Published on: Mar 3, 2016
Transcripts - NarayanaBK122014T
THREE-DIMENSIONAL OPTICAL MEASUREMENTS IN AN ETHYLENE
FUELLED MODEL SCRAMJET ENGINE
Submitted to the Graduate School
of the University of Notre Dame
in Partial Fulﬁllment of the Requirements
for the Degree of
Master of Science
Bhargava K. Narayana
Hyungrok Do, Director
Graduate Program in Aerospace and Mechanical Engineering
Notre Dame, Indiana
c Copyright by
Bhargava Kumar Narayana
All Rights Reserved
THREE-DIMENSIONAL OPTICAL MEASUREMENTS IN AN ETHYLENE
FUELLED MODEL SCRAMJET ENGINE
Bhargava K. Narayana
This work documents the development of non-intrusive optical diagnostic methods
towards a qualitative study of ethylene ﬂame dynamics in a laboratory scale model
scramjet engine. Planar laser Rayleigh scattering (PLRS) and OH based planar laser
induced ﬂuorescence (PLIF) have been successfully developed and applied.
Prior to understanding the turbulent ﬂame dynamics due to ethylene combustion
in the model scramjet, it is necessary to reveal the role played by turbulent struc-
tures in a combustion free environment. Also, shock/ turbulent boundary layers are
known to signiﬁcantly impact unstart dynamics. Hence, PLRS has been chosen to
be employed considering its relevancy to the present experimental subject.
Visualizing ﬂame structures in a transient combustion system is a key to estab-
lishing stable operational regimes. Imaging ground state OH is a proven, simple and
cost eﬀective method amongst the LIF based techniques. In addition, these laser
based techniques are instantaneous in nature with temporal resolution as high as
Flow physics in the scramjet model is complicated due to the interaction of tur-
bulence and ﬂame structures. High intensities of turbulence are expected at such
high Reynolds number ﬂows involving combustion. The high strain rates imposed
by turbulent structures might, in fact, contribute to ﬂame extinguishment. In view
Bhargava K. Narayana
of turbulence being a 3-dimensional phenomena, there exists a need to visualize the
ﬂow proﬁle in a 3-dimensional domain. However, a truly 3-dimensional study is be-
yond the scope of current research methods. A closer and more accessible alternative
would be to apply 2-dimensional ﬂow imaging techniques spanning over multiple
planes, provided that the ﬂow exhibits a quasi-stable behavior. Although optical
investigations in the combustor regions have been reported, this study, to the best
of the author’s knowledge, is the ﬁrst one to cater to the ﬂow ﬁeld investigation over
a signiﬁcant region beyond the combustor/cavity in supersonic ﬂows. Furthermore,
this study encompasses multiple planes to achieve a holistic reconstruction of the ﬂow
A unique optical arrangement to aid such a visualization has been developed.
The results obtained provide supportive evidence underlining the applicability of
these laser based techniques to the present combustion system.
FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
SYMBOLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
CHAPTER 1: INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 Fluorescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
22.214.171.124 Laser-induced Fluorescence . . . . . . . . . . . . . . 4
126.96.36.199 Linear Regime . . . . . . . . . . . . . . . . . . . . . 4
188.8.131.52 Saturated Regime . . . . . . . . . . . . . . . . . . . 4
184.108.40.206 OH PLIF . . . . . . . . . . . . . . . . . . . . . . . . 6
220.127.116.11 Disadvantages . . . . . . . . . . . . . . . . . . . . . . 7
1.2.2 Chemiluminescence . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.3 Rayleigh Scattering . . . . . . . . . . . . . . . . . . . . . . . . 8
CHAPTER 2: EXPERIMENTAL SETUP . . . . . . . . . . . . . . . . . . . . 10
2.1 Hypersonic Wind Tunnel and Associated Instrumentation . . . . . . . 10
2.2 Optical Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Timing Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4 Wavelength Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4.1 Theoretical Spectral Database . . . . . . . . . . . . . . . . . . 18
2.5 Transient Combustion System . . . . . . . . . . . . . . . . . . . . . . 19
2.6 Condensed Manual . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
CHAPTER 3: RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.1 PLRS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2 Chemiluminescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3 OH PLIF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
CHAPTER 4: CONCLUSIONS AND RECOMMENDATIONS . . . . . . . . 34
4.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.2 Recommendations for Future Work . . . . . . . . . . . . . . . . . . . 35
BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
1.1 Basic physical processes aﬀecting the ﬂuorescence rate depicted in a
two level system. Rate constants; b12 - stimulated absorption ; b21
-emission rate constants; A21 - spontaneous emission rate constant ;
Q21 quenching rate constant; W2i photoionization rate constant; P -
predissociation rate constant. Adopted from  . . . . . . . . . . . . 3
1.2 Depiction of LIF signal dependence on laser excitation energy. Signal
response is linear for low pulse energies. Signal response is highest for
saturated regime and doesn’t increase with increasing laser energy  5
1.3 OH (left) and CH2O (right) LIF signals from a co-axial burner from
Li. Note the post ﬂame existence of OH radicals denoting the
region of burned gases. Also notable is the prevalence of OH signature
over that of CH2O, denoting unburned gases. . . . . . . . . . . . . . . 6
1.4 Schlieren images of HyShotII combustor: (top) instantaneous; (mid-
dle) averaged over test time by Laurence  and (bottom) Rayleigh
scattering images from the present scramjet model . . . . . . . . . . . 9
2.1 Schematic of the OH PLIF and PLRS measurements . . . . . . . . . 10
2.2 Sheet generation optics . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Sheet generation optics for OH PLIF and PLRS measurements . . . . 13
2.4 Schematic of conventional laser beam expansion optics . . . . . . . . 14
2.5 View of the stainless steel optical enclosure . . . . . . . . . . . . . . 15
2.6 View of the stainless steel optical enclosure: (left) without and (right)
with streamlined deﬂector hood . . . . . . . . . . . . . . . . . . . . . 16
2.7 Timing diagram of the simultaneous operation of OH PLIF, fuel in-
jection valve and the ICCD camera . . . . . . . . . . . . . . . . . . . 17
2.8 Sample wavelength scan in the range 282-284 nm using the sirah dye
laser by Jalbert  . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.9 Emission spectra generated by LIFBASE. Transition lines of interest
are marked. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.10 Schematic (not to scale) of the scramjet model used in the experiments
depicting the fuel injection port and cavity combustor. . . . . . . . . 20
3.1 Detailed ﬂow features of Rayleigh scattering images in the scramjet
central plane: (top) with and (bottom) without active fuel jet opera-
tion. Free stream ﬂow is at Mach = 4.5 and from left to right. . . . . 23
