ABSTRACT
Title of dissertation: EXPERIMENTAL INVESTIGATIONS
AND SCALING ANALYSES OF
WHIRLING FLAMES
Sriram Bharath Hariharan
Doctor of Philosophy
2020
Dissertation directed by: Professor Michael J. Gollner
Department of Fire Protection Engineering
Professor Elaine S. Oran
Department of Aerospace Engineering
Swirling flows are ubiquitous in nature, occurring over a large range of length
scales ? on the order of many tens-of-thousands of kilometers in the case of Saturn?s
hexagonal polar vortex, to just a few centimeters in dandelion flight. Most instances
of swirling flow involve momenta competing in two different directions, axial and
azimuthal. Whirling flames (also known as fire whirls) occur at the intersection of
vortical flow fields and buoyant, reactive plumes, and they represent a general class
of flows that may be considered slender vortices involving axial momentum from
heat-release and tangential momentum from air entrainment.
In this work, two previously unexplored characteristics of whirling flames are
considered over a wide range of scales, spanning three orders of magnitude in length
and four orders in heat-release rate. First, emissions of particulate matter (PM)
from fire whirls (FW) were measured and compared to those from free-buoyant pool
fires (PF). For different pool diameters and fuels, FWs showed higher burning rate
and fuel-consumption efficiency, but lower PM-emission rate, leading to lower PM-
emission factors. The lower PM emissions from FWs is attributed to a feedback cycle
between higher oxygen consumption from improved entrainment, higher average
temperatures, increased heat feedback to the fuel pool, which in turn increases
burning rate and entrainment. A scaling analysis showed that the PM emission
factor decreased linearly with the ratio of inverse Rossby number to nondimensional
heat-release rate.
Second, the structure of the blue whirl (BW), a soot-free regime, was investi-
gated using dimensional analysis and non-intrusive optical diagnostics. Experimen-
tal data of heat-release rates and circulation for BWs and FWs from the literature
were used to define the nondimensional equivalents of buoyant and azimuthal mo-
menta. The combinations of these parameters showed that FWs primarily formed
in a buoyancy-dominated regime, and that a circulation-dominated regime was re-
quired for BW formation, corroborating hypotheses that the transition was caused
by the bubble mode of vortex breakdown, resulting in the formation of a recircula-
tion zone. Finally, OH- and PAH-PLIF, OH* and CH* chemiluminescence suggest
a triple-flame structure anchored at the blue ring region of the BW, with the rich
branch formed by the lower blue cone, and the lean branch by the upper purple
haze. These results show that the mixing process occurs upstream of the conical
region and that the recirculation zone is comprised of combustion products.
EXPERIMENTAL INVESTIGATIONS AND
SCALING ANALYSES OF WHIRLING FLAMES
by
Sriram Bharath Hariharan
Dissertation submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
2020
Advisory Committee:
Professor Michael J. Gollner, Chair & Advisor
Professor Elaine S. Oran, Co-Chair & Co-Advisor
Professor Arnaud Trouve?
Professor Peter B. Sunderland
Professor James H. Duncan
Professor Christopher Cadou
Professor Konstantina Trivisa, Dean?s Representative
?c Copyright by
Sriram Bharath Hariharan
2020

Acknowledgments
This dissertation was possible primarily because of the people who have sup-
ported me during my time in graduate school at the University of Maryland. First
and foremost, encouragement from my advisors, Professor Michael J. Gollner and
Professor Elaine S. Oran, was the primary reason I chose to pursue doctoral studies.
They have stood by me through the ups and downs as a graduate student over the
past five years. Michael showed me how to approach a problem meticulously and
provided ample room for me to experiment with my own ideas. His rigorous work
ethic and enthusiasm for new problems will serve as a personal goal for my career.
I cannot emphasize enough Elaine?s influence on my development as a researcher.
Her encouragement and insights have been crucial to broadening the scope of this
work asking the right questions at every stage.
Sincere thanks to Professors Arnaud Trouve?, Peter B. Sunderland, James H.
Duncan, Christopher Cadou, and Konstantina Trivisa for serving on the doctoral
committee. They were very supportive during early reviews of my research and
provided valuable comments to improve the dissertation.
I express my gratitude to my colleagues, Dr. Paul Anderson, who acted as an
unofficial mentor, and Dr. Yu Hu, whose experiments on the blue whirl served as a
basis for a part of this work.
I would like to acknowledge the Bureau of Safety and Environmental Enforce-
ment (BSEE), U.S. Department of Interior, Washington, D.C. for funding this work
under contract No. E17PC00016. Special thanks to Karen Stone, Program Manager,
ii
for her guidance and support throughout the duration of the project. This work was
also supported by the National Science Foundation (NSF, CBET-1507623,1554026).
A number of large-scale experiments were conducted at the Fire Protection
Engineering Performance Laboratory at the Worcester Polytechnic Institute, thanks
to the efforts of Dr. Ali Rangwala and Raymond T. Ranellone.
I would like to thank Drs. Waruna D. Kulatilaka and Yejun Wang at the the
Optical Diagnostics and Imaging Laboratory, Texas A&M University, for organizing
the optical diagnostics experiments of the blue whirl. Special thanks to Kelsie
Thompson, Ryan McDaniel, Department of Aerospace Engineering, for facilitating
my visits to College Station, TX.
Much gratitude to Dr. Hamed Farahani, Joseph Dowling, Michael R. Jones,
Xingyu Ren and Dr. Ali Tohidi for their important contributions to experiments
and analysis in this work. I am grateful to all my fellow lab mates in the De-
partment of Fire Protection Engineering (FPE), University of Maryland ? Evan
Sluder, Dr. Zhengyang Wang, Dr. Dennis Kim, Dr. Joseph Chung, Dr. Xiao
Zhang, Nicholas Chui, Alison Davis, Priya Garg, Lana Benny, Raquel Hakes, Mike
Heck and Dushyant Chaudhari. I convey my thanks to the staff at FPE ? Brian
Sully, Olivia Noble, Olga V. Zeller, Mary Lou Holt, Nicole Hollywood and Christine
O?Brien for making administrative formalities easy to navigate.
The past few years have presented many opportunities for collaborative work,
including microgravity experiments at the NASA Glenn Research Center. I thank
Dr. Paul Ferkul, Dr. Sandra Olson, Vittorio M. Valletta, Luke Ogorzaly and
Eric Neumann for arranging them. I also thank the Department of Mechanical
iii
Engineering, University of California, Berkeley for hosting me in the last stage of
graduate education.
Finally, my sincere and heartfelt thanks to my family and friends. Their
constant and unseen support during this time cannot be overstated.
College Park, Maryland
July 2020
iv
Table of Contents
Acknowledgements ii
Table of Contents v
List of Tables viii
List of Figures x
Nomenclature xviii
1 Introduction 1
1.1 Swirling Flows and Whirling Flames . . . . . . . . . . . . . . . . . . 1
1.2 The Blue Whirl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Motivation and Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Outline of the Dissertation . . . . . . . . . . . . . . . . . . . . . . . . 11
2 Review of Fire Whirls and Experimental Methods 12
2.1 Swirl and Whirl Combustion . . . . . . . . . . . . . . . . . . . . . . . 12
2.2 Fire Whirls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.1 Scaling Approach and Influential Parameters . . . . . . . . . 23
2.3 Pool Fires, Emissions and Marine Oil Spills . . . . . . . . . . . . . . 28
2.4 Soot Formation and Emission . . . . . . . . . . . . . . . . . . . . . . 32
2.4.1 Soot Diagnostics . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.5 Current Understanding of the Blue Whirl . . . . . . . . . . . . . . . . 39
2.6 Chemiluminescence and Planar Laser-Induced Fluorescence . . . . . . 43
3 Measurement of Particulate-Matter Emissions from Liquid-Fueled Fires 46
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.1.1 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.2 Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.2.1 General Apparatus . . . . . . . . . . . . . . . . . . . . . . . . 50
3.2.2 Exhaust Duct and Emissions Sampling . . . . . . . . . . . . . 54
3.2.3 Temperature Measurements . . . . . . . . . . . . . . . . . . . 59
v
3.2.3.1 Gas-Phase Temperature . . . . . . . . . . . . . . . . 59
3.2.3.2 Liquid-Phase Temperature . . . . . . . . . . . . . . . 60
3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.3.1 Physical Characteristics of PFs and FWs . . . . . . . . . . . . 62
3.3.2 Burning Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.3.3 Emissions: Particulate matter . . . . . . . . . . . . . . . . . . 69
3.3.4 Emissions: Gaseous species . . . . . . . . . . . . . . . . . . . . 74
3.3.5 Heat Feedback and Temperature . . . . . . . . . . . . . . . . 79
3.3.6 Air-Entrainment Velocity . . . . . . . . . . . . . . . . . . . . . 86
3.4 Discussion of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.6 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
4 Effect of Circulation and Heat-Release Rate on Particulate-Matter Emissions 96
4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
4.2.1 Emission Factors . . . . . . . . . . . . . . . . . . . . . . . . . 98
4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
4.3.1 Effects of D on Fire Behavior . . . . . . . . . . . . . . . . . . 100
4.3.2 Circulation and Heat Feedback . . . . . . . . . . . . . . . . . 102
4.3.3 PM Emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
4.4.1 Scaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.4.2 Boilover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4.4.3 Factors influencing PM emissions . . . . . . . . . . . . . . . . 120
4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
4.6 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
5 Quantifying the Effects of Circulation and Buoyancy on the Transition from
a Fire Whirl to a Blue Whirl 127
5.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
5.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.2.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
5.3 Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.4 Flame Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
5.5 Dimensional Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
5.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
5.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
5.8 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
6 Understanding the Blue Whirl using Optical Diagnostics 159
6.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
6.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
6.2.1 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
6.3 Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . . 165
vi
6.3.1 Apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
6.3.2 PLIF experiments . . . . . . . . . . . . . . . . . . . . . . . . . 167
6.3.3 Chemiluminescence experiments . . . . . . . . . . . . . . . . . 168
6.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
6.4.1 PLIF: OH and PAH . . . . . . . . . . . . . . . . . . . . . . . 172
6.4.2 Chemiluminescence . . . . . . . . . . . . . . . . . . . . . . . . 175
6.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
6.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
6.7 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
7 Conclusions and Future Directions 190
7.1 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
7.2 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
A Effect of Entrainment Conditions on PM Emissions from Fire Whirls 203
A.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
A.2 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
A.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 206
A.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
B Supplementary Material to Chapters 3 and 4 213
C Supplementary Material to Chapter 6 230
C.1 Chemiluminescence data of premixed methane-air flames . . . . . . . 230
C.2 Chemiluminescence and PLIF: BW . . . . . . . . . . . . . . . . . . . 234
C.3 Flow in the post-flame region of the blue whirl . . . . . . . . . . . . . 238
Bibliography 241
vii
List of Tables
3.1 Experimental conditions for pool fires and fire whirls formed using
heptane and ANS crude oil. . . . . . . . . . . . . . . . . . . . . . . . 52
3.2 Physical properties of the fuels discussed in this work. Data obtained
from Material Safety Data Sheet (MSDS) for each fuel. Crude oil
properties are a strong function of the level of evaporation and com-
position, resulting in significant variation in the reported properties. . 53
3.3 Tabulated data of average values of burning duration, burning rate,
and estimations for flame width (wf ) and flame height (Hf ) for the
different flame regimes and pool diameters. Experimental variability
for m? and residue mass are presented in Appendix B. . . . . . . . . . 68
4.1 Tabulated values of EF for the different species from PFs and FWs
in this study. For each parameter, the mean of three experiments is
shown for both fuel-mass based and fuel carbon-mass based estima-
tion methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
5.1 Experimental data and nondimensional quantities for the BW, TBW,
FW and LFW regimes. . . . . . . . . . . . . . . . . . . . . . . . . . . 145
5.2 Values of W? = (??f / Q??f ) for the BW, TBW, FW and LFW regimes. 152
6.1 Coefficients for the third-order polynomial functions relating I?Oct,cal
and ?, shown for the three methods used to normalize I. . . . . . . . 180
A.1 Experimental conditions for the different air entrainment conditions
for fire whirls formed using diesel fuel. . . . . . . . . . . . . . . . . . 205
B.1 Experimental variability in the estimation of ?fuel for ANS fires at
D = 70 cm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
B.2 Experimental variability in the estimation of ?fuel for ANS fires at
D = [10, 20, 30] cm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
viii
C.1 Consolidated velocity data (in [mm/s]) obtained by tracking soot in
three individual experiments. . . . . . . . . . . . . . . . . . . . . . . 240
ix
List of Figures
1.1 Examples of swirling flow in nature at different length scales. (Left)
Polar vortex within the hexagonal jetstream of Saturn, from [1].
(Right) Separated vortex structure in dandelion flight, from Cum-
mins et al. [2]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 An image of the blue whirl formed with heptane over a water surface.
The lower portion of the image is the reflection of the flame over the
water surface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1 Radial variation of azimuthal velocity in (A) a swirl combustor [3],
and (B) a fire whirl [4]. . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Laboratory configurations used to form fire whirls. Figure from To-
hidi et al. [5]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 Photograph of a fire whirl formed in a fixed-frame setup with a com-
bination of natural and forced air entrainment. The pool diameter
is 30 cm, placed in an enclosure of dimensions 3.6?3.6?1.5 m3. To
increase entrainment velocity through the four inlets, a suction sys-
tem was employed above the fire. In configurations with natural
entrainment, buoyancy and air entrainment are coupled. Buoyancy
within the enclosure must be capable of drawing sufficient air into
the enclosure. This minimum level of circulation required for fire-
whirl formation depends on many factors including pool diameter [6],
and small pools within large enclosures do not form strong fire whirls
without externally-forced entrainment. . . . . . . . . . . . . . . . . . 17
2.4 (A) Radial variation of temperature in a fire whirl, from [7]. (B)
Burning rate of acetone as a function of ambient circulation, from [8].
The difference in burning rate between the cases with and without the
flame holder may be attributed the level of disruption in the Ekman
layer by the vertical lip formed by the flame holder. . . . . . . . . . . 20
2.5 A graphical representation of the velocity field around a stationary
Type-I fire whirl. Figure adapted from [5]. . . . . . . . . . . . . . . . 22
x
2.6 Influence of Pe on nondimensional height of fire whirls under high
ambient circulation, (A) shown experimentally in [9] using a tilting
fire whirl apparatus, and (B) developed analytically in [10]. . . . . . . 25
2.7 (A) Dependence of Ucr on Frf for buoyancy-dominated fire whirls,
adapted from [11]. (B) Variation of nondimensional fire whirl height
with nondimensional circulation, adapted from [12]. . . . . . . . . . . 26
2.8 (A) Variation of flame height with Pe. The prediction of flame height
based on Pe was improved by using a modified axial velocity profile.
Figure adapted from [13]. (B) Dependence of Fr on nondimensional
circulation, adapted from [12]. . . . . . . . . . . . . . . . . . . . . . . 26
2.9 Photograph of a pool fire formed with n-heptane, D = 30 cm. . . . . 29
2.10 (A) Dependence of pool-fire puffing frequency on pool diameter, adapted
from [14]. (B) Influence of Fr on puffing St, adapted from [15]. . . . . 29
2.11 Comparison of PM-2.5 emission factor and Modified Combustion Effi-
ciency (MCE) for JP-5 and propane pool fires. Figure adapted from [16]. 31
2.12 Schematic showing soot formation and destruction zones in a laminar
jet diffusion flame. Figure from Turns [17]. . . . . . . . . . . . . . . . 34
2.13 Figure showing the evolution of soot structure in different regions of
a flame. Figure from [18]. . . . . . . . . . . . . . . . . . . . . . . . . 35
2.14 Schematic showing the different stages of soot inception and forma-
tion, beginning from PAHs to particle growth. Figure from [19]. . . . 36
2.15 Soot measurement techniques, adapted from [20]. . . . . . . . . . . . 37
2.16 Schematic of soot measurement in the TSI DustTrak, adapted from [21]. 38
2.17 (A) Image of a blue whirl formed using iso-octane over a flat metallic
plate. (B) The recirculation zone within the blue whirl is visualized
by incandescence of soot particles entrapped in the recirculation zone.
This image was captured with an exposure time of 1/25 s, allowing
sufficient time for steaks to develop in the axial direction. Image from
Hariharan et al. [22]. . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.18 Sequence of images showing the region immediately above the fuel
pool during different intermediary states before formation of the blue
whirl. Figure from Coenen et al. [23]. . . . . . . . . . . . . . . . . . . 40
2.19 (A) Thermocouple measurements in the blue whirl, showing peak
temperatures around 2000 K. (B) Temperature map obtained by thin-
filament pyroemtry, for a blue whirl formed using iso-octane formed
over a water surface. Figures adapted from [22]. . . . . . . . . . . . . 41
2.20 (A) The effect of gap size, S, and fuel flow rate, V? , on the fire whirl
regime formed. (B) Relationship between nondimensional circulation
and heat-release rate showing transition limits to the blue whirl. The
subscript ?D? refers to the reference length scale, which was chosen
as the enclosure diameter for nondimensionalization. Figures from [24]. 42
2.21 Schematic showing an experimental setup to perform PLIF. Figure
from Hanson et al. [25]. . . . . . . . . . . . . . . . . . . . . . . . . . . 44
xi
3.1 Schematics of the experimental configuration used to form FWs at
various scales. The ratio of gap width (W ) to enclosure side length
(S) was maintained at W/S = 1/4 for experiments at all scales.
Schematic in panel (B) from Tohidi et al. [5]. . . . . . . . . . . . . . . 51
3.2 Schematic showing suction hood and exhaust duct positioned above
the fire whirl apparatus. Flow rate in the duct was determined using
temperature measurements and a differential pressure sensor. Two
sampling tubes were used, one for gaseous species and one for partic-
ulate matter. Inlet velocity into the fire whirl enclosure was measured
using either a hot-wire or vane anemometer. Mass of the fuel dish
was measured using a Mettler-Toledo load cell. . . . . . . . . . . . . . 55
3.3 Exhaust-gas sampling tubes, pressure and temperature sensors posi-
tioned inside the exhaust duct. Exhaust flow direction is towards the
reader. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.4 Images of (A) CAI ZPA NDIR/O2 Analyzer [26], and (B) DustTrak
DRX Aerosol Monitor 8534 [27]. . . . . . . . . . . . . . . . . . . . . . 57
3.5 (A) Schematic showing locations of TCs above the fuel pool, used
for experiments with D = [10, 20, 30] cm. (B) Schematic showing
locations of TCs for gas- and liquid-phase temperature measurements
for experiments with D = 70 cm. . . . . . . . . . . . . . . . . . . . . 60
3.6 TCs for liquid-phase temperature measurements for D = 10, 20, 30
cm. (A) TCs to check for boiling at the fuel-water interface, without
the stirrer. (B) TCs to check for effect of the stirrer on temperature
stratification in the water sublayer. . . . . . . . . . . . . . . . . . . . 61
3.7 Photographs of PFs (panels A?D) and FWs (panels E?H) formed
using heptane at D = [10, 20, 30, 70] cm. . . . . . . . . . . . . . . . . 63
3.8 Photographs of PFs (panels A?C) and FWs (panels D?F ) formed
using ANS crude oil at D = [10, 20, 70] cm. . . . . . . . . . . . . . . . 64
3.9 Mass loss behavior of PFs and FWs formed at D = 10 cm. Data
from individual experiments using (A) heptane and (C) ANS crude
oil. The marker colors denote individual experiments. Averaged data
and linear fits are shown in (B) and (D). The slope of the linear fits
represents the average burning rate. . . . . . . . . . . . . . . . . . . . 66
3.10 Burning rate for PFs and FWs formed with D = [20, 30] cm. The
red and black markers in each panel represent the average of three
experiments. Overall burning rate is shown in panel (C) for heptane
fires and in panel (F) for ANS fires. compared for all medium-scale
heptane fires. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.11 Variation of PM-emission rate with time at D = 10 cm. The spikes
in (C) and (D) are instances of boilover towards the end of the burn. 70
3.12 Variation of PM-emission rate with time at D = 20 cm for PFs and
FWs formed using heptane (A, B) and ANS crude oil (C, D). The
different colors represent data from individual experiments. . . . . . . 71
3.13 Variation of PM emissions rate with time at D = 30 cm for PFs and
FWs formed using heptane (A, B) and ANS crude oil (C, D). . . . . . 72
xii
3.14 PM emissions rate from PFs and FWs formed using a 5 mm slick of
ANS crude oil at D = 70 cm. . . . . . . . . . . . . . . . . . . . . . . 73
3.15 Variation of O2 consumption rate with time for the different fires at
D = 70 cm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.16 Variation of heat-release rate (HRR) with time for the different fires
at D = 70 cm. HRR was estimated by O2-consumption calorimetry,
assuming a value of 13.1 kJ/g-of-O2 [28, 29]. . . . . . . . . . . . . . . 76
3.17 Variation of CO2 emission rate with time for the different fires at the
70 cm scale. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.18 Variation of CO emission rate with time for different fires at the 70
cm scale. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.19 Effect of using the magnetic stirrer on temperature in the water sub-
layer, (A) with the stirrer showing a decrease in temperature with
depth, and (B) without the stirrer, showing minimal stratification
and a single bulk temperature for the sublayer. . . . . . . . . . . . . . 80
3.20 Temporally and spatially averaged q??? for PFs and FWs at D =
[10, 20, 30] cm, formed using (A) heptane and (B) ANS crude oil. . . . 81
3.21 Variation of q??? to the center of a 70 cm pool for PFs and FWs formed
using (A) heptane and (B) ANS crude oil. . . . . . . . . . . . . . . . 81
3.22 Flame temperatures at different axial locations (H) above the fuel
surface, for PFs and FWs formed using heptane and crude oil, at D
= 10 cm. Here, data from a single experiment is shown so that there
is no smearing of trends due to averaging. . . . . . . . . . . . . . . . 83
3.23 Temperature measurements for fires formed at D = 20 cm. . . . . . . 84
3.24 Temperature measurements for fires formed at D = 30 cm. . . . . . . 85
3.25 Measurements of U? and estimates of ? for FWs formed using heptane
and ANS at different D. . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.26 EFs and CF (fuel-mass based) for heptane and ANS crude oil fires
at different D. Panels (A?D) show results for heptane, and panels
(E?H) for ANS crude oil fires. The black square markers denote
results for PFs and the red circles denote those for FWs. . . . . . . . 89
3.27 (A) Combination of peak Q?, and ? for the different FWs in this study.
(B) Variation of q? with D. . . . . . . . . . . . . . . . . . . . . . . . . 92
3.28 Burning rate and flame height for pool fires formed using various
liquid fuels. As radiative feedback begins to dominate in large fires,
burning rate tends to flatten with pool diameter. Figure adapted
from [30]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.1 Comparison of m? for PFs and FWs formed using (A) heptane and
(B) ANS crude oil. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.2 Variation of (C) m??? for heptane and ANS fires, and (D) ?fuel with D
for ANS fires. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.3 Combinations of Q? and ? for the FWs in this study, showing a loga-
rithmic relationship for both heptane and ANS FWs. . . . . . . . . . 103
xiii
4.4 Variation of (A) ? and Ro?1 with D, (B) q???f with D. Panels (C) and
(D) show the linear increase in q???f and q
?
f with ? for FWs. Solid lines
indicate linear fits for heptane FWs, and dashed lines indicate those
for ANS FWs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.5 PM emission rate, measured at 1 Hz, as a function of time for 70 cm
heptane PFs and FWs are shown in panels (A) and (B). Data for 70
cm ANS PFs and FWs are shown in panels (C) and (D). Boilover is
evident in the PM emission rate for both ANS PFs and FWs. . . . . 107
4.6 EFPM for the different values of D is shown for fires formed using
heptane and ANS crude oil in panels (A) and (B). Panel (B) shows
markers for overall EFPM, and those calculated for the pre- and post-
boilover periods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.7 Variation of different EFs with D. EFUHC is shown in panels (A)
and (B), EFCO2 in panels (C) and (D), and EFCO in panels (E) and
(F). Data for 70 cm ANS fires also shows markers for the pre- and
post-boilover periods. . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
4.8 Grashof number (Gr) at different values of D. . . . . . . . . . . . . . 111
4.9 (A) Variation of Q?? with D. (B) Variation of EF ?1 ?PM with (Ro /Q? ).
Burning efficiency for (C) heptane and (D) ANS fires. Oxidation
factor for (E) heptane and (F) ANS fires. . . . . . . . . . . . . . . . . 114
4.10 Variation of the value of (Ro?1/Q??) with D for heptane and ANS
FWs. The ratio un . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4.11 Images of a PF under (A) normal and (B) boilover burning, formed
using a 5 mm slick of ANS crude oil. The streaks seen in (B) are
formed by the fuel droplets emanating from the fuel pool. Variation
of carbon-emission rate with time (D = 70 cm) for (C) heptane PF
and (D) ANS PF. (D) Bar graphs showing the difference between PFs
and FWs (D = 70 cm) on the basis of time for onset of boilover, fuel
mass consumed before boilover onset, and fuel mass consumed during
boilover. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.1 Images of fire whirl regimes formed at a gap size of S = 35 mm. The
experimental configuration is shown in Figure 5.4. The fuel burning
rate, V? , varies from (A) to (I) as 0.5, 1.1, 1.5, 2, 3, 4.5, 6, 8 and
10 ml/min. Panels (A)?(B) show BWs, (C)?(E) show TBWs, and
(F)?(I) show FWs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
5.2 Images of the blue whirl where the RZ is visualized by streaks of
incandescent soot particles. Images (A) and (B) were captured with
exposure times of 1/60 s and 1/100 s, respectively. . . . . . . . . . . . 132
5.3 Influence of S and V? on the fire whirl regime, adapted from [24]. . . . 136
5.4 Schematic of experimental apparatus. . . . . . . . . . . . . . . . . . . 138
5.5 Variation of wf (or bw) with ? and Q?. Only a very minimal increase
in wf is seen with Q?. . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
xiv
5.6 Variation of H with ? and Q?. The hashed section in panels B and
D indicates the limited variation in H for the BW. In panel B, flame
regimes at S = 15 mm are not considered for the curve-fit shown and
follow a different trend. This difference at low S? is discussed in [24]. 141
5.7 Variation of H with Q?, shown for cases without (A) and with (B)
the added inlet channels for the BW. The vertical dashed lines, green
(left) and magenta (right), indicate limits of extinction and transition,
respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
5.8 (Top) Variation of H? with Q?? . The range of H?f for the BW and FW
do not overlap, showing that it varies only when the regime changes.
The colored stripes denote the range of H? for each regime. S? does
not influence H?, particularly in the BW regime. (Bottom) Variation
of ??f with H
?, with the different transition limits shown as dotted
lines. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
5.9 ??f as a function of Q?
?
f for the different regimes in this study. The blue
markers correspond to BWs, magenta to TBWs, yellow to FWs, and
black to LFWs. The dotted blue line represents ??f = Q?
?
f , roughly
the region where the FW?BW occurs. Quantities for LFWs were
calculated from raw data in Table 1 of [7], whose experiments used
a propane burner within a square cross-section setup, with natural
air entrainment at S? = 0.111 (calculated based on the definition in
this study) and Q? ? [25, 300] kW. Extinction occurs in the region
where ?? ?f > 4Q?f , and FW?LFW transition occurs when Q??f > 3??f ,
represented by the red line. . . . . . . . . . . . . . . . . . . . . . . . 149
5.10 Comparison of experimental data in this work with those for LFWs
in the literature [7]. (Top) The LFW regime occurs when ??f ? (0, 1).
The value of H? for LFWs in the literature are many times that of
the FWs in this study, and an order magnitude higher than the BWs.
(Bottom) The FW-to-BW transition occurs around W? ? 1, in the
circulation-dominated regime. . . . . . . . . . . . . . . . . . . . . . . 153
5.11 Proposed map of some whirling flames regimes as functions of nondi-
mensional circulation and heat-release rate, as defined in this chapter. 157
6.1 (A) Schematic of a fixed-frame fire whirl apparatus formed using two
half-cylinders. These were positioned slightly offset to create gaps to
allow natural entrainment of air into the enclosure. (B) Schematic of
the imaging setups for chemiluminescence intensity and PLIF mea-
surements. These techniques were performed separately for each
species of interest. Figure adapted from [31]. . . . . . . . . . . . . . . 167
6.2 Contour plots of 200-frame averages of (A) OH* and (B) CH*, in
methane-air flames at ? = 1 stabilized over a 3.81 cm Hencken burner.171
xv
6.3 Contour maps of ground-state OH in a radial plane of the (A) BW
formed with iso-octane, and (B) FW formed with methanol. (C) PAH
signals in the BW formed using a mixture of iso-octane and biacetyl.
(D) Background signals of PAH after extinction of the BW,caused by
scattering of evaporating droplets of fuel-dopant mixture. Additional
figures for each species are presented in Appendix C. . . . . . . . . . 173
6.4 Sequence of chemiluminescence images obtained at 1000 Hz, shown
in false color. Panels (i?v) show CH*, and panels (a?e) show OH*.
Processed images with intensity values are shown in Figures C.5 and
C.6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
6.5 Contour map of (CH?/OH?) in the BW. A radial location of r = 0
signifies a cross sectional plane through the center of the BW. . . . . 177
6.6 Effect of the different averaging methods on the OH* and CH* inten-
sity counts. The 7-pixel average was chosen for determining I?Meth,cal
and thus, I?Oct,cal. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
6.7 (Left) Variation of I?ref with ?, obtained using system responses for
methane and iso-octane in [32]. (Right) Calibration curve showing
the overall relationship between I?Oct,cal and ? ? [0.8, 1.3] for the three
different normalization methods. . . . . . . . . . . . . . . . . . . . . . 179
6.8 Contours of flame index (indicating level of premixing), equivalence
ratio and temperature heat-release rate in the blue whirl, obtained
using numerical simulations. Figure adapted from [33]. . . . . . . . . 184
6.9 A simplified schematic of the flame structure of the blue whirl, based
on measurements of OH, PAH, OH* and CH*. . . . . . . . . . . . . . 185
7.1 Variation of Aratio with D. Lineat fit shown has a R
2 value of 0.96. . 195
A.1 Dependence of PM emission rate on gap size for the naturally en-
trainment cases (top), and fan speed on the forced entrainment cases
(bottom) for fire whirls formed over a 70 cm diesel pool. . . . . . . . . 207
A.2 PM emission factors for fire whirls formed over a 70 cm diesel pool.
Dependence on gap size for naturally entrainment cases (top), and
dependence on fan speed for forced entrainment cases (bottom). . . . 208
A.3 Influence of equivalence ratio and in-cylinder temperature on soot
and NO emissions from an internal combustion reciprocating engine.
Figure from [34]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
B.1 O2 consumption, D = 10 cm. . . . . . . . . . . . . . . . . . . . . . . 215
B.2 O2 consumption, D = 20 cm. . . . . . . . . . . . . . . . . . . . . . . 216
B.3 O2 consumption, D = 30 cm. Panel B shows results from six different
experiments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
B.4 CO2 emission rate, D = 10 cm. . . . . . . . . . . . . . . . . . . . . . 218
B.5 CO2 emission rate, D = 20 cm. . . . . . . . . . . . . . . . . . . . . . 219
B.6 CO2 emission rate, D = 30 cm. . . . . . . . . . . . . . . . . . . . . . 220
B.7 CO emission rate, D = 10 cm. . . . . . . . . . . . . . . . . . . . . . . 221
xvi
B.8 CO emission rate, D = 20 cm. . . . . . . . . . . . . . . . . . . . . . . 222
B.9 CO emission rate, D = 30 cm. . . . . . . . . . . . . . . . . . . . . . . 223
B.10 Experimental variability in the measurement of m? at D = [10, 20, 30]
cm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
B.11 Average values of centerline temperature measured at different axial
distances for PFs and FWs formed with heptane and ANS crude oil. . 225
B.12 Maximum values of centerline temperature measured at different axial
distances for PFs and FWs formed with heptane and ANS crude oil. . 226
B.13 Rez, estimated as (m?
??D/?)/?, with ? and ? evaluated at Tfilm. The
values of Tfilm for heptane PF, heptane FW, ANS PF and ANS FW
were 675 K, 800 K, 650 K and 725 K, respectively. The ratio (m???/?)
is Uz=0, the spatially averaged value of fuel evaporation at the fuel
surface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
B.14 Re?, estimated as ?/?. . . . . . . . . . . . . . . . . . . . . . . . . . . 227
B.15 Riz, estimated as Gr/Re
2
z. . . . . . . . . . . . . . . . . . . . . . . . . 228
B.16 Ri?, estimated as Gr/Re
2
?. . . . . . . . . . . . . . . . . . . . . . . . . 228
B.17 Fuel Froude number, Frf , estimated as (m?
??2)/(?20gD) [11], with ?0
evaluated at T0 = 300K. . . . . . . . . . . . . . ?. . . . . . . . . . . . 229
B.18 Modified Peclet number, Pe, evaluated as (m?)/(? gD5) [35]. . . . . 229
C.1 Average contours of CH* chemiluminescence intensity, maximum in-
tensity and location indices of maximum intensity for ? = [0.7, 0.8, 0.9, 1.0].230
C.2 Average contours of CH* chemiluminescence intensity, maximum in-
tensity and location indices of maximum intensity for ? = [1.1, 1.2, 1.3, 1.4].231
C.3 Average contours of OH* chemiluminescence intensity, maximum in-
tensity and location indices of maximum intensity for ? = [0.7, 0.8, 0.9, 1.0].232
C.4 Average contours of OH* chemiluminescence intensity, maximum in-
tensity and location indices of maximum intensity for ? = [1.1, 1.2, 1.3, 1.4].233
C.5 CH* in the BW. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
C.6 OH* in the OH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
C.7 Individual frames showing PAH in the BW. . . . . . . . . . . . . . . 236
C.8 Individual frames showing OH in the BW. . . . . . . . . . . . . . . . 236
C.9 Sequence of frames showing OH in the FW. Two frames (top right,
bottom left) show a tendency to transition to the BW, evidenced by
the bulging of the flame sheet, but the flame stays in the FW regime
in the following frames. . . . . . . . . . . . . . . . . . . . . . . . . . . 237
C.10 Vectors showing the instantaneous velocity of two examples of soot
particles emanating from the blue ring and passing through the post-
flame region. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
xvii
Nomenclature
Symbols
A fuel pool area m2
C heat capacity J/kg-K
d, D diameter mm, cm, m
D diffusion coefficient m2/s
f function ?
f pool-fire puffing frequency s?1
g gravitational acceleration m/s2
G intensifier gain value %
h height or axial distance cm, m
H flame height cm
?h lower heating value kJ/g
I chemiluminescence intensity ratio ?
I species chemiluminescence intensity ?
L length scale m
m mass mg, g, kg
m? burning (mass-loss) rate g/s
q heat flux kW/m2
Q? heat-release rate kW
r radial distance mm
R vortex radius m
S enclosure side length cm
S Swirl number ?
t time s, min
T temperature K
?T excess temperature K
u, U flow velocity m/s
V? fuel volumetric flow rate ml/min
w flame width cm
W enclosure gap size cm
z axial distance mm
Nondimensional numbers
Da Damko?hler number
Fr Froude number
xviii
Gr Grashof number
Pe Peclet number
Ra Rayleigh number
Re Reynolds number
Ri Richardson number
Ro Rossby number
St Strouhal number
Stk Stokes number
W? whirl number
Greek symbols
? expansion coefficient K?1
? efficiency ?
? circulation m2/s
? kinematic viscosity (momentum diffusivity) m2/s
? equivalence ratio ?
? thermal conductivity W/mK
? density kg/m3
?? density change at the flame sheet kg/m3
? dynamic viscosity Ns/m2
? characteristic time scale s
? proportional to
Abbreviations
ANS Alaska North Slope crude oil
ASGSR American Society for Gravitational and Space
Research
BSEE Bureau of Safety and Environmental Enforcement
BW Blue Whirl
CF Consumption Factor
DAQ Data Acquisition
xix
EF Emission Factor
FE Forced Entrainment
FID Flame Ionization Detection
FPS Frames Per Second
FW Fire Whirl
HF Heat Flux
HRR Heat-Release Rate
ISB In-Situ Burn
LFW Large Fire Whirl
LII Laser-Induced Incandescence
LIFS Laser-Induced Fluorescence Spectroscopy
MCE Modified Combustion Efficiency
MLR Mass-Loss Rate
NASA National Aeronautics and Space Administration
NDIR Non-Dispersive Infrared
NE Natural Entrainment
PAH Polycyclic Aromatic Hydrocarbons
PF Pool Fire
PIV Particle Image Velocimetry
PLIF Planar Laser-Induced Fluorescence
PM Particulate Matter
RZ Recirculation Zone
SLPM Standard Liters Per Minute
TAMU Texas A&M University, College Station
TBW Transition Blue Whirl
TC Thermocouple
TEM Transmission Electron Microscopy
TPM Total Particulate Matter
UHC Unburned Hydro-Carbons
xx
UMD University of Maryland, College Park
VB Vortex Breakdown
VOC Volatile Organic Compounds
WPI Worcester Polytechnic Institute
Subscripts
0 nominal ambient conditions
b burning
c combustion
C carbon-mass based
cr critical
f flame feedback or fuel
h horizontal length scale
H Hencken burner
p constant pressure
r radial component
s species
? azimuthal (tangential) component
z axial component
Superscripts
? nondimensional quantity or excited-state radical species
?? per unit area
? per unit time
xxi
Chapter 1
Introduction
1.1 Swirling Flows and Whirling Flames
Swirling flows are ubiquitous in nature. They occur at a large range of length
scales ? on the order of many tens-of-thousands of kilometers in the case of Sat-
urn?s hexagonal polar vortex [1] and Jupiter?s Great Red Spot [36], a few hundred
meters in the case of secondary motions causing river meandering [37?39], to just a
few centimeters in dandelion flight [2]. Historically, eddies and vortices have been
the topic of academic [40, 41], technical [42], and even artistic interest in the case
of Vincent van Gogh?s The Starry Night [43], which has intrigued turbulence re-
searchers [44?46].
Most instances of swirling flow involve momenta competing in two different
directions, primarily axial and azimuthal. Whirling flames (also known as fire whirls)
represent a general class of flows that may be considered slender vortices involving
axial (buoyant) momentum from heat-release, and azimuthal momentum from air
1
Figure 1.1: Examples of swirling flow in nature at different length scales. (Left)
Polar vortex within the hexagonal jetstream of Saturn, from [1]. (Right) Separated
vortex structure in dandelion flight, from Cummins et al. [2].
entrainment. These structures are observed frequently in wildland fires and have
been studied for many decades in the context of fire spread and firefighter safety [47?
49]. Most recently, four fatalities were reported as the direct result of the formation
of a large fire whirl during the 2018 Carr Fire in California [50,51]. Fire whirls can
also accelerate wildland fire spread rates dramatically by increasing buoyancy and
helping loft firebrands long distances, causing fires to initiate prior to the arrival of
the fire front [52]. Various types of fire whirls can also occur in large urban fires [53].
Many approaches to classify, scale and predict fire whirl formation for the
purposes of safety are prevalent [5]. The simplicity of generating reacting fire whirls
in a laboratory and the complex interaction between combustion and fluid mechanics
[8] provides a platform to study phenomena such as turbulence suppression, mixing,
vortex stretching and dynamics. Various aspects such as formation methods in the
laboratory [5, 47, 48], thermal structure [8], velocity structure [54], scaling [11, 12],
flow structure [55,56], etc. have been reported in the literature. Byram and Martin
2
[48] listed three essential conditions required for thermally-driven vortex structures
such as fire whirls ? (i) an eddy generating mechanism, (ii) a fluid sink within the
eddy, and (iii) drag (or a boundary layer) at the bottom boundary of the slender
eddy.
