = 4 atm ( ─), Po = 5 atm ( ─), Po = 6 atm ( ─)

The experimental results for Mixture 2 for all initial pressures (Po = 3 - 6 atm) are also
presented in Figure 5.2 depicting the pressure oscillations in (a) and corresponding non-
monotonic behavior in the pressure gradient from the two-stage ignition process.

39

Figure 5.2 Experimental results for Mixture 2: (a) temporal evolution of P and (b) d(lnP)/dt as a
function of P/Po for Po = 3 atm ( ─), Po = 4 atm ( ─), Po = 5 atm ( ─), Po = 6 atm ( ─)

The 1
st
stage ignition, associated with a mild heat release, is manifested by the bump in
d(lnP)/dt (occurring between 3 ≤ 𝑃 𝑃 𝑜 ≤ 4 ⁄ ) , while the violent second-stage ignition, associated
with large heat release, corresponds to the rapid rise in d(lnP)/dt. After the point of autoignition,
there is a generation of pressure oscillations due to the short duration of the increase in pressure
compared to timescales of acoustic relaxation [27] which is discussed later.
LTORC was used to model the entire experiment, capturing flame propagation, isentropic
compression, end-gas reactivity upon compression, and near-wall TBL effects. Mixture 1 at Po =
6 atm will be used to analyze the observed pressure response.
Directly after a hot ignition kernel is initiated at the origin (r = 0) at t = 0 ms, a flame
propagates outwardly. During the initial stages, the unburned mixture is consumed almost
isobarically with the temperature remaining at ~ 468 K. Following this, effective compression
commences at ~
𝑅 𝑓 𝑅 𝑤 = 80% [38] as evident by the temporal evolution of temperature distribution

40

in Figure 5.3a at r ~ 7 cm up to 9.56 cm where the unburnt gas temperature (Tu) has risen to ~700 K
with negligible low-temperature reactivity and/ or ignition. A TBL forms near the chamber wall
[36] during flame propagation where heat loss occurs as seen by the insert in Figure 5.3a.

Figure 5.3 Computed radial variation of T for Mixture 1 at Po = 6 atm with corresponding time
sequence (a) a - 36.12 ms, b - 47.84 ms, c - 54.31 ms, d – 58.60 ms, e – 61.80 ms, (b) f – 53.72
ms, g – 57.20 ms, h - 59.94 ms, i – 61.58 ms, j -61.96 ms, k - 62.2040 m

For t > 61.58 ms and r ~ 9.6 cm, reactivity vigorously commences in the unburnt gas and
there is a significant unburned gas temperature, (Tu) jump to ~ 940 K, a manifestation of 1
st
stage
ignition. Further compression results in continuous temperature rise up to ~1040 K, at which point
the second-stage ignition takes place resulting in an exponential increase in temperature (Figure
5.3b).
Regarding the low-mach number approximation implemented in LTORC, the subject of
pressure wave development effects from the 1
st
stage ignition influencing the processes leading to
the 2
nd
stage ignition required some justification. These disturbances, depending on their
magnitude can initiate after the 1
st
stage heat release and can potentially affect the upstream flow
field through pressure, temperature and concentration gradients and result in compressible gas
dynamic features. To estimate and/ or quantify this degree of compressibility which can possibly

41

occur, a characteristic value, C, was calculated by comparing the acoustic time-scale (domain size
divided by the unburned gas speed of sound) divided by the inverse pressure-rise time-scale (dt/dP
* P) at both stages of ignition for both Mixtures 1 and 2 at Po = 6 atm. Results in Table 5.3 suggest
that the acoustic effects are negligible after 1
st
stage ignition since C at the 2
nd
stage is higher than
the former. Thus, the low Mach number approximation is apt, granted the ignition event is
relatively mild.
Mixture Ignition
stage
Acoustic Time-
scale (ms)
Pressure-rise Time-
scale (ms)
C

1
1
st

2.53 × 10
−4
3.19 × 10
−3
7.93 × 10
−2

2
nd

1.79 × 10
−4
8.05 × 10
−4
2.23 × 10
−1

2
1
st

2.56 × 10
−4
1.86 × 10
−3
1.38 × 10
−1

2
nd

1.90 × 10
−5
7.11 × 10
−4
2.66 × 10
−1

Table 5.3 Characteristic values, C, at 1
st
and 2
nd
stage ignition

Secondly, to investigate the influence of pressure waves on IDT, IDT numerical
simulations were performed using two different initial conditions. The first (IDT) involved P and
T right before 2
nd
stage ignition while the second ignition delay time (IDTo) simulations involved
the stagnation pressure, Po, and temperature, To. Implied here is the conversion of all the kinetic
energy to chemical energy (which is invariantly the definition of Mach number i.e. chemical
energy to kinetic energy). In an instance where the gas dynamics effects are substantial, the results
will be manifested through the Mach number and stagnation conditions. The IDT results in Table
5.4 differ by less than a 1% and it thus, was cautiously concluded that, IDT was not going to be
affected to a large extent by mild pressure waves and again for the conditions investigated.

42

Mixture P o (atm) T o (K) IDT (ms) IDT o (ms) Percentage Difference (%)
1 27.08

1507.58

8.167 × 10
−3

8.16 × 10
−3

0.099
2 27.24

1471.87

9.18 × 10
−3
9.14 × 10
−3

0.465

Table 5.4 IDT computation for dynamic and stagnation thermodynamic values

With the comparison of acoustic relaxation timescales and pressure rise timescales, it is
concluded that since the purpose of this work is to capture only the instant of autoignition,
considerations of compressibility are safely negligible.
LTORC was found also to capture satisfactorily the phenomena of flame propagation and
autoignition after comparing the experimental and computed dP/dt results during compression for
Mixture 1 at Po = 6 atm in Figure 5.4.

Figure 5.4 Comparison of experimental ( …) and computed ( ─) dP/dt as a function of P for
Mixture 1 at Po = 6 atm

There is agreement during the initial stages of flame propagation. However, when end-gas
reactivity initiates, significant discrepancies exist and could be caused by uncertainties in the
kinetic model. To that end, the oxidation of DME has been studied in a jet-stirred reactor from low
to high temperature (500 – 1100 K) conditions [83] and similar discrepancies were found using

43

the present kinetic model as well as that of Burke et al. [84]. Analysis showed that the results are
sensitive to the branching ratios involving the CH2OCH2OOH species. On the other hand, it should
be emphasized that ignition chemistry typically has much larger uncertainties than flame speed.
IDT uncertainties in measurements and models can nearly differ by a factor of 2 while flame speed
uncertainties are usually 10-20%, thus, the discrepancy does seem reasonable. Despite the obvious
discrepancies, the ability to qualitatively capture the trend by DNS provides encouraging evidence
that the CSEF approach is viable towards probing autoignition phenomena at low temperatures.

5.5 Characteristic ignition delay time and sensitivity analysis
To further investigate the underlying chemistry of end-gas autoignition, the experimental
P-t traces were fed as input into sensBVP, to determine the Tu, Yi evolution and 𝜏 𝐶𝐼𝐷𝑇 . During low-
temperature reactivity and ensuing 1
st
stage ignition, the fuel breakdown undergoes a chain-
branching sequence over a narrow thermodynamic range, generating OH radicals in the process
while further augmenting the fuel’s consumption along with attendant temperature rise. If low-
temperature reactivity is absent and ignition does not occur (chemically frozen), the end-gas will
only be isentropically compressed before being consumed entirely by the flame.

44

Figure 5.5 Computed P-Tu functions for an unreactive ( ─) and reactive ( ─) end-gas for Mixture 1
at Po = 6 atm

Figure 5.5 depicts two computed results of the end-gas P-Tu of Mixture 1 at Po = 6 atm. In
both cases there is isentropic compression caused by the pressure rise, however in one the end-gas
reactivity has been suppressed and a deviation of the results initiates at P ~ 23 atm. At Tu ~ 700 K,
low-temperature reactivity becomes significant resulting in 1
st
stage ignition where the end-gas
transitions to a different thermodynamic and chemical state. Despite the end-gas traversing the
NTC region featuring a known decreased reactivity with increasing temperature, the simultaneous
compression by the CSEF results in the P-Tu rise. Finally, with considerable build-up of relevant
chemical species promoting high temperature ignition, the 2
nd
stage occurs with an accompanying
rapid temperature rise.
Compared to RCMs, a key difference of the CSEF approach is that ignition is achieved in
a time-varying thermodynamic (P, Tu) condition. Thus, a “traditional” ignition delay time ( τign)
cannot be defined. Instead, a 𝜏 𝐶𝐼𝐷𝑇 is defined as the time elapsed from t = 0 to a point of rapid

45

reaction phase leading to a temperature and pressure rise in Fig 5.6; 𝜏 𝐶𝐼𝐷𝑇 values from sensBVP
for all Mixtures in Table 5.1 are displayed in Table 5.5.

Figure 5.6 Computed temperature profile for Mixture 1 at Po = 6 atm with 𝜏 𝐶𝐼𝐷𝑇 . Insert shows a
section of the zoomed temperature profile starting before 1
st
stage ignition to after 2
nd
stage ignition

Mixture ϕ Po (atm) Computed 𝜏 𝐶𝐼𝐷𝑇 (ms) Experimental 𝜏 𝐶𝐼𝐷𝑇 (ms)

1

0.9
3 95.54 N/A
*

4 97.79 96.99
5 102.03 101.42
6 105.39 104.77

2

0.6
3 77.49 76.75
4 82.15 81.49
5 86.62 86.04
6 90.60 90.02
Table 5.5 Computed and measured characteristic ignition delay times from sensBVP and
experiment

Formaldehyde (CH2O), a strong marker of low-temperature reactivity [10][36][85], largely
starts being produced in the end-gas. The CH2O time-evolution shown in Figure 5.7 also coincides
with the initial temperature rise during 1
st
stage ignition. At low-temperatures it is produced

46

mainly from the β-scission of CH3OCH2 and CH2OCH2OOH which competes with other reactions
leading to low-temperature chain branching (discussed in next chapter).
At relatively higher temperatures and pressures, DME and CH 2O are rapidly consumed as
OH build-up in the unburned mixture becomes prominent through H2O2 (+ M) → OH + OH (+ M)
and results in second-stage hot ignition.

Figure 5.7 Computed temperature profile, T ( ─) and DME ( ─), OH ( ─) and CH2O ( ─) mass
fractions in the unburned mixture as a function of the normalized time for Mixture 1 at Po = 6 atm

To assess the controlling reactions of end-gas autoignition, sensitivity analysis using
sensBVP was performed. CH2OCH2O2H is consumed through two main pathways as depicted in
the reaction path diagram in Figure 5.8 at P = 23 atm and T = 760 K for Mixture 1. The first
reaction of CH2OCH2O2H is CH2OCH2O2H + O 2 → O2CH2OCH2O2H with its product going
through the following reaction steps O2CH2OCH2O 2H → HO2CH2OCHO + OH and subsequently,
HO2CH2OCHO → OCH2OCHO + OH releasing two OH radicals in a sequence that promotes
low-temperature chain branching. The second being a β-scission reaction
CH2OCH2O2H → 2CH2O + OH inhibits autoignition. β-scission of CH3OCH2 can also produce

47

CH2O and CH3 radicals that inhibit autoignition. Thus, the competition between O2 addition and
β-scission reactions of CH2OCH2O2H at low-temperatures is largely responsible for the NTC
behavior [84][86][87]. The ranked logarithmic sensitivity coefficients (LSC) for Mixture 1 at
Po = 3 and 6 atm shown in Figure 5.8 depict the importance of the above reactions in end-gas
autoignition.

Figure 5.8 Ranked LSC of 𝜏 𝐶𝐼𝐷𝑇 to kinetics for Mixture 1 at Po = 3 (blue) and 6 atm (orange)

Similar to 𝜏 𝐶𝐼𝐷𝑇 , a characteristic 1
st
stage ignition delay time (𝜏 𝐶𝐼𝐷 𝑇 1
) is also defined, but
at the 1
st
stage ignition point with sensitivity analysis performed subsequently for Mixture 1 at
Po = 6 atm. The results in Figure 5.9 are compared with those of Figure 5.8 and as expected 𝜏 𝐶𝐼𝐷 𝑇 1

is not sensitive to high-temperature kinetics.
The results presented below indicate that the same set of reactions control both 𝜏 𝐶𝐼𝐷 𝑇 1
and
𝜏 𝐶𝐼𝐷𝑇 except H2O2 (+M) <=> 2OH (+M) and CH3 + HO2 <=> CH3O + OH which largely affect
the latter due to the end-gas marching through the low-temperature regime with its associated
-0.12
-0.07
-0.02
0.03
0.08
0.13
Logarithmic Sensitivity Coefficient

48

chemistry before high-temperature chemistry becomes rate-controlling. Secondly, at low-
temperatures, the slow kinetics permit the peroxy chemistry to proceed easily and result in fuel
species that promote the high-temperature ignition.

Figure 5.9 Ranked logarithmic sensitivity coefficient of 𝜏 𝐶𝐼𝐷𝑇 (blue) and 𝜏 𝐶𝐼𝐷 𝑇 1
(red) to kinetics
for Mixture 1 at Po = 6 atm

5.6 Unsteady effects of pressure on kinetics
An important aspect of the CSEF approach is that autoignition initiates in the presence of
dP/dt similar to piston engines. Furthermore, alkyl radical oxidation is highly dependent on the
prevailing thermodynamic conditions. Thus, it becomes imperative to explore the effects of this
transient pressure rise on kinetics at low temperatures.
To that end, five linear P-t traces shown in Figure 5.10a for Mixture 1 at To = 650 K and
Po = 25 atm with increasing dP/dt to simulate different compression rates were fed into sensBVP
to compute the Tu and Yi evolution. The temperature history in Figure 5.10b shows that with
-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
Logarithmic Sensitivity Coefficient

49

increasing dP/dt autoignition is affected notably. A kinetic analysis in Eqn. 5.1 involving DME
H-abstraction (CH3OCH3 + OH → CH3OCH2 + H 2O) and the first O2 addition to CH3OCH2
(CH3OCH2 + O2 ↔ CH3OCH2O2) reactions reveals that CH3OCH2 and thereby CH3OCH2O2
formations are progressively being quenched by increasing dP/dt assuming constant temperature.