3.2 Set of detailed Rayleigh scattering images arranged based on their
proximity to the central plane (x=0) of the model. Fuel jet injection
with N2 is enabled. Free stream ﬂow is at Mach = 4.5 and from left
to right. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3 Set of detailed ﬂow features of Rayleigh scattering images in the scram-
jet central plane: (top) with and (bottom) without active fuel jet op-
eration (N2). Free stream ﬂow is at Mach = 4.5 and from left to right. 25
3.4 Detailed chemiluminescence image of the combustion process. Brighter
(blue) regions are indicative of intense heat release reactions. Free
stream ﬂow direction is from left to right. . . . . . . . . . . . . . . . . 26
3.5 Sequence of PLIF images taken ∆T =100 ms apart from each other
(numbered), at one of the scramjet investigation planes. Free stream
ﬂow direction is from left to right. Fuel jet was active for 300 ms. . . 27
3.6 Detailed set of images comparing chemiluminescence (middle) and
PLIF measurements planes at x = 0 mm (top) and x = 17 mm (bot-
tom) in the model scramjet. Overall equivalence ratio (φ)= 0.83, M =
4.5, P0 =100 kPa, T0 = 2600. Images were acquired during the quasi-
stable state of the combustion process. Free stream ﬂow direction is
from left to right. The brighter (ﬂuorescing) contours are indicative of
higher OH concentrations. . . . . . . . . . . . . . . . . . . . . . . . . 28
3.7 A series of spatially varying OH distribution images obtained using
PLIF in the model scramjet. The bottommost image is at the scramjet
center plane and top image is closest to the side wall. All the images
were obtained at least 100 ms after fuel injection and can be considered
to be in stable mode of the quasi-steady combustion process. Free
stream ﬂow direction is from left to right. The brighter (ﬂuorescing)
contours are indicative of stronger OH concentrations. . . . . . . . . . 30
3.8 A depiction of ﬂame residence (in quasi-stable mode) on the bottom
wall of the scramjet model for conditions - φ)= 0.97, M = 4.5, P0 =100
kPa, T0 = 2600 . Free stream ﬂow direction is from left to right. . . . 31
3.9 Sequence of PLIF images with varying overall equivalence ratios ob-
tained at a planar section 2 mm from the center of the model. All the
images were obtained 100 ms after fuel injection and can be considered
to be in quasi-stable mode. Free stream ﬂow direction is from left to
right. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
d Diameter of a lens
f Focal length of a lens
ICCD Intensiﬁed charge-coupled device
O2 Mass ﬂow rate of O2
C2H4 Mass ﬂow rate of C2H4
φ Equivalence Ratio
FWHM Full width at half-maximum
χOH OH mole fraction
T0 Total temperature
P0 Total pressure
Spp Fluorescence signal per pixel
fB(T) Temperature-dependent Boltzmann fraction of the absorbing state
I am deeply grateful to my advisor, Dr. Hyungrok Do, to have given me this
opportunity to partake in an enriching research experience. I would like to thank
my committee members, Dr. Flint Thomas and Dr. Scott Morris for their eﬀorts in
going through the thesis and being part of the defense.
I would like to extend my thanks to my friends at Notre Dame without whom
this experience wouldn’t have been possible. Further, Qili Liu, Brian Neiswander and
Joanna have been extremely helpful with their advice and help. Stephen Hammack
and Constandinos have been very helpful in setting up the laser system and trans-
ferring operational skills to our team. The journey wouldn’t have been incredible
without the loving warmth of Eugene Heyse, Michael Sanders, Terry Jacobsen and
their amazing machining skills. I dedicate this work to my mother, father and sister
for supporting and motivating me with their annoying midnight phone calls from the
other side of the planet. This work would never have been possible for the love and
unrelenting support of my girlfriend, Jasmine.
With the development of laser based non-intrusive diagnostic techniques in re-
cent years, there has been renewed interest in combustion processes occurring in
Some groups have exploited the more economical methods like schlieren and
chemiluminescene  for the study of ﬂow inside scramjet models. However, sig-
nal quality of schlerien experiments is diminished in the absence of stark contrast in
the refractive index of the medium under observation. Furthermore, schlieren photg-
raphy is a line-of-sight technique. Rayleigh scattering technique oﬀers a much better
alternative. Its attractiveness lies in the fact that it does not require doping with
particles or tracers .
Although non-intrusive laser based diagnostics were thought to be the best way
to retrieve ﬂow parameters in hypersonic ﬂows as far back as 1990, the techniques
weren’t economical enough until recently with the advancements in the development
of reliable and compact UV lasers. Previous studies in the facility by Do  and Liu
 have focused on inlet unstart in a model scramjet engine phenomenon utilizing
chemi-luminescence (for visualization part). Rayleigh scattering has been extensively
applied by Do  for study of inlet unstart phenomenon in supersonic ﬂows. Work
presented in this thesis delves into the development and application of more advanced
ﬂow visualization techniques (PLRS and PLIF) in the facility.
Planar laser-induced ﬂuorescence of OH and CH radicals are commonly used for
experimental investigation of turbulent ﬂame structures. Burned gas has a high
concentration of OH radicals and OH PLIF signals that can be used to separate
burned gas from unburned constituents. One must be careful in the interpretation of
the results of OH radical distribution though. In the case of low Reynolds number
turbulent ﬂames, OH concentrations may correlate to the ﬂame fronts. But for high
Reynolds number cases, such as the ﬂow in a scramjet, this is less likely as the ﬂame
front is heavily distorted and folded.
Nevertheless, considering the economical incentive, higher concentration and con-
sequently, easier detection over CH PLIF, it was decided to qualitatively analyze the
ethylene ﬂame dynamics using OH PLIF.
When atoms or molecules spontaneously relax to a lower energy level, (typically)
due to vibrational and rotational energy transfer in the upper state and are accom-
panied by the emission of radiation, it is termed ﬂuorescence. Fluorescence does not
possess directionality. A spectrally resolved ﬂuorescence signal might contain more
than one wavelength even though excited at only one transition from a lower state
A simpliﬁed energy structure with two energy levels is illustrated in ﬁgure 1.1.
The amount of ﬂuorescence signal is aﬀected by various collisional and optical pro-
cesses. A prerequisite for spontaneous emission to occur is that the molecule must be
in an excited state. This can be achieved through absorption of photons, following
which, the molecule might relax into a lower energy state through spontaneous emis-
sion/ﬂuorescence. An alternate process is stimulated emission, wherein the excited
molecule is stimulated to emit a photon with the same energy, phase, polarization
and direction as the incoming one and settles into a lower state.