Swirling flow fields are used in engineering applications to enhance combustion
efficiency and reduce emissions [57,58]. They are typically used in aerospace appli-
cations and involve gaseous and liquid fuels, and both premixed and nonpremixed
modes of operation [59]. Whirling flames are distinct from swirling flames in the
manner that fuel and air interact [60]. In swirl flames, the momentum of fuel in-
jection is comparable with that of the air, both of which have axial and azimuthal
components, enabled by mechanical forcing of both to enhance turbulent mixing
prior to combustion. The reduced reaction length allows high heat-release rates to
be achieved in small burners. In whirling flames, however, momentum of the fuel
supply is negligible, as is azimuthal momentum of the entrained air. Turbulence is
suppressed in fire whirls [8], causing flame lengths to increase when compared to
non-whirling pool flames, even when the burning rate similar. Jet flames under the
influence of rotating flow fields posses different boundary conditions at the base of
the flame, but show similarity to fire whirls in the vortex structure of the continuous
flame region [61?63].
Most fire whirls, irrespective of formation method or fuel (except light, oxy-
genated fuels such as methanol or ethanol), exhibit a yellow color, which stems from
incandescent soot particles in the flame (see Figure 2.3). An aspect of fire whirls
that has not been explored in the literature is the quantification of particulate-
3
matter (soot) emissions. This is investigated in the first part of this dissertation by
comparing emissions from fire whirls and free-buoyant pool fires.
4
1.2 The Blue Whirl
In recent experiments by Xiao et al. [64], the blue whirl was discovered during
experiments with liquid-fueled fire whirls formed over a water surface. An abrupt
transition from a traditional fire whirl to a completely blue flame was seen at some
experimental conditions. The thermal characteristics of this flame structure was the
subject of the author?s M.S. thesis [65].
The blue whirl is a laminar regime of the fire whirl that burns with negligible
soot formation and stabilizes directly over a liquid fuel surface [22,65,66]. The flame
(Figure 1.2) has an inverted conical region below, a hazy region above, and a bright
blue ring at the edge. This flame regime is of interest since it transitions from the
sooty fire whirl regime without any externally-forced air entrainment.
Soot-free flames are preferred in many applications where combustion efficiency
and issues like coking are critical to performance. Achieving a soot-free flame typi-
cally requires either premixed or lean operation, which may not be possible in many
industrial applications due to the safety and control issues associated with premixed
flames. Non-premixed (diffusion) flames are the choice for most industrial applica-
tions, but are challenged by soot formation. Thus, a burner based on the blue whirl
could present opportunities to have control over the combustion process similar to
a non-premixed flame, and inherent soot-free operation even when burning liquid
fuels.
Swirl stabilization has been used as a technique to extend the lean-side op-
erating range of combustors [59, 67, 68] and to reduce emissions [58, 60, 69]. Since
5
Figure 1.2: An image of the blue whirl formed with heptane over a water surface.
The lower portion of the image is the reflection of the flame over the water surface.
such combustors typically operate at high power output, it is necessary for them
to operate in the turbulent regime and use various injection techniques to pro-
mote fuel-air mixing. The reduced residence times in combination with lean, low-
temperature combustion of light fuels, such as methane or propane, can lead to low
emissions [57, 59,61,70,71].
The blue whirl, however, is a laminar whirling flame. In all the experiments,
the blue whirl transitions from the fire whirl, and the transition requires buoyancy
and circulation to exist in specific ratios, and emits negligible quantities of soot
(PM). The thermal structure and experimental conditions required for blue whirl
formation were studied in previous work [65], but questions regarding the mecha-
nisms governing its formation and the nature of combustion still remained.
Questions regarding the transition and structure of the blue whirl flame are the
second aspect of whirling flames considered in this work. Using recent experimental
data on the formation limits of the blue whirl [24], dimensional analysis and scaling
6
are used to explain the fluid-dynamic factors that govern the transition process from
a fire whirl to a blue whirl. Non-intrusive optical diagnostics are used to present
preliminary insight into the flame structure. Some of the results presented here have
been used as boundary conditions for numerical simulations of the blue whirl [33,72].
7
1.3 Motivation and Scope
Emissions and near-limit aspects of fire whirls have not been given much con-
sideration in the literature. Emissions such as NOx, SOx, CO2 and particulate
matter (PM) from air-breathing engines are regulated since they can impact hu-
man health [73]. Emissions from in-situ burning (ISB), which are essentially large
pool fires, have been characterized in the literature [73?76]. ISBs are a trade-off
between marine pollution, caused by oil spreading over the sea water, and air-
borne particulate-matter (PM) emissions, which can impact human respiratory sys-
tems [77,78]. Reducing airborne particulate emissions could reduce the environmen-
tal impact of remediating offshore and onshore oil spills.
Reducing soot emissions from ISBs, incinerators and gas flares [79] offer moti-
vation to investigate the differences between the fire whirl and pool fire regimes. In
this work, the hypothesis that is tested is that for a burning pool of liquid fuel, fire
whirls emit a lower mass of particulate matter. Applicability to ocean environments
and practical ISB apparatus may be considered in the future using the fundamental
results presented here to design combustion devices.
While emissions from fire whirls in a wildland fire may not be of much practi-
cal interest, the intensification in combustion due to a fire whirl and the emissions
behavior are still fundamental questions that are a gap in the literature. The addi-
tion of swirl to gaseous-fuel burners has been shown to reduce emissions [60]. These
results, however, are not immediately applicable to free-burning fires [80] such as
pool fires and fire whirls ? regimes in which the role of turbulence is quite disparate.
8
In the first part of this dissertation (Chapters 3 and 4), particulate matter
(PM) emissions are measured from pool fires and fire whirls formed using different
fuels. Parameters such as fuel and slick diameter are varied to study behavior at
different length scales. These emissions are compared for fire whirls and free-buoyant
pool fires. If combustion efficiency is higher and emissions from fire whirls are lower
than that from pool fires, fire whirls could potentially be used in an engineering
device at an appropriate length scale such that it may be utilized for improving
incineration of spilled combustible materials.
The second part of this dissertation (Chapters 5 and 6) discusses the blue
whirl, the near-limit whirling flame that occurs at very low heat-release rates [24]
and is very sensitive to experimental conditions. These factors do not permit their
use in practical engineering devices in the near future since sufficient energy-release
rates may not be achieved for applications such as propulsion. The flame is, however,
of academic interest to identify the characteristics that enable soot-free combustion
while directly burning liquid fuels. It could also serve as a model to study reacting
vortex breakdown and edge flames under in a rotating flow field.
A scaling analysis is presented to understand the conditions that influence the
transition from the fire whirl to the blue whirl. This analysis builds on previous work
by Hu et al. [24], and uses flame geometry to determine the characteristic length
scales for nondimensionalization of the two parameters that influence whirling flames
? circulation and heat-release rate. Buoyancy and momentum dominated regimes
are common in many flame and fire problems [81] and these analogies are used here
to explain the transition.
9
Finally, the flame structure of the blue whirl is investigated using optical diag-
nostics including chemiluminescence and planar laser-induced fluorescence (PLIF).
Species such as OH, PAH, OH* and CH* are imaged to identify their locations in
the flame and comment on the general structure and relate them to previous work.
The results from both of these aspects have been used for numerical modeling of the
blue whirl [72, 82]. Scaling results were also used to design preliminary micrograv-
ity experiments of fire whirls, conducted in the 5.18 s drop tower at NASA Glenn
Research Center in May 2019 [83].
10
1.4 Outline of the Dissertation
This dissertation explores whirling flames at different length scales from an
experimental standpoint. The largest fire whirls were on the order of 5 m in height,
while the smallest blue whirls were on the order of 5 cm. A general literature re-
view on fire whirls, the motivation and background on the diagnostic techniques
are presented in Chapter 2. The chapters that follow are roughly organized based
on the journal articles that compose this dissertation, and each contains a separate
introduction and literature review. A comprehensive list of references is provided
as a bibliography. Chapter 3 presents the experimental data obtained for pool fires
and fire whirls. This raw data is analyzed in detail and the relevant nondimensional
relationships influencing PM emissions are presented in Chapter 4. The effect of
boilover on the burning characteristics and emissions from large-scale ANS crude oil
fires is also discussed. Chapter 5 presents the dimensional analysis used to explain
the onset of vortex breakdown and the transition to the blue whirl regime, fol-
lowed by non-intrusive optical diagnostics measurements in Chapter 6. A summary,
conclusion and future research directions are presented in Chapter 7. Additional
figures, raw data and preliminary data for future work are included in Appendices
A?C. Videos are included in a separate accompanying document as Supplementary
Figures.
11
Chapter 2
Review of Fire Whirls and Experimental Methods
2.1 Swirl and Whirl Combustion
Whirling flames, also known as fire whirls, are a class of thermally-driven
vortices which can be either reacting or non-reacting flows. A detailed classification
is given by Kuwana et al. [11] and Tohidi et al. [5]. Type-I fire whirls are stationary,
on-source (formed directly over the fuel source), reacting fire whirls. Reacting fire
whirls that are unsteady and shed periodically (Type-II) are typically formed when
a line fire is subject to angled crosswind. Type-III fire whirls can be reacting or
non-reacting (comprised only of hot combustion products), but are usually formed
off source [53]. They typically form close to one or more nearby reacting plumes.
The fire whirls investigated in this work are all classified as stationary, on-source,
and Type-I.
Whirl flames are distinct from swirl flames, with the definitions depending on
the manner in which fuel and air are introduced into the reaction zone [60, 84]. In
12
Figure 2.1: Radial variation of azimuthal velocity in (A) a swirl combustor [3], and
(B) a fire whirl [4].
whirl combustion, air is introduced with a strong azimuthal component and very
low axial component. Fuel is introduced separately and whirl flames are typically
nonpremixed flames. In contrast, in swirl combustion, air is introduced with a
combination of both axial and azimuthal components, and may be premixed with
fuel. The method of flame stabilization between these flames is different. Swirl
flames are stabilized by a recirculation zone that acts as a flame holder, and peak
azimuthal velocities are found at the edges of the enclosure, and don?t show the
existence of a free-vortex region due to wall effects.
Whirling flames have a radial distribution of azimuthal velocity that is well
described by a Burger?s vortex model [12], and have an inner core defined by solid-
body rotation and an external free vortex region. The differences are shown in
Figure 2.1, adapted from references [3,4]. The inner core in swirl flames consists of
13
reactants and products, whereas the inner core in whirling flames consists mostly of
fuel vapor [8]. Additionally, swirl flames typically operate in the turbulent regime,
even at low swirl numbers [85]; the reaction zone in fire whirls is typically in the
laminar or transitional regimes for a very large range of length scales, with the plume
being turbulent [86].
14
2.2 Fire Whirls
One of the necessary conditions for fire whirl formation according to Byram
and Martin [48] is the presence of drag along the surface over which the fire whirl is
formed (refer to section 1.1). This third condition, essentially a radial boundary layer
at the base of the fire whirl, is not expected to be applicable to some laboratory fire
whirl apparatus that are referred to as an ?Emmons-type rotating mesh? setup [8],
where circulation is externally imposed on the flame.
In fire whirls (Figure 2.3), the fluid sink is created by buoyancy that is gener-
ated by heat release from combustion. In a laboratory, the eddy is generated either
by arrangement of obstacles in the path of the entrained air using vanes or walls, or
mechanically swirling air by means of a rotating mesh or fans (Figure 2.2). During
wildland fires, concentrated regions of buoyant flow may be created by fires in con-
fined regions and eddy formation is enabled by topographical factors or opposing
winds. In a laboratory environment, fire whirls are generated in a multitude of ways,
all of which can be classified as either natural air entrainment [9, 12, 56, 87, 88] or
forced entrainment.
The design of natural entrainment configurations typically involves a set of
walls with multiple openings that permit tangential entrainment. Configurations
with two half-cylindrical shells are used for small fire whirls, while the four-walled
configuration is the most common [5]. When the number of panels used is above
four, they are sometimes referred to as vanes, although few studies utilize such
configurations [89]. Coenen et al. [23] used twelve vanes for their experimental study
15
on the blue whirl. Fire whirls can also be formed without the use of physical walls.
The shedding of fire whirls from line fires was shown to be dependent on a critical
ambient velocity by Zhou et al. [90], and Wang et al. [91] built a configuration using
air screens, and may be considered to be a variation of the four-walled apparatus.
The most popular of forced-entrainment apparatus is the Emmons-type rotat-
ing mesh [8,92,93]. Lee and Garris [94] modified this configuration to include a line
burner between two screens moving anti-parallelly, generating multiple equidistant
fire whirls. Byram and Martin also suggested the use of a combination of both nat-
ural and forced entrainment methods to generate fire whirls by using a combination
of vanes and air curtains [48]. The combination of natural and forced entrainment
systems are used in some laboratory apparatus for safety purposes. A combination
of the four-walled configuration with an exhaust suction system was used to generate
the fire whirl photographed in Figure 2.3.
16
Figure 2.2: Laboratory configurations used to form fire whirls. Figure from Tohidi
et al. [5].
Figure 2.3: Photograph of a fire whirl formed in a fixed-frame setup with a combina-
tion of natural and forced air entrainment. The pool diameter is 30 cm, placed in an
enclosure of dimensions 3.6?3.6?1.5 m3. To increase entrainment velocity through
the four inlets, a suction system was employed above the fire. In configurations
with natural entrainment, buoyancy and air entrainment are coupled. Buoyancy
within the enclosure must be capable of drawing sufficient air into the enclosure.
This minimum level of circulation required for fire-whirl formation depends on many
factors including pool diameter [6], and small pools within large enclosures do not
form strong fire whirls without externally-forced entrainment.
17
Combustion in a fire whirl occurs in a slender cylindrical structure with the
reaction occurring mostly at the edge of the vortex core where the shear is highest,
promoting mixing. The interior of the fire whirl core contains mostly fuel vapor
under solid body rotation [4,12,54]. Peak temperatures are found at the edge of the
vortex core (Figure 2.4 A). For the same pool diameter, the height of a fire whirl is
larger than that of a pool fire. This elongation and stretching of the fire whirl may
be attributed to
1. the increase in burning rate (Figure 2.4), caused by increased heat flux, both
convective [88] and radiative [87], feedback to the fuel surface.
2. turbulence suppression in turbulent fire whirls [95]. This is caused by increased
temperatures and lower Richardson number (Ri) in the continuous-flame re-
gion of the fire whirl, resulting in increased mixing length.
3. an absence of the puffing instability that is prevalent in pool fires (see section
2.3), where the instability is responsible for periodic mixing of fuel and air.
The puffing instability is replaced by a helical mode in fire whirls [23], also
sometimes associated with the helical mode of vortex breakdown in conical
fire whirls which are close to a transitional regime [64].
4. the vortex structure and the concentration of the vortex core through an-
gular momentum conservation and turbulence suppression [95]. Vortex core
concentration is balanced by pressure gradients generated by centripetal ac-
celeration [8]. This structure is present even in non-reacting swirling buoyant
plumes.
18
These ensure that in the absence of puffing, a central core of fuel vapor requires
a long mixing length (and time), increasing the length of the flame sheet. The
increased flame surface area essentially balances the higher burning rate in liquid-
fueled flames. The higher temperatures and larger surface area also lead to a ?40%
increase in radiative feedback (see Fig. 9 in [96]). Even if the burning rate (and
thus, heat-release rate) for a pool fire and fire whirl are the same, while the total
fire-whirl surface area may be the same, depending on the fuel pool diameter, fire
whirls may have a lower flame width and higher flame height owing to the vortex
structure. This needs to be verified in the future through systematic experiments
of pool fires and fire whirls, both formed at the same heat-release rate using gas
burners.
Drag at the bottom surface boundary is generated by a boundary layer formed
by radial entrainment along the surface over which the fire whirl is formed. When
circulation is not externally imposed on the flame, the presence of a radial boundary
layer is essential for, or rather a consequence of, fire whirl formation. Dobashi
et al. [87] performed preliminary experiments to elucidate the effect of the radial
boundary layer on fire whirl formation. If the height of the vertical obstruction
to inflow is higher than the radial boundary layer, fire whirl formation does not
occur and the flame remains similar to a pool fire regime. For this reason, fire whirl
apparatus with natural air entrainment typically have the surface of the burner
(for gaseous fuels) or the fuel dish (for liquid fuels) flush with the bottom surface
boundary [6, 7, 12,23,54,88,97].
For both natural and forced entrainment, when this radial boundary layer
19
Figure 2.4: (A) Radial variation of temperature in a fire whirl, from [7]. (B) Burning
rate of acetone as a function of ambient circulation, from [8]. The difference in
burning rate between the cases with and without the flame holder may be attributed
the level of disruption in the Ekman layer by the vertical lip formed by the flame
holder.
(Ekman layer) is affected by the presence of disruptions to the incoming flow such
as vertical projections formed by fuel pans etc., fire whirl behavior is affected. For
instance, the influence of fuel pool height on burning rate is seen in the results of
Emmons and Ying (Fig. 5 in [8], shown in Figure 2.4 B). Experiments by Byram
and Martin [48] also noted the effects of such obstructions on fire whirl formation,
and noted that when the
burning pool of alcohol is flush with the surrounding horizontal surface,
the burning rate will be drastically reduced in a strong vortex. The flame
will be blown over the liquid surface except in a small, central area where
a core of whirling flame a few inches in height remains. This does not
happen if the edge of the burner projects a short distance above the liquid.
20
These observations are very similar to the sensitivity of the blue whirl to
experimental conditions [65,66]. In an apparatus allowing natural entrainment of air,
fire whirl formation depends on the strength of the radial boundary layer that causes
Ekman layer formation [5, 87]. The radial boundary formed under swirling flow is
typically referred to as the Ekman layer in fire whirls, and a graphical representation
of the velocity field around a fire whirl is shown in Figure 2.5. For a given fuel
pool diameter, when the height of the step is sufficiently large to affect the radial
boundary layer, the fire whirl flame shape at the bottom surface is affected (see Fig.
11 in [87] and Fig. 8 in [5]). Similar effects are observed even in pool fires, where the
height of the lip above the bottom surface boundary affects burning rate [98?100].
For fire whirls formed in a 50 cm fuel pool, Lei et al. [97] show about a 10% difference
in burning rate between two different lip heights, although the authors claim that
the difference is not significant since only 1 repetition for each was performed.
Apparatus in which swirl is imposed on a jet flame, however, do not necessarily
need a boundary layer or drag at the bottom surface boundary to form a concen-
trated reaction zone. Laboratory experiments on fire whirls do not operate in a
regime where radial inflow does not exist at the base, although formation of a flame
with swirling characteristics is still possible. In these cases, the concentration mech-
anism may be expected to be similar to a buoyant plume under ambient circulation
that exhibits vortex stretching and concentration [101], and turbulence is actually
enhanced by mixing at the edge of the fuel jet and the swirling flow field [62].
Chigier et al. [62] noted that in some jet flames under imposed swirl (where
the radial boundary does not exist), the central reacting core does not exhibit ro-
21
Figure 2.5: A graphical representation of the velocity field around a stationary
Type-I fire whirl. Figure adapted from [5].
tation, and the external ambient circulation induces turbulence causing shortening
of the flame. This is fundamentally different from a fire whirl, where the central
core exhibits solid-body rotation and turbulence is suppressed, reducing mixing and
leading to flame lengthening [8, 95]. Thus, in apparatus that use a central gaseous
jet and externally imposed circulation to study fire whirls, the surface of the burner
is typically flush with the burner exit to promote the formation of a radial boundary
layer at the bottom surface boundary [6, 92,93,102].
For fire whirls formed with gaseous fuels supplied through a burner, burner
diameters typically above 35 mm show fluid dynamic similarity with larger fire
whirls. Apparatus with smaller burner diameters tend to form jets, which under
imposed rotating flow fields [103] tend to show both turbulence enhancement at the
lower region of the flame (similar to Chigier?s observations [62]) and suppression
22
further downstream (similar to fire whirls). This causes jet fires under rotating flow
fields to show flame shapes that are different from gas-fueled fire whirls (for ex. [92]),
and consequently scale differently from fire whirls [103]. Some of these differences
may be attributed to momentum from the fuel jet, which causes lift-off under high
swirl [103], an effect typically not seen in liquid-fueled fire whirls. Lift-off from the
surface of a liquid fuel can, however, occur when both heat-release rate and the
boundary layer conditions are met [24], such as in the case of the blue whirl [23,64].
2.2.1 Scaling Approach and Influential Parameters
Scaling methods on the basis of similarity arguments have been used to predict
the formation of fire whirls [11,53] and the geometric attributes of fire whirl structure
[9, 12, 13, 56]. The parameters influencing fire whirl structure, as summarized in
[5, 104] are
(Ur, Uz,?, H, m?) = ? (Lh, Q?, Cp,??, ?,?T, T, g, ?, ?, ?,Ds) (2.1)
where Ur and Uz are the radial and azimthual velocity components, ? is the circula-
tion, H is the fire-whirl height, and m? is the burning rate of fuel (either gaseous or
liquid). The horizontal length scale is given by Lh. Experimentally, the heat-release
rate (Q?) is determined either from m? or oxygen-consumption calorimetry. The spe-
cific heat capacity (Cp) density change (??), excess temperature (?T ), dynamic vis-
cosity (?), expansion coefficient (?), thermal conductivity (?) and molecular species
diffusion coefficient (Ds) are determined as a function of flame temperature at the
23
fire whirl flame sheet. The ambient air density and temperature are given by ? and
T .
This parameter space results in 13 nondimensional gr(oups, of wh)ich a few im-
portant numbers are mentioned here. The Froude number Fr = ?Uz( denote)s thegH
ratio of tangential velocity to buoyant velocity. The Rossby number Ro = UzH also
?
relates the buoyant and tangential momenta, and is analogous to the swirl number
for swirl combustors [57]. Dimensionally, it is the inverse of Froude number, but is es-
timated based on ?, which represents the strength of ambient swirl outside the flame
sheet. Chuah et al. [9] showed that the structure of fire whirls could be better de-
s(cribed by)Ro than by Fr [5].(Other traditio)nal quantities such as Reyn(olds numb)er
3
Re = ??
g??TL
, Grashof number Gr = h2 and Richardson number Ri =
Gr
2 ,? ? Re
all defined for the vortex core, have been used to discuss the important role of
turbulence suppression in fire whirl height [95]. Finally, the heat-release rate and
buoyant momen(tum from fu)el evaporation are given by Q?? = Q?? and the?Cp?T gL5h
Peclet number Pe = m?D , respectively. The axial momentum represented by?Lh s
these two quantities is typically applicable only to reacting fire whirls. Klimenko
and Williams [10] estimated Pe as (UzLh/Ds), and showed the strong influence of Pe
on the nondimensional height of fire whirls under high ambient circulation (Figure
2.6). Chuah et al. [9] also found a similar relationship, and estimated Uz as
?
m??? ??
Uz = ( )1 ??fuel + ? 22 air L ?h 1300 K
Apart from the quantities discussed above, parameters derived from experi-
24
Figure 2.6: Influence of Pe on nondimensional height of fire whirls under high ambi-
ent circulation, (A) shown experimentally in [9] using a tilting fire whirl apparatus,
and (B) developed analytically in [10].
mental conditions not directly related to the momenta within the vortex core have
also been studied. For instance, the critical ambient velocity (Ucr) that may result
in fire whirl formation, provided an existing fire is not dominated by momentum,
varies as Ucr ? Fr0.15f [11], where L is the horizontal length scale, and Frf is theg/L
fuel Froude number (Figure 2.7 A). For small scale fire whirls, fire whirl height
depends primarily on circulation, and this is consistent for naturally entrained con-
figurations as well for apparatus where circulation is imposed (Figure 2.7 B). Along
the height of a fire whirl that is not fully dominated by buoyancy, the circulation
remains relatively constant above the entrainment region at the bottom [4,97]. The
peak tangential velocity (and thus, circulation) reduces consistently from a height
of H = 0.4, to the bottom surface boundary where viscous effects cause the no-slip
Lh
condition [12,63].
25
Figure 2.7: (A) Dependence of Ucr on Frf for buoyancy-dominated fire whirls,
adapted from [11]. (B) Variation of nondimensional fire whirl height with nondi-
mensional circulation, adapted from [12].
Figure 2.8: (A) Variation of flame height with Pe. The prediction of flame height
based on Pe was improved by using a modified axial velocity profile. Figure adapted
from [13]. (B) Dependence of Fr on nondimensional circulation, adapted from [12].
The other parameter that influences fire whirl flame height is Pe, derived from
the fuel flow velocity (evaporation velocity for liquid pools, exit velocity for gas
burners). Pe for fire whirls (Chuah et al. [9]) is similar to the fuel Froude number
(Frf ) defined by Kuwana et al. [11] and the ratio of fuel flow rate to advection
rate [5,105]. The effect of Pe on fire whirl height is shown in Figure 2.8A. Pe defined
based on fuel mass flux is represented as m? , where L
?L D h
is the horizontal length
h s
26
?? 2
scale and Ds is the species diffusivity. Similarly, Frf is defined as
(m? ) . The overall
?0 g Lh
Fr for the fire whirl also depends on nondimensional circulation (Figure 2.8B). Hartl
and Smits [12] also noted that fire whirl diameter is the appropriate horizontal
reference length scales for fire whirls since fire whirls can form over an infinite fuel
pool. Consequently, their experiments using a gas burner did not conform to the
scaling between nondimensional circulation and Fr.
27
2.3 Pool Fires, Emissions and Marine Oil Spills
Pool fires (Figure 2.9) are a standard research problem applicable to a wide
range of fire scenarios, and are of interest for both fundamental and practical pur-
poses. Characteristics such as pulsation frequency, burning rate, heat flux feedback,
temperature, soot fraction and radiation have been studied for various fuels, and
detailed reviews are presented in [106?109]. Additionally, flame structure [98] and
scaling behavior [110] are also of interest for predictive purposes, including cases
where a free buoyant plume is affected by crosswinds etc. [111].
Entrainment of air towards pool fires causes necking of the flame at the base
and causes the initiation of Rayleigh-Taylor instabilities [112]. These instabilities
cause the development of the phenomenon known as ?puffing,? which is a periodic
oscillation and fluctuation of the flame height. The puffing frequency, f, decreases
with pool diameter as f ? D?0.49 [14], and the puffing Strouhal number (St), is
linearly dependent on the Froude number (Fr) [15] (see Figure 2.10). The onset
of puffing is attributed to a critical Rayleigh number (Ra) [113], supported by the
development of large-scale toroidal vortex rolls occurring at an axial distance within
one burner diameter above the fuel surface [114]. Cetegen et al. [114] also found that
the puffing frequency for thermal and isothermal plumes was similar (f ? D?0.5). Be-
jan [115] analyzed the phenomenon using the buckling of inviscid streams to explain
2
why the expression f2 ? 2.7 [m/s ]= worked well in predicting shedding frequency (f,D
[s?1]) over a very large range of pool diameters (D, [m]).
Interest in the studies of emissions from pool fires, particularly particulate
28
Figure 2.9: Photograph of a pool fire formed with n-heptane, D = 30 cm.
Figure 2.10: (A) Dependence of pool-fire puffing frequency on pool diameter,
adapted from [14]. (B) Influence of Fr on puffing St, adapted from [15].
29
matter, CO, and CO2 is recent, stemming mostly from the initial work of Aurell
et al. [116] on solid waste incineration. Measurements of emissions from pool fires
of gaseous and liquid fuels have also been performed [16]. Apart from particulate
matter, CO, and CO2, various PAHs and VOCs were also measured, each with
an emission factor in the range of hundreds of milligrams per kilogram of fuel.
Experiments involved sampling in the fire(plume, and the calcu)lated emission factors
and the modified combustion efficiency MCE = ?CO2 are shown in Figure
?CO2+?CO
2.11.
There have been a number of studies of oil spreading over open water [117,118],
and several of these have addressed the problems of burning spilled fuel [119]. Oil
spills can have devastating effects on both onshore and offshore ecosystems. Burning
spilled oil is a quick response method [76] since their long-term existence can cause
toxicity to both wildlife and humans [120]. The work of van Gelderen et al. [121]
identified properties such as initial slick thickness, vaporization order, slick diameter,
weathering state, heat flux feedback and heat losses to the water layer as factors
that influence burning efficiency in an in-situ burn (ISB).
During recovery efforts for accidental spills, the dispersed fuel is gathered us-
ing booms and ignited once the fuel layer is thick enough. This layer must be thick
enough because it must insulate the top layer of fuel from the water. This entire
approach is fundamentally different from traditional approaches and studies of pool
fires burning lighter fuels over water [122]. Wu et al. [123] performed experiments to
determine the ignition, flame spread, and mass burning characteristics of oil spilled
on water. Later they expanded to more detailed ignition experiments determining
30
Figure 2.11: Comparison of PM-2.5 emission factor and Modified Combustion Effi-
ciency (MCE) for JP-5 and propane pool fires. Figure adapted from [16].
the effects of weathering on crude oil [124], which is important because, as oil be-
comes more weathered, it becomes more difficult to ignite, and finally it emulsifies
to the point where ignition is no longer possible [125]. Based on these results, phe-
nomenological descriptions have been developed for combustion of oil on a water
layer [126,127].
More recent experiments have extended the experimental capability to mea-
sure the properties of crude oil layers and its effects, particularly applied to arctic
regions [128]. No prior studies, however, have been conducted to understand and
characterize the fundamentals of fire whirl behavior over open water, in particular
from small to large scales. More background on this is provided in Chapters 3 and
4.
31
2.4 Soot Formation and Emission
The study of soot formation and emission from flames has been a challenging
problem in combustion science [129]. Soot may be defined as a carbonaceous ma-
terial containing up to 1% hydrogen by mass, an empirical formula of C8H, with
diameters of aggregates ranging between 100 and 2000 A? [130]. Soot formation and
destruction processes occur simultaneously in flames. This was illustrated as early
as 1908 in ?The Chemical History of a Candle? by Michael Faraday [131,132]. The
process may be simplified to the following four processes [17]:
1. Formation of precursor species.
2. Particle inception.
3. Surface growth and particle agglomeration.
4. Particle oxidation.
While both premixed [133?135] and nonpremixed [130, 136] flames can form
and emit soot, from an engineering perspective, nonpremixed flames are primarily
studied since they are used in most engines (internal combustion such as recipro-
cating [137] or gas turbine [58]) and industrial burners and flares [79]. The term
particulate matter (PM) is more commonly used [58] and is a concern since they
can have significant impact on human health [138]. As a consequence, PM emis-
sions from commercial engines and burners are strictly regulated by environmental
agencies around the world [139].
32
The tendency for a flame to emit soot can be represented by an empirically
obtained property called the smoke point, an experimental condition that results
in visible emission of black soot from a flame. Experiments use a diffusion flame
and the smoke point is measured in terms of either the maximum flame height or
maximum fuel flow rate at which a stream of soot is first visible downstream of
the flame. The larger either of these values, the lower the propensity of the fuel
to soot [17]. The propensity for the different fuel families to soot, from highest to
lowest, is aromatics>alkynes>alkenes>alkanes [140].
Although soot may be formed in the flame, it is not necessary that soot is
emitted from the flame. Soot typically originates on the fuel side of a diffusion
flame and has to pass through the oxidation region at some point during the flow.
If the soot if fully oxidized in this region, no soot is emitted. If the soot is not fully
oxidized, it is emitted from the flame and may be visible as a streak. Laminar jet
diffusion flames typically have a pointed shape when the soot is completely oxidized
by the flame. The formation of soot, however, causes ?wings? to form, from which a
streak of soot emanates (refer to Fig. 4 in [141]). This is illustrated in Figure 2.12.
The primary factors controlling sooting are the nature of the fuel (number
of C?C bonds, C/H ratio etc. which are dependent on the whether the fuel is an
aromatic, alkyne etc.), diluents in the fuel stream, equivalence ratio (controlled by
mixing), local temperature, flow conditions such as turbulence, ambient pressure,
gravitational environment, the flame curvature and strain rate, and acoustic exci-
tations, [130, 142?149]. These factors also affect the structure of soot formed in
different regions of the flame [150]. Of course, these factors have multiple interde-
33
Figure 2.12: Schematic showing soot formation and destruction zones in a laminar
jet diffusion flame. Figure from Turns [17].
pendencies. For instance, fuel structure affects sooting tendencies in nonpremixed
flames, but has little effect in fully premixed flames since ?in premixed flames there
is a competition between the rate of pyrolysis to form the precursors and the rate
of oxidative attack; thus, since the oxidation rate increases faster with temperatures
than the pyrolysis rate, the higher the flame temperature, the lower is the tendency
to soot? [151]. In nonpremixed flames, however, soot inception begins around 1300?
1400 K and soot volume fraction depends on temperature and initial structure of
the fuel molecule [130].
Soot inception and growth in flames depends on flame structure [152] and
chemical kinetics which control both formation and oxidation [153, 154]. Processes
responsible for the conversion of gas-phase species to the condensed phase are con-
trolled by the chemical kinetic behavior of precursor species such as polycyclic aro-
matic hydrocarbons (PAHs) and aromers [18,155]. A schematic of soot formation in
different regions of a flame is shown in Figure 2.13. Johansson et al. further studied
34
Figure 2.13: Figure showing the evolution of soot structure in different regions of a
flame. Figure from [18].
the processes involved in the conversion of gas-phase PAHs to particle formation and
growth [19]. The stability of the formed soot was found to be from resonance stabi-
lization of the graphitic layers in the multi-layered structure of soot. The different
stages of soot inception, graphitization and growth are shown in Figure 2.14.
2.4.1 Soot Diagnostics
Measuring parameters such as soot number density, volume fraction and size
distribution are important to understand kinetics in flames, radiation in fires and
emissions. A classification of the different techniques is shown in Figure 2.15,
adapted from [20]. In-situ experimental techniques to measure such characteris-
tics may involve thermocouples and [156], dilution probes [157], thermophoretic
sampling [158?161], and deposition based techniques [20]. Transmission Electron
Microscopy (TEM) is used to measure soot-aggregate size distribution from the
35
Figure 2.14: Schematic showing the different stages of soot inception and formation,
beginning from PAHs to particle growth. Figure from [19].
samples, which may be collected in-situ and or in the post-flame regions [162,163].
Non-intrusive optical techniques involve either digital cameras [164,165] to perform
pyrometry [166,167], or laser-based diagnostics such as Laser-Induced Incandescence
(LII) [168, 169]. Optical diagnostics are more suitable when physical probing can
interfere with the physical or chemical processes in the flame. They may be used
to measure soot temperature, volume fraction and size distribution and are used in
laminar or turbulent flames [170], premixed [171] or nonpremixed flames [172].
36
Figure 2.15: Soot measurement techniques, adapted from [20].
The measurement of particulate matter emitted by flames in the post-flame
region uses a combination of different techniques including gravimetric approaches,
scattering and absorption [173]. A review of the different techniques are presented
in [174]. Flame Ionization Detection (FID) is a technique used to detect both soot
and unburned hydrocarbons [175]. Atmospheric particulate matter measurements
typically use a gravimetric approach to quantify the mass of PM emitted over a
stipulated time period and are deployed over large geographic areas [176] and during
events such as wildfires [177]. Measurement of PM specific to soot emissions from
engines require sensors to be portable and typically use optical techniques such as
infrared scattering since understanding size distribution in the exhaust is important
[178,179].
For the work considered in this dissertation, measurement of particulate mat-
ter is performed in the post-flame region using a laser scattering technique. Particle
size distributions are determined based on Mie theory [180]. The TSI DustTrak
37
Figure 2.16: Schematic of soot measurement in the TSI DustTrak, adapted from [21].
instrument [27] used in this work measures the mass concentration of PM-2.5 par-
ticles and estimates a particle size distribution based on a calibration aerosol mix
called Arizona Road Dust [21]. A schematic of the aerosol measurement by the TSI
DustTrak is shown in Figure 2.16. A separate gravimetric calibration was performed
to convert adjust for the difference in density between the calibration aerosol and
soot emitted from the fires. Only Total Particulate Matter (TPM) measurements
are reported since measurements of size distributions of the emitted soot are not
available.
38
2.5 Current Understanding of the Blue Whirl
The blue whirl (Figure 2.17) is a flame first observed when studying fire whirls
in a fixed-frame setup with natural air entrainment. The blue whirl evolved from
fire whirls formed with liquid fuels floating over a water surface, first reported by
Xiao et al. [64]. Figure 2.18 is a sequence of images showing intermediate states that
are visible in the transition to a blue whirl. Although the blue whirl was discovered
to transition from fire whirls formed over a water surface, experiments with other
surfaces showed that the ?presence of the water itself was not essential, but the
presence of a smooth boundary is critical? [66]. This unimpeded surface boundary
permits the formation of an Ekman layer, thought to be necessary for transition to
the blue whirl.
Figure 2.17: (A) Image of a blue whirl formed using iso-octane over a flat metallic
plate. (B) The recirculation zone within the blue whirl is visualized by incandescence
of soot particles entrapped in the recirculation zone. This image was captured with
an exposure time of 1/25 s, allowing sufficient time for steaks to develop in the axial
direction. Image from Hariharan et al. [22].
39
Figure 2.18: Sequence of images showing the region immediately above the fuel pool
during different intermediary states before formation of the blue whirl. Figure from
Coenen et al. [23].
During the transition from a fire whirl to a blue whirl, the presence of a recircu-
lation zone within the lower blue conical region is visible due to soot incandescence
(Figure 2.17 B). This internal flow structure is visible only during the transition
due to the temporary presence of soot, but it is expected that the recirculation
zone exists after stable blue whirl formation [66]. This recirculation zone was pos-
tulated to be the result of the onset of a bubble mode of vortex breakdown in a fire
whirl [65,181]. The transition process from a fire whirl to a blue whirl occurs in this
order ? onset of vortex breakdown in the fire whirl, appearance of the recirculation
bubble, lifting of the base of the flame to form the bright blue ring, formation of
the lower blue cone, disappearance of soot in the flame, and finally stable blue whirl
formation (see Fig. 10 in [66]). The bright blue ring was termed as the vortex rim
in [22], and it was suggested in [65] that it may be similar to a vortex ring, although
these have not yet been confirmed.
Hariharan et al. [22] obtained a thermal map of the blue whirl using thin-
filament pyrometry. The map showed a wide region of high temperatures above the
bright blue ring, and relatively lower temperatures below it. Peak temperatures,
40
Figure 2.19: (A) Thermocouple measurements in the blue whirl, showing peak
temperatures around 2000 K. (B) Temperature map obtained by thin-filament py-
roemtry, for a blue whirl formed using iso-octane formed over a water surface. Fig-
ures adapted from [22].
obtained from thermocouple measurements, were found to be around 2000 K in
the region above the ring, termed the purple haze. The temperature measurements
are shown in Figure 2.19. Formation of the blue whirl over surfaces other than
water, such as a flat metallic plate and a porous surface was shown in a follow-up
study [66], which also showed qualitative similarity in the temperature maps. Fur-
ther, images of OH* chemiluminescence in the blue whirl (UV, 310 nm) showed
significant intensity in the bright blue ring, and minimal intensity in the regions
below and above it, which suggested that a significant fraction of the reaction oc-
curred there, which also supports the thermal structure obtained from temperature
measurements. Obtaining a map of equivalence ratio distribution in the blue whirl
using chemiluminescence measurements is discussed later in Chapter 6.