𝑑 𝑋 𝐶𝐻 3𝑂𝐶𝐻 2
𝑑𝑡
= 𝐾 𝑓 ,1
𝑃 𝑅 𝑢 𝑇 𝑋 𝐶𝐻 3𝑂𝐶𝐻 3
𝑋 𝑂𝐻
− 𝐾 𝑓 ,2
𝑃 𝑅 𝑢 𝑇 𝑋 𝐶𝐻 3𝑂𝐶𝐻 2
𝑋 𝑂 2
+ 𝐾 𝑏 ,2
𝑋 𝐶𝐻 3𝑂𝐶𝐻 2𝑂 2
−
𝑋 𝐶𝐻 3𝑂𝐶𝐻 2
𝑃 𝑑𝑃
𝑑𝑡
(5.1)
More specifically, dP/dt contributes a subtractive term proportional to the XCH3OCH2/P.
Thus, there is an acceleration towards high-temperature kinetics and second-stage ignition. It is
also observed from the net rate of progress of CH3OCH2 + O2 ↔ CH3OCH2O2 in Figure 5.10c that
the production of CH3OCH2O2 first increases then peaks with increase in dP/dt before starting to
decrease for dP/dt = 73.33 atm/ms.

50

Figure 5.10 (a) P-t history, (b) Tu-t variation and (c) Net rate of progress of CH 3OCH2O2
production from sensBVP for Mixture 1 at Po = 25 atm, Tu,o = 650 K for dP/dt = 7.33 atm/ms ( ─),
dP/dt = 14.67 atm/ms ( ─), dP/dt = 36.67 atm/ms ( ─), dP/dt = 58.67 atm/ms ( ─), dP/dt = 73.33
atm/ms ( ─).

51

5.7 Conclusion
End-gas autoignition was investigated using the confined spherically expanding flame
method for DME/O 2/N2/He reactive mixtures. The distinctive features of the experimental
pressure-time were highlighted through the pressure gradient, specifically during the compression
region. Low-temperature ignition was identified by a ‘bump’ in the pressure gradient and high-
temperature ignition by a rapid rise in pressure followed by pressure oscillations whose amplitude
increased with higher initial pressure and for fuel lean conditions. For all experimental conditions,
it was seen that 1
st
stage ignition occurred approximately at 3.6 times the initial pressure.
A one-dimensional Lagrangian reacting flow code was developed to directly model the
unsteady spherical flame propagation including isentropic compression, end-gas reactivity and
autoignition at the same experimental conditions. The results showed that the expanding flame
induced 1
st
stage ignition during the compression followed immediately by the second-stage
ignition. Furthermore, it was shown that the boundary layer shrunk with time, thereby minimizing
the possibility of fluid mechanics interference with the adiabatic core.
Species and temperature time evolution were processed from the experimental pressure
time history using a 0-D code which in addition to calculating a characteristic ignition delay time
along with mathematically rigorous sensitivity analysis to kinetics to identify the controlling
reactions of end-gas autoignition. It was found that this characteristic ignition delay is largely
controlled by fuel-specific reactions, specifically, dimethyl-ether low-temperature chemistry.
Additionally, the sensitivity to kinetics increases as the fuel mixture becomes leaner due to the
increasing effect of oxygen in the low temperature chain branching.
The current study of autoignition using confined spherically expanding flames represents
a complementary low-cost approach to obtain experimental data which can be modeled for reliable

52

kinetic model validation under thermodynamic conditions relevant to engines. Coupled with
numerical tools, the physics leading to engine knock can be reasonably explored to understand the
controlling mechanisms.

53

Chapter 6: Assessment of observables from end-gas autoignition
spherically expanding flame experiments
6.1 Introduction
The use of the experimental thermodynamic pressure as a primary diagnostic tool has been
standard practice in RCMs and in certain instances, OH emissions have been employed as well
[10][88][89][90]. From the measured pressure-time history, the main ignition point (ignition delay
time - τign) in RCMs is typically defined as an induction period from a reference time (the end of
compression) until the largest rate of change of pressure (dP/dt) is realized. Similarly, for a two-
stage ignition, the 1
st
stage ignition ( τ1) is usually defined based on the location of the inflection in
the pressure trace or the first maximum in the rate of change of pressure.
Various authors have quantified the low temperature heat release as the integral of the heat
release rate curve during the 1
st
stage ignition peak. All these key features amongst others have
been summarized by Goldsborough et al. [10] in a comprehensive review of experimental and
computational efforts to study autoignition phenomena in RCMs.
Similarly, in this spherically expanding flame approach to investigate autoignition, dP/dt
has been used extensively for comparative analysis as discussed previously, where the key
experimental observables were used for characterization and subsequent comparative analysis.
The objective here is to highlight the relevance of the experimental observables using numerical
simulations in order to provide insight to experimental results.

54

6.2 Experimental parameter space and numerical approach
End-gas autoignition was assessed for flames of DME with initial thermodynamic
conditions and mole fraction compositions as listed in Table 6.1. The mixture dilution of He/N2
= 70/30 on a molar basis.
Mixture ϕ Tad (K) Po (atm) XCH3OCH3 XO2 XN2 XHE
1 0.9 1950 6 0.032967 0.109890 0.257143 0.60000
Table 6.1 Experimental thermodynamic condition and mixture composition used for current
investigation

To model the end-gas reactivity, sensBVP [55] was used to account for isentropic compression
of the unburned mixture. The DME Zhao et al. [71] molecular transport and chemical model was
also used.

6.3 Characterization of end-gas autoignition observables
1
st
stage ignition point
The experimental pressure gradient as a function of pressure during the compression region
for Mixture 1 at initial conditions stated in Table 6.1 is presented in Figure 6.1 to highlight the key
observables used for characterization and comparison.

55

Figure 6.1 Experimental pressure gradient for Mixture 1 in Table 6.1 during the compression
stage of CSEF propagation. Point ‘a’ – 1
st
stage ignition point (Pign_1); Point ‘b’ – dP/dtmax;
Hatched Area ‘c’ – Measure of 1
st
stage exothermicity.

The first experimental observable, point ‘a’, 1
st
stage ignition point (Pign_1), was shown as
the inflection point preceding the 1
st
stage ignition. A plot of the species and temperature profile
in Figure 6.2 below shows the profiles for DME, CH2O, OH, HO2 and HO2CH2OCHO
(hydroperoxymethyl formate) during the two-stage ignition event after submitting the
experimental pressure trace of Mixture 1 in Table 6.1 to sensBVP for thermo-chemical evolution
modeling.

56

Figure 6.2 Computed temperature profile T and species mass fraction in the unburned mixture as
a function of the normalized time.

HO2CH2OCHO evolution shows its highest peak close to the inflection point in
temperature preceding 1
st
stage ignition. It has been shown in a detailed study [86] that its
decomposition to OH + OCH2OCHO, after a brief buildup is the main driver towards autoignition
of DME below 700 K.

Maximum heat release during 1
st
stage ignition
The second experimental observable, Point ‘b’, noted as dP/dtmax, is a benchmark indicator
of the maximum heat release attainable during 1
st
stage ignition. As with point ‘a’, sensBVP
simulation results were used to help interpret its significance. The computed temperature profile
showing the end-gas temperatures during compression and the heat release rate normalized by that
at the 2
nd
stage ignition is shown below in Figure 6.3(a) with a section at only the 1
st
stage displayed
in Figure 6.3(b). Three pressures (x,y,z) were chosen during 1
st
stage ignition with corresponding
temperatures 760 K, 854 K and 880 K. These were points during the rise in temperature before the

57

peak of the normalized heat release rate (point x, 760 K), the peak (point y; 854 K) and after the
peak (point z; 880 K).

Figure 6.3 (a) Computed thermodynamic pressure and normalized heat release rate as a function
of unburned temperature for Mixture 1 (b). Section of computed thermodynamic pressure and
normalized heat release rate as a function of unburned temperature for Mixture 1 at 1
st
stage
ignition.

Investigations carried out into the exothermicity of specific DME low-temperature
oxidation at the peak (T = 854 K) indicated that the following reactions below each contributed
more than 5% to the overall heat release and approximately 83% to the total low-temperature heat
release at the peak:
CH3OCH3 + OH  CH3OCH2 + H2O (i)
CH3OCH2 + O2  CH3OCH2O2 (ii)
CH3OCH2O2  CH2OCH2O2H (iii)
CH2OCH2O2H + O2  O2CH2OCH2O2H (iv)
O2CH2OCH2O2H  HO2CH2OCHO+OH (v)
HO2CH2OCH2O  OCH2OCHO + OH (vi)
CH2OCH2O2H  2 CH2O + OH (vii)

58

The reaction sequence (i) – (vi) leading to chain-branching and subsequent heat release all
serve to generate OH radicals thereby promoting heat release and the 1
st
stage ignition [71][91].
However, at these three points in Figure 6.3(b), it was noted that there was a competition between
the second oxygen addition (iv) leading to chain branching and β-scission reaction (vii) leading to
a chain propagation and the entrance to the NTC regime. The transition from 760 K through 854
K to 880 K during the 1
st
stage ignition sees the shift from one controlling pathway to the other
and this was manifested as the curvature in the temperature and peaking of the heat release rate
profiles. Thus, Point ‘b’, could be interpreted as the maximum heat release attainable during 1
st

stage ignition before the commencement of dominating NTC regime typified by chain propagating
pathways. This is further displayed in the reactions pathways of DME at those particular
temperatures where reaction (vii) being an inhibitive radical propagation pathway, becomes
important as the temperature increases and competes with the chain branching sequence.

59

Figure 6.4 Reaction path analysis at T = 760 K, T = 854 K and T = 880 K for Mixture 1.

Measure of 1
st
stage ignition heat release
The hatched area ‘c’ in Figure 6.1 denoted as the measure of heat release (MHR) is
indicative of the total energy released during 1
st
stage ignition. As done above, for Points ‘a’ and
‘b’ the sensBVP simulation results were examined with respect to the exothermicity of all 290
reactions from the Zhao et al. DME kinetic model [71] during 1
st
stage heat release as depicted in
Figure 6.3(b) which spanned the temperature range from 715 K to 930 K. The exothermicity of
the all reactions contributing more than 1% to the total heat release in that range include the
following:
H + O2 (+M)  HO2 (+M) (viii)
H + HO2  2 OH (ix)
T = 760 K
T = 854 K
T = 880 K

60

HO2 + OH  H2O + O2 (x)
2 HO2  H2O2 + O2 (xi)
HCO + O2  CO + HO2 (xii)
CH2O + OH  H 2O + HCO (xiii)
CH3 + HO2  CH3O + OH (xiv)
CH3 + HO2  CH4 + O2 (xv)
CH3O + M  CH2O + H + M (xvi)
2 CH3OCH2O2  CH3OCH2OH + CH3OCHO + O2 (xvii)
OCH2OCHO  HOCH2OCO (xviii)
HOCH2OCO  CO + HOCH2O (xix)
HOCH2O  H + HCOOH (xx)
CH2O + OH  HOCH2O (xxi)

Also contributing to the heat release in this range are the aforementioned DME low
temperature oxidation reactions (i) to (vii) which accounted for approximately 54%. It is observed
that a few small molecule chemistry (H2, CH3 reactions) do contribute slightly to the heat release
as temperatures around 900 K are attained. Of importance is the CH2O kinetics which is a main
marker of 1
st
stage ignition as well as a stable intermediate species formed in the NTC region.
Thus, the hatched area ‘c’, predominantly encapsulates the entirety of low temperature
exothermicity and to a lesser extent, a few reactions characteristic of NTC region

6.4 Concluding remarks
Unique attributes of experimental end-gas autoignition pressure gradient observations
during the compression phase were assessed numerically to clarify their relevance towards
providing insights into the experimental results.
The time evolution of species mass fractions and temperature were processed from an
experimental pressure-time history. A 1
st
stage ignition point with respect to the 1
st
stage was
defined and the inflection point prior to 1
st
stage heat release and was found to coincide with the
decomposition point of the stable intermediate, hydroperoxymethyl formate.

61

The maximum heat release attained during 1
st
stage ignition was analyzed through the
oxidation sequence at three different temperature during; one prior to the point, at the point, and
after the point. Investigations carried out numerically suggested that the peak was attributed to the
competition between the beta scission pathway and second molecular oxygen addition to
hydroperoxyl-methoxymethyl radical (CH2OCH2O2H) which typically shuts down the 1
st
stage
chain branching sequence.
While the measure of heat release in the temperature range of 715 K to 930 K was due to
the exothermicity of DME specific low temperature reactions leading to chain branching,
contributing approximately 54%, other contributions stemmed from formaldehyde as well as the
small molecule chemistry reactions of H, CH3 and HO2.