In addition, the molecule might leave the excited state without emitting ﬂuores-
cence in the event of collisions with surrounding molecules. This process is colli-
sional quenching and its rate is higher for species at room temperature and pressure.
Photo-ionization and predissociation also contribute to increase in loss of sponta-
neous emission. Photo-ionization occurs when a molecule is ionized by a photon with
a large enough energy. Predissociation occurs when a molecule relaxes to an unbound
(dissociative) state from a bound state causing dissociation.
Figure 1.1. Basic physical processes aﬀecting the ﬂuorescence rate depicted
in a two level system. Rate constants; b12 - stimulated absorption ; b21
-emission rate constants; A21 - spontaneous emission rate constant ; Q21
quenching rate constant; W2i photoionization rate constant; P -
predissociation rate constant. Adopted from 
18.104.22.168 Laser-induced Fluorescence
Fluorescence can be conveniently achieved using lasers with the added features
of spatially, temporally and spectrally selective excitation. Owing to its simplicity
of operation, LIF has become one of the most widely used diagnostic techniques for
combustion studies in recent years. Also, it is well suited for pulsed ﬂow facilities
when compared to probe based methods like hot wire anemometry. Additionally, the
ﬂuorescence is usually at a longer wavelength than the laser radiation. This helps to
easily ﬁlter away the stray background radiation at the shorter wavelengths.
Application of LIF is limited to atoms or molecules which have bound states
accessible with laser radiation. Knowledge of emission spectrum of the atom or
molecule and rate of radiative decay of its excited state is a pre-requisite to LIF
measurements. For quantitative studies, losses in the form of non-radiative processes,
such as collisional quenching and predissociation, should also be accounted for.
22.214.171.124 Linear Regime
For low laser intensities, the ﬂuorescence signal obtained is proportional to the
laser radiation. This is called the linear regime. Here, quenching rate (Q21) and spon-
taneous emission rate constants (A21) deﬁne ﬂuorescence. Therefore, the quenching
rate must be estimated prior to quantitative concentation measurements. The ﬂuo-
rescence signal is relatively weak in the linear regime compared to saturated regime.
126.96.36.199 Saturated Regime
It is the aim of any LIF experiment to achieve full spatial and temporal saturation.
At suﬃciently high laser energies, the ﬂuorescence signal becomes independent of
the laser intensity and of quenching. The energy transfers in the upper state are
dominated by the absorption and stimulated emission rates. This is described as the
saturated regime. Quenching can be disregarded in this regime. The ﬂuorescence
signal is maximized, leading to maximized detection levels. Usually, the intensity in
the wings of the laser sheet are always low given the gaussian nature of the pumped
laser beam. Hence, full spatial resolution is never achieved. An optical setup to
achieve full spatial resolution is described later in this text. Also, because the laser
energy varies during the duration of a pulse, temporal saturation is very diﬃcult to
achieve . The dependence of ﬂuorescence signal on laser energy is illustrated in the
schematic ﬁgure 1.4 with the linear and saturated regimes identiﬁed.
Figure 1.2. Depiction of LIF signal dependence on laser excitation energy.
Signal response is linear for low pulse energies. Signal response is highest
for saturated regime and doesn’t increase with increasing laser energy 
188.8.131.52 OH PLIF
The ﬂuorescence signal measured by the intensiﬁed CCD camera is proportional
to the OH mole fraction, found in the region of interest, and a temperature dependent
function. Experimental eﬃciencies for the current setup, such as the electronic gain of
the camera and transmission eﬃciency of the collection optics are assumed constant.
Figure 1.3. OH (left) and CH2O (right) LIF signals from a co-axial burner
from Li. Note the post ﬂame existence of OH radicals denoting the
region of burned gases. Also notable is the prevalence of OH signature over
that of CH2O, denoting unburned gases.
OH mole fraction depends on numerous factors including pressure, strain, local
equivalence ratio, exhaust gas recirculation and fuel. Considering its highly non-
specifc nature, caution must be exercised in the interpretation of results. Because
detection of OH is easier compared to other radicals it is usually chosen to characterize
the combustion activity.
Spp = const · χOH ·
One of the main disadvantages identiﬁed with PLIF is the quenching of ﬂuores-
cence at higher pressures due to increased collisions of molecules. The key to avoid
quenching is to achieve short predissociation lifetimes, provided the ﬂuorescence is
emitted only during predissociation lifetime. This is based on the fact that for suf-
ﬁciently short predissociation lifetimes, molecular collisions are eliminated. For the
combustion reactions in the scramjet model, which can be treated as a semi-enclosed
system, the temperature and pressure increases are strongly correlated. Therefore,
quenching ceases to be a problem during PLIF measurements.
A brief description of chemiluminescence is provided as an overview of the pre-
existing ﬂow visualization capabilities in the facility. Chemiluminescence is based on
the chemical excitation of species as opposed to excitation due to laser radiation. For
example, radiation emitted by chemically excited OH, denoted by OH∗
, is captured
by the camera. The instrumentation required to perform chemiluminescence is eco-
nomical and therefore continues to still see regular application . Band pass ﬁlters
may be used to observe deﬁning spectral line of chemical species. For example, the
maximum spectral line for OH∗
2 are known to occur at 308 nm, 431
nm nd 513 nm respectively. However, CO∗
2 has emission over a broad spectral range.
Chemiluminescence, being a line-of-sight visualization, complicates the interpretation
of acquired images. Also, LIF is known to provide much more detailed information
due to its greater spatial and temporal resolution. Chemiluminescence, however, has
still been employed in this study because OH∗
are known to unambiguously
characterize the overall equivalence ratio of laminar and turbulent ﬂames.
1.2.3 Rayleigh Scattering
Rayleigh scattering imaging captures light emitted from particles illuminated by
the laser sheet, causing it to be more reliable than line-of-sight and path integrated
optical methods like schlieren photography. Therefore, it is reasonable alternative for
qualitative characterization of the shock and turbulence structures.
Filtered Rayleigh scattering and condensate enhanced Rayleigh scattering are
the variants considered for application in this study. Because condensate enhanced
Rayleigh scattering signals have the potential to be much stronger than molecular
Rayleigh scattering, it was chosen to be employed in this study. Condensate enhanced
Rayleigh scattering utilizing condensed CO2 particles was used for imaging the ﬂow.