The relationship between the combination of circulation (?), heat-release rate
(Q?) and the dimensions of the enclosure within which the blue whirl formed was
41
Figure 2.20: (A) The effect of gap size, S, and fuel flow rate, V? , on the fire whirl
regime formed. (B) Relationship between nondimensional circulation and heat-
release rate showing transition limits to the blue whirl. The subscript ?D? refers to
the reference length scale, which was chosen as the enclosure diameter for nondi-
mensionalization. Figures from [24].
explored by Hu et al. [24]. This work showed that the ? and Q? resulting in blue
whirl formation (Figure 2.20) was close to the extinction limit previously defined
for fire whirls by Lei et al. [92]. Although apparatus diameter or gap size are
not intrinsic properties affecting fire whirl behavior, these have been used in the
literature for experiments on fire whirls formed in both natural and forced air-
entrainment configurations [5, 11, 12, 54]. Following this approach, nondimensional
circulation (??D) and heat-release rate (Q?
?
D) were defined based on the enclosure
diameter, D. The relationship between these quantities is shown in Figure 2.20 B,
which shows that the blue whirl transition limit depends on the ratio of gap size, S,
to D.
42
2.6 Chemiluminescence and Planar Laser-Induced Fluorescence
The detection and measurement of species important to combustion processes
is an important diagnostic method to understand the behavior of flames and com-
bustion devices. A multitude of techniques exist to non-intrusively measure vari-
ous properties such as temperature, flow and chemistry, in a combustion process.
Optical methods are the primary non-intrusive technique, including Raman scat-
tering for major species and temperature [182], and Laser-Induced Fluorescence
Spectroscopy (LIFS) for minor species [183]. While Raman spectroscopy and ab-
sorption methods are also used for detecting and measuring concentrations of major
species [25, 184], techniques related to emission spectroscopy are used in this work.
The nature of emission and absorption by the species are governed by quantum
mechanics [25], and the kinetics governing the formation and destruction of these
radicals in hydrocarbon-air reactions has been studied previously [185?187], and are
used in numerical evaluation of local flame properties [188].
The important species for optical characterization of flames are OH, CH, CO2
and C2 [189], and the spontaneous emission by de-excitation of these radicals is
termed as chemiluminescence. Excited radicals are denoted as OH*, CH*, C2*,
CO2*, which contribute the most to visible and UV emissions in hydrocarbon-air
flames [187,189]. The concentration of these radicals and the corresponding intensity
of emission are a function of pressure [190], local equivalence ratio, strain rate [191],
nature of the fuel, etc. [32,192]. The relative intensity of chemiluminescence signals
may be used to determine local flame properties (in non-sooting flames [182]) by
43
Figure 2.21: Schematic showing an experimental setup to perform PLIF. Figure
from Hanson et al. [25].
evaluating the ratio of the relative emission intensities of the different radicals [32,
187,190,192?195].
The other important technique is induced fluorescence, of which the most
commonly used configuration is Laser Induced Fluorescence (LIF). Here, the radicals
are excited by a laser at a particular frequency, and the emission from the excited
state is observed using a camera, which is similar to observations of spontaneous
chemiluminescence. Compared to Raman scattering, the signal levels are higher
in LIF, and species measurement are not affected much by background noise from
other species [25]. When a 2-D laser sheet is used to excite the flame, a camera
is positioned 90? to the sheet to detect signals, the technique is called Planar LIF
(PLIF). Figure 2.21 shows a typical experimental setup to perform PLIF. The pulse
(or repetition) rate of the laser can be on the order of kHz, and PLIF is thus
used in the diagnostics of complex combustion environments such as high speed
flow [196, 197], turbulence [67, 198]. It is often used in engines and turbines since
it can often be coupled with other diagnostics such as Particle Image Velocimetry
44
(PIV) and/or temperature measurements [59,70,199?202], along with the ability to
image multiple species simultaneously.
45
Chapter 3
Measurement of Particulate-Matter Emissions from
Liquid-Fueled Fires
3.1 Introduction
Marine activities such as oil exploration and extraction can cause accidental
oil spills into inshore or offshore waters. Oil spills can pose a serious threat to sur-
rounding populations [78,203], response workers [77], and ecosystems [120,204,205].
Additional effects include substantial disruption to local oil exploration infrastruc-
ture, marine transportation, and potentially widespread economic impacts [206].
Rapidly advancing oil recovery technology as well as increased interest in local pro-
duction continues to increase the probability of oil spills? occurrence [207]. In the
case of such a spill, an efficient, effective, and robust treatment technique is crucial
to mitigate their impact. In this regard, in situ burning (ISB) is one of the most
reliable and effective oil spill treatment techniques [74,125,208], particularly in con-
46
ditions where remoteness, harshness of the climate, and scale of the incident make
it impossible for mechanical recovery or dispersant techniques to be deployed and
utilized immediately after the spill.
Decades of laboratory and meso-scale testing of in-situ combustion of oil have
shown that the composition and concentration of emitted emissions from ISBs is
an acceptable trade-off in relation to inshore and offshore contamination and its
environmental consequences and cleanup costs [75,208]. In fact, efficient ISBs have
been shown to effectively eliminate at least 90% of the released liquid oil in some
cases [209?211], and is a relatively fast and portable method of treating oil spills.
This is especially important since 70% of the untreated, spilled oil can emulsify
during the first 24 hours [125].
Despite their effectiveness, ISB techniques are still challenged by the airborne
emissions they release (especially near shorelines), the degree of weathering and
emulsification which decreases fuel flammability, and the intensity of ambient wind
and waves. In general, pool fire burns over water are oxygen-starved [206,212]. This
leads to the emission of black soot (particulate matter, PM) that is visible from as
dense smoke emanating from ISBs. In fact, one of the major operational limits
of current ISB practices is the after-burn emission concentrations, particularly at
the downwind locations that may lie close to populated areas. For instance, the
maximum 1-hour averaged concentration of PM-10 (PM of ten ?m or lower in di-
ameter) must not exceed 150 ?g/m3, according to ISB guidance from the United
States National Response Team, Science and Technology Committee [206]. Reduc-
tion of PM emission and high-efficiency burning requires the fire to reach ample
47
oxygen concentrations, high temperatures over the fuel surface and completeness of
combustion [208].
One potential method of enhancing these parameters is to induce swirl to
the air entrained to the fire, causing the formation of a fire whirl (FW). FWs are
structures which arise when the right conditions of wind and fire interact, and are
typically studied in the context of fire safety during wildland and urban fires [5].
The FW structure results in intensification of combustion, imposing significantly
higher heat feedback to the fuel surface, increasing the rate of burning, potentially
reducing airborne PM, and decreasing post-combustion residuals (on the water sur-
face). It is thought that high efficiency combustion may occur through higher of
oxygen availability and increased temperatures within the fire, ultimately reducing
the emission of PM from the plume. There are open questions, however regarding
this hypothesis. For instance, one important mechanisms of reducing emissions is
ensuring sufficient mixing, which is typically achieved by increasing turbulence in-
tensity. Turbulence, however, is suppressed in FWs [8, 95]. Thus, the hypothesis
that is tested is: the burning of a pool of liquid hydrocarbon fuel in a FW structure
emits lower quantities of PM as compared to burning as a PF.
3.1.1 Outline
This chapter details experimental investigations of emissions from pool fires
(PF) and fire whirls (FW) at various scales, conducted at the University of Mary-
land (UMD) and Worcester Polytechnic Institute (WPI). To demonstrate the ef-
48
fectiveness of using FWs over pool fires, experiments were conducted using two
fuels, n-heptane and Alaska North Slope (ANS) crude oil. For all experiments,
laboratory-grade heptane from Aqua Solutions Inc. or Fisher Scientific was used,
and ANS crude oil was supplied by the Bureau of Safety and Environmental En-
forcement (BSEE). Experiments were performed at four different fuel-pool diame-
ters (D = [10, 20, 30, 70] cm). Experiments at D = [10, 20, 30] cm were performed at
UMD, and those at D = 70 cm were performed at WPI. PFs and FWs are compared
on the basis of burning rate (m?), post-burn residual weight, emissions of PM, CO,
CO2, O2 consumption, flame temperature and heat feedback (q?
??) to the fuel pool.
The general experimental methods used for all experiments at all scales are
described in section 3.2. The experimental data are presented in detail in section
3.3. An analysis of the emission factors and preliminary explanation of the factors
controlling emissions from the different fires is presented in section 3.4, along with
a summary in section 3.5. A detailed analysis of the mechanisms and the role of
circulation and buoyancy are discussed in Chapter 4.
49
3.2 Experimental Methods
3.2.1 General Apparatus
A fixed-frame, four-wall configuration was used to generate FWs at all scales.
The four walls were positioned to form an enclosure, and the fuel pool was placed at
the center of the enclosure over the bottom surface. This four-walled setup permits
natural entrainment of air to the fire through the four gaps (see Figure 3.1), forming
an on-source, quasi-steady FW [5]. These are a stable and stationary form of FWs,
providing a good platform to compare them with PFs. Four-walled setups are also
simple to construct and allow flexibility in choosing the enclosure side (S) and gap
width (W ). For experiments at all scales, reliable and steady FW formation was
observed when the W ratio was kept constant at 1 . The dimensions of the enclosures
S 4
used at different scales in this study are tabulated in Table 3.1. To generate PFs,
the walls were removed to allow radially-symmetric air entrainment to the fire.
A circular dish made of aluminum was used to contain the fuel pool. The
dish was used to hold a water sublayer above which the fuel slick floated, mimicking
open-water conditions at a lab scale. This also eliminated any warping in the metal
dish due to high temperatures during burning of the fuel pool. The quantity of fuel
used for experiments at each scale were determined based on the thickness of the
resulting fuel slick. Since a minimum slick thickness around 2? 2.5 mm is required
to support combustion, a thickness of 5 mm was decided to allow sufficient time
for fire whirl development and adequate data from the various measurements. For
50
Figure 3.1: Schematics of the experimental configuration used to form FWs at vari-
ous scales. The ratio of gap width (W ) to enclosure side length (S) was maintained
at W/S = 1/4 for experiments at all scales. Schematic in panel (B) from Tohidi et
al. [5].
experiments at the 70 cm scale, some experiments were performed with both 5 and
7 mm slicks of ANS crude oil to study their effect on boilover. This is discussed in
detail in section 3.3.3 and Chapter 4.
The depth of the dish was dependent on the pool diameter, D. The values
of D, dish depth, fuel volume, fuel-slick thickness for the different experiments are
shown in Table 3.1. For a given experimental condition (defined by D, fuel type
and flame regime), the slick thicknesses were the same for both fuels to maintain
consistency. Similarly, all test conditions were kept consistent for both PF and FW
regimes.
Two liquid fuels, n-heptane and ANS crude oil, were used for experiments.
Experiments were performed at each scale with both fuels to allow for comparisons
and scaling. N-heptane (C7H16) is a single-component liquid hydrocarbon with a
low flash point (?4 ?C) and was chosen due to its steady burning rate in both the
pool fire and fire whirl regimes. Using this fuel eliminates the effect of time-varying
51
Table 3.1: Experimental conditions for pool fires and fire whirls formed using hep-
tane and ANS crude oil.
Pool Gap Enclosure Enclosure Fuel Slick Fuel dish
diameter, D width, W side, S height, H volume thickness depth
[cm] [cm] [cm] [cm] [ml] [mm] [cm]
10 15 60 60 40 5 10
20 25 100 100 160 5 5
30 25 100 100 350 5 20
2000 5
70 45 175 240 70
3000 7
fuel composition on the burning rate and aid in identifying general trends to explain
the behavior of more complex fuels like crude oil. ANS crude oil was used to mimic
a realistic spill. Multi-component fuels such as ANS, with their constituent species
having a wide range of flash points, provide an opportunity to study the influence
of a fire whirl on complex fuels and their residue after extinction of the fire. Table
3.2 shows important physical properties of the fuels used.
At D = [10, 20, 30] cm, the fuel dish was placed over a load cell, which was used
to measure the instantaneous mass of the fuel and determine the burning rate (m?).
The heat-release rate (HRR, Q?) is determined by m?. At D = 70 cm, however, the
excessive weight of the fuel-water container, the use of a load cell was not possible,
and thus m? was not directly measured. Two alternate methods were used instead to
measure Q?: (i) peak and instantaneous Q? was measured using oxygen-consumption
calorimetry; and (ii) the average Q? was measured using the average value of m?,
52
Table 3.2: Physical properties of the fuels discussed in this work. Data obtained
from Material Safety Data Sheet (MSDS) for each fuel. Crude oil properties are a
strong function of the level of evaporation and composition, resulting in significant
variation in the reported properties.
Density Flash point Boiling point
Fuel
[kg/m3] [?C] [?C]
n-Heptane 684 ?4 98
ANS Crude Oil 866 25? 35 >38
West Texas Intermediate 867 32 >35
Diesel 830 50 120? 350
determined as the ratio of fuel mass consumed by the fire and the burning duration
(tb).
A typical experiment involved the following steps:
(i) addition of water into the glass dish,
(ii) addition of of fuel (n-heptane or ANS) over the water surface,
(iii) beginning data logging using NI LabView, TSI TrakPro and/or BalanceLink,
(iv) ignition of the fuel using a butane torch,
(v) stopping data logging upon extinction. Residue was collected and weighed for
experiments with ANS crude oil.
All experiments were performed a minimum of three times to assess variability in
the measured quantities.
53
3.2.2 Exhaust Duct and Emissions Sampling
The exhaust from the fires were channeled through a duct (28 cm in diameter)
to facilitate sampling and measurement of the different gas species in the combustion
products. The entire fire whirl apparatus was placed under a suction hood as shown
in Figure 3.2. A fiberglass curtain was draped along the perimeter of the hood to
ensure all combustion products were collected. Gases were sampled from the duct
at a downstream location greater than 10 diameters from the entrance to the duct
to ensure for sufficient mixing.
The exhaust duct was instrumented with thermocouples (TCs) and a Verabar
V100 differential pressure transducer. The volumetric flow rate through the duct
was estimated to be 0.114 m3/s, based on a 0.5 V operating voltage in the differential
pressure sensor. A maximum flow rate of 0.8 m3/s could be achieved at ?3.5 V, with
a roughly linear dependence of flow rate on voltage. The TCs were positioned 5 cm,
10 cm and 14 cm from the wall of the duct. During operation, the maximum spread
of temperature between the three TCs was less than 2 K, justifying the assumption
of constant temperature and well-mixedness across the duct.
Two sampling tubes were used in the suction duct, each with 17 holes, each
2 mm in diameter, along the length of the tube. This ensured sufficient sampling
from across the diameter of the exhaust duct. In both sampling tubes, the sampling
holes were pointed in the downstream direction of the flow within the duct to avoid
clogging of the holes by soot. Combustion products encountered the differential flow
meter, thermocouples, the sampling tube for PM, and sampling tube for gases, in
54
Figure 3.2: Schematic showing suction hood and exhaust duct positioned above
the fire whirl apparatus. Flow rate in the duct was determined using temperature
measurements and a differential pressure sensor. Two sampling tubes were used,
one for gaseous species and one for particulate matter. Inlet velocity into the fire
whirl enclosure was measured using either a hot-wire or vane anemometer. Mass of
the fuel dish was measured using a Mettler-Toledo load cell.
that order. Due to the close proximity of the the two sampling tubes, they were
offset by 45? (around the axis of the duct) to prevent interference in sampling. An
image of the cross-section of the exhaust duct showing the described arrangement
is shown in Figure 3.3.
The two sampling tubes were used for measuring PM and gas species. The
sampling tube for PM was positioned upstream of the other. The upstream sam-
pling tube was connected directly to a DustTrak 8534 with the manufacturer-
recommended high-temperature rubber tubing. Exhaust from the duct was sampled
at a rate of 3 L/min using the internal pump in the instrument. In order to use
55
Figure 3.3: Exhaust-gas sampling tubes, pressure and temperature sensors posi-
tioned inside the exhaust duct. Exhaust flow direction is towards the reader.
the device on particulate matter (PM) from crude oil and heptane, a gravimetric
calibration of the DustTrak was performed.
The downstream sampling tube was connected to a California Instruments
CAI ZPA NDIR/O2 Analyzer. Sampling was enabled by a double-head vacuum
pump connected upstream of the gas analyzer to provide the required flow rate. A
particulate filter was also connected upstream to the vacuum pump to prevent any
clogs in the pump or gas sensor. The instrument contains IR sensors to measure the
CO and CO2 concentrations on one flow channel and a paramagnetic O2 sensor on
a second channel. CO2, CO and O2 were measured in the range 0?10%, 0?1%, and
0?25%, respectively. The DustTrak 8534 and CAI ZPA Analyzer used are shown in
Figure 3.4. For experiments conducted at WPI, the same arrangement was used,
with the DustTrak for measuring particulates and a Spectris Servomex 4200 gas
analyzer. The operating range of the Servomex and CAI gas analyzers was the
same for each of the gas species (CO2, CO, O2).
A two-point reference calibration for CO, CO2 and O2 using the gas analyzer
56
Figure 3.4: Images of (A) CAI ZPA NDIR/O2 Analyzer [26], and (B) DustTrak
DRX Aerosol Monitor 8534 [27].
was performed to determine the response of each sensor before each experiment.
Zero calibration was performed by supplying the sensors with 100% N2 and span
calibration was performed by using a gas combination of 1.007% CO, 7.855% CO2
and 20.01% O2, with the remainder composed of N2. The gas sensors were all
calibrated with their respective calibration gases on each day that experiments were
performed. Experiments conducted at WPI used a similar suite of gas analyzers,
while the same DustTrak was used to measure PM.
The measured soot mass and that measured by the DustTrak was compared
to determine the appropriate settings for future experiments. The instrument was
factory calibrated using Arizona Test Dust [21], and the baseline mass concentrations
calculated by the instrument were based on this density. To adjust for the density
variation between the test dust and the soot in our experiments, a gravimetric
calibration was performed. In parallel to the DustTrak, a separate pump sampling
system was operated at the same flow rate, and the exhaust gases were channeled
through a fine particulate filter, which collects all the PM greater than 0.8 ?m. The
57
filter elements were weighed on a high-precision mass balance before the experiment,
and the DustTrak and parallel pump system operated for the same amount of time.
Post experiment, the filter element was again weighed to determine the mass of
soot deposited in the test duration. A density adjustment was made according to
Equation 3.1 [213]. Once this correction factor is determined, the values were either
programmed into the instrument directly or applied during data processing.
[ ]
mg Filter post[-weig]ht? Filter pre-weight [mg]Concentration = (3.1)m3 2 ? Sampling rate [L/s] ? Sampling duration [s]
3 1000 L/m3
The instrument uses an inbuilt photometric calibration that separates the size
fractions of soot, classifying them as PM-1, PM-2.5, PM-4, PM-10, and Total PM.
The fundamental measurement in the instrument is based on PM-2.5, and the other
fractions are determined based on the size distribution in Arizona Road Dust. This
photometric calibration of size distribution is not directly applicable to soot since it
depends on fuel characteristics. In this work, however, the size calibration had little
effect since the fundamental measurement (PM-2.5) and the Total PM (TPM) were
nearly identical.
An E-instruments E-8500 multi-gas analyzer was used to try to measure other
gaseous species such as NOx, SOx and unburned hydrocarbons (UHC or CxHy) in
experiments with D = 10 cm. However, even when the instrument was used to probe
directly in the flame, the concentration of these species in the plume was lower than
the sensitivity of the instrument, and it was thus difficult to distinguish between
58
actual signals and electronic noise. These results are not presented in this work,
and the measurement of these species is suggested as future work. The emission
of SOx is especially important when burning crude oil. Based on the temperature
measurements, substantial emissions of NOx is not expected.
Regardless, the potential presence of these species in the combustion products
does not affect measurements of CO, CO2 and PM, which were reliably detected
and measured.
3.2.3 Temperature Measurements
3.2.3.1 Gas-Phase Temperature
Flame temperature was measured for each fire. For D = [10, 20, 30] cm, five K-
type thermocouples (TC, sheathed and grounded, 50 ?m bead size) were positioned
along the centerline of the fuel pool at different heights (H), such that the ratio H
D
varied between 1 and 5 (see Figure 3.5A). For D = 70 cm, a tree of 25 K-type TCs
was mounted in the middle of the pool. The first five TCs above the fuel surface
were spaced 0.5 cm apart, and the remaining were separated by 6 cm along the tree.
A schematic of the arrangement is shown in Figure 3.5B. Radiation correction was
not performed for any measurements since measured temperatures were generally
below 1300 K, and the flames were unsteady. Such flames or fires typically require
a two-TC approach to estimate correction [214].
59
Figure 3.5: (A) Schematic showing locations of TCs above the fuel pool, used for
experiments with D = [10, 20, 30] cm. (B) Schematic showing locations of TCs for
gas- and liquid-phase temperature measurements for experiments with D = 70 cm.
3.2.3.2 Liquid-Phase Temperature
Another important parameter in PFs and FWs is heat-flux feedback to the
fuel pool. The fraction of heat release that is fed back from the flame to the fuel
surface is a critical parameter influencing energy balance in the fuel layer and is
dependent on flame dynamics and geometry. Since a conventional Schmidt-Boelter
gauge could not be positioned within the fuel dishes with D = [10, 20, 30] cm, the
heat flux was calculated as the average rate of heating of the water sublayer. Prior
to each experiment, the amount of water in the fuel dish was measured, and four
TCs were positioned in the water sublayer (see Figure 3.19). Data from the TCs in
the water layer were used to determine the total heat-flux feedback to the pool.
For experiments at D = 70 cm, a water-cooled heat flux gauge was installed
at different locations in the fuel pool to measure heat feedback from the flame to
60
Figure 3.6: TCs for liquid-phase temperature measurements for D = 10, 20, 30 cm.
(A) TCs to check for boiling at the fuel-water interface, without the stirrer. (B) TCs
to check for effect of the stirrer on temperature stratification in the water sublayer.
the fuel.
Small pools with ANS burning were prone to boilover towards the end of
the experiments. Based on previous work on boilover [215], this was attributed to
temperature stratification in the water sublayer and near-boiling temperatures at the
water-fuel interface. To check this, one TC was placed at the fuel-water interface
(3.6A). Results from this experiment are shown in Figure 3.19. One method of
avoiding boilover is ensuring that the water sublayer heats up uniformly. To do
this, a magnetic stirrer was introduced at the bottom of the fuel dish to mix the
sublayer and prevent water at the interface from reaching boiling point. To check
for stratification after positioning of the TCs, all four TCs were positioned within
the sublayer (Figure 3.6B).
A stirrer could not be employed at D = 70 cm, and boilover could not be
avoided. To delay the onset of boilover, however, a pump was used to recirculate
water between the water-fuel interface and the bottom of the dish.
61
3.3 Results
3.3.1 Physical Characteristics of PFs and FWs
For each experiment, photographs of PFs and FWs were obtained using a
digital camera. Images of the fires at difference scales are shown in Figures 3.7 and
3.8. For PFs, the average flame height, Hf ? [1, 2]D. At the base of PFs, the flame
width, wf , is equal to the pool diameter, D. PFs have a conical flame shape as wf
reduces with axial distance. Entrainment of air along the boundary layer toward
a PF causes necking of the flame at the base and the initiation of Rayleigh-Taylor
instabilities [112]. These instabilities cause rolling-up of the flame sheet (see Figure
3.8A) and a periodic fluctuation of the flame height [14], a phenomenon referred to
as ?puffing?. In contrast to PFs, Hf ? [3, 6]D for FWs, which have a cylindrical
shape. FWs also show stronger necking at the base of the flame, resulting in reduced
average flame width, wf ? [0.5, 1)D. FWs do not show puffing, which is replaced
by a helical instability that causes the flame to wrap around a vortex core [23,102].
3.3.2 Burning Rate
The fuel mass-loss behavior for fires at D = [10, 20, 30] cm is presented in
Figures 3.9 and 3.10. For experiments with heptane fuel, the total mass loss was
the same for both PFs and FWs as no residue was left behind after extinction. For
all conditions considered here, m? of the FW regime is higher than the PF regime.
Generally, mass loss is linear for heptane fires, but the rate declines with time for
62
Figure 3.7: Photographs of PFs (panels A ? D) and FWs (panels E ? H) formed
using heptane at D = [10, 20, 30, 70] cm.
63
Figure 3.8: Photographs of PFs (panels A ? C) and FWs (panels D ? F ) formed
using ANS crude oil at D = [10, 20, 70] cm.
64
ANS fires. This non-linear behavior of ANS crude oil is seen at smaller scales (see
Figure 3.9). The physical characteristics of the different fires are tabulated in Table
3.3.
A linear fit is used to determine the average m? (in [g/s]) from the slope of the
line. For the purposes of comparison, these are shown in Figure 3.9 D and Figure
3.10 C& F. For both heptane and ANS crude oil, the FW regime consumes fuel at
approximately twice the rate of the corresponding PF. For heptane fires, the entire
quantity of initial fuel was consumed, while there was a residue left behind after
extinction of ANS crude oil fires.
For experiments at D = 70 cm, since a load-cell could not be installed beneath
the large fuel container, only an overall average of m? could be determined based on
the burn time of each fire and the fuel mass consumed. For a 5 mm slick of fuel,
the mass of heptane and ANS for each experiment was ? 1250 g and ? 1710 g,
respectively. Since boilover was always observed for ANS at this scale, the slick was
increased to 7 mm, corresponding to an average fuel mass of 2450 g. The post-burn
residual ANS crude oil was collected with adsorbent pads (3M Petroleum Sorbent
Static Resistant Pad HP-556). Table 3.3 shows the initial mass, residual weight and
average burning rate for the different fires at the 70 cm scale. While boilover was
delayed with the 7 mm slick as compared to the 5 mm slick, the intensity was visibly
higher, and is evidenced by the higher average mass-loss rate of the 7 mm slick fires.
65
Figure 3.9: Mass loss behavior of PFs and FWs formed at D = 10 cm. Data from
individual experiments using (A) heptane and (C) ANS crude oil. The marker colors
denote individual experiments. Averaged data and linear fits are shown in (B) and
(D). The slope of the linear fits represents the average burning rate.
66
Figure 3.10: Burning rate for PFs and FWs formed with D = [20, 30] cm. The red
and black markers in each panel represent the average of three experiments. Overall
burning rate is shown in panel (C) for heptane fires and in panel (F) for ANS fires.
compared for all medium-scale heptane fires.
67
Table 3.3: Tabulated data of average values of burning duration, burning rate,
and estimations for flame width (wf ) and flame height (Hf ) for the different flame
regimes and pool diameters. Experimental variability for m? and residue mass are
presented in Appendix B.
Flame D tb m? Residue Approx. Approx.
Fuel
Regime [cm] [s] [g/s] [g] wf Hf
10 182 0.00015
20 249 0.00044
PF - 1D (1?1.5)D
30 239 0.001
Heptane 70 165 0.0083
10 91 0.0003
20 100 0.0011
FW - ?0.5D (3.5?6)D
30 133 0.0018
70 137 0.01
10 291 0.00005
20 278 0.00024 -
PF 30 257 0.00067 1D (1?2)D
(5 mm) 70 180 0.0094 301
ANS (7 mm) 70 169 0.01 382
10 161 0.00011
20 167 0.00046 -
FW 30 206 0.001 (0.5?1)D (3?4)D
(5 mm) 70 161 0.0106 290
(7 mm) 70 167 0.013 382
68
3.3.3 Emissions: Particulate matter
Time-varying PM concentration was obtained directly from the DustTrak,
after calibration using Equation 3.1. This value was then converted to a PM emission
rate according to Equation 3.2.
[ ] [ ]
mg mg
PM emission rate = Measured concentration
s m3 [ ]
? m
3
Exhaust flow rate (3.2)
s
The variation of PM-emission rate with time for the different fires at D =
[10, 20, 30] cm is shown in Figures 3.11?3.13. Consistent with the behavior of m?
for PFs and FWs, the duration of PM emission from FWs is roughly half that of
PFs. For D = 10 cm, the peak emission rate is similar for PFs and FWs, and the
steep spikes in PM-emission rate due to boilover are visible in Figure 3.11 C, D. For
D = 10 cm, a few experiments are shown with boilover only to show the effectiveness
of stirring in the water sublayer. At D = [10, 20, 30] cm, experiments that resulted
in boilover were not used to calculate emission factors. The PM-emission rate for
D = [20, 30] cm is similar to those at the D = 10 cm, with some difference in the
peak emission rate between PFs and FWs at each scale. Comparing the different
fuels, for each experimental condition, the PM-emission rate from ANS fires is an
order of magnitude higher than that from heptane fires. Data for D = 70 cm is
discussed in detail in Chapter 4.
69
Figure 3.11: Variation of PM-emission rate with time at D = 10 cm. The spikes in
(C) and (D) are instances of boilover towards the end of the burn.
70
Figure 3.12: Variation of PM-emission rate with time at D = 20 cm for PFs and
FWs formed using heptane (A, B) and ANS crude oil (C, D). The different colors
represent data from individual experiments.
71
Figure 3.13: Variation of PM emissions rate with time at D = 30 cm for PFs and
FWs formed using heptane (A, B) and ANS crude oil (C, D).
72
For the 70 cm scale, the emission rate of fires formed using a 5 mm slick of
ANS crude oil are shown in Figure 3.14. For both PFs and FWs, the PM rate
increases linearly initially, but a sharp spike occurs after about 85 s in PFs and 70
s in FWs. This rise in PM rate is due to boilover. To delay the onset of boilover,
a 7 mm slick of ANS was used. Results for fires formed using the 7 mm ANS slick
and 5 mm heptane slick are presented in Chapter 4.
The total mass of PM emitted in each case was calculated according to Equa-
tion 3.3. This value is used to calculate an emission factor, described in section
3.4.
[ ] [ ] ?
m3 t
[ ] [ ]
b mg
TPM mg = V? TPM dt s (3.3)
s 0 m
3
Figure 3.14: PM emissions rate from PFs and FWs formed using a 5 mm slick of
ANS crude oil at D = 70 cm.
73
3.3.4 Emissions: Gaseous species
The gaseous species measured were CO, CO2 and O2. O2 measurements were
also used to estimate the instantaneous heat-release rate of each fire. Data and
figures for D = 70 cm are presented in this section for brevity, and the figures for
D = [10, 20, 30] cm are included in Appendix B.
The emission of CO2 is directly proportional to the consumption of O2. The
emission rate of CO2 and consumption rate of O2 with time for 70 cm fires are
shown in Figure 3.17 and Figure 3.15. For both fuels, peak CO2 emission and O2
consumption are higher for the fire whirl. Consequently, the peak Q? of FWs is also
higher than that of PFs (Figure 3.16), which is also evident from the higher m? of
FWs (Table 3.3). By assuming ?hc = 44.9 kJ/g for heptane [17], the average Q?
derived from m? was found to be 0.373 MW for PFs and 0.45 MW for FWs. The
peak Q? obtained from O2-consumption calorimetry for these fires was around 0.3
MW and 0.45 MW, respectively.
Similarly, using ?hc = 42 kJ/g for ANS crude oil [216], the average Q? cal-
cualted from m? was 0.42 MW and 0.55 MW for PFs and FWs, and the peak ex-
perimental values are in a similar range. The sharp spike in CO2 production and
O2 consumption denotes the onset of boilover in ANS fires. This also indicates that
the instantaneous value of m? rises significantly during the boilover phase, leading to
the higher Q?. This behavior has been observed in previous studies on boilover [215].
The trends in CO emission rate (Figure 3.18) are similar to CO2 and O2. The CO
emission rate from ANS crude oil fires is an order magnitude higher than that from
74
heptane fires.
Figure 3.15: Variation of O2 consumption rate with time for the different fires at
D = 70 cm.
75
Figure 3.16: Variation of heat-release rate (HRR) with time for the different fires at
D = 70 cm. HRR was estimated by O2-consumption calorimetry, assuming a value
of 13.1 kJ/g-of-O2 [28, 29].
76
Figure 3.17: Variation of CO2 emission rate with time for the different fires at the
70 cm scale.
77
Figure 3.18: Variation of CO emission rate with time for different fires at the 70 cm
scale.
78
3.3.5 Heat Feedback and Temperature
The efficacy of the stirrer in the sublayer was tested by comparing the temper-
ature measurements with and without the stirrer. When the stirrer was not used, the
temperature at the water-fuel interface reached about 400 K, resulting in boilover.
This is shown in Figure 3.19A, which also shows that the temperature within the
sublayer reduces with depth. This stratification in the sublayer, however, did not
exist When the stirrer was used, and the sublayer temperature showed a linear in-
crease with time (Figure 3.19B). This shows that the stirrer was effective in avoiding
boilover.
Heat feedback to the fuel pool was obtained in two different ways depending
on D. For [D = 10, 20, 30] cm, heat flux was calculated using temperature mea-
surements of the water sublayer since a heat-flux gauge could not be positioned
within these pools. Thus, the heat-flux feedback estimated is averaged temporally
and spatially at these scales.
The total heat feedback was calculated as q = mCp,water ?T , where m is the
mass of water in the sublayer, Cp,water is the heat capacity of water, and ?T is the
total rise in temperature of the water sublayer. Since the water sublayer was well
mixed, the temperature readings from all TCs in the sublayer showed very similar
readings. Thus, ?T represents the rise in bulk temperature. Then, the averaged
heat flux is given by
q???
q
=
Apool tb
79
Figure 3.19: Effect of using the magnetic stirrer on temperature in the water sub-
layer, (A) with the stirrer showing a decrease in temperature with depth, and (B)
without the stirrer, showing minimal stratification and a single bulk temperature for
the sublayer.
where tb is the burn time and Apool is the pool area. The results for this scale are
shown in Figure 3.20. q??? increases wtih D, and at each scale, FWs exhibit higher
feedback than PFs. This behavior is consistent for both fuels, and ANS fires have
higher feedback than heptane fires. The difference in values at these scales is small,
but the averaging causes peaks in heat flux to be smeared.
At the 70 cm scale, heat-flux gauges were installed in the pool and direct
measurements from a Schmidt-Boelter gauge are shown in Figure 3.21. The heat
flux from heptane PFs tends to peak immediately after ignition, stabilize around 25
kW/m2, and then spike to about 45 kW/m2 prior to extinction. Heptane FWs, on
the other hand, have a relatively stable heat flux feedback around 60 ? 75 kW/m2
throughout the burn, peaking around 90 kW/m2. ANS pool fires show relatively
stable heat flux around 40 kW/m2, and fire whirls peak just below 120 kW/m2.
Fire temperature was measured above the fuel pool, with axial distances in
80
Figure 3.20: Temporally and spatially averaged q??? for PFs and FWs at D =
[10, 20, 30] cm, formed using (A) heptane and (B) ANS crude oil.
Figure 3.21: Variation of q??? to the center of a 70 cm pool for PFs and FWs formed
using (A) heptane and (B) ANS crude oil.
81
the range of H/D ? [1, 5] (see Figure 3.5). Measurements along the centerline
give an understanding of the qualitative differences in the heat-flux feedback to the
fuel pool. The temporal change in temperature at the different axial locations for
D = [10, 20, 30] cm is shown in Figures 3.22?3.24.
At D = 10 cm, the peak temperatures are similar for both PFs and FWs,
however, the temperature drops sharply with height in case of PFs. Temperatures
are highest at H/D = 1, and the decline in temperature with height is more gradual
in the case of fire whirls. For the 20 and 30 cm scales, the peak temperatures of
fire whirls is higher by about 200 K, but the decline in temperature with height
is still evident in the pool fires. The temperatures measured here cannot be used
to directly calculate heat-flux feedback, but they help in explaining the qualitative
differences between pool fires and fire whirls. These measurements do not represent
temperatures at the flame sheet, and have not been corrected for radiation since the
maximum value is below 1300 K.
Temperature data for D = 70 cm are shown in Appendix B.
82
Figure 3.22: Flame temperatures at different axial locations (H) above the fuel
surface, for PFs and FWs formed using heptane and crude oil, at D = 10 cm. Here,
data from a single experiment is shown so that there is no smearing of trends due
to averaging.
83
Figure 3.23: Temperature measurements for fires formed at D = 20 cm.
84
Figure 3.24: Temperature measurements for fires formed at D = 30 cm.
85
3.3.6 Air-Entrainment Velocity
The configuration used here for forming FWs is termed as a fixed-frame ap-
paratus with natural air entrainment. Air entrainment to the fire is controlled
by the buoyancy generated within the enclosure and the gap size of the inlets. Air-
entrainment (inlet) velocity measurements using either a hot-wire or vane anemome-
ter are reported here. The ambient circulation, ?, was estimated as ? = 4S U?,
where U? is the average tangential inlet velocity measured using the anemometer.
The values of U? and ? for the different cases are shown in 3.25.
Values for U? were averaged over time and between several repetitions, with
one standard deviation measurements shown as error bars in Figure 3.25 A,B. U?
was measured at only one H for FWs at D = [20, 30] cm, shown in panels C and
D. At all values of D, U? for heptane FWs was higher than that for ANS FWs.
The inlet velocity decreases with height (see panels A and E ). For D = 70 cm, a
constant gap size of W = 45 cm was used, and U? was measured using a combination
of hot-wire and vane anemometers. The average U? for heptane FWs was between
0.7 and 0.8 m/s, and showed a slight decrease with H. U? for ANS FWs decreased
from about 0.76 m/s (at H = 20 cm) to 0.45 m/s (at H = 100 cm).
86
Figure 3.25: Measurements of U? and estimates of ? for FWs formed using heptane
and ANS at different D.
87
3.4 Discussion of Results
To compare the emissions from the two flame regimes formed at different D,
the total mass of each emissions species emitted (or consumed, in the case of O2)
was calculated according to Equation 3.3. For PM, CO2, CO, the emission factor
(EF) was defined as
mass of emission species [g]
EFspecies = (3.4)
mass of fuel burned [kg]
Similarly, a consumption factor (CF) of O2 by each fire was defined as
Total O2 consumed [g]
CFO2 = (3.5)
Mass of fuel consumed [kg]
The above definitions are based on the total mass of each species and the mass
of fuel burned. A similar definition based on only the carbon content is presented
in Chapter 4. The calculation of emission factors at D = [10, 20, 30] cm does not
include experiments that showed boilover. The EFs and CF for the different fires
are presented in Figure 3.26. In general, the EFPM of the FW regime is lower than
that of the PF regime, for both fuels and all D. For both fuels and flame regimes,
EFPM reduces from D = 10 cm to D = 30 cm, then increases again at D = 70 cm.
The rise in EFPM at D = 70 cm cannot be attributed only to boilover since it is
observed even for heptane.
In the case of heptane fires, EFCO2 and CFO2 are similar for PFs and FWs.
EFCO from PFs is slightly higher for FWs. EFCO for heptane fires are on the order
88
Figure 3.26: EFs and CF (fuel-mass based) for heptane and ANS crude oil fires at
different D. Panels (A?D) show results for heptane, and panels (E?H) for ANS
crude oil fires. The black square markers denote results for PFs and the red circles
denote those for FWs.
89
of 1 g/kg-of-fuel, which is 3 orders of magnitude below that of EFO2 and EFCO2 .
While the same amount of O2 is consumed by both PFs and FWs, the total amount
of PM and CO emitted by FWs is lower. In the case of ANS crude oil, the amount
of O2 used for combustion by FWs is higher, and consequently the amount of CO2
emitted is also higher. EFCO for ANS fires are on the order of 10 g/kg-of-fuel, and
EFCO for FWs is higher than for PFs.
The difference in PM emissions behavior between PFs and FWs stems from
multiple factors including fluid dynamics of air entrainment to these fires, the global
equivalence ratio (determined by oxygen consumption), effect of D on turbulence
and mixing, flame temperature, radiation heat transfer, and the nature of the fuel.
These factors are intrinsically coupled and constitute a complex heat-feedback loop.