62

Chapter 7: Preliminary investigations of the pressure rise rate
influence on end-gas reactivity
7.1 Introduction
One of the primary advantages of using this CSEF experimental configuration is the
ability to explore the autoignition process occurring in the presence of steep gradients of
thermodynamic pressure. An initial exploratory study on this was performed in [67] to
investigate the unsteady pressure effects on kinetics where it was concluded that, a modification
of the rate of change of pressure could potentially result in a shortening of the 1
st
stage ignition
and an acceleration to high-temperature kinetics (2
nd
stage ignition). Despite this being a realistic
scenario considering the various engine operating conditions, the need to methodically isolate
various influencing parameters is required. This study seeks to further the previous investigation
(Section 5.6) systematically by understanding the effect of the flame burning rate on the end-gas
kinetics and in so doing, draw conclusive remarks about how the thermodynamic history matters
during end-gas reactivity.

7.2 Experimental approach and parameter space
End-gas autoignition experiments were performed in the completely heated spherical
chamber under constant volume conditions, following all the essential procedures stipulated in
Chapter 3. The dynamic pressure transducer was calibrated in the 0-500 psi pressure range of direct
relevance to the experiments with an uncertainty of (±0.7 psi) in the pressure measurement.
Reactive mixtures of DME with an inert dilution of He and Ar, which spanned 100%He
(0% Ar) to 70% He (30% Ar) for the sake of affecting the speed of flame propagation was used

63

for this investigation. He and Ar have identical heat capacities, Cp, thus the Tad will remain constant
assuming one inert is replaced with an equal proportion of the other. To that effect, initial
thermodynamic conditions (Po, To), Xo, ϕ and T ad are listed in Table 7.1.
Mixture ϕ T ad (K) P o (atm) T o (K) % He X DME X O2 X AR X HE
1 0.8 2001 4 468.15 100 0.0301 0.1130 0 0.8569
2 0.8 2001 4 468.15 90 0.0301 0.1130 0.0857 0.7712
3 0.8 2001 4 468.15 80 0.0301 0.1130 0.1714 0.6855
4 0.8 2001 4 468.15 70 0.0301 0.1130 0.2571 0.5998
Table 7.1 Thermodynamic conditions considered in the present investigation

7.3 Numerical approach
End-gas reactivity was modeled using the 0-D adiabatic batch reactor model code
sensBVP which accounts for the temporal isentropic compression of the unburned mixture.
Additionally, the thermo-chemical evolution for various mixtures and sensitivity of τCIDT to
kinetics was performed using sensBVP. Details of the code are provided in Section 4.2.

7.4 Thermal characteristics of the end-gas
Experimental rates of pressure change from CSEF under constant volume conditions for
the initial thermodynamic space listed in Table 7.1 is displayed in Figure 7.1. One striking
observation is that with the gradual replacement of He with Ar, the heat release at the 1
st
stage
tends to occur earlier, despite them having the same “measure” or magnitude. This corresponds
with the trend of slower rates of pressure change with inert content changing from 100% He (0%
Ar) to 70% He (30% Ar). Thus, faster burning rates tend to have a delayed ignition event at both
stages and vice versa.

64

Figure 7.1 Experimental rate of change of pressure as a function of P/Po for Mixtures 1 - 4.

Mixture 1 from Table 7.1, consisting of an inert entirely of He, had the highest rate of
pressure change while the lowest rate of pressure change was experienced by Mixture 4 with a
70% He (30% Ar) inert composition. Higher dP/dt values are obtained for He abundant diluent
mixtures compared to Ar in the initial phase of flame propagation in the compression region
before 1
st
stage ignition due the higher unburned gas temperatures. From first principles, 𝑆 u
o

varies with and heat capacity according to the relation, 𝑆 u
o
~ ( 𝜆 𝐶 𝑝 ⁄ )
1 2 ⁄
[2], as such, the higher
rates of dP/dt results for He abundant diluent mixtures, and this trend subsides as the He
percentage reduces.
Assuming a chemically frozen unburned gas where no chemical reactions and subsequent
two stage ignition occur, the computed results of unburned gas pressure and temperature (P-Tu)
for all Mixtures in Table 7.1 is depicted in Figure 7.2 and shows the CSEF traverses similar
thermodynamic conditions (P-Tu) ideally. In all cases, there exists isentropic compression by

65

virtue of pressure rise from the expanding flame. Therefore, the differences in dP/dt in Figure 7.1
is ideally a marker of the substantial changes in the burning rate.

Figure 7.2 Computed P-Tu functions for an frozen unburned gas for Mixtures 1 – 4 in Table 7.1.

Using this P-Tu variation and initial mole fraction in Table 7.1, while still assuming a
frozen unburned gas, the flame speed, (Su) as a function of Tu does indeed depict the changes in
burning speed shown in Figure 7.3.

66

Figure 7.3 Comparison of computed Su vs Tu for flames of Mixtures 1 - 4 in Table 7.1 for a
chemically frozen unburned gas

Despite the assumption of a chemically frozen unburned gas, a bit of reactivity might occur in
the unburned gas, hence, the notation of flame speed (Su) in contrast to laminar flame speed (𝑆 u
o
) .
Nevertheless, with the reduction in the He percentage in the inert, a 10% difference in Su does
ensue per 10% reduction in one inert component.
7.5 Effect of flame burning rate on end-gas reactivity
Given the experimental and numerical results obtained, the experimental pressure traces
were all submitted as input to sensBVP to monitor the temporal evolution of the species mass
fraction, temperature and perform sensitivity analysis of the characteristic end-gas autoignition to
kinetics. Although, the results in Figure 7.2 show similar thermodynamic (P-Tu) trajectory,
albeit, for a chemically frozen end-gas, substantial changes for each reactive isentrope in terms
of ignition for both stages are observed in Figure 7.4(a), initiating at differing pressures.

67

Figure 7.4 Computed P-Tu functions for Mixtures 1 – 4 in Table 7.1

A zoomed portion of the reactivity commencement and 2
nd
stage ignition in Figure 7.4(b)
displays show similar trend to Figure 7.1. As the burning rate reduces (slower compression rate)
by the decreased He content from 100% (0% Ar) to 70% (30% Ar) by mole, the results suggest
that substantial time is afforded the chemical inductive and radical initiation processes to occur
in the end-gas, in addition to an accumulated heat release, thus, an earlier 1
st
stage ignition is
achieved. Consequently, the radical pool has sufficient time to build up and promote the chain-
branching sequence considering their respective timescales of reactions. As the 2
nd
stage ignition
is directly coupled to the 1
st
stage ignition through its heat release, an earlier 2
nd
stage ignition is
achieved for the slower flame 70% He (30% Ar).
The results of the computed key radical (OH, HO 2) profiles controlling 1
st
stage ignition
in Figure 7.5 indicate similar orders of magnitude in relation to the thermodynamic pressure.
However, there is an offset due to the earlier ignition at both stages for the 70% He (30% Ar)
case compared to the 100% He (0% Ar) case.

68

Similarly, the ranked logarithmic sensitivity coefficients of the characteristic ignition
delay time to kinetics show the same set of controlling reactions and magnitudes for all mixtures.
This points to the fact that the kinetics are not affected to a large extend and the time history of
the compression affects only the time at which ignition occurs, or the radical pool builds up.

Figure 7.5 Computed temperature, OH and HO2 mass fractions profiles in the unburned mixture
as a function of the normalized pressure for 100% He (0%Ar) [ ─] and 70% He (30% Ar) [- -].

69

Figure 7.6 Ranked logarithmic sensitivity coefficient for Mixtures 1-4 in Table 7.1

.
7.6 Concluding remarks
In conclusion, the effects of the rate of pressure change on kinetics was explored to
understand the potential changes in the unburned gas with respect to low temperature ignition.
This was accomplished by modifying the flame propagation speed of reactive DME mixtures
through inert compositions of Helium and Argon possessing the same specific heat capacities by
different thermal conductivities.
The experimental results of four mixtures at similar initial thermodynamic conditions but
with varying ratios of Helium and Argon showed identical ignition characteristics at both stages
with an offset due to the different flame burning rates. Computed isentropic compression results
-0.08 -0.07 -0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03
CH3OCH2O2 <=> CH2OCH2O2H
CH2OCH2O2H + O2 <=> O2CH2OCH2O2H
HO2CH2OCHO <=> OCH2OCHO + OH
CH2OCH2O2H <=> 2 CH2O + OH
O2CH2OCH2O2H <=> HO2CH2OCHO + OH
CH3OCH3 + O2 <=> CH3OCH2 + HO2
HOCH2OCO <=> CO + HOCH2O
H + O2 (+M) <=> HO2 (+M)
CH3OCH3 + HO2 <=> CH3OCH2 + H2O2
CH3OCH3 + H <=> CH3OCH2 + H2
CH3OCH3 + OH <=> CH3OCH2 + H2O
Logarithmic Sensitivity Coefficients
100% He
90% He
80% He
70% He

70

of the chemically frozen unburned gas showed the unburned gas traversed the same pressure-
temperature trajectory but at different burning rates.
Zero-dimensional modeling of the end-gas thermo-chemical evolution displayed an
earlier commencement of the 1
st
and 2
nd
ignition stages with lower flame burning rates
suggesting sufficient time was given for the radical pool to build up. Sensitivity of characteristic
ignition delay time to kinetics displayed the exact controlling chemical pathways and associated
magnitudes of the logarithmic sensitivity coefficients.

71

Chapter 8: Fuel additive effects on end-gas autoignition in
spherically expanding gasoline flame experiments
8.1 Introduction
Low temperatures and high compression ratios are desirable conditions for achieving the
goal of increasing efficiency, decreasing hydrocarbon and NOx emissions to improve internal
combustion engine (ICE) efficiency while conforming to stringent societal regulations [8]. A major
factor limiting these next generation high density ICE’s is the tendency of the end-gas to autoignite
leading to the onset of undesirable knock that produces a loss in engine power output and
potentially results in engine damage in the long term. Implemented advanced engine strategies and
combustion modes including direct fuel injection, exhaust gas recirculation and fuel lean burning
conditions have yielded encouraging results and provided the impetus towards knock mitigation
[49][92][93]. Equally important to these technologies have been the development and integration
of fuel additives by harnessing their associated chemical effects to significantly alter the unburnt
fuel mixtures’ thermo-chemical state for knock suppression.
Additives, a vital part of today’s fuels, are carefully blended up to a maximum treat rate of
2% by volume into a formulated multicomponent base-fuel (gasoline or diesel) composition to
improve performance, efficiency and emissions. Clarification is hereby made with fuel
components which usually have a treat rate greater than 2% by volume. Nevertheless, for all
purposes related to this study, fuel components will herein be referred to as additives. Fuel
additives generally improve combustion by directly modifying the radical pool or altering key
kinetic pathways [94] to enhance the flame propagation and ignition characteristics.
Ignition promoters, germane to compression ignition engines, undergo rapid thermal
decomposition under low-temperature conditions that release radicals participating in chain

72

initiation and branching processes to advance the reactive mixture’s ignition [95]. Anti-knock
additives, employed in spark-ignition engines increase the octane number and thereby, the fuel
mixture’s resistance to knock [7][96] by scavenging chain initiating radicals or interrupting chain
branching processes, thereby suppressing autoignition [94][96].
The iterative developmental process [43] for fuel additives principally involves the initial
formulation, base-fuel blending and finally engine testing. During the latter stage, hundreds of
additives require critical examination; even more so as modern engines typically tend to exhibit
unique operational conditions. Assessing the additives’ essential performance, by the hundreds,
does prove costly, laborious, impractical and necessitates a pragmatic, expedited and guided
methodology prior to engine testing. Ideally, such inexpensive screening tools should be
qualitatively capable of capturing engine-relevant combustion behavior to help identify additives
and provide the requisite guidance that triggers detailed studies requiring the proper scrutiny and
significant investment.
To date, the detailed effects of fuel additives on ignition propensity have been
experimentally investigated in rapid compression machines, shock tubes, and constant volume
chambers under low-temperature conditions to elucidate their kinetics at engine-like conditions.
Various studies on ignition promoters and anti-knock additives have been summarized [10] while
Boot et al. [97] have briefly highlighted their mode of operation and working principles; both
studies were performed under low-temperature conditions. 2-Ethylhexyl nitrate (2-EHN), a
predominant ignition promoter, has been extensively investigated in [95][98][99]. Results from
these studies indicate that its rapid decomposition at temperatures less than 550 K, produces a
radical pool having a kinetic and thermal interaction which culminates in accelerating ignition.
Aromatics, a class of anti-knock additives have been known to display exceptional ignition

73

suppression qualities. Walsh in a study of aromatic compounds in [100], found a correlation
between the antiknock effect and the side chains influence based upon the benzene ring’s electronic
properties. He suggested an increased antiknock effect with a corresponding decrease in the
binding of the ring electrons with a few exceptions. An experimental study of 104 aromatic amines
by Brown et al. [101] found many effective anti-knock compounds particularly, 4-sec-Butyl-o-
phenylenediamine, p-ethylaniline, n-methylaniline and n-nitrosodiphenylamine. The authors
attributed the strong knock suppression mechanism to the formation of the resonantly stabilized
aromatic amine radicals.
The CSEF configuration does also offer itself as a complementary screening technique to
the aforementioned legacy approaches for fuel additive investigation as the autoignition process
can be evaluated in the presence of pressure gradients applicable to the compression stroke in
ICE’s while being low-cost since relatively small quantities of fuel (< 20 mL) and other gaseous
components are required. The progressive sweep through varying thermodynamic states with its
associated flow-field and physical phenomena, simplify the resulting physics for tractable
modeling with detailed kinetics.
Flame-induced, end-gas autoignition using the CSEF approach has been experimentally
investigated and demonstrated using flames of dimethyl ether (DME) reactive mixtures in [67],
where the distinct features from the pressure gradient were observed and characterized, based off
the initial proposition by Hu and Keck in [41][42]. Similar experimental investigations conducted
in [74][75] as well as numerical works in [21][22][23][24][25][26][27][31] have also elucidated
the key fundamental physics of flame-induced autoignition and overall, provided sufficient
evidence this technique remains a complementary low-cost approach for autoignition studies at
practical engine conditions. Experimental investigations into autoignition and detonation