The size distribution of carbon dioxide clusters is shown to follow a very narrow trend
with a mean diameter of 6-10 nm . As long as the molecular clusters satisfy the
Rayleigh criterion (diameter less than 1/10 the wavelength of incident light), Rayleigh
scattering is viable. Condensation of residual water vapour and CO2 are known to
provide a favorable medium in high speed test facilities, satisfying this criteria. Also,
the small size of clusters allows for faithful and rapid response to ﬂow changes.
Rayleigh scattering is highly dependent on the thermal response of the particles
to the ﬂow. The CO2 particles are prone to sublimation in regions with increased
local ﬂow temperatures. Such regions are predominantly ones containing features like
shocks and boundary layers. Sublimation can lead to reduction or even elimination of
scattering signals. Boundary layers present a high temperature condition favorable for
sublimation and are accordingly marked by the mismatch in scattering signals from
the clusters present in cold core ﬂow and their absence in the boundary layer. Also,
the presence of shocks in the ﬂow causes clusters to sublimate due to strong changes
in ﬂow temperature and consequently, eliminate the scattering signals downstream.
Figure 1.4 shows a set of schlieren images compared to Rayleigh scattering results
obtained in the facility. Rayleigh scattering is very eﬀective in capturing the boundary
layers, but not as eﬀective with shock propagation. However, because the excitation
and scattering occur at the same wavelength, imaging in near wall regions might be
an issue at higher wavelengths (∼ 532 nm).
Supercooling rates aﬀect the equilibrium of condensation produced by nozzles.
Supercooling rates are lower in long, large, slow expanding nozzles compared to
their short, high expansion counterparts . A process closer to equilibrium can be
expected for lower levels of supercooling.
Figure 1.4. Schlieren images of HyShotII combustor: (top) instantaneous;
(middle) averaged over test time by Laurence  and (bottom) Rayleigh
scattering images from the present scramjet model
2.1 Hypersonic Wind Tunnel and Associated Instrumentation
Experiments described here were performed in the hypersonic wind tunnel facility
at University of Notre Dame. The wind tunnel is a pulsed-arc-heated facility. For
a more detailed description of the facility see . A schematic of the experimental
setup is shown in ﬁgure 2.1.
Figure 2.1. Schematic of the OH PLIF and PLRS measurements
The OH PLIF laser system constitutes a Nd:YAG laser (Spectra-Physics, Quanta
Ray PRO, 532 nm, 450 mJ/pulse) and a dye laser (Sirah Precision Scan). Rhodamine
dye in ethanol solvent was used in the frequency doubled dye laser, which is pumped
by the Nd:YAG laser and emits 283.22 nm light corresponding to the Q1(7) line within
the OH A2
Π(1 − 0) transition band. The dye laser energy was varied in
the range 13.5 - 20 mJ/pulse and it was deduced that the laser intensities used in
the present study were in the saturated regime, as no marked increase was detected
in the ﬂuorescence signal levels. Hence, it was decided to conduct experiments with
the dye laser power at 13.5 mJ/pulse.
Time-sequential chemiluminescence images were obtained using a high-speed movie
camera (Casio, Exilim Pro EX-F1) at 60 fps through a quartz optical access. A CCD
camera (LaVision Imager Intense) coupled with an intensiﬁer (LaVision IRO) was
mounted near the other optical access window to capture the PLIF and PLRS sig-
nals. Fluorescence signals were focused onto the intensiﬁer. The intensiﬁer gain was
set at 7 for all experiments. A f/2.8 UV lens (Sodern Cerco) ﬁtted with a band-
pass ﬁlter 306-320 nm (Asahi Spectra) was mounted on the intensiﬁer. The acquired
ﬂuorescence signals by the camera were digitized to 12 bits (equivalent to 4095 gray
levels). Windows on either side of the test section allowed optical access to the ﬂow
conditions. DaV is 7.2 software was used for recording images acquired by the camera
and for controlling image acquisition. Image processing was done using Matlab.
Free stream ﬂow of Mach 4.5 was generated with an axisymmetric converging
diverging nozzle 60 mm in diameter. The scramjet model has a ﬂow channel cross-
section of 15mm × 40 mm (height × width). Hence, the scramjet can be safely
assumed to be in the core ﬂow region of the nozzle. The maximum test time of the
facility was 1 s. Total pressure was ﬁxed at 100 kPa for the tests. Fuel concentration,
controlled with fuel jet injection pressure, in the scramjet was varied through a wide
range (φ = 0.2 − 5.5). Free stream conditions of total pressure and temperature for
all the tests were kept constant.
2.2 Optical Setup
A collimated laser sheet of measured minimum thickness of 1 mm was gener-
ated using a unique setup. The sheet generation optics included a cylindrical plano-
concave lens of focal length f = −30mm, and cylindrical plano-convex lenses of focal
length f = 100mm and f = 700mm. An arrangement of the optics is shown in
ﬁgure 2.2. This setup was mounted on a traversable bread board operated on by
a computerized stepper motor drive. The design was such that the laser beam of
height around 20 mm was generated downstream of the scramjet model and directed
upstream towards the nozzle.
Figure 2.2. Sheet generation optics
Figure 2.3. Sheet generation optics for OH PLIF and PLRS measurements
The optics are enclosed in a stainless steel enclosure with a slit wide enough to
allow the ejection of the laser sheet. The enclosure is shown in the ﬁgures 2.3 − 2.4.
The enclosure has a triangular protrusion for streamlining the ﬂow around it. Owing
to space constraints in the enclosure, a Gallilean conﬁguration was chosen for the
pair of f = 100mm plano-convex and f = −30mm plano-concave lenses. Care was
taken to acquire optics made from fused silica for the experiments. The enclosure also
features a slit to receive the laser beam from the dye laser. A circular UV fused silica
window was mounted behind the ejection slit of the enclosure to prevent any dust
accumulation on the optics within and also to prevent any burnt gases emanating
from the scramjet adversely aﬀecting the sheet generation optics. This window was
replaced from time to time. This optical setup was used for both PLRS and PLIF
measurements in the facility.
Figure 2.4. Schematic of conventional laser beam expansion optics
The intensity of the laser beam emanating from the dye laser assumes a circular
Gaussian distribution. When this circular laser beam is molded into a sheet, the
Gaussian distribution is preserved and maintained, causing the wings of the sheet to
not possess the intensities required to excite ﬂuorescence. Prior to the setup described
above, the laser sheet was generated using conventional means as shown in ﬁgure 2.5.
This setup required large divergence of the laser sheet, causing the sheet generation
optics to be spaced farther from the test subject, and therefore did not seem practical
for achieving full spatial saturation.