For instance, radial air entrainment increases oxidizer concentration, and reduces
flame standoff distance [87] causing convective heat transfer to increase due to higher
velocity gradients in the boundary layer [5]. This causes an increase in m?, Q? and
?T , increasing the buoyancy within the enclosure, in turn sustaining high values of
U?.
The above mechanism need to be studied in detail at small scales to understand
the dynamics of mixing. A general explanation of PM reduction in fire whirls may be
approached by analyzing the increase of average plume temperatures in the presence
of additional oxygen (see section 3.3.5), potentially causing partial re-oxidation of
carbon in the soot into CO. Higher temperatures potentially contribute to lower PM
emissions in two ways:
90
(i) increasing radiative feedback from soot incandescence. This is visible in Figure
3.21 where the peak heat flux feedback from fire whirls is higher than that in
pool fires. The difference is especially high in case of ANS fires which have a
higher soot concentrations than heptane fires. The higher values of q??? leads
to higher m?.
(ii) enabling partial oxidation of soot formed in the lower part of the fire whirl.
While the mechanism of the soot oxidation or mixing cannot be established
from the data presented here, it is clear that a simultaneous reduction in EFPM
and an increase in EFCO for fire whirls is the result of higher EFO2 and ?T .
These higher temperatures exist over a longer length scale, which potentially
provides sufficient time for mixing and soot oxidation.
The combinations of Q? and ? for the different FWs in this study are shown
in 3.27A. The range of Q? in the study spans three orders of magnitude, and q?
increases linearly with Q? (Figure 3.27B). Parameters such as m? and q??? are expected
to stabilize beyond 70 cm [217] and the results at D = 70 cm may be used to
determine applicability to field trials (Figure 3.28).
Chemical-kinetic mechanisms that control soot formation and destruction have
not been explored in this work. Parameters such as residence time, flame surface
area, turbulence intensity at the different regions of the FW and PF structures
determine the emissions of PM. These details of these parameters, however, have
not been addressed in this work. The hypothesis that FWs indeed result in lower PM
emissions has been confirmed, and predicting the reduction in PM emissions requires
91
Figure 3.27: (A) Combination of peak Q?, and ? for the different FWs in this study.
(B) Variation of q? with D.
Figure 3.28: Burning rate and flame height for pool fires formed using various liquid
fuels. As radiative feedback begins to dominate in large fires, burning rate tends to
flatten with pool diameter. Figure adapted from [30].
92
establishing relationships between the EFPM factor and parameters such as ?, Q?
and q???. These global variables define the structure of a FW, and they parameters
vary significantly with D (Figure 3.27). Thus, the explanation of the reduction in
emissions across length scales can be based on a phenomenological approach, using
scaling analysis to relatE EFPM to nondimensional circulation, heat-release rate and
heat-flux feedback. This is explored in Chapter 4.
93
3.5 Summary
Experiments were performed at different length scales, defined by fuel-pool
diameter, to quantify emissions from liquid-fueled pool fires and fire whirls. Two
fuels were used ? heptane, a single-component hydrocarbon, and ANS crude oil, a
multi-component mixture of hydrocarbons with large differences in molecular weight
and boiling point among the components.
Results show that at all pool diameters, FWs have higher burning rate and
lower PM emission factor compared to corresponding PFs, confirming the original
hypothesis. PM emission factors from ANS crude oil fires were an order of magnitude
higher than those from heptane fires. In the case of ANS crude oil fires, fire whirls
also had higher fuel consumption efficiency, although the difference between pool
fires and fire whirls was negligible when boilover occurred.
Preliminary explanations for this reduction in PM emissions are based on
a feedback mechanism between heat-flux feedback, heat-release rate, entrainment
and gas temperature. A scaling approach is necessary to relate the different fire
parameters to emissions. Further, the effect of boilover needs to be investigated
further using data from the 70 cm ANS crude oil experiments. These aspects are
discussed in detail in Chapter 4.
94
3.6 Acknowledgments
This chapter, in part, was published as report 1094AA, ?Efficient Remediation
of Oil Spills over Water Using Fire Whirls,? Bureau of Safety and Environmental
Enforcement (BSEE) in the In Situ Burn Research category in May 2019 [218].
The final report can be accessed at https://www.bsee.gov/research-record/
efficient-remediation-of-oil-spills-over-water-using-fire-whirls.
Experimental results were also included in the proceedings of the 42nd AMOP
Technical Seminar, Halifax, Canada, June 2019 [219]; the 11th US National Com-
bustion Meeting, Pasadena, California, March, 2019 [220]; and in a manuscript
titled ?Comparison of Particulate Emissions from Liquid-Fueled Pool Fires and Fire
Whirls at Different Length Scales,? by S.B. Hariharan, H.F. Farahani, A.S. Rang-
wala, J. Dowling, E.S. Oran, M.J. Gollner, in preparation for submision and peer
review in August 2020.
95
Chapter 4
Effect of Circulation and Heat-Release Rate on
Particulate-Matter Emissions
4.1 Overview
The addition of swirling flow in a combustor improves flame stability, enhances
combustion efficiency and reduces emissions [57]. In most swirl combustors, reduc-
tion in emissions typically depends on increasing mixing by increasing turbulence
intensity [59, 60]. While there are similarities between swirl combustion and fire
whirls, they differ in the manner of air entrainment. In swirl combustion, air is
typically premixed with fuel, or injected with considerable axial momentum around
a central fuel port or spray injector. There is negligible axial momentum from air
entrainment at the bottom surface boundary over which fire whirls are formed, and
combustion occurs in a structure that is defined by the Burger?s vortex [4]. Com-
pared to pool fires (PF), fire whirls (FW) show higher fuel consumption rates [88]
96
and burn at higher temperatures [109]. In contrast to swirl combustion, turbulence
is generally suppressed in FWs [95]. A fundamental question needs to be answered
to evaluate the potential for using FWs in ISBs ? when compared to PFs, do FWs
emit lower quantities of airborne soot? The experimental data in Chapter 3 showed
that the PM emission factor (EFPM) was lower for FWs at a given pool diameter.
In this chapter, the effect of length scale (pool diameter) on burning rate,
fuel-consumption efficiency, emissions of PM, unburned hydrocarbons (UHC), CO2,
heat-flux feedback to the fuel pool, and emission factors for each species are explored
in detail. Then, two scaling parameters relevant to FW behavior are defined ?
inverse Rossby number (Ro?1), which represents the effect of ambient circulation,
and nondimensional heat-release rate (Q??), the effect of buoyancy. EFPM of ANS
FWs decreases linearly with the ratio (Ro?1/Q??), which quantifies the competition
between tangential momentum and axial buoyant momentum. Finally, for ANS
fires, boilover was found to cause most of the fuel to be burned in the post-boilover
period, although the overall EFPM was not affected significantly by boilover.
97
4.2 Methods
4.2.1 Emission Factors
The raw experimental data from Chapter 3 is used as the basis for analysis.
The instantaneous concentration measured for each species and the total volumetric
flow rate through the duct were used to calculate the emission rate in units of
[mass/time]. The total mass, M , of each species was measured as an integral over
time,
[ ] [ ] ? t [ ] [ ]m3 b mg
M mg = V? C dt s (4.1)
s m30
where V? is the volumetric flow rate measured in the exhaust duct, C is the con-
centration measured by either the gas analyzer or PM sensor, and tb is the burn
duration.
The concentration of unburned hydrocarbons (UHC) was not directly mea-
sured, but was estimated using the carbon mass balance approach [116], assuming
the different emissions to be composed of four species, CO2, CO, PM and UHC. The
carbon mass emitted as UHC was estimated as the difference between carbon mass
in the fuel consumed and the total carbon mass measured as CO2, CO and PM. The
carbon mass in heptane and ANS crude oil was 84% and 85.9% [221], respectively.
For each of the emitted species, an emission factor was defined as
Mass of carbon in emission species [g or kg]
EFspecies = (4.2)
Mass of carbon in fuel consumed [kg]
98
based on the carbon mass in the species and carbon mass in burned fuel, similar to
previous work [116,222]. The burning efficiency (?b) was defined as
Mass of carbon emitted as CO2
?b = (4.3)
Total mass of carbon in CO2, CO, PM, UHC
to compare the relative amount of C emitted as CO2 to the total mass of C
emitted from the fuel as CO2, CO, UHC and PM. In this way, both ?b and EF
are bounded between 0 and 1, and also cause ?b to be numerically equal to EFCO2 .
Finally, oxidation factor (OF) is defined for each case as
Mass of O2 emitted as CO2
OF = (4.4)
Total mass of O2 consumed
which quantifies the fraction of total oxygen consumed that is emitted as CO2. An
oxidation factor of 1 implies all the O2 consumed is emitted in the form of CO2.
99
4.3 Results
4.3.1 Effects of D on Fire Behavior
The variation of burning rate (m?) with D is shown in Figure 4.1. For both
heptane and ANS crude oil, m? is higher for FWs. The experimental variability in
the measurement of m? was negligible, evidenced by the very small error bars in
Figure 4.1. For fires formed using each fuel at D = [10, 30] cm, the ratio of m? of
FWs to PFs is about 2, and at D = 70, the ratio is 1.3. Comparing each flame
regime individually for the D = [10, 20, 30] cm, m? for heptane fires is larger than
that for ANS fires. This trend, however, reverses at D = 70 cm, where m? of ANS is
higher than that of heptane. This is also evident in Figure 4.2A, which shows the
variation of fuel mass flux (m???). For heptane FWs, m??? declines with scale, while for
heptane PFs, m??? decreases from 10? 30 cm and then increases at the 70 cm scale.
For ANS fires, however, m??? increases linearly in the 10?30 cm scales, but increases
sharply between the 30 and 70 cm scales. The characteristics described above were
also tabulated in Table 3.3.
Upon extinction of an ANS crude oil fires, some residue remained over the
water surface. This residue is primarily composed of heavier hydrocarbons and tars
in the fuel [221]. Measurable residues were not observed with heptane fuel. For
ANS fires, fuel-consumption efficiency (?fuel) was defined as
mass of fuel burned
?fuel =
initial mass of fuel
100
Figure 4.1: Comparison of m? for PFs and FWs formed using (A) heptane and (B)
ANS crude oil.
Figure 4.2: Variation of (C) m??? for heptane and ANS fires, and (D) ?fuel with D
for ANS fires.
101
and is shown in Figure 4.2B. In experiments at D = 10 cm, ?fuel of FWs about
53%, while that of PFs was just over 40%. At D = 70 cm, efficiencies above 80%
were observed for both PFs and FWs. Removal efficiencies above 90% have been
reported in field trials [208]. ?fuel is higher for FWs at D = [10, 20, 30] cm, but this
difference is negligible at D = 70 cm. The sharp rise in both m??? and the negligible
difference in ?fuel between for ANS PFs and FWs at D = 70 cm is attributed to
boilover (section 4.4.2).
4.3.2 Circulation and Heat Feedback
The combination of circulation (?) and heat-release rate (Q?) defines the state
and shape of a FW [92]. Values of ? and Q? for the FWs in this study are shown
in Figure 4.3. The Q? was estimated as Q? = m??hc, where ?hc is the lower heating
value of the fuel, and ? was estimated as ? = 4S U?. The Q? of the FWs in this study
spans three orders of magnitude, wider than most previous studies. According to
previous definitions by Lei et al. [92], the shapes of the FWs observed in this study
may generally be classified as cylindrical FWs.
The ? calculated for the FWs in this study is compared to the inverse of the
Rossby number in Figure 4.4A. The quantity Ro?1 describes the ratio of azimuthal
(tangential) momentum from circulation to the buoyant (axial) momentum from
heat release. Following [9], Ro?1 was estimated as (?D)/(m???/?b), where ?b is the
air density at T = 1300 K. While ? increases continuously with pool diameter, Ro?1
increases from the 10?30 cm scales, but is relatively unchanged between the 30 and
102
Figure 4.3: Combinations of Q? and ? for the FWs in this study, showing a logarith-
mic relationship for both heptane and ANS FWs.
70 cm scales. Both ? and Ro?1 are generally higher for FWs formed using heptane
when compared to those formed using ANS crude oil.
The variation of average heat-flux feedback, q???, with D is shown in Figure
4.4B. q??? is higher for FWs at all scales, although the difference is very small in the
10 ? 30 cm scales. At the 70 cm scale, however, q??? for FWs is more thna twice
that for PFs. Also, for the two flame regimes, the value of q??? for is higher for fires
formed with ANS crude oil than those formed with heptane, partly due to the higher
concentration of soot particles in the flame. This is discussed in section 4.3.3. For
FWs, since ? increases linearly with D, q??? also increases roughly linearly with ?
(Figure 4.4C). For each case, nondimensional heat flux was estimated from [223]
?
as q?f = q?
?? / (?0 cp ?T gD) and its variation with ? is shown in Figure 4.4D. In
general, an increase in circulation corresponds to an increase in heat flux feedback
to the pool, similar to previous studies [87, 224]. The experimental variation is
103
Figure 4.4: Variation of (A) ? and Ro?1 with D, (B) q???f with D. Panels (C) and
(D) show the linear increase in q???f and q
?
f with ? for FWs. Solid lines indicate linear
fits for heptane FWs, and dashed lines indicate those for ANS FWs.
104
higher for fires at D = 70 cm since these measurements were made directly using
a Schmidt-Boelter gauge, whereas experiments in the smaller scales are averaged
temporally and spatially, and instantaneous fluctuations are not apparent.
4.3.3 PM Emissions
The PM-emission rate for fires at D = 70 cm is shown in 4.5. Panels (A)
and (B) compare heptane fires, and panels (C) and (D) compare ANS fires, with
the different line colors indicating individual experiments. As FWs have a higher
burning rate as compared to PFs, the duration of PM emission is lower. The peak
emission rate of PM is higher for PFs (250 mg/s for heptane PF, 3000 mg/s for
ANS PF) than for FWs (175 mg/s for heptane FW, 2300 mg/s for ANS FW). The
instantaneous PM emission rate from ANS fires is an order magnitude higher than
that from the heptane fires. For heptane fires, the peak emission rate is observed
roughly around the middle of the burn duration. For ANS fires, however, there
is an initial linear increase in the emission rate from ignition, with a sudden spike
in the rate, followed by a quick drop towards extinction. The sharp spike in the
emission rate (at ?65 s for ANS FWs, ?110 s for ANS PFs) is caused by the onset
of boilover. Boilover was seen only with ANS crude oil at D = 70 cm, and hence
the sharp rise in PM emission rate was not seen in ANS fires for D = [10, 20, 30] cm
(refer to Figures 3.11?3.13).
The PM emission factor, EFPM, calculated according to Equation 4.2 is pre-
sented in Figure 4.6. The definition of EF based on carbon content of the fuel also
105
shows that a lower fraction of carbon in the fuel is emitted as soot by FWs. For
PFs and FWs formed with both heptane and ANS crude oil, the lowest EFPM is at
the 30 cm scale. For PFs and FWs formed using ANS crude oil, the value of EFPM
in the pre-boilover and post-boilover periods are similar to the overall EF (Figure
4.5B).
The EFs of UHC (EFUHC) and of CO2 (EFCO2) are shown in Figure 4.7.
The lowest EFUHC is seen at D = 20 cm for both heptane and ANS crude oil. For
heptane fires (Figure 4.7A), EFUHC for FWs is slightly lower than from PFs, and this
simultaneously corresponds to slightly higher EFCO2 (Figure 4.7C). For ANS fires at
D = [10, 70] cm, EFUHC is slightly lower for FWs, along with slightly higher EFCO2 .
At D = [20, 30] cm, however, FWs show slightly higher EFUHC but lower EFCO2
(Figure 4.7D). Compared to CO2 and UHC, CO is found only in trace amounts.
EFCO is shown in Figure 4.7 (E) and (F). For heptane, fires, FWs show lower EFCO
than PFs. This is reversed for ANS fires, with FWs showing higher EFCO than PFs.
The EFs discussed above were calculated based on the mass of carbon in the fuel.
The EFs determined on the basis of the total fuel mass burned are tabulated for all
species in Table 4.1. EFs calculated by the two methods vary numerically, but show
the same trend with D. The mass of UHC emitted was calculated as the product
of the molecular weight of propane (C3H8) and the estimate of carbon emitted as
UHC.
106
Figure 4.5: PM emission rate, measured at 1 Hz, as a function of time for 70 cm
heptane PFs and FWs are shown in panels (A) and (B). Data for 70 cm ANS PFs
and FWs are shown in panels (C) and (D). Boilover is evident in the PM emission
rate for both ANS PFs and FWs.
Figure 4.6: EFPM for the different values of D is shown for fires formed using heptane
and ANS crude oil in panels (A) and (B). Panel (B) shows markers for overall EFPM,
and those calculated for the pre- and post-boilover periods.
107
Figure 4.7: Variation of different EFs with D. EFUHC is shown in panels (A) and
(B), EFCO2 in panels (C) and (D), and EFCO in panels (E) and (F). Data for 70 cm
ANS fires also shows markers for the pre- and post-boilover periods.
108
109
Table 4.1: Tabulated values of EF for the different species from PFs and FWs in this study. For each parameter, the mean of
three experiments is shown for both fuel-mass based and fuel carbon-mass based estimation methods.
Emission Factor based on total fuel mass Emission Factor based on carbon mass in fuel
Scale EFPM EFCO2 EFCO CFO2 EFUHC EFPM EFCO2 EFCO CFO2 EFUHC
[cm] [kg/kg] [kg/k]g [kg/kg] [kg/kg] [kg/kg] [kg/kg] [kg/kg] [kg/kg] [kg/kg] [kg/kg]
10 0.021 2.238 0.001 3.193 0.254 0.025 0.727 0.001 3.801 0.248
Heptane 20 0.023 2.805 0.004 3.236 0.062 0.027 0.911 0.002 3.853 0.060
PF 30 0.010 2.790 0.003 3.067 0.083 0.012 0.906 0.002 3.651 0.080
70 0.018 1.921 0.003 2.178 0.364 0.022 0.624 0.002 2.593 0.354
10 0.011 2.403 0.001 2.909 0.211 0.014 0.780 0.001 3.464 0.205
Heptane 20 0.007 2.975 0.003 3.237 0.025 0.008 0.966 0.001 3.854 0.024
FW 30 0.005 2.867 0.003 3.168 0.062 0.007 0.931 0.002 3.772 0.061
70 0.008 1.965 0.002 2.196 0.369 0.009 0.638 0.001 2.615 0.359
10 0.148 0.912 0.003 3.426 0.575 0.161 0.290 0.001 3.989 0.548
ANS 20 0.141 2.851 0.032 3.016 0.058 0.144 0.787 0.014 3.511 0.055
PF 30 0.049 1.215 0.002 1.764 0.584 0.057 0.386 0.001 2.054 0.557
70 0.113 1.501 0.018 1.629 0.415 0.131 0.468 0.009 1.874 0.395
Pre-boilover 0.119 1.426 0.011 0.910 0.416 0.138 0.453 0.006 1.874 0.396
70
Post-boilover 0.109 1.492 0.021 2.419 0.484 0.127 0.474 0.010 1.874 0.396
10 0.097 1.524 0.002 4.112 0.423 0.112 0.484 0.001 4.787 0.403
ANS 20 0.042 2.571 0.032 2.590 0.125 0.049 0.816 0.016 3.016 0.119
FW 30 0.013 1.037 0.002 1.299 0.672 0.015 0.344 0.001 1.512 0.640
70 0.062 1.741 0.029 1.997 0.381 0.072 0.551 0.014 2.290 0.363
Pre-boilover 0.064 1.995 0.035 0.550 0.261 0.074 0.633 0.017 2.290 0.249
70
Post-boilover 0.061 1.693 0.028 3.379 0.467 0.071 0.538 0.014 2.290 0.382
4.4 Discussion
The experimental results presented above show that a pool of liquid fuel when
burned as a FW, in comparison to a PF, has a higher burning rate and emits lower
peak concentrations of airborne PM. These factors together result in a lower EFPM
for FWs. Additionally, when ANS crude oil is used as the fuel, FWs had a higher
?fuel since the residual mass upon extinction was lower. A minimum slick thickness
of 2 ? 2.5 mm is required to sustain burning over water surfaces [121, 208] due to
heat loss to the water surface beneath the fuel. Consequently, the consumption
efficiency for a given flame regime and pool diameter is directly proportional to the
initial slick thickness, since all fires will eventually burn down to the minimum slick
thickness.
The behavior of EFPM also shows significant dependence on the fuel type.
The EFPM for ANS fires is an order of magnitude higher than those for heptane.
Lighter fuels are known to have lower EFs than fuels that are contain high-molecular
weight, complex components [16, 116]. For a given flame regime, higher values of
EFPM are seen for crude oil, a multi-component fuel with long-chain hydrocarbons
with complex structures [221].
The addition of swirl to jet flames results in the reduction of emissions from
swirl burners [60]. In swirl burners, the reduction in emissions is enabled by enhanc-
ing mixing caused by increased levels of turbulence intensity, which causes combus-
tion to approach premixed conditions. Turbulence, however, is suppressed in FWs,
and is one of the reasons for elongation of the flame as compared to a PF formed
110
Figure 4.8: Grashof number (Gr) at different values of D.
over a pool at the same D [8,95]. The reduced effect of turbulence is evident in the
lower values of the Grashof number (Gr) for PFs in Figure 4.8. Gr was estimated
as Gr = g??TD3/?2, where g is gravitational acceleration, ? is the ideal-gas ex-
pansion coefficient, ?T is the excess temperature, and ? is the kinematic viscosity.
Gr increases significantly with D, but only the fires at D = 70 cm approach the
turbulent regime (Gr > 109) [225]. Thus, the lower value for EFPM for FWs is not
explicable based on turbulence or mixing effects.
The values of EFUHC and EFCO2 are at least an order of magnitude higher
than EFPM, showing that a majority of the carbon in the fuel is emitted in the form
of gaseous components, and that only a small fraction is emitted as particulate soot.
So, small differences in the values of EFUHC and EFCO2 between the PF and FW
regimes can cause significant differences in EFPM. This is evident when comparing
111
Figure 4.5 (C,D) with Figure 4.7 ? the value of EFPM for FWs is roughly half that
of PFs, but the values of EFUHC and EFCO2 are within 10% of each other for the
different cases.
In the following section, the behavior of EFPM with FW-scaling parameters
is analyzed to provide preliminary explanations of the differences between PFs and
FWs that lead to reduced emissions.
4.4.1 Scaling
In a FW setup with natural entrainment, ? is intrinsically coupled with Q?
[24]. This is because the buoyancy, that stems from Q?, controls the strength of
air entrainment into the enclosure, U?, which in turn determines ?. The effect of
circulation on FWs is represented by either the Froude number (Fr = U2? /gD),
which is analogous to the Richardson number (Ri = g??TD/U2? ). Both Fr and Ri
represent the competing effects of tangential momentum from ambient circulation
and buoyant momentum from heat release. While Ri has previously been used to
describe Hf [93], Chuah et al. [9] showed that the effect of circulation on FWs and
Hf was better represented by the Rossby number (Ro), a nondimensional quantity
that is equivalent to Ri?1. The effect of buoyancy on the FW is represented in the
form of a nondimensional heat-release rate, Q??, defined for FWs as [5]
Q??
Q?
= ? (4.5)
? 50 ?T gD
112
where ?0 is the ambient air density [17], Q? is the average heat-release rate. The
variation of Q?? with D is shown in Figure 4.9A. We now define the ratio (Ro?1/ Q??)
to compare the effects of buoyancy and circulation for FWs formed at different
length scales. The effect of this ratio on EFPM for heptane and ANS FWs is shown
in Figure 4.9B.
For ANS FWs, Figure 4.9A shows a linear relationship between EFPM and
(Ro?1/Q??). EFPM decreases as (Ro
?1/Q??) increases, indicating that as the effect
of circulation increases relative to buoyancy, conditions are favorable for lower PM
emissions. The ratio (Ro?1/Q??) has a similar effect on EFPM from heptane fires,
with the lowest emissions seen at (Ro?1/Q??) ? 150. Beyond this value, however,
there is an increase in EFPM for heptane FWs.
For a given fuel and D, with the increasing effect of circulation, EFPM reduces
linearly with (Ro?1/Q??), but there exists an upper limit beyond which EFPM rises
again. The increase in EFPM for heptane fires beyond this threshold value agrees
with previous observations of FW behavior under high values of ? ? above a certain
value of ambient ?, FWs show reduced Hf , increased wf , and increased soot emis-
sions that are visible as black smoke [61, 92, 226]. Recent experiments conducted
by Lei et al. [92] showed that as ? is increased at a constant value of Q?, Hf first
increases, then decreases. This change in Hf and wf resulting from high ? is re-
ferred to as ?over-rotation? [226]. Above (Ro?1/Q??) ? 150, the increase in EFPM
may be attributed to over-rotation in heptane FWs. It appears that there may be a
different value of this threshold for different fuels, since over-rotation is not observed
for ANS FWs in Figure 4.9B.
113
Figure 4.9: (A) Variation of Q?? with D. (B) Variation of EFPM with (Ro
?1/Q??).
Burning efficiency for (C) heptane and (D) ANS fires. Oxidation factor for (E)
heptane and (F) ANS fires.
114
For ANS FWs, the highest value of the ratio (Ro?1/Q??) is found forD = 30 cm.
Going from the 30 to 70 cm scales, the value of Ro?1 remains relatively unchanged
(see Figure 4.4A), but Q?? increases (Figure 4.9A), causing (Ro?1/Q??) to decrease
from 213 to 141. The increase in Q?? at 70 cm is a direct consequence of the steep
rise in m??? caused by boilover. In contrast, for heptane FWs, the largest value of the
ratio (Ro?1/Q??) is 231, occurring at D = 70 cm. Similar to ANS FWs, heptane FWs
at D = 30, 70 cm have similar values for Ro?1, but Figure 4.9A shows that heptane
FWs have the lowest value of Q?? for fires at D = 70 cm. The significant reduction
in Q?? from the D = 30 to 70 cm is the major contributing factor to increasing the
effect of buoyancy (Figure 4.9B) on the FW and leads to an increase in the value
of the ratio Ro?1/Q??. Thus, over-rotation in heptane FWs at D = 70 cm is a
consequence of a decline in Q?? rather than an increase in Ro?1. It is hypothesized
that if boilover did not occur, the value of Q?? would have been lower, causing the
value of the ratio (Ro?1/Q??) at D = 70 cm to be higher than that at D = 30
cm, potentially presenting the opportunity for over-rotation to occur. This may
not, however, be a preferred FW state for practical applications since it leads to an
increase in EF . The variation of (Ro?1/Q??PM ) with D is shown in Figure 4.10.
The variation of ?b and OF, calculated according to Equations 4.3 and 4.4,
with D are shown in Figure 4.9 (C?E) for the all the different cases. Generally,
FWs have higher values for both ?b and OF as compared to PFs, showing that a
higher fraction of carbon from the fuel is oxidized to CO2, causing EFPM to be
lower for FWs than for PFs. Based on the expressions in Equations 4.2 and 4.3, ?b
is numerically equivalent to EFCO2 . Again, the metrics ?b and OF are calculated
115
Figure 4.10: Variation of the value of (Ro?1/Q??) with D for heptane and ANS FWs.
The ratio un
based on gaseous components, which are primarily composed of CO2. So even
slight improvements in converting carbon in the fuel to CO2 can result in significant
reduction in soot (PM) emissions. Both these parameters are generally higher for
heptane fires than for ANS fires. The high values for ?b and OF for ANS fires at
the 20 cm scale leads to lower EFPM at these scales for both fuels (see Figure 4.6).
4.4.2 Boilover
The violent burning of thin layers of oil over water is referred to as boilover
[215]. Figure 4.11A shows a comparison of normal burning (panel A) and boilover
burning (panel B). This regime of burning is caused by a sudden expansion of
superheated water-vapor bubbles that nucleate at the fuel-water interface. Char-
acteristics such as burning rate and flame height in this regime depend on many
factors including slick thickness, D, fuel type and vaporization order, weathering,
116
q???, etc. [121, 227]. Generally, boilover causes a significant increase in both Hf and
m?.
In this work, boilover was observed only with ANS crude oil, and was actively
avoided in the 10? 30 cm experiments by using a magnetic stirrer to continuously
mix the water sublayer. This ensured that the water bulk temperature increased
only about 10 ? 30 K, depending on D and the flame regime. This prevented the
fuel-water interface temperature from reaching boiling point, avoiding any bubble
nucleation. At the 70 cm scale, however, a stirrer could not be used. Instead, a
recirculating pump system was installed to recirculate the water from the water-fuel
interface to the bottom of the sublayer. However, the recirculation system could
not completely prevent boilover due to the large value of D, but could delay the
onset. For ANS experiments at D = 70 cm, a slick thickness of 7 mm was used
instead of the 5 mm that was used at the smaller scales. The recirculation system
and increased slick thickness delayed boilover onset by about 40 s, allowing sufficient
time for capturing data in both normal burning and boilover burning regimes.
During boilover, the bubbles shoot through the fuel-water interface, ejecting
oil droplets into the flame [215]. Streaks of these evaporating fuel droplets are visible
in Figure 4.11B, significantly increasing the effective surface area for combustion.
This results in the large value of both m??? (Figure 4.2B) and ?fuel (Figure 4.2C) for
ANS fires at the 70 cm scale when compared to the smaller scales. Boilover burning
is also responsible for increased q??? (Figure 4.4), partially through enhanced radiative
feedback, which in turn sustains a high temperature at the fuel-water interface.
The onset of boilover can be determined visually, but was also determined by
117
Figure 4.11: Images of a PF under (A) normal and (B) boilover burning, formed
using a 5 mm slick of ANS crude oil. The streaks seen in (B) are formed by the
fuel droplets emanating from the fuel pool. Variation of carbon-emission rate with
time (D = 70 cm) for (C) heptane PF and (D) ANS PF. (D) Bar graphs showing
the difference between PFs and FWs (D = 70 cm) on the basis of time for onset of
boilover, fuel mass consumed before boilover onset, and fuel mass consumed during
boilover.
118
the rate of emission of carbon emitted by the fire. This was determined from the
cumulative measurements of the carbon emission from CO2, CO and PM. Carbon-
emission rate is almost constant during the normal burning, shown in Figure 4.11C
for heptane PFs (D = 70 cm). When boilover occurs, however, the slope of this
curve changes during the burning period, seen in Figure 4.11D, showing the carbon-
emission rate for ANS PFs. The time at which the slope of the carbon-emission
increases to a new value is denoted as the onset of boilover. For the 7 mm slicks
used in these experiments, the mean boilover-onset time for PFs and FWs was 111
s and 63.5 s, respectively.
The carbon-emission rate is higher during the boilover phase, resulting in the
higher overall value of m? and m??? for ANS fires (Figure 4.2 A and B). The mass
of fuel consumed in each phase was calculated using the boilover onset time, the
fraction of O2 consumed in the pre-boilover regime, and then extrapolating the
fraction to the total fuel mass consumed. For FWs, only 14% of the fuel is burned
prior to boilover onset, compared to 25% for PFs (Figure 4.11E). Thus, a majority
of the fuel consumed by the FW is during the boilover phase, resulting in the steeper
carbon-emission rate in Figure 4.11D.
Figure 4.6 and Table 4.1 show that the value of EFPM for the pre- and post-
boilover periods is very similar to the overall EFPM, meaning that the fraction
of total carbon in the fuel that is emitted as soot is similar in these periods. This
suggests that in terms of soot emissions, the effectiveness of the FW regime is rather
similar during normal burning and boilover burning. Present understanding of ISBs
suggests that boilover does not occur over open water [208]. Given some of the
119
benefits of boilover over normal burning such as increased m???, ?b, radiative feedback,
and an EFPM similar to normal burning, it may be advantageous to purposely induce
boilover during open-water ISBs.
4.4.3 Factors influencing PM emissions
A preliminary analysis of the differentiating factors between PFs and FWs,
and their relation to EFPM is discussed briefly. First, the higher oxygen consump-
tion (and OF) for FWs suggests that their global equivalence ratio is leaner than
that of PFs. This causes higher average temperatures in the FW structure, which
has a larger surface area as compared to PFs. The high temperatures and improved
view factor results in higher radiative heat feedback to the fuel pool [87]. The ra-
diative feedback is higher in ANS fires than heptane fires (Figure 4.4) due to the
higher concentration of soot particles in the flame. The convective heat feedback is
also enhanced by the strong radial inflow at the bottom boundary layer [88]. When
compared to PFs, increased heat feedback (q???) leads to higher fuel-evaporation rate
(m?) and thus, higher heat-release rates (Q?). In a FW apparatus with natural air en-
trainment, increased Q? corresponds to increased U? and ?, which in turn contribute
to decreasing overall equivalence ratio, increasing OF, ?T and q???, completing the
feedback loop. The increase in q?f with ?, and the higher ?fuel values for ANS
FWs indicate that the high-molecular-weight components in ANS crude oil require
higher heat feedback to evaporate [227]. Overall, the lower EFPM from FWs is a
consequence of higher burning efficiency as compared to PFs.
120
The detailed mechanisms leading to a reduction of EFPM, when a pool of
fuel is burned as a FW rather than a PF, are not immediately apparent from the
data presented here. The complex feedback mechanism between the fluid dynamics
of air entrainment and combustion in the whirling flames need to be investigated
further. More detailed experiments at small length scales will shed light on the role
of the vortex structure, turbulence and mixing characteristics, and their interplay
with chemical kinetics of soot formation and destruction. Additionally, the effects
of boilover are not fully understood in the context of emissions and is dependent on
the flame regime. For instance, considering EFCO from PFs, post-boilover emissions
are higher than pre-boilover, but this is reversed for FWs. Still, boilover has only a
minimal effect on the overall EFPM. The data presented here serves as a platform
for more detailed investigations in the future by incorporating the effects of chemical
kinetics and soot formation using a combination of non-intrusive diagnostics, scaling
methods and numerical simulations.
Another parameter of that influences burning efficiency is the flame surface
area, which determined by wf and Hf . With increasing ? and Ro
?1, Hf increases
and wf decreases, and the overall flame area increases [92]. PFs have the lowest
flame area since Ro?1 for PFs is 0. As the ratio (Ro?1/Q??) increases above 0, flame
surface area increases as FWs are formed, combustion efficiency increases, resulting
in the lowering of EFPM. Beyond a certain upper limit of ?, however, Hf begins to
decrease. Apart from Ro?1, Hf is also controlled m?. For the FWs in this study,
although m? increases continuously with D, but Hf does not change linearly with D.
This is a consequence of Gr; as D increases from 10 to 70 cm, the values of Gr fall
121
within the laminar-turbulent transition regime, which exists between D = 10 cm
and D = 100 cm [30,217]. Two effects cause this ? (i) the nonlinear in increase in Gr
with D causes improved mixing, leading to Hf values being lower than expected.
This is similar to the stabilization of the flame height of jet diffusion flames at
the onset of turbulence (Fig. 5.10 in [81]); (ii) more importantly, the height of
the enclosure walls also impact Hf . The value of Hf is only slightly higher than
the wall height because ? = 0 above the limits of the enclosure walls. When the
effect of ambient circulation is lost, the flame-lengthening process ceases, eliminating
vortex concentration, and eventually the turbulence suppression that is necessary for
flame lengthening. This enables buoyant momentum to dominate the fire, causing a
transition to turbulence, and the resulting mixing ensures that the flame height ends
immediately above the height of the enclosure walls. The higher the wall height,
the higher the maximum value of Hf can be.
?b also depends on the the flame surface area, which in turn depends on the
height of the enclosure, which must be capable of supporting the maximum value
of Hf possible for a given combination of m? and D. For instance, Figures 2.3 and
3.7G were both formed at D = 30 cm using heptane. But Hf is different because of
the height of the enclosure walls and the effect of suction used in the case of Figure
2.3, enabling a strong vortex to be formed.
122
4.5 Summary
An experimental investigation of airborne emissions from pool fires and fire
whirls was conducted at fuel-pool diameters of 10, 20, 30 and 70 cm. Two liquid
fuels were used, heptane and ANS crude oil. Free-buoyant pool fires were formed
in a quiescent background, and fire whirls were formed in a fixed-frame setup that
allowed natural entrainment of air towards the fire. Attributes such as burning rate,
burning efficiency, emissions of particulate matter, carbon dioxide, carbon monoxide,
and oxygen consumption were measured for the different pool diameters and flame
regimes. Results showed that for both fuels, as compared to pool fires, fire whirls
burned fuel at higher rates. When burning ANS crude oil, the mass of unburned
residue was lower for fire whirls, showing that the fuel consumption efficiency of fire
whirls was also higher.
Although the burning rate was higher for FWs, the peak emission rate of
particulate matter (soot) form FWs was lower than from PFs. The emission factor
of particulate matter emissions from fire whirls was ? 50% lower than that from
pool fires. This was true for both heptane and ANS crude oil, across all of the
pool diameters. For pool fires and fire whirls, the emissions behavior of the two
fuels was significantly different. The PM emission factor for ANS crude oil fires
was an order magnitude higher than that for heptane fires. The burning efficiency
also showed that a over half of the carbon in the crude oil was emitted as unburned
hydrocarbons. This is attributed to the significant differences in the composition of
the fuels and the fires they supported.
123
For the fire whirls in the study, a ratio of the inverse of the Rossby number
and nondimensional heat-release rate, Ro?1/Q??, was defined to compare the effects
of buoyancy and circulation on the fire. As the value of Ro?1/Q?? increased, the
emission factor of particulate matter decreased linearly for ANS fire whirls. For
heptane fire whirls, the linear relationship existed for diameters between 10 and
30 cm. The emission factor increased again at 70 cm due to a phenomenon called
?over-rotation?. Generally, the flame height of a fire whirl increases with Ro?1.
But beyond an upper limit, the flame height reduces and flame width begins to
increase, a state referred to as over-rotation, which also corresponds to an increase
in particulate emissions. While over-rotation has been reported qualitatively in the
literature [61], the ratio Ro?1/Q?? quantifies the phenomenon and helps explain the
increase in emissions as a consequence of the decrease in flame surface area.
Over-rotation was not observed in fire whirls formed using ANS crude oil at
the 70 cm scale. This is attributed to the phenomenon of boilover, which occurs
during burning of thin films of dense hydrocarbon fuels floating over a water surface.
Boilover caused a rapid increase in burning rate, which resulted in an increase in
heat-release rate and buoyancy. This was reflected in a significant decrease in the
ratio Ro?1/Q??, corresponding to an increase in the buoyant momentum relative
to tangential momentum, ensuring that over-rotation does not occur. The high
burning rate during boilover also resulted in significant increase in the flame height
and fuel-consumption efficiency. The onset-time of boilover was lower for fire whirls,
and fire whirls also consumed a higher fraction of the initial fuel during the boilover
period. Additionally, the overall particulate emission factor was similar to that in
124
the pre-boilover and post-boilover periods, showing that it may be advantageous to
induce boilover during in-situ burns.
The reduction in particulate-matter emissions from fire whirls is attributed to
the following factors. Compared to pool fires, fire whirls consume more oxygen per
unit mass of fuel, leading to a lower global equivalence ratio. They also have higher
burning efficiency, and a higher fraction of the oxygen consumed was converted to
carbon dioxide. Collectively, these result in higher temperatures in the fire whirl
structure, leading to the higher heat feedback to the fuel surface. Heat feedback
increased linearly with pool diameter and circulation, and led to higher burning
rate, which in turn resulted in higher air-entrainment velocities, which in turn leads
to higher oxygen supply to the fire whirl, completing the feedback loop.
A number of open questions regarding the emissions from fire whirls remain to
be answered before practical applications are considered. While soot-free whirling
flames such as the ?blue whirl? are known in the literature [22, 64], they form at
very small length scales [24] and are highly sensitive to experimental conditions [66].