74

phenomena via flame-induced compression have also been studied in [76][77][78], albeit, in the
RCM and constant volume configuration.
Much alike to experimental techniques for screening fuel additives, a computational tool
should be capable of making inexpensive qualitative predictions while precisely capturing reliable
trends over a wide range of conditions through numerical simulations. RMG [68][69], an open-
source software package, purposely achieves this by automatically constructing chemical kinetic
models using a flux-based algorithm for model expansion in the shortest possible time for reactive
mixtures to assess and make predictions of the ignition tendency of various fuel additives. In [44],
RMG was applied to construct chemical models for mixtures of six substituted phenols as 2%
additives in n-butane to rank their anti-knock performance, thus serving as an economic means to
practically screen fuel additives. Considering the current tools available and challenges involved
in producing high fidelity kinetic models over a large thermodynamic space, RMG does proffer a
unique solution to provide guidance and feedback to the fuel and engine community for identifying
potential additives which warrant further detailed analysis.
Besides, RMG has been employed in the quantitative modeling of aromatic formations in
fuel-rich methane oxy-combustion [102] and investigation of alkylaromatics [103]. Furthermore,
it has been extended to account for nitrogen species for the nitrogenous systems pyrolysis and
oxidation [104], building reaction networks for di-tert-butyl sulfide thermal decomposition
[105]and developing a detailed comprehensive model of JP-10’s combustion and pyrolysis kinetics
[106].
In view of these, this present study seeks to demonstrate a methodology to systematically
screen fuel additives for engine testing. This was achieved by investigating the end-gas
autoignition of a gasoline/additive/oxygen/inert mixtures by leveraging the CSEF approach at the

75

following conditions: equivalence ratio ϕ = 0.8 and 1.0, initial temperature To = 468 and 493 K,
and initial pressure, Po = 9 atm. Commercially available fuel additives investigated include 2-
EHN, decalin, aniline and 1,3,5-trimethylbenzene (135-TMB). The experimental observables from
the two-stage ignition process were characterized similar to [67] and used as comparative analysis
for the fuel additive ignition tendency. The anti-knock effects are qualitatively investigated
through 0-D numerical simulation predictions using RMG developed chemical kinetic models.
Finally, interpretation was performed to provide insights into effects of additives on autoignition
for screening/selection purposes towards future fuel and engine design.

8.2 Experimental approach
End-gas autoignition for flames of gasoline were investigated using the CSEF approach
alongside all relevant procedures and details described. The experimental P-t trace was used for
analysis as such, the dynamic pressure transducer was calibrated in the 0 – 750 psi pressure range
of relevance to the experiments to reduce the uncertainty (±1.2 psi) of the pressure measurement.
Gasoline fuels were supplied by Shell Global Solutions (US) while O2, Ar, and He from Gilmore
Gases had respective purities of 99.8%, 99.998% and 99.999%.

8.3 Parameter space
Flames of gasoline/additive blends with Po, To, initial adiabatic flame temperature (Tad),
mole fractions (Xi) compositions, ϕ, fuel additive and its percentage (in the fuel) for each mixture
as listed in Table 8.1 was used for end-gas autoignition investigations.

76

Mixture T o (K) P o (atm) ϕ T ad (K) Additive (%) X fuel X O2 X AR X HE
1 468 9 1.0 1950 - 0.009021 0.090435 0.315191 0.585354
2 468 9 1.0 1951 2-EHN (0.1) 0.009020 0.090435 0.315191 0.585354
3 468 9 1.0 1952 Decalin (20) 0.008288 0.090502 0.315424 0.585787
4 493 9 0.8 1975 - 0.009166 0.114857 0.306592 0.569385
5 493 9 0.8 1978 Decalin (20) 0.008421 0.114944 0.306822 0.569813
6 493 9 0.8 1976 Aniline (1) 0.009186 0.114855 0.306586 0.569373
7 493 9 0.8 1985 135TMB (20) 0.008821 0.114897 0.306698 0.569583
Table 8.1 Thermodynamic conditions and mixture summary considered in the present
investigation.

Per molar basis, an inert dilution ratio of He/Ar = 65/35 was used for all experiments whiles
the ratio of oxidizer to inert ratio was also fixed. It is hereby noted that for additives that suppress
ignition (135-TMB and aniline), experimental investigations were only performed at ϕ = 0.8,
Po = 9 atm and To = 493 K, due to the end-gas autoignition phenomenon interacting with the wall
and this is further discussed in a later section.

8.4 Modeling approach
Gasoline fuel composition and its properties
The gasoline base-fuel with its key properties including the average molecular weight and
RON were listed in Table 8.2. The composition of the gasoline base-fuel is specified in Figure 8.1
Fuel RON MW (g/mol)
Gasoline Base-fuel 80 92.5
Table 8.2 Key properties of gasoline base-fuel.

77

Figure 8.1 Mass percentages of the fuel components.

For modeling the combustion chemistry of a real-fuel, including gasoline, jet and rocket
fuels, the hybrid chemistry (HyChem) approach [107][108][109] has in recent times been used as
a viable alternative to the surrogate approach. The former incorporates a fuel pyrolysis lumped-
step model under high temperature conditions and a foundation fuel chemistry model, specifically,
H2/CO/C1-C4/benzene/toluene species to describe the oxidative products of pyrolysis. Compared
to the traditional surrogate fuel approach for combustion chemistry modeling, where a variety of
neat hydrocarbon components (up to 10 components) are used to mimic both the physical and
chemical properties, HyChem employs the use of experimental data from shock tubes and flow
reactors to derive key parameters of the lumped model of specific products of the fuel oxidative
and pyrolysis process.
With the current study seeking to make predictions of additive effects on end-gas
autoignition, the gasoline HyChem model [109] was initially proposed to be used in conjunction
with RMG. However, a limitation in RMG is the number of product species for reactions which
posed a challenge considering the lumped reactions of HyChem. Secondly, each species ought to

78

be represented by a unique molecular structure which was unavailable for the HyChem gasoline
related species.
In view of these restraints, a surrogate [110][111] approach was opted for, to model the
gasoline base-fuel. A three-component gasoline surrogate made up of 39.85% iso-octane, 28.58%
n-heptane and 31.57% toluene by volume was chosen to represent the base-fuel which matched
the RON of 80 based on the work of Javed et al. [112]. It is hereby emphasized that, numerical
modeling served to capture reasonable qualitative trends and insights into the additive chemistry
as opposed to providing experimental data validation. The gasoline kinetic model by Wu et al. [72]
containing 165 species and 1403 reactions including both low and high temperature chemistry was
used as a basis for numerical simulations.

Ignition delay time
Ignition delay time simulations of particular anti-knock mixtures in Table 8.1, (using air as
oxidizer) at a constant pressure of 30 atm and temperatures from 700 – 950 K were performed
using the sensBVP [55] code and RMG chemical kinetic models (discussed in the next section).
Kinetic model generation and refinement
The 2EHN and 135-TMB models were automatically generated using RMG version 2.4.1
while the aniline model was generated and refined using a combination of four software tools. The
Tandem Tool (T3) [113] version 0.1.0 was used to iteratively call the RMG software package
version 3.0.0 to generate kinetic models and the Automated Rate Calculator (ARC) [114] version
1.0.0 for automating electronic structure calculations and attaining refined thermodynamic
properties of selected chemical species. ARC used the Automated Reaction Kinetics and Network

79

Exploration (Arkane) [68] tool to perform the necessary statistical mechanics computations to
derive partition functions.
A species was selected for refinement by T3 if its thermodynamic properties assigned by
RMG were derived from a group additivity estimation [68] and if it is ranked among the top 10
species with respect to the highest absolute normalized sensitivity coefficient of [OH] to the
respective standard Gibbs energy of formation. The RMG methodology was described elsewhere
in details [44][68][69][102][103][104][105][106]; briefly, RMG utilizes a rate-based algorithm to
automatically discover important species and associated reactions to generate chemical kinetic
models. Thermodynamic parameters and rate coefficients data are obtained from a database where
available or estimated mainly using the group additivity method and kinetic rate rules using pre-
defined reaction family templates. An isothermal isobaric gas-phase batch reactor was used for
generating the model at 30 atm and 700-950 K. Kinetic and thermodynamic data were taken
primarily from [72]. Thermodynamic data was augmented from a library in the RMG database
repository with values computed at the CCSD(T)F12A/cc-pVTZ-F12//B3LYP/6-311++g(d,p)
level of theory. Reaction rate coefficients were augmented from three literature sources
[115][116][117] where available. Phenomenological rate coefficients for unimolecular and well-
skipping reactions were computed by RMG using the Modified Strong Collision [118]
approximation. Electronic structure calculations done by ARC were at the CBS-QB3 [119][120]
level of theory using the rigid rotor harmonic oscillator approximation [119][120] and corrected
using bond additivity corrections implemented in Arkane. A zero-point energy scale factor of 0.99
[119] was used. Many computed species exhibited coupled internal torsional degrees of freedom.
Since conducting multidimensional coupled torsional scans was computationally impractical, the
common approximation of N-dimensional coupled torsional system as N uncoupled, 1-

80

dimensional torsional modes were used. Hence, anharmonic torsional modes where treated using
torsional scans at 8º increments at the B3LYP/ 6-311G(2d,d,p) level of theory, which is the CBS-
QB3 optimization and frequency level.
The gasoline surrogate described earlier was used to represent the base-fuel with 0.1%, 1%
and 20% by mole of 2EHN, aniline and 135-TMB, respectively, of the fuel additive included in
the reaction system under conditions of temperature 700 – 950 K and pressure 30 atm using air as
oxidizer. The gasoline surrogate model hereafter called Model 2 was seeded into RMG.
Independent RMG models for mixtures with gasoline surrogate/additive blends were generated
with their final specifics listed in Table 8.3.
Fuel additive Model Number of species Number of reactions
Aniline Model 2-Aniline 183 1660
135-TMB Model 2-135TMB 180 1767
2-EHN Model 2 – 2EHN 305 4521
Table 8.3 RMG Model nomenclature and size.

8.5 Fuel additive effects on experimental observables
As stated earlier, end-gas autoignition has previously been investigated using this
experimental configuration [67], albeit, using flames of DME where the time-resolved
experimental pressure derivative (dP/dt) was used for analysis. Results from the authors showed
dP/dt in the compression region initially displayed linearity signifying flame propagation without
autoignition followed by a ‘bump’ and finally a rapid rise in dP/dt, attributed to the DME two-
stage ignition. 1-D DNS further buttressed the experimentally observed two-stage ignition
characteristics and highlighted the kinetically controlled end-gas autoignition at low-temperatures,
the versatility of this experimental configuration to offer the requisite timescales for low-
temperature chain-branching processes to significantly advance autoignition and the ability to
attain engine-relevant end-gas conditions. It is noted that in the absence of any autoignition, this

81

non-monotonic behavior in the experimental dP/dt is not observed but rather, a near linear trend
ensues.
With the degree of low-temperature chemistry progression dictating the magnitude of 1
st

stage ignition manifested through dP/dt, different fuel blends will exhibit distinct behaviors
sufficient to make reasonable comparison and conclusions. It should be noted that dP/dt changes
are key indicators of temperature and the unburnt gas evolution owing to the chemical energy
release from exothermic reactions. All experimental comparative analysis hereafter is scaled with
the base-fuel to indicate the percentage improvement in all characteristics.
To that end, the experimental dP/dt for Mixture 4 at initial conditions stated in Table 8.1
is presented in Figure 8.2 to highlight the key observables used for characterization and
comparison. The time axis was re-scaled to begin (t = 0 ms) at the commencement of considerable
end-gas compression (P/Po > 2.5) [38][48][122] by the CSEF. P instead of t was chosen as the
independent variable, given that it categorically described the unburned gas state while
experimental issues such as temporal delay resulting from ignition effects and jitter in experimental
time, t = 0 ms were circumvented.

82

Figure 8.2 Ignition characteristics for Mixture 4 for Po = 9 atm and To = 493 K during the
compression stage of CSEF propagation. Point ‘a’ – 1
st
stage ignition point (Pign_1); Point ‘b’ – 1
st

stage pressure rise rate; Point ‘c’ – dP/dtmax; Point ‘d’- 2
nd
stage ignition point (Pign_2); Point ‘e’ –
2
nd
stage pressure rise rate; Hatched Area ‘f’ – Measure of 1
st
stage exothermicity; τ1 –
Characteristic 1
st
stage ignition delay time.

The 1
st
stage ignition point presented in Figure 8.2 as Point ‘a’ was the inflection point
indicating substantial commencement of low-temperature chain branching activity. Likewise,
Point ‘d’, the 2
nd
stage ignition point was the inflection point after 1
st
stage ignition but preceding
the exponential rise in dP/dt.
These two ignition point (“a” and “d”) values, scaled with that of the base-fuel, were
compared among all fuel Mixtures in Table 8.1 for both ϕ’s. It was observed from Figure 8.3a that
Mixtures 2 and 3, containing the additives, 2-EHN and decalin, respectively, promoted reactivity
development by a decrease in the ignition points ‘a’ by ~ 2% and ‘d’ by ~ 4%. Similarly, decalin’s
effect in Mixture 5 in Figure 8.3b was by 3% and 6% at points ‘a’ and ‘d’, respectively. Of the
additives tested, aniline and 135-TMB present in Mixtures 6 and 7 both saw a ~ 5 – 7% increase

83

at point ‘a’. However, at point ‘d’, there was a substantial increase of suppression by ~12%
exhibited by 135-TMB compared to the base-fuel.