Figure 2.5. View of the stainless steel optical enclosure
With the current design, the sheet spanning across the scramjet is given access to
the intensities suﬃcient to excite OH radicals and elicit ﬂuorescence and consequently,
achieve full spatial saturation. The only downside to the laser sheet generated in this
fashion is the depletion of laser intensity due to interaction with ﬂow particles and
subsequent obstruction as the laser sheet traverses upstream. However, this could be
easily overlooked as the study was qualitative in nature.
Figure 2.6. View of the stainless steel optical enclosure: (left) without and
(right) with streamlined deﬂector hood
2.3 Timing Circuit
Laser ﬁring was synchronized with the ICCD camera exposure as illustrated in
ﬁgure 2.6. The laser Q-switches at a frequency of 10 Hz. A combination of relays
(built by Qili Liu as part of his dissertation) helped delay the trigger controlling the
tunnel injection valves to coincide with the Q-switching signal. The tunnel signal then
triggered a signal to the fuel jet injection valve, which could be altered temporally
as desired. For the current set of experiments, the fuel valve was opened at 100 ms
after the free stream ﬂow was triggered. The fuel injection signal was also relayed
to the Programmable Timing Unit (PTU), triggering the ICCD camera which was
gated to 100 ns. The jet injection was controlled by a solenoid valve triggered by the
fuel injection signal.
Figure 2.7. Timing diagram of the simultaneous operation of OH PLIF,
fuel injection valve and the ICCD camera
2.4 Wavelength Selection
Wavelength selection is key to PLIF experimentation. A wavelength pertaining to
Q1(7) (283.222 nm) transition of the OH spectra was selected for excitation because
it is strong and relatively temperature insensitive. Although some transitions might
appear to be much stronger in intensity during a peak ﬁnding scan, such as the peak
Q1(6) as seen in ﬁgure 2.7, the intensity is bound to vary relative to other peaks
due to pressure and temperature of the environment the OH radicals ﬂuoresce in.
Therefore, peak ﬁnding scans must be run only to tune the dye laser wavelength to
a desired value.
Figure 2.8. Sample wavelength scan in the range 282-284 nm using the
sirah dye laser by Jalbert 
2.4.1 Theoretical Spectral Database
Figure 2.8 shows the variation of emission spectra of OH excitation LIF at a
temperature of 2600 K in a thermalized system between 282.8 - 283.0 nm generated
in the software LIFBASE. Note the similar strengths of Q1(7) and Q1(6) transition
lines. For more information on how LIFBASE simulates LIF spectra see .
Figure 2.9. Emission spectra generated by LIFBASE. Transition lines of
interest are marked.
2.5 Transient Combustion System
A cross-sectional view of the scramjet is provided in the ﬁgure 2.9. The scramjet
model was made of stainless steel and had constant internal ﬂow channel dimensions
of 15 × 40 mm (height × width) stretching to a length of 600 mm. Sharp leading
edges are provided on the inlet lips of the model. The inner side of the upper inlet
lip had a 12 deg wedge to produce an incident shock into the scramjet for ﬂow
deceleration. The fuel jet was injected obliquely 100 mm downstream of the inlet
lip at a 60 deg inclination from the centerline of the bottom wall of the model. A
solenoid valve, attached to a fuel reservoir, controlled the fuel jet injection. The
stagnation temperature of the ﬂows was around 2600 K for the PLIF tests, which
is suﬃcient to auto ignite the partially premixed ﬂames downstream of the fuel jet.
The bottom wall had a wall cavity located 100 mm downstream of the fuel jet nozzle
with dimensions of 3 mm in depth and 12 mm in length. The ﬂame front position
in the model was assumed to be behaving in a quasi-stable mode of this transient
Figure 2.10. Schematic (not to scale) of the scramjet model used in the
experiments depicting the fuel injection port and cavity combustor.
2.6 Condensed Manual
One of the purposes of this thesis is to also provide tips in areas of laser operation
and tuning to minimize the time spent on producing the eﬃcient ﬂuorescence signal.
Tips on laser operation and maintenance were adopted from literature review and
conversations with graduate students from UIUC and LaVision experts. A good part
of the combustion community uses the ND:YAG pump laser with the Sirah dye laser.
The author sincerely hopes the information provided here will be beneﬁcial for any
beginner trying to use these systems.
Peak ﬁnding scans were run everyday prior to the experiments. As the tempera-
ture and humidity levels change during the day, multiple peak ﬁnding scans were ran
to keep track of the desired wavelength. Ideally, the peak ﬁnding scans need to be
performed using a Bunsen burner for credibility, as the fuel/oxidizer ﬂow rate is con-
stant. However, reliable results could also be obtained using commercial-oﬀ-the-shelf
handheld butane burners. A notorious problem associated with such burners is the
loss of fuel ﬂow rate/pressure within a few minutes of operation, rendering the results
of the scans inconclusive beyond the time frame of a few minutes. One such burner
was used for peak ﬁnding. Normally, the peak ﬁnding scans would take at least 15-20
minutes to cover the entire spectrum around the desired transition wavelength. One
way to overcome this problem would be to perform scans over small segments of the
spectrum and compare the peaks. It was found during our initial scans that the
peaks might have comparable strengths. As mentioned earlier, the pre-determined
peak must be adhered to. Once the peak ﬁnding scan has been performed, the energy
of the dye laser should be optimized near the desired peak.
DaV is 7.2 software and Sirah control software were used simultaneously for peak
ﬁnding scans. DaV is allows for real time integral counting of ﬂuorescence signals and
hence was used in plotting the peaks. Usually a course step size was chosen and once
vicinity of peaks were identiﬁed, step sizes were tuned down to a ﬁner scale.
The quartz windows on the scramjet model are prone to soot accumulation over
a few runs. This was found to cause considerable loss in the ﬂuorescence signals. A
frequently used and well established way to remove the soot is to clean the quartz
surface with a cloth dipped in dilute hydrocloric acid. Glass windows are also known
to attenuate ﬂuorescence signals. Installing glass windows either on the scramjet
model or wind tunnel test section should be avoided during LIF tests.
Overtime, there might be a decrease in the pump laser power output due to
buildup of condensation inside the ﬂashlamp assemblies, rendering their reﬂective
surfaces cloudy. This aﬀects their eﬃciency to reﬂect photons, and in turn, their
ability to generate a strong beam, ultimately aﬀecting the ﬂuorescence signal. Wiping
down each ﬂashlamp assembly will help increase the dye laser output .
Because self-luminosity of the reacting ﬂow in the scramjet can be overwhelming,
spatial ﬁltering methods and electronic shuttering of the intensiﬁed detector might
be necessary for success of PLIF.