Fire whirls, on the other hand, can be formed more robustly across a large range
of length scales. While fire whirls do emit some quantities of soot, the results in
this work demonstrated that they can still offer opportunities to significantly reduce
the environmental impact of ISBs, which are currently burned as pool fires. The
range of heat-release rates and circulation for fire whirls in this study is wider than
those in the literature, although these factors were not varied independently in this
study and only natural-entrainment configurations were used. To understand the
independent effect of these parameters on emissions, systematically varying these
125
quantities by controlling air entrainment could help isolate the effects. The effect of
boilover on emissions needs to be investigated further since its onset does not seem
to significantly affect the particulate matter emission factor. The metrics presented
here provide a global explanation for the reduction in particulate emissions compared
to pool fires, and future work will focus on the exact mechanisms leading to efficient
combustion.
4.6 Acknowledgments
Data from this chapter, in part, is included in a manuscript titled ?Comparison
of Particulate Emissions from Liquid-Fueled Pool Fires and Fire Whirls at Different
Length Scales,? by S.B. Hariharan, H.F. Farahani, A.S. Rangwala, J. Dowling, E.S.
Oran, M.J. Gollner, in preparation for submission and peer review in August 2020.
126
Chapter 5
Quantifying the Effects of Circulation and Buoyancy on the
Transition from a Fire Whirl to a Blue Whirl
As demonstrated in Chapters 3 and 4, the addition of circulation (or, a swirling
flow field) to a free-buoyant pool fire can lead to an increase in combustion efficiency,
reducing the emission of particulate matter (soot). The most ideal case is when no
soot is emitted from the flame. There exist two possibilities for the scenario to occur:
(i) soot is formed in the flame, but is completely oxidized; this is the case with a non-
sooting candle flame [131], where soot nucleates on the rich side, but is consumed
on the lean side of the flame, and the products of combustion are gaseous; (ii) soot
nucleation does not occur; this is a possibility even in nonpremixed flames, under
some specific configurations such as flame strain rate, low pressure, etc. [143, 145].
The blue whirl is a regime where soot does not nucleate, and this is attributed to
a combination of partial premixing and strain rate [64, 65]. This chapter discusses
the fluid dynamics of transition from a fire whirl to a blue whirl.
127
5.1 Overview
The relative influence of circulation and buoyancy on fire whirls (FW), blue
whirls (BW), and the transition between these regimes of a whirling flame is inves-
tigated using a combination of experimental data and scaling analyses. FWs are
whirling, turbulent, cylindrical yellow (sooting) flame structures that form naturally
in fires and are here created in laboratory experiments. In contrast, a BW is a lam-
inar, blue flame (non-sooting) with an inverted conical shape. Prior measurements
of the circulation and heat-release rate are now combined with additional measure-
ments of the flame geometry, particularly the width and height, to provide char-
acteristic length scales for these flames. Using these, a nondimensional circulation
(??) and heat-release rate (Q??f f ) were defined and shown to correspond to azimuthal
(buoyancy driven) and axial momenta, respectively. The ratio W? = ??f/Q??f , a
quantity analogous to the swirl number used to characterize swirling jets, was eval-
uated for FWs and BWs. For FWs, W? < 1, so that axial momentum is greater
than azimuthal momentum and the flame is dominated by buoyant momentum. For
BWs, W? > 1, so that the flame is circulation dominated. This is argued to be
consistent with vortex breakdown being an important part of the transition of FWs
to BWs. This work presents a basis for predicting when a BW will form and remain
a stable regime.
128
5.2 Introduction
The interaction of reacting buoyant plumes and swirling flow fields gives rise
to structures called fire whirls. The formation and dynamics of fire whirls are im-
portant because of the devastating effects they can have in urban and wildland
fires [8,47,48]. Both the structure and the dynamic behavior of fire whirls are domi-
nated by the effects of buoyancy, which arises from heat release, and by circulation,
determined by the level of ambient swirl. In laboratory-scale experimental investiga-
tions, the combination of heat-release rate (Q?) and circulation (?) can be controlled
to determine the shapes of fire whirls, which have been classified into a number of
regimes [92].
The blue whirl is a small, soot-free flame that was first observed in an exper-
imental study of fire whirls formed on a water surface [64]. The transition to the
blue whirl occurred naturally, as air was entrained through tangential inlets and
without the aid of externally forced air entrainment. The fire whirls that exist be-
fore blue-whirl formation are much larger in height than the blue whirl and exist at
the laminar-turbulent transition flame regime, evidenced by wrinkled laminar flames
at the flame base. In contrast, the blue whirl regime is characterized by a much
smaller length scale (see Figure 5.1) and is a laminar flame with no visual or aural
indications of turbulence [64]. The unexpected transition from a fire whirl to a blue
whirl involves laminarization of the flow, which suggests that the interplay between
Q? and ? is different from that in fire whirls.
Lei et al. [92] showed the influence of Q? and ? on the shape of fire whirls formed
129
130
Figure 5.1: Images of fire whirl regimes formed at a gap size of S = 35 mm. The experimental configuration is shown in Figure
5.4. The fuel burning rate, V? , varies from (A) to (I) as 0.5, 1.1, 1.5, 2, 3, 4.5, 6, 8 and 10 ml/min. Panels (A)?(B) show BWs,
(C)?(E) show TBWs, and (F)?(I) show FWs.
in an Emmons-type [8] fire-whirl apparatus where swirl is generated by a spinning
mesh around a central flame. Their Q? ? ? map showed the strong influence of ?
on the overall shape of fire whirls, especially on the flame shape at the base. This
map showed an ?extinction limit jump,? a sharp increase in the threshold Q? below
which fire whirls were extinguished for all values of ?. In recent work, the Q? ? ?
map was extended to include the blue whirl regime [24]. This extended map showed
that the combinations of Q? and ? that led to blue-whirl formation was within the
extinction limit defined in [92]. The formation of blue whirls in this region also
requires experimental apparatus to satisfy surface-boundary conditions [66] that
were absent in previous work.
In the literature, scaling analyses have often been used to identify parameters
and quantities to describe the occurrence of fire whirls [5, 11, 12, 228]. The recent
extension of this analysis to blue whirls [24] showed that blue whirls exist in a
regime distinct from traditional fire whirls. The transition from the fire whirl to
the blue whirl was previously hypothesized to be the result of the onset of vortex
breakdown [23, 64], which is characterized by the presence of a recirculation zone
(RZ) [229]. Subsequently, the presence of a RZ was shown to exist in the blue
whirl [66], qualitatively confirming these suggestions. The presence of the RZ was
observed using streaks of incandescent soot within the conical region of the flame.
Since soot particles are not present in a stable blue whirl, the RZ is visible fleetingly
during the transition process.
The onset of vortex breakdown is driven by the relative magnitude of local
momenta in the axial and azimuthal directions. Different types of vortex breakdown
131
Figure 5.2: Images of the blue whirl where the RZ is visualized by streaks of incan-
descent soot particles. Images (A) and (B) were captured with exposure times of
1/60 s and 1/100 s, respectively.
can occur in both nonreacting [230,231] and reacting flows [57,59]. The momenta in
these directions were compared using a scalar quantity, called the swirl number, S,
defined as the ratio of axial flux of azimuthal momentum to the axial flux of axial
momentum [232]
? R?u 2S 0 z u? r dr= R
R u2z r dr0
where uz and u? are the axial and azimuthal velocities, and R is the radius of the
vortex core. For an experimental apparatus, however, the definition is simplified
and defined based on the geometry of the swirler, with different expressions for
axial and radial swirlers [59]. In swirl burners, the RZ develops at a threshold value
of S ? 0.6 [232].
In fire whirls, the axial and azimuthal momenta are represented by buoyancy
132
and circulation, respectively. The purpose of this work is to test this hypothesis:
The flow field of the blue whirl is a state of vortex breakdown, that occurs when the
local azimuthal momentum becomes higher than the axial momentum, and this blue
whirl retains properties of the bubble mode. We approach this from the experimental
point of view by reinvestigating the scaling, which helps us quantitatively distinguish
between the fire- and blue-whirl regimes. Through this approach, the entrainment
conditions required to generate the blue whirl at other length scales are also explored.
Here, we use the raw data for Q? and ?, taken from previous experiments, and
define new nondimensional quantities to quantify the role played by buoyancy and
circulation in whirling-flames. These quantities are of the same order of magni-
tude and can be used to differentiate between buoyancy- and circulation-dominated
regimes. A nondimensional ratio analogous to S is used to establish a threshold
value for the onset of vortex breakdown in whirling flames, corresponding to forma-
tion of the blue whirl. We find that when the flow field is dominated by buoyancy,
the traditional fire whirl develops, and when circulation dominates, there is a tran-
sition to the blue whirl. The scaling analysis is used to reinterpret fire whirl data in
the literature to give an explanation of why this unique transition was not discovered
in previous fire whirl experiments.
5.2.1 Background
To date, formation of a blue whirl has always been preceded by a fire whirl. In
addition, a blue whirl can be formed from a variety of different liquid hydrocarbon
133
fuels. Experimental measurements show that it has peak temperatures of about
2000 K [22, 31, 72] and can also be formed on other types of smooth surfaces such
as metals [66]. There is intense combustion in the bright blue ring, as evidenced by
high concentrations of OH, OH* and CH* radicals in this region [31,66].
Recent experiments [24] allow us to extend previous scaling analyses to include
the blue whirl, providing useful context to the material in the current article. In [24],
the experimental setup was a fixed-frame fire-whirl apparatus formed by two semi-
cylindrical quartz segments. The semi-cylinders form an enclosure within which
combustion occurs, and the offset distance between the pieces creates a gap that
allows air entrainment into the enclosure. Various combinations of fuel flow rate
(V? ) and gap size (S) were used to generate fire whirls and blue whirls. The flow
velocity into the enclosure was measured at selected axial (vertical) locations using
a DANTEC 54T42 Mini CTA anemometer attached to a DANTEC 55P16 hot-wire
probe, which was positioned at the center of one gap. These velocity measurements
were used to find an approximate circulation for each flame regime according to
? = ?U?DC , where U? is the tangential velocity measured at the inlet, and DC
is the diameter of the enclosure. Assuming complete combustion of the supplied
fuel, the heat-release rate was calculated as Q? = V? ?hc ?f , where ?hc is the lower
heating value (LHV) and ?f is the density of liquid n-heptane.
The combination of S and V? influences the shape of the flame, and three
flame regimes were identified: the blue whirl (BW), the stable fire whirl (FW), and
the transitional blue whirl (TBW). Images of the different regimes for S = 35 mm
are shown in Figure 5.1, and the full data set from [24] is summarized graphically
134
in Figure 5.3. For high values of V? , unsteady FWs were formed with very large
diameters, termed as Large Fire Whirls (LFWs). These impeded safe operation
of the hot-wire anemometer and were thus avoided in experiments. These raw
data were used in a scaling analysis to determine nondimensional quantities for
global scaling using the apparatus diameter (DC) as the characteristic length scale.
Although the apparatus diameter or gap size are not intrinsic properties of the FW
itself, they have been used in the literature [11, 12] for scaling purposes since they
influence ambient circulation in FW apparatus with natural air entrainment.
This analysis showed that the BW was close to the extinction limit defined
previously [92]. The BW formation limits, defined on the basis of Q? and ?, as well as
the relationship between nondimensional heat-release rate (Q??D) and nondimensional
circulation (??D), were both sensitive to the gap size between the half-cylinders. The
extinction limit for FW regimes was extended by the presence of the BW, and this
was attributed to the experimental conditions at the bottom surface over which they
formed [66].
Using fixed length scales (such as apparatus or burner diameter) in scaling
analyses [92, 97] results in quantities that denote general large-scale effects. For
instance, the predictions of flame height of FWs for known values of ? and Q?. The
use of characteristic length scales derived from flame geometry represent effects of
the local flow field. In the present work, the raw data for ? and Q? is combined with
new data on flame geometry. To understand the local effects and the controlling
factors in the transition to the BW, flame height, H, and flame width, wf , are used
as characteristic length scales to normalize ? and Q?.
135
Figure 5.3: Influence of S and V? on the fire whirl regime, adapted from [24].
The approach presented in this article was taken to help quantify the effects
of the primary competing forces in whirling flames, buoyancy and circulation. By
incorporating H and wf into ? and Q?, local buoyancy and circulation in the flame
are represented by their respective nondimensional quantities, Q??f and ?
?
f . Their
ratio then present a basis for distinguishing buoyancy- and circulation-dominated
regimes in whirling flames.
136
5.3 Experimental Methods
The experimental apparatus (Figure 5.4), similar to that used in previous
work [22,24,64,66], consisted of two quartz half-cylinders (310 mm diameter, 600 mm
height) suspended on an aluminum frame. The two quartz pieces were positioned
over a water pan and offset from each other, forming gaps for natural entrainment
of air. Liquid fuel, here n-heptane, was supplied to the water surface using a syringe
pump at a constant volumetric flow rate. Upon ignition of the fuel, the buoyant
flow due to heat release within the enclosure draws air into the enclosure through
the two gaps. For all experiments, a short ?overlap? region was constructed from
thin sheets of aluminum to form a channel at each inlet (see Figure 5.4). The effect
of this inlet channel is discussed in section 5.4. All data were obtained with these
overlapping inlet channels.
A typical experiment involved the injection of 10 ml of fuel onto the water
surface, followed by ignition using a propane torch. Initially, a pool fire was formed
and lasted for a few seconds before evolving to a FW. Upon transition to a fire
whirl, fuel was supplied at a constant rate using the syringe pump. After about 60
s, the rate of fuel supply (from the syringe pump) and consumption (by the FW)
were nearly equal, and a stable FW or BW formed, depending on the gap size and
fuel supply rate. For a given gap size, S, the regime of the FW depended on the fuel
supply rate, V? , which was varied in the range of [0.5, 10.0] ml/min. S was varied in
the range [15.0, 55.0] mm. Due to the natural-entrainment configuration, the only
regimes that formed resulted from a balance between these parameters.
137
Figure 5.4: Schematic of experimental apparatus.
Measurements were made of flame geometry for each regime. Videos (1280 x
720 pixel resolution) of each experiment were recorded at 60 fps for 60 s using a
Sony RX10II, with an aperture of f/5.6. The flame width (wf ) and flame height
(H) were obtained by averaging the flame contour on multiple frames. From each
60 s video, three separate periods of 0?12 s, 24?36 s and 48?60 s were chosen, and
720 individual frames were extracted from each period. The resulting 2160 RGB
images for each experimental condition were converted into grayscale images, which
were then converted to a binary format in ImageJ [233] using Otsu?s method [234]
to determine a binarization threshold. This method was reliable since exposure in
the videos was set such that there was a dark background in contrast to the bright
flame, resulting in maximum intensity separation at the edge of the flame sheet. The
mean binary value for each pixel was obtained by stacking all 720 binary images for
each time period. Dividing the mean binary value by 256 (the range of intensity
values in gray scale), the probability of the flame appearance for each pixel was
138
determined, giving the probability contours of the flame. The mean flame contour
was defined by a probability of 0.5 based on previous literature [235].
Based on this mean contour, the flame width measured at the widest cross-
section, was found ?10 mm above the water surface for FWs, and the width of the
bright blue ring (also called vortex rim in [22]) was used as the flame width for both
the BW and TBW. The flame height was measured at the highest position of the
continuous flame region above the water surface. For each time period, one set of
wf and H was obtained, and an overall average was obtained by calculating a mean
from the values for the three time periods recorded and binarized.
139
5.4 Flame Geometry
Figures 5.5 and 5.6 shows the variation of the flame geometry parameters, wf
and H with both ? and Q?. The narrow error bars for the FW and BW regimes
reflects the relative stability of their flame geometry during the 60 s period analyzed
(Figures 5.5A and 5.6A). TBWs, on the other hand, show significant changes in
flame geometry parameters due to their repeated transitions between the FW and
BW regimes. The FW shows the highest values of wf , which is not a strong function
of ? or Q?. The dashed line in Figure 5.5B shows a slight decrease from the BW to
the TBW, followed by an increase for FWs.
The value of H, however, varies with both ? and Q?. The increase in H with
? is shown in Figure 5.6A. The curve is approximated by the fit in Equation 5.1,
which has a R2 value of 0.55. The fit in Equation 5.1 does not include data points for
S = 15 mm in Figure 5.6A. The data points for S = 15 mm (solid square markers)
follow a trend different from the other conditions. This bifurcation behavior was
discussed in previous work [24]. The relationship between H and Q?, shown in Figure
5.6B, is given by Equation 5.2, which has a R2 (coefficient of determination) value
of 0.96.
H = 52.03 ?3.89 (5.1)
H = 2.91 Q?1.19 (5.2)
140
Figure 5.5: Variation of wf (or bw) with ? and Q?. Only a very minimal increase in
wf is seen with Q?.
Figure 5.6: Variation of H with ? and Q?. The hashed section in panels B and D
indicates the limited variation in H for the BW. In panel B, flame regimes at S = 15
mm are not considered for the curve-fit shown and follow a different trend. This
difference at low S? is discussed in [24].
141
Figure 5.7: Variation of H with Q?, shown for cases without (A) and with (B) the
added inlet channels for the BW. The vertical dashed lines, green (left) and magenta
(right), indicate limits of extinction and transition, respectively.
The effect of the inlet channels (see Figure 5.4) on the fluctuation in H for BWs
is shown in Figure 5.7. The BW is particularly sensitive to ambient perturbations,
which cause H to fluctuate significantly when the inlet channel is not present. This
results in the large variations in H, seen in Figure 5.7A, particularly near the ex-
tinction and transition limits. When the inlet channel was present, the variation in
H for BWs decreased even in conditions close to the extinction and transition limits
(Figure 5.7B). Furthermore, the BW formation limits were also extended when the
inlet channels are present. The differences between panels (A) and (B) in Figure 5.7
reflect the role of the inlet channels in stabilizing the BW. The experimental data
obtained with the inlet channel was used below in the scaling analysis since they
correspond to more stable experimental conditions.
142
5.5 Dimensional Analysis
Based on a review of previous scaling methods used for FWs [5], the parameters
governing circulation (?) of a FW in a fixed-frame type apparatus are
( )
? = f1 Q?, S, wf , H,DC , T0, ?0,?T,??, Cp,0, g (5.3)
where ?, Q?, S, wf , H and DC are as defined previously. The quantities T0, ?0,
?T , ??0, Cp,0 and g are ambient temperature, ambient density, specific heat of air
at T0, and gravitational acceleration. Equation 5.3 assumes that (i) circulation is
independent of axial location of the FW, (ii) viscosity is negligible relative to inertial
and buoyant forces, (iii) heat conduction and mass diffusion terms are negligible,
and (iv) combustion is infinitely fast and steady.
As shown in Figure 5.1, the geometry of the flame may be used to distinguish
among the different flame regimes. Specifically, H strongly depends on ? and Q?
(Figure 5.6). Using H and wf as characteristic length scales, DC , T0, ?0 and g were
chosen as the basic physical parameters to apply the Buckingham?? theorem [236]
on Equation 5.3, yielding 8 nondimensional ? terms shown below.
( ) ( )
? ? wf ?T ?? g wf?f , H , = f
? ?
2 Q?f , S , , , (5.4)DC T0 ?0 Cp,0?T
143
? ?
The dimensionless circulation is defined as ??f = ? /(DC gH) = (?U?/ gH),
which is analogous to the Froude number, defined as Fr = (U2/gd). The quantity
?
Q??f= Q?/(Cp,0?T? w
2
0 f gH) is the dimensionless heat-release rate, representing the
?
ratio of the actual heat-release to a reference combustion enthalpy (V? /(w2f gH)),
where the subscript ?f ? denotes normalization by a flame dimension. In the quantity
Q??
?
f , the denominator (C
2
p,0 ?T ?0wf gH) denotes the volumetric distribution of the
heat-release rate for a non-swirling diffusion flame of width wf and height H. Thus,
Q??f reflects an enhancement in combustion intensity due to the presence of swirl,
and also represents the buoyancy due to heat release. This is discussed further in
section 5.6.
The ratio S? = S/DC is the dimensionless gap size, a geometric feature of
the setup. The quantity (wf/DC) is nearly constant for a given regime (see Figure
5.5). The quantities (?T/T0) and (??/?0) can be assumed to be nearly constant for
most fires [9], and (g wf/Cp,0 ?T ) indicates the ratio of potential energy to thermal
energy, which is small enough to be neglected here. Hence, Equation 5.4 reduces to
( ) ( )
??f , H
? = f3 Q?
? ?
f , S (5.5)
H?
(
= f ?? ?
)
, Q??4 f f , S (5.6)
where H?=H/wf . The different experimental data and nondimensional quantities
for each flame regime are shown in Table 5.1.
According to Equation 5.6, H? depends on ??f , Q?
?
f and S
?. Figure 5.8A shows
144
Table 5.1: Experimental data and nondimensional quantities for the BW, TBW,
FW and LFW regimes.
S Flame D S?C ? Q? wf H ?
?
f Q?
?
f
[mm] Regime [m] [-] [m2/s] [kW] [cm] [cm] [-] [-]
BW 0.685 0.436 2.314 2.56 4.1 1.21
BW 0.726 0.599 1.81 1.83 5.15 3.22
15 TBW 0.333 0.05 0.847 1.634 1.934 11.74 2.37 3.04
FW 0.918 2.451 2.532 20.69 1.94 2.73
FW 0.969 3.268 3.033 27.62 1.77 2.19
BW 0.479 0.327 2.345 2.056 3.11 0.99
BW 0.521 0.436 2.037 2.329 3.18 1.64
25 TBW 0.343 0.073 0.631 1.003 1.786 10.574 1.81 2.3
TBW 0.686 1.091 2.035 11.554 1.88 1.85
FW 0.741 1.178 3.174 32.495 1.21 0.67
FW 0.739 1.175 3.142 36.943 1.31 0.64
BW 0.503 0.436 1.625 1.951 3.26 2.81
FW 0.635 1.634 2.024 16.197 1.43 3.22
35 FW 0.353 0.1 0.715 2.451 2.846 30.198 1.18 1.79
FW 0.731 3.268 3.126 34.813 1.12 1.84
FW 0.744 4.357 3.405 39.551 1.07 1.94
BW 0.433 0.327 2.609 2.28 2.52 0.76
BW 0.581 0.599 2.762 2.635 3.15 1.15
45 TBW 0.363 0.124 0.669 1.089 1.382 13.43 1.61 3.71
FW 0.743 1.634 1.899 19.9 1.46 3.3
FW 0.79 2.451 2.391 24.265 1.41 2.83
FW 0.789 3.268 2.701 30.543 1.26 2.63
BW 0.593 0.599 1.611 1.413 4.27 4.62
BW 0.636 0.708 2.596 2.208 3.67 1.68
55 TBW 0.373 0.15 0.623 1.089 1.586 15.172 1.37 2.65
FW 0.69 1.634 1.892 17.692 1.41 3.52
FW 0.758 2.451 2.627 30.561 1.17 2.09
2.62 25 5.6 117 0.43 2.39
3.22 50 7.6 201 0.4 1.98
4.12 100 9.4 316 0.41 2.07
30 [7] LFW 1.8 0.111 4.9 150 10.4 419 0.42 2.2
5.87 200 11.1 500 0.47 2.36
6.27 250 12 543 0.48 2.42
6.6 300 12.7 553 0.5 2.57
145
that Q??f has minimal influence on H
?, in that H? varies only when the flame regime
changes. Also, for each flame regime, the influence of Q??f is limited, particularly for
the BW regime. For a given flame regime, H? does not show any trend with S?.
The quantity H?, however, does depend on ??f , as shown in Figure 5.8B. Figure
5.8A shows that there is a difference of an order magnitude in H? between the BW
and FW regimes. The solid line in Figure 5.8B shows the power-law relationship
between H? and ??f , approximated by Equation 5.7 with an R
2 of 0.9.
H? = 15.49/??2f (5.7)
When compared to FWs, BWs form in a region of relatively high ??f and low H
?
(Figure 5.8B). BW formation occurs for ??f ? [2.4, 5.2], while FWs generally form
in a much narrower region where ??f ? [1.0, 1.4]. The FWs for S? = 0.045 deviate
slightly from this trend and occur closer to ??f ? 2. Below ??f = 2.4, which is the
BW transition limit in Figure 5.8B, H? begins to increase rapidly. Generally, these
results show that the dimensionless circulation, ??f , has a significant impact on H
?,
and appears in this analysis as an important parameter to understand the transition
between the FW and BW flame regimes.
The LFW limit is near ??f = 1. The LFW was unsteady and its diameter
was comparable to that of the enclosure, which did not permit any measurements.
Consequently, below we use data for LFWs from the literature to calculate the
required nondimensional quantities. In the following section, we will focus on the
relationship between ??f and Q?
?
f , both of which are O(1).
146
Figure 5.8: (Top) Variation of H? with Q??f . The range of H
? for the BW and
FW do not overlap, showing that it varies only when the regime changes. The
colored stripes denote the range of H? for each regime. S? does not influence H?,
particularly in the BW regime. (Bottom) Variation of ?? with H?f , with the different
transition limits shown as dotted lines.
147
5.6 Discussion
The three nondimensional quantities, ?? ?f , Q?f and H
? have different relation-
ships for the FW and BW regimes and may be used to provide insight on the local
effects governing the transition from FWs to BWs. Going from the BW to the FW,
while wf is relatively constant with V? , H increases continuously. Thus, H and H
?
distinguish these regimes. The near-linear increase in H with V? is similar in behav-
ior to laminar jet diffusion flames [81]. Within the BW regime, however, H varies
little with V? . Figures 5.5B and 5.6B show that the BW forms in a region of low
Q?. Since entrainment velocity is directly proportional to Q? (see Fig.5 in [24]), this
leads to low ? for BWs. Q??f , however, lies in the same range for both FWs and BWs
(Figure 5.8A).
Figure 5.6A indicates that the BW forms in a region where ??f is relatively high.
As ??f decreases, there is a steep increase in H
?, leading to the FW. With further
decrease in ??f , the effect of circulation on the buoyant plume reduces. LFWs are
formed when ??f ? (0, 1) and flame height varies roughly linearly with circulation
[12, 92]. Eventually, when ??f = 0, a free-convection pool-fire is formed [80, 102].
Here, H? depends on other factors such as Q? and the pool diameter [105,237].
Plotting Q??f vs. ?
?
f allows for the different regimes to be more clearly distin-
guished, presented in Figure 5.9. Since measurements of the LFW (at ??f < 1) are
not available from this work, data of LFWs from the literature are used for com-
parison. Fire whirls similar to the LFW were investigated by Lei et al. [7] using a
propane burner in a square fixed-frame setup. The LFWs in their experiments form
148
Figure 5.9: ??f as a function of Q?
?
f for the different regimes in this study. The blue
markers correspond to BWs, magenta to TBWs, yellow to FWs, and black to LFWs.
The dotted blue line represents ?? = Q??f f , roughly the region where the FW?BW
occurs. Quantities for LFWs were calculated from raw data in Table 1 of [7], whose
experiments used a propane burner within a square cross-section setup, with natural
air entrainment at S? = 0.111 (calculated based on the definition in this study) and
Q? ? [25, 300] kW. Extinction occurs in the region where ??f > 4Q??f , and FW?LFW
transition occurs when Q??f > 3?
?
f , represented by the red line.
in a narrow range of ??f , lower than the FWs in this study, and are also included for
comparison in Figures 5.9 and 5.10.
The quantities Q?? and ??f f are of the same order and represent buoyancy and
circulation, respectively. The dashed line in Figure 5.9 represents ?? ?f = Q?f , where
the influence of circulation and buoyancy is roughly balanced. For the flame regimes
above this line, circulation dominates (i.e., ?? > Q??f f ), and for those below, buoyancy
dominates (i.e., ??f < Q?
?
f ). In general, the BW lies above the ?
?
f = Q?
?
f line, and FWs
in the region below. Data points for LFWs are well within the buoyancy-dominated
region. Figure 5.10A and Table 5.2 show that for naturally-entrained FWs, with
149
increasing scale, buoyancy tends to dominate circulation (see markers for LFW).
This is a consequence of larger Q?, which for liquid-fueled FWs is partially controlled
by the fuel pool diameter.
Of the five data points available for the TBW regime, three lie between the FW
and BW regimes. This transition regime is expected to occur in the neighborhood
of ?? = Q??f f , although Figure 5.9 shows two instances deviating more towards the
FW side (these deviations are observed for one instance of the BW at S? = 0.147
and two FWs at S? = 0.073). This spread in TBW data is expected from the large
fluctuations in H in this regime (see Figures 5.5 and 5.6). Additionally, only five
data points exist for the TBW regime in our work, and the trend may become more
apparent when there is more data available.
The quantity (??f )
2 is the Froude number, Fr. While Fr represents the com-
petition between buoyant and external momentum, it does not fully represent the
effect of circulation on H. Chuah et al. [9] pointed out that the value of Ro (Rossby
number) in relation to Fr was important for quantifying the effect of circulation on
burning rate in whirling flames. This is especially applicable to configurations with
natural air entrainment where ? and Q? are not independent.
While ??f varies significantly for the BW and FW regimes, it does not help
in comparing the local relationship between axial and buoyant momenta. Thus for
whirling flames, it is more useful to represent the competition between circulation
(tangential component) and buoyancy (axial component) in the form of the ratio,
W? = (??f/Q??f ), the whirl number. The value of the whirl number for the different
regimes is shown in Table 5.2. Figure 5.10(B) shows the relationship between H?
150
with (W?) for all the regimes. The relationship is approximated by a least-squares
fit as H? = 4.14/(W?)1.32. This graph shows that transition to the BW occurs in
the neighborhood of W? = 1, when circulation begins to dominate buoyancy. In
other reacting or nonreacting swirl flows, this condition leads to the formation of
a recirculation zone (RZ), and to vortex breakdown [57, 232]. In swirl burners, the
threshold value for RZ formation is S = 0.6 [59].
Earlier work showed the presence of a RZ within the BW and suggested that
the shape of the BW regime may be governed by the bubble mode of vortex break-
down [22,181]. While the RZ is visible during the transition process, it is not visible
in a stable BW due to the absence of soot tracers [66]. Assuming that the develop-
ment of a RZ is necessary for BW formation [23],W? is analogous to S in predicting
the onset of vortex breakdown in whirling flames, and may therefore represent an
appropriate nondimensional scale to distinguish whirling flames. Table 5.2 shows
that the threshold value of W? leading to vortex breakdown and BW formation is
? 1.
The large difference in the values ofW? for FWs and BWs provides hints as to
why the BW regime was not observed in previous FW experiments. In addition to
the smooth bottom boundary without obstructions to the incoming radial flow [66], a
circulation-dominated regime (??f > Q?
?
f ) is required for transition to the BW. FWs
subject to strong vorticity have been studied previously [9, 10] and it was found
that elongation in FW flame length, when compared to a pool fire, cannot be fully
attributed to increasing burning rates. This suggests that that the vortex structure,
within which fuel fractions are high, plays a significant role in the increased flame
151
Table 5.2: Values of W? = (?? ?f / Q?f ) for the BW, TBW, FW and LFW regimes.
W? = ?? ?f/Q?f
S?
BW TBW FW LFW
0.045 3.39 0.78 0.71
(S = 15 mm) 1.6 0.81
0.073 3.15 0.78 1.82
(S = 25 mm) 1.94 1.02 1.78
0.099 1.16 0.44
(S = 35 mm) 0.66
0.61
0.55
0.124 3.33 0.43 0.44
(S = 45 mm) 2.73 0.5
0.48
0.147 0.92 0.52 0.4
(S = 55 mm) 2.18 0.56
0.111 0.18
(S = 300 mm) 0.2
From [7] 0.2
0.19
0.2
0.2
0.19
152
Figure 5.10: Comparison of experimental data in this work with those for LFWs in
the literature [7]. (Top) The LFW regime occurs when ??f ? (0, 1). The value of H?
for LFWs in the literature are many times that of the FWs in this study, and an
order magnitude higher than the BWs. (Bottom) The FW-to-BW transition occurs
around W? ? 1, in the circulation-dominated regime.
height.
In the case of transition from FWs to BWs, increasing ??f results in suppression
of H?, eventually leading to BW formation (Figures 5.8B and 5.10). The BW does
not show much variation in H? (see Figure 5.8A), which may be attributed to
the existence of a RZ upon vortex breakdown. The RZ potentially aids in better
mixing, leading to a smaller volume required for the reaction to occur and causing
suppression of H?. This is similar to the reduction in flame height of laminar jet
diffusion flames upon transition to turbulence [81].
Coenen et al. [23] noted that for a given fuel pool diameter, BW formation
required evaporation of fuel from a fraction of the total pool area such that Q? and H
decrease simultaneously. The contraction in evaporation area is a consequence of in-
tensification in vorticity as the width of the FW vortex core reduces. This behavior
153
agrees with the data from the present work, where decreasing H? and Q??f is enabled
by increasing ??f . Since circulation is conserved radially [8], a reduced vortex core
diameter allows tangential momentum to overcome axial momentum locally, result-
ing in the conditions required for a vortex breakdown bubble to form and results
in the BW. The limits presented in this work are determined from experiments on
regimes formed by natural air entrainment. Controlling the transition process or
further expanding the envelope of BW formation conditions, if possible by means of
external forced entrainment, will require independent control of Q?, ? and the fuel
pool area to ensure ??f > Q?
?
f .
For all of the flame regimes, the influence of ? and Q? is weak on wf but strong
on H. The flame dimensions are determined by the axial flux of fuel vapor, which is
determined by V? , and thus the diameter of the fuel pool over the water surface. For
a given flame regime, the fuel mass flux upon evaporation is nearly constant, and the
increase in specific volume upon combustion of the fuel vapors shows a near-linear
growth in H rather than wf . This is similar to the behavior of jet diffusion flames,
for which the flame height in the laminar regime depends on the volumetric flow
rate of fuel from the fuel port, before a transition to turbulence reduces the flame
height (see Fig. 5.10 in [81]).
154
5.7 Summary
The transition from fire whirls to blue whirls was studied by a scaling approach.
Using raw data from previous work and new measurements of flame geometry, an
investigation of the scaling parameters now allows us to understand the relative
influences of circulation and buoyancy in determining the different fire whirl regimes.
In addition to data from previous work, new measurements of the flame geometry
(height, H, and width, wf ) were obtained from videos.
The flame width did not vary significantly with circulation (?) or heat-release
rate (Q?) for the different flame regimes. The relationship between H and ? was
exponential, and between H and Q? was nearly linear. Using H and wf as the
characteristic length scales, two primary nondimensional quantities were defined:
dimensionless heat-release rate, Q??f , and dimensionless circulation, ?
?
f . These quan-
tities were of the same order of magnitude, and represent the role of buoyancy and
circulation in each regime.
The ratio of these quantities, W? = (?? ?f/Q?f ), the whirl number, represents
the relative influences of circulation and buoyancy on the flame. This is analogous to
the swirl number for swirling jets, where the relative magnitudes of axial Reynolds
number and azimuthal swirl determine the flame regime. The influence of this ratio
on the tranistion may be summarized as
155
? ?
??????? ?< 1, buoyancy dominated; FW ?????
W? = ?????
?
? 1, transitional; TBW
?? ?????> 1, circulation dominated; BW ??
A value of W? < 1 represents a flow field dominated by buoyancy, and most
fire whirls belong to this regime, with values in the range [0.4, 0.8]. Two instances
of fire whirls (referred to as ?conical fire whirls? in [92]) showed an average value
of 1.8. The value of this ratio was calculated to be ?0.2 for large fire whirls in the
literature, placing them well within the buoyancy-dominated regime.
For the blue whirls in this study, W? ? [0.9, 3.4], and generally W? > 1. This
indicates a flow regime where circulation dominates over the buoyancy locally. The
transitional blue whirl is defined as a regime that continuously switches between the
blue and fire whirl, and may theoretically be expected to occur when the ratio is
in the neighborhood of 1, where the effects of buoyancy and circulation are roughly
equal. The number of data points for transitional blue whirls was limited and
calculated to be in the range of [0.43, 1]. This range extends more towards the
buoyancy-dominated side, and is attributed to the large fluctuations in H, caused
by repeated alternation between the fire whirl and blue whirl regimes.
The transition from the fire whirl to the blue whirl occurs at a threshold value
of W? = 1, roughly where circulation begins to dominate buoyancy. This favors the
formation of a recirculation zone, leading to the onset of vortex breakdown. This is
one reason why the blue whirl was not discovered in previous apparatus. Previous
fire whirl apparatus generated only buoyancy-dominated regimes by design, and did
156
Figure 5.11: Proposed map of some whirling flames regimes as functions of nondi-
mensional circulation and heat-release rate, as defined in this chapter.
not provide the strong radial inflow at the bottom boundary surface [66]. The right
combination of low Q? and natural entrainment at the right length scale provided
the optimal conditions for its discovery in the experiments performed by Xiao et
al. [64]. With the results presented here, future experiments of the blue whirl can
be designed with forced entrainment apparatus to control Q? and ? independently
at different length scales to directly hone in on the BW regime.
This work provides a quantitative basis to explain the factors controlling the
transition from the fire whirl to the blue whirl and lays the foundation for future
experimental and numerical efforts on fire whirls and blue whirls. Open questions
still remain regarding the flow structure within the recirculation zone and the mixing
conditions that stabilize the flame at the bright blue ring. Future measurements
using non-intrusive laser diagnostics of the flow-field in the vicinity of the flame
157
and the distribution of radical species will be helpful in further understanding these
better. Finally, the nondimensional quantities presented here may be used to obtain
a complete map of all whirling flame regimes as shown in Figure 5.11.
5.8 Acknowledgments
This chapter, in part, was submitted as a research manuscript to Physical
Review Fluids, titled ?Effects of circulation and buoyancy on the transition from a
fire whirl to a blue whirl,? by S.B. Hariharan, Y. Hu, M.J. Gollner and E.S. Oran,
and was accepted for publication in July 2020. The thesis author is the primary
author in this publication.
158
Chapter 6
Understanding the Blue Whirl using Optical Diagnostics
6.1 Overview
The blue whirl (BW) regime of whirling flames shows suppressed soot for-
mation. Observed as a regime that develops from a traditional fire whirl (FW), it
stabilizes over a pool of liquid fuel, it is a near-limit flame that occurs at low heat-
release rates. Initial studies focused on the general experimental conditions leading
to BW formation, the thermal structure, and the range of circulation and heat-
release rates leading to BW formation. The results in Chapter 5 showed that the
transition from the FW to the BW was dependent on the magnitudes of tangential
and axial momentum; a circulation-dominated regime explained the occurrence of a
recirculation zone necessary for stabilizing the blue-whirl regime. It is also under-
stood that the reaction front is a ring-like structure, observed as a bright, blue ring,
which is lifted from the fuel surface and shows poloidal motion in the same sense
of the ambient swirl. Open questions regarding the flame structure and mixing still
159
remain, and a partially-premixed triple flame has been proposed [23,64].
In this chapter, these open questions are approached by studying the flame us-
ing non-intrusive optical diagnostics. BWs were formed with iso-octane and probed
using planar laser-induced fluoresecence (PLIF) and chemiluminescence. First, the
distribution of OH and PAH radicals was obtained using PLIF, which showed that
the highest concentration of ground-state OH existed in a conical structure above
the ring. The highest concentration of OH was at the ring, and diminished continu-
ously in the axial direction. Negligible presence of OH was found towards the conical
region below the ring. PAH was observed surrounding the lower conical region, with
maximum intensity immediately upstream of the blue ring.