Figure 8.3 Scaled 1
st
and 2
nd
stage ignition point for all fuels ignition at (a) To = 468 K, ϕ = 1.0
and (b) To = 493 K, ϕ = 0.8.

In ICEs, a key metric of engine performance indicating the rate of exothermic heat release
during combustion is the maximum rate of pressure rise that usually occurs prior to the peak
pressure at ignition. For a two-stage ignition process, a higher rate of pressure-rise during 1
st
stage
ignition translates to a higher temperature afterwards during the combustion cycle ending with a
shorter overall ignition delay and vice versa. In essence, a very small slope change can directly
translate into a substantial combustion duration across the entire engine cycle. This maximum rate
of heat release indicator has been quantified through a characteristic timescale, τc for maximum
pressure rise as ( d𝑃 d𝑡 ⁄ . 1 𝑃 ⁄ )
−1
, during either ignition stage and subsequently scaled with the
base-fuel value. Indicated as Points ‘b’ and ‘e’ on Figure 8.2 at the 1
st
and 2
nd
stage ignition event,
respectively, it was interpreted as the ‘characteristic time’ taken for ignition (signified by pressure
rise) to occur. Thus, in general, additives significantly retarding ignition have a longer τc, while
those promoting ignition possess a shorter τc as observed below in Figure 8.4

84

Figure 8.4 To = 493 K, ϕ = 0.8 (a) Section of experimental pressure rise rates up to 1
st
stage
ignition (b) Scaled characteristic time τc during 1
st
stage pressure rise. (c) Section of
experimental pressure rise rates during both ignition stages (d) Scaled τc during 2
nd
stage pressure
rise.

dP/dt in Figure 8.4 (a) and (c) provides a comparative representation of the steepness of
the slope during the 1
st
and 2
nd
stage ignition, respectively, with only a section displayed up to the
maximum ‘heat release’ for the 1
st
stage in Figure 8.4 a while in (c) also, a section of both ignition
stages are shown. It is also emphasized that after 1
st
stage ignition, flame propagation still
continues before 2
nd
stage ignition. However, depending on the radicals and intermediates formed
in the unburned gas, as well as the extent of heat release, this results in a different thermo-chemical
state for each mixture and thus, the isentropes traversed before 2
nd
stage ignition are bound to
differ. This can be observed in Figure 8.4c.

85

It can be inferred from Figure 8.4 that additives, successful at scavenging or diverting
active radicals to delay ignition will largely have a sluggish exothermic heat release process; an
example being Mixture 7 containing 135-TMB having τc showing the longest characteristic time
for ignition by a factor of 2 and 7 at the 1
st
and 2
nd
stage, respectively. This can visually be seen in
Figures 8.4 (a) and (c) where the gentle slope compared to the base-fuel ensued suggesting a mild
heat release ignition process. Of interest is aniline in Mixture 6, where the change in τc was
miniscule indicating aniline hardly suppressed the rate of pressure rise. Mixture 5 containing
Decalin, was seen to also have a slightly lower τc at both stages.

Figure 8.5 To = 468 K, ϕ = 1.0 (a) Section of experimental pressure rise rates up to 1
st
stage
ignition (b) Scaled characteristic time τc during 1
st
stage pressure rise. (c) Section of
experimental pressure rise rates during both ignition stages (d) Scaled τc during 2
nd
stage pressure
rise.

86

The results for To = 468 K, ϕ = 1.0 in Figures 8.5 (a-d) above displayed a similar trend to
the cases of To = 493 K, ϕ = 0.8 in Figure 8.4, but were investigated for the ignition promoting
additives 2-EHN and decalin for Mixtures 2 and 3, respectively. Compared to the base-fuel results
of Mixture 1, both additives showed a slight decrease in τc at both ignition stages which was
expected.
Point ‘c’ in Figure 8.2 displays the maximum pressure rise (dP/dtmax) during 1
st
stage
ignition. This experimental observable is particularly important as it is a good benchmark of the
maximum heat release attainable that will be translated to the 2
nd
stage of flame propagation and
subsequent autoignition. As observed in Figure 8.6a, the ignition improvers 2-EHN and decalin in
Mixtures 2 and 3, respectively, for the conditions To = 468 K, ϕ = 1.0, attain higher maximum
exothermic heat release, i.e. ~2-6% compared to the base-fuel in Mixture 1. For the results
presented in Figure 8.6b at conditions of To = 493 K, ϕ = 0.8, 135-TMB is seen to substantially
have a lower dP/dtmax by ~16% while decalin increased it by ~4%. Interestingly, aniline which
supposedly acts as an anti-knock compound, however, displays a relatively high scaled dP/dtmax
suggesting it largely plays a significant role between the 1
st
and 2
nd
stage ignition. Thus, aniline or
an associated fuel radical is an extremely good radical scavenger after the 1
st
stage process.

Figure 8.6 Scaled maximum dP/dt at 1
st
stage ignition (a) To = 468 K, ϕ = 1.0 (b) To = 493 K,
ϕ = 0.8.

87

Due to the propagating flame front and autoignition simultaneously contributing to
combustion and heat release, the scaled measure of heat release (MHR) due to 1
st
stage ignition is
the integrated semi-hemispherical area under the bump (area ‘f’) scaled by that of the base-fuel.
Computed in this manner, it indicates the percentage improvement compared to the base-fuel. The
MHR is indicative of the total energy release during low-temperature oxidation and is the main
driver towards arrival at the 2
nd
ignition stage. As a result, the larger this area is, the shorter the
total ignition delay and the smaller it is, the longer the total ignition delay. In general, reducing the
intensity of 1
st
stage ignition may well be the key to effective anti-knock action. The result in
Figure 8.7 displayed against τ1 reveals substantial differences in the 1
st
stage ignition.

Figure 8.7 Scaled measure of heat release as a function of 1
st
stage ignition delay time (a)
To = 468 K, ϕ = 1.0 (b) To = 493 K, ϕ = 0.8.

As expected, for the ignition promoting additives, 2-EHN and decalin (i.e. Mixtures 2 and
3), at conditions of To = 468 K, ϕ = 1.0, saw a 13% and 24% increase in MHR compared to the
base-fuel while at To = 493 K, ϕ = 0.8, aniline and 135-TMB achieved a 12% and 67% decrease,
respectively, in Figure 8.7 above. Decalin was also seen to achieve a slightly higher increase of
30% at this condition.

88

These differences seen in MHR suggest that the additives are capable of significantly
interrupting the exothermicity itself. From chemical kinetic theory, the main heat release steps are
mostly from decomposition of the ketohydroperoxdes and the hydrogen abstraction initiation step
from the parent fuel. Nevertheless, this remains an important characteristic as it dictates the
magnitude of the successive 2
nd
stage ignition.
As the knock resistance of practical fuels is measured in terms of its RON, various studies
(e.g.,[123][124]) have sought to develop correlations between this (RON) and ignition delay times
in Ignition Quality Testers as well as legacy experiments. Since, fuel mixtures which readily ignite
easily will have a lower ignition and vice versa, this trend was shown using the MHR and the 1
st

stage ignition delay time for only the decalin-gasoline (Mixture 5) and 135-TMB-gasoline
(Mixture 7) fuel mixtures compared with the base-fuel. RON values for these fuel blends were
estimated based on the expression of Foong et al. [125] with the RON of the neat fuels obtained
from [96][126]. The resulting RON for decalin-gasoline surrogate and 135-TMB-surrogate fuel
blend were computed as 72.4 and 91.4, respectively, with the results of their characteristic trends
during 1
st
stage ignition shown in Figure 8.8. The general observation expected is that as the RON
increases, the MHR decreases while the 1
st
stage ignition delay time increases.

89

Figure 8.8 Scaled measure of heat release and 1
st
stage ignition delay time as a function of the
estimated RON at To = 493 K, ϕ = 0.8.

The presented results in Figures 8.3 – 8.7 had the additive 2-EHN only included for
investigations at a lower initial temperature, To = 468 K in Mixture 2. The oxidation of reactive
fuel mixtures enhanced by 2-EHN has been known to vary due to the decomposition chemistry
over a narrow temperature range of the latter, yet, producing significant effects. To assess this
experimentally, flames of gasoline/2-EHN/oxygen/diluent mixtures were investigated at
thermodynamic conditions as shown in Table 8.4. Similar to Table 8.1, T ad and the oxidizer/diluent
ratio were fixed for all the test Mixtures for the two initial temperatures To = 468 and 493 K. It is
noted that for the additive containing fuel, 2-EHN was only present at 0.1% concentration in the
fuel.

90

Test
Mixture
T o (K) P o
(atm)
ϕ T ad (K) Additive
(%)
X fuel X O2 X AR X HE
1 468.15 9 0.8 1975 - 0.009337 0.117006 0.305780 0.567877
2 468.15 9 0.8 1975 2-EHN
(0.1)
0.009336 0.117006 0.305780 0.567877
3 493.15 9 0.8 1975 - 0.009166 0.114857 0.306592 0.569385
4 493.15 9 0.8 1975 2-EHN
(0.1)
0.009165 0.114857 0.306592 0.569385
Table 8.4. Parameter space considered in the present investigation.

The experimental pressure gradient presented in Figure 8.9a depicts the case at To = 468 K
where 2-EHN is observed to promote ignition taking into consideration, all ignition characteristics
as described previously in Figure 8.2. However, for To = 493 K in Figure 8.9b, the ignition
promoting characteristics are diminished and displaying similar ignition characteristics as the base-
fuel firmly suggesting a temperature range of effectiveness and/or a different set of intermediate
species formed not enhancing chain radical branching.

Figure 8.9. Experimental comparison of the Basefuel and 2-EHN for (a) Test Mixtures 1 and 2
at To = 468 K (b) Test Mixtures 3 and 4 at To = 493 K.

91

An experimental study by Stein et al. [127] investigated the oxidation of surrogate diesel
fuels with and without 2-EHN and found that the full decomposition and vaporization occurred
during a narrow temperature range. This could possibly suggest that, at higher temperatures
To = 493 K, the decomposition rate is significantly reduced and its kinetic effect is not realized in
accelerating low-temperature reactivity and ignition.
In this experimental configuration, heat loss occurs as the propagating flame approaches
the chamber wall during unburnt gas compression. At the same time, it is this flame-induced
compression that triggers ignition of the reactive fuel mixture. Thus, experimental conditions to
promote end-gas autoignition ought to be tailored through mixture conditions such that both
ignition stages occur devoid of any heat loss to the wall.
From Table 8.1, experimental investigations were not conducted for aniline and 135-TMB
mixtures at a lower initial temperature, To = 468 K, due to this heat loss occurrence as the 2
nd
stage
of ignition was masked. This phenomenon is evident by a decreasing pressure gradient before
ignition and thus, that data was deemed unhealthy and discarded entirely.

8.6 Ignition delay time predictions
As stated in Section 8.4.3, RMG Models (Table 8.3) were generated for the gasoline
surrogate additive (2-EHN, aniline and 1,3,5-trimethylbenzene) fuel blend mixtures. The seeded
gasoline model, Model 2, (reduced from [72]) was used as the basis for RMG to produce new
models. IDT simulations under constant pressure were performed for gasoline-surrogate-additive
air mixtures at initial conditions of Po = 30 atm, To = 700-950 K and ϕ = 1.0 for both RMG Models
and compared with results of gasoline-surrogate-air mixture using Model 2. For the additive
blends, Model 2-2EHN had 0.1% of the fuel replaced by 2EHN, Model 2-Aniline had 1% of the

92

gasoline surrogate replaced by aniline while Model 2-135TMB had 20% of the fuel replaced by
135-TMB.

Figure 8.10. Predicted ignition delay time modeling for mixtures in air for gasoline surrogate
/additive blend ϕ = 1.0, P = 30 atm.

Results presented in Figure 8.10 satisfactorily predict IDT enhancement with the addition
of anti-knock additives especially at T < 870 K, similar to experimental trends in Section 8.5.
Approximately, a 1.5-4 fold increase in IDT was observed with the addition of 20% 135-TMB to
the fuel using Model 2-135TMB. This was expected as the aromatic composition of the fuel was
increased from 40% to 52%. With the fuel mixture, comprising a 48% of paraffinic composition,
it was expected the NTC region governed by their low-temperature chemistry will be well
observed. Model 2-Aniline predictions of 1% aniline addition to the gasoline surrogate resulted in
an IDT increase by a factor of ~ 3 at 700 K. This effect gradually waned as higher temperatures
were approached until T > 900 K where it was less effective. The additive 2EHN was particularly
effective in promoting ignition especially at T < 850 K by as much as a factor of ~ 2 (at 700 K)
with its effect declining with higher temperatures.

93

Comprehensive descriptions of the gasoline surrogate, ignition controlling pathways are
provided in [128][129][130][131], however, an in-depth kinetic analysis including the initial
thermal decomposition, ignition augmenting/inhibition oxidative pathways of the additive, bond
dissociation energy calculations, maximum concentrations to suppress ignition, steric effects and
inductive effects are not presented as the main endeavor of this study was the experimental
assessment and model predictions and trends.
Considering the fuel additives displaying various desired experimental characteristics and
numerical trends, the key question to be answered is how the experimental data and predictions
translate to a real engine test and what the selection criteria for fuel additives are. Assuming a 5%
increase is reported in the MHR compared to the base-fuel, is this performance enhancement
significant enough to be recorded in an engine test? It should be noted, engine tests capture
additional phenomena such as the fuel mixing, different compression ratios and turbulence, among
other factors. As such, engine-data with these gasoline additives blends will be required to form
correlations and draw conclusions.