Rayleigh scattering was performed under two cases of ﬂow conditions to ascertain
the extent of fast fuel jet mixing in the turbulent ﬂow structures. The ﬁrst case was
run without fuel jet to determine the ﬂow structures. The second case was repeated
with an operational fuel jet under the same free stream conditions. Nitrogen was used
to simulate the eﬀect of ethylene as both of the molecules possess similar molecular
weights. Longitudinal scans were run along the width of the scramjet cross-section
ranging from the central plane to the side walls in increments of 1 millimeter. Rayleigh
scattering at two wavelengths, 532 nm and 283 nm, were attempted. Typically,
532 nm is selected for Rayleigh scattering applications. However, higher signal-to-
noise ratios are achieved with UV light due to a larger Rayleigh cross-section of the
clusters, as well as, subdued reﬂectivity of metallic surfaces at UV wavelengths. It
was later realized that the lower pulse energies (∼ 20 mJ/pulse) at 283 nm were
not suﬃcient enough to elicit a strong Rayleigh scattering response. Hence, only the
results pertaining to experiments conducted at 532 nm light would be discussed in
Figure 3.1 is an illustration of the basic ﬂow features generated in the model
scramjet. The cavity combustor and fuel jet injection port have been depicted in the
images for reference.
Figure 3.1. Detailed ﬂow features of Rayleigh scattering images in the
scramjet central plane: (top) with and (bottom) without active fuel jet
operation. Free stream ﬂow is at Mach = 4.5 and from left to right.
The presence of a shock train was visualized near the inlet. The shock train
decelerates the ﬂow to supersonic speeds, and the cavity combustor aids in holding
the ﬂame. Additionally, a possible boundary layer transition due to shock/boundary
layer interaction was seen near the inlet lip due to the reﬂected shock. The successive
reﬂected shocks lost their strength, as perceived in the region prior to the fuel jet
injector. An expansion fan is created at the end of the cavity. Also, in the case of
an injected fuel jet, it can be seen that the turbulence characteristics downstream
cannot be ﬁnely deﬁned. This is due to the loss of ﬂuorescence signals from a strong
sublimation of CO2 clusters at the fuel jet, owing to the bow shock created by it.
Figure 3.2. Set of detailed Rayleigh scattering images arranged based on
their proximity to the central plane (x=0) of the model. Fuel jet injection
with N2 is enabled. Free stream ﬂow is at Mach = 4.5 and from left to right.
Figure 3.2 shows a collection of Rayleigh scattering images taken at diﬀerent pla-
nar locations in the scramjet model. Prominent amongst the features is the tripping
of boundary layers (marked in red) by the fuel jet, even at locations as far as the
side wall of the model. The supersonic fuel jet injection was believed to cause an
increase in downstream pressure and temperature, which could have the potential to
trigger unstart. However, since the conditions selected for the experiments are not
conducive for unstart, no unstart was observed. As expected, the boundary layer at
the side wall is much thicker than at the central plane. Note the stark diﬀerence in
the ﬂow structure between images at x=0 mm and x=18,19 m, as a result of the side
wall on the boundary layer development. The turbulent boundary layers developing
on the top and bottom walls seemed to merge beyond the cavity.
Figure 3.3. Set of detailed ﬂow features of Rayleigh scattering images in
the scramjet central plane: (top) with and (bottom) without active fuel jet
operation (N2). Free stream ﬂow is at Mach = 4.5 and from left to right.
A series of Rayleigh scattering images captured in diﬀerent planes of the scramjet
model is shown in Figure 3.3. Each set has an image taken with and without the
operation of the jet injection. Previously discussed rapid growth of the top wall
boundary layer triggered by the fuel jet injection, can be conﬁrmed from the images.
A gradual decrease in the presence of the reﬂected shock near the inlet lip can be
noticed as one translates towards the side walls of the scramjet model. For ﬂow
proﬁles near the wall, x = 18 mm and 19 mm, no shock train can be perceived in the
images. This indicates the strong eﬀect of the side walls. Also, similar ﬂow features
in each set of images can be observed until the fuel injection port. Fuel jet injection
does not seem to have any eﬀect on the ﬂow features upstream of the injection port.
Images were acquired over a shutter period of 1/60 s. However, as no ﬁlters
were used, the chemiluminescence can be assumed due to the combined presence of
chemically excited species, mainly consisting of OH∗
2 and CO∗
the region with relatively higher intensities of chemiluminescence can be associated
with the strongest combustion reactions.
Shown in the ﬁgure 3.4 is a detailed, typical chemiluminescence image acquired
in the facility. All of the chemiluminescence images were focused onto the central
plane of the scramjet.
Figure 3.4. Detailed chemiluminescence image of the combustion process.
Brighter (blue) regions are indicative of intense heat release reactions. Free
stream ﬂow direction is from left to right.
3.3 OH PLIF
Prior to application of the discussed optical measurements in a reacting environ-
ment, it is quintessential to verify the quasi-stable assumption for the chosen test
conditions. Figure 3.5 substantiates the assumption that combustion reactions were
in fact in a quasi-stable mode in the model. The fuel jet was active for a period of 300
ms and the images seem to show similar characteristics, most conspicuous of which
is the strong OH signature around the cavity region; with the brighter (ﬂuorescing)
contours indicative of stronger OH concentrations. OH distribution is an indicator
of intermediate reactions characteristic of ignition. It is well known that ignition
reactions are a precursor to the heat release reactions, and therefore, presence of
OH ﬂuorescence may be suggestive of negligible heat release in respective regions of
Figure 3.5. Sequence of PLIF images taken ∆T =100 ms apart from each
other (numbered), at one of the scramjet investigation planes. Free stream
ﬂow direction is from left to right. Fuel jet was active for 300 ms.
Figure 3.6 shows the diﬀerences in images acquired using simultaneous chemilu-
minescence and OH PLIF for runs under similar test conditions. The PLIF images
were taken at separate planes during diﬀerent runs. The chemiluminescence and
topmost PLIF images complement each other. A striking feature is the absence of
signals in the central region of the longitudinal plane in both the images. This pecu-
liarity, apart from suggestive of a truly stable behavior of the ﬂame, also indicates the
ﬂame containment to regions sporting high mixing environments and consequently,
favoring ﬂame residence. However, a contrasting perspective is obtained when the
chemiluminescence is compared to a PLIF measurements closer to the wall at x =17
mm (bottom image). A possible explanation could be that chemiluminescence, being
a line-of-sight technique, absorbs the signals in the planes encompassed in its depth
of ﬁeld. Here, the planes in discussion are ones closest to the scramjet central plane.