Next, chemiluminescence images of spontaneous emission from OH* and CH*
in the BW were captured at 1 kHz. The intensity ratio of OH*/CH* around the
blue ring was computed using Abel inversion. A radial map of the intensity ratio
shows possibility for the existence of a triple flame anchored at the blue ring. The
results presented here are largely qualitative, since an exact relationship between
intensity ratio and equivalence could not be obtained. Such a calibration is necessary
to identify distinct rich and lean fronts, quantify the mixture fraction around the
ring and determine the exact location of the stoichiometric region. A method to
determine this relationship is presented using OH* and CH* images of adiabatic
methane-air flames stabilized over a 3.81? cm2 Hencken burner.
These results indicate that there is potential for the blue ring to be a triple
flames, no reaction occurs within the recirculation zone, and that mixing occurs
primarily near the blue ring. Future investigations using simultaneous imaging
160
are necessary to further understand the structure and the dynamics of the mixing
process.
161
6.2 Background
The BW is a regime of the fire whirl, first observed by Xiao et al. [64] while
studying the behavior of FWs formed in a fixed-frame setup that allowed natural
entrainment of air into the enclosure. The laminar flame structure forms directly
over a liquid fuel pool and transitions from the fire whirl when specific conditions
of buoyancy and vorticity exist (Chapter 5 and [238]). The conical shape of the
flame has been attributed to the bubble mode of vortex breakdown in reacting flow
[64], supported by visual observations of a recirculation zone within the flame [66].
Current understanding of the BW includes the experimental conditions required for
its formation [66], temperature maps across a radial cross-section of the flame [22],
and the combination of nondimensional heat-release rate and circulation that are
required for the transition to occur [24]. Open questions remain regarding the exact
nature of combustion and the importance of the blue ring region of the flame.
The nature of combustion in the BW is of interest because of negligible emis-
sions of particulate matter (soot) from the flame. Mechanisms of flame stabilization
using swirling flow fields have been studied for many decades because of their impor-
tance in jet engines and industrial combustors, particularly for stabilization when
operating lean [57,59]. These flames typically operate in a turbulent regime, where
reduction in emissions is caused by enhanced mixing between light gaseous fuels and
air that are injected as jets into the combustor [60,84]. In FWs, however, the effect
of circulation is to suppresses turbulence [8,61,95], which causes the flame height to
increase with the level of ambient circulation [8,10,11,92]. In this context, the BW
162
emits negligible soot even while burning directly above a pool of liquid fuels that
traditionally form considerable amounts of soot. It is of interest, therefore, to study
the flame structure that leads to negligible soot emissions, and the mechanisms that
permit optimal mixing between air and fuel evaporating from a liquid-fuel pool.
For FWs formed over a liquid-fuel pool, the flame diameter above the fuel
surface depends on the flow conditions immediately surrounding the fuel pool [66].
If any vertical projections are in the immediate vicinity of the fuel pool, the flame is
stabilized at the edge of the fuel pool and the FW diameter at the base is the same as
the pool diameter. If the radial boundary layer is uninterrupted and the circulation
is sufficiently high, FW diameter can be smaller than the fuel-pool diameter [47].
In the case of the BW, it is stabilized at the blue ring.
In the transition from a FW to a BW, the blue ring lifts off the fuel surface.
This is suggested to be caused by a combination of finite-rate chemistry and the
large axial momentum at the base of the flame generated by the radial collision
in the Ekman layer [23, 86]. The shape of the bottom blue conical region and
the upper purple haze are likely a consequence of the recirculation zone setup by
vortex breakdown, with the strain rates at the stagnation zone being low enough
to support a weak, inverted conical flame. Based on temperature maps and OH*
chemiluminescence, it is clear that a large fraction of the heat release occurs at the
ring [22,66]. It has also been suggested that the ring is a rotating triple flame, with
the the lower blue cone forming the rich front, the purple haze the lean front, and
the ring being stoichiometric [23,65].
163
6.2.1 Scope
While the techniques used in previous studies provide global measurements of
the flame, measurements of chemical species require non-intrusive diagnostics since
the blue whirl is very sensitive to the flow perturbations caused by probes positioned
in the vicinity of the flame [22]. In this chapter, non-intrusive optical diagnostics
are used to image the distribution of important combustion species. Using PLIF
measurements of OH and PAH, rich and lean regions around the reaction front are
identified; chemiluminescence imaging of spontaneous emission from OH* and CH*
radicals presents a qualitative picture of the triple-flame structure of the ring. A
framework for quantitatively determining equivalence ratio around the ring using
the emission intensity of OH* and CH* radicals is also presented. These results
agree with previous measurements in the literature as well as recent postulations
about the partially-premixed nature of the flame.
164
6.3 Experimental Methods
First, OH-PLIF was used to visualize ground-state OH radical distribution
in a radial cross-section of the flame. In this technique, single-photon radiation
at 282.750 nm was used to excite OH radical in the Q1(5) rotational line of the
OH A2?+ ?X2?(1,0) band, followed by fluorescence emission detection in the OH
A2?+ ? X2?(0,0)+(1,1) bands. The nature of the reaction zone is compared for
regular conical FWs and BWs using OH-PLIF. To locate the mixing region with
respect to the reaction zone, PAH-PLIF was also performed to identify polycyclic
aromatic hydrocarbons (PAH). Since iso-octane does not fluoresce, the addition of
a dopant such as toluene in small quantities is necessary to visualize fuel [239].
OH* and CH* radicals can be used to locate the reaction front, sometimes even
in the visible range [240]. Excited radicals such as OH*, CH*, C2*, CO2* contribute
the most to visible and UV radiation in hydrocarbon-air flames [187, 189]. The
concentration of these radicals and the corresponding intensity of spectral emission
are a function of pressure [190], local equivalence ratio, strain rate [191], nature
of the fuel [32, 192] etc. Local flame properties such as equivalence ratio or heat-
release rate in non-sooting flames may be calculated by evaluating the ratio of
the relative intensities of spontaneous chemiluminescence signals from the different
radicals [32, 187, 190, 192?194]. In this work, the intensity ratio of OH* and CH*
were chosen to determine the equivalence ratio in the region around the blue ring.
165
6.3.1 Apparatus
BWs were formed using iso-octane fuel in a fixed frame, self-entraining fire
whirl apparatus, similar to that used in previous work [22, 24, 64, 66]. In place of
a water layer, a flat steel surface (45?45 cm2) was used as the bottom boundary
surface over which FWs and BWs were formed. It was shown by Hariharan et al. [66]
that the presence of a water surface was not critical to cause transition from the
FW to the BW. The metallic surface was part of a hollow steel cuboid (45?45?2.5
cm3), which allowed cooling water to be passed below the surface and maintain a
constant temperature over the surface. A constant temperature reservoir was setup
using a recirculating water chiller (Fisher Scientific Recirculating Heater Isotemp
4100 H7) set to operate at a constant temperature of 36 ?C. Two half cylinders (30
cm diameter, 60 cm height) were positioned over the surface and offset such that
the gap width was 25 mm. Fuel was supplied at 0.65 ml/min to the center of the
metal surface through a threaded stainless-steel tube (3 mm inner diameter) using
a syringe pump (NE-300 JustInfusionTM Syringe Pump). The water reservoir and
the quartz half-cylinders were positioned over an optical table, and a schematic of
the setup is shown in Figure 6.1A.
The setup for optical diagnostics is shown in Figure 6.1B. For PLIF, the laser
system consisted of a 10 Hz nanosecond duration Nd:YAG laser (Continuum Power-
lite 8000) operating at 532 nm, and used to pump a dye laser (Continuum ND6000)
with circulating Rhodamine-590 dye diluted in pure methanol. The output beam
of the dye laser was frequency doubled in a beta-barium borate (BBO) crystal to
166
Figure 6.1: (A) Schematic of a fixed-frame fire whirl apparatus formed using two
half-cylinders. These were positioned slightly offset to create gaps to allow natural
entrainment of air into the enclosure. (B) Schematic of the imaging setups for chemi-
luminescence intensity and PLIF measurements. These techniques were performed
separately for each species of interest. Figure adapted from [31].
generate a UV beam at 282.750 nm with an energy of 4 mJ/pulse. The UV beam
was then guided to the blue whirl flame region using several 45? dielectric mirrors
(LEO TWB-275-285-45-UF-1025). A plano-concave cylindrical lens (?25 mm focal
length) coupled with a +400 mm focal length plano-convex lens was used to gener-
ate a laser sheet. The central portion of the laser sheet (uniform within 80%) with
dimensions of 53 mm ? 0.15 mm was used to image the distribution of OH radicals.
The emitted fluorescence was collected at a 90? angle with respect to the beam
path using an intensified CCD (ICCD) camera (Princeton Instruments PI-MAX4)
fitted with the same UV camera lens fitted with an OH bandpass filter. The ICCD
detection gate width was fixed at 100 ns to mitigate flame luminosity.
6.3.2 PLIF experiments
For OH-PLIF experiments, pure iso-octane was used as the fuel. For com-
parison of FWs and BWs, OH-PLIF was also performed on traditional cylindrical
167
FWs [8, 47]. Fire whirls formed using iso-octane form soot, which results in their
yellow color, and the blackbody radiation can interfere OH fluorescence [66]. Thus,
liquid methanol was used as the fuel to avoid soot formation. For PAH-PLIF, a
50%-50% mixture (by volume) of iso-octane and biacetyl was supplied at the same
flow rate. Addition of this dopant was necessary since pure iso-octane does not
show significant PAH fluorescence [239,241,242]. Smaller volume concentrations of
biacetyl were tested, but even with very high laser power, a concentration of at least
40% was needed to obtain strong fluorescence signals. The PLIF images obtained
from the PI-MAX4 were 14-bit, with intensity values ranging between 0 and 16383.
For both OH- and PAH-PLIF, background images were obtained. An average of
the background was then subtracted from each frame of interest to ensure sufficient
dynamic range.
6.3.3 Chemiluminescence experiments
High-speed OH* and CH* chemiluminescence imaging was performed using a
complementary metal-oxide semiconductor (CMOS) camera (Photron SA-Z) cou-
pled to a high-speed intensifier (LaVision HS-IRO). A UV camera lens (+100 mm
focal length, f/2, B. Halle Nachfl) was used for focusing chemiluminescence signals
onto the detection system. Bandpass filters, 50.8 mm in diameter, for OH* (315
nm, Semrock FF01-315/15-50) and CH* (434 nm, Semrock FF01-434/17-50) were
mounted at the front end of camera lens, allowing greater than 90% transmission
of the respective signals and blocking all other interferences. The data acquisition
168
rate was fixed at 1 kHz, with an exposure time and gain of 6 ms and 38% for OH*,
and 9 ms and 45% for CH*, respectively.
To obtain calibration images for the OH*/CH* intensity ratio, the BW appa-
ratus was replaced by a standard 38.1?38.1 mm2 Hencken burner [243]. CH4-air
flames were stabilized over the burner with the flow rate of dry air maintained at
14.28 slpm and varying the CH4 flow rate between 1.2 to 1.95 slpm to obtain equiv-
alence ratios (?) between 0.8 and 1.3. Each flame was imaged for 0.2 s with the
settings described above. Obtaining the correlation between the OH*/CH* intensity
ratio and local equivalence ratio [32] is discussed in the rest of this section.
The relationship between chemiluminescence intensity ratio (I? = OH?/CH?)
and local equivalence ratio (?) for methane and iso-octane was experimentally deter-
mined using Cassegrain optics and spectroscopic detection by Hardalupas et al. [192]
and Orain et al. [32]. The resulting expressions are given by
( )
OH?? ?(??0.6)IOct, ref = ? = 0.344 + 0.999 e
0.314 (6.1)
CH
Oct, ref
( )
? OH
? ?(??0.7)
I 0.26Meth, ref = ? = 0.597 + 2.107 e (6.2)CH
Meth, ref
While the blue whirls in this work were formed using iso-octane, methane was used
in the calibration Hencken burner. The relationship needed is that between I? and ?
for iso-octane using the optics used in this study. First, OH* and CH* measurements
of methane-air flames over the Hencken burner were obtained to calculate I?Meth,cal
at different values of ?, given as
169
( )
OH?
I?Meth,cal = ? (6.3)CH
Meth,cal
The quantity I?ref is defined as
?
? IOct,refIref = (6.4)I?Meth,ref
and represents the behavior of the imaging system to methane-air and iso-octane-air
flames. Combining the measured response I? ?Meth,cal with the reference response Iref ,
we obtain the relationship
I? ? ?Oct,cal = IrefIMeth,cal (6.5)
where I?Oct,cal is the required relationship between intensity ratio and ? in the imaging
system described above. Now, by capturing images of OH* and CH* in the BW, this
function can be used to obtain local ? in the blue whirl from I?Oct,cal by extrapolation.
To calculate I?Meth,cal, a set of 200 images of the Hencken-burner flame were
captured using the OH* and CH* filters for each value of ? ? [0.8, 1.3]. An average
was obtained using the 200 images, shown in Figure 6.2 for ? = 1. For each value
of ?, the maximum intensity in the streamwise direction (in the flow direction) was
obtained to determine the pixel location of the center of the flame in the image
frame. Using this center, an average of the intensity values was calculated for an
area of 200 pixels in the transverse direction and 7 pixels in the streamwise direction.
The sensitivity of the intensity to the streamwise pixel count is discussed in Figure
170
Figure 6.2: Contour plots of 200-frame averages of (A) OH* and (B) CH*, in
methane-air flames at ? = 1 stabilized over a 3.81 cm Hencken burner.
6.6.
The average intensity values of each species for every equivalence ratio was
then normalized by the total flow rate of the air-fuel mixture and the intensifier
gain values used for each species. The relationship between I?Oct,cal and ? depended
significantly on the normalization method, and the details are discussed in section
6.4.2.
171
6.4 Results
6.4.1 PLIF: OH and PAH
The distribution of ground-state OH in the BW is shown in Figure 6.3A. The
radial cross section through the center of the flame shows the highest signal intensity
at the edge of the blue ring, the anchoring point of the flame. OH exists in a hollow,
conical region above the blue ring, and the concentration diminishes continuously in
the axial post-flame direction. Previous chemiluminescence measurements of OH*
in the BW showed significant intensity only near the blue ring, and not in the post-
flame region [66]. The absence of excited-state OH* in the post-flame region is
attributed to the lower temperatures. In contrast, the higher temperatures in the
intense reaction zone near the blue ring permit the presence of OH* radicals. As
the temperature falls in the axial direction in the post-flame region [22], the excited
state OH* radicals are not present in sufficient concentration to show chemilumi-
nescence, but the ground-state OH is still present. The approximate width of the
ring, estimated as the peak-to-peak distance in this map, is about 12.5 mm. The
lift-off distance of the ring from the fuel surface is about 10? 15 mm.
In contrast to the BW, ground-state OH is present in a cylindrical structure
in the traditional FW. Figure 6.3B shows the lower portion of a tall FW formed
with methanol. The FW forms immediately over the fuel surface, showing negligible
flame lift-off distance. This cylindrical burning structure agrees with previous mea-
surements of temperature in the FW [7,8,91,93,97]. Previously, chemiluminescence
172
Figure 6.3: Contour maps of ground-state OH in a radial plane of the (A) BW
formed with iso-octane, and (B) FW formed with methanol. (C) PAH signals in
the BW formed using a mixture of iso-octane and biacetyl. (D) Background signals
of PAH after extinction of the BW,caused by scattering of evaporating droplets of
fuel-dopant mixture. Additional figures for each species are presented in Appendix
C.
173
measurements of sooty fire whirls suggested a similar structure [66], but the data
presented here eliminates any interference from broadband soot incandescence, rein-
forcing previous hypotheses that relied only on temperature measurements. Emmons
and Ying [8] suggested that the interior of the FW was a fuel-rich region consisting
of mostly fuel vapor, with a nonpremixed flame forming the flame sheet in a cylin-
drical structure. Lei et al. [97] also measured tangential velocity in the fire whirl,
which showed that the peak values were at this flame sheet.
PAH signals from the blue whirl are shown in Figure 6.3C. The bottom of the
image frame is ?10 mm above the fuel surface. The peak-to-peak distance of PAH
signals at the bottom of the frame is ?18 mm, showing that the two streaks with
high signal intensity are in the region immediately outside the blue conical region. In
these streaks, higher concentrations are found closer the fuel surface, upto a height
of ?15 mm from the fuel surface (?5 mm from the bottom of the image frame).
This is expected to be the the approximate location of the blue ring, with high fuel
concentrations below the ring and low concentrations above it. Between 15 and 25
mm from the bottom of the frame, the signals drop progressively to background
levels. The shape of the PAH signals shows that the fuel vapors are distributed in
a contour that is similar that of the BW, surrounding it. Background PAH signals
were quite high due to substantial scattering from droplets above the fuel pool. This
was visible even after blue whirl extinction as shown in Figure 6.3D. It is supposed
that this experimental challenge was not encountered in previous work [242] where
the fuel-dopant mixture was injected as a spray.
174
6.4.2 Chemiluminescence
Individual frames of OH* and CH* chemiluminescence in the BW are shown
in Figure 6.4. Since the imaging rate is 1000 Hz, a sufficiently large number of
images exist for both OH* and CH* where the blue ring is captured as a straight,
horizontal line. The straight line represents the intensity observed as integrated
along the line-of-sight.
A number of such images were averaged, and Abel inversion was performed to
deconvolve the intensity data to the (r, ?, z) coordinate system (see Fig. 2 in [194]).
The deconvolved data was used to calculate the intensity ratio I? = (CH?/OH?)
at each r ? ? location. A map of I? for the BW is shown in Figure 6.5. Detailed
descriptions of the image processing methods including criteria for image selection,
averaging, deconvolution, alignment of OH* and CH* images, and background sub-
traction are discussed in a forthcoming publication [244].
The map of I? around the blue ring shows that, qualitatively, there exists the
possibility of a triple-structure along the ring. At the outer edge of the ring, the
values of I? suggest that the rich branch is above the ring, and the lean branch
below it. On the inner edge, however, the location of the rich and lean branches are
reversed. The ring may be considered an edge flame since it shows poloidal motion,
propagating continuously in the azimuthal direction due to the effect of ambient
circulation.
The goal is to use the ratio I? to determine the local equivalence ratio (?)
around the blue ring. This necessitates a calibrated relationship between the quan-
175
Figure 6.4: Sequence of chemiluminescence images obtained at 1000 Hz, shown in
false color. Panels (i?v) show CH*, and panels (a?e) show OH*. Processed images
with intensity values are shown in Figures C.5 and C.6.
tities I? and ?. The calibration process using premixed methane-air flames stabilized
over a Hencken burner was described in section 6.3.3. I? = (CH?/OH?) and I?Oct,cal
are related as ( )
CH?
? ? f(I
? ?1
OH Oct,cal
)
where I?Oct,cal is as defined in section 6.3.3.
The variation of OH* and CH* signal intensity at different values of ? is
shown in Figure 6.6. As shown in Figure 6.2, the methane-air flames formed over
the Hencken burner are quite thin in the streamwise direction. Five different averag-
ing methods were evaluated for their sensitivity on the value of chemiluminescence
intensity. For each method, averaging in the transverse direction was done over 200
pixels around the center of the frame. The five methods are
(i) The maximum signal intensity of each species in the streamwise direction
(ii) 1-pixel average, centered at the pixel with maximum transverse intensity at
176
Figure 6.5: Contour map of (CH?/OH?) in the BW. A radial location of r = 0
signifies a cross sectional plane through the center of the BW.
the center of the frame (area of 1? 200 px2)
(iii) 3-pixel average, with an area of 3? 200 px2
(iv) 5-pixel average, with an area of 5? 200 px2
(v) 7-pixel average, with an area of 7? 200 px2
As evident in Figure 6.6, the line-of-sight intensity values are weak functions of the
area over which the intensity data was averaged. The 7-pixel average was used to
calculate I?Meth,cal.
The relationship between I?ref and ?, calculated from Equations 6.1, 6.1 [32], is
shown in Figure 6.7A. The overall relationships between the (OH?/CH?) intensity
ratio in iso-octane and ? is shown in Figure 6.7B. The three curves represent the
three methods of normalization of the OH* and CH* intensity values (I, depicted
177
Figure 6.6: Effect of the different averaging methods on the OH* and CH* intensity
counts. The 7-pixel average was chosen for determining I? ?Meth,cal and thus, IOct,cal.
in Figure 6.6. Normalization is required to take into account the two variables in
the imaging process: intensifier gain values (G) used for each species, and the total
mass flow rate (m?H) of the fuel-air mixture flowing through the Hencken burner
at each value of ?. Additionally, normalization by the maximum value measured
was recommended by Jeong et al. [194], which helps bounds intensity ratio between
0 and 1. The maximum intensity measured (Imax) for both species was at ? = 1
(Figure 6.6). I was normalized by three methods:
Method A:
I
Imax m?H
Method B:
I
Imax m?H G
178
Figure 6.7: (Left) Variation of I?ref with ?, obtained using system responses for
methane and iso-octane in [32]. (Right) Calibration curve showing the overall re-
lationship between I?Oct,cal and ? ? [0.8, 1.3] for the three different normalization
methods.
where GOH = 38 and GCH = 45, and
Method C:
I
I 2max m?H G
The three methods resulted in the three curves shown in Figure 6.7. The curve
fit for each normalization method is a third-order polynomial of the form
I? 3 2Oct,cal(?) = a? + b? + c?+ d (6.6)
The values of a, b, c and d are shown in Table 6.1. The normalization methods
result in similar shapes of the polynomial functions, and may be considered to be
just displaced in the ordinate direction, with differences in the I?Oct,cal intercept. Dif-
179
Table 6.1: Coefficients for the third-order polynomial functions relating I?Oct,cal and
?, shown for the three methods used to normalize I.
Normalization
method a b c d R2
A 3.383 -9.056 8.291 -2.126 1
B 4.007 -10.72 9.818 -2.517 1
C 4.745 -12.7 11.63 -2.981 1
ferences in response between different normalization procedures has been previously
been shown to significantly affect the calibration curve (see Fig. 9 in [245]).
In spite of the similar shape for each of the calibration curves for the different
normalization methods, the calibration curve cannot be directly used to determine ?
in the BW. This is because the normalization methods necessitate normalization by
the total mass-flow rate to account for the population density of the reactive species,
and an arbitrary maximum, which was found at ? = 1 for both OH* and CH*.
Determining the local flow rate around the blue whirl and the arbitrary maximum
are challenges since they present an open ended problem ? no assumptions can
be made about either of these parameters since doing so will fix either the global
equivalence ratio or the stoichiometric conditions (? = 1) at some arbitrary value
of I?Oct,cal.
One potential solution is to use data from numerical simulations of the BW
[33,72] to first determine an iso-surface of ? = 1 near the blue ring, then identify the
corresponding region in the map of I?. The value of I?Oct,cal at that radial location
in the BW will represent the known value of ? = 1, which then provides a way to
quantify the distribution of ? around the blue ring. This is outside of the scope of
180
this work and requires additional experiments with simultaneous imaging to verify
the results.
181
6.5 Discussion
OH-PLIF images show that ground-state OH concentration is highest at the
blue ring, where the flame is anchored. The concentration diminishes continuously in
the post-flame region, forming a hollow conical structure. The high concentration of
OH at the ring supports previous observations of high OH* concentration there? the
high temperatures enable the existence of OH* at the ring, but as the temperature
reduces in the post-flame region (termed ?purple haze? [23, 65]), only ground-state
OH exists. Heat release is expected to occur in the haze region, but at much lower
rates than at the ring. Measurements of HCO, or simultaneous imaging of OH and
HCHO, are required to quantify the heat-release rates in the different regions.
The concentration of fuel vapor increased with axial distance and peaked at
the region immediately upstream of the blue ring. PAH-PLIF images also showed
that no fuel was present within the recirculation zone. The absence of OH and PAH
fluorescence signals in the recirculation zone indicates that the region is dominated
by products, potentially aiding the ignition process and stabilizing the flame by
acting as a flame holder.
Occasionally, small particles of soot emanating from the ring are observable
due to faint incandescence (see Appendix C, and Supplementary Figures, Figures
S.1 and S.2 ). They generally accelerate in the axial direction due to buoyancy, and
the visible incandescence reduces as the particles move through the purple haze.
The reduction in incandescence is postulated to be because of the availability of OH
to, at least partially, oxidize the soot particles. This leads to a situation similar to a
182
perfect candle flame, where soot is formed in the flame, but is completely oxidized
in the leaner regions of the flame [131,132]. Of course, the production of soot is not
continuous in the BW, owing to the partially-premixed nature of the flame.
Traditional triple flames are typically studied using opposed flow burners [246?
248], and chemiluminescence measurements of the BW suggest that the blue ring is
likely to be a propagating triple (edge) flame formed under the influence of ambient
circulation and within the radial boundary layer of the whirling flame. The locations
of the rich and lean branches depend on the radial location. A major portion of
the ring shows the rich branch located below (in the axial direction) the ring, with
the lean branch above it. The lower blue cone is then expected to be a partially-
premixed rich front, as postulated by Coenen et al. [23]. Numerical simulations by
Chung [33] generally agree with this structure, although a nonpremixed (diffusion)
flame was proposed to exist in the fuel-lean post-flame region with heat-release rates
comparable to that at the blue ring. Experimental data does not show significant
presence of OH* or PAH to support a separate reaction front in the purple haze. A
?wing-like? lean front is more likely, visible when imaging total chemiluminescence
(Supplementary Figures, Figures S.1 and S.2 ). Numerical simulations of the blue
whirl (Figure 6.8) show significant heat release in the purple haze region, which
agrees with measurements of OH, but not with OH*. Finally, the location of the
rich and lean branches seem to reverse at the very edge of the ring, possibly a
part of the ?wing? seen in the numerical simulations and total chemiluminescence
mesaurements. This aspect needs to be investigated further in future work.
Quantities of less than 5% by volume of biacetyl were recommended in previous
183
Figure 6.8: Contours of flame index (indicating level of premixing), equivalence
ratio and temperature heat-release rate in the blue whirl, obtained using numerical
simulations. Figure adapted from [33].
work [239], but sufficient signal levels were not observed for these small quantities
when used in the blue whirl. The fraction had to be increased to 50% biacetyl
in the fuel mixture for PAH-PLIF ? this affects both evaporative characteristics
of the fuel and chemical kinetics in the blue whirl. A significant fraction of the
biacetyl was not consumed by the flame and remained over the metallic surface
over which the fuel was supplied. The results presented here are sufficient to make
conjectures that the lower blue cone is the rich front and the purple haze is the
lean front, but probing mixing dynamics will require a technique to visualize fuel
evaporation without the dopant interfering with either the diagnostic technique or
combustion chemistry. Additionally, since PAH is basically a precursor to soot, and
no soot is eventually formed in the blue whirl, the measurements are only qualitative.
Determining density counts will require the utilization of other optical techniques.
Using these results, a 2-D representation of fuel and air flow around the blue
whirl is represented in Figure 6.9. The blue ring is the widest part of the flame,
and the diameter is always smaller than the fuel pool over which the flame stabilizes
184
Figure 6.9: A simplified schematic of the flame structure of the blue whirl, based
on measurements of OH, PAH, OH* and CH*.
[22,23]. The strong radial entrainment at the bottom surface helps concentrate the
evaporating fuel into a small region above the fuel pool. The strong radial boundary
layer influences the conical shape of the lower part of the flame and the lift-off of the
ring [23], which stabilizes at the region where the fuel-air mixture has the highest
reactivity. Mixing begins in the region outside the lower blue cone, and the highest
concentration is seen immediately below the ring. Some fuel vapor may react in
the lean flame of the purple haze, although there is a possibility that some vapors
are not entrained towards any of the reaction zones. This is postulated from PAH-
PLIF signals that peak near the ring, but decrease in intensity in the axial direction
(Figure 6.9). The shape of the vortex-breakdown bubble and the recirculation zone
determines the flow path around the bubble towards the blue ring region. The
lower blue cone, in addition to being the rich branch, may support heat release
in the form of low-temperature hydrocarbon cracking, causing low-intensity visible
radiation. The details of the recirculation zone are not depicted in Figure 6.9 since
185
the dynamics are not evident from the diagnostics used in this study.
The composition of the region within the conical region is still unclear, and
previous work on vortex breakdown in both reacting [59,232] and non-reacting [229,
231] bubble modes of vortex breakdown suggest that the interior of the bubble may
be composed mostly of combustion products. This may also be diluted by additional
entrainment by flow in the reverse axial direction. This needs to be verified by
imaging CO?2 or H2O. The proposed structure is for a stable blue whirl and detailed
diagnostics are required to study the mixing dynamics during a transition from a
fire whirl to a blue whirl, as well as the lift-off of the ring from the surface of the
fuel pool.
186
6.6 Summary
An experimental investigation of the blue whirl was performed to identify
the distribution of different radicals in the flame. PLIF was used to identify OH
and PAH, and chemiluminescence to image OH* and CH*. The results present
qualitative insight to the flame structure, which is expected to be a lifted triple-
flame under ambient circulation. The flame is stabilized at the blue ring, with the
rich branch formed by the lower blue cone, and the lean branch formed by the purple
haze region. A large fraction of the heat release occurs at the ring, due to the high
concentration of OH in the region, with the possibility of some heat release in the
post-flame (purple haze) region where OH concentration diminished continuously in
the axial direction. Quantifying the relative rates of heat release in these regions
requires either HCO imaging, or simlutaneous measurements of OH and HCHO.
The ratio of OH* and CH* signal intensities used to determine the rich and
lean edges are as yet uncalibrated due to the effects of data normalization. Data
normalization could be performed for standard, premixed, methane-air flames, but
the procedure could not be extended to images of the blue whirl due to the necessity
of knowing the local flow rates in the blue whirl and the requirement of an arbi-
trary maximum value. Previous studies that used the chemiluminescence-intensity
ratios of different radical species to determine heat-release rate or equivalence ratio
typically use the calibration process that is validated by numerical modeling [187]
or limit the procedure to observing only the calibration flame [194]. Calibration
using premixed flames to extrapolate properties to partially-premixed flames has
187
also not been attempted earlier. This approach needs to be investigated further
to determine applicability to complex flames such as the blue whirl. Additionally,
the subtraction of chemiluminescence from CO2* also needs to be accounted for in
future simultaneous OH*/CH* experiments [249].
The absence of OH or PAH signals in the recirculation zone indicates that it
is composed of combustion products, which act as a flame holder, stabilizing the
blue ring, enabling thermal cracking of the fuel at the edge of the lower blue cone,
which exhibits visible faint blue signals. CO2 and H2O imaging could be performed,
possibly in the IR range, to confirm this. The dynamics of the mixing process
outside the blue cone are still unclear, and certain experimental challenges need to
be overcome to probe this. For instance, the wandering of the flame needs to be
suppressed such that simultaneous imaging of OH/PAH-PLIF can be performed for
a sufficient period of time at high speed to track the flow of fuel towards the reaction
zone.
The qualitative results obtained in this work present explanations for the ther-
mal structure previously established in the literature. They also present sufficient
qualitative understanding of the flame structure and mixing that can be supple-
mented by concurrent efforts in numerical simulations, and also a foundation for
future quantitative, simultaneous, multi-species measurements.
188
6.7 Acknowledgments
The contents of this chapter, in part, were presented as a talk titled ?Un-
derstanding Combustion in the Blue Whirl using Optical Diagnostics?, by S.B.
Hariharan, Y. Wang, P.M. Anderson, M.J. Gollner, W.D. Kulatilaka, E.S. Oran at
the 72nd Annual Meeting of the APS Division of Fluid Dynamics, Seattle, Washing-
ton, November 2019. Some data is also included in a manuscript in preparation for
submission to Combustion and Flame by the same authors.
189
Chapter 7
Conclusions and Future Directions
Whirling flames have been a subject of academic interest for many decades
as they present a phenomenon involving complex interactions between combustion,
heat transfer and fluid instabilities. Experimental, theoretical and numerical ap-
proaches have focused on each of the above aspects with the aim of understanding
the mechanisms that lead to their formation. Early studies were devoted to pre-
dicting fire-whirl occurrence during urban or wildland fires, and although progress
has been made in this front, limitations still exist in extending the knowledge to
real-world scenarios.
In contrast to whirling flames, the field of swirl combustion is quite mature,
given the extensive relevance to applications in propulsion and energy conversion.
One crucial difference between these fields is the difference between length scales of
experimental investigations in the laboratory and that of the real-world applications;
in swirl combustion, the length scales of laboratory study are very close to that
in actual applications: in the case of fire whirls, however, laboratory studies can
190
sometimes be orders of magnitude smaller than the real-world scenarios they seek
to understand and explain. Despite this, substantial progress has been made in the
understanding of fire whirls.
Emissions from fire whirls
This dissertation explored aspects of whirling flames that were not considered
previously in the literature and is motivated by potential applications in environ-
ments that are relatively controlled when compared to wildfires. First, the emissions
of particulate matter (soot) were measured for fire whirls and compared to those
from free-buoyant pool fires to evaluate the potential for fire whirls to reduce the im-
pact of airborne emissions during in-situ burning of spilled oil. Measurements were
performed for pool fires and fire whirls formed with two liquid fuels, n-heptane and
ANS crude oil. The length scales chosen ranged from typical typical laboratory scale
pools (10, 20, 30 cm in diameter) to length scales that can represent characteristics
of field trials (70 cm).
Fire whirls had higher burning rate and fuel-consumption efficiency, and lower
emission rates of particulate matter, all of which led to lower emission factors. This
was consistent at the four pool diameters and for both fuels. Phenomenological
relationships between global scaling parameters such as Rossby number and nondi-
mensional heat-release rate showed that the emission factor of particulate matter
from a whirling flame decreases as the effect of ambient circulation on the fire in-
creases. These relationships do not provide a predictive methodology for estimating
191
the reduction in emissions by burning a pool of liquid fuel as a fire whirl rather
than a pool fire, but present a foundation for further improving developing these
approaches by studying the effects of circulation and heat-release rate on emissions.
Further, challenges in the domain of engineering and implementation were not ex-
plored; rather, a fundamental hypothesis has been tested and confirmed, providing
strong justification to subsequently consider more detailed research.
Studies on the blue whirl
Next, the blue whirl, an ideal state of a whirling flame producing no particulate
emissions was analyzed. The transition from the traditional fire-whirl regime to the
blue whirl was explored using a scaling analysis to quantify the roles of local buoyant
and tangential momenta. The transition from a buoyancy-dominated regime to a
circulation-dominated regime correlated to the establishment of a recirculation zone
and the onset of the bubble mode of vortex breakdown. Here, dependent variables
based on flame geometry were used as characteristic length scales for normalization
of heat-release rate and circulation. At this stage, this limits the potential to predict
transition to the blue whirl at length scales larger than those tested. Still, the results
presented help definitively answer why the transition occurs, and to quantitatively
differentiate between different regimes of whirling flames, with scope to improve the
deterministic capabilities by developing simple empirical relationships between fixed
dimensions of the experimental apparatus and those of flame geometry. Finally, the
distribution of OH, PAH, OH* and CH* radicals in the flame indicate potential for
192
the blue ring to be a triple-flame, with the lower blue cone forming the rich branch
and the purple haze forming the lean branch.
193
7.1 Future Work
Fire-whirl formation and emissions
Emissions from fire whirls were consistently lower for all pool diameters. But,
as mentioned above, the formation conditions where not the same for fire whirls
formed at the different length scales. This affects the geometry of fire whirls, and
consequently their burning characteristics and efficiency. One aspect that may be
standardized in future investigations is the area ratio, defined as
Apool
Aratio =
Aenclosure
The values of Aratio for the different values of D in this study are shown in Figure
7.1. Aratio needs to be constant for fire whirls formed at different values of D to
ensure that the entrainment conditions are similar. Since flame area is important
parameter influencing burning efficiency, ensuring the maximum flame height at
each length scale will permit evaluation of the maximum possible value of burning
efficiency at each scale.
From this work, it is evident that FWs have higher flame surface area as
compared to PFs. But the data at the different scales are vastly different in terms
of the effects of Gr and external Re. So, experiments in a given regime, preferably
the laminar regime with pool diameters below 10 cm, are required to study the
influence of circulation on emission factor.
194
Figure 7.1: Variation of Aratio with D. Lineat fit shown has a R
2 value of 0.96.
Secondly, the effects of circulation and buoyancy need to be investigated inde-
pendently. In the literature, flame geometry data for pool fires and fire whirls formed
at similar values of Q? are available [106], but the experimental apparatus can vary
significantly. Particularly, the bottom boundary over which the fires are formed are
important ? the presence of a lip can cause large differences in m? (see Fig. 5 in [8]).
Independent control over ? and Q? requires a setup with forced entrainment, enabled
by a spinning mesh or a bank of fans to control entrainment velocity, thus fixing the
value of ?. Results from preliminary experiments comparing the effects of natural
and forced entrainment on liquid-fueled fire whirls are included in Appendix A. Q?
can be controlled separately by the use of a gaseous burner. While experiments
with such arrangements have been performed in the past, emissions data for these
conditions do not exist. Experiments with a spinning mesh can also help study the
effects of flame stretch and residence time on soot formation.
More generally for fire whirls, the linear relation between heat-release rate and
195
circulation exists only up to a certain point in natural entrainment configurations.
Additionally, the linear relationship between flame height and circulation exists only
if ??f ? (0, 1). Over-rotated FWs reduce in height, with corresponding increase in
flame width. For all these cases, the relationship between heat-release rate and
flame area needs to be investigated. A systematic study of pool fires, and fire whirls
formed over a range of externally-controlled circulation, are needed over a wide
range of heat-release rates to observe the effect of flame area on emissions.
Other gaseous species such as SOx and NOx need to be measured for comparing
pool fires and fire whirls. Given that temperatures are lower than 1800 K, it is
unlikely that NOx formation is a major concern. Unburned hydrocarbons need
to be measured directly using methods such as Flame Ionization Detection (FID)
rather than estimation using the carbon balance method, which requires assumptions
regarding the composition of combustion products. Additionally, an analysis of the
post-combustion residue can help understand the effects of high heat feedback on
the residue.
Another aspect of this work is to investigate the effect of fuel slick thickness and
boilover on emissions. Characteristics such as burning rate can change significantly
with time when burning crude oil, and the composition of the emission species
is also expected to change over extended periods of time. Boilover can occur in
laboratory-scale experiments, but is not observed in field conditions. By varying
the fuel layer thickness, regular burning can be sustained over a longer duration,
resulting in combustion of a broader range of hydrocarbons from the crude oil.
Increased slick thicknesses can also delay boilover, causing a longer duration
196
of normal burning, making emissions less dependent on test conditions such as pool
diameter. This is a more practical approach to evaluate fire whirl use in ISBs.
Other practical concerns also include evaluating the efficacy of fire whirls on crude
oil under varying degrees of weathering.
Finally, the exact mechanisms of mixing and combustion in the fire whirl are of
immense importance to understanding its emissions behavior. The dynamics at the
base of the fire-whirl structure are important since most of the entrainment occurs
there, and the mixture fractions and turbulence intensity in that region determine
the shape and regime of the fire whirl. These characteristics are best explored using
non-intrusive laser diagnostics, both absorption and emission spectroscopy. Such
experiments can then serve as a foundation to validate numerical simulations with
detailed chemistry.
Future studies of the blue whirl
Similar to the fire whirl, the mixing dynamics and the mechanisms of flame
stabilization at the blue ring are important to advancing our understanding the blue
whirl. Here too, there is a need for using non-intrusive and high-speed diagnostics
to probe multiple reactive species simultaneously. This will help understand the role
of the ring, the recirculation zone and the blue conical region. More importantly,
visualization of the change in the nature of the flow during the transition from a fire
whirl a blue whirl is necessary to understand the development of the recirculation
zone and the ring lift-off. The optical-diagnostic results presented in this work
197
need to be quantified: density counts of HCO, or OH and HCHO, are needed to
quantify the local heat-release rate. These can also help validate and calibrate
existing numerical models of the blue whirl.
Before such experiments are possible, there is a need to develop a burner that
can stably form the blue whirl using a gaseous fuel and eliminates flame wander.
Flame wander is intrinsic to whirling flames, but limiting it to within one burner
diameter will aid in obtaining a sufficient amount of data to analyze the transition
processes.
For the bubble mode to appear, axial momentum must be overcome by tan-
gential momentum [229, 230, 250], and this is more likely to occur when ??f > Q?
?
f
(Chapter 5). Although vortex breakdown is a phenomenon occurring in the vicinity
of the flame, the axial profile of tangential velocity at the inlet for the blue whirl [24]
is quite different from the typical velocity profile measured for a quasi-steady on-
source fire whirl [251]. While the measurements in this study are made at the inlet
section, the flow field in the vicinity of the blue whirl is unclear and deserves further
study.
Practical applications of the blue whirl are still an open question. The lami-
nar nature of the flame, in addition to high sensitivity to experimental conditions,
make it unlikely to be used in applications such as power generation or propulsion.
It could, however, potentially serve as a canonical model to study vortex break-
down under the influence of heat release at low Reynolds number (low turbulence
intensity). A preliminary study of whirling flames under microgravity was recently
performed [83]. Microgravity conditions may offer an alternative method of study-
198
ing the blue whirl since the effect of buoyancy is suppressed, providing the impetus
for vortex breakdown to occur in whirling flames
199
7.2 Contributions
This work has contributed to whirling-flame literature in three areas not ex-
plored previously. First, the potential for practical utilization of whirling flames
have been shown, which advances our knowledge of whirling flames beyond tra-
ditional considerations of safety during unwanted urban and wildland fires. The
lower particulate-matter emissions from whirling flames present a platform to eval-
uate considerations of practicality and the engineering of devices for remediation or
even energy conversion applications. Second, circulation- and buoyancy-dominated
regimes were identified in whirling flames, and the transition to a circulation-dominated
regime was shown to correspond with the occurrence of the bubble mode of vortex
breakdown, resulting in blue-whirl formation. These results, while presented in the
context of reacting flows, can be extended to non-reacting flows as well. Finally, the
near-limit blue whirl was shown to have a flame structure similar to that of a triple
flame, anchored at the blue ring region. The above contributions have been dissem-
inated in the form of project reports, conference presentations and proceedings, and
peer-reviewed journal articles, listed below.
1. S.B. Hariharan, Y. Hu, M.J. Gollner, E.S. Oran, ?Effects of circulation and
buoyancy on the transition from a fire whirl to a blue whirl,? Accepted for
publication in Physical Review Fluids
2. S.B. Hariharan, H.F. Farahani, A.S. Rangwala, J. Dowling, E.S. Oran, M.J.
Gollner, ?Comparison of Emissions from Liquid-Fueled Pool Fires and Fire
200
Whirls at Different Length Scales,? In preparation for submission and peer
review in August 2020
3. S.B. Hariharan, Y. Wang, P.M. Anderson, M.J. Gollner, W.D. Kulatilaka,
E.S. Oran, ?On the nature of combustion in the blue whirl using optical diag-
nostics,? In preparation for submission to Combustion and Flame
4. M.J. Gollner, E.S. Oran, S.B. Hariharan, J. Dowling, H.F. Farahani, A.S.
Rangwala, ?Efficient Remediation of Oil Spills over Water Using Fire Whirls,?
BSEE Project Number 1094, Washington, DC, 2019.
5. S.B. Hariharan, Y. Wang, P. Anderson, W.D. Kulatilaka, M.J. Gollner, E.S.
Oran, ?Understanding Combustion in the Blue Whirl using Optical Diagnos-
tics,? 72nd Annual Meeting of the APS Division of Fluid Dynamics, Session
C06.3, Seattle, Washington, November 2019.
6. S.B. Hariharan, J. Dowling, M.R. Jones, V.M. Valletta, M.J. Gollner, E.S.
Oran, A. Ogorzaly, E.S. Neumann, S. Olson, P. Ferkul, ?Fire Whirls in Micro-
gravity,? 35th Annual Meeting, American Society for Gravitational and Space
Research, Denver, Colorado, November 2019.
7. S.B. Hariharan, J. Dowling, H.F. Farahani, Y. Hu, A. Tohidi, K. Stone,
A. Rangwala, E.S. Oran, M.J. Gollner, ?Demonstration of Reduction in Emis-
sions by Fire Whirls through Small Scale Experiments,? 42nd AMOP Technical
Seminar, Halifax, Nova Scotia, Canada, June 2019.
8. S.B. Hariharan, ?Fire Whirls and Blue Whirls: Emissions Reduction and
201
Vortex Breakdown,? 2019 Research Symposium on Environmental and Applied
Fluid Dynamics, Johns Hopkins University, May 2019.
9. S.B. Hariharan, J. Dowling, H.F. Farahani, M.J. Gollner, E.S. Oran, ?Com-
parison of emissions from liquid-fueled pool fires and fire whirls,? 11th US
National Combustion Meeting, Pasadena, California, March 2019.
10. Y. Hu, S.B. Hariharan, H. Qi, M.J. Gollner, E.S. Oran, Conditions for
formation of the blue whirl, Combustion and Flame. 205 (2019) 147?153.
11. Y. Hu, S.B. Hariharan, M.J. Gollner, E.S. Oran, Conditions for Formation
of the Blue Whirl, Eastern States Section of the Combustion Institute Spring
Technical Meeting, The Pennsylvania State University, Abstract 3B08, Penn-
sylvania, March 2018.
12. S.B. Hariharan, P.M. Anderson, H. Xiao, M.J. Gollner, E.S. Oran, The blue
whirl: Boundary layer effects, temperature and OH* measurements, Combust.
Flame. 203 (2019) 352?361.
13. S.B. Hariharan, E.T. Sluder, M.J. Gollner, E.S. Oran, Thermal structure of
the blue whirl, Proc. Combust. Inst. 37 (2019) 4285?4293.
202
Appendix A: Effect of Entrainment Conditions on PM Emis-
sions from Fire Whirls
A.1 Overview
The experimental results discussed in Chapters 3 and 4 represent conditions
where fire whirls were formed under natural entrainment, i.e., circulation is not
imposed externally on the fire. Rather, the circulation is intrinsically coupled with
the heat-release rate of the fire whirl and depends, to a certain extent on the gap
size in a four-walled setup. This means that only one combination of heat-release
rate and circulation are permitted for each scale, similar to experiments with the
blue whirl discussed in Chapter 5.
To decouple the effect of heat-release rate and circulation, heat-release rate
is maintained constant by controlling fuel-flow rate in a gaseous-fuel burner, and
externally-controlled air entrainment is be imposed by means of a rotating mesh
(such as the early Emmons-type apparatus [8]) or using a bank of fans to control
the inlet velocity. Most experimental setups in the literature do not use the fan-
based setup, although recent work on micrgravity fire whirls used this setup [83].
In such apparatus with forced entrainment, heat-release rate and circulation remain
203
coupled if a condensed-phase fuel is used.
For a fuel pool of given diameter, it is of interest to understand the effect
of different entrainment conditions on PM emissions from fire whirls such that the
ideal level of ventilation may be designed for a practical device. Hence, preliminary
experiments were performed with 70 cm pools to determine the effect of enclosure
gap size and different levels of forced entrainment. The results discussed here are
primarily qualitative since U? was not measured, and therefor an estimate of cir-
culation was not possible. Still, these results still help in understanding that an
optimum condition exists, and provide insight on how to design future experiments
to identify the combinations of circulation and heat-release rate that result in the
low values of EFPM for a given D.
A.2 Experiments
Experiments with diesel fuel were conducted only at the 70 cm scale, and
details of the general setup and PM measurements have been presented in sections
3.2 and 4.2. For experiments with heptane and ANS crude oil, gap size in the four-
walled configuration was kept constant for all experiments. For the 70 cm fuel pool,
the enclosure side was 175 cm with a fixed gap size of 45 cm. The fuel used was
automotive-grade diesel due to its low cost and easy accessibility.
The different experimental conditions in this study are shown in Table A.1.
Four natural-entrainment (NE) conditions were studied by varying the gap sizes
between 35 and 65 cm. Four forced-entrainment (FE) conditions were studied with
204
Table A.1: Experimental conditions for the different air entrainment conditions for
fire whirls formed using diesel fuel.
Air Gap Enclosure side Enclosure wall
Fuel
entrainment width, W length, S height, H
volume [ml]
mechanism [cm] [cm] [cm]
35
Natural 45 175 240 2000
55
65
Natural: Level 0
Forced: Level 1 55 175 240 2000
Forced: Level 2
Forced: Level 3
a fixed gap size of 55 cm and positioning a box fan at each of the four inlets. The box
fan (50? 50 cm2, Lasko Model 3733) could be operated at three speeds, although a
characterization of the flow profile or outlet velocity are not available at this time.
The fans supplied air only to the lower part of the enclosure, and future experiments
at smaller scales will be required to design inlets that supply air along the entire
height of the enclosure. For comparison, a pool fire experiment was also performed.
For all experiments, 2L (5 mm slick) of diesel was used. For each of the conditions
listed in Table A.1, only one experiment was performed. Thus, there is no estimation
of error, and the results from this preliminary study may be used as a platform to
design systematic multi-scale experiments in the future.
205
A.3 Results and Discussion
Similar to ANS crude oil, pool burning of diesel left behind a residue. This
residue, however, could not be collected using the 3M adsorbent pads for mass
measurement. Right before extinction, a distinct crackling noise was also audible,
similar to boilover, but existed for only a few seconds. The PM emission rate
for each of the different conditions in Table A.1 is shown in Figure A.1, and the
corresponding PM emission factors are shown in Figure A.2. The burning duration
for the NE conditions are slightly higher than for the FE conditions. The peak
emission rate under NE occurs at a gap size of 35 cm, and is higher than the
peak emission rate from the pool fire. The average emission rate is lowest for a
55 cm gap, and is lower than that for the pool fire, similar to previous results for
non-boilover burning of heptane and ANS crude oil. In the case of FE, the peak
emission rate increases continuously with fan speed, with the highest concentration
being exceeding the DustTrak?s measurement limit after ?100 s.
For the 70 cm pools burning heptane and ANS crude oil, EFPM of fire whirls
was around ?50 % of that of pool fires (Figure 4.5). For diesel, however, the
reduction in EFPM by fire whirls is not as significant. EFPM reduces continuously
with gap size, and the value levels off after 55 cm. With FE, the lowest EFPM occurs
at the lowest inlet velocity, but then increases continuously with fan speed. At the
highest fan speed, the EFPM for the FW is higher than that for a pool fire . These
results suggest that a unique combination of gap size and inlet velocity (or ?), that
results in the lowest EFPM, exists for a given fuel pool diameter.
206
Figure A.1: Dependence of PM emission rate on gap size for the naturally entrain-
ment cases (top), and fan speed on the forced entrainment cases (bottom) for fire
whirls formed over a 70 cm diesel pool.
207
Figure A.2: PM emission factors for fire whirls formed over a 70 cm diesel pool.
Dependence on gap size for naturally entrainment cases (top), and dependence on
fan speed for forced entrainment cases (bottom).
208
Results in Chapter 4 showed the importance of the relative influences of buoy-
ancy and circulation on EFPM. The primary effect of changing gap size or adding
forced entrainment is to change the level of circulation on the fire. These results
show that for a given fuel pool diameter, natural air entrainment does not offer the
ideal conditions for lowest PM emissions. Some level of external forcing is required,
which increases circulation, and for liquid fuel pools, alters the heat-release rate.
Gabler?s work on asymmetrically whirling flames formed using gaseous fuels
[252] presents a method to analyze the influence of varying momenta in different
directions on NOx and CO emissions. The nondimensional Damko?hler number (Da)
was used to explain some of the emissions behavior. The effect of Damko?hler number
(Da) is typically associated with extinction [253], and has recently been shown to
significantly affect the transition from a fire whirl to a blue whirl [82].
Similar to Gabler?s approach [60,84,252], it is proposed that the mechanism of
soot emission from fire whirls may need to be related to flow characteristics defined
based on the relative time scales (derived from local flow velocities) in the two
directions of interest - axial, due to buoyancy, and azimuthal, due to air entrainment.
These quantities may be used to define a Froude number (Fr) and analyze the effects
on EFPM. For this, an estimate both of axial [7] and azimuthal velocity [54] in the
vicinity of the flame sheet at different heights. For more global descriptions, a scaling
approach still needs to be used by defining nondimensional quantities representing
circulation and heat-release rate. Characteristic axial and azimuthal time scales
may be determined based on appropriate length scales which may be either the fuel
pool diameter, fire whirl height and width, ratio of of gap size to enclosure side etc.,
209
depending on the type of forced entrainment used in the experiment.
A.4 Conclusions
Circulation controls many properties of the fire whirl including flame height
and width, temperature, heat-flux feedback etc. Additionally, for condensed-fuel
flames, it also affects burning rate. Lei et al. [92] showed that fire whirl height
first increases with ?, then decreases, while flame width follows the opposite trend.
Circulation and heat-release rate are intrinsically coupled in apparatus with natural
air entrainment and only one combination of these quantities can form stably for
a given set of dimensions such as pool diameter, enclosure side (or diameter), gap
width and enclosure height. Preliminary results have shown that this combination
of circulation and heat-release rate under natural entrainment does not lead to the
lowest possible particulate matter emission factor. Rather, a small level of externally
forced entrainment leads to a reduction in the emission factor. This reduction is
attributed to some additional air available for combustion at the lower region of the
fire whirl, and enhanced mixing.
The preliminary experiments involved only one fuel pool diameter, and only
one test for each of the different entrainment conditions. A more systematic ap-
proach, similar to those discussed in Chapters 3 and 4 is needed to fully understand
the effects of forced entrainment on pollutant emission at different length scales.
This would also provide an estimate of the experimental errors associated with the
measurements. For this, suitable apparatus need to be designed to identify differ-
210
Figure A.3: Influence of equivalence ratio and in-cylinder temperature on soot and
NO emissions from an internal combustion reciprocating engine. Figure from [34].
ences between the popular rotating mesh setup and the fan-based setup. This would
also help generate a fire-whirl behavior map for liquid-fueled fire whirls, following
the approach of Lei et al. [92] for a gas burner.
Effects of Da and Fr could be incorporated into the analysis to include the
effects of local flow properties and fuel chemistry to generate a topological map of
the influence of important parameters on soot emissions from fire whirls. Similar
approaches are used in other fields such as IC engines, for which in-cylinder pollu-
tant production is controlled by local equivalence ratio and temperature [17, 254].
High temperatures and low equivalence ratios favors production of NO, while low
temperatures and high equivalence ratios favors production of soot (see Figure A.3,
adapted from [34] and pertaining to pressures above atmospheric conditions). For
fire whirls and jet flames under swirling flow fields, characteristic Da and Fr num-
211
bers may be used to explain particulate matter emission by including the effects
of chemistry and identifying the right operating conditions that are required for an
engineering device. One possibility for field use may be a burner based on a large
pilot-ignited spray flame, described in detail by Tuttle et al. [255].
212
Appendix B: Supplementary Material to Chapters 3 and 4
This appendix contains additional data and figures as supplement to Chapters
3 and 4. The different line colors in Figures B.1?B.9 depict individual experimental
trials.
Table B.1: Experimental variability in the estimation of ?fuel for ANS fires atD = 70
cm.
Initial mass Residue ?fuel
[g] [g] [%]
Test 1 2400 388 83.8
ANS PF Test 2 2450 375 84.7
(7 mm) Test 3 2550 390 84.7
Test 4 2400 377 84.3
Test 1 2360 362 84.67
ANS FW Test 2 2360 386 83.6
(7 mm) Test 3 2400 401 83.3
Test 4 2400 382 83.9
Test 1 1710 284 83.4
ANS PF
Test 2 1710 304 82.2
(5 mm)
Test 3 1710 315 81.6
Test 1 1710 284 83.4
ANS FW Test 2 1710 285 83.3
(5 mm) Test 3 1710 295 82.7
Test 4 1710 298 82.6
213
Table B.2: Experimental variability in the estimation of ?fuel for ANS fires at D =
[10, 20, 30] cm.
D Regime Initial mass Residue ?fuel
[m] [g] [g] [%]
Test 1 33 18.1 45.1
PF Test 2 33 19.7 40.4
10 Test 3 33 61.4 40.4
Test 1 33 16.35 50.45
FW Test 2 33 16.3 50.7
Test 3 33 15.8 52.2
Test 1 132.5 83.1 37.3
PF Test 2 128.5 58.1 54.8
20 Test 3 129.3 61.4 52.5
Test 1 132.9 64.2 51.7
FW Test 2 132.9 53 60.1
Test 3 132.9 60.3 54.6
Test 1 313.3 163.5 47.8
PF Test 2 306.8 126.1 58.7
30 Test 3 310 112 63.9
Test 1 310.4 99.7 67.9
FW
Test 2 314 182.3 41.9
214
Figure B.1: O2 consumption, D = 10 cm.
215
Figure B.2: O2 consumption, D = 20 cm.
216
Figure B.3: O2 consumption, D = 30 cm. Panel B shows results from six different
experiments.
217
Figure B.4: CO2 emission rate, D = 10 cm.
218
Figure B.5: CO2 emission rate, D = 20 cm.
219
Figure B.6: CO2 emission rate, D = 30 cm.
220
Figure B.7: CO emission rate, D = 10 cm.
221
Figure B.8: CO emission rate, D = 20 cm.
222
Figure B.9: CO emission rate, D = 30 cm.
223
Figure B.10: Experimental variability in the measurement of m? at D = [10, 20, 30]
cm.
224
Figure B.11: Average values of centerline temperature measured at different axial
distances for PFs and FWs formed with heptane and ANS crude oil.
225
Figure B.12: Maximum values of centerline temperature measured at different axial
distances for PFs and FWs formed with heptane and ANS crude oil.
226
Figure B.13: Rez, estimated as (m?
??D/?)/?, with ? and ? evaluated at Tfilm. The
values of Tfilm for heptane PF, heptane FW, ANS PF and ANS FW were 675 K, 800
K, 650 K and 725 K, respectively. The ratio (m???/?) is Uz=0, the spatially averaged
value of fuel evaporation at the fuel surface.
Figure B.14: Re?, estimated as ?/?.
227
Figure B.15: Riz, estimated as Gr/Re
2
z.
Figure B.16: Ri?, estimated as Gr/Re
2
?.
228
Figure B.17: Fuel Froude number, Frf , estimated as (m?
??2)/(?20gD) [11], with ?0
evaluated at T0 = 300K.
?
Figure B.18: Modified Peclet number, Pe, evaluated as (m?)/(? gD5) [35].
229
Appendix C: Supplementary Material to Chapter 6
C.1 Chemiluminescence data of premixed methane-air flames
Figure C.1: Average contours of CH* chemiluminescence intensity, maximum inten-
sity and location indices of maximum intensity for ? = [0.7, 0.8, 0.9, 1.0].
230
Figure C.2: Average contours of CH* chemiluminescence intensity, maximum inten-
sity and location indices of maximum intensity for ? = [1.1, 1.2, 1.3, 1.4].
231
Figure C.3: Average contours of OH* chemiluminescence intensity, maximum in-
tensity and location indices of maximum intensity for ? = [0.7, 0.8, 0.9, 1.0].
232
Figure C.4: Average contours of OH* chemiluminescence intensity, maximum in-
tensity and location indices of maximum intensity for ? = [1.1, 1.2, 1.3, 1.4].
233
C.2 Chemiluminescence and PLIF: BW
Figure C.5: CH* in the BW.
234
Figure C.6: OH* in the OH.
235
Figure C.7: Individual frames showing PAH in the BW.
Figure C.8: Individual frames showing OH in the BW.
236
Figure C.9: Sequence of frames showing OH in the FW. Two frames (top right,
bottom left) show a tendency to transition to the BW, evidenced by the bulging of
the flame sheet, but the flame stays in the FW regime in the following frames.
237
C.3 Flow in the post-flame region of the blue whirl
Close observation of the BW shows the emanation of small, short-lived soot
particles from the blue ring. They are visible briefly before being consumed in the
post-flame region. The is visible when total chemiluminescence was imaged at high
frame rates (see Supplementary Material, Figures S.1A and S.2A). These soot traces
may be used to estimate roughly the residence time of soot within the post-flame
region and the Damko?hler number (Da). Since the diameter of each particle cannot
be measured from the incandescence, it is assumed here that the Stokes number
(Stk) is low enough that it closely follows the local flow path.
The path lines of some soot particles tracked using the Fiji version of Im-
ageJ [256]. Each instance of soot-particle emanation was tracked using the ?Manu-
alTrack? feature in Fiji. The overlayed tracks are shown in Supplementary Material,
Figures S.1B and S.2B). The vectors for some cases are shown in Figure C.10. Most
instances involved a particle following a trajectory in the axial direction, but some
particles first showed an azimuthal trajectory before eventually exiting the flame in
the axial direction. One such instance is shown in Figure C.10B.
Although the motion of the soot particles is in 3 dimensions, the projected
velocity of each particle was determined using the tracking-coordinate data and the
inter-frame time (0.04 s). The magnitudes of horizontal, vertical and total velocity
(u,v, |~u|) are summarized in Table C.1.
The Damko?hler number is defined as [17]
238
Figure C.10: Vectors showing the instantaneous velocity of two examples of soot
particles emanating from the blue ring and passing through the post-flame region.
?flow
Da = (C.1)
?rxn
where ?flow and ?rxn are the characteritic flow and reaction time scales. The two
time scales for the BW are defined as
BW diameter
?flow = (C.2)|~u|
ring thickness
?rxn = (C.3)Vf,iso?octane
where ? is the characteristic time scale, Vflame is the laminar flame speed of iso-
octane at 1 atm [257]. Using the limits of |~u| from Table C.1 and an average
BW diameter of 20 mm, ?flow was calculated to be in the range (0.5, 1) s. Using
values of Vflame from [257] and a ring thickness of 1.5 mm, ?rxn was estimated as
239
Table C.1: Consolidated velocity data (in [mm/s]) obtained by tracking soot in three
individual experiments.
u v |~u|
Overall Mean 12.6 23 27.3
Video 1
Std. deviation 8.1 7.6 8.2
Overall Mean 9.6 15.6 19
Video 2
Std. deviation 6.6 12.7 12.9
Overall Mean 16 27.7 34.5
Video 3
Std. deviation 7.8 23.7 21.3
? 0.004 s. Thus, the estimate of Da is in the range of (125, 250). The velocity
measurements are in the post-flame region, and the upstream flow velocity can be
calculated by adjusting for the excess temperature. Assuming the average postflame
temperature as 1500 K, the upstream velocity reduces by a factor of 1500 , which
300
results in ?flow ? (0.1, 0.2) and Da ? (25, 50).
240
Bibliography
[1] NASA/JPL-Caltech/Space-Science-Institute. Basking in Light, 2016.
[2] Cathal Cummins, Madeleine Seale, Alice Macente, Daniele Certini, Enrico
Mastropaolo, Ignazio Maria Viola, and Naomi Nakayama. A separated vortex
ring underlies the flight of the dandelion. Nature, 562(7727):414?418, 2018.
[3] Daniel Durox, Jonas P. Moeck, Jean Franc?ois Bourgouin, Pascal Morenton,
Marc Viallon, Thierry Schuller, and Se?bastien Candel. Flame dynamics of a
variable swirl number system and instability control. Combustion and Flame,
160(9):1729?1742, 2013.
[4] Pengfei Wang, Naian Liu, Katherine Hartl, and Alexander Smits. Measure-
ment of the Flow Field of Fire Whirl. Fire Technology, 52(1):263?272, 1 2016.
[5] Ali Tohidi, Michael J. Gollner, and Huahua Xiao. Fire Whirls. Annual Review
of Fluid Mechanics, 50(1):187?213, 1 2018.
[6] Jiao Lei, Naian Liu, and Kohyu Satoh. Buoyant pool fires under imposed
circulations before the formation of fire whirls. Proceedings of the Combustion
Institute, 35(3):2503?2510, 2015.
[7] Jiao Lei, Naian Liu, Linhe Zhang, and Kohyu Satoh. Temperature, veloc-
ity and air entrainment of fire whirl plume: A comprehensive experimental
investigation. Combustion and Flame, 162(3):745?758, 2015.
[8] Howard W. Emmons and Shuh-Jing Ying. The fire whirl. Symposium (Inter-
national) on Combustion, 11(1):475?488, 1 1967.
[9] Keng Hoo Chuah, Kazunori Kuwana, Kozo Saito, and Forman A. Williams.
Inclined fire whirls. Proceedings of the Combustion Institute, 33(2):2417?2424,
2011.
241
[10] A. Y. Klimenko and F. A. Williams. On the flame length in firewhirls with
strong vorticity. Combustion and Flame, 160(2):335?359, 2013.
[11] Kazunori Kuwana, Kozo Sekimoto, Kozo Saito, and Forman A. Williams.
Scaling fire whirls. Fire Safety Journal, 43(4):252?257, 5 2008.
[12] K.A. Hartl and A.J. Smits. Scaling of a small scale burner fire whirl. Com-
bustion and Flame, 163:202?208, 1 2016.
[13] Kazunori Kuwana, Satoshi Morishita, Ritsu Dobashi, Keng H. Chuah, and
Kozo Saito. The burning rate?s effect on the flame length of weak fire whirls.
Proceedings of the Combustion Institute, 33(2):2425?2432, 2011.
[14] A. Hamins, J.C. Yang, and T. Kashiwagi. An experimental investigation of
the pulsation frequency of flames. Symposium (International) on Combustion,
24(1):1695?1702, 1 1992.
[15] Hiroyuki Sato, Kenji Amagai, and Masataka Arai. Scale Modeling of Puffing
Frequencies in Pool Fires Related with Froude Number. In Progress in Scale
Modeling, pages 133?147. Springer Netherlands, Dordrecht, 2008.
[16] Johanna Aurell, David Hubble, Brian K. Gullett, Amara Holder, Ephraim
Washburn, and Dennis Tabor. Characterization of Emissions from Liquid
Fuel and Propane Open Burns. Fire Technology, 53(6):2023?2038, 11 2017.
[17] Stephen R. Turns. An Introduction to Combustion: Concepts and Applica-
tions. McGraw-Hill, 3rd edition, 2012.
[18] Hai Wang. Formation of nascent soot and other condensed-phase materials in
flames. Proceedings of the Combustion Institute, 33(1):41?67, 2011.
[19] K. O. Johansson, M. P. Head-Gordon, P. E. Schrader, K. R. Wilson, and
H. A. Michelsen. Resonance-stabilized hydrocarbon-radical chain reactions
may explain soot inception and growth. Science, 361(6406):997?1000, 9 2018.
[20] Chun Lou, Chen Chen, Yipeng Sun, and Huaichun Zhou. Review of soot mea-
surement in hydrocarbon-air flames. Science China Technological Sciences,
53(8):2129?2141, 8 2010.
[21] TSI Incorporated. Theory of Operation - DustTrak DRX Aerosol Monitor.
Technical report, TSI Incorporated, 2012.
[22] Sriram Bharath Hariharan, Evan T. Sluder, Michael J. Gollner, and Elaine S.
Oran. Thermal structure of the blue whirl. Proceedings of the Combustion
Institute, 37(3):4285?4293, 2019.
[23] Wilfried Coenen, Erik J. Kolb, Antonio L. Sa?nchez, and Forman A. Williams.
Observed dependence of characteristics of liquid-pool fires on swirl magnitude.
Combustion and Flame, 205:1?6, 7 2019.
242
[24] Yu Hu, Sriram Bharath Hariharan, Haiying Qi, Michael J. Gollner, and
Elaine S. Oran. Conditions for formation of the blue whirl. Combustion
and Flame, 205:147?153, 2019.
[25] Ronald K. Hanson, R. Mitchell Spearrin, and Christopher S. Goldenstein.
Spectroscopy and Optical Diagnostics for Gases. Springer International Pub-
lishing, Cham, 2016.
[26] CAI-Instruments. Non-Dispersive Infrared (NDIR) Analyzers.
[27] TSI-Incorporated. DUSTTRAK DRX AEROSOL MONITOR 8534.
[28] W M Thornton. XV. The relation of oxygen to the heat of combustion of
organic compounds. The London, Edinburgh, and Dublin Philosophical Mag-
azine and Journal of Science, 33(194):196?203, 2 1917.
[29] C. Beyler, P. Croce, C. Dubay, P. Johnson, and M. McNamee. Oxygen con-
sumption calorimetry, William Parker: 2016 DiNenno Prize. Fire Science
Reviews, 6(1):1?7, 2017.
[30] Morgan J. Hurley, Daniel Gottuk, John R. Hall, Kazunori Harada, Erica
Kuligowski, Milosh Puchovsky, Jose? Torero, John M. Watts, and Christopher
Wieczorek, editors. SFPE Handbook of Fire Protection Engineering. Springer
New York, New York, NY, 2016.
[31] Sriram Bharath Hariharan, Yejun Wang, Paul M Anderson, Waruna D. Ku-
latilaka, Michael J. Gollner, and Elaine S. Oran. Understanding Combustion
in the Blue Whirl using Optical Diagnostics. In 72nd Annual Meeting of the
APS Division of Fluid Dynamics, Seattle, Washington, 2019.
[32] Mikae?l Orain and Yannis Hardalupas. Effect of fuel type on equivalence ratio
measurements using chemiluminescence in premixed flames. Comptes Rendus
Me?canique, 338(5):241?254, 2010.
[33] Joseph Dongil Chung. Numerical Simulation of the Blue Whirl: A Reacting
Vortex Breakdown Phenomenon. PhD thesis, University of Maryland, College
Park, 2020.
[34] Suyin Gan, Hoon Kiat Ng, and Kar Mun Pang. Homogeneous Charge Com-
pression Ignition (HCCI) combustion: Implementation and effects on pollu-
tants in direct injection diesel engines. Applied Energy, 88(3):559?567, 2011.
[35] James G. Quintiere. Fundamentals of Fire Phenomena. 2006.
[36] Philip S Marcus. Jupiter?s Great Red Spot and Other Vortices. Annual Review
of Astronomy and Astrophysics, 31(1):523?569, 9 1993.
[37] A. Einstein. Die Ursache der Maanderbildung der FluBlaufe und des sogenan-
nten Baerschen Gesetzes. Die Naturwissenschaften, 14(11):223?224, 1926.
243
[38] C. S. Yogananda and Albert Einstein. The cause of the formation of meanders
in the courses of rivers and of the so-called Baer?s law. Resonance, 5(3):105?
108, 2000.
[39] R. D. HEY. Geometry of river meanders. Nature, 262(5568):482?484, 8 1976.
[40] H Helmholtz. U?ber Integrale der hydrodynamischen Gleichungen, welche den
Wirbelbewegungen entsprechen. Journal fu?r die reine und angewandte Math-
ematik (Crelles Journal), 1858(55):25?55, 1 1858.
[41] H Helmholtz. LXIII. On Integrals of the hydrodynamical equations, which ex-
press vortex-motion. The London, Edinburgh, and Dublin Philosophical Mag-
azine and Journal of Science, 33(226):485?512, 1 1867.
[42] Martin Kemp. Leonardo?s laboratory: studies in flow. Nature, 571:322?323,
2019.
[43] Vincent Van Gogh. The Starry Night, 1889.
[44] Katherine Wright. Arts & Culture: Turbulence in The Starry Night. Physics,
4 2019.
[45] J L Arago?n, Gerardo G Naumis, M Bai, M Torres, and P K Maini. Turbu-
lent Luminance in Impassioned van Gogh Paintings. Journal of Mathematical
Imaging and Vision, 30(3):275?283, 2008.
[46] James Beattie and Neco Kriel. Is The Starry Night Turbulent?
(arXiv:1902.03381v2). arXiv, (2):1?6, 2 2019.
[47] George M. Byram and Robert E. Martin. Fire Whirlwinds in the Laboratory.
Fire Control Notes, 23(1):13?17, 1962.
[48] George M. Byram and Robert E. Martin. The Modeling of Fire Whirlwinds.
Forest Science, 16:386?399, 1970.
[49] Jason M. Forthofer and Scott L. Goodrick. Review of vortices in wildland fire.
Journal of Combustion, 2011(Figure 1), 2011.
[50] N. P. Lareau, N. J. Nauslar, and J. T. Abatzoglou. The Carr Fire Vortex:
A Case of Pyrotornadogenesis? Geophysical Research Letters, 45(23):107?13,
2018.
[51] CalFIRE. CalFire Green Sheet: Burn Over Fatalities July 26, 2018, Carr In-
cident 18-CA-SHU-007808, 18-CA-SHU-007962, California Northern Region.
Technical report, CalFire, 2018.
[52] A. Muraszew, J.B. Fedele, and W.C. Kuby. Trajectory of firebrands in and
out of fire whirls. Combustion and Flame, 30:321?324, 1 1977.
244
[53] Kazunori Kuwana, Kozo Sekimoto, Kozo Saito, Forman a. Williams, Yoshihiko 
Hayashi, and Hideaki Masuda. Can We Predict the Occurrence of Extreme 
Fire Whirls? AIAA Journal, 45(1):16?19, 1 2007.
[54] K. A. Hartl and A. J. Smits. Stereo PIV measurements in fire whirls. Exper-
iments in Fluids, 60(1):1?16, 2019.
[55] Y. Hayashi, K. Kuwana, and R. Dobashi. Influence of Vortex Structure on Fire 
Whirl Behavior. Fire Safety Science, 10:671?679, 2011.
[56] M. Hassan, Kazunori Kuwana, Kozo Saito, and Fengjuan Wang. Flow Struc-
ture Of A Fixed-frame Type Firewhirl. Fire Safety Science, 8:951?962, 2005.
[57] N. Syred and J.M. Bee?r. Combustion in swirling flows: A review. Combustion 
and Flame, 23(2):143?201, 1974.
[58] Randal G. McKinney and James B. Hoke. Aero Gas Turbine Combustion: 
Metrics, Constraints, and System Interactions. In Timothy C. Lieuwen and 
Vigor Yang, editors, Gas Turbine Emissions, chapter 1, pages 3?23. Cambridge 
University Press, 2013.
[59] Se?bastien Candel, Daniel Durox, Thierry Schuller, Jean-Franc?ois Bourgouin, 
and Jonas P. Moeck. Dynamics of Swirling Flames. Annual Review of Fluid 
Mechanics, 46(1):147?173, 1 2014.
[60] Richard A Yetter, Irvin Glassman, and H Clay Gabler. Asymmetric Whirl 
Combustion: A New Low NOx Approach. Proceedings of the 
Combustion Institute, 28(x):1265?1272, 2000.
[61] N.A. Chigier, J.M. Bee?r, D. Grecov, and K. Bassindale. Jet flames in rotating 
flow fields. Combustion and Flame, 14(2):171?179, 4 1970.
[62] N. A. Chigier and A. Chervinsky. Aerodynamic study of turbulent burning 
free jets with swirl. Symposium (International) on Combustion, 11(1):489?499, 
1967.
[63] Rikiya YAMADA and Kazunori KUWANA. Scaling laws of flow structures 
around geometrically similar fire whirls. Journal of Thermal Science and Tech-
nology, 14(1):1?7, 2019.
[64] Huahua Xiao, Michael J. Gollner, and Elaine S. Oran. From fire whirls to blue 
whirls and combustion with reduced pollution. Proceedings of the National 
Academy of Sciences, 113(34):9457?9462, 8 2016.
[65] Sriram Bharath Hariharan. The Structure of the Blue Whirl: A Soot-Free 
Reacting Vortex Phenomenon. MS thesis, University of Maryland, College 
Park, 2017.
245
[66] Sriram Bharath Hariharan, Paul M Anderson, Huahua Xiao, Michael J Goll-
ner, and Elaine S Oran. The blue whirl: Boundary layer effects, temperature
and OH* measurements. Combustion and Flame, 203:352?361, 5 2019.
[67] Ying Huang and Vigor Yang. Dynamics and stability of lean-premixed swirl-
stabilized combustion. Progress in Energy and Combustion Science, 35(4):293?
364, 2009.
[68] V. Tangirala, R. H. Chen, and J. F. Driscoll. Effect of heat release and swirl
on the recirculation within swirl-stabilized flames. Combustion Science and
Technology, 51(1-3):75?95, 1987.
[69] Sandipan Chatterjee and O?mer L. Gu?lder. Soot concentration and primary
particle size in swirl-stabilized non-premixed turbulent flames of ethylene and
air. Experimental Thermal and Fluid Science, 95(January):73?80, 2018.
[70] P. Palies, D. Durox, T. Schuller, and S. Candel. Experimental study on the
effect of swirler geometry and swirl number on flame describing functions.
Combustion Science and Technology, 183(7):704?717, 2011.
[71] David G. Lilley. Swirl Flows in Combustion: A Review. AIAA Journal,
15(8):1063?1078, 1977.
[72] Joseph Chung, Xiao Zhang, Carolyn Kaplan, and Elaine S. Oran. Blue whirl
structure revelead. In 72nd Annual Meeting of the APS Division of Fluid
Dynamics, Seattle, Washington, 2019.
[73] D. D. Evans, G. W. Mulholland, H. R. Baum, W. D. Walton, and K. B.
McGrattan. In situ burning of oil spills. Journal of Research of the National
Institute of Standards and Technology, 106(1):231?278, 2001.
[74] John L Ross, Ronald J. Ferek, and Peter V. Hobbs. Particle and Gas Emissions
from an In Situ Burn of Crude Oil on the Ocean. Journal of the Air & Waste
Management Association, 46(3):251?259, 3 1996.
[75] Alan A Allen and Ronald J Ferek. Advantages and Disadvantages of Burning
Spilled Oil. International Oil Spill Conference Proceedings, 1993(1):765?772,
3 1993.
[76] David D. Evans and William D. Walton. Burning of Oil Spills. Technical
report, National Institute of Standards and Technology, Gaithesburg, MD.
[77] Jan Paul Zock, Gema Rodr??guez-Trigo, Francisco Pozo-Rodr??guez, Joan A.
Barbera?, Laura Bouso, Yolanda Torralba, Josep M. Anto?, Federico P. Go?mez,
Carme Fuster, and He?ctor Verea. Prolonged respiratory symptoms in clean-up
workers of the Prestige oil spill. American Journal of Respiratory and Critical
Care Medicine, 176(6):610?616, 2007.
246
[78] Gina M. Solomon and Sarah Janssen. Health effects of the gulf oil spill.
Disaster Medicine and Public Health Preparedness, 4(4):273?276, 2010.
[79] Olusegun G. Fawole, X. M. Cai, and A. R. Mackenzie. Gas flaring and resultant
air pollution: A review focusing on black carbon. Environmental Pollution,
216:182?197, 2016.
[80] H.W. Emmons. Fundamental problems of the free burning fire. Symposium
(International) on Combustion, 10(1):951?964, 1 1965.
[81] Amable Linan and Forman A. Williams. Fundamental Aspects of Combustion.
Oxford University Press, New York, 1993.
[82] Jaime Carpio, Wilfried Coenen, Antonio L Sanchez, Elaine Oran, and For-
man A Williams. Numerical description of blue whirls over liquid-fuel pools
(Under Review). Proceedings of the Combustion Institute, 2020.
[83] Sriram Bharath Hariharan, Joseph Dowling, Michael R. Jones, Vittorio M.
Valletta, Michael J. Gollner, Elaine S. Oran, Anthony Ogorzaly, Eric S. Neu-
mann, Sandra Olson, and Paul Ferkul. Fire Whirls in Microgravity. In 2019
Meeting of the American Society for Gravitational and Space Research, Den-
ver, Colorado, 2019.
[84] H Clay Gabler, Richard A Yetter, and Irvin Glassman. Asymmetric Whirl
Combustion : A New Approach for Non-Premixed Low NOx Gas Turbine
Combustor Design. AIAA Journal, pages 1?11, 1998.
[85] Aaron Alex Frank. Effects of Downscaling on the Low Swirl Burner. PhD
thesis, University of California, Berkeley, 2019.
[86] W. Coenen, P. Rajamanickam, A. D. Weiss, A. L. Sa?nchez, and F. A. Williams.
Swirling flow induced by jets and plumes. Acta Mechanica, 230(6):2221?2231,
2019.
[87] Ritsu Dobashi, Tetsuya Okura, Ryosuke Nagaoka, Yasuhiro Hayashi, and
Toshio Mogi. Experimental Study on Flame Height and Radiant Heat of
Fire Whirls. Fire Technology, 52(4):1069?1080, 7 2016.
[88] Jiao Lei, Naian Liu, Linhe Zhang, Zhihua Deng, Nelson Kudzo Akafuah,
Tianxiang Li, Kozo Saito, and Kohyu Satoh. Burning rates of liquid fuels
in fire whirls. Combustion and Flame, 159(6):2104?2114, 6 2012.
[89] Hanyuan Yu, Song Guo, Minjun Peng, Quanwei Li, Jifeng Ruan, Wei Wan,
and Chen Chen. Study on the influence of air-inlet width on fire whirls com-
bustion characteristic. Procedia Engineering, 62:813?820, 2013.
[90] Kuibin Zhou, Naian Liu, and Xieshang Yuan. Effect of Wind on Fire Whirl
Over a Line Fire. Fire Technology, 52(3):865?875, 2016.
247
[91] Pengfei Wang, Naian Liu, Linhe Zhang, Yueling Bai, and Kohyu Satoh. Fire
Whirl Experimental Facility with No Enclosure of Solid Walls: Design and
Validation. Fire Technology, 51(4):951?969, 2015.
[92] Jiao Lei, Naian Liu, Yan Jiao, and Shaojie Zhang. Experimental investiga-
tion on flame patterns of buoyant diffusion flame in a large range of imposed
circulations. Proceedings of the Combustion Institute, 36(2):3149?3156, 2017.
[93] Jiao Lei, Congcong Ji, Naian Liu, and Linhe Zhang. Effect of imposed cir-
culation on temperature and velocity in general fire whirl: An experimental
investigation. Proceedings of the Combustion Institute, 37(3):4295?4302, 2019.
[94] Shao Lin Lee and Charles A. Garris. Formation of multiple fire whirls. Sym-
posium (International) on Combustion, 12(1):265?273, 1969.
[95] Jiao Lei, Naian Liu, and Ran Tu. Flame height of turbulent fire whirls: A
model study by concept of turbulence suppression. Proceedings of the Com-
bustion Institute, 36(2):3131?3138, 2017.
[96] Kuibin Zhou, Naian Liu, Linhe Zhang, and Koyu Satoh. Thermal Radiation
from Fire Whirls: Revised Solid Flame Model. Fire Technology, 50(6):1573?
1587, 2014.
[97] Jiao Lei, Naian Liu, Linhe Zhang, Haixiang Chen, Lifu Shu, Pu Chen, Zhihua
Deng, Jiping Zhu, Kohyu Satoh, and John L. De Ris. Experimental research
on combustion dynamics of medium-scale fire whirl. Proceedings of the Com-
bustion Institute, 33(2):2407?2415, 2011.
[98] A Bouhafid, J.P. Vantelon, P Joulain, and A.C. Fernandez-Pello. On the flame
structure at the base of a pool fire. Symposium (International) on Combustion,
22(1):1291?1298, 1 1989.
[99] Vytenis Babrauskas. Estimating large pool fire burning rates. Fire Technology,
19(4):251?261, 11 1983.
[100] Chen Kuang, Longhua Hu, Xiaolei Zhang, Yujie Lin, and Larry W. Kostiuk.
An experimental study on the burning rates of n-heptane pool fires with var-
ious lip heights in cross flow. Combustion and Flame, 201:93?103, 2019.
[101] E. J. Hopfinger and G J F V Heijst. Vortices in Rotating Fluids. Annual
Review of Fluid Mechanics, 25(1):241?289, 1 1993.
[102] Jiao Lei and Naian Liu. Reciprocal transitions between buoyant diffusion
flame and fire whirl. Combustion and Flame, 167:463?471, 5 2016.
[103] Kuibin Zhou, Xiaole Qin, Le Zhang, and Yueqiong Wu. An experimental
study of jet fires in rotating flow fields. Combustion and Flame, 210:193?203,
2019.
248
[104] S Soma and K Saito. Reconstruction of fire whirls using scale models. Com-
bustion and Flame, 86(3):269?284, 8 1991.
[105] James G. Quintiere. Fundamentals of Fire Phenomena. John Wiley & Sons,
Ltd, Chichester, UK, 3 2006.
[106] Pierre Joulain. The behavior of pool fires: State of the art and new insights.
Symposium (International) on Combustion, 27(2):2691?2706, 1998.
[107] Zhigang Wang, Wai Cheong Tam, Ki Yong Lee, and Anthony Hamins. Tem-
perature Field Measurements using Thin Filament Pyrometry in a Medium-
Scale Methanol Pool Fire - NIST Technical Note 2031. Technical report, NIST,
2018.
[108] M Klassen and Jay P Gore. Structure and radiation properties of pool fires.
Technical report, National Institute of Standards and Technology, Gaithes-
burg, MD, 1994.
[109] A. Hamins, S. J. Fischer, T. Kashiwagi, M. E. Klassen, and J. P. Gore. Heat
Feedback to the Fuel Surface in Pool Fires. Combustion Science and Technol-
ogy, 97(1-3):37?62, 4 1994.
[110] R. I. EMORI and K. SAITO. A Study of Scaling Laws in Pool and Crib Fires.
Combustion Science and Technology, 31(5-6):217?231, 5 1983.
[111] Longhua Hu. A review of physics and correlations of pool fire behaviour in
wind and future challenges. Fire Safety Journal, 91:41?55, 2017.
[112] Longhua Hu, Junjun Hu, and John L. de Ris. Flame necking-in and instability
characterization in small and medium pool fires with different lip heights.
Combustion and Flame, 162(4):1095?1103, 2015.
[113] Daniel Moreno-Boza, Wilfried Coenen, Jaime Carpio, Antonio L. Sa?nchez,
and Forman A. Williams. On the critical conditions for pool-fire puffing.
Combustion and Flame, 192:426?438, 2018.
[114] Baki M. Cetegen and Tarek A. Ahmed. Experiments on the periodic instability
of buoyant plumes and pool fires. Combustion and Flame, 93(1-2):157?184, 4
1993.
[115] A. Bejan. Predicting the Pool Fire Vortex Shedding Frequency. Journal of
Heat Transfer, 113(1):261?263, 2 1991.
[116] Johanna Aurell, Brian K. Gullett, and Dirk Yamamoto. Emissions from Open
Burning of Simulated Military Waste from Forward Operating Bases. Envi-
ronmental Science & Technology, 46(20):11004?11012, 10 2012.
[117] J.A. Fay. The Spread of Oil Slicks on a Calm Sea. In Oil on the Sea - Ocean
Technology, chapter The Spread, pages 53?63. Springer, Boston, MA, Boston,
MA, 1969.
249
[118] D P Hoult. Oil Spreading on the Sea. Annual Review of Fluid Mechanics,
4(1):341?368, 1 1972.
[119] P. Cooperman. Some criteria for the in situ combustion of crude oil. Journal
of Applied Physics, 30(9):1376?1380, 1959.
[120] Charles H. Peterson, Stanley D. Rice, Jeffrey W. Short, Daniel Esler, James L.
Bodkin, Brenda E. Ballachey, and David B. Irons. Long-Term Ecosystem
Response to the Exxon Valdez Oil Spill. Science, 302(5653):2082?2086, 2003.
[121] Laurens van Gelderen and Grunde Jomaas. The Parameters Controlling the
Burning Efficiency of In-Situ Burning of Crude Oil on Water. In Proceedings
of the 40th AMOP Technical Seminar on Environmental Contamination and
Response, pages 817?832, Calgary, Canada, 2017.
[122] Howard D Ross. Ignition of and flame spread over laboratory-scale pools of
pure liquid fuels. Progress in Energy and Combustion Science, 20(1):17?63,
1994.
[123] Neil Wu, Michael Baker, Gilles Kolb, and Jose L. Torero. Ignition, flame
spread and mass burning characteristics of liquid fuels on a water bed. Spill
Science and Technology Bulletin, 3(4):209?212, 1996.
[124] Neil Wu, Gilles Kolb, and Jose L. Torero. Piloted ignition of a slick of oil
on a water sublayer: The effect of weathering. Symposium (International) on
Combustion, 27(2):2783?2790, 1998.
[125] Ian Buist. Window-of-opportunity for in situ burning. Spill Science and
Technology Bulletin, 8(4):341?346, 2003.
[126] J. L. Torero, Stephen M. Olenick, J. P. Garo, and J. P. Vantelon. Determina-
tion of the burning characteristics of a slick of oil on water. Spill Science and
Technology Bulletin, 8(4):379?390, 2003.
[127] A. Y. Walavalkar and A. K. Kulkarni. Combustion of water-in-oil emulsion
layers supported on water. Combustion and Flame, 125(1-2):1001?1011, 2001.
[128] Nicholas L. Brogaard, Martin X. S?rensen, Janne Fritt-Rasmussen, Ali S.
Rangwala, and Grunde Jomaas. A new experimental rig for oil burning on
water - Results for crude and pure oils. Fire Safety Science, 11:1481?1494,
2014.
[129] Irvin Glassman. Supersonic flight and cooking over wood-burning stoves:
Challenges to the combustion community. Proceedings of the Combustion In-
stitute, 28(1):1?10, 2000.
[130] I. Glassman and R. A. Yetter. Combustion. Elsevier, 4th edition, 2008.
250
[131] Michael Faraday. The Chemical History of a Candle. Chatto and Windus,
London, 1908.
[132] Bill Hammack and Don DeCoste. Michael Faraday?s The Chemical History of
a Candle (Edited). Articulate Noise Books, 2016.
[133] U. Bonne, K. H. Homann, and H. G.G. Wagner. Carbon formation in premixed
flames. Symposium (International) on Combustion, 10(1):503?512, 1965.
[134] Fang. Xu. Soot growth in laminar premixed flames. PhD thesis, University of
Michigan, 1999.
[135] W Flower. Optical Measurements of Soot Formation in Premixed Flames.
Combustion Science and Technology, 33:17?33, 4 2007.
[136] Peter Bradford Sunderland. Soot formation in laminar jet diffusion flames.
PhD thesis, University of Michigan, Ann Arbor, 1995.
[137] Charles A. Amann and Donald C. Siegla. Diesel particulates?What they are
and why. Aerosol Science and Technology, 1(1):73?101, 1982.
[138] Jonathan O. Anderson, Josef G. Thundiyil, and Andrew Stolbach. Clearing
the Air: A Review of the Effects of Particulate Matter Air Pollution on Human
Health. Journal of Medical Toxicology, 8(2):166?175, 2012.
[139] EPA. Regulations for Emissions from Vehicles and Engines
(https://www.epa.gov/regulations-emissions-vehicles-and-engines).
[140] J. H. Kent. A quantitative relationship between soot yield and smoke point
measurements. Combustion and Flame, 63(3):349?358, 1986.
[141] J. De Ris and X. Cheng. The Of Role Of Smoke-point In Material Flamma-
bility Testing. Fire Safety Science, 4:301?312, 1994.
[142] Robert J. Santoro and Thomas F. Richardson. Concentration and Temper-
ature Effects on Soot Formation in Diffusion Flames. In Henning Bockhorn,
editor, Soot Formation in Combustion, chapter 13, pages 221?238. Springer-
Verlag, Berlin, 1994.
[143] K.-C. Lin and G. M. Faeth. Hydrodynamic suppression of soot emissions in
laminar diffusion flames. Journal of Propulsion and Power, 12(1):10?17, 1996.
[144] K C Lin and G M Faeth. Effects of hydrodynamics on soot formation in lami-
nar opposed-jet diffusion flames. Journal of Propulsion and Power, 12(4):691?
698, 1996.
[145] K.-C Lin and G M Faeth. State Relationships of Laminar Permanently-Blue
Opposed-Jet Hydrocarbon-Fueled Diffusion Flames. Int. J. Environ. Combust.
Tech. Vol I, pages 53?79, 2000.
251
[146] S. M. Mahmoud, G. J. Nathan, Z. T. Alwahabi, Z. W. Sun, P. R. Medwell,
and B. B. Dally. The effect of exit strain rate on soot volume fraction in
turbulent non-premixed jet flames. Proceedings of the Combustion Institute,
36(1):889?897, 2017.
[147] B. A. Strayer, D. Dunn-Rankin, and F. Jabbari. A comparison between
frequency- and amplitude-modulated adaptive control of a non-premixed
flame. Symposium (International) on Combustion, 27(1):1247?1254, 1998.
[148] Akshit Markan, Peter B. Sunderland, James G. Quintiere, John L. de Ris,
Dennis P. Stocker, and Howard R. Baum. A Burning Rate Emulator (BRE)
for study of condensed fuel burning in microgravity. Combustion and Flame,
192:272?282, 2018.
[149] M. S. Cha, S. M. Lee, K. T. Kim, and S. H. Chung. Soot suppression by
nonthermal plasma in coflow jet diffusion flames using a dielectric barrier
discharge. Combustion and Flame, 141(4):438?447, 2005.
[150] R. A. Dobbins and C. M. Megaridis. Morphology of Flame-Generated Soot
As Determined by Thermophoretic Sampling. Langmuir, 3(2):254?259, 1987.
[151] Irvin Glassman. Soot formation in combustion processes. Symposium (Inter-
national) on Combustion, 22(1):295?311, 1 1989.
[152] J. Du and R. L. Axelbaum. The effect of flame structure on soot-particle
inception in diffusion flames. Combustion and Flame, 100(3):367?375, 1995.
[153] K H Homann and H G Wagner. Chemistry of Carbon Formation in Flames.
Proceedings of the Royal Society of London. Series A, Mathematical and Phys-
ical Sciences, 307(1489):141?152, 3 1968.
[154] B. S. Haynes and H. Gg Wagner. Soot formation. Progress in Energy and
Combustion Science, 7(4):229?273, 1981.
[155] M. Bachmann, W. Wiese, and K. H. Homann. PAH and aromers: Precursors
of fullerenes and soot. Symposium (International) on Combustion, 26(2):2259?
2267, 1996.
[156] Charles S. McEnally, U?mit O? Ko?ylu?, Lisa D. Pfefferle, and Daniel E. Rosner.
Soot volume fraction and temperature measurements in laminar nonpremixed
flames using thermocouples. Combustion and Flame, 109(4):701?720, 1997.
[157] Samuel L. Manzello, David B. Lenhert, Ahmet Yozgatligil, Michael T. Dono-
van, George W. Mulholland, Michael R. Zachariah, and Wing Tsang. Soot
particle size distributions in a well-stirred reactor/plug flow reactor. Proceed-
ings of the Combustion Institute, 31 I(1):675?683, 2007.
252
[158] U? O? Ko?ylu?, C. S. Mcenally, D. E. Rosner, and L. D. Pfefferle. Simultaneous
measurements of soot volume fraction and particle size / microstructure in
flames using a thermophoretic sampling technique. Combustion and Flame,
110(4):494?507, 1997.
[159] Chiara Saggese, Alberto Cuoci, Alessio Frassoldati, Sara Ferrario, Joaquin
Camacho, Hai Wang, and Tiziano Faravelli. Probe effects in soot sampling
from a burner-stabilized stagnation flame. Combustion and Flame, 167:184?
197, 2016.
[160] Nathan J. Kempema, Bin Ma, and Marshall B. Long. Investigation of in-flame
soot optical properties in laminar coflow diffusion flames using thermophoretic
particle sampling and spectral light extinction. Applied Physics B: Lasers and
Optics, 122(9):1?13, 2016.
[161] Alex M. Vargas and O?mer L. Gu?lder. A multi-probe thermophoretic soot
sampling system for high-pressure diffusion flames. Review of Scientific In-
struments, 87(5), 2016.
[162] Randall L. Vander Wal. A TEM Methodology for the Study of Soot Particle
Structure. Combustion Science and Technology, 126(1-6):333?351, 7 1997.
[163] Paul M. Anderson, Haiqing Guo, and Peter B. Sunderland. Repeatability and
reproducibility of TEM soot primary particle size measurements and compari-
son of automated methods. Journal of Aerosol Science, 114(October):317?326,
2017.
[164] Haiqing Guo, Jose A. Castillo, and Peter B. Sunderland. Digital camera
measurements of soot temperature and soot volume fraction in axisymmetric
flames. Applied Optics, 52(33):8040?8047, 2013.
[165] Catherine Hamel, Fernando Raffan-Montoya, and Stanislav I. Stoliarov. A
method for measurement of spatially resolved radiation intensity and radiative
fraction of laminar flames of gaseous and solid fuels. Experimental Thermal
and Fluid Science, 104:153?163, 6 2019.
[166] Peter B. Kuhn, Bin Ma, Blair C. Connelly, Mitchell D. Smooke, and Mar-
shall B. Long. Soot and thin-filament pyrometry using a color digital camera.
Proceedings of the Combustion Institute, 33(1):743?750, 2011.
[167] Dennis K. Kim and Peter B. Sunderland. Fire ember pyrometry using a color
camera. Fire Safety Journal, 106(February):88?93, 2019.
[168] H. A. Michelsen, C. Schulz, G. J. Smallwood, and S. Will. Laser-induced
incandescence: Particulate diagnostics for combustion, atmospheric, and in-
dustrial applications. Progress in Energy and Combustion Science, 51:2?48,
2015.
253
[169] R. L. Vander Wal and K. J. Weiland. Laser-induced incandescence: Devel-
opment and characterization towards a measurement of soot-volume fraction.
Applied Physics B Laser and Optics, 59(4):445?452, 1994.
[170] Seong Young Lee, Stephen R. Turns, and Robert J. Santoro. Measurements
of soot, OH, and PAH concentrations in turbulent ethylene/air jet flames.
Combustion and Flame, 156(12):2264?2275, 2009.
[171] Chen Gu, He Lin, Joaquin Camacho, Baiyang Lin, Can Shao, Ruoxin Li,
Hao Gu, Bin Guan, Zhen Huang, and Hai Wang. Particle size distribution of
nascent soot in lightly and heavily sooting premixed ethylene flames. Com-
bustion and Flame, 165:177?187, 2016.
[172] R.J. Santoro, H.G. Semerjian, and R.A. Dobbins. Soot particle measurements
in diffusion flames. Combustion and Flame, 51(3):203?218, 1 1983.
[173] M. Y. Choi, G. W. Mulholland, A. Hamins, and T. Kashiwagi. Comparisons
of the soot volume fraction using gravimetric and light extinction techniques.
Combustion and Flame, 102(1-2):161?169, 1995.
[174] Simone Amaral, Joa?o de Carvalho, Maria Costa, and Cleverson Pinheiro.
An Overview of Particulate Matter Measurement Instruments. Atmosphere,
6(9):1327?1345, 9 2015.
[175] J. H. Sun and S. H. Chan. A time-resolved measurement technique for partic-
ulate number density in diesel exhaust using a fast-response flame ionization
detector. Measurement Science and Technology, 8(3):279?286, 1997.
[176] Florentin M.J. Bulot, Steven J. Johnston, Philip J. Basford, Natasha H.C.
Easton, Mihaela Apetroaie-Cristea, Gavin L. Foster, Andrew K.R. Morris,
Simon J. Cox, and Matthew Loxham. Long-term field comparison of mul-
tiple low-cost particulate matter sensors in an outdoor urban environment.
Scientific Reports, 9(1):1?13, 2019.
[177] Nathaniel May, Evan Ellicott, and Michael Gollner. An examination of fuel
moisture, energy release and emissions during laboratory burning of live wild-
land fuels. International Journal of Wildland Fire, 28(3):187?197, 2019.
[178] Andreas Petzold, Richard Marsh, Mark Johnson, Michael Miller, Yura Sev-
cenco, David Delhaye, Amir Ibrahim, Paul Williams, Heidi Bauer, Andrew
Crayford, William D. Bachalo, and David Raper. Evaluation of methods for
measuring particulate matter emissions from gas turbines. Environmental Sci-
ence and Technology, 45(8):3562?3568, 2011.
[179] Jerzy Merkisz and Jacek Pielecha. Nanoparticle Emissions From Combustion
Engines, volume 8 of Springer Tracts on Transportation and Traffic. Springer
International Publishing, Cham, 2015.
254
[180] Hans Grimm and Delbert J. Eatough. Aerosol measurement: The use of
optical light scattering for the determination of particulate size distribution,
and particulate mass, including the semi-volatile fraction. Journal of the Air
and Waste Management Association, 59(1):101?107, 2009.
[181] S.B. Hariharan, Y. Hu, H. Xiao, M.J. Gollner, and Elaine S. Oran. The
Structure of the Blue Whirl. In 70th Annual Meeting of the APS Division of
Fluid Dynamics, Denver, Colorado, 2017.
[182] J. Warnatz, Ulrich Maas, and Robert W. Dibble. Combustion. Springer, 4th
edition, 2006.
[183] A C Eckbreth. Laser Diagnostics for Temperature and Species in Unsteady
Combustion. In F Culick, M V Heitor, and J H Whitelaw, editors, Unsteady
Combustion, pages 393?410. Springer Netherlands, Dordrecht, 1996.
[184] Christof Schulz, Andreas Dreizler, Volker Ebert, and Ju?rgen Wolfrum. Com-
bustion Diagnostics BT - Springer Handbook of Experimental Fluid Mechan-
ics. In Cameron Tropea, Alexander L Yarin, and John F Foss, editors, Springer
Handbook of Experimental Fluid Mechanics, pages 1241?1315. Springer Berlin
Heidelberg, Berlin, Heidelberg, 2007.
[185] T. Kathrotia, M. Fikri, M. Bozkurt, M. Hartmann, U. Riedel, and C. Schulz.
Study of the H+O+M reaction forming OH*: Kinetics of OH* chemilumines-
cence in hydrogen combustion systems. Combustion and Flame, 157(7):1261?
1273, 2010.
[186] Jun Kojima, Yuji Ikeda, and Tsuyoshi Nakajima. Basic aspects of OH(A),
CH(A), and C2(d) chemiluminescence in the reaction zone of laminar methane-
air premixed flames. Combustion and Flame, 140(1-2):34?45, 2005.
[187] Venkata Nori and Jerry Seitzman. Evaluation of Chemiluminescence as a
Combustion Diagnostic Under Varying Operating Conditions. In 46th AIAA
Aerospace Sciences Meeting and Exhibit, number January, pages 1?14, Reston,
Virigina, 1 2008. American Institute of Aeronautics and Astronautics.
[188] C. S. Panoutsos, Y. Hardalupas, and A. M K P Taylor. Numerical evalua-
tion of equivalence ratio measurement using OH* and CH* chemiluminescence
in premixed and non-premixed methane-air flames. Combustion and Flame,
156(2):273?291, 2009.
[189] Thomas P Clark. Studies of OH, CO, CH, and C(sub 2) Radiation from Lam-
inar and Turbulent Propane-air and Ethylene-air Flames. Technical report,
National Advisory Committee for Aeronautics. Lewis Flight Propulsion Lab,
Cleveland, Ohio, 1958.
[190] Yuji Ikeda, Jun Kojima, and Hideki Hashimoto. Local chemiluminescence
spectra measurements in a high-pressure laminar methane/air premixed flame.
Proceedings of the Combustion Institute, 29(2):1495?1501, 2002.
255
[191] B. Higgins, M. Q. McQuay, F. Lacas, and S. Candel. An experimental study
on the effect of pressure and strain rate on CH chemiluminescence of premixed
fuel-lean methane/air flames. Fuel, 80(11):1583?1591, 2001.
[192] Y. Hardalupas and M. Orain. Local measurements of the time-dependent
heat release rate and equivalence ratio using chemiluminescent emission from
a flame. Combustion and Flame, 139(3):188?207, 2004.
[193] Jun Kojima, Yuji Ikeda, and Tsuyoshi Nakajima. Spatially resolved measure-
ment of OH, CH, and C2 chemiluminescence in the reaction zone of lami-
nar methane/air premixed flames. Proceedings of the Combustion Institute,
28(2):1757?1764, 2000.
[194] Yong Ki Jeong, Chung Hwan Jeon, and Young June Chang. Evaluation of the
equivalence ratio of the reacting mixture using intensity ratio of chemilumi-
nescence in laminar partially premixed CH4-air flames. Experimental Thermal
and Fluid Science, 30(7):663?673, 2006.
[195] T. M. Muruganandam, B. H. Kim, M. R. Morrell, V. Nori, M. Patel, B. W.
Romig, and J. M. Seitzman. Optical equivalence ratio sensors for gas turbine
combustors. Proceedings of the Combustion Institute, 30(1):1601?1608, 2005.
[196] Chadwick C. Rasmussen, Sulabh K. Dhanuka, and James F. Driscoll. Visu-
alization of flameholding mechanisms in a supersonic combustor using PLIF.
Proceedings of the Combustion Institute, 31 II:2505?2512, 2007.
[197] Z. S. Li, B. Li, Z. W. Sun, X. S. Bai, and M. Alde?n. Turbulence and combustion
interaction: High resolution local flame front structure visualization using
simultaneous single-shot PLIF imaging of CH, OH, and CH2O in a piloted
premixed jet flame. Combustion and Flame, 157(6):1087?1096, 2010.
[198] Mamoru Tanahashi, Shinichirou Murakami, Gyung Min Choi, Yuichi Fukuchi,
and Toshio Miyauchi. Simultaneous CH-OH PLIF and stereoscopic PIV mea-
surements of turbulent premixed flames. Proceedings of the Combustion Insti-
tute, 30(1):1665?1672, 2005.
[199] Mingfa Yao, Zhaolei Zheng, and Haifeng Liu. Progress and recent trends in
homogeneous charge compression ignition (HCCI) engines. Progress in Energy
and Combustion Science, 35(5):398?437, 2009.
[200] Swetaprovo Chaudhuri, Stanislav Kostka, Michael W. Renfro, and Baki M.
Cetegen. Blowoff dynamics of bluff body stabilized turbulent premixed flames.
Combustion and Flame, 157(4):790?802, 2010.
[201] Bikram R. Chowdhury and Baki M. Cetegen. Effects of free stream flow
turbulence on blowoff characteristics of bluff-body stabilized premixed flames.
Combustion and Flame, 190:302?316, 2018.
256
[202] S. Bo?ckle, J. Kazenwadel, T. Kunzelmann, and C. Schulz. Laser-diagnostic
multi-species imaging in strongly swirling natural gas flames. Applied Physics
B: Lasers and Optics, 71(5):741?746, 2000.
[203] S. R. Pezeshki, M. W. Hester, Q. Lin, and J. A. Nyman. The effects of oil
spill and clean-up on dominant US Gulf coast marsh macrophytes: A review.
Environmental Pollution, 108(2):129?139, 2000.
[204] Robin J Law and Jocelyne Hellou. Contamination of fish and shellfish following
oil spill incidents. Environmental Geosciences, 6(2):90?98, 6 1999.
[205] Michael J. Moran and Howard N. Shapiro. Engineering Thermodynamics.
John Wiley & Sons, Inc., 5th edition, 2006.
[206] National-Response-Team. Guidance on Burning Spilled Oil in Situ. Technical
Report December 1995, Office of Response and Restoration, 1995.
[207] Michael A. Peters. Oil geopolitics and eco-nightmares. Educational Philosophy
and Theory, 49(5):435?438, 2017.
[208] Mervin F Fingas, Greg Halley, Francine Ackerman, Robert Nelson, Martine
Bissonnette, Nanci Laroche, Zhendi Wang, Patrick Lambert, Ken Li, Paula
Jokuty, Gary Sergy, Edward J Tennyson, Joe Mullin, Larry Hannon, Wayne
Halley, John Latour, Roy Galarneau, Bill Ryan, Rod Turpin, Phil Campagna,
Don V Aurand, and Robert R Hiltabrand. The Newfoundland Offshore
Burn Experiment?NOBE. International Oil Spill Conference Proceedings,
1995(1):123?132, 2 1995.
[209] Ian Buist, Ken Trudel, Jake Morrison, and Don Aurand. Laboratory Studies
of the Properties of In-Situ Burn Residues. International Oil Spill Conference
Proceedings, 1997(1):149?156, 4 1997.
[210] Ian Buist, Steve Potter, Tim Nedwed, and Joseph Mullin. Herding surfactants
to contract and thicken oil spills in pack ice for in situ burning. Cold Regions
Science and Technology, 67(1-2):3?23, 2011.
[211] Qianxin Lin, Irving A. Mendelssohn, Kenneth Carney, Scott M. Miles, Nel-
son P. Bryner, and William D. Walton. In-situ burning of oil in coastal
marshes. 2. Oil spill cleanup efficiency as a function of oil type, marsh type,
and water depth. Environmental Science and Technology, 39(6):1855?1860,
2005.
[212] IPIECA and IOGP. In-Situ Burning of Spilled Oil - Good practice guidelines
for incident management and emergency response personnel. Technical report,
IPIECA, IOGP, London, UK, 2016.
[213] TSI Incorporated. DustTrak DRX Aerosol Monitor Model 8534 Operation
and Service Manual. Technical Report January, TSI Incorporated, 2017.
257
[214] S. Brohez, C. Delvosalle, and G. Marlair. A two-thermocouples probe for
radiation corrections of measured temperatures in compartment fires. Fire
Safety Journal, 39(5):399?411, 2004.
[215] J. P. Garo, J. P. Vantelon, and A. C. Fernandez-Pello. Boilover burning of oil
spilled on water. Symposium (International) on Combustion, 25(1):1481?1488,
1994.
[216] C.S. Cragoe. Thermal Properties of Petroleum Products. Technical report,
Bureau of Standards, 1933.
[217] Benjamin D. Ditch, John L. de Ris, Thomas K. Blanchat, Marcos Chaos,
Robert G. Bill, and Sergey B. Dorofeev. Pool fires - An empirical correlation.
Combustion and Flame, 160(12):2964?2974, 2013.
[218] Michael J. Gollner, Elaine S. Oran, Sriram Bharath Hariharan, Joseph Dowl-
ing, Hamed F. Farahani, and Ali S. Rangwala. Efficient Remediation of Oil
Spills over Water using Fire Whirls (Project Number 1094). Technical report,
Bureau of Safety and Environmental Enforcement, Washington, DC, 2019.
[219] Sriram Bharath Hariharan, Joseph Dowling, Hamed F. Farahani, Yu Hu, Ali
Tohidi, Karen Stone, Ali S. Rangwala, Elaine S. Oran, and Michael J. Goll-
ner. Demonstration of reduction in emissions by fire whirls through small scale
experiments. In 2019 AMOP Technical Seminar on Environmental Contami-
nation and Response, Halifax, Nova Scotia, Canada, 2019.
[220] Sriram B Hariharan, Joseph Dowling, Hamed F Farahani, Michael J Goll-
ner, Elaine S Oran, and Karen Stone. Comparison of emissions from liquid-
fueled pool fires and fire whirls. In 11th U.S. National Combustion Meeting,
Pasadena, California, 2019.
[221] ExxonMobil. Alaska North Slope Material Safety Data Sheet, 2018.
[222] R. J. Yokelson, J. G. Goode, D. E. Ward, R. A. Susott, R. E. Babbitt, D. D.
Wade, I. Bertschi, D. W. T. Griffith, and W. M. Hao. Emissions of formalde-
hyde, acetic acid, methanol, and other trace gases from biomass fires in North
Carolina measured by airborne Fourier transform infrared spectroscopy. Jour-
nal of Geophysical Research: Atmospheres, 104(D23):30109?30125, 12 1999.
[223] Wei Tang, Colin H. Miller, and Michael J. Gollner. Local flame attachment and
heat fluxes in wind-driven line fires. Proceedings of the Combustion Institute,
36(2):3253?3261, 2017.
[224] Kuibin Zhou, Naian Liu, Jesse S. Lozano, Yanlong Shan, Bin Yao, and Ko-
hyu Satoh. Effect of flow circulation on combustion dynamics of fire whirl.
Proceedings of the Combustion Institute, 34(2):2617?2624, 2013.
258
[225] Katsuhisa Noto, Kenji Teramoto, and Tsuyoshi Nakajima. Spectra and crit-
ical grashof numbers for turbulent transition in a thermal plume. Journal of
Thermophysics and Heat Transfer, 13(1):82?90, 1999.
[226] RUI ZHOU and ZI-NIU WU. Fire whirls due to surrounding flame sources
and the influence of the rotation speed on the flame height. Journal of Fluid
Mechanics, 583:313?345, 7 2007.
[227] Laurens Van Gelderen, Linus M V Malmquist, and Grunde Jomaas. Vapor-
ization order and burning efficiency of crude oils during in - situ burning on
water. Fuel, 191:528?537, 2017.
[228] Keng Hoo Chuah, Kazunori Kuwana, and Kozo Saito. Modeling a fire whirl
generated over a 5-cm-diameter methanol pool fire. Combustion and Flame,
156(9):1828?1833, 2009.
[229] Sidney Leibovich. The Structure of Vortex Breakdown. Annual Review of
Fluid Mechanics, 10(11):221?246, 1978.
[230] J H Faler and S Leibovich. Disrupted states of vortex flow and vortex break-
down. Physics of Fluids, 20(9):1385, 1977.
[231] Turgut Sarpkaya. On stationary and travelling vortex breakdowns. Journal
of Fluid Mechanics, 45(03):545?559, 1971.
[232] A K Gupta, D G Lilley, and N Syred. Swirl flows. Abacus Press, Tunbridge
Wells, Kent, 1984.
[233] Michael D. Abra?moff, Paulo J. Magalha?es, and Sunanda J. Ram. Image pro-
cessing with imageJ. Biophotonics International, 11(7):36?41, 2004.
[234] Nobuyuki Otsu. A Threshold Selection Method from Gray-Level Histograms.
IEEE Transactions on Systems, Man, and Cybernetics, 9(1):62?66, 1 1979.
[235] Gunnar Heskestad. Luminous heights of turbulent diffusion flames. Fire Safety
Journal, 5(2):103?108, 1983.
[236] Bruce R. Munson, Donald F. Young, Theodore H. Okiishi, and Wade W.
Huebsch. Fundamentals Sixth Edition of Fluid Mechanics. John Wiley &
Sons, Inc., 6th edition, 2010.
[237] E. E. Zukoski. Properties of fire plumes. In Combustion Fundamentals of Fire,
pages 101?219. Academic Press Ltd., 1995.
[238] Sriram Bharath Hariharan, Yu Hu, Michael J. Gollner, and Elaine S. Oran.
Conditions for Formation of the Blue Whirl. In 71st Annual Meeting of the
APS Division of Fluid Dynamics, volume 63, Atlanta, Georgia, 2018.
259
[239] James D. Smith and Volker Sick. High-speed fuel tracer fluorescence and OH
radical chemiluminescence imaging in a spark-ignition direct-injection engine.
Applied Optics, 44(31):6682?6691, 2005.
[240] Zhi Wang, Yunliang Qi, Xin He, Jianxin Wang, Shijing Shuai, and Chung K.
Law. Analysis of pre-ignition to super-knock: Hotspot-induced deflagration
to detonation. Fuel, 144:222?227, 2015.
[241] J. D. Smith and V. Sick. Crank-angle resolved imaging of biacetyl laser-
induced fluorescence in an optical internal combustion engine. Applied Physics
B: Lasers and Optics, 81(5):579?584, 2005.
[242] James D. Smith and Volker Sick. Quantitative, dynamic fuel distribution
measurements in combustion-related devices using laser-induced fluorescence
imaging of biacetyl in iso-octane. Proceedings of the Combustion Institute,
31(1):747?755, 1 2007.
[243] R. S. Barlow, R. W. Dibble, J. Y. Chen, and R. P. Lucht. Effect of Damko?hler
number on superequilibrium OH concentration in turbulent nonpremixed jet
flames. Combustion and Flame, 82(3-4):235?251, 1990.
[244] Sriram Bharath Hariharan, Yejun Wang, Paul M. Anderson, Michael J. Goll-
ner, Waruna D. Kulatilaka, and Elaine S. Oran. On the nature of combustion
in the blue whirl using optical diagnostics (In preparation). 2020.
[245] P. G. Aleiferis, Y. Hardalupas, A. M.K.P. Taylor, K. Ishii, and Y. Urata. Flame
chemiluminescence studies of cyclic combustion variations and air-to-fuel ratio
of the reacting mixture in a lean-burn stratified-charge spark-ignition engine.
Combustion and Flame, 136(1-2):72?90, 2004.
[246] J.W. Dold. Flame propagation in a nonuniform mixture: Analysis of a slowly
varying Triple Flame. Combustion and Flame, 76(1):71?88, 4 1989.
[247] J. Buckmaster. Edge-flames. Journal of Engineering Mathematics, 31(2-
3):269?284, 1997.
[248] J. Buckmaster. Edge-flames. Progress in Energy and Combustion Science,
28(5):435?475, 2002.
[249] Zhengyang Wang, Peter B. Sunderland, and Richard L. Axelbaum. Double
blue zones in inverse and normal laminar jet diffusion flames. Combustion and
Flame, 211:253?259, 2020.
[250] J. H. Faler and S Leibovich. An experimental map of the internal structure
of a vortex breakdown. Journal of Fluid Mechanics, 86(1978):313?335, 1978.
[251] Kazunori Kuwana, Kozo Sekimoto, Takeaki Minami, Takahiro Tashiro, and
Kozo Saito. Scale-model experiments of moving fire whirl over a line fire.
Proceedings of the Combustion Institute, 34(2):2625?2631, 2013.
260
[252] Hampton Clay Gabler. An Experimental and Numerical Investigation of
Asymmetrically-Fueled Whirl Flames. PhD thesis, Princeton University, 1998.
[253] F. A. Williams. Urban and wildland fire phenomenology. Progress in Energy
and Combustion Science, 8(4):317?354, 1982.
[254] John B Heywood. Internal Combustion Engine Fundamentals, volume 21.
McGraw-Hill, 1988.
[255] Steven G. Tuttle, Katherine Hinnant, Thomas N. Loegel, and Brian T. Fisher.
Development of a Low-Emission Spray Combustor for Emulsified Crude Oil.
Technical report, Bureau of Safety and Environmental Enforcement, Wash-
ington, DC, 2017.
[256] Johannes Schindelin, Ignacio Arganda-Carreras, Erwin Frise, Verena Kaynig,
Mark Longair, Tobias Pietzsch, Stephan Preibisch, Curtis Rueden, Stephan
Saalfeld, Benjamin Schmid, Jean-Yves Tinevez, Daniel James White, Volker
Hartenstein, Kevin Eliceiri, Pavel Tomancak, and Albert Cardona. Fiji: an
open-source platform for biological-image analysis. Nature Methods, 9(7):676?
682, 7 2012.
[257] S. G. Davis and C. K. Law. Laminar flame speeds and oxidation kinetics of
iso-octane-air and n-heptane-air flames. Symposium (International) on Com-
bustion, 27(1):521?527, 1998.
261