8.7 Concluding remarks
The present study details the development of a low-cost methodology to evaluate the
effects and trends of fuel additives in reactive mixtures at engine-relevant conditions through a
comparative analysis of their ignition tendency. This was accomplished experimentally using
confined spherically expanding flames by characterizing the low-temperature ignition features of
fuel additives reactive mixtures of the end-gas and numerically through zero-dimensional
numerical simulation predictions with the aid of detailed kinetic models developed from an
automatic mechanism generator.

94

End-gas autoignition for confined spherically expanding flames of gasoline-additive
blends/oxygen/argon/helium reactive mixtures at an initial pressure of 9 atm, initial temperature
of 468 and 493 K and equivalence ratio 0.8 and 1.0 were investigated. Commercially available
fuel additives examined consisted of 2-ethylhexyl nitrate, decalin, aniline and 1,3,5-
trimethylbenzene. Distinct attributes from the pressure gradient derived from the experimental
pressure-time history specifically during the compression stage, was used to characterize the
autoignition phenomena applicable to engines. For all the reactive mixtures investigated, those
consisting of 2-ethylhexyl-nitrate and decalin fuel additives exhibited ignition augmenting effects
while the others, 1,3,5-trimethylbenzene and aniline demonstrated ignition inhibition tendencies.
Zero-dimensional numerical simulations were performed to predict the knock suppression
of reactive mixtures of gasoline surrogate-additive fuel blends using kinetic models automatically
developed with Reaction Mechanism Generator. Ignition trends similar to the experimental results
were captured. Ignition delay times were simulated at constant pressure (30 atm) and low-
temperatures (700–950 K). It was found that, the predictions of mixtures in air of blended gasoline
surrogate and fuel additives followed similar trends to experimental results, particularly, aniline
and 1,3,5-trimethylbenzene retarded ignition by factor of ~3 and 1.5-4, respectively.
This autoignition study thus represents a complementary low-cost methodology to provide
an effective and efficient screening tool when coupled with numerical tools to investigate the
mechanistic effects of fuel additives on end-gas autoignition. RMG can be a useful numerical tool
to predict qualitative and semi-quantitative trends of fuel additive effects in reactive mixtures.

95

Chapter 9: Conclusions and Recommendations
9.1 Concluding Remarks
The confined spherically expanding flame approach under constant volume conditions was
employed to investigate end-gas autoignition of dimethyl-ether/oxygen/inert reactive mixtures
using an experimental facility that had previously been utilized for the measurement of laminar
flame speeds at engine-relevant conditions. The dynamic pressure measurement served as the only
experimental observable for analysis, thus, its distinctive features were highlighted during the
compression region through the pressure gradient with low-temperature ignition identified by a
‘bump’ and high-temperature ignition by an exponential rise. For all experimental conditions
investigated, the 1
st
stage ignition occurred approximately 3.6 times the initial pressure.
Unsteady flame propagation, end-gas reactivity upon compression and autoignition were
modeled using a one-dimensional direct numerical simulation code, LTORC, which adequately
captured all the essential aforementioned physics. LTORC results indicated that, the flame induced
1
st
-stage ignition followed directly by 2
nd
stage ignition via compression. Boundary layer growth
was also found to shrink with time, thereby, minimizing the probability of fluid mechanics
interferences with the adiabatic core.
Thermochemical modeling of the end-gas from experimental pressure traces and
calculating a characteristic ignition delay time with sensitivity analysis to kinetics were performed
using a zero-dimensional sensBVP code. It was found that this characteristic ignition delay time
was largely controlled by fuel -specific reactions, specifically, dimethyl-ether low-temperature
chemistry. Additionally, the sensitivity to kinetics increases as the fuel mixture becomes leaner
due to the increasing effect of oxygen in the low-temperature chain branching process.

96

The experimental observables from the dimethyl-ether two-stage ignition process were also
assessed using zero-dimensional numerical simulations. Heat release computations and reaction
path analysis were used to reveal the key low-temperature pathways responsible for low-
temperature ignition. Numerical results strongly suggested that the competition between the
molecular oxygen addition to CH2OCH2O2H and its beta-scission reaction to produce CH2O was
what informed the negative temperature coefficient behavior and its ensuing decreased reactivity.
Furthermore, low temperature exothermicity was primarily dictated by DME low-temperature
chain branching sequence.
The effect of the flame propagation rate on end-gas kinetics for reactive mixtures of DME
was also assessed to understand the significance of the time-history of compression on the buildup
of the radical pool. Experimental results suggested faster burning flames slightly delayed the onset
of both ignition stages and slower burning flames promoted earlier ignition. Numerical results
showed that a similar set of kinetic pathways controlling end-gas autoignition and magnitudes of
key radical species produced in the end-gas, albeit, at a later time, signifying the time-history might
play a role in ignition.
A low-cost methodology was developed to evaluate the effects and trends of fuel additives
in gasoline reactive mixtures through a comparative analysis of their ignition tendency. This was
accomplished experimentally using confined spherically expanding flames by characterizing the
low-temperature ignition features of fuel additives reactive mixtures of the end-gas applicable to
engines and numerically through zero-dimensional numerical simulation predictions with the aid
of detailed kinetic models developed from an automatic mechanism generator.

97

For all the reactive mixtures investigated, those consisting of 2-ethylhexyl-nitrate and
decalin fuel additives exhibited ignition augmenting effects while the others, 1,3,5-
trimethylbenzene and aniline demonstrated ignition inhibition tendencies.
Zero-dimensional numerical simulations were performed to predict the ignition propensity
of reactive mixtures of gasoline surrogate-fuel additive blends using kinetic models automatically
developed with Reaction Mechanism Generator. Ignition delay times were also simulated at
constant pressure (30 atm) and low-temperatures (700–950 K). It was found that, the predictions
of mixtures in air of blended gasoline surrogate and fuel additives followed similar trends to
experimental results, particularly, aniline and 1,3,5-trimethylbenzene retarded ignition by factor
of approximately 2.5 and 3-8, respectively. 2-EHN was also seen to promote ignition at lower
temperatures by a factor of 2.
In conclusion, the current study represents a complementary low-cost approach to obtain
experimental data which can be modeled for reliable kinetic model validation under
thermodynamic conditions relevant to engines. Additionally, it serves to provide an effective and
efficient screening tool when coupled with numerical tools to investigate the mechanistic effects
of fuel additives on end-gas autoignition. Finally, the automatic reaction mechanism generator,
RMG, can be a useful numerical tool to predict qualitative and semi-quantitative trends of fuel
additive effects in reactive mixtures.

9.2 Recommendations for Future Work
Considering the results presented which are encouraging, additional work ought to be done
to complete the entire scope of this proposal. While the advantages of employing this technique
have been explored, i.e., the use of the flame as a piston, reducing boundary layer effects and its

98

simpler incorporation in a one-dimensional model, studying kinetics in a variable pressure
environment which is of direct relevance to knock, it is imperative to compare the Livengood-Wu
approach that utilizes RCM data to predict the conditions of engine knock with this approach.
In order for this approach to be effectively used in capturing the instance of autoignition,
more accurate and qualitative analysis and data over a large thermodynamic space ought to be
obtained to be useful for kinetic model validation. Since this is the first demonstration of an
alternate method to experimentally study ignition, a robust uncertainty quantification ought to be
carried out. The uncertainty in experimental characteristic ignition delay time is directly related to
the sampling frequency, pressure transducer uncertainty and mixture composition strategy (partial
pressure method): these factors are identical to those of legacy experiments (shock tubes and rapid
compression machines).
However, in order to constrain rate coefficients for a chemical kinetic model, the flame
speed will have to be correctly captured, as this is what primarily dictates pre-ignition dP/dt in an
experiment: the high temperature chemistry description of the model will have to be accurate.
Fortunately, the laminar flame speed can be measured using this same approach; a fairly robust
technique which has been effectively developed with this type of data useful to optimize chemical
kinetic models. Once the high temperature chemistry has been established, the low temperature
chemistry can be investigated/verified/optimized using this approach.
With the resolution afforded from this experimental approach, sufficient efforts can be
channeled to investigate gaseous additives that have unique characteristics for natural gas ignition
as ground transportation seem to gravitate towards that domain.

99

References
[1] U.S Energy Information Administration International Energy Outlook 2014 DOE/EIA-
0484 (2014)
[2] C.K. Law, Combustion physics, Cambridge University Press, 2008.
[3] R.K. Hanson, D.F. Davidson, Recent advances in laser absorption and shock tube methods
for studies of combustion chemistry, Prog. Energy Combust. Sci. 44 (2014) 103-114.
[4] C.J. Sung, H.J. Curran, Using rapid compression machines for chemical kinetics studies,
Prog. Energy Combust. Sci. 44 (2014) 1-18.
[5] Y. Zhang, J. Cai, L. Zhao, J. Yang, H. Jin, Z. Cheng, Y. Li, L. Zhang, F. Qi, An
experimental and kinetic modeling study of three butene isomers pyrolysis at low
pressure, Combust. Flame 159 (2012) 905-917.
[6] F.N. Egolfopoulos, N. Hansen, Y. Ju, K. Kohse-Höinghaus, C.K. Law, F. Qi, Advances
and challenges in laminar flame experiments and implications for combustion
chemistry. Prog. Energy Combust. Science, 43 (2014) 36-67.
[7] J.B. Heywood, Internal Combustion Engines Fundamentals, McGraw-Hill, New York
(1988).
[8] N. Fraser, H. Blaxill, G. Lumsden, M. Bassett, Challenges for increased efficiency through
gasoline engine downsizing, SAE Int. J. Engines 2 (2009) 991-1008.
[9] Y. Zhang, Y. Yang, A.L. Boehman, Premixed ignition behavior of C 9 fatty acid esters: A
motored engine study, Combust. Flame 156 (2009) 1202-1213.
[10] S.S. Goldsborough, S. Hochgreb, G. Vanhove, M.S. Wooldridge, H.J. Curran, C.J. Sung,
Advances in rapid compression machine studies of low-and intermediate-temperature
autoignition phenomena, Prog. Energy Combust. Sci. 63 (2017) 1-78.

100

[11] R.L. Belford, R.A. Strehlow, Shock tube technique in chemical kinetics, Annu. Rev. Phys.
Chem. 20 (1969) 247-272.
[12] E.L. Petersen, R.K. Hanson, Measurement of reflected-shock bifurcation over a wide range
of gas composition and pressure, Shock Waves 15 (2006) 333-340.
[13] H. Mark, The interaction of a reflected shock wave with the boundary layer in a shock tube,
NACA Technical Memorandum - 1418 (1958).
[14] M. Honda, K. Takayama, O. Onodera, Y. Kohama, Motion of reflected shock waves in
shock tube, Proc. 10
th
Intern. Shock Tube Symp. (1975) 312-319.
[15] A.B. Mansfield, M.S. Wooldridge, H. Di, X. He, Low-temperature ignition behavior of
iso-octane, Fuel 139 (2015) 79-86.
[16] K.P. Grogan, M. Ihme, Weak and strong ignition of hydrogen/oxygen mixtures in shock-
tube systems, Proc. Combust. Inst. 35 (2015) 2181-2189.
[17] J.W. Meyer, A.K. Oppenheim, On the shock-induced ignition of explosive gases, Symp.
(Int.) Combust. 13 (1971) 1153-1164.
[18] R. Blumenthal, K. Fieweger, K.H. Komp, G. Adomeit, Gas dynamic features of self-
ignition of non-diluted fuel/air mixtures at high pressure, Combust. Sci. Tech. 113 (1996)
137-166.
[19] K.P. Grogan, S.S. Goldsborough, M. Ihme, Ignition regimes in rapid compression
machines, Combust. Flame 162 (2015) 3071-3080.
[20] F. Bozza, D. Siano, M. Costa, Heat transfer, knock modeling and cyclic variability in a
“downsized” spark-ignition turbocharged engine, Adv. App. Math. Mech. 3 (2011) 310-
326.

101

[21] Y.B. Zeldovich, Regime classification of an exothermic reaction with nonuniform initial
conditions, Combust. Flame 39 (1980) 211-214.
[22] X.J. Gu, D.R. Emerson, D. Bradley, Modes of reaction front propagation from hot
spots, Combust. Flame 133 (2003) 63-74.
[23] J.B. Martz, G.A. Lavoie, H.G. Im, R.J. Middleton, A. Babajimopoulos, D.N. Assanis, The
propagation of a laminar reaction front during end-gas auto-ignition, Combust. Flame 159
(2012) 2077-2086.
[24] J.H. Chen, E.R. Hawkes, R. Sankaran, S.D. Mason, H.G. Im, Direct numerical simulation
of ignition front propagation in a constant volume with temperature inhomogeneities: I.
Fundamental analysis and diagnostics, Combust. Flame 145 (2006) 128-144.
[25] J.H. Chen, E.R. Hawkes, R. Sankaran, S.D. Mason, H.G. Im, Direct numerical simulation
of ignition front propagation in a constant volume with temperature inhomogeneities: I.
Parametric study, Combust. Flame 145 (2006) 145-159.
[26] A. Ansari, J. Jayachandran, F.N. Egolfopoulos, Parameters influencing the burning rate of
laminar flames propagating into a reacting mixture, Proc. Combust. Inst. 37 (2019) 1513-
1520.
[27] H. Yu, Z. Chen, End-gas autoignition and detonation development in a closed
chamber, Combust. Flame 162 (2015) 4102-4111.
[28] Z. Chen, M.P. Burke, Y. Ju, On the critical flame radius and minimum ignition energy for
spherical flame initiation, Proc. Combust. Inst. 33 (2011) 1219-1226.
[29] Z. Chen, Effects of radiation and compression on propagating spherical flames of
methane/air mixtures near the lean flammability limit, Combust. Flame 157 (2010) 2267-
2276.