If the depth of ﬁeld were to also include the planes closer to the walls, then it can
be concluded that the reactions are stronger in the planes closer to the central plane,
and hence have more bearing on the chemiluminescence image.
Figure 3.6. Detailed set of images comparing chemiluminescence (middle)
and PLIF measurements planes at x = 0 mm (top) and x = 17 mm
(bottom) in the model scramjet. Overall equivalence ratio (φ)= 0.83, M =
4.5, P0 =100 kPa, T0 = 2600. Images were acquired during the quasi-stable
state of the combustion process. Free stream ﬂow direction is from left to
right. The brighter (ﬂuorescing) contours are indicative of higher OH
These can be seen predominantly in downstream regions of the cavity, which, as
expected, can be attributed to enhanced mixing.
PLIF measurements were acquired for ﬂows at overall equivalence ratios varying
from lean to rich. PLIF images were helpful in ascertaining the interaction of tur-
bulence structures with ﬂame fronts. Since the ﬂame fronts have a 3D structure,
visualization of planar sections of the scramjet during fuel operation helped reveal
interactions unperceived in a single plane. Figure 3.4 shows a set of PLIF images
taken in various cross-sectional planes of the scramjet model.
Flame front fading towards the cavity was observed as one travels away from
the central plane of the model towards the side wall. However, there seems to be
combustion activity registered near the wall, (16-19 mm) downstream of the inlet
lip. This phenomenon could be attributed to development of dense boundary layers
at the side walls. A simple 2D scan at only one plane would not have provided
suﬃcient information to reach this conclusion. The OH distribution was seen to be
more concentrated in regions surrounding the lower boundary layers.
Figure 3.7. A series of spatially varying OH distribution images obtained
using PLIF in the model scramjet. The bottommost image is at the
scramjet center plane and top image is closest to the side wall. All the
images were obtained at least 100 ms after fuel injection and can be
considered to be in stable mode of the quasi-steady combustion process.
Free stream ﬂow direction is from left to right. The brighter (ﬂuorescing)
contours are indicative of stronger OH concentrations.
Figure 3.8. A depiction of ﬂame residence (in quasi-stable mode) on the
bottom wall of the scramjet model for conditions - φ)= 0.97, M = 4.5, P0
=100 kPa, T0 = 2600 . Free stream ﬂow direction is from left to right.
A possible explanation of the ﬂame residence in regions upstream of the fuel jet
could be due to a separation region induced by either mass loading or combustion
downstream of the fuel jet. In cases where the overall equivalence ratio is close to
stoichiometric ratio, as in the case just discussed, mass loading may be ruled out. As
the ﬂow near the walls and upstream of the fuel jet is already separated, the presence
of ﬂame might be ascribed to the pressure buildup due to combustion occurring
downstream and subsequently increased ﬂame propagation speeds.
In the image sequence shown in ﬁgure 3.8, high concentrations of OH can be
seen near and downstream of the cavity for leaner and stoichiometric fuel mixtures
(based on overall equivalence ratios). This is indicative of the role the cavity plays
in anchoring and stabilizing the ﬂame for these mixture fraction regimes. Detectable
OH distributions are noticeable in the shear layers of the fuel jet, although these only
seem to become prominent during auto-ignition in fuel rich scenarios. In these cases,
the combustion activity is shifted to upstream locations of the cavity, indicative of
its subdued role in assisting combustion.
Figure 3.9. Sequence of PLIF images with varying overall equivalence
ratios obtained at a planar section 2 mm from the center of the model. All
the images were obtained 100 ms after fuel injection and can be considered
to be in quasi-stable mode. Free stream ﬂow direction is from left to right.
The total temperature was such that the ﬂame was auto-ignited in the windward
region of the fuel jet, and stretched downstream. A bow-shock was induced by the fuel
jet and the fuel is auto-ignited in the jet wake region. Strongest OH concentrations
were detected in the lower boundary layers in the periphery of the jet, supporting the
auto-ignition hypothesis. Another notable characteristic of the ﬂames in the fuel rich
regime was the OH distribution, ﬂanking, and what seems to be, a highly fuel rich
mixture convecting downstream of the fuel jet and residing in the central portion of
the longitudinal plane.
CONCLUSIONS AND RECOMMENDATIONS
Planar Laser Rayleigh Scattering and OH Planar Laser Induced Fluorescence
techniques have been successfully developed and demonstrated to investigate ﬂow
physics in a model scramjet engine in a free stream ﬂow of Mach 4.5. These techniques
were extended to 3-dimensional ﬂow domain for qualitative measurements with the
development of an unique optical system. Full spatial ﬂuorescence (in the case of
PLIF) was achieved with the application of this optical arrangement. Particular
examples were discussed detailing the ﬂow features.
Rayleigh scattering images were acquired using the second harmonic (532 nm)
of 1064 nm and at 283 nm (dye laser output). The incentive to utilize light at 283
nm was due to the fact that shorter wavelengths help achieve better signal-to-noise
ratio as a result of a larger Rayleigh cross-section, as well as, subdued reﬂectivity
of metallic surfaces; UV light is very strongly scattered and consequently, Rayleigh
light possesses highly exploitable contrast compared to surface scattered light. How-
ever, the laser intensity was not suﬃcient enough (∼ 13.5 mJ/pulse) to invoke strong
Rayleigh signals. Further work in this domain is possible. The results of the experi-
ments utilizing 532 nm light have been discussed. Rayleigh scattering has shown the
capability of highlighting detailed ﬂow structures such as shock and expansion waves,
as well as, boundary layers in cold, non-reacting ﬂows.
Chemiluminescence has been the preferred means of ﬂow visualization in the fa-
cility. Diﬀerences between the chemiluminescence and PLIF have been delved into to
strengthen the case for PLIF. Chemiluminescence did help strengthen some deduc-
tions obtained through PLIF measurements as a result its long exposure time scales.
Compounded PLIF measurements from multiple planes and over various mixture
fractions have been examined to further establish the signiﬁcance of this technique.
Results point to the eﬀectiveness of cavities in the deceleration of the ﬂow injection
and subsequent quasi-stable combustion in stoichiometric and fuel lean regimes based
on overall equivalence ratio. The examined results also attested to the fact that the
combustion reactions were in a quasi-stable mode, and that auto-ignition hypothesis
was valid for fuel rich scenarios. Fuel jet injection might be a deﬁning operation in
generation of turbulence features downstream of the injection port. Hence, PLIF
and PLRS measurements can be combined to provide a holistic means for deduc-
ing combustion physics, discussed using examples. The boundary layer eﬀect has
been shown to dominate upstream regions of the fuel jet (close to the walls), even in
A small section composed of tips helpful for a trouble-free experimentation were
provided for beginners using the system. The measurement techniques discussed have
helped underline the reliability of instantaneous and very high spatially resolvable op-
tical methods for a complicated combustion system like a scramjet. These techniques
have to be extended to multiple planes for appropriate interpretation of results.
4.2 Recommendations for Future Work
Since the production of CH radicals occurs at the ﬂame front, they can be re-
liably associated with the reaction zones , ,. Also, CH radical distribution
is narrower, and the lifetime much shorter than that of OH. Further, the heat re-
lease rate correlates betterl with the CH radical distribution than that of OH .
Nevertheless, CH PLIF doesn’t suﬃce to diﬀerentiate between unburned and burned
gaseous zones. Therefore, simultaneous OH and CH PLIF, or OH and CH2O PLIF
measurements could be applied to future experiments in the facility.
Bandpass ﬁlters corresponding to the maximum spectral lines of various chemical
2) could be used with the Casio camera to selectively and
cost eﬀectively observe chemiluminescence in future experiments.
The best use of the current laser sheet generation scheme can only be exploited
if the camera is capable of imaging the entire length of the scramjet model. Un-
fortunately, due to the presence of struts in the optical access window this is not
possible with a single camera. A second imaging camera could be required to image
ﬂuorescence in the part of the scramjet model obstructed by the strut.
Additionally, a pulse delay generator integrated into the timing circuit could po-
tentially help access time sequential Rayleigh signals and provide temporally resolved
(resolution as high as 0.5 ms) information on the development of turbulent and shock
For the same laser pulse intensities, Rayleigh scattering measurements at shorter
wavelengths help achieve better signal-to-noise ratio as a result of a larger Rayleigh
cross-section, as well as, subdued reﬂectivity of metallic surfaces. Therefore, the
third harmonic of 1064 nm light would be highly suited for application in the facil-
ity. Coupled with a multi-kHz, high output (∼ 120 - 200 mJ) pulsed laser system,
experiments with high temporal resolution could be attained. For convenience, these
experiments could be conducted post OH PLIF measurements by removing the dye
laser from the beam path.
 MG Allen, Robert D Howe, and Ronald K Hanson. Digital imaging of reaction
zones in hydrocarbon–air ﬂames using planar laser-induced ﬂuorescence of CH
and C2.Opticsletters, 11(3) : 126 − −128, 1986.
 CD Carter, JM Donbar, and JF Driscoll. Simultaneous CH planar laser-induced
ﬂuorescence and particle imaging velocimetry in turbulent nonpremixed ﬂames.
Applied Physics B: Lasers and Optics, 66(1):129–132, 1998.
 RJ Cattolica, RL Schmitt, and RE Palmer. Feasibility of non-intrusive optical
diagnostic measurements in hypersonic boundary layers for ﬂight experiments.
In AIAA, Aerospace Sciences Meeting, volume 1, 1990.
 Hyungrok Do, Seong-kyun Im, M Godfrey Mungal, and Mark A Cappelli. The
inﬂuence of boundary layers on supersonic inlet ﬂow unstart induced by mass
injection. Experiments in ﬂuids, 51(3):679–691, 2011.
 Hyungrok Do, Seong-kyun Im, M Godfrey Mungal, and Mark A Cappelli. Vi-
sualizing supersonic inlet duct unstart using planar laser Rayleigh scattering.
Experiments in ﬂuids, 50(6):1651–1657, 2011.
 Hyungrok Do, Andrea Passaro, and Damiano Baccarella. Inlet Unstart of an
Ethylene-Fueled Model Scramjet with a Mach 4.5 Freestream Flow. AIAA Paper
No. 2012-5929, 2012.
 Hyungrok Do, Andrea Passaro, Tonghun Lee, and Damiano Baccarella. Ethylene
Flame Dynamics in an Arc-Heated Hypersonic Wind Tunnel. AIAA Paper No.
 Alan C Eckbreth. Laser diagnostics for combustion temperature and species,
volume 3. CRC Press, 1996.
 PJ Erbland. Filtered Rayleigh scattering and homogeneous nucleation of CO2 in
supersonic ﬂows. PhD thesis, Princeton University, 2000.
 Adrienne Murphy Jalbert. A study of quantitative concentrations of hydroxyl
(OH) in laminar ﬂat ﬂames using planar laser induced ﬂuorescence (PLIF). 2011.
 Martin Lauer and Thomas Sattelmayer. On the adequacy of chemiluminescence
as a measure for heat release in turbulent ﬂames with mixture gradients. Journal
of Engineering for Gas Turbines and Power, 132(6):061502, 2010.
 SJ Laurence, D Lieber, J Martinez Schramm, K Hannemann, and J Larsson.
Incipient thermal choking and stable shock-train formation in the heat-release
region of a scramjet combustor. Part I: Shock-tunnel experiments. Combustion
and Flame, 2014.
 ZS Li, Bo Li, ZW Sun, Xue-Song Bai, and Marcus Ald´en. 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.
 Qili Liu, Andrea Passaro, Damiano Baccarella, and Hyungrok Do. Ethylene
Flame Dynamics and Inlet Unstart in a Model Scramjet. Journal of Propulsion
and Power, pages 1–9, 2014.
 J. Luque and D.R. Crosley. Lifbase: Database and spectral simulation program
(version 2.1.1). SRI International Report MP 99-009, 1999.
 Elin Malmqvist. Thermometry using OH laser-induced ﬂuorescence excitation
spectra: A feasibility study. 2013.
 Mohy S Mansour, Norbert Peters, and Yung-Cheng Chen. Investigation of scalar
mixing in the thin reaction zones regime using a simultaneous CH-LIF/Rayleigh
laser technique. In Symposium (International) on Combustion, volume 27, pages
767–773. Elsevier, 1998.
 Daniel J Micka and James F Driscoll. Reaction zone imaging in a dual-mode
scramjet combustor using CH-PLIF. AIAA, 5071:2008, 2008.
 Phillip H Paul and Habib N Najm. Planar laser-induced ﬂuorescence imag-
ing of ﬂame heat release rate. In Symposium (International) on Combustion,
volume 27, pages 43–50. Elsevier, 1998.
This document was prepared & typeset with pdfLATEX, and formatted with
nddiss2ε classﬁle (v3.2013[2013/04/16]) provided by Sameer Vijay and updated
by Megan Patnott.