102

[30] P. Dai, Z. Chen, S. Chen, Y. Ju, Numerical experiments on reaction front propagation in
n-heptane/air mixture with temperature gradient, Proc. Combust. Inst. 35 (2015) 3045-
3052.
[31] H. Yu, C. Qi, Z. Chen, Effects of flame propagation speed and chamber size on end-gas
autoignition, Proc. Combust. Inst. 36 (2017) 3533-3541.
[32] J.F. Griffiths, S.K. Scott, Thermokinetic interactions: Fundamentals of spontaneous
ignition and cool flames, Prog. Energy Combust. Sci., 13 (1987) 161-197.
[33] W. Jost, Explosion and combustion processes in gases, McGraw-Hill, New York, 1946.
[34] J.F. Griffiths, Reduced kinetic models and their applications to practical combustion
systems, Prog. Energy Combust. Sci. 21 (1995) 25-107.
[35] J.F. Griffiths, D.J. Rose, M. Schreiber, J. Meyer, K.F. Knoche. Novel features of end-gas
autoignition revealed by computational fluid dynamics, Combust. Flame 91 (1992) 209-
212.
[36] J. Jayachandran, F.N Egolfopoulos, Thermal and Ludwig–Soret diffusion effects on near-
boundary ignition behavior of reacting mixtures, Proc. Combust. Inst. 36 (2017) 1505-
1511.
[37] B. Lewis, G. von Elbe, Determination of the speed of flames and the temperature
distribution in a spherical bomb from time - pressure explosion records, J. Chem. Phys. 2
(1934) 283-290.
[38] C. Xiouris, T. Ye, J. Jayachandran, F.N. Egolfopoulos, Laminar flame speeds under
engine-relevant conditions: uncertainty quantification and minimization in spherically
expanding flame experiments, Combust. Flame 163 (2016) 270-283.

103

[39] Faghih, Z. Chen, The constant-volume propagating spherical flame method for laminar
flame speed measurement, Sci. Bull. 61 (2016) 1296-1310.
[40] A. Omari, L. Tartakovsky, Measurement of the laminar burning velocity using the confined
and unconfined spherical flame methods–A comparative analysis, Combust. Flame 168
(2016) 127-137.
[41] H. Hu, J. Keck, Autoignition of adiabatically compressed combustible gas mixtures, SAE
Technical paper 872110 (1987).
[42] J.C. Keck, H. Hu, Explosions of adiabatically compressed gases in a constant volume
bomb, Symp. (Int.) Combust. 21 (1988) 521-529.
[43] S.M. Sarathy, A. Farooq, G.T. Kalghatgi, Recent progress in gasoline surrogate fuels, Prog.
Energy Combust. Sci. 65 (2018) 67-108.
[44] P. Zhang, N.W. Yee, S.V. Filip, C.E. Hetrick, B. Yang, W.H. Green, Modeling study of
the anti-knock tendency of substituted phenols as additives: an application of the reaction
mechanism generator (RMG), Phys. Chem. Chem. Phys. 20 (2018) 10637-10649.
[45] E.L. Ranzi, A. Frassoldati, R. Grana, A. Cuoci, T. Faravelli, A.P. Kelley, C.K. Law,
Hierarchical and comparative kinetic modeling of laminar flame speeds of hydrocarbon
and oxygenated fuels, Prog. Energy Combust. Sci. 38 (2012) 468-501.
[46] M.J. Pilling, Low-temperature combustion and autoignition. Elsevier, 1997.
[47] H.J. Curran, P. Gaffuri, W.J. Pitz, C.K. Westbrook, A comprehensive modeling study of
n-heptane oxidation, Combust. Flame 114 (1998) 149-177.
[48] A. Movaghar, R. Lawson, F.N. Egolfopoulos, Confined spherically expanding flame
method for measuring laminar flame speeds: Revisiting the assumptions and application to
C1C4 hydrocarbon flames, Combust. Flame, 212 (2020) 79-92.

104

[49] Z. Wang, H. Liu, R.D. Reitz, R.D., 2017. Knocking combustion in spark-ignition
engines. Prog. Energy Combust. Sci. 61 (2017) 78-112
[50] R. J. Kee, J. F. Grcar, M. D. Smooke, J. A. Miller, A FORTRAN Program for Modeling
Steady Laminar One-Dimensional Premixed Flames, Sandia Report.SAND85-8240,
Sandia National Laboratories, 1985.
[51] R. J. Kee, F. M. Rupley, J. A. Miller, Chemkin-II: A Fortran Chemical Kinetics Package
for the Analysis of Gas-Phase Chemical Kinetics, Sandia Report SAND89-8009, Sandia
National Laboratories, 1989.
[52] R. J. Kee, J. Warnatz, J. A. Miller, A Fortran Computer Code Package for the Evaluation
of Gas-Phase Viscosities, Conductivities, and Diffusion Coefficients, Sandia Report
SAND83-8209, Sandia National Laboratories, 1983.
[53] F.N. Egolfopoulos, Geometric and radiation effects on steady and unsteady strained
laminar flames, Symp. (Int.) Combust. 25 (1994) 1375 – 138.
[54] Y. Dong, A.T. Holley, M.G. Andac, F.N. Egolfopoulos, S.G. Davis, P. Middha, H. Wang,
Extinction of premixed H2/air flames: chemical kinetics and molecular diffusion effects,
Combust. Flames 142 (2005) 374-387.
[55] V. Gururajan, F.N. Egolfopoulos, Direct sensitivity analysis for ignition delay times,
Combust. Flame 209 (2019) 478-480
[56] D. Goodwin, H.K. Moffat, R.L. Speth, Cantera: An object- oriented software toolkit for
chemical kinetics, thermodynamics, and transport processes, 2009.
[57] A.C. Hindmarsh, P.N. Brown, K.E. Grant, S.L. Lee, R. Serban, D.E. Shumaker, C.S.
Woodward, SUNDIALS: suite of nonlinear and differential/algebraic equation solvers,
ACM Trans. Math. Software 31 (2005) 363-396.

105

[58] http://poli5.usc.edu:3000/USC/LTORC
[59] D. Bradley, P.H. Gaskell, X.J. Gu, Burning velocities, Markstein lengths, and flame
quenching for spherical methane-air flames: a computational study, Combust. Flame 104
(1996) 176-198.
[60] R.J. Kee, L.R. Petzold, A differential/algebraic equation formulation of the method-of-
lines solution to systems of partial differential equations, In Technical Report SAND86–
8893. Sandia National Laboratories, 1986.
[61] J. Jayachandran, R. Zhao, F.N. Egolfopoulos, Determination of laminar flame speeds using
stagnation and spherically expanding flames: molecular transport and radiation
effects, Combust. Flame, 161 (2014) 2305-2316.
[62] D.B. Spalding, P.L. Stephenson, R.G. Taylor, A calculation procedure for the prediction
of laminar flame speeds, Combust. Flame 17 (1971) 55-64.
[63] G. Dixon-Lewis, I.G. Shepherd, Some aspects of ignition by localized sources, and of
cylindrical and spherical flames, Symp. (Int.) Combust. 15 (1975) 1483-1491.
[64] R. Stauch, S. Lipp, U. Maas, Detailed numerical simulations of the autoignition of single
n-heptane droplets in air, Combust. Flame 145 (2006) 533-542.
[65] S.Y. Cho, R.A. Yetter, F.L. Dryer, A computer model for one-dimensional mass and
energy transport in and around chemically reacting particles, including complex gas-phase
chemistry, multicomponent molecular diffusion, surface evaporation, and heterogeneous
reaction, J. Comput. Phys. 102 (1992) 160-179.
[66] R.D. Richtmyer, K.W. Morton, Difference Methods for Initial-Value Problems, (2
nd
ed.)
Interscience, New York, (1967)

106

[67] R. Lawson, V. Gururajan, A. Movaghar, F.N. Egolfopoulos, Autoignition of reacting
mixtures at engine-relevant conditions using confined spherically expanding
flames, Proc. Combust. Inst. 38 (2020) 2285-2293.
[68] C.W. Gao, J.W. Allen, W.H. Green, R.H. West, Reaction Mechanism Generator:
Automatic Construction of Chemical Kinetic Mechanisms, Comput. Phys. Commun. 203
(2016) 212–225.
[69] R.G. Susnow, A.M. Dean, W.H. Green, P. Peczak, L.J. Broadbelt, Rate-based construction
of kinetic models for complex systems, J. Phys. Chem. A, 101 (1997) 3731-3740
[70] S.W. Benson, J.H. Buss, Additivity rules for the estimation of molecular properties.
thermodynamic properties, J. Chem. Phys. 29 (1958) 546–572.
[71] Z. Zhao, M. Chaos, A. Kazakov, F.L. Dryer, Thermal decomposition reaction and a
comprehensive kinetic model of dimethyl ether, Int. J. Chem. Kinet. 40 (2008) 1-18.
[72] Y. Wu, P. Pal, S. Som, T. Lu, A skeletal chemical kinetic mechanism for gasoline and
gasoline/ethanol blend surrogates for engine CFD applications, 10
th
International
Conference on Chemical Kinetics, Chicago (2017).
[73] M.T. Donovan, X. He, B.T. Zigler, T.R. Palmer, M.S. Wooldridge, A. Atreya,
Demonstration of a free-piston rapid compression facility for the study of high temperature
combustion phenomena, Combust. Flame 137 (2004) 351-365.
[74] A. Moghaddas, C. Bennett, K. Eisazadeh-Far, H. Metghalchi, Measurement of Laminar
Burning Speeds and Determination of Onset of Auto-Ignition of Jet-A/Air and Jet
Propellant-8/Air Mixtures in a Constant Volume Spherical Chamber, J. Energy Resour.
Technol. 134 (2012) 022205.

107

[75] O. Askari, E. Mimmo, M. Ferrari, H. Metghalchi, Auto-ignition characteristics study of
gas-to-liquid fuel at high pressures and low-temperatures, J. Energy Res. Technol. 139
(2017) 012204.
[76] H. Quintens, C. Strozzi, R. Zitoun, M. Bellenoue, Deflagration/Autoignition/Detonation
Transition Induced by Flame Propagation in an N-Decane/O2/Ar Mixture, Flow, Turbul.
Combust. 102 (2019) 735-755.
[77] J. Santner, S.S. Goldsborough, Hot-spot induced mild ignition: Numerical simulation and
scaling analysis, Combust. Flame, 209 (2019) 41-62.
[78] L. Kagan, G. Sivashinsky, Hydrodynamic aspects of end-gas autoignition. Proc. Combust.
Inst. 34 (2013) 857-863.
[79] Q. Fan, Y. Qi, Y. Wang, Z. Wang, 2020. Investigation into pressure dependence of flame
speed for fuels with low and high octane sensitivity through blending ethanol. Combust.
Flame, 212 (2020) 252-269.
[80] G. Jomaas, C.K. Law, J.K. Bechtold, On transition to cellularity in expanding spherical
flames, J. Fluid Mech. 583 (2007) 1-26.
[81] G.H. Markstein, Experimental and theoretical studies of flame-front stability, J. Aeronaut.
Sci. 18 (1951) 199-209.
[82] G. Rozenchan, D.L. Zhu, C.K. Law, S.D. Tse, Outward propagation, burning velocities,
and chemical effects of methane flames up to 60 atm, Proc. Combust. Inst. 29 (2002) 1461-
1470.
[83] A. Rodriguez, O. Frottier, O. Herbinet, R. Fournet, R. Bounaceur, C. Fittschen, F. Battin-
Leclerc, Experimental and modeling investigation of the low-temperature oxidation of
dimethyl ether, J. Phys. Chem. A 119 (2015) 7905-7923.

108

[84] U. Burke, K.P. Somers, P. O’Toole, C.M. Zinner, N. Marquet, G. Bourque, E.L. Petersen,
W.K. Metcalfe, Z. Serinyel, H.J. Curran, An ignition delay and kinetic modeling study of
methane, dimethyl ether, and their mixtures at high pressures, Combust. Flame 162 (2015)
315-330.
[85] Y. Ju, C.B. Reuter, S.H. Won, Numerical simulations of premixed cool flames of dimethyl
ether/oxygen mixtures. Combust. Flame 162 (2015) 3580-3588.
[86] S.S. Merchant, C.F. Goldsmith, A.G. Vandeputte, M.P. Burke, S.J. Klippenstein, W.H.
Green, Understanding low-temperature first-stage ignition delay: Propane, Combust.
Flame 162 (2015) 3658-3673.
[87] V. Gururajan, F.N. Egolfopoulos, Transient plasma effects on the autoignition of
DME/O2/Ar and C3H 8/O2/Ar mixtures, Proc. Combust. Inst. 36 (2017) 4165-4174
[88] D.M. Karwat, S.W. Wagnon, M.S. Wooldridge, C.K. Westbrook, Low-temperature
speciation and chemical kinetic studies of n-heptane, Combust. flame 160 (2013) 2693-
2706.
[89] H. Di, X. He, P. Zhang, Z. Wang, M.S. Wooldridge, C.K. Law, C. Wang, S. Shuai, J.
Wang, Effects of buffer gas composition on low temperature ignition of iso-octane and n-
heptane, Combust. Flame 161 (2014) 2531-2538.
[90] C.L. Barraza-Botet, M.S. Wooldridge, Combustion chemistry of iso-octane/ethanol blends:
Effects on ignition and reaction pathways, Combust. Flame 188 (2018) 324-336.
[91] E.E. Dames, A.S. Rosen, B.W. Weber, C.W. Gao, C.J. Sung, W.H. Green, A detailed
combined experimental and theoretical study on dimethyl ether/propane blended oxidation,
Combust. Flame 168 (2016) 310-330.

109

[92] A.C. Alkidas, Combustion advancements in gasoline engines, Energy Convers. Manag. 48
(2007) 2751-2761.
[93] A.M. Taylor, Science review of internal combustion engines, Energy Policy 36 (2008)
4657-4667.
[94] S.W. Benson, The mechanism of inhibition of knock by lead additives, a chain debranching
reaction, J. Phys. Chem. 92 (1988) 1531-1533.
[95] S. Tanaka, F. Ayala, J.C. Keck, J.B. Heywood, Two-stage ignition in HCCI combustion
and HCCI control by fuels and additives, Combust. Flame 132 (2003) 219-239.
[96] A. Demirbas, M.A. Balubaid, A.M. Basahel, W. Ahmad, M.H. Sheikh, Octane rating of
gasoline and octane booster additives, Petrol. Sci. Technol. 33 (2015) 1190-1197.
[97] M.D. Boot, M. Tian, E.J. Hensen, S.M. Sarathy, Impact of fuel molecular structure on auto-
ignition behavior–Design rules for future high-performance gasolines, Prog. Energy
Combust. Sci. 60 (2017) 1-25.
[98] J.C.G. Andrae, Semidetailed kinetic model for gasoline surrogate fuel interactions with the
ignition enhancer 2-ethylhexyl nitrate, Energy Fuels 29 (2015) 3944-3952.
[99] A.M. Ickes, S.V. Bohac, D.N. Assanis, Effect of 2-ethylhexyl nitrate cetane improver on
NOx emissions from premixed low-temperature diesel combustion, Energy Fuel. 23 (2009)
4943-4948.
[100] A.D. Walsh, The effect of aromatic compounds on the vapour-phase oxidation of fuels.
Part II. The anti-knock effect of aromatic compounds in engines, Trans. Faraday Soc. 45
(1945) 1043-1048.
[101] J.E. Brown, F.X. Markley, H. Shapiro, Mechanism of aromatic amine antiknock action,
Ind. Eng. Chem. 47 (1955) 2141-2146.

110

[102] T.C. Chu, Z.J. Buras, P. Oßwald, M. Liu, M.J. Goldman, W.H. Green, Modeling of
aromatics formation in fuel-rich methane oxy-combustion with an automatically generated
pressure-dependent mechanism, Phys. Chem. Chem. Phys. 21 (2019) 813-832.
[103] L. Lai, S. Gudiyella, M. Liu, W.H. Green, Chemistry of Alkylaromatics Reconsidered,
Energy Fuels 32 (2018) 5489-5500.
[104] A.G. Dana, B. Buesser, S.S. Merchant, W.H. Green, Automated reaction mechanism
generation including nitrogen as a heteroatom, Int. J. Chem. Kinet. 50 (2018) 243-258.
[105] C.A. Class, M. Liu, A.G. Vandeputte, W.H. Green, Automatic mechanism generation for
pyrolysis of di-tert-butyl sulfide, Phys. Chem. Chem. Phys. 18 (2016) 21651-21658.
[106] C.W. Gao, A.G. Vandeputte, N.W. Yee, W.H. Green, R.E. Bonomi, G.R. Magoon, H.W.
Wong, O.O. Oluwole, D.K. Lewis, N.M. Vandewiele, K.M. Van Geem, JP-10 combustion
studied with shock tube experiments and modeled with automatic reaction mechanism
generation, Combust. Flame 162 (2015) 3115-3129.
[107] H. Wang, R. Xu, K. Wang, C.T. Bowman, R.K. Hanson, D.F. Davidson, K. Brezinsky,
F.N. Egolfopoulos, A physics-based approach to modeling real-fuel combustion chemistry-
I. Evidence from experiments, and thermodynamic, chemical kinetic and statistical
considerations, Combust. Flame 193 (2018) 502-519.
[108] R. Xu, K. Wang, S. Banerjee, J. Shao, T. Parise, Y. Zhu, S. Wang, A. Movaghar, D.J. Lee,
R. Zhao, X. Han, A physics-based approach to modeling real-fuel combustion chemistry–
II. Reaction kinetic models of jet and rocket fuels, Combust. Flame 193 (2018) 520-537.
[109] R. Xu, C. Saggese, R. Lawson, A. Movaghar, T.C. Parise, J. Shao, R. Choudhary, J.W.
Park, T. Lu, R.K. Hanson, D.F. Davidson, F.N. Egolfopoulos, A. Aradi, A. Prakash, V.R.R.
Mohan, R. Cracknell, H. Wang, A physics-based approach to modeling real-fuel

111

combustion chemistry - VI. Predictive kinetic models of gasoline fuels, Combust. Flame
220 (2020) 475-487.
[110] A. Violi, S. Yan, E.G. Eddings, A.F. Sarofim, S. Granata, T. Faravelli, E. Ranzi,
Experimental formulation and kinetic model for JP-8 surrogate mixtures, Combust. Sci.
Technol. 174 (2002) 399-417.
[111] F.L. Dryer, Chemical kinetic and combustion characteristics of transportation fuels, Proc.
Combust. Inst. 35 (2015) 117-144.
[112] T. Javed, C. Lee, M. AlAbbad, K. Djebbi, M. Beshir, J. Badra, H. Curran, A. Farooq,
Ignition studies of n-heptane/iso-octane/toluene blends, Combust. Flame 171 (2016) 223-
233.
[113] A.G. Dana, K.A. Spiekermann, W.H. Green, The Tandem Tool (T3) for automated
kinetic model generation and refinement, version 0.1.0,
https://github.com/ReactionMechanismGenerator/T3
[114] A.G. Dana, D. Ranasinghe, H. Wu, C. Grambow, X. Dong, M. Johnson, M. Goldman, M.
Liu, W.H. Green , ARC - Automated Rate Calculator, version 1.1.0,
https://github.com/ReactionMechanismGenerator/ARC
[115] M.P. Burke, M. Chaos, Y. Ju, F.L. Dryer, S.J. Klippenstein, Comprehensive H 2/O2 kinetic
model for high‐pressure combustion, Int. J. Chem. Kinet 44 (2012) 444-474.
[116] P. Glarborg, J.A. Miller, B. Ruscic, S.J. Klippenstein, Modeling nitrogen chemistry in
combustion, Prog. Energy Combust. Sci. 67 (2018) 31-68.
[117] H. Hashemi, J.M. Christensen, S. Gersen, H. Levinsky, S.J. Klippenstein, P. Glarborg,
High-pressure oxidation of methane, Combust. Flame 172 (2016) 349-364.

112

[118] A.Y. Chang, J.W. Bozzelli, A.M. Dean, Kinetic analysis of complex chemical activation
and unimolecular dissociation reactions using QRRK theory and the modified strong
collision approximation, Z. Phys. Chem. 214 (2000) 1533-1568.
[119] J.A. Montgomery, M.J. Frisch, J.W. Ochterski, G.A. Petersson, A complete basis set model
chemistry. VI. Use of density functional geometries and frequencies, J. Chem. Phys. 110
(1999) 2822-2827.
[120] J.A. Montgomery, M.J. Frisch, J.W. Ochterski, G.A. Petersson, A complete basis set model
chemistry. VII. Use of the minimum population localization method, J. Chem. Phys. 112
(2000) 6532-6542.
[121] D.A. McQuarrie, Statistical Mechanics, University Science Books, 2000. C.J. Cramer,
Essentials of Computational Chemistry: Theories and Models, Wiley, New York, 2002.
[122] Z. Chen, M.P. Burke, Y. Ju, 2009. Effects of compression and stretch on the determination
of laminar flame speeds using propagating spherical flames, Combust. Theory Model 13
(2009) 343-364.
[123] D. Bradley, R.A. Head, Engine autoignition: The relationship between octane numbers and
autoignition delay times, Combust. Flame 147 (2006) 171–184.
[124] P.L. Perez, A.L. Boehman, Experimental investigation of the autoignition behavior of
surrogate gasoline fuels in a constant-volume combustion bomb apparatus and its relevance
to HCCI combustion, Energy Fuels 26 (2012) 6106–6117.
[125] T.M. Foong, K.J. Morganti, M.J. Brear, G. da Silva, Y. Yang, F.L. Dryer, The octane
numbers of ethanol blended with gasoline and its surrogates, Fuel 115 (2014) 727-739.
[126] Y. Yang, A.L. Boehman, Oxidation chemistry of cyclic hydrocarbons in a motored engine:
Methylcyclopentane, tetralin, and decalin, Combust. Flame 157 (2010) 495-505.

113

[127] Y. Stein, R.A. Yetter, F.L. Dryer, A. Aradi, The autoignition behavior of surrogate diesel
fuel mixtures and the chemical effects of 2-ethylhexyl nitrate (2-EHN) cetane improver,
SAE Trans. (1999) 1029-1045.
[128] M. Mehl, W.J. Pitz, C.K. Westbrook, H.J. Curran, Kinetic modeling of gasoline surrogate
components and mixtures under engine conditions, Proc. Combust. Inst. 33 (2011) 193-
200.
[129] W.K. Metcalfe, S. Dooley, F.L. Dryer, Comprehensive detailed chemical kinetic modeling
study of toluene oxidation, Energy Fuels 25 (2011) 4915-4936.
[130] H.J. Curran, P. Gaffuri, W.J. Pitz, C.K. Westbrook, A comprehensive modeling study of
n-heptane oxidation, Combust. Flame 114 (1998) 149-177.
[131] H.J. Curran, P. Gaffuri, W.J. Pitz, C.K. Westbrook, A comprehensive modeling study of
iso-octane oxidation, Combust. Flame 129 (2002) 253-280.

Abstract (if available)

Abstract Propagation of a confined spherically expanding flame induces isentropic compression that can culminate in autoignition and/or detonation under conducive thermodynamic conditions. This relatively simple technique measures a distinct ‘characteristic ignition delay time’ to identify the onset of autoignition and complements other established legacy approaches. The present study details this methodology by examining the experimental autoignition characteristics of reactive dimethyl ether (DME)-oxygen-inert-mixtures. Experimental results displayed the classic two-stage ignition typical of dimethyl-ether oxidation at low temperatures with 1st stage ignition occurring approximately 3.6 times the initial pressure. ❧ Direct numerical simulation tools were used to model with results adequately capturing the appropriate physics of unsteady flame propagation, end-gas reactivity upon compression, end-gas autoignition and heat loss to the walls, all in a bid to aid in the interpretation of the experimental results. Furthermore, 0-D thermochemical evolution simulations, sensitivity analysis and heat release calculations using a detailed DME kinetic model were performed to assess the controlling reactions of end-gas autoignition and provide substantial insight into the unsteady flame propagation effects on end-gas kinetics. ❧ Upon demonstration of this approach, a methodology was developed to screen and predict the mechanistic effects of fuel additives/components on the ignition propensity of reactive gasoline-additive (2-ethylhexyl nitrate, aniline, decalin, 1,3,5-trimethylbenzene) mixtures using this technique. Distinctive features from the two-stage ignition process end-gas autoignition for these reactive gasoline-additive mixtures were characterized in order to gain relevant information, applicable to engine testing for screening purposes. ❧ The role of additives on end-gas autoignition at low-temperatures was probed by developing detailed chemical kinetic models of gasoline surrogateㅡfuel additives using an open-source automatic reaction mechanism generator software package for qualitative predictions through ignition delay time simulations. Specifically, aniline, a known anti-knock gasoline additive was found to effectively suppress the 2nd stage ignition by a factor of 3. Simulation predictions closely followed experimental trends and highlighted the versatility of using this technique in gleaning information towards engine applications while demonstrating the immense capabilities of Reaction Mechanism Generator to explore the qualitative performance and chemical kinetics of fuel additives on autoignition.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Experimental studies of high pressure combustion using spherically expanding flames

PDF

Modeling investigations of fundamental combustion phenomena in low-dimensional configurations

PDF

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

PDF

Determination of laminar flame speeds under engine relevant conditions

PDF

Pressure effects on C₁-C₂ hydrocarbon laminar flames

PDF

Accuracy and feasibility of combustion studies under engine relevant conditions

PDF

Studies of methane counterflow flames at low pressures

PDF

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

PDF

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

PDF

Investigations of fuel effects on turbulent premixed jet flames

PDF

Experimental investigation of the propagation and extinction of edge-flames

PDF

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

PDF

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

PDF

Direct numerical simulation of mixing and combustion under canonical shock turbulence interaction

PDF

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

PDF

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

PDF

Studies of siloxane decomposition in biomethane combustion

PDF

Aerodynamics at low Re: separation, reattachment, and control

PDF

Development of a novel heterogeneous flow reactor: soot formation and nanoparticle catalysis

PDF

The use of carbon molecule sieve and Pd membranes for conventional and reactive applications

Asset Metadata

Creator Lawson, Robert (author)

Core Title End-gas autoignition investigations using confined spherically expanding flames

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Aerospace Engineering

Degree Conferral Date 2021-08

Publication Date 07/18/2021

Defense Date 04/05/2021

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

Tag end-gas autoignition,laminar flames,low-temperature chemistry,OAI-PMH Harvest,spherically expanding flames

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Egolfopoulos, Fokion (committee chair), Bermejo-Moreno, Ivan (committee member), Jessen, Kristen (committee member), Ronney, Paul (committee member)

Creator Email rlawson@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC15602409

Unique identifier UC15602409

Legacy Identifier etd-LawsonRobe-9763

Document Type Dissertation

Format application/pdf (imt)

Rights Lawson, Robert

Type texts

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

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu