258
ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar Flames by Optical Methods Andrew M. Schaffer 2001 Optical techniques are used to characterize steady and time-varying laminar flames in order to verify computational models and non-optical measurements. The first set of measurements determines major species concentration, temperature, and flame front profiles in a steady and flow-modulated laminar methane diffusion flame through Rayleigh and Spontaneous Raman scattering techniques. These experimental results are compared to the computational profiles of the group of Professor Mitchell Smooke. The next set of measurements determines temperature and soot particle size and volume fraction in a sooting ethylene, laminar diffusion flame through laser-induced incandescence techniques. A model is developed to extract particle size information from the incandescence signal. Soot particle size is compared with particle sizes obtained from soot sampling measurements. The final set of measurements determines particle and aggregate information of nanoparticles synthesized in a premixed laminar flame through laser-induced incandescence and laser light scattering techniques.

ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

  • Upload
    others

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

Page 1: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

ABSTRACT

Quantitative Characterization of Species, Temperature, and Particles in

Steady and Time-Varying Laminar Flames by Optical Methods

Andrew M. Schaffer

2001

Optical techniques are used to characterize steady and time-varying laminar flames in

order to verify computational models and non-optical measurements. The first set of

measurements determines major species concentration, temperature, and flame front

profiles in a steady and flow-modulated laminar methane diffusion flame through

Rayleigh and Spontaneous Raman scattering techniques. These experimental results are

compared to the computational profiles of the group of Professor Mitchell Smooke. The

next set of measurements determines temperature and soot particle size and volume

fraction in a sooting ethylene, laminar diffusion flame through laser-induced

incandescence techniques. A model is developed to extract particle size information from

the incandescence signal. Soot particle size is compared with particle sizes obtained from

soot sampling measurements. The final set of measurements determines particle and

aggregate information of nanoparticles synthesized in a premixed laminar flame through

laser-induced incandescence and laser light scattering techniques.

Page 2: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

Quantitative Characterization ofSpecies, Temperature, and Particles in

Steady and Time-Varying Laminar Flames by Optical Methods

A DissertationPresented to the Faculty of the Graduate School

ofYale University

in Candidacy for the Degree ofDoctor of Philosophy

by

Andrew M. Schaffer

Dissertation Director: Professor Marshall B. Long

December 2001

Page 3: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

2001 by Andrew M. Schaffer

All rights reserved.

Page 4: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

ACKNOWLEDGEMENTS

I would like to thank my advisor, Professor Marshall Long, for his guidance, insight, and

friendship throughout my dissertation research. I would also like to thank Professor

Daniel Rosner for inspiration and good ideas throughout our collaborative efforts with his

group. Thanks to Barbara LaMantia and Charles McEnally for their friendship during our

collaborative efforts.

I would like to thank Professor Mitchell Smooke and his group, who performed much of

the computational work compared to the experimental work in this thesis. Thanks to

Mikhail Noskov and Beth Bennett for the computational counterpart to a major part my

work, and for the constant insightful discussion about numerical computations.

I would like to express my appreciation to the other members of my thesis committee,

Professor Richard Chang and Professor Kurt Gibble, who have both had a positive impact

on my career as a graduate student, and Professor Kevin Lyons, who has graciously

agreed to be the outside reader of my thesis.

This work is dedicated to my wife, Julie, and our son, Evan, who have given me

overwhelming support and joy, and have shown an infinite amount of patience.

Page 5: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

ii

TABLE OF CONTENTS

LIST OF FIGURES................................................................................... vi

LIST OF TABLES ..................................................................................... x

TABLE OF NOMENCLATURE.............................................................. xi

CHAPTER 1 INTRODUCTION ............................................................. 1

CHAPTER 2 LIGHT SCATTERING TECHNIQUES ......................... 62.1 Introduction.......................................................................................62.2 Rayleigh and Raman scattering..........................................................72.3 Laser-induced incandescence...........................................................162.4 Chemiluminescence.........................................................................202.5 Laser light scattering .......................................................................212.6 Laser absorption ..............................................................................23

CHAPTER 3 CHARACTERIZATION OF A STEADY AND TIME-VARYING, AXISYMMETRIC, LAMINARDIFFUSION FLAME...................................................... 26

3.1 Introduction.....................................................................................263.2 Flame and Burner Characterization..................................................283.3. Boundary Conditions......................................................................30

3.3.1 Steady Flame.........................................................................303.3.2 Particle Image Velocimetry ...................................................313.3.3 Time-varying Flame ..............................................................36

3.4 Computational Modeling.................................................................413.4.1 Unforced ...............................................................................413.4.2 Forced ...................................................................................42

3.5 Measurement of CH* via Chemiluminescence.................................423.5.1 Introduction...........................................................................423.5.2 Experimental Setup and Acquisition ......................................433.5.3 Image processing...................................................................453.5.4 CH* profiles for the steady and time-varying flame...............47

Page 6: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

iii

3.6 Fuel Concentration, Temperature, andMixture Fraction Measurement........................................................483.6.1 Theory and introduction.........................................................483.6.2 Experimental Setup ...............................................................553.6.3 Acquisition ............................................................................57

3.6.4 Data Processing .....................................................................583.6.5 Calculation of fuel concentration, temperature, and mixture

fraction..................................................................................613.7 Multi-species Measurement using Difference Raman

and Rayleigh Scattering...................................................................623.7.1 Difference Scattering.............................................................633.7.2 Multi-species technique in calculation of temperature and

species number density..........................................................673.7.3 Setup .....................................................................................683.7.4 Unforced Case Acquisition ....................................................723.7.5 Forced Case Acquisition........................................................74

3.7.6 Processing .............................................................................743.7.7 Determination of the temperature dependence of the

bandwidth factor τm(T) ..........................................................82

3.7.8 Temperature and Species Concentration Calculation.............863.8 Discussion on experimental techniques............................................88

3.8.1 Effectiveness of two scalar technique ....................................88

3.8.2 Effectiveness of difference Raman technique.........................923.9 Comparison of experimental and computational profiles .................94

3.9.1 Steady Flame.........................................................................943.9.2 Forced Flame.......................................................................104

3.10 Summary.....................................................................................124

CHAPTER 4 SOOT AND TEMPERATURE CHARACTERIZATIONA SOOTING, LAMINAR, ETHYLENEDIFFUSION FLAME................................................... 126

4.1 Introduction...................................................................................1264.2 Flame and Burner Characterization................................................1274.3 Computational Model....................................................................1284.4 Probe measurement of temperature and soot volume fraction ........128

Page 7: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

iv

4.5 Experimental determination of temperature using thetwo scalar technique ......................................................................1294.5.1 Optical imaging setup for temperature measurement............1294.5.2 Processing ...........................................................................1314.5.3 Calculation of Temperature .................................................132

4.5.4 Two scalar temperature comparison with probemeasurements and computations .........................................132

4.6 Determination of the soot volume fraction profile usinglaser-induced incandescence..........................................................1344.6.1 Introduction and theory........................................................1344.6.2 LII imaging setup ................................................................1364.6.3 Data acquisition...................................................................1384.6.4 Processing ..........................................................................1424.6.5 Error estimates of the LII technique in determining

soot volume fraction............................................................1434.7 LII soot volume fraction comparison to probe measurements

and computations ..........................................................................1484.8 Primary soot particle size using time-resolved LII .........................148

4.8.1 Introduction.........................................................................1484.8.2 Time-resolved LII setup.......................................................1534.8.3 Time-resolved LII data acquisition ......................................1564.8.4 Qualitative analysis of the time-resolved LII signals ............1564.8.5 Calculation of particle size distribution from LII data ..........1614.8.6 Grid sampling of soot particles ............................................1654.8.7 Comparison of LII-derived and grid sampling particle

size distributions .................................................................1674.8.8 Sensitivity analysis of LII-derived particle sizing technique.167

4.9 Time-resolved laser light scattering and laser absorption ...............1714.9.1 Introduction.........................................................................1714.9.2 Setup for LLS and laser absorption experiment....................1724.9.3 Acquisition of LLS and laser absorption ..............................1734.9.4 LLS/Absorption results........................................................175

4.10 Error estimates of the absorption/scattering technique .................1794.11 Summary.....................................................................................179

Page 8: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

v

CHAPTER 5 CHARACTERIZATION OF NANOPARTICLESTRUCTURES SYNTHESIZED IN A PREMIXED,METHANE/AIR FLAT FLAME.................................. 183

5.1 Introduction...................................................................................1835.2 Burner and Flame..........................................................................1845.3. Measurement of the LII spectrum .................................................187

5.3.1 LII spectrum setup and acquisition ......................................1875.3.2 LII Spectrum processing......................................................1905.3.4 LII spectrum results .............................................................192

5.4 Sampling of iron oxide particles ....................................................198

5.5 X-ray diffraction of the particle material........................................2005.6 Time-resolved LII and laser light scattering experiment ................202

5.6.1 Experimental setup and acquisition......................................2025.6.2 Qualitative analysis of time-resolved LII data ......................2065.6.3 Particle sizing model, parameters, and procedure.................2095.6.4 Particle distribution results from LII data analysis ...............2135.6.5 Results of LLS.....................................................................220

5.7 Conclusion ....................................................................................223

CHAPTER 6 SUMMARY AND CONCLUSIONS............................. 226

REFERENCES....................................................................................... 231

Page 9: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

vi

LIST OF FIGURES

Figure 3.1 Forced flame burner. ................................................................................. 29Figure 3.2 Computational and PIV velocity profiles 1 mm above the burner. ............. 35Figure 3.3 Computational and PIV fuel tube centerline velocity for 30% modulation

of flow in the fuel tube as a function of forcing phase.. ............................. 39Figure 3.4 Fuel tube centerline velocity and speaker forcing signal for 30% modulation

in the fuel flow as a function of forcing phase. .......................................... 40Figure 3.5 Experimental setup for CH* chemiluminescence....................................... 44Figure 3.6 CH* profiles for 30% and 50% flow modulation....................................... 46

Figure 3.7 Burke-Schumann flame configuration ....................................................... 51Figure 3.8 Experimental setup for two scalar imaging................................................ 56Figure 3.9 Methane Raman intensity profile (arbitrary scale) from the two scalar

experiment ................................................................................................ 59Figure 3.10 Multi-species experimental setup .............................................................. 69Figure 3.11 Sample images from multi-species/ difference Raman experiment ............ 76Figure 3.12 Intensity spectrum (arbitrary units) taken from the region of maximal C2

fluorescence interference of the images in Figure 3.11 (marked with avertical white rectangle in Fig. 3.11) ......................................................... 78

Figure 3.13 Intensity profiles (arbitrary units) of Ids for oxygen Ramanand Iyz for C2 fluorescence. ........................................................................ 80

Figure 3.14 Experimental and simulated Raman spectra for nitrogen atT = 300 K and T = 2000 K ........................................................................ 84

Figure 3.15 Difference Raman signal temperature dependence .................................... 85Figure 3.16 Methane Raman intensity profiles (arbitrary units) from experiments........ 90Figure 3.17 Two-scalar calculation of mixture fraction and temperature based on

computational data, compared to ξCHO and computational temperature based

on computations........................................................................................ 91

Figure 3.18 Measured (multi-species technique) and computedtemperature (degrees Kelvin) for the steady flame..................................... 95

Figure 3.19 Measured (multi-species technique) and computedcarbon dioxide mole fractions for the steady flame.................................... 96

Figure 3.20 Measured (multi-species technique) and computed water mole fractionsfor the steady flame................................................................................... 97

Figure 3.21 Measured (multi-species technique) and computed carbon monoxidemole fractions for the steady flame............................................................ 98

Page 10: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

vii

Figure 3.22 Measured (multi-species technique) and computed hydrogenmole fractions for the steady flame............................................................ 99

Figure 3.23 Measured (multi-species technique) and computedmethane mole fractions for the steady flame............................................ 100

Figure 3.24 Measured (multi-species technique) and computed

nitrogen mole fractions for the steady flame............................................ 101Figure 3.25 Measured (multi-species technique) and computed

oxygen mole fractions for the steady flame. ............................................ 102Figure 3.26 Radial and centerline plots of temperature, water, and carbon dioxide for

steady flame experiments (multi-species technique) and computations.... 103Figure 3.27 Temperature profiles of the multi-species and two scalar techniques

for 30% flow modulation ........................................................................ 105Figure 3.28 Temperature lineplots of the two-scalar and multispecies

technique for 30% flow modulation. ....................................................... 106Figure 3.29 Mixture fraction profiles of the multi-species and two scalar techniques

for 30% flow modulation ........................................................................ 109

Figure 3.30 Two scalar and multi-species mixture fraction plots for30% flow modulation.............................................................................. 110

Figure 3.31 Temperature profiles (degrees K) of experiments (multi-species) andcomputations for 30% flow modulation................................................... 112

Figure 3.32 Temperature lineplots (degrees K) of experiments (multi-species) andcomputation for 30% flow modulation. ................................................... 113

Figure 3.33 Carbon dioxide mole fraction profiles of experiments (multi-species) andcomputations for 30% flow modulation................................................... 114

Figure 3.34 Carbon dioxide mole fraction lineplots of experiments (multi-species) andcomputations for 30% flow modulation................................................... 115

Figure 3.35 Water mole fraction profiles of experiments (multi-species) and

computations for 30% flow modulation................................................... 116Figure 3.36 Water mole fraction lineplots of experiments (multi-species) and

computations for 30% flow modulation................................................... 117Figure 3.37 Temperature profiles (degrees K) of experiments (multi-species)

and computations for 50% flow modulation.. .......................................... 118Figure 3.38 Temperature lineplots (degrees K) of experiments (multi-species)

and computations for 50% flow modulation.. .......................................... 119Figure 3.39 Carbon dioxide mole fraction profiles of experiments (multi-species) and

computations for 50% flow modulation................................................... 120Figure 3.40 Carbon dioxide mole fraction lineplots of experiments (multi-species) and

computations for 50% flow modulation................................................... 121

Page 11: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

viii

Figure 3.41 Water mole fraction profiles of experiments (multi-species) andcomputations for 50% flow modulation................................................... 122

Figure 3.42 Water mole fraction lineplots of experiments (multi-species) andcomputations for 50% flow modulation................................................... 123

Figure 4.1 Temperature profiles from experiments and computations-32% ethylene flame................................................................................133

Figure 4.2 LII Imaging Setup..................................................................................137Figure 4.3 Variation of time-integrated LII signal with laser fluence

from LII imaging. ..................................................................................140Figure 4.4 Calculated LII response to variations in laser fluence across

the height of the laser sheet.. ..................................................................141Figure 4.5 Shot-to-Shot LII Fluctuation.. ................................................................144Figure 4.6 Interference on ethylene Raman data......................................................146Figure 4.7 Experimental and Computational Soot Volume Fraction-

32% Ethylene Flame. .............................................................................147Figure 4.8 Select properties of carbon and nitrogen.................................................150Figure 4.9 Time-resolved LII setup. ........................................................................155Figure 4.10 Time-resolved LII curves at various laser fluences. ................................157Figure 4.11 Time-integrated LII signals vs. laser fluence.. ........................................158Figure 4.12 Time-resolved LII signals for various laser fluences...............................160Figure 4.13 Curve fit to the soot LII data ..................................................................166Figure 4.14 Primary soot particle size distributions from grid sampling

measurements and from the LII-derived particle size distribution. ..........168Figure 4.15 Effect of change in parameter value on the predicted particle size ..........170Figure 4.16 Time-resolved Scattering/Absorption Setup ...........................................174

Figure 4.17 Time-resolved change in elastic scattering and absorption of thesooty region (measured with the Ar+ laser) due to the YAG laser pulse atvarious YAG laser fluences....................................................................176

Figure 4.18 Time-resolved change in elastic scattering of the sooty region (measuredwith Ar+ laser) due to YAG laser pulse with laser fluence = 0.15 J/cm2. 177

Figure 5.1 Burner for iron oxide particle production ................................................ 185Figure 5.2 Experimental LII spectrum setup............................................................. 188Figure 5.3 Flame emission spectrum. ....................................................................... 191

Page 12: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

ix

Figure 5.4 Raw LII spectrum and the LII spectrum corrected for optical throughputand detector efficiencies.......................................................................... 193

Figure 5.5 LII spectrum for several laser fluences. ................................................... 194Figure 5.6 Delayed and prompt detection of LII spectrum........................................ 196Figure 5.7 LII spectrum for several heights above the burner ................................... 197

Figure 5.8 TEM images of thermophoretically sampled particles forflame #1 and flame #2............................................................................. 199

Figure 5.9 Xray diffraction peaks of sample (top graph) and ofpure hematite (bottom graph) .................................................................. 201

Figure 5.10 Time-resolved LII and laser light scattering setup. .................................. 203Figure 5.11 Time-resolved LII at several laser fluences. ............................................ 205Figure 5.12 Time-resolved LII at several laser fluences. ............................................ 207Figure 5.13 Time-resolved LII at two different laser fluences for

flame #1 and flame #2............................................................................. 208Figure 5.14 Time-integrated LII signals for flame #1 and flame #1

as a function of laser fluence. .................................................................. 210

Figure 5.15 Select hematite and nitrogen properties ................................................... 212Figure 5.16 Least-squares fits to the experimental LII curves using a

lognormal (1 and 2 mode) and normal (1 and 2 mode) particle sizedistribution for flame #1 and flame #2..................................................... 214

Figure 5.17 Comparison of LII-derived particle size distributions with grid samplingparticle size distributions ........................................................................ 216

Figure 5.18 LII curves generated from grid sampling data.......................................... 219Figure 5.19 LLS vs. fluence for flame #1 and flame #2.............................................. 221

Page 13: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

x

LIST OF TABLES

Table 4.1 Parameters used in the soot LII analysis ................................................... 151Table 5.1 Flow and flame conditions for the seeded premixed methane flame.......... 186Table 5.2 Parameters used in nanoparticle LII analysis............................................. 213Table 5.3 χ values for the fit to the LII data using various size distributions............. 215

Page 14: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

xi

Table of Nomenclature

A21 Einstein A coefficient (s-1)a particle radius

mean of the classical polarizability tensor

a' mean of the quantum mechanical polarizability tensora0 initial particle radiusa0 particle radius where the particle size distribution is a maximuma0,1 a0 for mode #1 of a multi-modal size distributiona0,2 a0 for mode #2 of a multi-modal size distributionaes particle radius for a volume-equivalent spherebj,j Placzek-Teller coefficientsbj j±2,

c speed of lightC constantCa absorption cross section for isolated spherules

Cs total scattering cross section for isolated spherulesCp

νν scattering cross section for light perpendicular to the incident light

polarization for an isolated spheruleCa

νν scattering cross section for light perpendicular to the incident light

polarization for an aggregate of spherulescv

* mean specific heat (at constant volume) between Tg and Tp

cp specific heat (at constant pressure) of the particle materialD coefficient of diffusionDf fractal dimension

rE electric field

eb Planck functionE(m) refractive index function Im[(m2 - 1)/(m2 + 1)]F(m) refractive index function |(m2 – 1)/(m2 + 2)|2

fv volume fractiong(λ) spectral detection efficiency

h Planck constant

h h/2πH sensible enthalpy cpT/QI intensity of incident lightI0 laser intensityIem particle emission intensity

Page 15: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

xii

Iiz scattering intensity polarized in the i for an input light source polarizealong the z direction

Iiz,Ray Rayleigh scattering intensity polarized in the i for an input light sourcepolarized along the z direction

Iiz,Ram,m Raman scattering intensity of species m polarized in the i for an input light

source polarized along the z directioni (-1)0.5

J rotational quantum numberk wave number of light 2π/λkB Boltzmann constantkƒ fractal prefactor

Kabs absorption cross section of an isolated spheruleKe extinction coefficientKn Knudsen numberlg mean free pathm complex refractive indexmg mass of a gas particlemp rate of mass change of a particle

M magnification of the optical systemnp number of particles per aggregateNa number density of aggregatesNp particle number densityNm number density of species mN* number density of an excited-state speciesNtot total number density of all speciesn number of particlesp(a0) particle size distribution function

P pressurepg gas pressurepv

* vapor pressure at a reference temperaturepO2

partial pressure of oxygen

rp dipole moment

Q lower calorific value of the fuelq 2ksin(θ/2)

R gas constantRg radius of gyrationri location of particle i within an aggregate with origin at the center of mass

Sem emission intensity relative to background emission

Page 16: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

xiii

rs displacement vectorT temperatureTp particle temperatureTp,0 initial particle temperatureTg gas temperature

Tp* particle temperature at a reference point

V volumeVp individual particle volumeVem emission volume (cm3)

rv velocityvg mean thermal speed of the gas molecules (Maxwellian)

vv mean thermal speed of the vapor

Wf molecular weight of the fuelWmix molecular weight of the mixtureWi molecular weight of species iWv molecular weight of the particle vapor speciesx' coordinate of the in-plane emission distributionxpixels number of pixel columns in the image

rX1 position of the particle image for the first exposure

rX2 position of the particle image for the second exposure, at a time ∆t after

the first exposureX mole fractionXm mole fraction of species mY mass fractionYF fuel mass fraction

Greek terms

α thermal accommodation coefficient

αv evaporation coefficient

α electronic polarizability tensorα iz polarizability components

αzz on-axis polarizability components

αyz off-axis polarizability components

αxz

β volumetric thermal expansion coefficientconserved scalar

γ* mean value of (cv + R)/cv between Tg and Tp

Page 17: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

xiv

γ anisotropy of the classical polarizability tensor

γ' anisotropy of the quantum mechanical polarizability tensor

∆Hv heat of vaporization

ξ mixture fractionξFT two scalar mixture fraction based on YF and cpT/Q

ξCHO mixture fraction based on mass fraction of C, H, and O

ε spectral emissivity

ε0 dielectric constant

ƒ form factorκ imaginary part of the refractive index

λ wavelength

λem emission wavelength

λex excitation wavelength

η detector efficiency in counts per photon

dimensionless parameter hc/λkBT

ηem η(λ = λem)

ν vibrational quantum number

νi stoichiometric coefficients of ν ν νF O PF O P+ →νO

νF

ρm depolarization ratio of species m

ρ density

ρp particle density

σ size distribution spread parameter

σ1 σ for mode #1 of a multi-modal size distribution

σ2 σ for mode #2 of a multi-modal size distribution

σSB Stefan-Boltzmann constant

σRay Air, Rayleigh scattering cross section for air

σRay He, Rayleigh scattering cross section for helium

∂σ∂Ω

m iz,

differential scattering cross section for species m, collection of light

polarized along the i axis, and incident light polarized along the z axis∂σ∂Ω

Ram m iz, ,

differential Raman scattering cross section for species m, collection of

light polarized along the i axis, and incident light polarized along the zaxis

Page 18: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

xv

∂σ∂Ω

eff iz Ray, ,

differential Rayleigh scattering cross section for collection of light

polarized along the i axis and incident light polarized along the z axis,where the contribution of each species is weighted by its mole fraction

θ angle (degrees)

τ integration time (s)

bandwidth factor correctionω frequency of light (s-1)

chemical production rateΩ detection solid angle

χ least-squares error parameter

Page 19: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

1

Chapter 1

Introduction

Optical diagnostic techniques are used effectively in combustion systems as a method of

quantifying the system without disturbing the system itself. Light emitted from and

scattered off of these systems gives information of species concentrations, temperature,

velocity, and particle size and concentration. These signals are often spectral and

temporal signatures that are specific to a particular species or possibly the dimensions of

particles or particle aggregates. Use of a monochromatic, coherent light source along with

fast optical detection equipment allows easy detection and interpretation of these signals

due to the high spatial, temporal, and spectral resolution achieved. The lack of divergence

of lasers allows for remote measurements in systems where this would not otherwise be

possible with physical probes.

There is a recognized need in the world today for tighter controls on pollutant emissions

for industrial factories, automobiles and power plants. Laser diagnostic techniques offer

long term monitoring of these emissions, whereas a physical device will eventually

corrupt due to deposits and corrosion. Monitoring can be done on not only current

combustion facilities, but can provide feedback for the construction of more efficient,

cleaner burning combustion facilities.

Page 20: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

2

As the level of detail and sophistication of numerical modeling increases, laser

measurements are often the only means available with the spatial resolution and

sensitivity needed to check the computational results. Experimental confirmation of

these simulations provides the necessary confidence to extend the computational models

to systems of increasing complexity. With better models, it is easier to develop and

evaluate new, more complex practical devices that are both more efficient and have a

lesser impact on the environment.

Certain combustion-generated materials have properties that make them of considerable

economic importance. For an example, thin films created by deposition of combustion-

synthesized particles have special magnetic and optical properties are used in data storage

and communications technologies. The monitoring of the production of these materials is

critical to the special properties of the films. Laser diagnostics provide real-time

monitoring of the synthesized materials and thus a feedback loop in the production of

these materials.

This dissertation introduces the optical techniques of Raman and Rayleigh scattering,

laser-induced incandescence (LII), laser light scattering (Mie scattering), absorption,

flame emission (chemiluminescence), and particle image velocimetry, and applies these

Page 21: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

3

techniques to simple, ideal combustion systems (i.e. systems which have an axis of

symmetry and are repeatable over any length of time or at least over a specified period).

These techniques are used in conjunction with modeled quantities to quantify the systems

in terms of velocity, temperature, gaseous species and particle concentrations, and

particle and aggregate dimensions. Results obtained are compared to computational

results and with the results of non-optical experiments.

Chapter 2 describes the fundamental theory behind the diagnostic techniques used. The

fundamental principles behind the optical techniques used to relate the optical signals to

the underlying physical quantities such as temperature, concentration, size, etc.

Chapter 3 presents the quantitative characterization of a non-sooting laminar methane,

coflowing diffusion flame using several techniques. The techniques are performed on a

flame with steady fuel flow and on a time varying flame where the fuel flow is

modulated. Chemiluminescence is used to determine the variation in flame structure due

to the flow modulation. Next, Rayleigh and fuel Raman scattering are used as a two

scalar measurement to determine temperature and mixture fraction for the forced flame.

Finally, a technique using Rayleigh and multi-species Raman scattering is used to

determine major species concentration, temperature and mixture fraction in the steady

and time-varying flames. A sub-technique of this approach involves using the orthogonal

Page 22: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

4

polarized Raman scattering signals to eliminate fluorescence interference on the Raman

signals. The results of the two-scalar technique are compared to the mixture fraction and

temperature obtained from the multi-species technique. The results of the multi-species

technique are compared to computations of temperature and major species

concentrations. The effectiveness and error estimates of the two techniques are discussed.

Chapter 4 presents the quantitative characterization of a sooting, laminar, ethylene,

coflowing diffusion flame. Rayleigh and fuel Raman scattering are used in the two scalar

technique to obtain a temperature image. The two scalar temperature is compared to

computations and thermocouple probe measurements. Time-integrated LII is used to

quantify soot volume fraction in the flame. LII results are compared to probe sampling

measurements of soot volume fraction along with computations. Time-resolved LII is

used along with modeling to obtain soot particle size distributions and to estimated mass

vaporization limits of the soot particles. LII-derived particle size distributions are

compared to particle-sampling derived particle size distributions. Time-resolved

absorption and laser light scattering are used to study the effect of the laser pulse on the

particles/aggregates.

Chapter 5 presents the characterization of inorganic nanoparticles created downstream of

a premixed laminar methane/air flame. Time-resolved LII is used along with modeling to

Page 23: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

5

obtain primary particle size distributions and to estimate mass vaporization limits of the

particles. Time-resolved Mie scattering of the particles is used to obtain aggregate size

information and to study the effect of the laser on aggregates. LII-derived particle size

distributions and aggregate information inferred from laser light scattering measurements

are compared to grid sampling particle/aggregate distributions.

Page 24: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

6

Chapter 2Light Scattering Techniques

2.1 Introduction

Optical techniques used today in combustion are based on optical principles known for

some time. Emission spectroscopy dates back to 1857 when Swan observed C2 emissions

from flames [Gaydon 1974]. In 1871, Lord Raleigh formalized earlier observations that

light preferentially scatters in the blue. In 1928, Raman and Krishnan observed a

modified scattering of light in a medium that occurs at an altered wavelength from the

incident light [Long 1977].

Optical measurements are remote and non-intrusive, allowing for measurements at the

specific location and time of interest, as opposed to waiting for products which exit the

volume of interest in a sampling approach. Optical techniques often detect spectral

signatures of atoms or molecules which can be detected by no other available method.

In practical combustion systems, the hostile environment may prohibit the use of physical

probes.

Page 25: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

7

The use of lasers offers high spatial and temporal resolution, allowing for instantaneous

multi-dimensional measurements. A pulsed laser is often used to permit the region of

interest to be frozen in time and space. The high repetition rate of some lasers allows for

good frequency tracking of time dependent events such as turbulence. Lasers offer the

opportunity to study the fundamentals of atoms/molecules by probing specific

atomic/molecular states. With a very fast laser pulse along with a fast repetition rate,

molecular phenomena such as energy transfer/chemical reactions can be studied.

2.2 Rayleigh and Raman scattering

When an electric field is incident on a medium, it induces electric dipoles in the medium

which mainly line up in the polarization direction of the electric field. The degree of

induced polarization is related to the strength of the electric field. For a molecule, the

dipole moment p→

induced by an electric field E→

is

r rp E= ε α0

˜ (2.1)

where the electronic polarizability α of a molecule is a tensor in general. This

Page 26: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

8

polarizability characterizes how easily the light will induce a dipole moment for a given

molecule.

A medium can have different polarizabilites along different axes. This means that for a

given input wave with E→

along a defined axis, some of the induced dipoles will line up in

orthogonal directions to E→

, as well as along the direction of E→

. For many of the gases

important in combustion, the moment along the incident E→

field direction is

approximately two orders of magnitude greater than off axis moments. This principle is

the motivation behind the difference Raman and Rayleigh scattering techniques discussed

in Chapter 3. Defining the incident E→

along the z axis, α zz is the polarizability for

dipoles pointing in the same direction as the input E→

field (i.e. on axis terms), where

( α αyz xz, ) are the polarizabilities terms for dipoles with components pointing in

orthogonal directions to E→

(i.e. off axis terms).

Polarizability can be expanded in powers of E→

, which results in a term linear with E→

field and terms proportional to a higher power of the E→

field. Higher order processes,

even at high input E→

fields ( such as a laser with power density ~ 109 W/cm2) have a

dipole moment that is approximately 10-3 times the preceding lower order dipole moment.

Page 27: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

9

If the polarization is time varying, electromagnetic radiation is emitted from the medium

with the same time variation as the incident polarization of the electromagnetic field. This

new EM wave combines with the incident wave- however, there is a phase lag between

the old and new waves as there is some response time of the medium to produce

oscillating dipoles in response to the incident light.

One resulting combination is a wave scattered with the same frequency as the oscillation

of the incident EM wave (thus an elastically scattered wave). This is called Rayleigh

scattering. It is a linear process and therefore the induced dipole moment of the medium

is linear with the strength of the incident E→

field.

Another combination of the waves is the result of the interaction of the induced EM wave

from the oscillating dipoles with the oscillation or rotation of a molecule about its

equilibrium position. This interaction produces a scattered wave that is shifted from the

oscillation frequency by the frequency of a particular vibrational and/or rotational mode

of the molecule. The resulting frequency of the scattered wave can be higher than the

initial EM wave frequency (Stokes shifted) or lower in frequency (anti-Stokes). This is

therefore an inelastic process, and is termed Raman scattering.

Page 28: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

10

The total power of a scattered wave from an induced dipole is

P p E~

~

ω ω ε α42

402

2 2→

=

( )r

(2.2)

For a linear process as Raman and Rayleigh scattering the dipole moment is linear in E→

and therefore P ~ E I→

2

~ , where I is the intensity of the input radiation source.

In a gas, molecules are randomly oriented with respect to one another. Therefore, one

must average α~

2

over all orientations of the molecules. Defining E→

along the z-axis,

the average of the square of the polarizability terms

α γzz a( ) = +( )2 2 21

4545 4

(2.3)

α α γyz xz( ) = ( ) = ( )2 2 21

15

where a is called the mean and γ is the anisotropy. The mean consists only of on axis

polarizability terms, and the anisotropy consists of only off axis terms. For most

scattering processes in typical combustion gases, a is almost two orders of magnitude

larger than γ.

In a typical experiment, scattered signals are collected at an angle from the laser axis to

avoid scattering interferences. An ideal location is perpendicular to the axis. This permits

Page 29: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

11

easy two-dimensional imaging, since detection at another angle would require a

transformation of the projected image.

In this experiment, a linearly polarized laser interacts with multiple gaseous species to

produce Raman and Rayleigh scattering. If the polarization axis of the laser is defined as

the z-axis, the laser propagation direction is defined as the y-axis, and the scattering is

collected along the x-axis, the contribution from species m to the radiant intensity of the

scattered light polarized along each of these axes is

I N Iyz mo

m yz m, ,~90 0

2 4( ) ( )

α ω (2.4)

I N Izz mo

m zz m, ,~90 0

2 4( ) ( ) α ω

I xz mo

, 0 0( ) =

The above scattered intensity equations can be arranged in the general form

I CN VIiz m mm iz

,,

=

0

∂σ∂Ω

(2.5)

The depolarization ρm of species m is defined as the ratio of the scattered light intensity

perpendicular to the incident light polarization divided by the scattered light intensity

parallel to the incident light polarization. Combining (2.3) and (2.4),

ραα

γγm

yz m

zz m

yz m

zz m

m

m m

I

I a= = =

+,

,

,

,

( )

( )

2

2

2

2 2

345 4

(2.6)

Page 30: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

12

Therefore a completely depolarized molecular transition ( am = 0) gives ρm = 3/4. This

ratio is molecule specific. Tabulated values of ρm are obtained from other work [Penney

1972, Woodward 1967, Murphy 1977, Rowell 1971, Schrötter 1979, Holzer 1973].

In Rayleigh scattering, the scattering intensities of all species m in the probe volume

spectrally overlap, since Rayleigh scattering is an elastic phenomenon. The resultant

Rayleigh signal is therefore the sum of the scattering intensities of all species. Using the

general scattering intensity of (2.5):

I I CVIdd

N

CVI Ndd

X

iz Ray iz Ray mm m iz Ray

mm

totm iz Ray

mm

, , ,, ,

, ,

= ∑ =

=

0

0

σ

σ

Ω

Ω

(2.7)

=

CVI N

ddtot

eff iz Ray0

σΩ , ,

Through the ideal gas law, NP

k TtotB

= , one can relate the Rayleigh signal to temperature,

I CVIPkT

ddiz Ray

eff iz Ray,

, ,

=

0

σΩ

(2.8)

If one calibrates the Rayleigh scattering signal of the experiment with a Rayleigh signal

of known composition and temperature (labeled "ref"), and assuming constant pressure as

in the simple systems we will be studying, one can relate the Rayleigh signal to the

temperature T from (2.8):

Page 31: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

13

TI ref

II

I ref

dd

dd

refT ref

aI

iz Ray

iz Ray

eff iz Ray

eff iz Ray

T

iz Ray

=

=,

,

, ,

, ,

,

( )

( ) ( )( )0

0

σ

σΩ

Ω

(2.9)

where aT depends upon the local composition and is independent of temperature.

In Raman scattering, one detects scattering at a frequency shifted from the laser

frequency by an amount specific to a particular vibrational-rotational mode of a

molecular species. Thus the Raman scattering signal for a specific molecular species can

in general be isolated from elastic scattering and from the Raman scattering signals of

other species. The Raman scattering intensity for species m is related to the number

density of species m through (2.5) I CN VIiz m mm iz

,,

=

0

∂σ∂Ω

I CN VIRam iz m mRam m iz

, ,, ,

=

0

∂σ∂Ω

(2.10)

If one calibrates the experimental Raman signal of species m with a Raman signal of a

known quantity of a species n at a known temperature ("ref"), one can relate the Raman

signal to the number density Nm using (2.10):

NI

I refI ref

IN ref

dd

T

dd

Tm

iz Ram m

iz Ram nn

iz Ram nref

iz Ram m

=

, ,

, ,

, ,

, ,

( )( )

( )( )

( )

0

0

σ

σΩ

Ω

(2.11)

The Raman cross section does have a temperature dependence. This temperature

dependence is the result of the quantized levels of the rotational/vibrational states of a

Page 32: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

14

molecule. Higher rotational/vibrational energy levels are populated when the temperature

increases, thus decreasing the signal in each level. From a full quantum mechanical

model, the temperature dependence of the Raman cross section of a molecule which

behaves as an ideal harmonic oscillator is:

11− −( )−

exp( / )hω k TB (2.12)

Rotational states (if any) of a molecule may interact with the vibrational states of a

molecule, to produce Raman transitions dependent upon ν and J, the respective

vibrational and rotational levels. It is also possible to have purely rotational transitions.

Rotational levels are much closer together than vibrational levels (≈ 10 cm-1 separation

compared to ≈ 1000 cm-1 separation typically for vibrational levels). To be able to resolve

rotational levels spectrally, one needs spectral resolution on the order of 10 cm-1 to

separate out the different rotational transitions. It is quite a bit easier to resolve

vibrational-rotation Raman shifts. Quantum mechanics yields certain selection rules of

the Raman transitions. For vibrational/rotational Raman (referred to as just vibrational

Raman), these rules are ∆ν=+/-1 and ∆J=0,+/-2. From Placzek polarizability theory, the

average of the square of the derived polarizability tensor components over all orientations

for individual transitions in diatomic molecules are

Q branch (∆J = 0, ∆ν = +1) ( ) ( ' ) ( ),'ν γ+ +

1445

2 2a bJ J (2.13)

Page 33: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

15

O, S branch (∆J = +/-2, ∆ν = +1)445

1 22( ) ( ),

'ν γ+ ±bJ J

where bJ,J and bJ±2,J are Placzek-Teller coefficients and γ' and a' are the mean and

anisotropy of the derived polarizability tensor. Q branch transitions depend upon the

mean and anisotropy while O and S transitions depend only on the anisotropy of the

derived polarizability tensor. This implies that the Q branch transition typically have a

much larger Raman cross section than O and S branch transitions. This also implies that

O and S transitions are completely depolarized, and Q transitions are highly polarized.

Not all vibrational/rotational modes of a molecule are Raman active- this is highly

dependent on the symmetry of the molecule.

In theory, all Q-branch (∆ν = ± 1, ∆J=0) transitions originating from different vibrational

levels overlap perfectly since vibrational energy levels are equally spaced. In practice,

the Q-branch spreads out slightly due to the anharmonicity of the vibration, as well as

some coupling between rotation and vibration [Eckbreth 1996]. The O and S branches

(∆ν = ± 1, ∆J=± 2) are far more diffuse than the Q-branch since the energy difference

between rotational levels increases with increasing J. Since the number of rotational

levels populated increases with increasing temperature, the spectral width of the O and S

branches broaden significantly as the temperature rises.

Page 34: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

16

2.3 Laser-induced incandescence

Laser-induced incandescence is the emission of blackbody-like radiation from particles

that are heated to temperatures well above ambient by a high intensity laser source. The

qualitiative theory and early experiments on laser-induced incandescence (LII) were done

by Eckbreth [Eckbreth, 1977]. The conservation equation for a particle heated by a laser

was first given by Melton [Melton 1984]. The model presented in this work adapts the

original model to the free molecular regime, and includes terms that are significant yet

unaccounted for in the Melton model [Rosner 2001, Filippov and Rosner 2000a, Rosner

2000]. From energy conservation, the equation for a particle that is subject to laser

heating is:

K a I ap v

T T mH

WT T

K deabs

g gp g p

v

vSB p g

abs

em

π απ γγ

π σ η η ηη

η

20

2 4 4 43

211

1 151

− +−

− + − −−∫

*

* ( / ) ˙ ( / ) ( )( )∆

− =43

03π ρa cdTdtp p (2.14)

In the first term of (2.14), Kabs(a, λ) is determined in the Rayleigh limit, where the radius

of the particle, a, is small compared to the excitation wavelength, λex, of the light

absorbed (2

1πλ

a << ), yielding

laser energyabsorbed/ time

rate of heattransfer tomedium

rate of energyused for particlevaporization

rate of blackbodyradiation loss

rate of internalenergy rise ofthe particle

0

Page 35: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

17

K aa m

mabsex

( , ) Imλ πλ

= −+

8 12

2

2 (2.15)

In general, the real and imaginary parts of m will depend upon temperature and excitation

wavelength. The first term of (2.14) is time dependent since I0 is time dependent.

The second term in (2.14) is the rate of heat transfer in the free molecular limit, where the

mean free path in the surrounding gas, lg, is taken to be large compared to the particle

radius- i.e. the Knudsen number is large (Knl

ag= >>1). The mean thermal speed of

molecules in the gas, vg , is derived from a Maxwellian distribution of particles

vk T

mg

B g

g

=

8

1 2

π

/

(2.16)

The average adiabatic constant, γ*, equals (cv*+R)/cv

*, where cv* is the mean value of the

specific heat (at constant volume) between the gas and particle temperatures. The thermal

accommodation coefficient, α, equals 1 if reflection of the gas molecules off the particle

surface is completely diffuse.

The third term of (2.14) is the rate of energy lost due to mass vaporization. This term

competes with particle heating to limit the maximum temperature a particle can reach.

Page 36: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

18

The fourth term in (2.14) is generally neglected since it is small compared to the other

terms below 10000 K. Maximum particle temperatures are typically 4000 K for

carbonaceous particles, as vaporization begins to severely limit the temperature rise

above this point.

The particle density, ρp, in the last term of (2.14) is approximately constant, although the

thermal expansion of the particle is taken into account. From mass balance, the mass flux

is

m

adadt

a dTdt

WRT

p vpp v

v

pv p4 32π

ρ β α= − = − (2.17)

The mean thermal velocity of the particle vapor, vv , is calculated from a Maxwellian

distribution of particles. The vapor pressure of the particle material, pv, is assumed to take

the form of the Clapeyron equation

p pH

RT

T

Tv vv

p

p

p

= −

**

*

exp∆

1 (2.18)

The particle density changes with temperature according to

ρ ρ βp p g p gT T T T( ) ( )exp ( )= − −( ) (2.19)

where Tg is the reference gas temperature and ρp(Tg) is known.

Page 37: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

19

The imaginary part of the index of refraction, κ, will scale with the particle density and

will therefore change with particle temperature

κ κ β( ) ( )exp( ( ))T T T Tg p g= − − (2.20)

For a known initial particle temperature Tp,0 and radius a0, and a known time dependent

laser power density I0, (2.14) and (2.17) may be numerically integrated to determine

Tp(t,a0,I0) and a(t,a0,I0), the time dependence of the particle temperature and particle

radius.

The emission intensity of a particle at a detection wavelength λ em is

I a e Tem b em p= 4 2π ε λ( , ) (2.21)

where e T

e

b em p em hc

kTem p

( , ) ~λ λλ

5 1

1

(2.22)

Since ε λ= K aabs em( , ), (2.21) becomes

I a K a e Tem em b em p= 4 2π λ λ( , ) ( , ) (2.23)

Iem is time dependent as both a and Tp vary with time. The relative incandescence signal

at time t for a particle is then

S I T t a t I T aem em p em p= −( ( ), ( )) ( , ),00

(2.24)

where Tp,0 and a0 are the particle temperature and radius before the laser pulse arrives.

Page 38: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

20

The relative incandescence from a distribution of particles, with initial particle size

distribution p(a0) centered at a0 and ranging from a1 to a2, over a spectral region ranging

from λ1 to λ2, within a volume V, at time t is

J t N S a t T t p a g da d dVp ema

a

Vp( ) ( ( ), ( )) ( ) ( )= ∫∫∫

1

2

1

20 0 0

λ

λ

λ λ (2.25)

The only quantity with inherent spatial dependence is the laser intensity. When

numerically integrating (2.14) and (2.17) to get a(t) and Tp(t) for different values of a0,

the spatial variation in I0 must be taken into account.

2.4 Chemiluminescence

Chemiluminescence is the emission of photons by an atom or molecule that has been

excited by chemical interaction to an electronically-excited state. The signal is directly

proportional to the rate at which the atom/molecule spontaneously emits photons, called

the Einstein A coefficient. The measured emission signal is given by [Hertz 1988]

S A V Nem em= 1

4 21πτ εη*Ω (2.26)

The spectral dependence of the emission is a signature of the excited state molecule,

making molecules/atoms with known emission curves easy to detect.

Page 39: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

21

Since there is no laser sheet to define a measurement plane, the emission signal collected

is an integration of the emission over the line of sight between the flame and a

corresponding pixel location of the two-dimensional imaging system. Since intermediates

in laminar, diffusion flames such as CH and OH occur in a very thin region of the flame,

largely away from the centerline (except near the flame tip), the emission intensity has

cylindrical symmetry. A tomographic inversion technique, called Abel inversion,

converts the line- of-sight integrated emission signal into a two-dimensional in-plane

emission intensity image. The Abel technique assumes a perfectly axisymmetric signal

distribution, and the collection of only infinitely thin and parallel rays. Therefore,

collection must be done as far away from the flame as possible, along with largest lens f/

possible, to allow approximately parallel rays to be collected [Hughey 1982, Dasch 1994,

Walsh 2000].

2.5 Laser light scattering

Laser light scattering (LLS) of particles refers to the elastic scattering of laser light off of

particles. LLS gives information on particle size, number density, and morphology. For

aggregate particle structure, LLS determines the radius of gyration, Rg, of an aggregate

Page 40: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

22

[ Rn

rgp

ii

np2 2

1

1==∑ ]. Early soot experiments attempted to infer soot paticle sizes and number

densities using Mie theory [Kent and Wagner 1982, Santoro et al. 1983]. This theory

assumes particles are spherical. Thus to determine the scattering from an aggregate using

this theory, a volume-equivalent radius (aes) of an aggregate must be determined, and

used in place of the particle radius of a single particle (a):

aes/a = np1/3 (2.27)

This implies that an aggregate, no matter how complex in structure, will scatter the same

amount of light as a spherical particle that has a volume equal to that of the aggregate.

More recent work has shown the equivalent sphere-Mie theory to inaccurately predict aes

for large (np>10) aggregates. [Köylu, 1996].The predictions become more inaccurate for

larger np. From Rayleigh theory (2πa/λ<<1), the scattering cross section for scattering off

an individual particle with polarization parallel to the light source polarization is [Köylu

1993]

Ca F m

kpνν θ θ( ) ~

( )cos ( )

6

22 (2.28)

where F(m) is a function of the index of refraction, and θ is the angle at which the

scattering is collected relative to the forward scattering direction. For aggregates, a more

realistic theory than the equivalent sphere Mie theory is the Rayleigh-Debye-Gans theory,

which accounts for the closely spaced particles within the aggregate. Taking into account

Page 41: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

23

the effects of phase differences in the scattering from individual particles, the parallel

scattering cross section for an aggregate is

C n C qRap

pgνν νν= ƒ2 ( ) (2.29)

where ƒ( )qRg is the form factor of aggregates of any shape and q k= 2 2sin( / )θ . If one

models the aggregates as fractals with fractal dimension Df , the form factor is [Dobbins

and Megaridis 1991]

ƒ =−

( ) expqRq R

gg

2 2

3q2Rg

2<3Df/2 Guinier regime (2.30)

= ( )− ƒq Rg

D2 2 2/q2Rg

2>3Df/2 Power law regime

The relationship between np and Rg for a fractal is

n kR

apg

D

=

ƒ

ƒ

(2.31)

Using (2.29-2.31) one arrives at an expression for the parallel scattering cross section of a

fractal-like aggregate:

C n C aqn

ka

pp p

DF

νν νν= −

ƒ

2 2

2

13

exp ( )

/

q2Rg2<3Df/2 Guinier regime (2.32)

C n Ck

aqa

pp

DFνν νν= ƒ

( )q2Rg

2>3Df/2 Power law regime

2.6 Laser absorption

Page 42: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

24

The measurement of laser absorption in a medium involves recording the intensity of

laser light entering a medium and recording the intensity of light after it has passed

through the medium. The ratio of the two intensities gives information about the bulk

absorption properties of the material, which depends upon the refractive index, m,

primary particle diameter, a, and particle volume fraction of the material. In the Rayleigh

limit, (2πa /λ<<1), the absorption cross section for isolated primary spherical particles is

Given by (2.15) Ca m

maex

= −+

8 12

2

2

πλ

Im (Note Kabs is renamed Ca here to be consistent

with the spectral extinction literature). The extinction coefficient, Ke, is related to Ca and

the total scattering cross section, Cs, through [Köylu 1996]

K N C Ce p a s= +( ) (2.33)

For an input light intensity, I0 , the output intensity, I, through a medium of length L is

given by Beer's law :

I I K Lo e= −exp( ) (2.34)

The condition for applicability of this equation is an optically thin medium, i.e.

KeL<<1. This condition can be relaxed as long as absorption is the dominant mechanism,

i.e., NpCsL<<1 [Bohren 1983]. In this limit,

K N Ce p a= (2.35)

Page 43: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

25

The basic principles of the optical techniques used in the experiments have been

described in this chapter. The final equations used to determine physical quantities from

the measurements are developed in the experimental sections as a given technique is

used.

Page 44: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

26

Chapter 3Characterization of a Steady and Time-Varying,

Axisymmetric, Laminar Diffusion Flame

3.1 Introduction

Laminar, two-dimensional flames (with one axis of symmetry) are well suited to laser-

diagnostic studies and computational modeling due to their stability ( i.e. remaining

frozen in space) and their symmetry. This allows for signal averaging in the experiment,

which will greatly improve the signal-to-noise ratio for a weak optical process used as the

diagnostic tool. The simple, symmetric and predictable flows of these flames allow for

computational modeling that includes both detailed chemistry and full fluid mechanics.

In turbulent combustion modeling, it is impossible computationally to incorporate both

the turbulent fluid mechanical models and detailed chemistry, so reduced reaction

mechanisms need to be developed. Another class of flames has been studied to bridge

the gap between laminar and turbulent flames: time-varying laminar flames. This flame

may be formed by imposing a periodic fluctuation in the flow of a laminar flame. This

flame offers the advantage a repeatable interaction of chemistry and fluid mechanics.

Therefore a detailed chemistry model along with the well-defined fluid mechanical model

can be applied to computationally model this flame. Information on chemistry and flow

interaction in these flames can be used to help modelers develop better models for

turbulent combustion. The repeatable environment and detailed chemistry calculations

Page 45: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

27

make it possible to evaluate different possible reduced mechanisms, which in turn may be

applied to the turbulent combustion models. The turbulent combustion simulations may

then be used to help design practical combustion devices, where combustion is almost

always turbulent. Since many different reactions comprise the detailed chemistry model,

each of which is described by an uncertain reaction rate, the results obtained in the

numerical simulations must be verified experimentally.

3.2 Flame and Burner Characterization

The flame studied here is a lifted, axisymmetric laminar diffusion flame. The fuel is

methane, diluted with nitrogen (35% dilution by volume) to reduce soot production in the

flame. Methane is used for its simple structure that allows for detailed computational

modeling. The flame is lifted to prevent heat loss to the burner, which will simplify the

computations. The suppression of soot production allows us experimentally to measure

species concentrations and temperature in every part of the flame, while also allowing

less complex modeling. The flame is surrounded by a coflowing annular region of air

which prevents dust particles from entering the system, as well as helping to lift the flame

from the burner. This flame has been computationally modeled with full C2 and nitrogen

chemistry [Smooke 1996], and has been the basis for a number of computational and

experimental comparisons [Smooke 1996, 1992, 1990, Xu 1993], as well as previous

Page 46: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

28

theses [Marran 1997, Lin 1995, Xu 1991]. Major species and temperature measurements

have been taken on this system using spontaneous Raman scattering [Lin 1995, Xu 1993,

Smooke 1990]. These experiments produced significant interferences from fluorescence

of fuel fragments just rich of the flame zone. Difference Raman scattering [Marran 1997]

is effectively applied to eliminate these interferences.

The burner is shown in Figure 3.1. The burner consists of a cylindrical fuel tube

surrounded by an annular region containing a 1/64" cell honeycomb, which is flush with

the top of the fuel tube. The honeycomb straightens the flow of air to produce a radially

constant coflowing region. The 6.8 cm long fuel tube is almost 3 times longer than the

length needed for fully developed pipe flow under the conditions studied. The fuel tube

inlet is open, producing a parabolic radial velocity profile across the tube (Poseille flow),

with a maximum velocity along the tube centerline. The fuel tube has an inner diameter

of 4 mm and a wall thickness of 0.5 mm, and is attached at the bottom to a plenum. The

coflow region has a diameter of 5 cm. A loudspeaker is attached to the bottom of the

plenum. This speaker is used to generate the time-varying laminar flame by modulating

the fuel flowrate. Similar time-varying diffusion flames have been studied

Page 47: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

29

Figure 3.1 Forced flame burner.

Fuel Jet

Air

Fuel

Loudspeaker

4 mm Glass Beads

Fine Steel Wool

80 Mesh Screen

1/64" Honeycomb

Page 48: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

30

experimentally [Skaggs and Miller 1996, Smyth et al. 1993, Mohammed et al. 1998].

This study represents the most complete characterization of a flow-modulated flame in

the current literature.

3.3. Boundary Conditions

3.3.1 Steady Flame

The flow of the fuel has a constant, average fuel exit velocity (at the burner surface)

across the fuel tube of 35 cm/s with a peak velocity along the tube centerline of 70 cm/s.

In the steady flame, the flow remains constant in time. The coflow region maintains a

radially constant flow of 35 cm/s. The two coflowing regions have a matched average

inlet velocity to minimize sheer effects in the boundary between them. The flow

conditions are chosen to lift the flame off the burner while remaining in the laminar

regime (i.e. flame is constant in space and time), and have fully developed Poseille flow

in the fuel tube. Flow boundary conditions near the burner surface are verified and

matched with computational flow boundary conditions using Particle Image Velocimetry

(PIV).

Page 49: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

31

3.3.2 Particle Image Velocimetry

Particle image velocimetry determines velocity in a gas or liquid by acquiring multiple

consecutive, planar exposures of Mie-scattered light from particles seeded into the flow.

Techniques used fall into 1 of 2 categories:(1) acquiring multiple consecutive exposures

with a single image, such that the image has multiple images of the same particle, or

(2) acquiring separate consecutive images, such that each image contains one particle

image for each particle. The distance and direction of the flow in the measurement plane

are determined. Since the time between exposures is known, one can determine the flow

velocity by the following expression:

rr r r

vs

M tX X

M t= = −

∆ ∆1 2 (3.1)

where M is the magnification of the image, ∆t is the time between exposures, and r rX X1 2,

are the image positions of Mie scattering off a particle in exposure 1 and exposure 2

respectively. PIV requires the consecutive particle images to be similar in intensity and

size in order to correlate the image pair. This requires either the same light source for

consecutive exposures or two equal intensity, spatially overlapping light sources for

consecutive images. Rather than identify individual image pairs, the average particle

displacement is computed over an interrogation window of the PIV images using

correlation algorithms. For technique (1), under the assumption that the imaging device

Page 50: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

32

only detects intensity and not color, for a single particle it is not known which particle

image refers to rX1 and which to

rX2 . This leads to a directional ambiguity in the

measurement. Also, zero velocity cannot be measured as the particle images will be right

on top of each other. Therefore technique (2) is considered an improved technique, and is

used in the velocity measurements here. The drawback of technique (2) is the

requirement for equipment capable of acquiring consecutive images separated by ∆t,

appropriate for the flow studied.

In this experiment, Mie scattering images of sugar particles seeded into cold flowing

gases are used for the PIV analysis to verify velocities near the burner inlet for the fuel

tube and coflow regions. A flow of room temperature air with flow rates matching that of

the flows for the experiment is seeded with sugar particles (TSI particle generator Model

9306, with a concentration of 1 g/L sugar in a 50/50 water/methanol mixture). Under

these conditions the atomizer produces sugar particles approximately 1-2 µm in diameter.

It is important that the particles effectively track the gas flows for PIV to give desirable

results. Using the relationship of Melling [Melling 1997] that takes into account the

maximum frequency of the gas motion, seed particle density, gas density, and gas

viscosity, one can determine the maximum allowable particle diameter that will track the

flow exactly. For the forced flame, 20 Hz is the maximum frequency of motion, which

Page 51: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

33

yields a maximum allowable particle diameter of 20 µm. Therefore the 1-2 µm sugar

particles should track the flow exactly. The fuel tube and coflow regions both are seeded.

Seeding densities are chosen to produce at least 8 particle images within the interrogation

region of the correlation [Keane 1992]. A frequency doubled Nd:YAG laser (532nm

wavelength, 10 Hz rep rate) is Q-switched twice for each repetition to produce two

consecutive green light pulses of 8 ns duration. These consecutive laser pulses are needed

for the consecutive Mie-scattering images. Laser pulse separation is 250 µs, and each

pulse has approximately the same energy and energy distribution. This time separation is

found to give the most reliable results using a cross correlation algorithm for interpreting

flow velocities. The laser is focused over the burner with a cylindrical lens, producing a

vertical laser sheet 10 mm tall. The laser sheet is located 1 mm off the burner surface, as

moving the beam closer to the surface caused significant elastic scattering interference in

the images. The consecutive laser pulses produce Mie scattering from sugar particles at

two consecutive instances. The Mie scattering is collected at 90 degrees to the laser by a

fast CCD camera (Cooke Sensicam), which can record two images less than 1 µs apart.

Consecutive Mie scattering-images from the seeded particles are obtained. The

magnification of the images corresponds to 140 pixels/mm. Each imaged region is 9 mm

wide and 7 mm tall, including regions above the jet and above the coflow on both sides of

the jet. A cross correlation algorithm is applied to the image pairs to determine flow

velocities. The size of the FFT interrogation region for the cross correlation is 64 pixels x

Page 52: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

34

64 pixels (or 450 µm x 450 µm). The PIV algorithm produces velocity vectors with a

vector separation of 32 pixels (or 225 µm) in both the horizontal and vertical directions.

The interrogation region in image 2 is displaced from image 1 by the average particle

displacement vector from particles on image 1 to particles on image 2. This eliminates

the undesired affect of particles leaving the interrogation region from image 1 to image 2.

Also, the beam thickness is chosen to be thick enough to eliminate particles leaving the

image plane from image 1 to image 2. Conversely, the beam thickness needs to be thin

enough such that the image volume does not produce out of focus particle images, which

would limit the accuracy of the velocity determination. The laser beam waist is measured

by replacing the cylindrical lens with a spherical lens of same focal length, focusing the

laser beam to a line across the burner, and imaging the Mie scattering from sugar

particles onto the CCD camera. The measured beam waist is 400 µm.

PIV results indicate a parabolic axial flow above the fuel tube at the burner surface as

well as downstream from the surface, along with a uniform axial flow in the coflow

region. There is no detected radial velocity components from the PIV measurements.

These results are compared to the computational boundary conditions at the burner

surface. Figure 3.2 shows the computational velocity profile at 1 mm above the burner

surface compared to the experimental profile obtained by PIV 1 mm above the burner

Page 53: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

35

Figure 3.2 Computational and PIV velocity profiles 1 mm above the burner.The velocity vectors are parallel to the burner axis.

Burner Inlet Velocity

0

100

200

300

400

500

600

700

radial position(mm)

Tube wall

Coflow Fuel Tube

Centerline

02.0 1.03.04.0

PIV

Computational

velo

city

(m

m/s

)

Page 54: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

36

surface. The two profiles show good agreement, with a slightly larger dead zone (region

above the wall of the fuel tube where the velocity is minimal and flat) in the experiment

than in the computation boundary condition. Also the experimental velocity along

centerline is slightly lower (5%). The dead zone length difference may be attributed to

the physical geometry of the honeycomb mating with the fuel tube, while the slight

centerline difference may be due to non-ideal Poseille flow above the fuel tube, or slight

inaccuracy in flow metering.

Slight variation in the dead zone length and slight variation in centerline inlet velocity

produces at most a 1 mm difference in flame lift-off height in the computations.

However, the flame structure remains unaltered. Therefore, slight differences in

experimental and computational boundary conditions should not produce flame profile

differences.

3.3.3 Time-varying Flame

The same volumetric flow rates used in the unforced case are applied to the time-varying

flame. The loudspeaker is driven with a 20 Hz sine wave from a function generator

(HP 33120A). The modulation frequency is chosen as a convenient multiple of the laser

repetition rate rate (10 Hz). The peak-to-peak amplitude of the sine wave is set to 0.7V

Page 55: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

37

to produce 30% flow modulation and to 1.225V to produce 50% flow modulation along

the fuel tube centerline at burner inlet. The reason for studying these specific modulations

is to have a system with enough modulation to make a significant difference from the

steady case but not to create an overly modulated system where the local strain rates may

be too high and computations would be very difficult. This flame oscillates at the

modulation frequency, but does not naturally flicker when there is no modulation

frequency applied to the speaker. Natural flickering in diffusion flames has been studied

by several researchers [Cetegen 1993, Chen 1988, Hamins 1992], and are observed to

flicker at single frequencies. This oscillation frequency is inversely proportional to the

square root of the burner diameter. The oscillation occurring in these jet flames is thought

by several researchers to be caused by the instability of the buoyant plume generated by

the flame and by the interaction of the flame with the plume-generated vortices [Chen

1993, Hamins 1992]. Recent work suggests the oscillations to be caused by flow

instabilities near the wall of the fuel jet [Maxworthy 1999].

The PIV experiment to verify boundary conditions is described in the previous section. In

this experiment, the function generator is synchronized with the laser, camera, and a

delay generator to allow acquisition of any particular phase of the forcing modulation.

PIV images are taken over 1 period of modulation at 10 equally spaced phases of the

forcing. To obtain the most reliable velocity measurements at each phase, the delay

Page 56: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

38

between consecutive laser pulses is varied from 0.1 to 0.4 ms, depending on the specific

phase (or fuel tube exit velocity). The effect is to have similar particle displacements

between the consecutive Mie scattering images for each phase. The maximum variation

in flow velocity over the time between consecutive images produces a maximum particle

image displacement fluctuation, ∂x, of 3 µm or 0.4 pixels. Since particle images are 1-2

pixels in diameter (di), and interrogation regions are 64 pixels in length (d1), the criteria

of Keane for reliable velocity determination using the cross correlation algorithm are

satisfied :

∂x/d1 < .03 (3.2)

∂x/di < 1

Peak centerline velocities near the burner inlet, as well as velocity profiles across the

burner near the surface and downstream, are compared with the computational boundary

conditions over each point in phase space where data is acquired.

The results of the PIV indicate a sinusoidal variation in centerline

velocity closely resembling the computational modulation condition for 30% and 50%

modulation (see Figure 3.3 for the 30% modulation case). The only difference is the

average centerline exit velocity, which is 67 cm/s, as measured by PIV in the unforced

flame. In Figure 3.4, notice the phase lag between the forcing function to the speaker and

Page 57: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

39

Figure 3.3 Computational and PIV fuel tube centerline velocity for 30% modulation of flow in the fuel tube as a function of forcing phase. Points are taken 1 mm above the burner surface.

Centerline Inlet Velocity30% modulation

40.0

50.0

60.0

70.0

80.0

90.0

100.0

0 20 40 60

time(ms)

PIVComputational

velo

city

(cm

/s)

Page 58: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

40

Figure 3.4 Fuel tube centerline velocity and speaker forcing signal for 30% modulation

in the fuel flow as a function of forcing phase. Notice the phase difference between thevelocity (upper curve) and the forcing signal (lower curve)

Centerline Inlet velocity and Speaker Forcing Signal30%flow modulation

Forcing Signal(V)

0

20

40

60

80

100

0 10 20 30 40 50 60

time(ms)

forcing signalvelocity

+0.70

-0.70

Vel

ocit

y(m

m/s

)

Page 59: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

41

the velocities for the 30% modulation case. No radial velocity components are detected

with PIV. The velocity profile across the fuel tube remains parabolic over the observed

phases of the forcing, while the coflow velocity remains constant in time and spatially

uniform over the modulation.

3.4 Computational Modeling

Computations on the flame are obtained by the group of Professor Mitchell Smooke at

Yale University. Two-dimensional species concentration and temperature profiles are

obtained and compared with experiments.

3.4.1 Unforced

The conservation equations are formulated with velocity-vorticity variables (containing

vorticity terms). The equations have first order accuracy in space for diffusive and

convective terms and second order accuracy in space for viscous terms. The steady-state

equations are discretized on a non-uniform, non-adaptive grid. The chemical mechanism

contains C1 chemistry and involves 20 chemical species.

Page 60: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

42

3.4.2 Forced

The conservation equations are formulated with velocity-vorticity variables (containing

vorticity terms). The equations have first order accuracy in space for diffusive and

convective terms and second order accuracy in space for viscous terms. The equations

have second order accuracy in time. The time-dependent equations are discretized on a

non-uniform, non-adaptive grid. The chemical mechanism used involves 15 chemical

species.

3.5 Measurement of CH* via Chemiluminescence

3.5.1 Introduction

Chemically-excited CH (or CH*, the A2∆ molecular state of CH) is the source of blue

light in hydrocarbon diffusion flames. CH* occurs in a very spatially thin, small high

temperature region in diffusion flames, and is a good marker of the flame front. CH*

concentration in flames is spatially coincident with CH concentration [Walsh 2000].

Therefore, measurements of CH* indicate the spatial distribution of CH in the flame. The

radical CH* is also known to play a role in the C2 reaction chain [Najm 1998].

Page 61: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

43

CH* chemiluminesces via the the A2∆ -> X2Π transition. From (2.26)

S A V Nem em= 1

4 21πτ εη*Ω the CH* emission signal Sem is proportional to the number

density N* of CH*. In this experiment there is only determination of relative CH* number

density. The purpose of this experiment is to observe the effect of the flow modulation on

the flame front at different phases of the forcing before more quantitative measurements

are performed.

3.5.2 Experimental Setup and Acquisition

The setup is shown in Figure 3.5. CH* flame emission is collected with a f/8 50mm lens

at a distance of 50 cm from the burner. The collected light passes through an interference

filter centered at 430 nm (10 nm bandpass) and is focused onto a gated, image intensified

(1 ms gate time) CCD camera (Photometrics CC200). This spectral region contains the

maximum CH* emission (from the (0,0) bandhead of the A2∆ -> X2Π transition). This

spectral region also has minimal interferences from flame emission signals of other

species. Images are obtained for the steady and time-varying flames. For the latter,

images are obtained for 10 equally spaced phases (5 ms intervals) over one period (50

ms) of the forcing. Five of the phases correspond to phases of acquisition for the two

Page 62: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

44

Inte

nsif

ied

CC

DC

amer

a

Inte

rfer

ence

Filt

er

Len

ses

Fun

ctio

nG

ener

ator

Mec

hani

cal S

hutt

er

Com

pute

r

Bur

ner

Lou

dspe

aker

Exp

erim

enta

l Set

up

Fig

ure

3.5

Exp

erim

enta

l set

up f

or C

H*

emis

sion

mea

sure

men

t

Page 63: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

45

scalar and multispecies measurements. Acquisitions are phase locked and integrated over

200 intensifier gates. The 1 ms gate time is small enough compared to the 50 ms period

of forcing that there there is little change in the flame over the gate time, and large

enough to acquire sufficient signal.

For several of the phases there is significant interference from soot emission. For these

phases, data is acquired at a separate time in the same manner as described above, but

with a sharp cut colored glass filter (03FCG-115, Melles Griot - cuts off light with a

wavelength below 680 nm) replacing the interference filter. There is no detectable CH*

emission through this filter- the light which passes through is blackbody emission from

soot. This data is used to correct the raw CH* images.

Images of dark current and fixed pattern detector noise are acquired for the same

integration time as the data. These images serve as the background on the CH* images

and soot images.

3.5.3 Image processing

Page 64: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

46

The detector dark current image ID is subtracted from both the soot emission and CH*

emission images ICH*(x,y) and Isoot(x,y). The background subtracted soot image is scaled

by an empirical factor c, and subtracted from the background subtracted CH* emission

image:

I x y I x y I c I x y ICH corr CH D soot D*, *( , ) ( , ) ( , )= −( ) − −( ) (3.3)

The image is cropped such that one half of the axisymmetric image remains (x=0 now

corresponds to the centerline of the flame). For each vertical pixel height y of the

corrected emission image, an Abel inversion is performed to convert the integrated line-

of-sight collection of CH* emission into an in-plane two-dimensional CH* emission

profile. The large object distance and large f/ helps eliminate foreshortening and parallax-

this is critical since the Abel inversion assumes that parallel light rays are collected. The

inverted two-dimensional emission profile is calculated by the following equation:

I x yI x y

xx x x

x x xCH Abelx x

xpixelsCH corr

*,'

*,/

/( ' , )( , )

log( ' )

( ) (( ) ' )= ∑

+ −+ + + −

=

11 1

2 2 1 2

2 2 1 2π∂

∂(3.4)

The derivative ∂

∂I x y

xCH corr*, ( , )

is calculated using a least-squares approximation to three

consecutive pixel values [Dasch 1994, Walsh 2000]. The two-dimensional emission

Page 65: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

47

profile is mirrored to obtain a full symmetric profile, and the image size is scaled and

cropped.

3.5.4 CH* profiles for the steady and time-varying flame

Shown in Figure 3.6 are the CH* profiles for the steady flame, 30% and 50% flow-

modulated flames. Five equally spaced phases ("a"-"e") of the forced flame that span one

period are shown in the figure. Also indicated is the centerline fuel exit velocity for each

phase of the forcing. The flame length is the shortest at phase "a". Then the flame length

increases at successive phases "b"-"e". The degree of flame curvature is minimal at phase

"e" and maximal at phase "b". From the profiles it appears that the flame starts to "pinch

off" at phase "a" and continues to "pinch off" at phase "b", while the downstream flame

portion convects upward and out of the measurement area. As expected, the 50% flow

modulation shows greater curvature in flame profiles and more drastic modulation than

the 30% case. There is no apparent phase lag between the 30% and 50% modulations.

3.6 Fuel Concentration, Temperature, and Mixture Fraction Measurement

using the Two Scalar Technique

3.6.1 Theory and introduction

Page 66: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

48

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

r(m

m)

r(m

m)

r(m

m)

r(m

m)

r(m

m)

min

max

0.04

40

0

6080100

120 0

0.01

0.02

0.03

20

0.05

0.06

stea

dy f

lam

e

Cen

terl

ine

fuel

exi

t vel

ocity

vs.

tim

e

v(cm/s)

time(

s)d

cb

ae

(ii)

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)

(a)

(b)

(c)

(d)

(e)

(a)

(b)

(c)

(d)

(e)

(i)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

stea

dy

r(m

m)

30% 50

%

Fig

ure

3.6

(i)

Exp

erim

enta

l CH

* pr

ofile

s fo

r 30

% a

nd 5

0% m

odul

atio

n.

(i

i) C

ente

rlin

e fu

el e

xit v

eloc

ity v

s. ti

me

(for

50%

mod

ulat

ion)

. Pha

ses

whe

re d

ata

is ta

ken

are

mar

ked

by a

,b,c

,d,e

on

the

grap

h an

d co

rres

pond

to th

e im

ages

labe

lled

likew

ise.

Ste

ady

flam

e im

age

is s

how

n on

the

far

left

. The

ste

ady

flam

e fu

el e

xit v

eloc

ity is

indi

cate

d in

the

grap

h.

Page 67: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

49

Mixture fraction is a useful non-dimensional parameter that can be useful in comparing

the behavior of flames of different geometries and stoichiometries. It is defined as the

mass fraction of all atoms originating in the fuel stream. For practical purposes, reduced

formulations of mixture fraction are desired to limit the number of measurements needed

to calculate mixture fraction.

The two scalar method determines the mixture fraction and temperature from the

measurement of fuel concentration and Rayleigh scattering. The two scalar measurement

approach has been applied to the study of laminar as well as turbulent nonpremixed

flames [Stårner 1994, 1996, Kelman et al. 1994, Frank 1994]. A recent improvement in

the two-scalar method [Fielding 2001] improves the measurement of mixture fraction

near the flame front.

One method for obtaining reduced definitions of mixture fraction is outlined below:

The energy and chemical species conservation equations may be combined into a single

equation through the use of coupling functions, α, that obey the equation

L( )α ω= (3.5)

where L is an operator defined by

L v D( ) [ ]α ρ α ρ α= ∇ ⋅ − ∇r(3.6)

Page 68: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

50

A conserved scalar ß is a linear combination of coupling functions which satisfies

L ß( ) = 0 (3.7)

independently of the knowledge of chemical reaction rates. If one assumes

equal mass and thermal diffussivities (i.e. Lewis number = mass diffusivity/ thermal

diffusivity = 1), equal species diffusivities, and a 1-step reaction between fuel F and

Oxidant O

ν ν νF O PF O P+ → (3.8)

the fuel mass fraction YF and the sensible enthalpy H=cpT/Q may be combined into a

conserved scalar

βFT F pY c T Q= + / (3.9)

Alternatively, Bilger formulated a conserved scalar based on the mass fractions Zi of C,

H, and O [Bilger 1990]:

βν ν νCHO

C

C C

H

H H

O

O O

ZW

ZW

ZW

= + −2 (3.10)

where νi are the stoichiometric coefficients of the atom balance equation

ν ν νC H OC H O P+ + → (3.11)

This definition of β allows for intermediate reactions to take place between reactants and

products, different diffusion coefficients for the involved chemical species, and preserves

the stoichiometric mixture from the stoichiometric equation (F+O -> H2O+ CO2).

Therefore βCHO gives a more rigorous representation of the physical process than βFT .

Page 69: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

51

Figure 3.7 Burke-Schumann flame configuration. Thefuel and oxidizer streams are denoted 1 and 2 respectively.

Oxidizer OxidizerFuel

2 21

Page 70: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

52

The advantage of βFT is the requirement of two measurements as opposed to

measurement of every major species containing C, H, and O for βCHO.

For a simple flame geometry described by Burke and Schumann, shown in Fig. 3.7, the

fuel exits a cylindrical tube, which is surrounded by a concentric tube through which

oxidizer exits. The fuel and oxidizer streams are denoted 1 and 2 respectively.

Then the mixture fraction may be defined in terms of ß:

ξ β ββ β

= −−

2

1 2

(3.12)

Therefore, ξ = 1 in the fuel stream and ξ = 0 in the oxidizer stream. Using the ideal gas

law, and assuming constant pressure, fuel mass fraction is related to fuel number density

by:

YNN

WW

NN ref

N refN

WW

NN ref

X ref TT ref

WWf

f

tot

f

mix

f

tot

f

mix

f f

mix

= = =( )

( )( )

( )( )

(3.13)

where "ref" refers to a reference species of known concentration and temperature. Fuel

mass fraction and enthalpy can be expressed in terms of Rayleigh and fuel Raman

scattering intensities. Assuming the reference species is fuel and the Raman cross section

for the fuel does not vary much with temperature,

(2.11) NI

I refI ref

IN ref

dd

T

dd

Tm

iz Ram m

iz Ram nn

iz Ram nref

iz Ram m

=

, ,

, ,

, ,

, ,

( )( )

( )( )

( )

0

0

σ

σΩ

Ω

becomes

Page 71: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

53

NI

I refI ref

IN reff

iz Ram f

iz Ram f

≈ , ,

, , ( )( )

( )0

0

(3.14)

Combining with (3.13) YNN

WW

NN ref

N refN

WW

NN ref

X ref TT ref

WWf

f

tot

f

mix

f

tot

f

mix

f f

mix

= = =( )

( )( )

( )( )

and (2.9) TI ref

II

I ref

dd

dd

refT ref

aI

iz Ray

iz Ray

eff iz Ray

eff iz Ray

T

iz Ray

=

=,

,

, ,

, ,

,

( )

( ) ( )( )0

0

σ

σΩ

Ω

gives

Y CI

WaIf

Ram f

mix

T

Ray

= 1, (3.15)

H Cc

QaI

p T

Ray

= 2 (3.16)

where CX ref W I ref

I reff Ray

Ram f1 =

( ) ( )

( ),

and C I refRay2 = ( ) are constants. (Note the subscript "iz"

is dropped as the signals measured are integrated over all polarization directions.)

Combining (3.9) βFT F pY c T Q= + / , (3.12) ξ β ββ β

= −−

2

1 2

, (3.15), and (3.16), the

mixture fraction formulation based on the conserved scalar βFT can be expressed in terms

of fuel concentration and Rayleigh scattering intensity:

ξ FTT Ram f

mix Ray

p T Ray pCa I

W IC

c a I c T

Q= +

−3 4

2 2, ,( / )(3.17)

where CC

Y c T c T QF p p3

1

1 1 1 2 2

=+ −, , ,( ) /

and CQ

Y c T c T QF p p4

1 1 1 2 2

1=+ −

/( ) /, , ,

are constants.

Page 72: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

54

Therefore one can determine ξ and T from two scalar measurements in a flame: Rayleigh

scattering and fuel Raman scattering (one can use an alternative scalar measurement for

fuel concentration, as long as the signal intensity is proportional to fuel concentration.)

The terms aT,Wmix, and cp are all functions of ξ and reactedness, r, which is defined by

rT T

T Tad

= −−

2

2

(3.18)

It has been shown by recent work that defining aT,Wmix, and cp as a function of ξ FT gives

a two scalar mixture fraction which is closer to ξCHO than if aT,Wmix, and cp are defined as

funtions of ξCHO. In the unreacted (frozen) regions, aT, Wmix, and Cp (with subscript "fr")

are all linear with ξ , and usually nonlinear with ξ in the reacting regions (with subscript

"r"). The terms in the reacted region are modeled from computations. The resulting terms

aT, Wmix, and Cp are linear functions of reactedness and take the form (showing aT as an

example)

a r a raT T fr T r= − +( ) , ,1 (3.19)

With reference data at known concentrations and temperatures for the Rayleigh and fuel

Raman signal, C1 and C2 can be determined. Initially, ξ and r are set equal to zero. Then

all the functional terms (aT,Wmix, and Cp) are calculated. Using these terms, ξ FT is

calculated from (3.17), and temperature is calculated from (2.9) T aI ref

ITRay

Ray

=( )

. From

Page 73: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

55

(3.18) r is calculated. This new ξ and r are used to calculate aT,W, and Cp again, and the

process is repeated until there is convergence.

3.6.2 Experimental Setup

The setup is shown in Figure 3.8. A frequency doubled, Q-switched Nd:YAG laser ( 532

nm wavelength, 10 Hz rep rate), producing green light pulses of 8 ns in duration and

energy/pulse of 200 mJ is focused by a cylindrical lens into a vertical sheet 18 mm tall

over the burner surface. Rayleigh scattering and Raman scattering are collected

perpendicular to the laser with a 50 mm focal length camera lens (with f/ 1.8 for Raman

and f/ 5.6 for Rayleigh imaging). The acquisition of the Rayleigh and Raman signal is

done independently of one another, justified by the repeatability of the flame. An

interference filter is used to spectrally isolate the Rayleigh and Raman scattering. For

Rayleigh imaging, an interference filter centered at 532 nm (10 nm bandpass) is placed

just behind the lens. A bandpass of 10 nm implies the throughput of the filter decreases to

50% of its maximum at +/- 5 nm from the wavelength of maximum throughput, and

typically less than 5% of maximum throughput at +/- 10 nm . For fuel Raman imaging,

an interference filter centered at 630 nm (10 nm bandpass) is used, which corresponds to

the spectral region of the Stokes-shifted methane Raman scattering using the specific

Page 74: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

56

Figure 3.8 Experimental setup for two scalar imaging

laser

burner

50 mm camera lens

interference filter

CCD camera

delaygenerator

image intensifier

Computer

Two scalar setup

Page 75: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

57

laser wavelength of 532 nm. The collected signals are imaged onto an intensified CCD

camera (Photometrics-CC200), where the data is digitized and transferred to a computer.

Pixel magnification is 12.5 pix/mm, and the corresponding imaged region contains the

fuel jet centerline out to a region of ambient air above the coflow. The laser beam

waist is estimated by replacing the cylindrical lens with a spherical lens of same focal

length, focusing the laser down to a line over the burner, and imaging the Rayleigh

scattering of air onto the CCD. A beam waist of 240 µm is estimated.

3.6.3 Acquisition

The laser is synchronized with the intensifier gate and camera with digital delay

generators to allow acquisition only when a laser pulse is present. This minimizes broad

spectral interferences from the flame such as luminosity. Since the flame is repeatable in

time, the scattering from many laser pulses can be averaged together on the CCD chip.

This greatly improves the signal-to-noise ratio of the images, especially for the much

weaker Raman process. Raman images are integrated over 600 laser pulses and Rayleigh

images are integrated over 100 laser pulses. Data are acquired at two different heights

above the burner at separate times. The first height has the bottom edge of the laser sheet

5 mm off of the burner surface, as moving closer to the surface caused significant elastic

scattering interferences from the burner. The next height had the bottom edge of the sheet

Page 76: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

58

18 mm above the burner surface. These images, after processing, are tiled together to

form single Rayleigh and Raman images.

In the raw methane Raman images at an approximate downstream location of 20 mm, one

begins observing significant interferences in the hot regions of the flame where the fuel

starts to disappear (see Figure 3.9), possibly from laser-induced incandescence of soot.

Therefore, data are only acquired with the laser beam at the height closest to the burner

for the Raman images. The laser, intensifier, and camera are synchronized to a delay

generator, such that data can be acquired at a specific phase of the forcing. Data are taken

for the 30% modulation case over one period of the forcing at five equally spaced

intervals (10 ms apart) such that one phase corresponds to the minimum fuel tube exit

velocity for the forcing cycle.

3.6.4 Data Processing

The raw images are downloaded onto a computer, where one can perform mathematical

operations on them at each pixel location (x,y). An image is acquired to measure detector

dark current and fixed pattern noise by closing the camera shutter. This signal ID is

subtracted from the raw Raman signal IRam,Flame. Background elastic scattering

Page 77: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

59

Figure 3.9 Methane Raman intensity profile (arbitrary scale) from the two scalarexperiment. The color scale is chosen to saturate most of the methane Raman signal toeasily see the interference signal (inside the white box).

min

max

0 4 8-8 -4

5

10

15

20

25

30

35

40

r(mm)

z(m

m)

Page 78: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

60

interferences in the Rayleigh signal are corrected for using the Rayleigh signal from a

uniform field, room temperature calibration of pure helium I x yRay He, ( , ) and the Rayleigh

signal from a uniform field, pure air room temperature calibration I x yRay Air, ( , ) . The

background signal I x yRay b, ( , ) on the raw Rayleigh flame data I x yRay Flame, ( , ) is [Long

1993]

I x y I x y I x y I x yRay b Ray HeRay He

Ray Air Ray HeRay Air Ray He, ,

,

, ,, ,( , ) ( , ) ( , ) ( , )= −

−−( )σ

σ σ(3.20)

Optical throughput and variation in laser beam intensity profile are corrected for by

dividing the background subtracted images by Raman and Rayleigh images of uniform

field, room temperature flows of pure air and pure methane, I x yRay Air, ( , ) and

I x yRam Meth, ( , ) respectively. These signals are also corrected for background noise and

interference. The pure fuels also serve as reference data for the Rayleigh signal in

ambient air and the methane Raman signal at ambient temperature. The corrected

Rayleigh and Raman signals are

I x yI x y I x y

I x y I x yRay corrRay Flame Ray b

Ray Air Ray b,

, ,

, ,

( , )( , ) ( , )

( , ) ( , )=

−−

(3.21)

I x yI x y I x y

I x y I x yRam corrRam Flame D

Ram Meth D,

,

,

( , )( , ) ( , )

( , ) ( , )=

−−

Since the reference data and flame data are taken at different times, there is a chance that

the laser intensity profile changes. This will produce horizontal stripes of signal variation

Page 79: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

61

most noticeable in the Rayleigh image. The stripes are corrected for using a rectangle in

the ambient air region of the corrected Raleigh image that extends from the top to the

bottom of the image. This rectangle is converted into a stripe image that is the same size

as the corrected Rayleigh image. The stripes are removed when the corrected Rayleigh

image is divided by the stripe image.

3.6.5 Calculation of fuel concentration, temperature, and mixture fraction

For Wmix, aT, and Cp, computations of the steady flame for 65/35 CH4/N2 are used to

determine these quantities in the reacted part of the flame as a function of local mixture

fraction. The mixture fraction is calculated from the computations at each location using

the mixture fraction formulation based on the conserved scalar made up of fuel mass

fraction and enthalpy ( ξ FT ). When calculating ξ FT from two scalar experimental

measurements, the ξ dependence of Wmix, aT, and Cp based on ξ FT gives better agreement

with ξCHO near the flame front than if the ξ dependence of Wmix, aT, and Cp is based on

ξCHO [Fielding 2001].

The experimentally measured quantities IRay, IRam, the calculated functional dependencies

W FT( )ξ , aT FT( )ξ , cp FT( )ξ , and the known quantities C3, C4, Q are substituted into (3.17)

Page 80: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

62

ξ FTT Ram

Ray

p T Ray pCa IWI

Cc a I c T

Q= +

−3 4

2 2( / ), . Initially, ξ FT is set equal to zero and T is set

equal to 300K (i.e. r=0) in the initial calculation of Wmix, aT, and cp. Next , ξ FT is

calculated in the above equation (3.17). Temperature is calculated from (2.9) TaI

T

Ray

= ,

and used to calculate r from (3.18) rT T

T Tad

= −−

2

2

. The calculated values of ξ FT and r are

substituted into Wmix, aT, and cp and the process is repeated until there is convergence in

ξ FT . The final values of ξ FT and r are substituted into aT and T is calculated from (2.9).

This entire iterative procedure is done for each pixel location (x,y) in the images. Final

images are cropped and mirrored to obtain a full symmetric profile about the fuel tube

centerline.

3.7 Multi-species Measurement using Difference Raman and Rayleigh Scattering

For hydrocarbon diffusion flames, hydrocarbon fragments and soot precursors are

generated in the hot fuel rich areas of the flame. C2 species produce broadband

fluorescence in the visible when excited by a visible laser wavelength. [Beretta et al.

1985, Masri et al. 1987, Osborne et al. 1996]. This makes vibrational Stokes-shifted

Raman signals from species that exist slightly to the rich side of the flame front hard to

discriminate against this interference. The difference Raman scattering technique has

Page 81: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

63

been applied in the past to reduce the C2 fluorescence interferences from this flame in the

unforced case [Marran 1997].

In the present experiment, simultaneous two-dimensional profiles of temperature and

mole fractions of N2, CO2, CH4, H2, O2, CO, and H2O in the steady and time-varying

flame are measured. Temperature and species concentrations are calculated with

measurements of vibrational Stokes-shifted Raman scattering and Rayleigh scattering.

Similar techniques have been demonstrated [Reckers et al. 1993]. Using difference

Raman scattering, the C2 fluorescence interference in this flame are eliminated down to

shot noise levels of the fluorescence signal. The steady flame profiles are compared to

computations as a verification of the multi-species/difference Raman technique. Next,

this technique is applied to the forced flame, and the results are compared to

computations.

3.7.1 Difference Scattering

Difference scattering is the collection of vibrational Q-branch Raman scattered light

intensities under two orthogonal linear polarizations: Izz parallel and Iyz perpendicular to

the laser polarization, using a linear polarized laser source. The difference scattering

signal is

Page 82: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

64

Ids= Izz- Iyz (3.22)

In practice, the laser is not completely linearly polarized and the ability of the collection

system to reject one of the linear polarizations is not perfect. This will lead to a scale

factor on Ids that is eliminated when the difference scattering intensity from the data is

normalized to reference data. From the quantum mechanical treatment of Raman

transitions, each Raman Q-branch transition has a rotational dependence [Holzer 1973],

and thus a temperature dependence as well. In general, the depolarization of the Q-branch

is so small that this J dependence will be insignificant on the scale of experimental error.

For Q-branch Raman transitions ρm is typically 0.02-0.05 [Penney et al. 1972,

Woodward 1967, Murphy 1977], and therefore vibrational Q-branch Raman scattering

(and Rayleigh scattering) are highly polarized along the polarization axis of the laser

(assuming a linearly polarized laser).

For laser-induced fluorescence, there is an initial preferred polarization of the dipole

radiation. In the case of a molecule, if there are “fast” rotations of the molecule which

occur in a time smaller than the decay time of fluorescence, the fluorescence is randomly

polarized if the signal is averaged over its decay time. If the fluorescence interference

from C2 is determined to be randomly polarized when compared to the gate time of signal

acquisition, one should be able to eliminate this interference through difference Raman

scattering.

Page 83: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

65

From (2.7) I CVI Ndd

Xiz Ray totm iz Ray

mm

,, ,

=

∑0

σΩ

, (3.22), and the ideal gas law

NP

kTtot = , the difference Rayleigh scattering signal is inversely proportional to

temperature

I CVI Ndd

X CVIP

kTddds Ray tot

m zz Raym Ray m

m eff ds Ray,

, ,,

, ,

( )=

∑ =

0 01

σ ρ σΩ Ω

(3.23)

Normalizing the difference Rayleigh signal by reference data,

TI ref

II

I ref

dd

dd

refT ref

aI

ds Ray

ds Ray

eff ds Ray

eff ds Ray

T

ds Ray

=

=,

,

, ,

, ,

'

,

( )

( ) ( )( )0

0

σ

σΩ

Ω

(3.24)

where aT' depends on the difference Rayleigh cross section (as opposed to aT).

From (2.8) I CN VIRam iz m mRam m iz

, ,, ,

=

0

∂σ∂Ω

and (3.22), the difference Raman scattering

intensity of species m is proportional to the number density of that species

I CVI N CVI NRam ds m mRam m zz

m Ram mRam ds m

, ,, ,

,, ,

( )=

=

0 01

∂σ∂

ρ ∂σ∂Ω Ω

(3.25)

Normalizing the difference Raman signal with reference data

NI

I refI ref

IN ref

dd

T

dd

Tm

ds Ram m

ds Ram nn

ds Ram nref

ds Ram m

=

, ,

, ,

, ,

, ,

( )( )

( )( )

( )

0

0

σ

σΩ

Ω

(3.26)

Page 84: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

66

The scattered signal intensities are integrated over a particular spectral detection

bandwidth. The bandwidth factor, or fraction of scattered signal detected, depends upon

the spectral detection bandwidth and spectral lineshape of a signal, as well as collection

efficiency. For Raman scattering, changes in temperature can significantly change the

spectral lineshape of the signal. In this experiment, the spectral detection bandwidths only

contain part of the spectrum of the scattered signals. Therefore, as a Raman lineshape

changes, the amount of scattered signal within the detection bandwidth changes.

Prediction of the change in Raman lineshapes with temperature is needed for accurate

determination of the temperature dependent part of the bandwidth factor, τm(T), for each

Raman species. (See section 3.7.7 for calculation of τm(T).) The number density in (3.26)

becomes

NI

I refI ref

IN ref

dd

T

dd

Tm

ds Ram m

ds Ram nn

ds Ram nref

ds Ram m

=

, ,

, ,

, ,

, ,

( )( )

( )( )

( )

0

00

σ

σΩ

Ω

τm(T) (3.27)

dd ds Ram m

σΩ

, ,

is calculated at a reference temperature To and the temperature dependence

of dd ds Ram m

σΩ

, ,

is incorporated into τm(T). The ratio

dd

T

dd

T

ds Ram nref

ds Ram m

σ

σΩ

Ω

, ,

, ,

( )

( )0

can be determined

with calibration data of species m at known concentrations and fixed temperature T0.

Page 85: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

67

Once τm(T) and

dd

T

dd

T

ds Ram nref

ds Ram m

σ

σΩ

Ω

, ,

, ,

( )

( )0

are determined then one can calculate number density

of species m from (3.27).

3.7.2 Multi-species technique in calculation of temperature and species number density

In this technique one measures Rayleigh scattering and Stokes Raman scattering

intensities for all the major species on a horizontal line through the center of the desired

region. The laser is focused down with a large focal length lens that produces an

approximately uniform diameter beam through the region of interest. Since the Raman

and Rayleigh signals have different spectral locations, Raman and Rayleigh signals can

be spatially isolated from each other by spectrally dispersing the scattered light, allowing

simultaneous measurement of Rayleigh scattering and Raman scattering of the major

species. Raman scattering is known to be proportional to Nm, and Rayleigh scattering is

known to be proportional to Ntot (and inversely proportional to T from the ideal gas law).

Since Nm is measured for the major species, then Ntot can be approximately calculated

since N Ntot mm

≈ ∑ . With Nm measured for the major species , one can calculate the

effective Rayleigh scattering cross section. The combination of the Rayleigh signal with

the Raleigh cross section gives temperature. This value of T is used to give a better value

Page 86: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

68

of Nm for the major species since this term does depend on temperature. The calculation

is then iterated until convergence is achieved.

Calculations are performed separately at each spatial location (usually defined by a pixel

of a digital detector) along the horizontal line at a given height. In this experiment, line

measurements are performed at many different heights above the burner, producing a

two-dimensional composite slice through the flame. The difference scattering technique

is performed for the measurement of Rayleigh and each Raman scattering signal.

3.7.3 Setup

Shown in Figure 3.10 is the experimental setup. The second harmonic of a flashlamp

pumped, Q-switched Nd:YAG laser (10 Hz rep rate, 532 nm) is focused into a line over

the center of the burner. The output polarization direction of the laser is vertical. The

laser is Q-switched twice per flashlamp pulse producing two consecutive green laser

pulses 8 ns in duration and separated by 95 µs with an average energy of 300 mJ per

double pulse- this prevents breakdown of air. An energy meter (Laser Probe RjP-734)

records the single-shot laser energy using a reflection off the lens. The computer

downloads the laser energy for each shot, and sums them as the data are acquired for a

single image. This total laser energy is used to correct for the effects of laser energy drift

Page 87: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

69

Inte

nsif

ied

CC

DC

amer

a

Q-s

wit

ched

Nd:

YA

GL

aser

(53

2 nm

)

Spec

trog

raph

f/4

Col

ored

Gla

ss F

ilter

Len

ses

Fun

ctio

nG

ener

ator

Pol

ariz

atio

n se

lect

or

Exp

erim

enta

l Set

up

Mec

hani

cal S

hutt

er

Pow

er M

eter

Com

pute

r

Bur

ner

Lou

dspe

aker

Fig

ure

3.10

Exp

erim

enta

l set

up f

or th

e m

ulti-

exp

erim

ent.

The

pol

ariz

atio

n se

lect

ortr

ansm

its li

ght t

hat i

s ei

ther

pol

ariz

ed p

erpe

ndic

ular

or

para

llel t

o th

e la

ser

pola

riza

tion.

Page 88: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

70

over the course of an experiment. Light from the Raman and Rayleigh scattering is

collected at 90 degrees to the incident beam with a single 85 mm f/1.8 camera lens that

provides a magnification of 0.8 and focuses the light on the horizontal entrance slit of an

imaging spectrograph (Spex 270M, 0.27 m focal length, f/4, 300 groves/mm ruled

grating, 500 nm blaze angle, 300 µm entrance slit). This magnification matches the

effective f/# of the lens to the spectrograph, and also allows imaging from the flame

centerline out to ambient air. A modified liquid crystal shutter (Displaytech PV100AC)

passes the desired polarization, enabling us to measure Izz and Iyz independently. This

modified shutter consists of a crystal that is oriented such that when +/- 15 Volts is

placed across it, the polarization of incident light is rotated by 90 degrees. The light then

passes through a linear polarizer oriented to pass only vertically polarized light. Since the

light always enters the spectrograph with vertical polarization, this eliminates the need to

correct for differences in spectrograph grating efficiency for different polarizations. As a

check to the effectiveness of the technique at eliminating randomly polarized light, a

randomly polarized mercury light source is collected under the two orthogonal

polarizations. The difference between the two orthogonal polarizations eliminates the

signal down to its shot noise limit.

Upon entering the imaging spectrograph the light is dispersed into its component

wavelengths, while preserving the spatial information along the entrance slit. This spatial

Page 89: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

71

/ spectral information is collected with an 18 mm gated Gen-II image intensifier (DEP

XX1450DH) which is optically coupled with f/1.4 camera optics to a cooled slow scan

15-bit dynamic range 512 x 512 pixel CCD (Princeton Instruments TE/CCD-512 TKM).

The intensifier gain is adjusted to optimize the signal-to-noise of the data, and is gated on

for 1 µs around the laser pulse to eliminate background luminosity from the flame. The

intensifier, CCD, and laser are synchronized with digital delay generators to only allow

acquisition only when a laser pulse is present. The beam waist is measured by imaging

Rayleigh scattered light of pure methane. The slit width is shortened until there is a

noticeable change in width of the Rayleigh line image, which is observed at a slit width

of 240 µm. With a magnification of 0.8, the correspondind beam waist is determined to

be 300 µm. Spatial resolution is approximately 200 µm (horizontal) x 300 µm (depth),

and spectral resolution is approximately 3 nm. Spatial resolution is estimated by imaging

a 50 µm diam. wire oriented vertically above the burner. Spectral resolution is estimated

by closing the slit down to 50 µm and measuring the width of the methane Rayleigh line.

The resultant 512 x 256 pixel images cover approximately 170 nm in the spectral

dimension by 14 mm in the spatial dimension. The spectral coverage of each flame

image (vertical direction) was large enough to permit observation of the Rayleigh line

(@ 532 nm) simultaneously with the hydrogen Raman line (@ 683 nm), while the spatial

axis (horizontal direction) imaged a radial profile from the centerline of the jet out to the

Page 90: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

72

ambient air at all downstream locations. Each image is transferred and stored in a

computer.

3.7.4 Unforced Case Acquisition

The signal-to-noise of the Raman data is increased by integrating the scattered light

resulting from 1200 consecutive laser pulses (two per 100 ms period) on the CCD over a

60 second period. Integration of the scattered light over the 1200 consecutive laser shots

produced a Rayleigh signal just above the fuel jet center that was close to the maximum

dynamic range of the CCD (14bit, or 32767 levels), so the full dynamic range of the CCD

was exploited. Scattered light with polarizations parallel and perpendicular to the laser

polarization (Izz and Iyz) is imaged onto the CCD in separate and consecutive 60 second

periods for a given height above the burner.

In order to obtain the radial information at the different downstream locations, the burner

described in Section 3.2 is placed on a computer controlled z-translation stage. Data are

acquired at heights above the burner ranging from 2 mm to 52 mm, in steps of 0.5 mm

closer to the burner and 1 mm farther downstream, since the length scales in the flame

increased significantly at downstream locations.

Page 91: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

73

The reference signal used for carbon dioxide, carbon monoxide, water, hydrogen, and the

C2 fluorescence is nitrogen Raman from room temperature, pure air. The spectrograph

grating is rotated such that the nitrogen Raman line in clean air overlaps with a particular

Raman line for the flame data. This will equalize the optical throughput of the Raman

lines and reference data. The reference signal for oxygen and nitrogen are the O2 and N2

Raman signals in room temperature, pure air. The reference signal for methane is the

Raman signal of room temperature, pure methane. Signal integration time of the

reference data is the same (or scaled to be an equivalent integration) as the flame data.

Calibrations are recorded for each of the major species: methane, oxygen, nitrogen,

carbon dioxide, water, carbon monoxide, and hydrogen for a known quantity of each

species at a known temperature. The difference Raman signals are measured directly at

room temperature for CO2, O2, N2, CH4, and H2. The difference Raman signal for H2O is

measured in the post flame region of several premixed methane / air flat flames (φ = 0.8,

1.0, 1.2) using equilibrium calculations to calibrate the water Raman signal. The room

temperature difference Raman signal for CO is calculated from RAMSES code [Hassel

1996] by determining the ratio of difference Raman signals of CO and N2 at room

temperature (for an equal concentration), and then scaling by the difference Raman signal

of N2 from the calibration data.

Page 92: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

74

Images are obtained for an equivalent acquisition time as the flame data with the camera

shutter closed to measure detector dark current and fixed pattern detector noise.

3.7.5 Forced Case Acquisition

Since the time-varying flame is cyclical in time, one can phase lock the measurements

over many forcing cycles to measure quantities at various phases of the forcing. This is

achieved by synchronizing the function generator with the laser and intensifier with

digital delay generators. Data are acquired as described above for 5 equally spaced

phases of the forcing spanning one period of the forcing (10 ms intervals). Each phase

acquisition is done at separate times from other phase acquisitions. The measurement

gate times are determined by the intensifier gate time of 1 µs. There is no variation in

flows or flame structure over this short as time; thus the measurements are instantaneous

with respect to the forcing.

3.7.6 Processing

The images of dark current are subtracted from each spectral/spatial image of flame data

at each pixel location (x,y). A resulting sample image is shown in Figure 3.11 for

scattering intensities with polarization of the scattering parallel Izz and perpendicular Iyz to

Page 93: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

75

the laser polarization and the difference scattering intensity of these signals Ids . This data

is acquired 18 mm above the burner.

The Stokes-shifted Raman lines for the species measured in the figure are labeled, along

with the Rayleigh line. The jet centerline and coflow region are labelled in each image.

Note the significant C2 fluorescence interference in the two vertical and horizontal

scattering images. Note that the color scale is purposely chosen to saturate the stronger

transitions in order to view the weaker transitions. Each species (and Rayleigh) are

integrated over a spectral window (marked with horizontal rectangles). Spectral window

size is chosen to be large enough to account for spectral broadening of the lines due to

temperature increase, but small enough to minimize line overlap with other species (a.k.a.

crosstalk). Also note the CO2 Raman vibrational modes. The Raman transition of the

2ν2 vibrational mode is used for the CO2 concentration measurements because of less

crosstalk with the O2 line as compared with the transition of the ν1 symmetric stretch

vibrational mode. Shown in Figure 3.12 is a lineplot taken along the vertical line of the

images of Figure 3.11 marked by the white vertical rectangles. This represents the region

of maximal C2 interference in the images of Fig. 3.11. Notice the large C2 fluorescence

interference from the C2(0,1) fluorescence bandhead (near 1560 nm) in Izz and Iyz

that is eliminated in the difference scattering signal Ids. Also notice the broadband

Page 94: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

76

-400

-800

-120

0

-160

0

-200

0

-240

0

-280

0

-320

0

-360

0

-400

0

-440

0H

ydro

gen

Wat

er

Nitr

ogen

Oxy

gen

Ray

leig

h

CO

Cof

low

48

12

Jet

Cen

ter

r(m

m)

I zz

I yz

Ids

Max

Min

Cof

low

48

12

Jet

Cen

ter

r(m

m)

Cof

low

48

12

Jet

Cen

ter

r(m

m)

0

C2

fluo

resc

ence

CO

2(2ν

2)C

O2(

ν 1)

Met

hane

(ν1)

Fig

. 3.1

1 Sa

mpl

e im

ages

of

I zz,

Iyz

, Ids

(ar

bitr

ary

units

) ta

ken

18 m

m d

owns

trea

m. T

he h

oriz

onta

l whi

te r

ecta

ngle

sre

pres

ent t

he s

pect

ral w

indo

w o

ver

whi

ch th

e R

aman

spe

cies

(an

d R

ayle

igh)

sig

nals

are

inte

grat

ed. T

he c

olor

sca

lein

this

fig

ure

is c

hose

n to

sat

urat

e st

rong

tran

sitio

ns in

ord

er to

vie

w w

eake

r tr

ansi

tions

.

Shift(cm-1)

Page 95: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

77

fluorescence interference that is present in Izz and Iyz and absent in Ids. This point in the

flame is on the rich side of the flame front, and no oxygen should be present, yet a signal

persists for Ids within the spectral region of the O2 Raman transition. This unexplained

signal can also be seen in Ids between the Raman lines of O2 and CO in

Figure 3.12. There also appears to be an interfering signal in Ids on the CO2 and CO

Raman lines. These signals that persist in Ids may possibly be resonance Raman scattering

of PAH species formed just rich of the flame front. Resonance Raman occurs when the

laser frequency is near an electronic transition of a particular species, and can produce

Raman scattering intensities several orders of magnitude greater than non-resonant

Raman scattering. PAH concentrations in methane diffusion flames are on the order of

parts-per-million [Petarca 1989] which is two to three orders of magnitude under the

detection limits for non-resonant Raman transitions in this experiment.

The C2 fluorescence signal is monitored near the C2(0,1) fluorescence bandhead and

images are obtained for Izz and Iyz in the same manner as the images for the flame species.

The Cc fluorescence signal should have approximately the same spatial distribution as the

PAH resonance Raman signal. Therefore the C2 fluorescence image Izz or Iyz may be used

Page 96: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

78

Figure 3.12 Intensity spectrum (arbitrary units) taken from the region ofmaximal C2 fluorescence interference of the images in Figure 3.11 (marked

with a verticalwhite rectangle in Fig. 3.11). The light gray rectangles representthe spectral region over which the signals are integrated. Notice the largeinterference fromthe C2(0,1) fluorescence bandhead region which is present inIzz and Iyz but eliminated in Ids. Also note the residual signals in Ids near the O2

and CO Raman transitions, which is believed to be resonance Raman fromPAH species.

Lineplot of Spectral Intensities Izz, Iyz, and Ids

50

150

250

350

450

550

525 550 575 600 625 650 675

wavelength(nm)

Rayleigh CO2(ν1)CO2(2ν2)

O2 N2CO CH4 H2O H2

IzzIyzIds

C2(0,1)

PAH

inte

nsit

y(ar

bitr

ary

unit

s)

Page 97: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

79

as an empirical correction for the resonant PAH resonance Raman signals which interfere

on Ids for O2, CO2, and CO. The 'yz' fluorescence image is used as there will be less

interference from highly polarized Raman signals than in the 'zz' fluorescence image.

Shown in Figure 3.13 is Ids for O2 along with the 'yz' fluorescence image. Notice the

interference on the O2 signal just rich of the flame front. This interference region has

approximately the same spatial distribution as the 'yz' fluorescence image.

The integrated signals for each species (and Rayleigh) at each height are tiled together to

form raw spatial images for the vertically and horizontally polarized scattered light. The

difference between the two polarizations is taken for each species (and Rayleigh) spatial

image. The difference scattering images are then normalized by appropriate reference

data to correct for optical throughput, laser intensity variation, and spectral detection

efficiencies.

Page 98: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

80

Figure 3.13 Intensity profiles (arbitrary units) of Ids for oxygen Raman and Iyz

for C2 fluorescence. Notice the similar spatial profile of the interference region onthe O2 Raman signal and the C2 fluorescence profile.

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60 Ids (Oxygen)

z(m

m)

z(m

m)

r(mm)

interference

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60

r(mm)

Iyz (C2 fluorescence)

max

min

Page 99: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

81

The correction for the PAH Raman interference is made for the O2, CO2, and CO images.

The C2 fluorescence 'yz' image is scaled by an empirical factor and subtracted from each

image.

The images now are corrected for the spectral overlap of Raman signals, or crosstalk. The

spectral dispersion of the Raman lines is in the vertical direction, represented by the

variable y. The spectral window used to integrate the Raman signal of species i is

centered at yi and the integrated signal over the spectral window of species i is

I y I y I yi i i j i( ) ( ) ( )= + (3.28)

where Ii(yi) is the Raman signal from species i in the spectral window of species i, and

Ij(yi) is the Raman signal from species j in the spectral window of species i. The crosstalk

ratio

cI y T

I y Tj ij i

j j→ =

( , )

( , )0

0 (3.29)

is calculated with reference gases at room temperature T0. In general, c j i→ depends on

temperature, but a room temperature value gives a reasonable correction for most species.

Using (3.28) and (3.29), I yi i( ) can be extracted from I yi( )

I y I y c I yi i i j i j j( ) ( ) ( )= − → (3.30)

The room temperature crosstalk ratio of N2->CO undercorrects for the N2->CO crosstalk

in the hot flame regions (T=Th) just lean of the flame where CO signal should disappear

Page 100: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

82

in the absence of crosstalk. With this knowledge and the knowledge that CO signal

should disappear in the ambient region (T=Tc) in the absence of crosstalk, the N2-> CO

crosstalk can be approximated as a linear funtion of I yN( )2

I y I y I y TI y T I y TI y T I y T

I y I y TCO CO CO CO hCO CO h

N N hN N h( ) ( ) ( , )

( , ) ( , )( , ) ( , )

( ) ( , )= − − −−

−( )0

02 2

2 2 (3.31)

The most significant crosstalk is between CO2 and O2 due to the proximity of their

respective Raman lines (see Fig. 3.11). The cross talk correction for CO2-> O2 is done in

succession to the crosstalk correction for O2-> CO2 [Dibble,1987]. With the knowledge

that the O2 signal should be zero at T=Th

I y I y c I yI y T

I y T c I y TCO CO CO O CO OCO h

CO h O CO O h2 2 2 2 2 2

2

2 2 2 2

( ) ( ) ( )( , )

( , ) ( , )= −( ) −

→→

(3.32)

I y I yI y T

I y TI yO O O

O h

CO hCO CO2 2 2

2

2

2 2( ) ( )

( , )

( , )( )= −

(3.33)

3.7.7 Determination of the temperature dependence of the bandwidth factor τm(T)

In this experiment, only part of the Raman line for a specific species falls within the

spectral window for that species. As temperature increases, the Raman lines spread out

spectrally due to the population of higher rotational J states (for non-Q-branch

Page 101: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

83

transitions), so even less of the Raman signal is captured in the window. In addition, the

Raman cross sections have a temperature dependence that must be accounted for.

To obtain the temperature dependence of the bandwidth factor τm(T) for O2, CO, N2, and

H2, a numerical simulation of the spectra that incorporates the depolarization ratio for

each Raman branch (O, Q, and S), accurate spectrograph dispersion, and a realistic slit

function is used [Hassel 1996]. The slit function is obtained by imaging the Rayleigh

scattered line of methane, since the methane Rayleigh line should be the thinnest spectral

line measured, as methane does not posses a rotational Raman spectrum. The Raman

spectra at various temperatures are integrated over their respective spectral windows used

in the flame measurements, and τm(T) is determined. For CO2, spectral line data [Miles

1996] is convolved with the slit function to get the CO2 Raman spectra at various

temperatures. This spectrum is integrated over the CO2 spectral window, and τm(T) is

determined. τm(T) for H2O and CH4 is approximated by the simple Boltzmann term of

(2.12) 11− −( )−

exp( / )hω k TB .

Page 102: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

84

Figure 3.14 Experimental and simulated Raman spectra for nitrogen atT = 300 K and T = 2000 K. The simulated nitrogen spectra are obtainedusing RAMSES code.

00.10.20.30.40.50.60.70.80.9

1

606.00 607.00 608.00 609.00 610.00 611.00 612.00 613.00 614.00

wavelength (nm)

Inte

nsit

y (A

rb. u

nits

)

Simulated and measured nitrogen Raman spectrum at 300 K and 2000 K

experiment, 2000 Ksimulated, 2000 K

experiment, 300 Ksimulated, 300 K

Page 103: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

85

Figure 3.15 Difference Raman signal temperature dependence. Signals arenormalized to unity at 300 K. Signals are obtained from simulationsof Raman spectra, except for CH4 and H2O which are given a Boltzmanntemperature dependence.

Difference Raman Signal Temperature Dependence

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

300 800 1300 1800

Temperature(K)

O2N2CH4H2OCO

H2CO2 (2ν2 )

CO2 (ν1 )

I ds(

arbi

trar

y un

its)

Page 104: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

86

The experimental and simulated Raman spectra for nitrogen are compared at

temperatures of 300K and 2000K.(Fig. 3.14) Excellent agreement is found between the

observed and calculated spectra. Figure 3.15 shows the computed temperature

dependencies of the difference Raman signals of each species. This data has been

normalized to unity at 300 K, and the spectral windows used to evaluate the simulated

temperature dependence are identical to the windows used in the experiment. Carbon

dioxide has the largest temperature correction, corresponding to an increase in the signal

by over 160% at 2000 K. Hydrogen and oxygen show negative trends with temperature,

indicating that the Raman signal is spreading out of the spectral window for these

species.

3.7.8 Temperature and Species Concentration Calculation

Calculations are performed separately at each pixel location (x,y) of the images. As a

starting point, one sets τm T( ) = 1 for all species m. With Ids,Ram,m determined from the

experiment for all major species and appropriate reference data Ids,Ram,n(ref), one gets a

first order calculation of each species number density from (3.27)

NI

I refI ref

IN ref

dd

T

dd

Tm

ds Ram m

ds Ram nn

ds Ram nref

ds Ram m

=

, ,

, ,

, ,

, ,

( )( )

( )( )

( )

0

00

σ

σΩ

Ω

τm(T)

Page 105: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

87

(The ratio

dd

T

dd

T

ds Ram nref

ds Ram m

σ

σΩ

Ω

, ,

, ,

( )

( )0

is calculated from calibration data.) With these values for

Nm, one now calculates N Ntot mm

≈ ∑ . One then calculates the effective difference

Rayleigh cross section dd N

dd

Neff ds Ray tot m zz Ray

m Ray mm

σ σ ρΩ Ω

, , , ,,( )

11 , where one

knows the 'zz' Rayleigh cross sections, a first order number density of each species (and

Ntot), and the Rayleigh depolarizations. (This same calculation is performed for

dd

refeff ds Ray

σΩ

, ,

( ) .) With Ids,Ray determined from the experiment along with appropriate

reference data Ids,Ray(ref), one calculates a temperature from (3.24)

TI ref

II

I ref

dd

dd

refT ref

aI

ds Ray

ds Ray

eff ds Ray

eff ds Ray

T

ds Ray

=

=,

,

, ,

, ,

'

,

( )

( ) ( )( )0

0

σ

σΩ

Ω

. With this temperature, one

calculates τm T( ) for each species. This gives a more refined estimate of species number

density in (3.27). Then the process is iterated until convergence of the number densities

and temperature is achieved. Final values of species number densities are converted into

mole fractions.

3.8 Discussion on experimental techniques

Page 106: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

88

3.8.1 Effectiveness of two scalar technique

As previously mentioned there is an interference signal on the fuel Raman images that

occurs just to the rich side of the flame. It is identified as an interference signal since it

develops downstream of the burner, where the fuel has just disappeared; therefore it is

not methane Raman signal. The likely source of this interference is fluorescence from C2

species. As seen in Figure 3.12 from the flame spectrum, there is a significant

fluorescence signal from the C2(0,1) bandhead at 560 nm. This signal can be picked up

through the 630 nm interference filter, although at a modest spectral throughput. Also

contributing to this signal is the broadband C2 and PAH fluorescence signal also seen in

Fig. 3.12.

Although data was only taken for Raman images at a maximum height downstream of 20

mm, the interference signal starts to be seen in the data at around 15 mm (see Figure 3.9).

The fuel Raman data in these locations is therefore unreliable. In addition, there is the

fuel Raman signal downstream of 20 mm that is not recorded. Although this signal is

small (relative to the fuel Raman signal at burner inlet), it will still significantly

contribute to the calculation of mixture fraction and temperature.

Page 107: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

89

Figure 3.16 shows a comparison of corrected methane Raman images obtained with the

two experimental techniques for a particular phase of the 30% modulation case. With the

exception of the interference region, there is good agreement of the two profiles less than

20 mm downstream. The methane Raman signal in the multi-species measurement is

shown to extend well beyond 20 mm downstream. Therefore the mixture fraction and

temperature for the two scalar data have unreliable values in regions above the 20 mm

height where methane is present but the methane Raman signal is not recorded.

To understand the effectiveness of the two scalar algorithm in calculating mixture

fraction and temperature, the two scalar calculation is performed on the computational

data. The temperature and mixture fraction derived from the two scalar method are

compared to the mixture fraction ξCHO in ( 3.1) (based on the mass fractions of carbon,

hydrogen, and oxygen) and computational temperature in the lineplots of Figure 3.17. In

the axial centerline plot of mixture fraction the curves show good agreement upstream,

but the curves separate downstream of 20 mm. The two scalar mixture fraction in this

region appears to be artificially low (by 0.05 at peak deviation). The range of mixture

fractions where the curves deviate is 0.07 to 0.2 (0.1 is the stoichiometric mixture

fraction). This range is also where the mixture fraction curves deviate in the radial

lineplot of mixture fraction taken 9 mm downstream of the burner. The centerline

Page 108: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

90

Figure 3.16 Methane Raman intensity profiles (arbitrary units)from experiments. Notice the interference (white box) in thetwo scalar measurement.

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60Difference Raman Two Scalar

Methane Raman Signal

z(m

m)

r(mm)r(mm)

z(m

m)

min

max

Page 109: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

91

Figure 3.17 Two-scalar calculation of mixture fraction and temperature based oncomputational data, compared to ξCHO and computational temperature based on

computations.

Mixture fraction- radial

00.10.20.30.40.50.60.70.80.9

0 2.5 5 7.5 10

radial distance(mm)

two-scalar

Mixture fraction- centerline

0

0.2

0.4

0.6

0.8

1

1.2

0 20 40 60

height(mm)

two-scalar

ξ CHO ξ CHO

Tem

pera

ture

(K

)

Tem

pera

ture

(K

)

Mix

ture

fra

ctio

n

Mix

ture

fra

ctio

n

Temperature-radial

0200400600800

100012001400160018002000

0 2.5 5 7.5 10

radial distance(mm)

comptwo-scalar

Temperature-centerline

300

550

800

1050

1300

1550

1800

0 20 40 60

height(mm)

comptwo-scalar

Page 110: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

92

temperature lineplot shows good agreement upstream, but the curves separate

downstream of 20 mm (as does the mixture fraction centerline plots). The two scalar

temperature appears to be artificially low (by 100 K at peak temperature). The radial

temperature lineplot taken 9 mm downstream, however, shows good agreement at each

radial flame position.

The inaccuracies of the two scalar formulation of temperature and mixture fraction can be

attributed to the assumption of single step chemistry. Intermediate species will affect the

temperature and mixture fraction near stoichiometric. The advantage of this technique is

the determination of temperature, fuel concentration, and mixture fraction from only two

scalar measurements.

3.8.2 Effectiveness of difference Raman technique

One needs to determine how effective the difference Raman/ multi-species technique is in

discriminating the relatively weak Raman signals from the large fluorescence

interferences. If the C2 fluorescence is determined to be randomly polarized over the gate

time of the image intensifier, one should be able to eliminate the fluorescence signal

down to shot noise levels of the fluorescence.

Page 111: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

93

The spatial and spectral region of highest interference from the C2 fluorescence in the

flame is determined, and the time dependence of the polarization of this signal is

investigated. The camera-intensifier system is replaced with a photomultiplier tube

(PMT) (Hamamatsu R928). With an exit slit placed on the spectrograph, light only in the

spectral region of interest is passed to the PMT. The fluorescence signal is digitized on a

digital oscilloscope (Techtronix, 500MS/s sampling rate), terminated with 50 ohms. The

time resolved C2 fluorescence signal intensities Izz and Iyz are recorded independently.

PMT and oscilloscope response time is approximately 2 ns. The time decay of the laser

pulse (1/e intensity point) τ laser = 4ns, and the time decay of the fluorescence τC210= ns.

The peak signal and time decay of each signal are almost identical. When these signals

are integrated over their lifetimes and are subtracted from one another, the result is within

noise levels of the fluorescence signals. In conclusion, the C2 fluorescence is randomly

polarized over the lifetime of the fluorescence. Therefore the difference Raman technique

reduces the C2 fluorescence interference on Raman data to shot noise of the fluorescence.

To determine the effectiveness of the difference Raman technique one must compare the

noise of the C2 fluorescence interference on the difference Raman signals. The noise is

defined as the rms of a signal over a region where the signal is approximately constant.

The weakest Raman signal, which also has significant interference from C2 fluorescence,

Page 112: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

94

is CO. Let I x yyz C CO, ( , )2 → represent the C2 fluorescence interference on the raw 'yz' image

of CO (Iyz,CO(x,y)). Let σyz C CO dx dy, ( , )2 → represent the noise of the C2 fluorescence

interference calculated in the small region (dx,dy) in the 'yz' CO image. Since the peak

intensity of Ids,CO(x,y) is 90 counts (arbitrary units) and the Raman depolarization of CO

ρCO = 0.038, then Iyz,CO(x,y) should have negligible CO signal. Thus,

I x y I x yyz C CO yz CO, ,( , ) ( , )2 → ≈ . Note that σ σds C CO yz C CO, ,2 2

2→ →≈ (due to the subtraction of

one noisy signal from another). One finds that I x y x yds CO yz C CO, ,( , ) ( , )> −>22

σ for all points

(x,y) containing CO, and therefore the CO difference Raman signal is effectively

discriminated from the C2 fluorescence noise.

3.9 Comparison of experimental and computational profiles

3.9.1 Steady Flame

Steady flame profiles of temperature, CO2 , H2O, CO, N2, O2, H2, and CH4 mole fractions

are shown in Figures 3.18-3.25 for the multi-species experiment and computations.

Profiles of temperature, CH4, CO2, N2, and O2, and H2O show very good agreement in

terms of spatial variation in flame structure and peak concentrations. The only apparent

difference is the difference in flame lift-off height. Because of the extremely small signal,

Page 113: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

95

Figure 3.18 Measured (multi-species technique) and computed

temperature (degrees Kelvin) for the steady flame.

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60Experimental Numerical

Temperature

z(m

m)

r(mm)r(mm)300 300

1950 1950

z(m

m)

Page 114: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

96

Figure 3.19 Measured (multi-species technique) and computedcarbon dioxide mole fractions for the steady flame.

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60Experimental Numerical

Carbon Dioxide

z(m

m)

r(mm)r(mm)

0.097

0

0.097

0

z(m

m)

Page 115: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

97

Figure 3.20 Measured (multi-species technique) and computed watermole fractions for the steady flame.

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60Experimental Numerical

Water

z(m

m)

r(mm)r(mm)

0.162

0

0.162

0

z(m

m)

Page 116: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

98

Figure 3.21 Measured (multi-species technique) and computed carbonmonoxide mole fractions for the steady flame.

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60Experimental Numerical

Carbon Monoxide

z(m

m)

r(mm)r(mm)

0.042 0.042

0 0

z(m

m)

Page 117: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

99

Figure 3.22 Measured (multi-species technique) and computedhydrogen mole fractions for the steady flame.

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60Experimental Numerical

Hydrogen

z(m

m)

r(mm)r(mm)

0.023

0

0.023

0

z(m

m)

Page 118: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

100

Figure 3.23 Measured (multi-species technique) and computedmethane mole fractions for the steady flame.

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60Experimental Numerical

Methane

z(m

m)

r(mm)r(mm)

0 0

0.65 0.65

z(m

m)

Page 119: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

101

Figure 3.24 Measured (multi-species technique) and computednitrogen mole fractions for the steady flame.

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60Experimental Numerical

Nitrogen

z(m

m)

r(mm)r(mm)

0.79

0.35 0.35

0.79

z(m

m)

Page 120: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

102

Figure 3.25 Measured (multi-species technique) and computedoxygen mole fractions for the steady flame.

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60

0 4 8-8 -4

5

10

15

20

25

30

35

40

45

50

55

60Experimental Numerical

Oxygen

z(m

m)

r(mm)r(mm)

0.210 0.210

0 0

z(m

m)

Page 121: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

103

Figure 3.26 Radial and centerline plots of temperature, water, and carbondioxide for steady flame experiments (multi-species technique) andcomputations.The radial plots are taken 10.5 mm above burner for thecomputations and 9 mm above burner for the experiment to account for lift-off

differences.

Temperature (K)-center

0

500

1000

1500

2000

2500

0 10 20 30 40 50 60

height (mm)

Temperature (K)-radial

0

500

1000

1500

2000

2500

0 2 4 6 8 10

radial (mm)

Carbon Dioxide (mole fraction)-radial

0

0.02

0.04

0.06

0.08

0.1

0 2 4 6 8 10

radial (mm)

Water (mole fraction)-center

0

0.05

0.1

0.15

0.2

0.25

0 10 20 30 40 50 60

height (mm)

Water (mole fraction)-radial

0

0.05

0.1

0.15

0.2

0 2 4 6 8 10

radial (mm)

Carbon Dioxide (mole fraction)-center

0

0.02

0.04

0.06

0.08

0.1

0 10 20 30 40 50 60

height (mm)

experiment computations

Page 122: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

104

the CO and H2 profiles are quite noisy. They are smoothed by using a gaussian smoothing

technique (σx= σy=1.71 pixels, smoothed over a 7x7 pixel area) to improve the signal-to

noise-ratio. The profile comparison for CO is remarkably good considering the relatively

low signal-to-noise of the CO image compared to the other species. The experimental

hydrogen profile is not as good a comparison to computations, but the peak signal and

general shape is the same in both the experiment and computations. Shown in Figure 3.26

are lineplots of radial and centerline lineplots of CO2, H2O, and temperature for the multi-

species experiment and computations. The radial profiles are taken at a height of 9 mm

above the burner. The radial plots show good agreement in terms of spatial variation and

peak values. The centerline plots show good agreement in spatial variation and peak

values as well. The most apparent difference is the flame lift-off height between the

computations and experiment.

3.9.2 Forced Flame

Figure 3.27 shows profiles of temperature for the multi-species experiment and the two

scalar experiment for the case of 30% flow modulation. Images are taken at five equally

spaced phases spanning one period of forcing, where the figure labeled "d" is the phase

where the centerline fuel tube exit velocity is at its minimum over the forcing period. The

spatial region where interference is observed on the two scalar methane Raman signals as

Page 123: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

105

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)

0.04

40 0

6080100

120 0

0.01

0.02

0.03

20

0.05

0.06

stea

dy f

lam

e

Cen

terl

ine

fuel

exi

t vel

ocity

vs.

tim

e

v(cm/s)

time(

s)d

cb

ae

(iii)

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)

(a)

(b)

(c)

(d)

(e)

(a)

(b)

(c)

(d)

(e)

(ii)

stea

dy

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

300

K

(i)

1960

K

Fig

ure

3.27

Tem

pera

ture

pro

file

s fo

r 30

% m

odul

atio

n (i

n de

gree

s K

).

(i

) M

ulti-

spec

ies

mea

sure

men

t. (i

i) T

wo

scal

ar. (

iii)

Cen

terl

ine

fuel

exit

velo

city

vs.

tim

e. P

hase

s w

here

dat

a is

take

n ar

e m

arke

d by

a,b,

c,d,

e on

the

grap

h an

d co

rres

pond

to th

e im

ages

labe

lled

likew

ise.

Stea

dy f

lam

e im

age

is s

how

n on

the

far

left

. The

ste

ady

flam

e

fu

el e

xit v

eloc

ity is

indi

cate

d in

the

grap

h. U

nrel

iabl

e da

ta r

egio

n fr

om

tw

o sc

alar

ram

an d

ata

is c

onta

ined

in th

e da

rk b

lue

boxe

s.

Page 124: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

106

Figure 3.28 Temperature lineplots of the two-scalar and multispecies technique for 30%flow modulation. Radial data is taken 9 mm above burner.

phase

a

b

c

d

e

Temperature(K)-Phase dcenterline

0

500

1000

1500

2000

2500

0 20 40 60

height(mm)

two-scalarmulti-species

Temperature(K)-Phase ccenterline

0

500

1000

1500

2000

0 20 40 60

height(mm)

two-scalarmulti-species

Temperature(K)-Phase bcenterline

0

500

10001500

2000

2500

0 20 40 60

height(mm)

two-scalarmulti-species

Temperature(K)-Phase bradial

0

500

1000

1500

2000

0 5 10

radial(mm)

two-scalarmulti-species

Temperature(K)-Phase cradial

0

500

1000

1500

2000

0 5 10

radial(mm)

two-scalarmulti-species

Temperature(K)-Phase dradial

0

500

1000

1500

2000

0 5 10

radial(mm)

two-scalarmulti-species

Temperature-Phase acenterline

0

500

1000

1500

2000

0 20 40 60

height(mm)

two-scalarmulti-species

Temperature(K)-Phase aradial

0

500

1000

1500

2000

0 5 10

radial(mm)

two-scalar

multi-species

Temperature(K)-Phase ecenterline

0

500

1000

1500

2000

0 20 40 60

height(mm)

two-scalarmulti-species

Temperature(K)-Phase eradial

0

500

1000

1500

2000

0 5 10

radial(mm)

two-scalarmulti-species

Page 125: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

107

well as the region where the methane Raman signal is present in the multi-species

measurement (downstream of 20 mm) but not recorded for the two scalar experiment is

blocked out in the images with a dark blue box. The oval region blocked out in phase "a"

for the multi-species temperature is a region where there is large elastic scattering

interference on the Rayleigh image. The most likely cause of this interference is

scattering from soot particles. The experimental images indicate similar modulations in

flame structure over the forcing period, with slightly less of a modulation in the two

scalar experiment, as is most apparent in phases (d) and (e). This is a good check on the

reproducibility of the forcing by the loudspeaker used. The profiles show good agreement

in lift off height, flame height, general flame shape, and absolute peak temperatures.

Plotted in Fig. 3.28 is the axial centerline temperature as well as the radial temperature 9

mm downstream. The broken part of the centerline curves for the two-scalar experiment

correspond to the blocked out regions in Figure 3.27. Radial temperatures 9 mm

downstream agree well, especially near stoichiometric and on the lean side. The major

difference in the radial plots is along the centerline, where the temperature is consistently

lower in the multi-species experiment. The centerline plots show good agreement in peak

temperature and temperature variation near stoichiometric, but the sharp rise in

temperature on the rich side of the flame front (along the centerline) occurs significantly

farther downstream in the multi-species experiment, which is most apparent in phase "c"

and "d". There is possibly some elastic scattering interference on the Rayleigh data in the

Page 126: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

108

multi-species experiment in the region where the fuel disappears, which is not corrected

for in the Rayleigh images. This will produce an artificially high Rayleigh intensity and

thus an artificially low temperature in this region.

Figures 3.29 and 3.30 show mixture fraction plots and profiles for the multi-species and

two-scalar experiments. Shown in red outline is the stoichiometric contour. Profiles and

lineplots show good agreement. The major difference in the two techniques is seen along

the centerline near stoichiometric. The mixture fraction for the two-scalar experiment is

consistently lower than the multi-species mixture fraction in this region. The discrepancy

here is due to the assumptions of one-step chemistry in the two-scalar formulation of

mixture fraction. (This is seen in the comparison of mixture fraction formulations on the

computations in Figure 3.15.)

Page 127: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

109

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)

0.04

40 0

6080100

120 0

0.01

0.02

0.03

20

0.05

0.06

stea

dy f

lam

e

Cen

terl

ine

fuel

exi

t vel

ocity

vs.

tim

e

v(cm/s)

time(

s)d

cb

ae

(iii)

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)

(a)

(b)

(c)

(d)

(e)

(a)

(b)

(c)

(d)

(e)

(ii)

(i)

stea

dy

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

01

Fig

ure

3.29

Mix

ture

fra

ctio

n pr

ofile

s fo

r 30

% m

odul

atio

n.

(i

) M

ulti-

spec

ies

mea

sure

men

t. (i

i) T

wo

scal

ar. (

iii)

Cen

terl

ine

fuel

exit

velo

city

vs.

tim

e. P

hase

s w

here

dat

a is

take

n ar

e m

arke

d by

a,b,

c,d,

e on

the

grap

h an

d co

rres

pond

to th

e im

ages

labe

lled

likew

ise.

Stea

dy f

lam

e im

age

is s

how

n on

the

far

left

. The

ste

ady

flam

e

fu

el e

xit v

eloc

ity is

indi

cate

d in

the

grap

h. U

nrel

iabl

e da

ta r

egio

n fr

om

tw

o sc

alar

ram

an d

ata

is c

onta

ined

in th

e da

rk b

lue

boxe

s.

Page 128: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

110

Figure 3.30 Two scalar and multi-species mixture fraction plots for30% flow modulation. Radial data is taken 9 mm above burner.

Mixture fraction-Phase acenterline

0

0.20.4

0.60.8

1

0 20 40

height(mm)

two-scalar

Mixture fraction-Phase aradial

00.20.40.60.8

1

0 5 10

radial distance(mm)

two-scalar

Mixture fraction-Phase e centerline

0

0.2

0.4

0.6

0.8

1

0 10 20 30 40

height(mm)

two-scalar

Mixture fraction-Phase eradial

0

0.2

0.4

0.6

0.8

1

0 5 10

radial distance(mm)

two-scalar

Mixture fraction-Phase cradial

00.20.40.60.8

11.2

0 5 10

radial distance(mm)

two-scalar

phase

a

b

c

d

e

Mixture fraction-Phase b centerline

00.20.40.60.8

11.2

0 20 40 60

height(mm)

two-scalar

Mixture fraction-Phase c centerline

0

0.2

0.4

0.6

0.8

1

0 20 40 60

height(mm)

two-scalar

Mixture fraction-Phase dcenterline

0

0.2

0.4

0.6

0.8

1

0 20 40 60

height(mm)

two-scalar

Mixture fraction-Phase bradial

00.20.40.60.8

11.2

0 2.5 5 7.5 10

radial distance(mm)

two-scalar

Mixture fraction-Phase dradial

0

0.2

0.4

0.6

0.8

0 2.5 5 7.5 10

radial distance(mm)

two-scalar

ξ CHO

ξ CHO

ξ CHO

ξ CHO

ξ CHO

ξ CHO

ξ CHO

ξ CHO

ξ CHO

ξ CHO

Page 129: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

111

Figures 3.31-3.36 show profiles and lineplots of temperature, CO2, and H2O for the

computations and the multi-species experiment for the case of 30% modulation.

Temperature profiles indicate a greater modulation for the experiment than observed in

the computations. Phase "b" in the experiment indicates a region where the flame begins

to "pinch off" at a height of 35 mm, a phenomenon which is not observed in phase "b" of

the computations. Temperature lineplots indicate good agreement in peak centerline

temperatures and in flame lift-off height. The difference in the variation of centerline

temperatures also indicates the more modest computational modulation. Radial

temperature profiles indicate good agreement of temperature variation, but peak radial

temperatures are lower in the experiment. Profiles and plots of both CO2 and H2O

indicate good agreement in peak values, and centerline and radial variation.

Figures 3.37-3.42 show profiles and lineplots of temperature, CO2, and H2O for the

computations and the multi-species experiment for the case of 50% modulation. As in the

30% modulation case, the computations show a significantly smaller modulation than the

experiment. One observes a significant degree of "pinching off" of the flame in phases

Page 130: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

112

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)30

0 K

1950

K

0.04

40

0

6080100

120 0

0.01

0.02

0.03

20

0.05

0.06

stea

dy f

lam

e

Cen

terl

ine

fuel

exi

t vel

ocity

vs.

tim

e

v(cm/s)

time(

s)d

cb

ae

(iii)

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)

(a)

(b)

(c)

(d)

(e)

(a)

(b)

(c)

(d)

(e)

(ii)

(i)

stea

dy

z(mm)

Fig

ure

3.31

Tem

pera

ture

pro

file

s fo

r 30

% m

odul

atio

n (i

n de

gree

s K

).

(i

) M

ulti-

spec

ies

mea

sure

men

t. (i

i) C

ompu

tatio

ns. (

iii)

Cen

terl

ine

fuel

exit

velo

city

vs.

tim

e. P

hase

s w

here

dat

a is

take

n ar

e m

arke

d by

a,b,

c,d,

e on

the

grap

h an

d co

rres

pond

to th

e im

ages

labe

lled

likew

ise.

Stea

dy f

lam

e im

age

is s

how

n on

the

far

left

. The

ste

ady

flam

e

fu

el e

xit v

eloc

ity is

indi

cate

d in

the

grap

h. U

nrel

iabl

e da

ta r

egio

n fr

om

tw

o sc

alar

ram

an d

ata

is c

onta

ined

in th

e da

rk b

lue

boxe

s.

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

stea

dy

r(m

m)

r(m

m)

Page 131: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

113

Figure 3.32 Temperature lineplots (degrees K) of experiments (multi-species) andcomputation for 30% flow modulation. Radial data is taken 9 mm above burner.

Temperature(K)-radial

0

500

1000

1500

2000

2500

0 5 10

radial distance(mm)

compexp

Temperature(K)-radial

0

500

1000

1500

2000

2500

0 5 10

radial distance(mm)

compexp

a

b

c

d

e

phase

Temperature (K)-centerline

0

500

1000

1500

2000

2500

0 20 40 60 80

height(mm)

compexp

Temperature(K)-radial

0

500

1000

1500

2000

2500

0 5 10

height(mm)

compexp

Temperature(K)-centerline

0

500

1000

1500

2000

2500

0 20 40 60 80

height(mm)

compexp

Temperature(K)-centerline

0

500

1000

1500

2000

2500

0 20 40 60 80

height(mm)

compexp

Temperature(K)-radial

0

500

1000

1500

2000

2500

0 5 10

radial distance(mm)

compexp

Temperature(K)-centerline

0

500

1000

1500

2000

2500

0 20 40 60 80

height(mm)

compexp

Temperature(K)-radial

0

500

1000

1500

2000

2500

0 5 10

radial distance(mm)

compexp

Temperature(K)-centerline

0

500

1000

1500

2000

2500

0 20 40 60 80

height(mm)

comp

Page 132: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

114

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)0

0.09

6 (iii)

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)

(a)

(b)

(c)

(d)

(e)

(a)

(b)

(c)

(d)

(e)

(ii)

(i)

stea

dy

z(mm)

z(mm)

z(mm)

z(mm)

Fig

ure

3.33

Car

bon

diox

ide

prof

iles

for

30%

mod

ulat

ion

(mol

e fr

actio

n).

(i)

Mul

ti-sp

ecie

s m

easu

rem

ent.

(ii)

Com

puta

tions

. (iii

) C

ente

rlin

e fu

el

ex

it ve

loci

ty v

s. ti

me.

Pha

ses

whe

re d

ata

is ta

ken

are

mar

ked

by

a,

b,c,

d,e

on th

e gr

aph

and

corr

espo

nd to

the

imag

es la

belle

d lik

ewis

e.

St

eady

fla

me

imag

e is

sho

wn

on th

e fa

r le

ft. T

he s

tead

y fl

ame

fuel

exi

t vel

ocity

is in

dica

ted

in th

e gr

aph.

Unr

elia

ble

data

reg

ion

from

two

scal

ar r

aman

dat

a is

con

tain

ed in

the

dark

blu

e bo

xes.

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

stea

dy

0.04

40

0

6080100

120 0

0.01

0.02

0.03

20

0.05

0.06

stea

dy f

lam

e

Cen

terl

ine

fuel

exi

t vel

ocity

vs.

tim

e

v(cm/s)

time(

s)d

cb

ae

r(m

m)

r(m

m)

Page 133: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

115

Figure 3.34 Carbon dioxide mole fraction lineplots of experiments (multi-species) andcomputations for 30% flow modulation. Radial data is taken 9 mm above burner.

a

b

c

d

e

phase

Carbon dioxide mole fraction-radial

0

0.02

0.04

0.06

0.08

0.1

0 5 10

radial distance(mm)

compexp

Carbon dioxide mole fraction-centerline

0

0.02

0.04

0.06

0.08

0.1

0.12

0 20 40 60 80

height(mm)

compexp

Carbon dioxide mole fraction-radial

0

0.02

0.04

0.06

0.08

0.1

0 5 10

radial distance(mm)

compexp

Carbon dioxide mole fraction-centerline

00.020.040.060.08

0.10.12

0 20 40 60 80

height(mm)

compexp

Carbon dioxide mole fraction-radial

0

0.02

0.04

0.06

0.08

0.1

0 5 10

radial distance(mm)

compexp

Carbon dioxide mole fraction-radial

00.020.040.060.08

0.1

0 5 10

radial distance(mm)

compexp

Carbon dioxide mole fraction-centerline

00.020.040.060.08

0.10.12

0 20 40 60 80

height(mm)

compexp

Carbon dioxide mole fraction-centerline

00.020.040.060.08

0.10.12

0 20 40 60 80

height(mm)

compexp

Carbon dioxide mole fraction-radial

00.02

0.040.06

0.080.1

0 5 10

radial distance(mm)

compexp

Carbon dioxide mole fraction-centerline

00.020.040.060.08

0.10.12

0 50 100

height(mm)

compexp

Page 134: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

116

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)0

0.16

0

(iii)

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)

(a)

(b)

(c)

(d)

(e)

(a)

(b)

(c)

(d)

(e)

(ii)

(i)

stea

dy

z(mm)

z(mm)

z(mm)

z(mm)

Fig

ure

3.35

Wat

er p

rofi

les

for

30%

mod

ulat

ion

(mol

e fr

actio

n).

(i)

Mul

ti-sp

ecie

s m

easu

rem

ent.

(ii)

Com

puta

tions

. (iii

) C

ente

rlin

e fu

el

ex

it ve

loci

ty v

s. ti

me.

Pha

ses

whe

re d

ata

is ta

ken

are

mar

ked

by

a,

b,c,

d,e

on th

e gr

aph

and

corr

espo

nd to

the

imag

es la

belle

d lik

ewis

e.

St

eady

fla

me

imag

e is

sho

wn

on th

e fa

r le

ft. T

he s

tead

y fl

ame

fuel

exi

t vel

ocity

is in

dica

ted

in th

e gr

aph.

Unr

elia

ble

data

reg

ion

from

two

scal

ar r

aman

dat

a is

con

tain

ed in

the

dark

blu

e bo

xes.

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

stea

dy

0.04

40

0

6080100

120 0

0.01

0.02

0.03

20

0.05

0.06

stea

dy f

lam

e

Cen

terl

ine

fuel

exi

t vel

ocity

vs.

tim

e

v(cm/s)

time(

s)d

cb

ae

r(m

m)

r(m

m)

Page 135: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

117

Figure 3.36 Water mole fraction lineplots of experiments (multi-species) and computations for 30% flow modulation. Radial data is taken 9 mm above burner.

a

b

c

d

e

phase

Water mole fraction-radial

0

0.05

0.1

0.15

0.2

0 5 10

radial distance(mm)

compexp

Water mole fraction-centerline

0

0.05

0.1

0.15

0.2

0.25

0 20 40 60 80

height(mm)

compexp

Water mole fraction-radial

0

0.05

0.1

0.15

0.2

0 5 10

radial distance(mm)

compexp

Water mole fraction-centerline

0

0.05

0.1

0.15

0.2

0.25

0 20 40 60 80

height(mm)

compexp

Water mole fraction-centerline

0

0.050.1

0.150.2

0.25

0 20 40 60 80

height(mm)

compexp

Water mole fraction-radial

0

0.05

0.1

0.15

0.2

0 5 10

radial distance(mm)

compexp

Water mole fraction-centerline

0

0.05

0.1

0.15

0.2

0 20 40 60 80

height(mm)

compexp

Water mole fraction-radial

0

0.05

0.1

0.15

0.2

0 5 10

radial distance(mm)

compexp

Water mole fraction-centerline

0

0.05

0.1

0.15

0.2

0 20 40 60 80

height(mm)

compexp

Water mole fraction-radial

0

0.05

0.1

0.15

0.2

0 5 10

radial distance(mm)

compexp

Page 136: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

118

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)30

0 K

1950

K

0.04

40

0

6080100

120 0

0.01

0.02

0.03

20

0.05

0.06

stea

dy f

lam

e

Cen

terl

ine

fuel

exi

t vel

ocity

vs.

tim

e

v(cm/s)

time(

s)d

cb

ae

(iii)

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)

(a)

(b)

(c)

(d)

(e)

(a)

(b)

(c)

(d)

(e)

(ii)

(i)

stea

dy

z(mm)

z(mm)

z(mm)

Fig

ure

3.37

Tem

pera

ture

pro

file

s fo

r 50

% m

odul

atio

n (i

n de

gree

s K

).

(i

) M

ulti-

spec

ies

mea

sure

men

t. (i

i) C

ompu

tatio

ns. (

iii)

Cen

terl

ine

fuel

exit

velo

city

vs.

tim

e. P

hase

s w

here

dat

a is

take

n ar

e m

arke

d by

a,b,

c,d,

e on

the

grap

h an

d co

rres

pond

to th

e im

ages

labe

lled

likew

ise.

Stea

dy f

lam

e im

age

is s

how

n on

the

far

left

. The

ste

ady

flam

e

fu

el e

xit v

eloc

ity is

indi

cate

d in

the

grap

h. U

nrel

iabl

e da

ta r

egio

n fr

om

tw

o sc

alar

ram

an d

ata

is c

onta

ined

in th

e da

rk b

lue

boxe

s.

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

stea

dy

r(m

m)

r(m

m)

Page 137: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

119

Figure 3.38 Temperature lineplots (degrees K) of experiments (multi-species) and computations for 50% flow modulation. Radial data is taken 9 mm above burner.

a

b

c

d

e

Temperature(K)-radial

0

500

1000

1500

2000

0 5 10

radial distance(mm)

compexp

Temperature(K)-radial

0

500

1000

1500

2000

0 5 10

radial distance(mm)

compexp

Temperature(K)-radial

0

500

1000

1500

2000

0 2 4 6 8 10

radial distance(mm)

compexp

Temperature(K)-centerline

0

500

1000

1500

2000

2500

0 20 40 60 80

height(mm)

compexpc

phase

Temperature(K)-radial

0

500

1000

1500

2000

0 5 10

radial distance(mm)

compexp

Temperature(K)-centerline

0

500

1000

1500

2000

2500

0 20 40 60 80

height(mm)

compexp

Temperature(K)-radial

0

500

1000

1500

2000

2500

0 5 10

radial distance(mm)

compexp

Temperature(K)-centerline

0

500

1000

1500

2000

2500

0 20 40 60 80

height(mm)

compexp

Temperature(K)-centerline

0

500

1000

1500

2000

2500

0 20 40 60 80

height(mm)

compexp

Temperature(K)-centerline

0

500

1000

1500

2000

2500

0 20 40 60 80

height(mm)

compexp

Page 138: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

120

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)0

0.04

40

0

6080100

120 0

0.01

0.02

0.03

20

0.05

0.06

stea

dy f

lam

e

Cen

terl

ine

fuel

exi

t vel

ocity

vs.

tim

e

v(cm/s)

time(

s)d

cb

ae

(iii)

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)

(a)

(b)

(c)

(d)

(e)

(a)

(b)

(c)

(d)

(e)

(ii)

(i)

stea

dy

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

stea

dy

r(m

m)

r(m

m)

Fig

ure

3.39

Car

bon

diox

ide

prof

iles

for

50%

mod

ulat

ion

(mol

e fr

actio

n).

(i)

Mul

ti-sp

ecie

s m

easu

rem

ent.

(ii)

Com

puta

tions

. (iii

) C

ente

rlin

e fu

el

ex

it ve

loci

ty v

s. ti

me.

Pha

ses

whe

re d

ata

is ta

ken

are

mar

ked

by

a,

b,c,

d,e

on th

e gr

aph

and

corr

espo

nd to

the

imag

es la

belle

d lik

ewis

e.

St

eady

fla

me

imag

e is

sho

wn

on th

e fa

r le

ft. T

he s

tead

y fl

ame

fuel

exi

t vel

ocity

is in

dica

ted

in th

e gr

aph.

Unr

elia

ble

data

reg

ion

from

two

scal

ar r

aman

dat

a is

con

tain

ed in

the

dark

blu

e bo

xes.

Page 139: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

121

Figure 3.40 Carbon dioxide mole fraction lineplots of experiments (multi-species) andcomputations for 50% flow modulation. Radial data is taken 9 mm above burner.

a

b

c

d

e

Carbon dioxide mole fraction-radial

0

0.02

0.04

0.06

0.08

0.1

0.12

0 5 10

radial distance(mm)

compexp

Carbon dioxide mole fraction-radial

0

0.02

0.04

0.06

0.08

0.1

0 5 10

radial distance(mm)

compexp

Carbon dioxide mole fraction-radial

0

0.02

0.04

0.06

0.08

0.1

0 5 10

radial distance(mm)

compexp

Carbon dioxide mole fraction-centerline

0

0.02

0.04

0.06

0.08

0.1

0.12

0 20 40 60 80

height(mm)

compexp

c

Carbon dioxide mole fraction-centerline

0

0.02

0.04

0.06

0.08

0.1

0.12

0 20 40 60 80

height(mm)

compexp`

Carbon dioxide mole fraction-centerline

0

0.02

0.04

0.06

0.08

0.1

0.12

0 20 40 60 80

height(mm)

compexp

Carbon dioxide mole fraction-radial

0

0.02

0.04

0.06

0.08

0.1

0 5 10

radial distance(mm)

compexp

Carbon dioxide mole fraction-centerline

0

0.020.04

0.060.08

0.10.12

0.14

0 20 40 60 80

height(mm)

compexp

Carbon dioxide mole fraction-radial

0

0.02

0.04

0.06

0.08

0.1

0 5 10

radial distance(mm)

compexp

Carbon dioxide mole fraction-centerline

0

0.02

0.04

0.06

0.08

0.1

0.12

0 20 40 60 80

height(mm)

compexp

phase

Page 140: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

122

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)0

0.16

0

0.04

40

0

6080100

120 0

0.01

0.02

0.03

20

0.05

0.06

stea

dy f

lam

e

Cen

terl

ine

fuel

exi

t vel

ocity

vs.

tim

e

v(cm/s)

time(

s)d

cb

ae

(iii)

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4

5101520253035404550

04

8-8

-4r(

mm

)r(

mm

)r(

mm

)r(

mm

)r(

mm

)

(a)

(b)

(c)

(d)

(e)

(a)

(b)

(c)

(d)

(e)

(ii)

(i)

stea

dy

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

z(mm)

stea

dy

r(m

m)

r(m

m)

Fig

ure

3.41

Wat

er p

rofi

les

for

50%

mod

ulat

ion

(mol

e fr

actio

n).

(i)

Mul

ti-sp

ecie

s m

easu

rem

ent.

(ii)

Com

puta

tions

. (iii

) C

ente

rlin

e fu

el

ex

it ve

loci

ty v

s. ti

me.

Pha

ses

whe

re d

ata

is ta

ken

are

mar

ked

by

a,

b,c,

d,e

on th

e gr

aph

and

corr

espo

nd to

the

imag

es la

belle

d lik

ewis

e.

St

eady

fla

me

imag

e is

sho

wn

on th

e fa

r le

ft. T

he s

tead

y fl

ame

fuel

exi

t vel

ocity

is in

dica

ted

in th

e gr

aph.

Unr

elia

ble

data

reg

ion

from

two

scal

ar r

aman

dat

a is

con

tain

ed in

the

dark

blu

e bo

xes.

Page 141: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

123

Figure 3.42 Water mole fraction lineplots of experiments (multi-species) and

computations for 50% flow modulation. Radial data is taken 9 mm above burner.

a

b

c

d

e

phase

Water mole fraction-radial

0

0.05

0.1

0.15

0.2

0 5 10

radial distance(mm)

compexp

Water mole fraction-centerline

0

0.05

0.1

0.15

0.2

0 20 40 60 80

height(mm)

compexp

Water mole fraction-radial

0

0.05

0.1

0.15

0.2

0 5 10

radial distance(mm)

compexp

Water mole fraction-centerline

0

0.05

0.1

0.15

0.2

0 20 40 60 80

height(mm)

compexp

Water mole fraction-radial

0

0.05

0.1

0.15

0.2

0 5 10

radial distance(mm)

compexp

Water mole fraction-centerline

0

0.050.1

0.150.2

0.25

0 20 40 60 80

height(mm)

compexp

Water mole fraction-radial

0

0.05

0.1

0.15

0.2

0 5 10

radial distance(mm)

compexp

Water mole fraction-centerline

00.05

0.10.15

0.20.25

0 50 100

height(mm)

compexp

Water mole fraction-radial

0

0.05

0.1

0.15

0.2

0 5 10

radial distance(mm)

compexp

Water mole fraction-centerline

0

0.05

0.1

0.15

0.2

0 20 40 60 80

height(mm)

compexp

Page 142: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

124

"a", "b" and "c" in the experiment. The computations do show this "pinching off"

phenomenon in phases "b" and "c", but to a lesser extent than seen in the experiment. The

experiment and computations show good agreement in lift off height and in peak values

(except a lower peak radial temperature for the experiments). Radial plots indicate a

slightly wider flame in the experiment, although the variation in species and temperature

on the lean side of the flame agree well.

3.10 Summary

A steady and time-varying laminar methane diffusion flame is quantified in terms of

flame structure, species concentrations, mixture fraction, and temperature.

Chemiluminescence gives a quick and inexpensive way of accurately determining the

flame structure and lift off height of a flame. CH* chemiluminescence produces CH*

number density images that give a preliminary picture of the effect of the flow

modulation on the flame shape over the period of the forcing. The flame front stretches

and then "pinches off," shrinking the upstream portion of the flame back down while the

downstream portion convects farther downstream.

The two scalar technique produces images of the mixture fraction and temperature of the

steady and time-varying flame through the measurement of only two scalar quantities in

Page 143: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

125

the flame. This technique produces good agreement away from stoichiometric when

compared with a major species measurement of temperature and mixture fraction. The

two-scalar technique gives a lower than expected mixture fraction near stoichiometric

than a major species measurement, due to the approximations in the two-scalar derivation

of mixture fraction. Interference on the fuel Raman image prevents quantitative

measurements in the interference regions. The two-scalar technique does offer a much

quicker and easier way to get temperature and mixture fraction than a measurement of all

of the major flame species. In addition, the two scalar technique can be applied much

more readily to a flame where there is less or no repeatability.

The multi-species technique produces images of temperature and major species

concentrations for the steady and time-varying flames. The difference Raman sub-

technique eliminates C2 and PAH interference on the weak Raman signals. In the steady

flame, the multi-species experiment produces results that agree well with the

computations in terms of spatial distribution and absolute values. For the forced flame,

the computations indicate a more modest forcing than experiments- further refinement in

the computations is needed. However, there is good agreement in peak values, lift-off

height, and in radial distributions 9 mm above the burner.

Page 144: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

126

Chapter 4Soot and temperature characterization of

a sooting, laminar, ethylene diffusion flame

4.1 Introduction

There has been an increase in concern over the past decade for the environmental impact

of combustion generated soot. Soot is a major pollutant from diesel engines. Studies have

shown inhalation of soot particles as a health risk, leading to respiratory problems and

possibly some types of cancer. Tighter emission standards will require these engines to

produce less soot. Soot is important in the transfer of heat in flames, and can

significantly contribute to thermal loads on combustor walls. Measuring soot emission

will lead to better models that predict soot creation and destruction in combustion

environments. These models can then be used to help build combustors that reduce or

control soot production.

Although most practical combustors involve complex interactions of fluid mechanics and

chemistry, an axisymmetric, laminar flame allows for the ability to model the interaction

of soot with detailed flame chemistry. Since soot plays a key role in radiative heat

transfer in flames, determination of local temperatures along with soot characterization is

Page 145: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

127

vitally important. In order to test these models, experiments are needed that produce high

spatial and temporal resolution in the detection of soot and measurement of temperature.

4.2 Flame and Burner Characterization

The same burner and configuration used in Chapter 3 is used in this experiment. The

flame is an axisymmetric, coflowing laminar diffusion flame. The flame is lifted to

prevent uncertain inlet boundary conditions due to heat loss to the burner. The fuel is

32% ethylene diluted with nitrogen. Ethylene is chosen for its high sooting tendencies

and its relatively simple structure for modeling purposes as compared to larger

hydrocarbons. This flame and similar ethylene diffusions flames have been well

characterized by experiments and computations [McEnally et al. 1998, Ni et al. 1995,

Wainnier and Seitzman 1999, Smyth et al. 1985]. This work represents the most complete

characterization of a sooting ethylene flame in the literature. The average fuel flow

velocity at inlet is 35 cm/s and is matched to the coflow velocity to minimize sheer

effects. This velocity is enough to produce a lifted flame but small enough to allow a

stable flame. From PIV measurements of velocity, there is no detected radial velocity

component.

Page 146: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

128

4.3 Computational Model

Computations for this flame are done by Professor Mitchell Smooke [McEnally et al.

1998] at Yale University. The model utilizes velocity -vorticity formulation (i.e.

containing vorticity terms) of the gas-phase conservation equations along with transport

conservation equations for soot. The chemical mechanism for ethylene has 45 species and

233 reactions, and is based on GRI Mech 1.2. The conservation equations are initially

solved on a two dimensional mesh for a particle-free flame using a time-dependent

approach. Then the soot equations are incorporated and an adaptive gridding method is

used.

4.4 Probe measurement of temperature and soot volume fraction

Thermocouple measurements are performed by McEnally [McEnally et al.1998] at Yale

University. Temperature is measured with a 75 µm wire thermocouple. Rapid insertion of

the thermocouple into the flame minimizes soot deposition onto the thermocouple. A

correction is made for the radiative heat transfer to the thermocouple.

Soot volume fraction is measured with the same thermocouple using a technique where

the volume fraction is determined by the soot mass transfer rate to the thermocouple

Page 147: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

129

junction [Eisner and Rosner 1985]. The full flame profile is obtained by translating the

thermocouple to known positions in the flame.

4.5 Experimental determination of temperature using the two scalar technique

The temperature field is determined optically using the two scalar approach of Stårner

[Stårner et al 1996] as a comparison to the probe measurement of temperature, and to

achieve a temperature measurement with better spatial resolution. This approach involves

the measurement of fuel concentration and Rayleigh scattering. In this experiment, fuel

Raman scattering is used as a measure of fuel concentration. The two scalar technique is

described in detail in Chapter 3.

4.5.1 Optical imaging setup for temperature measurement

The setup is shown in Figure 3.7 in Chapter 3. The second harmonic of a Q-switched

Nd:YAG laser emits 532 nm pulses of 8 ns in duration, at a 10 Hz rep rate. The laser is

magnified by a 3 X telescope and focused by a cylindrical lens of focal length 50 cm into

a 15.2 mm tall vertical sheet over the center of the burner. The scattered light is collected

perpendicular to the laser axis with a 50 mm camera lens( f/1.4 for Raman imaging, f/5.6

for Rayleigh imaging). Resulting pixel magnification is 12.5 pix/mm. The light passes

Page 148: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

130

through an appropriate interference filter and then is focused onto an image intensified

CCD camera (Photometrics CC200). The laser, intensifier, and camera are synchronized

with digital delay generators such that data is only acquired when a laser pulse is present.

The intensifier gate time is 1 µs, which is small enough to prevent the acquisition of large

interfering flame luminosity. Since the flame is laminar, the scattering signal from many

laser pulses can be integrated on the CCD chip. Data is acquired for Rayleigh and Stokes-

shifted Q-branch Raman scattering at separate times, as justified by the repeatability of

the flame. The images are then downloaded to a computer for processing.

For Rayleigh scattering, images at two downstream locations are acquired. Rayleigh

images are integrated over 100 laser pulses. For the first set of images the laser sheet is 3

mm off the surface of the burner, as going any closer to the burner caused significant

elastic scattering interference. The next height had the bottom edge of the laser sheet at

the visible flame tip (28 mm off the burner surface). Data could not be acquired in the

intermediate flame heights because of the significant elastic scattering interference in this

region from soot particles. A 532 nm interference filter (10 nm bandpass) is used to

collect the Rayleigh scattering and reject signals outside this spectral range. Laser energy

is set to 195 mJ/pulse. Data are acquired similarly for calibrations and background signals

of uniform, pure gas flows of helium, air, and ethylene at room temperature. The laser

beam waist is estimated by replacing the cylindrical lens with a spherical lens of equal

Page 149: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

131

focal length, and focusing the laser down to a line over the burner. Rayleigh scattered

light from the beam is imaged onto the camera, and the FWHM beam waist is measured

to be 300 µm.

For the ethylene Raman imaging, the laser is focused into a line across the burner with

approximate beam waist of 300 µm. Laser energy is reduced to 36 mJ/pulse to prevent

laser breakdown since the laser is now focused down to a point. Raman signals are

integrated over 200 laser pulses. Twenty Stokes-shifted ethylene Raman signals are

acquired at different heights above the burner, in steps of 0.5 mm. The line data are then

tiled together to form images. A 630 nm (10 nm FWHM) interference filter collects the

Stokes-shifted ethylene Raman scattering from the C-H stretch vibrational mode. Images

are also acquired from a uniform ethylene flow at room temperature and from

background flame luminosity present in the Raman data.

4.5.2 Processing

The Raman and Rayleigh images are corrected for throughput, detector dark counts, and

background as described in Section 3.4.2. Uniform fields of pure air and pure ethylene at

room temperature serve as calibration to the Rayleigh and Raman images.

Page 150: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

132

4.5.3 Calculation of Temperature

Two scalar temperature calculation is described in Section 3.3.4. The functional

dependence of aT ,Cp, and Wmix on mixture fraction are obtained from computations of

this flame. The iterative process converges in four or less iterations. Final temperature

images are cropped, then mirrored and scaled to obtain a full symmetric temperature

profile. These profiles are then compared to temperature profiles from computations and

thermocouple measurements.

4.5.4 Two scalar temperature comparison with probe measurements and computations

Figure 4.1 shows the two-dimensional profile of temperature as determined by

thermocouple, computational, and the two scalar techniques. Good agreement is seen in

the profiles downstream, and upstream of the soot region. No meaningful Rayleigh data

can be obtained in the sooty region due to the significant elastic scattering from the soot,

and C2 fluorescence. The two scalar temperature shows similar temperature distribution,

lift off height, and peak temperatures as the probe measurements and computations. The

two scalar technique produces a slightly lower peak flame temperature than in the

thermocouple measurement and computations. Thermocouple measurements near the

Page 151: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

133

05

-5r

(mm

)

2044

Cal

cula

ted

The

rmoc

oupl

e

298

Tw

o sc

alar

05

-5r

(mm

)0

5-5

r (m

m)

10203040 0

z (mm)

Fig

ure

4.1

Tem

pera

ture

Pro

file

Com

pari

sons

(K

)

Page 152: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

134

fuel tube seem to overpredict the temperature here, as compared to the two scalar and

computational temperature profiles.

4.6 Determination of the soot volume fraction profile using laser-induced incandescence

4.6.1 Introduction and theory

Laser-induced incandescence (LII) has been used effectively to determine soot volume

fraction and particle size in sooting environments such as laminar flames [Mewes and

Seitzmann 1997, Ni 1995, Vander Wal 1994, 1996], ideal soot generators [Seitzmann

1999], and in diesel engines [Pinson et al. 1993]. The advantages of the LII technique are

high spatial resolution, good detection limits, and the non-intrusive interaction with the

combustion environment. At sufficient laser intensities, LII signal is shown to be

proportional to soot volume fraction. In this case, the measurement of soot volume

fraction involves simply the luminosity-subtracted, time-integrated LII signal along with

an appropriate calibration constant.

If one assumes a monodisperse distribution of particles with diameter a0, particle volume

fraction can be defined as

Page 153: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

135

f N av po= ( )π

6

3(4.1)

In the limit of high laser power, and when the maximum particle temperature is reached

(dT/dt=0), (2.17) m

adadt

a dTdt

WRT

p vpp v

v

pv p4 32π

ρ β α= − = − can be substituted into

(2.14)

K a I ap v

T T mH

WT T

K deabs

g gp g p

v

vSB p g

abs

em

π απ γγ

π σ η η ηη

η

20

2 4 4 43

211

1 151

− +−

− + − −−∫

*

* ( / ) ˙ ( / ) ( )( )∆

− =43

03π ρa cdTdtp p to yield:

K a IH

RTp vabs v

v

pv pπ α2

0 0,max − =∆(4.2)

With (2.15) K aa m

mabsex

( , ) Imλ πλ

= −+

8 12

2

2 one can solve (4.2) for Tp in terms of a, I0,max.

Substituting this expression for Tp into the expression for LII intensity (2.25)

J t N S a t T t p a g da d dVp ema

a

Vp( ) ( ( ), ( )) ( ) ( )= ∫∫∫

1

2

1

20 0 0

λ

λ

λ λ , J reduces to

J C N p a a dapa

ax= ∫1

0 0 0

1

2

( )( ) (4.3)

where x c em= + −3 1λ , c is a constant dependent upon particle material properties (0.154

for carbonaceous particles), and λ em is in microns. If one assumes a very narrow particle

size distribution, (4.3) becomes

J C N ap

c em= ( ) + −

10 3 1λ

(4.4)

Page 154: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

136

Comparing (4.4) and (4.1), the incandescence in the high temperature part of the LII

curve is approximately linear with fv. Inaccuracies in the assumption of linearity of LII

signal with fv are discussed in Seitzman [Mewes and Seitzman 1997] for various λ em ,

various time integration gates around the LII signal, and variations in particle size.

4.6.2 LII imaging setup

The setup for the LII imaging measurement of soot volume fraction is shown in

Figure 4.2. The second harmonic of a Q-switched Nd:YAG (10Hz rep rate) laser is

magnified by a 3 x telescope, and then focused with a cylindrical 50 mm focal length lens

into a vertical sheet across the center of the burner, with the bottom edge of the sheet 8

mm off the burner surface. The magnification and vertical placement of the laser sheet is

intended to capture most of the LII signal within a region of the laser sheet where the

laser fluence variations are small, as variations in laser fluence will complicate the LII

analysis. The LII signal is collected perpendicular to the laser axis by a f/1.4 50 mm focal

length camera lens. The collected light then passes through a 450 nm (10 nm bandpass)

interference filter. This spectral region is known to have minimal interference from flame

emission and C2 fluorescence with a 532 nm excitation wavelength in hydrocarbon

flames [Vander Wal and Weiland 1994, Wainner and

Page 155: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

137

Figure 4.2 LII Imaging Setup

Nd:YAG Laser532 nm

30 cm f.l.

f/1.8 50 mm f.l. lens

CCD450 nm

ComputerEnergy Meter

interference filter

delaygenerator

imageintensifier

Page 156: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

138

Seitzman 1999], and less error in soot volume fraction measurements from variation in

particle size and temperature as compared with longer detection wavelengths [Mewes and

Seitzman 1997]. Although a green excitation laser is known to produce significant

fluorescent interference from C2 species in the visible, this occurs only at fluences > 5

J/cm2, well above the fluences at which data is acquired in this experiment [Seitzmann

1999]. The incandescence is then focused onto an image intensified, cooled CCD camera.

Pixel magnification of the images is 10 pix/mm.

4.6.3 Data acquisition

The intensifier gate is widened to 2 µs to detect the entire incandescence signal over its

decay. The intensifier, laser, and camera are synchronized with digital delay generators to

capture the prompt rise of the LII signal along with the full decay of the signal. Although

it has been shown that a prompt 50 ns gate would result in less error in volume fraction

measurement for a change in particle diameter [Mewes and Seitzman 1997], the time

response of the intensifier gating may cause greater problems. The intensifier time

response is approximately 50 ns to turn on and 50 ns to turn off, for a step function input

signal to the cathode. Therefore if the intensifier turned off during the LII signal decay,

there would be a substantial chance for inconsistency as to when the intensifier is off and

Page 157: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

139

no longer intensifying the incandescence signal. Therefore the signal is acquired over the

entire LII signal when the intensifier is fully "on".

In the region of greatest incandescence signal a survey is conducted of incandescence

signal versus laser fluence. At lower laser fluences, the incandescence signal increases

with laser fluence. This dependence starts to flatten out at sufficient fluences where

vaporization competes with particle heating. The laser fluence range used in the final

image acquisition of the incandescence signal is the fluence range where the

incandescence signal variation with laser fluence is small compared with the LII signal

variation with laser fluence at lower fluences. The resulting LII signal vs. laser fluence

curve is shown in Figure 4.3. The threshold laser fluence (the fluence at which the LII

signal becomes independent of laser fluence) of 1 J /cm2 is observed here, which is

approximately the same threshold fluence determined by Santoro [Quay et al. 1994] for a

similar ethylene diffusion flame and excitation scheme.

Shot-to-shot variation in LII images is recorded for 10 laser shots. Each single shot is

processed separately, then compared to each other. Then the average of these 10 is

compared to a single shot.

Page 158: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

140

Figure 4.3 Variation of time-integrated LII signal with laser fluencefrom LII imaging. 532nm laser excitation. A 2 µs integration

window is used, including the rise of the LII signal.

0

500

1000

1500

2000

2500

3000

3500

4000

4500

0 1 2 3

Laser Fluence(J/cm^2)

LII

Sig

nal (

Arb

. uni

ts)

Page 159: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

141

Figure 4.4 Calculated LII response to variations in laser fluence acrossthe height of the laser sheet. Shown in light gray is the region where LIIsignal is detected in the experiment.

Calculated LII response to variation in laser fluence vs. position within laser sheet

00.20.40.60.8

11.21.41.6

0 5 10 15 20 25 30 35 40

vertical position(mm)

laser fluenceLII response

00.460.921.381.842.32.763.223.68

Cal

cula

ted

LII

res

pons

e (A

rb.)

laser fluence (J/cm2)

region ofdetectable LII

Page 160: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

142

Images of flame luminosity, detector dark current, and uniform field are acquired to

correct the raw LII images for luminosity background and optical throughput. Raman

scattering of a uniform field of pure ethylene at room temperature is acquired to

determine the variation in laser fluence vs. height above burner (HAB). This curve is

used along with the LII vs. laser fluence curve (Figure 4.3) to obtain the expected

variation in LII signal vs. position within the laser sheet. Figure 4.4 shows the variation in

laser fluence across the sheet as well as the expected variation in LII response due to the

variation in laser fluence. The region of detectable LII signal is marked in the figure. The

maximum expected variation in LII signal due to laser fluence variation within the sheet

is 7% within the region containing LII signal. Therefore no correction to the LII signal is

made.

4.6.4 Processing

Images of flame luminosity Ilum(x,y) are subtracted from the raw LII images ILII,flame (x,y)

to correct for background luminosity interference. A uniform field, diffuse white light

image Idiff(x,y) is used to correct the background corrected LII images for optical

throughput. Detector dark current ID is used for the background of the uniform field

image. The throughput and background corrected LII image is given by

Page 161: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

143

I x yI x y I x y

I x y I x yLII corLII flame lum

diff D,

,( , )( , ) ( , )

( , ) ( , )0 =−

−(4.5)

Final images are then scaled and compared with probe measurements and computations.

The LII image intensity is calibrated with the probe measurement such that the peak soot

volume fraction of the probe measurement matches the peak LII signal.

4.6.5 Error estimates of the LII technique in determining soot volume fraction

Figure 4.5 shows the single shot LII images. Shot-to-shot, the LII profile does not vary

appreciably. The profiles are structurally similar compared to each other, and to the

average over ten shots. Also shown is the peak LII signals shot to shot. There is not more

than a 7% variation in peak signal. Therefore, it is reasonable to integrate the LII signal

over many shots, and therefore increase the detection sensitivity. One may observe a

slight asymmetry in the soot profiles in the main soot-containing region. This is most

likely due to asymmetry in the flows.

The LII signal is assumed to be proportional to soot volume fraction where there is

measurable signal. However, particle size has a significant effect on LII signal, since

larger particles reach higher peak temperatures, and thus emit more per soot volume

fraction. It is assumed for this flame that the primary particle size distribution is small.

Page 162: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

144

Figure 4.5 Shot-to-Shot LII Fluctuation. The far rightimage is an integration of 15 shots. The graph shows the LIIintensity taken from the region of maximum LII signal.

single shots

15 shots

0

200

400

600

800

1000

1200

1400

1600

0 5 10

Shot-to-Shot LII Signal

min max

10

20

30

0 4-4 0 4-4 0 4-4 0 4-4 0 4-4

z(mm)

r(mm)

LII

Int

ensi

ty (

Arb

.)

Shot number

Page 163: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

145

This is verified by TEM images from a carbon grid deposition technique. Grid samples

are taken along the flame centerline at heights where there is significant soot volume

fraction as measured by LII.

The temperature of the particles before laser heating will affect the peak temperature

reached, the cooling decay, and thus the LII signal. As seen with the computational

temperature data, the variation in gas temperature over the soot-containing regions is at

most 100 K. Compared to the temperature to which the particles are heated

(~ 2500 K)[Filippov 1999] there will be an insignificant difference in the LII signal decay

for a gas temperature difference of 100 K.

Other than for a small flame luminosity, which is subtracted off of the LII signal, there is

no apparent interfering signal within the spectral window of the 450 nm filter. At very

high fluences, it is known that a 532nm excitation will excite C2 species, which fluoresce

across the visible, and would interfere with this signal [Seitzman 1999]. The wing tips of

an interference signal can be seen on the ethylene Raman data (Fig. 4.6). This

interference is either laser-induced incandescence of the soot or C2 fluorescence. It has

been stated that using the fundamental of the YAG laser (1064 nm) is more desirable as it

produces less C2 Swan band interference [Vander Wal 1994]. A good check would be to

compare the LII images at excitation wavelength of 532 nm to that at 1064 nm.

Page 164: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

146

Figure 4.6 Interference on ethylene Raman data.The region of interference is indicated in the image.

Ethylene Raman

jet center Interference

min max

r(mm)

4 8

z(mm)

4

8

12

Page 165: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

147

Figure 4.7 Experimental and Computational Soot Volume Fraction.32% Ethylene Flame.

LII

10

20

30

0

0 4-4

r (mm)

0 4-4 0 4-4

ProbeComputed

10

20

30

0

z (m

m)

7.7 E-7

0

9.9 E-7

0

r (mm) r (mm)

z (m

m)

Page 166: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

148

4.7 LII soot volume fraction comparison to probe measurements and computations

Figure 4.7 shows a comparison of soot volume fraction profiles for LII, probe, and

computational techniques. The probe and LII measurement have peak soot volume

fractions along the centerline, at the same height above burner, approximately 21 mm.

The computations peak along the wings and extend closer to the burner than is the case

for the probe measurements. The LII soot volume fraction does not extend down as far as

the probe measurements in the wings of the volume fraction profiles. One explanation for

this is that the probe can detect translucent particles such as soot precursors (which

appear closer to the burner surface than mature soot) which do not absorb the visible laser

energy, and thus the LII technique would not detect.

4.8 Primary soot particle size using time-resolved LII

4.8.1 Introduction

Time-resolved LII (tires-LII) is used to determine the size distribution of soot in the

flame. Others have used tires-LII to determine particle size in flames with other

techniques, but all based on the model initially proposed by Melton [Will et al. 1995,

Mewes and Seitzman 1997]. An improved model is developed to analyze the time decay

Page 167: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

149

of the time resolved LII curves to obtain the primary particle size distribution. This model

is applicable in the free molecular regime and incorporates effects unaccounted for in

previous models such as thermal particle swelling and gas impingement cooling [Rosner

2001, Filippov and Rosner 2000a, Rosner 2000]. The model is used by a novel search

procedure to determine particle size distributions.

One can simultaneously numerically integrate the energy flux equation (2.14)

K a I ap v

T T mH

WT T

K deabs

g gp g p

v

vSB p g

abs

em

π απ γγ

π σ η η ηη

η

20

2 4 4 43

211

1 151

− +−

− + − −−∫

*

* ( / ) ˙ ( / ) ( )( )∆

− =43

03π ρa cdTdtp p

and the mass flux balance equation (2.17) m

adadt

a dTdt

WRT

p vpp v

v

pv p4 32π

ρ β α= − = − to

obtain the time dependence of the particle radius a(t,a0,I0) and temperature Tp(t,a0,I0) for

an initial particle radius a0 and a laser power density I r t0 ( , )r

where rr represents a

position within the probed volume.

The relevant properties of soot and the surrounding gas are summarized in Table 4.1 and

Figure 4.8 on the following page. For calculation of Kabs the index of refraction of soot

(or carbon) must be known. The index of refraction of carbon is taken from the work of

Lee and Tien [Lee and Tien 1981]. The refractive index is dependent on both

Page 168: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

150

Figure 4.8 Select properties of carbon and nitrogen

20

22

24

26

28

30

0 2000 4000 6000

00.20.40.60.8

11.21.41.61.8

2

300 500 700 900 1100

wavelength (nm)

Index of Refraction - Carbon [Lee and Tien 1981] Temperature = 1400 K

real

imaginary

index ofrefraction (m)

Temperature (K)

Cp( J/mol-K)

Specific Heat of Carbon [JANAF 1985]

7.37.47.57.67.77.87.9

8

1000 2000 3000 4000 5000

Temperature(K)

Adiabatic Ratio -Nitrogen [JANAF 1985]

γγ

**

+−

11

γ * +1

γ * −1

Page 169: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

151

Table 4.1 Parameters used in the soot LII analysis

αv [Thorn and Winslow 1957] ∆Hv(kJ/g-mole) [JANAF 1985]

( soot -> C1 ) 0.4 711.932( soot -> C2 ) 0.3 818.136( soot -> C3 ) 0.1 765.315

Tp*(K) 1300

pv*(atm) (C1) 5.0582x10-5 [Rosner 2001a]

pv*(atm) (C2) 2.4547x10-5

pv*(atm) (C3) 2.5351x10-4

Tg(K) 1580βgraphite (K

-1) 2.1x10-5 [Avallone and Baumeister 1996]

α 1

temperature and wavelength of the incident light. The carbon index of refraction vs.

wavelength for a temperature of 1600 K is used in the model.

Vaporized soot will contain C1, C2, C3, etc., each with a different evaporation coefficient.

For the mass vaporization term of (2.17), the mass vaporizations of all possible vapor

species Ci (weighted by its respective evaporation coefficient) are summed together:

˙,

,, ,

m

adadt

a dTdt

W

RTp vp

p p v iv i

pv i p i

Ci4 32πρ ρ β α= − = −∑

The literature only has evaporation data for C1, C2, C3 [Thorn and Winslow 1957], and

therefore these three species are used in the sum. Vapor pressure curves for each of the

Page 170: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

152

species are estimated using the form of (2.18) p pH

RT

T

Tv vv

p

p

p

= −

**exp

∆1 and JANAF

tables [JANAF 1985, Rosner 2001a] for the relevant parameters.

The gas temperature in the region of peak soot concentration is unknown from the two

scalar measurement as this region produced elastic scattering interference on the

Rayleigh data. The gas temperature measured just outside this region is used as an

approximate gas temperature within the sooty region. The temperature dependence of the

specific heat of carbon is obtained from JANAF [JANAF 1985]. The temperature

dependence of γ* for the gas surrounding the soot is estimated from JANAF [JANAF

1985], based on the properties of nitrogen since the majority of the surrounding gas is

nitrogen. The thermal accommodation coefficient is assumed to be unity. A value for α

of 0.3 is quoted in the literature for nitrogen initially at 300 K near a flat graphite surface

at 1200 K [LeRoy et al. 1997]. The affect on the predicted particle size due to uncertainty

in α is investigated in section 4.8.8.

From numerical integration of (2.25) J t N S a t T t p a g da d dVp ema

a

Vp( ) ( ( ), ( )) ( ) ( )= ∫∫∫

1

2

1

20 0 0

λ

λ

λ λ

one is able to calculate the relative time resolved LII intensity J(t), for a particle

distribution p(a0) ranging from a1 to a2 and integrated over spectral region from λ1 to λ2

Page 171: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

153

with a spectral detection efficiency g(λ). Only a relative intensity curve is calculated,

since an absolute intensity curve would involve knowledge of Np and optical efficiencies

of the collection system.

J(t) is known from the experimental data. The only unknown is p(a0). To obtain p(a0)

from the above equation, the integral must be inverted. If one defines the kernel function

K(a0,t) by

K a t N S a t a I T t a I g dVdp em pV

( , ) ( ( , , ), ( , , )) ( )0 00

00

1

2

= ∫∫ λ λλ

λ

(4.6)

(2.25) becomes a Fredholm integral equation of the first kind

J t K a t p a daa

a

( ) ( , ) ( )= ∫ 0 0 0

1

2

(4.7)

The solution to this equation for p(a0) is a non-linear iterative procedure developed by

Twomey [Twomey 1977] and improved by Markowski and others [Markowski 1987,

Filippov 1990]. This technique assumes the particle distribution is smooth and

continuous. A novel technique presented in this work utilizes the least squares method by

generating the best fit to J(t) by successive guesses at p(a0) until the error of the fit to J(t)

is minimized.

4.8.2 Time-resolved LII setup

Page 172: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

154

Setup for the experiment is shown in Fig 4.9. The fundamental wavelength of the

Nd:YAG laser (1064nm) produces near infrared laser pulses of 8 ns in duration. The laser

beam is unfocused, and is sent through a 0.5 mm diameter pinhole. A uniform intensity

cross section region of the YAG beam is steered through the pinhole. The intent is to get

a uniform beam intensity cross section (i.e., a top hat profile) to minimize spatial

variation in laser fluence which will complicate the analysis of the LII curves. The

pinhole is positioned as close as possible to the burner center (10 cm away), as diffraction

effects are minimized at shorter distances after the aperture. The resulting beam cross

section above the burner is calculated from diffraction theory. The resulting

beam profile can be approximated by a gaussian function with σ = 100 µm. The laser

passes through the burner centerline at a height of 21 mm above the burner, the region of

maximal soot concentration as determined by LII imaging. The spatial profile of soot

volume fraction is known from LII imaging. The LII signal is collected perpendicular to

the laser with a f/1.8 50 mm focal length lens. The collected incandescence is focused

onto a 0.5 mm vertical slit. The signal that passes through the slit is collected by a

photomultiplier tube (PMT) (HammamatsuR928). The signal is not spectrally filtered

before entering the PMT because the cathode of the PMT is not sensitive to the elastically

scattered light at 1064 nm. The object magnification is 1, and the resulting optical sample

volume is 500x200x200 µm3. The PMT signal is sent to a digital oscilloscope

(Techtronix, 500 Msamples/s) where the signal is digitized. The PMT

Page 173: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

155

Figure 4.9 Time-resolved LII setup.

Nd:YAG Laser106 4 nm

0.5 mm Pinhole

0.5 mm Slit

PMT

Oscilloscope

Photodiode

f / 1 .0

Energy Meter

Time Resolved

Comput er

Page 174: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

156

response time is 2 ns, which should be adequate for the LII curves with expected decay

times of approximately 1000 ns for 20-30 nm soot particles [Will et al. 1995]. The shot-

to-shot pulse energy is recorded by a power meter after the beam has passed though the

flame. The temporal laser pulse is recorded with a 2 ns rise time photodiode and digitized

on the scope.

4.8.3 Time-resolved LII data acquisition

Laser fluence is varied from approximately 0.05 J/cm2 to 5 J/cm2, and time-resolved LII

curves are acquired. Single-shot time-resolved LII signals are acquired along with the

temporal laser pulse and laser energy. Weaker signals are averaged over 8 and 64 pulses

and are compared to the single shot LII signals at the same fluence to make sure there is

no loss of information or signal distortion caused by integrating signals.

4.8.4 Qualitative analysis of the time-resolved LII signals

Figure 4.10 shows a sample of time-resolved LII decay curves at various laser fluences.

The peak LII signal is greater as fluence is increased. This makes sense because the

particles are heated by the laser to higher peak temperatures as the laser fluence is

increased, producing an increased amount of incandescence. These signals are then time

Page 175: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

157

Figure 4.10 Time-resolved LII curves at various laser fluences.

Laser fluence (J/cm2)

0.120.61.1

2.1

time (ns)

LII

sig

nal (

Arb

.)

0

200

400

600

800

1000

0 100 200 300 400 500

Page 176: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

158

laser fluence (J/cm2)

Figure 4.11 Time-integrated LII signals vs. laser fluence. Integration gate is 1 µs and includes the rise of the LII signal.

0

10000

20000

30000

40000

50000

60000

70000

80000

0 0.5 1 1.5 2 2.5

tim

e-in

tegr

ated

L

II (

Arb

.)

Page 177: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

159

integrated with a gate of 1 µs to include the peak signal as well as the decay. Figure 4.11

shows the time-integrated LII signals versus laser fluence. (Note that the LII imaging and

time-resolved measurements are performed with different excitation wavelengths so

their respective LII vs. fluence curves will not in general be the same.) The time-

integrated LII signals increases with fluence below a fluence of 1 J/cm2. Between a

fluence of 1 and 2 J/cm2, the curve begins to flatten out. This is due to the competition of

mass vaporization with laser heating. The particle vaporization will limit the maximum

temperature a laser-heated particle can reach. The integration time window is changed to

50 ns and is gated around the peak LII signals. This curve shows the same dependence on

laser fluence as the curve with a 1 µs gate.

The time-resolved LII signals are normalized by their peaks as shown in Figure 4.12. As

the laser fluence is increased, the initial decay rate of the LII signal increases. This makes

sense because particles with a higher peak temperature will cool at a faster rate, and

therefore the respective LII signal will decay at a faster rate. The difference in decay rates

of the LII curves becomes smaller as laser fluence is increased. This indicates a limited

maximum temperature value as laser fluence is increased. Also shown in Fig. 4.12 is the

temporal laser profile. The temporal laser pulse width will limit the particle sizes that can

be detected, as smaller particles will cool at a faster rate. Particles with

Page 178: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

160

Figure 4.12 Time-resolved LII signals for various laserfluences. Signals are normalized to their peaks.Also shown is the time-resolved laser pulse intensity.

0

2 5

5 0

7 5

100

125

0 100 200 300 400 500

Laser fluence (J/cm2)

0.120.61.1

2.1

laser

Nor

mal

ized

LII

(A

rb. u

nits

)

time (ns)

Page 179: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

161

cooling rates equal to or greater than the laser pulse decay rate cannot be extracted from

the model.

The model used to extract primary particle size information from LII data does not

account for aggregate break-up or aggregate restructuring. Some recent experiments have

shown that a laser with a particular laser fluence can sometimes alter the structure of the

soot aggregates without inducing substantial vaporization, i.e. at fluences below

vaporization threshold [Vander Wal 1998]. If restructuring and break-up occurs, particle

heating may compete with the restructuring or break-up, leading to a lower than

predicted maximum temperature (and thus a longer relative decay time for peak

normalized LII curves).

4.8.5 Calculation of particle size distribution from LII data

As an initial approach to finding the particle distribution, a least-squares procedure is

developed. This approach allows for single mode and bi-modal particle distributions. The

form of the distribution of a single mode is assumed to be either lognormal or normal

with parameters a0 and σ where p(a0) is the peak of the mode and σ is the characteristic

spread of the mode. A least squares approach is used to find the optimal curve fit to the

data, where the minimization parameter χ is defined by

Page 180: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

162

χ = −∑=

1 2

1NJ t J td i c i

i

N

( ( ) ( )) (4.8)

where N is the number of points, Jd(t) represents the time resolved LII intensity curve

obtained from experiments, and Jc(t) represent the time-resolved LII intensity curve

generated from (2.25) J t N S a t T t p a g da d dVp ema

a

Vp( ) ( ( ), ( )) ( ) ( )= ∫∫∫

1

2

1

20 0 0

λ

λ

λ λ for a given

particle distribution. The curves are normalized to their peak values. Data from the rise

portion of the LII signals are not used in the curve analysis. Criteria for the value of χ

that must be achieved for a “good” fit is determined from the sensitivity of χ to the fit

parameters.

The parameters that must be determined are a0 and σ for a single mode distribution and

a0,1, a0,2, σ1, σ2, and g12 for a bi-modal distribution, where g12 is the ratio of the peaks of

mode 1 and mode 2. A single mode distribution is used first, as it will involve less

computational time, and most particle synthesis processes produce a single mode

distribution.

A large part of the procedure is based on inspection of the computed curves versus the

data, and the knowledge of the effect on the time decay of the curves when a0 and σ are

varied. The curve can be broken into two parts- the initial fast decay of the curve (part 1),

Page 181: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

163

and then the slower decay region (part 2). The parameter σ is fixed at a small value to

approximate a monodisperse distribution. The parameter a0 is adjusted until the fit to

part 1 of the curve produces a minimum χ ( call it χ(1)). Then the value of a0 is held

constant while σ is increased gradually until the fit to part 2 of the curve produces a

minimum χ (call it χ(2)). The χ value of the entire curve is also calculated here. An

increase of σ will decrease the curve decay rate (increase the slope of the curve) in part 1

of the curve. Now σ is fixed, and a0 is gradually decreased, which has the effect of

increasing the curve decay rate (as smaller particles cool faster than large ones). The

value of a0 is decreased until the fit to part 1 of the curve produces a minimum in χ(1). The

procedure is repeated until there is convergence in the minimum value for χ.

If the procedure for a single mode distribution does not produce a reasonable fit to the

data, a bi-modal distribution algorithm is applied. The LII data curve is broken into two

parts. The first part of the curve has an initial steep drop in intensity. This part of the

curve will be affected by both distribution modes. The second part of the LII curve has a

relatively slow decay in signal. Since small particles have a faster LII decay than larger

particles, the contribution to the LII signal in the second part of the curve from larger

particles dominates. Thus the curve fit in the second part of the curve will be largely

unaffected by the parameters in the first part of the curve (as long as particle distributions

Page 182: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

164

do not begin to overlap). Two separate fits are done using the single mode procedure for

the first and second parts of the curve. After a reasonable fit is achieved for each of the

two parts of the curve, the two distributions are combined linearly with the parameter g12,

and a new fit is generated for the entire curve (and χ of the entire curve is calculated).

The value of g12 is adjusted to produce a minimal value of χ for the second part of the

curve, call it χ2. (Note the difference between χ2 and χ(2). The minimization terms χ(1) and

χ(2) are determined in the single mode procedure used to obtain χ1 and χ2 separately.) The

value of a0 and σ for the second part of the curve is fixed, and the value of a0 for the first

part of the curve is varied to achieve a minimal value of χ for the first part of the curve,

call it χ1. The value g12 is adjusted again to produce a minimal χ2. This process is

repeated until there is convergence in the value of χ for the entire curve.

LII experimental data is analyzed to extract a soot particle size distribution from the

model. An LII experimental curve taken 22 mm above the burner is analyzed. The

measured laser fluence is 0.12 J/cm2. Laser spatial intensity profile is gaussian with σ =

100 µm. Laser temporal intensity profile is gaussian with σ = 3.7 ns. The particle size

distribution is assumed to be normal for one case and lognormal for another case. The

detection sensitivity g(λ) is known from the PMT spectral sensitivity. Convergence of the

Page 183: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

165

iterative procedure that minimizes χ is achieved with less than twenty calculations of

Jc(t).

4.8.6 Grid sampling of soot particles

Results of the LII-derived soot particle distributions are compared to thermophoretic

probe samples of soot obtained from the flame. This experiment is performed by the

group of Professor Daniel Rosner at Yale University. It has been shown that the

orientation-averaged thermophoretic properties of aggregates are insensitive to aggregate

size and structure [Rosner et al. 1991]. In the free-molecule regime, the orientation-

averaged thermophoretic diffusivity of an aggregate varies by only 8% from that of a

single particle. Even though soot forms aggregates under the present conditions,

thermophoretic sampling should give a true representation of the particles/aggregates in

the flame. A rapid grid insertion technique [Köylu et al. 1997, Xing et al. 1999] is used to

extract soot particles at the same location in the flame as the LII measurement. The grids

are imaged by a transmission electron microscope (TEM) giving image magnifications on

the order of 100,000 X. Statistical analysis is performed on the particle images.

Page 184: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

166

Figure 4.13 Curve fit to the soot LII data. Z = 22 above burner.Laser fluence = 0.12 J/cm2. For the normal particle size distribution,a0 = 10 nm, σ = 6 nm, χ = 1.0. For the lognormal distribution,

a0 = 10 nm, σ = .13, χ = 1.2.

LII

sig

nal (

Arb

.)

LII data

Soot LII data fit to curves of modelLaser Fluence = 0.12 J/cm2

0

10

20

30

40

50

60

0 100 200 300 400

time(ns)

fit-normal

fit-lognormal

Page 185: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

167

4.8.7 Comparison of LII-derived and grid sampling particle size distributions

Figure 4.13 shows the fit of the model to the LII data for the conditions above. A value of

χ = 1.0 is achieved for a generated curve using a normal particle size distribution, with a0

= 10 nm and σ = 6 nm. A value of χ = 1.2 is achieved for a generated curve using a

lognormal particle size distribution with a0 = 10 nm and σ = 0.13.

The noise of the LII signal is approximately equal to the error of the two fits, and

therefore there is good agreement between the experimental and computed LII curves.

Figure 4.14 compares particle diameter (i.e., 2a) distributions obtained from time-

resolved LII and TEM images of the grids containing soot samples. The mean of the

lognormal and normal particle diameter distributions is identically 2a0= 20 nm, as

compared to 2a0 = 22 nm for TEM-derived particle sizes. All three distributions are

narrow and have similar distribution widths.

4.8.8 Sensitivity analysis of LII-derived particle sizing technique

A sensitivity analysis is done to see the effect of the change in parameters on the change

in predicted particle size from the model and search procedure. A reference LII curve is

Page 186: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

168

Figure 4.14 Primary soot particle size distributions from grid samplingmeasurements and from the LII-derived particle size distribution.Height = 22 mm above the burner.

PDF sampled soot

Soot size distribution comparison

0

0.2

0.4

0.6

0.8

1

1.2

0 10 20 30 40

diam(nm)

normal

lognormal

Page 187: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

169

generated using the same parameter values for the soot LII curve analysis in Section

4.8.6. The mean particle diameter is selected to be 20 nm as is determined in the soot LII

data analysis. For simplicity, a monodisperse distribution is used. The LII curve

generated with these parameters is used as the reference LII signal. One parameter is

changed at a time. The least-squares procedure is used to find the mean particle diameter

that gives the LII curve with the best fit to the reference LII curve.

Graphs illustrating the effect on predicted particle size due to variation in a parameter

value are shown in Figure 4.15. Shown in light gray are the reference values used to

calculate the reference LII signal. Since the gas temperature uncertainty is less than ± 50

K, the corresponding predicted particle diameter is in the range from 19 nm to 21.5 nm.

The volume expansion coefficient, β, for graphite is in the range from 2.1x10-5/K to

6.6x10-5/K [Rosner 2001a]. This range in β leads to a negligible change in predicted

particle diameter. The thermal accommodation coefficient, α, is not well known for

nitrogen impingement on carbon for Tg = 1600 K and Tp = 2500 K [Rosner 2001b]. The

value for α can be anywhere from 0.3 ( for Tg = 300 K and Tgraphite= 1200 K [LeRoy et al.

1997 and Rosner 2001a]) to 1. This gives a range of predicted particle diameter from 6

nm to 20 nm. The value of α must be known with better certainty to give more

confidence in the prediction of particle size from the model. The refractive index vs.

Page 188: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

170

Figure 4.15 Effect of change in parameter values on predicted particlesize. Shown in light gray are the reference values used to generate theLII reference signal.

Predicted particle diam. vs.Gas temperature

10

15

20

25

30

35

40

1000 1200 1400 1600 1800 2000

Gas Temperature(K)

Pre

dict

ed p

arti

cle

diam

. (nm

)

Pre

dict

ed p

arti

cle

diam

. (nm

)

Pre

dict

ed p

arti

cle

diam

. (nm

)P

redi

cted

par

ticl

e di

am. (

nm)

Pre

dict

ed p

arti

cle

diam

. (nm

)

α (thermal accommodation coeff.)

reference valuesfor soot LII analysis

Predited particle diam. vs. Im[m] at 1064 nm (soot)

15

17

19

21

23

25

0 0.2 0.4 0.6 0.8 1

Im[m] at 1064 nm

Predicted particle diam. vs.Thermal accommodation coefficient

0

5

10

15

20

25

0 0.5 1 1.5

Predicted particle diameter vs.Volume expansion coefficient (soot)

15

17

19

21

23

25

0.0E+00 2.0E-04 4.0E-04 6.0E-04

ß (volume expansion coefficient)

Predicted particle diam. vs.Re[m] at 1064 nm (soot)

5

10

15

20

25

30

1.3 1.5 1.7 1.9 2.1 2.3

Re[m] at 1064 nm

Page 189: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

171

wavelength curves in Figure 4.8 are shifted by the same amount at each wavelength for

the sensitivity analysis. Values for the imaginary part of the refractive index of soot

range in the literature [Lee and Tien 1981] from 0.5 to 0.8 at 1064 nm. This gives a range

in predicted particle diameter from 18 nm to 20 nm. Values for the real part of the

refractive index of soot range in the literature from 1.6 to 1.9 at 1064 nm. This gives a

range of predicted particle diameter from 16 nm to 20 nm.

4.9 Time-resolved laser light scattering and laser absorption

4.9.1 Introduction

Recently, researchers have shown that for aggregated particles such as soot, the laser

used to heat the soot for LII experiments may cause structural changes in the aggregates

and aggregate breakup [Vander Wal 1998]. This phenomenon can occur independently of

particle vaporization if the proper laser fluence is chosen.

In order to see if there are structural changes in the soot due to the laser pulse, time-

resolved laser light scattering (LLS) and laser absorption (or extinction) measurements

are performed. Time-resolved absorption in similar flames has been performed by others

Page 190: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

172

but on longer time scales. From (2.32) C

aE m

kabs =

4 3

2

πλ

( ) (in the Rayleigh limit) the

extinction coefficient is proportional to Npa3. As seen from (2.34) I I K Le= −0 exp( ),

any change in either Np or a will change the absorption ratio I/Io. If particles are

vaporized, this ratio will decrease. This ratio is independent of aggregate parameters such

as fractal dimension Df and radius of gyration Rg. Therefore I/Io should be insensitive to

the restructuring/break-up of aggregates. From Chapter 2, scattering cross sections are

dependent upon Rg and Df, as well as np, and therefore the LLS signal will change if any

of these parameters change. Ideally, one would like to isolate aggregate

restructuring/break-up from vaporization by finding a laser fluence large enough to cause

aggregate restructure/break-up and small enough to not cause significant particle

vaporization.

4.9.2 Setup for LLS and laser absorption experiment

Time-resolved LLS and laser extinction are measured with a pump-probe setup shown in

Figure 4.16. The fundamental of a Q-switched ND:YAG laser (1064 nm) is used to heat

the soot particles in the sooty flame region. A CW Ar+ laser is used for measurement of

extinction and laser light scattering. The Ar+ laser is tuned to the 488 nm transition

producing a CW blue laser beam. The two beams are made collinear and pass over the

Page 191: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

173

center of the burner in opposite directions, at a beam center height of 22 mm above the

burner. The YAG beam is apertured with an iris with a diameter of 2.5 mm. The Ar+

laser is focused down over the center of the burner. The scattering signal is collected at

90 degrees to the laser axis by a 50 cm f.l. f/1.4 lens. The collected light passes through

an interference filter centered at 488 nm and is focused onto the same aperture/PMT

configuration as for the tires-LII measurement. The PMT is terminated with 93 ohms and

the output signal is digitized on the scope, and sent to the computer for analysis. A

beamsplitter is placed in the path of the beams, on the side of the burner away from the

Ar+ laser. This reflects a fraction of the energy of the YAG beam and Ar+ laser in

opposite directions. The reflected YAG beam is collected by a energy meter to monitor

the shot-to-shot YAG energy. The reflected Ar+ laser passes through a 488nm (10 nm

bandpass) interference filter and onto a PIN silicon photodiode. This photodiode records

the change in Ar+ laser intensity as the flame absorption changes. The photodiode is

connected to the digital scope with a 93 ohm termination, where the signal is digitized

and downloaded onto the computer.

4.9.3 Acquisition of LLS and laser absorption

The laser fluence of the YAG laser is varied, and the time-resolved scattering and

absorption signal are recorded just before, during, and after the YAG pulse. Signals are

Page 192: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

174

Figure 4.16 Time-resolved Scattering/Absorption Setup

Aperture

Mirror

Photomultiplier Tube

Ar+ Laser

Nd:YAG LaserEnergy Meter

Burner

f/1.8 50 mm lens

Interference Filter

Photodiode

50 cm f.l. lens

0.5 mm slit

Beamsplitter

ND FilterInterference Filter

Scattering Absorption

Page 193: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

175

integrated over 256 YAG pulses to improve the signal-to-noise ratio of the

measurements. A correction to the scattering signal is made to account for LII

interference. This correction is acquired by turning off the Ar+ laser while leaving on the

YAG laser. The shot-to-shot laser energy is also monitored as mentioned in section 4.9.2.

4.9.4 LLS/Absorption results

Time-resolved scattering and absorption for several fluences are shown in Fig. 4.17.

Marked in the charts is the time of maximum YAG laser intensity (the YAG laser pulse

width is approximately 8 ns). As one can see from the scattering graph, there is

significant change in scattering signal after the YAG pulse (-50% for a fluence of 0.15

J/cm2) at fluences well below vaporization threshold (i.e. << 1 J/cm2). The change in

scattering signal increases for increased laser fluence. The absorption graph indicates no

significant change in absorption after the YAG pulse at fluences below threshold. At a

fluence of 0.9 J/cm2 (near vaporization threshold), the absorption signal just begins to

show absorption changes (-5%). At a laser fluence well above vaporization threshold (4

J/cm2), there is a significant change in absorption (-20%). Figure 4.18 shows the time-

resolved scattering signal on longer time scales than Figure 4.17, for the YAG laser

Page 194: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

176

Figure 4.17 Time-resolved change in elastic scattering and absorption of the sootyregion (measured with the Ar+ laser) due to the YAG laser pulse at various YAG laserfluences. The time of peak YAG pulse intensity is marked with a dashed line.

0.15

0.9

4.0

0.15

0.56

time of YAG laser pulse peak intensity(pulse width= 8 ns)

laser fluence (J/cm2)

Cha

nge

in a

bsor

btio

n (%

)

laser fluence (J/cm2)

Cha

nge

in e

last

ic s

catt

erin

g(%

)

% Change in Scattering vs time

- 1 0 0

- 8 0

- 6 0

- 4 0

- 2 0

0

2 0

0 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5

time(µs)

%Change in absorption vs. time

- 1 0 0

- 8 0

- 6 0

- 4 0

- 2 0

0

2 0

0 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5

time(µs)

Page 195: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

177

Figure 4.18 Time-resolved change in elastic scattering of the sooty region (measuredwith the Ar+ laser) due to a YAG laser pulse with laser fluence = 0.15 J/cm2.

- 1 0 0

- 8 0

- 6 0

- 4 0

- 2 0

0

2 0

- 5 0 0 5 0 100 150 200 250 300

t0peak YAGlaser intensity(8 ns pulse width)

%Change in elastic scattering vs time Laser fluence = 0.15 J/cm2

Cha

nge

in e

last

ic s

catte

ring

(%)

time (µs)

t0

t1

Ar+

YAG

2.5 mm

flow

Ar+

soot region heated by YAG

soot region heated by YAG

flow

t1

Page 196: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

178

fluence of 0.15 J/cm2. Marked in the figure (in addition to the time of maximum YAG

pulse intensity) is the time, t1, when the volume of soot heated by the YAG pulse has

convected 0.3 mm downstream. This distance is estimated by knowing the approximate

flow velocities and apertured YAG beam dimensions. (The flow velocity along the

centerline and 20 mm above the burner is estimated from computations to be ~ 130

cm/s.) After the YAG pulse, the scattering suddenly drops, and then increases gradually.

At t1, the scattering signal has recovered to its initial level before the YAG pulse ( at t <

t0). Note that the probe volume of the Ar+ laser is still well within the heated soot region

at t1. For the next 90 µs after t1, there is a positive change in scattering signal as compared

to the scattering before the YAG pulse (with a peak of ~+8% change). The scattering then

returns to its initial level before the YAG pulse.

Results indicate significant aggregate breakup/restructuring at laser fluences well below

vaporization threshold. The positive change in scattering at time t1 in Figure 4.18 may be

indicative of restructuring after aggregate breakup, resulting in a larger aggregate radius

(Rg) than that of the unheated soot. Vander Wal indicated from grid sampling and

probe/pump LII experiments that the LII pulse can cause a re-structuring of the soot after

aggregate breakup [Vander Wal 1998].

Page 197: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

179

4.10 Error estimates of the absorption/scattering technique

The scattering detector signal magnitude of unheated soot in the flame corresponds to a

13.5 mV deflection on the oscilloscope. The noise on that signal is 0.4 mV giving a lower

detection limit of 3% change in scattering. The absorption detector signal in the presence

of the flame subtracted from the detector signal without the flame corresponds to a 4mV

deflection on the oscilloscope. The noise on that signal is 0.3 mV giving a lower

detection limit of 7.5% change in absorption. This relatively high value is due to the

small path length through the sooting flame (5 mm) and the peak soot volume fraction of

10-6. Absorption measurements could be improved with flames of higher soot levels (e.g.

an ethylene flame with less N2 dilution), or multiple passes of the laser through the flame.

4.11 Summary

A temperature profile is determined using the two scalar technique. The two scalar

temperature profile agrees with the thermocouple measurement and computational

temperature profile in general flame structure, flame length, and lift off height. Peak

temperature in the two scalar measurement is 10% lower than in the computations and

thermocouple measurement. No meaningful measurement can be made in the soot-

containing region due to the elastic scattering of the soot particles.

Page 198: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

180

Soot volume fraction profiles from LII imaging are measured in the ethylene flame.

Comparison with soot volume fraction profiles obtained by probe measurements and

computations indicates a peak in soot volume fraction at 22 mm above the burner for the

probe and LII measurements. Computational and probe measurements of soot volume

fraction show longer wings in the soot volume fraction profiles. The LII technique is not

sensitive to translucent soot that is formed closer to the burner than mature soot.

A primary particle size distribution of soot particles is obtained from analysis of the time-

resolved LII curves. The distribution obtained agrees with the distribution from TEM

pictures of grid sampling measurements. The good agreement of particle size

distributions obtained from LII curve analysis with TEM measurements verifies the

model and iterative procedure used in the LII signal analysis for laser fluences

substantially below particle vaporization threshold. The model may not produce as

accurate a result for laser fluences that cause significant vaporization. A sensitivity

analysis indicates that parameters such as refractive index and thermal accommodation

coefficient must be known to a high degree of certainty for the predicted particle sizes to

be known with high certainty. Parameters such as thermal expansion coefficient and gas

temperature only need to be known approximately for high certainty of the predicted

particle sizes.

Page 199: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

181

Time-resolved scattering and absorption measurements have shown a significant initial

decrease in elastic scattering signal due to the morphological effects of the LII laser on

the soot, at laser fluences well below the soot vaporization threshold. No decrease in

absorption is observed until the LII laser fluence approaches the soot vaporization

threshold. Results indicate aggregate break-up and restructuring due to the LII laser,

which is not accounted for in the model used in the LII analysis. Also not included in the

model are several other effects related to the aggregation of particles that may be

significant in the LII analysis. One is a net shielding effect that reduces the rate of gas

impingement on the particles. This effect is estimated to reduce the accommodation

coefficient by a factor of (Na)1/8 [Rosner 2001a]. For Na = 100, the reduction factor is

estimated to be 1.7. Another effect is a net energy focusing by aggregates that increases

the laser energy absorbed by the average particle. There is not uniform agreement on the

significance of this effect in the literature. Mackowski states that for Na = 100, the energy

absorbed by the average particle increases by 24%, while Farias estimates only a 8%

increase in energy absorbed [Mackowski 1995 and Farias et al. 1995]. These effects

should be incorporated into the model in the future.

In future experiments, LII/LLS/absorption measurements could be used to estimate soot

aggregate bond strength and radius of gyration Rg. These in-situ measurement techniques

Page 200: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

182

may be applicable to industry where they could eventually be used to help monitor and

control particulate formation in combustion environments.

Page 201: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

183

Chapter 5Characterization of nanoparticle structures

synthesized in a premixed, methane/air flat flame

5.1 Introduction

Ultrafine powders like carbon black and silica are produced by condensation of vapor

which is formed by combustion synthesis. These and other ultrafine powders have special

magnetic, optical, or electrical properties that make them of economic importance. The

nanometer grain size and special nanostructures of the powders is what gives these

substances their special properties. Therefore, control of production of the particulates

which make up the powders is vitally important. This requires in-situ monitoring of the

particulates. The use of laser diagnostics offers an instantaneous, remote, non-intrusive

method of monitoring the formation and characterizing these particulates. In particular,

laser light scattering (LLS) and laser-induced incandescence (LII) can be used to study

particle/aggregate size, number density, and volume fraction. LLS and LII may be able to

give an estimate of bond strength between particles, and particle vaporization thresholds

in some cases. LLS and LII can also be used to monitor production of undesirable

combustion-synthesized particulates such as fly ash and soot when burning oil or coal.

Page 202: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

184

In the present experiment, we produce iron oxide particles in a flame (γ-Fe2O3, or

hematite). Iron oxide is chosen for its absorption properties in the green part of the

spectrum, since the laser of choice is the Nd:YAG, which can produce plenty of energy at

its second harmonic wavelength of 532 nm. In order to get the most repeatable conditions

possible, and allowing for lean enough conditions for the iron atoms to oxidize, a laminar

premixed flat flame (essentially a one dimensional flame) is used to produce the

particulates.

5.2 Burner and Flame

The burner is shown in Figure 5.1. The premixed flame is stabilized on an inner cylinder

of diameter 6.35 cm. This cylinder contains glass beads and wire mesh screens to diffuse

the flow. A honeycomb of cell size 1/32" sits on top of the inner cylinder and is used to

straighten the flow. This cylinder is surrounded by another cylinder through which flows

an air coflow, to help minimize the radial flow component. The flame is a lean premixed

flat flame with a premixture of methane, air, and water vapor containing dissolved FeCl2.

The flame sits 1-2 mm off the burner surface, and is lean to allow enough oxygen and

time for the Fe particles to oxidize post flame to produce Fe2O3. The flame emission is

dominated by a yellowish orange color that is distinct from the color of soot emission.

Page 203: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

185

Figure 5.1 Burner for iron oxide particle production

Fuel/Air/Seeding

4 mm Glass Beads

80 Mesh Screen

1/32" Honeycomb

Air Air

6.35 cm

2 mm 2 cm

6 cm

Page 204: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

186

Two flow conditions are used to produce two flames (#1 and #2) of differing

stoichiometry, temperature, and flow velocity:

Table 5.1 Flow and flame conditions for the seeded premixed methane flame

flows:___________________#1 ____________________#2__________

methane: 2.61 lpm 0.90 lpm

air+particles: 36.9 lpm 15.7 lpm

(air+particles)/fuel: ________14.1 17.4___________

equilib. flame temp. 1804 K 1572 K

inlet flow velocity 20.8 cm/s 8.7 cm/s

equivalence ratio 0.71 0.57

FeCl2 particles are introduced into the flame by seeding particles into the gas flows. A

solution of FeCl2 (actually iron(II) chloride tetrahydrate) in water is placed into an

atomizer (TSI model 9606). The atomizer is pressurized to 30 psi of air. A high pressure

jet of air passes over the top of several vertical tubes that are in contact with the solution

at the bottom of the tubes, drawing the solution into the flow. The solution and air impact

Page 205: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

187

onto a sphere at high speeds, allowing the smallest droplets to pass and rejecting the

larger droplets. The droplets are dried by the air flow and only particles remain. The

FeCl2 (tetrahydrate) solution concentration is 625 g FeCl2-4H2O/ 1L H2O. The solution is

drawn into the flow at a rate of 0.68 ml/min for (#1) and 0.29 ml/min for (#2).

5.3. Measurement of the LII spectrum

Time-integrated LII spectra for flame #1 are acquired at various laser fluences, heights

above burner, and acquisition gate delays. A flame emission spectrum (no laser) is first

acquired to determine the spectral region with minimal interference on the LII signal.

Based on this information, a suitable spectral region is chosen for the LII measurement.

5.3.1 LII spectrum setup and acquisition

The experimental setup is shown in Figure 5.2. The second harmonic of a Q-switched

Nd:YAG laser (532 nm, 8 ns pulse width) produces a laser beam of which a portion

passes through the rectangular aperture of height 3.81 mm and width 1.52 mm. A

uniform intensity of the beam is steered through the aperture, producing an output beam

with an approximately "top hat" profile, as observed by the elastic scattering from the

beam. The resulting beam passes over the center of the burner at heights ranging from 2-

Page 206: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

188

Figure 5.2 Experimental LII spectrum setup

Aperture

Mirror

Spectrograph f/4

Nd:YAG Laser

Energy Meter

Burner

f/1.8 50 mm lens

0.2 mm slit

Colored GlassCCD Camera

Time-integrated LII spectrum experimental setup

Page 207: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

189

12 cm. The incandescence, flame, emission, and elastically scattered light are collected

perpendicular to the beam by a 50 mm f.l. lens. For the LII measurement, an appropriate

colored glass filter is placed behind the collection lens. The spectral region over which

the filter transmits light is chosen for its minimal flame emission and other interference,

good spectral detection efficiency and complete attenuation of the elastically scattered

light. The filter transmits in the 380 - 475 nm range, and is used in the time-resolved LII

experiment as well. The incandescence is focused onto the vertical entrance slit (width =

200 µm) of a spectrometer (Spex 270, f/4, 0.27 m f.l.). The light is then dispersed by the

grating and focused onto an image intensified CCD camera. The digitized image is then

downloaded to a computer.

The intensifier, camera, and laser are synchronized to only record data near the time of a

laser pulse. First, the LII spectrum is recorded at various laser fluences. The integration

time gate on the intensifier is set to 1 µs to acquire signal over the rise, peak, and full

decay of the incandescence signal. The data is acquired 5 cm downstream of the burner,

where the flame and LII signal are stable. The acquisition is integrated over 100 laser

pulses to improve signal-to-noise ratio.

Page 208: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

190

Second, the laser fluence is fixed and the height above the burner is varied from 5 cm to

11 cm. Third, the integration gate on the intensifier is set to its minimum (100 ns), the

height and laser fluence are fixed, and the delay of the intensifier gate is varied with

respect to the rise of the laser pulse.

5.3.2 LII Spectrum processing

Flame emission images (no laser) are acquired with the same integration times as the LII

data. This data is subtracted from the raw LII images to account for the interference of

flame emission on the LII signal. The background-subtracted LII images are then

corrected for spatial throughput, spectral detection efficiency of the intensifier, and

spectral throughput of the colored glass filter. Images are integrated over their spatial

dimension (as the signal should not vary spatially at a given height above burner) and

reduced to lineplots of signal intensity over a spectral range.

5.3.3 Flame emission results

Shown in Figure 5.3 is the corrected lineplot of flame emission intensity vs. wavelength.

The flame emission signal appears to peak around 594 nm and decays at the same rate

toward the blue or red. Based on the flame emission curve, the ideal spectral filter for the

Page 209: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

191

Figure 5.3 Flame emission spectrum. Shown in gray is the spectral regionchosen for the time-resolved LII experiments

0

10000

20000

30000

40000

50000

60000

70000

80000

90000

370 470 570 670

wavelength(nm)

spectral region chosenfor LII experiments

Flame emission scan

flam

e em

issi

on in

tens

ity

(arb

. uni

ts)

Page 210: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

192

LII experiment could either transmit far toward the red or blue part of the spectrum,

where the flame emission is minimal. A filter that transmits in the blue is chosen because

the detection sensitivity of the intensifier as well as the PMT peak in the blue part of the

spectrum. The spectral transmission of the filter chosen for the LII experiments (380-470

nm, Corning 5-57 colored glass filter) also rejects any measurable elastically scattered

light (@ 532 nm).

5.3.4 LII spectrum results

Figure 5.4 shows a raw LII spectrum for a given fluence, and the signal corrected for

throughput, filter response, and detection sensitivities. Figure 5.5 shows the LII scans for

several laser fluences. The LII spectra indicate broad, blackbody emission-like curves.

The incandescence increases toward the red part of the spectrum, which makes sense

because the peak emission wavelength for a blackbody radiating at 4000 K, an estimate

for Tmax of a particle heated by a laser [Filippov 1999], is in the infrared. One does notice

a slight decrease in LII signal to the red of 450 nm. A similar drop in LII signal to the red

of 450 nm is observed by Vander Wal for a 532 nm excitation of soot from a premixed

ethylene flame [Vander Wal and Weiland 1994]. The overall LII signal increases with

laser fluences from 0.33 to 0.47 J/cm2, then decreases at higher laser fluences. This

decrease in signal at higher laser fluences is the result of mass loss due to particle

Page 211: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

193

Figure 5.4 Raw LII spectrum and the LII spectrum corrected for optical throughputand detector efficiencies. Also shown is the spectral response of the collection.

spectral responseuncorrected LIIcorrected LII

Corrected and uncorrected LII spectrum

0

1000

2000

3000

4000

5000

6000

380 400 420 440 460

wavelength(nm)

LII

sig

nal (

Arb

. uni

ts)

Page 212: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

194

Figure 5.5 LII spectrum for several laser fluences.

LII

Sig

nal (

Arb

. uni

ts)

LII spectrum at various laser fluences

0

500

1000

1500

2000

2500

380 400 420 440 460

wavelength(nm)

0.47 J/cm2

0.33 J/cm2

0.69 J/cm2

1.09 J/cm2

1.57 J/cm2

Page 213: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

195

vaporization. Beginning at a laser fluence of 0.69 J/cm2 one observes an intensity peak

near 405 nm that increases in intensity at higher fluences. Also apparent at

0.69 J/cm2 is an intensity peak near 380 nm that intensifies at higher fluences. At a laser

fluence of 1.09 J/cm2 and higher one observes two distinct peaks near

380 nm. In his study of iron nanostructures with LII, Vander Wal [Vander Wal 1999]

noticed a sharp peak in LII signal at high laser fluences (compared to the blackbody-like

variation of incandescence with wavelength) at 370 nm for excitation with a laser at 1064

nm. At a delayed signal detection of 50 ns from the rise of the 8 ns width of the laser

pulse, this peak disappeared, indicating that it is likely to be emission from an excited-

state of a species formed by particle vaporization (since these events are typically 10 -9-

10-8 s long). Figure 5.6 shows the LII spectrum for a laser fluence of 1.57 J/cm2, with the

detection gate set to capture the peak LII signal, and with the detection gate delayed by

65 ns from the rise of the laser pulse. One sees the drastic reduction of the unknown

peaks near 308 nm and 405 nm, indicating these peaks to be excited-state emission from

a species formed during vaporization.

Figure 5.7 shows LII spectra at three different heights above the burner. One sees an

increase in overall incandescence as one goes from 5.25 cm to 8.25 cm downstream, but

no increase from 8.25 cm to 11.25 cm downstream. The initial increase indicates either

an increase in number density or increase in primary particle size. From a comparison of

Page 214: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

196

Figure 5.6 Delayed and prompt detection of LII spectrum.Laser fluence = 1.57 J/cm2.

prompt gate

65 ns delayed gate

LII spectrum(corrected)Delayed gate vs non-delayed gate

0

500

1000

1500

2000

380 400 420 440 460

wavelength(nm)

LII

Sig

nal (

Arb

. Uni

ts)

Page 215: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

197

Figure 5.7 LII spectrum for several heights above the burner

5.25 cm8.25 cm11.25 cm

LII spectrum at various heights

0

1000

2000

3000

380 400 420 440 460

wavelength(nm)

LII

Sig

nal (

Arb

. Uni

ts)

Page 216: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

198

(4.1) f N av po= ( )π

6

3 and (4.4) J C N ap

c em= ( ) + −

10 3 1λ

, the spectrally and temporally

integrated LII signal at sufficient laser fluences is linear with ƒv and therefore goes as

~Npa3. One can intuitively see that longer residence times in the flame may allow the

creation of bigger or more numerous Fe2O3 particles. At far distances downstream, the

post flame gases spread out, and the number density will decrease.

5.4 Sampling of iron oxide particles

The sampling experiment and data reduction are performed by the group of Professor

Daniel Rosner at Yale University. Iron oxide particles are sampled from the post-flame

region and statistically analyzed. Deposition of the particulates onto a thin carbon grid is

performed at the same height above the burner as the LII data acquisition. The grids are

circular and approximately 3 mm in diameter. The deposition technique is identical to the

technique used by Köylu and others [Köylu 1997, Xing 1999] and involves fast insertion

(80 ms in the flame) of the carbon grid into the flame region, minimizing oxidation and

damage to the grid. The grids are then imaged with a TEM, producing particle images at

magnifications on the order of 100,000 X. From the images, particle size distributions are

obtained, along with information on particle aggregation.

Page 217: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

199

Figure 5.8 TEM images of thermophoretically sampled particles forflame #1 and flame #2.

flame #1

flame#2

100 nm

100 nm

Page 218: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

200

TEM images of the particles from flame #1 are shown in Figure 5.8. The magnification is

indicated in the image. Aggregation is present only for flame #2.

There appears to be a large number of particles within a narrow size distribution, and

there also appears to be a few relatively large-size particles. For flame #2, particle

aggregation is seen in the TEM images (Figures 5.8), where the estimated average value

of np ~ 3-5 particles/aggregate. The majority of primary particles in the images from

both flame #1 and #2 are 5-15 nm in diameter. Dividing the particles into bins of

various particle size ranges yields particle size distributions that are compared to the

results of the time-resolved LII experiment. (Results are shown along with the LII-

derived particle size distributions in Section 5.6.)

5.5 X-ray diffraction of the particle material

The X-ray diffraction experiment and data analysis is performed by the group of

Professor Wei Tong at Yale University. X-ray diffraction is used to determine the

stoichiometry of the particulate matter produced in the flame. A metal plate is placed

above the flame, and the particulate material is deposited onto the plate for twelve hours.

The deposited material is scraped off of the plate, forming a powder to be used by the X-

ray diffraction apparatus. The diffraction scan produces diffraction peaks at various

Page 219: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

201

Figure 5.9 Xray diffraction peaks of sample (top graph) and of pure hematite

(bottom graph). Three unknown peaks in the sample are marked.

unknown peakssample

pure hematite

angle (degrees)

Inte

nsit

y (A

rb.)

Page 220: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

202

angles of the sample with respect to the incident X-rays. These peaks are then compared

to a database of diffraction peaks of pure compounds.

Results of X-ray diffraction indicate hematite to be the most likely molecule in the

sample. Shown in Figure 5.9 are the diffraction peaks of the sample compared with the

diffraction peaks of pure hematite. Three unknown peaks are marked on the graph, which

may be other iron compounds, or possibly some chlorinated compounds (as the seed

material used is FeCl2).

5.6 Time-resolved LII and laser light scattering experiment

5.6.1 Experimental setup and acquisition

Time-resolved laser light scattering (LLS) and laser-induced incandescence (LII) are used

to obtain primary particle size, detect particle aggregation and estimate aggregate size,

and to study the effects of laser fluence on the particles/aggregates [Rosner et al. 2001b].

The experimental setup is shown in Figure 5.10. The second harmonic of a Q-switched,

Nd:YAG laser (10 Hz rep rate, 532 nm, 8ns pulse width) produces a laser beam of which

a portion passes through a rectangular aperture of height 3.81 mm and width 1.52 mm.

The output beam has approximately a top hat intensity profile in height and width, which

Page 221: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

203

Figure 5.10 Time-resolved LII and laser light scattering setup.

Aperture

Mirror

Photomultiplier Tube

Nd:YAG Laser

Energy Meter

Burner

f/1.8 50 mm lens

Interference Filter

Photodiode

0.5 mm slit

Colored Glass

LIIScattering

Page 222: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

204

is observed from the sharp, well-defined edges of the elastic scattering. The beam passes

over the center of the burner at heights varying from 2-12 cm above the burner.

The burner is placed on a z translation stage to easily change the height above burner. For

LII, the incandescence is collected at 90 degrees to the laser with a 20 cm f.l. lens, then

through a sharp cut colored glass filter that passes light in the spectral region 380-475

nm, and highly rejects the elastically scattered light (@ 532 nm). The incandescence is

focused onto a 3 mm square aperture, and the output light is collected by a

photomultiplier tube (Hamamatsu R928), which has a response time of 2 ns. For LLS, the

elastic scattering signal is collected at 45 degrees to the laser axis by a 30 mm f.l. lens,

and then through an interference filter centered at 532 nm (10 nm bandpass). The

transmitted signal then is collected by a photodiode, which has a 2 ns response time. Both

the time-resolved LII and LLS signals are digitized on a fast digital oscilloscope (500

Msamples/sec). Single-shot data are acquired along with signals integrated over 8 and 64

laser shots. The laser fluence is varied over a wide range of fluences and recorded by an

energy meter. The LLS and LII signals are recorded at each fluence for conditions (#1)

and (#2).

No processing is needed, as the background flame emission signal is negligible. Spectral

and optical sensitivities are incorporated into the model for the time-resolved LII curves.

Page 223: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

205

Figure 5.11 Time-resolved LII at several laser fluences.

.09 J/cm2

.11 J/cm2

.14 J/cm2

.17 J/cm2

.23 J/cm2

.29 J/cm2

.40 J/cm2

tires-LII vs laser fluence-flame#1

0

500

1000

1500

2000

2500

3000

0 100 200 300 400

time (ns)

LII

inte

nsit

y (A

rb.)

Page 224: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

206

5.6.2 Qualitative analysis of time-resolved LII data

Figure 5.11 shows time-resolved LII curves for flame #1 for varying laser fluences. The

characteristic sharp rise in LII signal is followed by a relatively slow decay in signal. The

rise in signal is the region where the laser is heating the particles, and the signal decay

region is where the laser pulse is gone, and the particles cool by convection. Shown in

Figure 5.12 are the same curves normalized by their peaks. Included is a time response of

the PMT to the elastic scattering of the particles, which is representative of the PMT time

response to the laser pulse. The time response of the PMT is important as it affects the

time decay of the LII curves, while also limiting the size of the particles that one can

analyze. As noted in Chapter 2, smaller particles will cool faster, and therefore this limits

this technique to analysis of particles large enough to have a characteristic decay time

greater than the PMT response time. In this figure, the LII signals exactly follow the laser

pulse on the signal rise portion of the curves, while the decay time of the signal decreases

with increasing laser fluence. This makes sense since particles that see larger fluences

will heat to higher temperatures, and thus their cooling rate will be initially faster. There

does seem to be a limiting curve as one gets higher and higher in fluence, however. This

is due to the particles having a maximum temperature limited by particle vaporization

effects which increases at higher fluences. Thus the vaporization of the particles

competes with the laser heating effect at high fluences.

Page 225: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

207

Figure 5.12 Time-resolved LII at several laser fluences. Signals are normalized to thepeak intensity. Also indicated is the time-varying laser pulse intensity, measured fromelastic scattering with the PMT.

.09 J/cm2

.11 J/cm2

.14 J/cm2

.17 J/cm2

.23 J/cm2

.29 J/cm2

.40 J/cm2

laser

tires-LII (normalized) vs. laser fluence-flame#1

0

500

1000

1500

2000

2500

0 100 200 300 400

time(ns)

LII

inte

nsit

y (A

rb.)

Page 226: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

208

Figure 5.13 Time-resolved LII at two different laser fluences for flame #1 and flame #2.Curves are normalized to the peak intensity. Also shown is the normalized LII signal forsoot with a laser fluence of 0.12 J/cm2, and the time-varying laser pulse intensity.

nanoparticle tires-LII (normalized)for flame #1 and flame #2

compared to soot LII

0

10

20

30

40

50

60

70

80

90

100

0 100 200 300 400

time(ns)

flame #2, 0.14 J/cm2flame #1, 0.14 J/cm2

flame #1, 0.09 J/cm2

flame #2, 0.09 J/cm2

laser

soot, 0.12 J/cm2

LII

(ar

b. u

nits

)

Page 227: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

209

Figure 5.13 compares the peak-normalized curves for two different laser fluences of

flame #1 and #2, and a soot LII curve for a fluence of 0.12 J/cm2. The signal decay time

for a given laser fluence is similar for flame #1 and #2. The curves indicate that the

primary particle size from flame #1 and #2 should be roughly the same, and smaller than

soot particles. Shown in Figure 5.14 is the time integration of the LII curves (not

normalized by the peaks) over a 50 ns gate around the laser pulse, plotted vs. laser

fluence for flame#1 and #2. The time-integrated LII curves for the two flame conditions

show similar laser fluence dependence below 0.2 J/cm2. Both curves begin to flatten near

0.2-0.25 J/cm2, where vaporization begins to affect the LII curves and competes with

particle heating.

5.6.3 Particle sizing model, parameters, and procedure

The model and procedure used to obtain primary soot particle size distributions in

Chapter 4 from time-resolved LII curves is used here as well. Parameters specific to the

particle material and flame conditions are discussed here.

Since data is taken in the post-flame region of a premixed flame, the gas temperature, Tg,

is determined via thermodynamic equilibrium calculations for the mixture. All gas

properties are based on the properties of N2, as this species dominates the equilibrium

Page 228: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

210

Figure 5.14 Time-integrated LII signals for flame #1 and flame #1 as afunction of laser fluence. The integration gate is 50 ns, and includes the riseof the time-resolved LII signal.

Integrated LII (50 ns gate) vs laser fluenceflame #1 vs. flame #2

0

2000

4000

6000

8000

10000

12000

14000

0 0.1 0.2 0.3 0.4 0.5

laser fluence(J/cm 2)

flame #1

flame #2

Page 229: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

211

state (75% by volume). No data is available for vapor pressure (or heat of vaporization)

of Fe2O3. Instead, the heat of vaporization is calculated from the chemical equilibrium

equation

Fe2O3(s) = 2 FeO(g) + (1/2) O2(g) (5.1)

using the enthalpies of FeO and O2 (see Table 5.2) , obtained from JANAF tables

[JANAF 1985]. Therefore there will be a ( ) /pO2

1 4− dependence in the effective FeO

vapor pressure. The O2 pressure in the post flame region, is estimated from

thermodynamic equilibrium calculations. In the above equation, it is assumed that FeO is

the dominant Fe carrier. The effective FeO vapor pressure is [Rosner 2001a]

p E pTv FeO O,

/( . )( )exp( . ( ))= − −−2 774 9 44 29 11700

2

1 4 (atm) (5.2)

The mass flux of FeO is calculated from (2.17)

m

adadt

a dTdt

WRT

p vpp v

v

pv p4 32π

ρ β α= − = − (5.3)

The mass flux of Fe2O3 can be related to the mass flux of FeO from mass conservation of

(5.1),

˙ ˙m

am

a

W

WFe O FeO Fe O

FeO

2 3 2 3

4 4122 2π π

= (5.4)

This expression is used for the mass flux term in the mass flux and energy flux equations.

The following graphs in Figure 5.15 and the thermodynamic quantities in Table 5.2 are

used in the model:

Page 230: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

212

Figure 5.15 Select hematite and nitrogen properties

γ*+1

γ*−1

Adiabatic Ratioγ*+1

γ*−1-Nitrogen

realimaginary

7.37.47.57.67.77.87.9

8

1000 2000 3000 4000 5000

Temperature(K)

Heat capacity(at constant pressure)-hematite

142

144

146

148

150

1000 2000 3000 4000 5000

Temperature(K)

Cp

(J/m

ol-K

)in

dex

of r

efra

ctio

n

Index of refraction (m)hematite(T=300K)

00.5

11.5

22.5

33.5

300 350 400 450 500 550

wavelength(nm)

[JANAF 1985]

[Hsu and Matijevic 1985]

[JANAF 1985]

Page 231: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

213

Table 5.2 Parameters used in nanoparticle LII analysis

Flame#1 Flame#2

H (T) - H (kJ/mol)O0

O0

2 2 ,298K 51.7 44.3 [JANAF 1985]

H (T) - H (kJ/mol)FeO0

F0

eO K,298 54.4 46.8 [JANAF 1985]

∆Hf FeO,0 (kJ/mol) 221.3 226.4 [JANAF 1985]

Tg(K) 1804 1572p(O2) (atm) 0.0547 0.0807∆Hv Fe O, 2 3

( J/g) 7,845 7,845

βFe O K2 3

1( )− 3.0x10-5

pv,FeO(1700 K) (atm) 2.774x10-9 [Rosner 2001]Tp

* (K) 1700

5.6.4 Particle distribution results from LII data analysis

LII data curves for a laser fluence of 0.09 J/cm2 are analyzed with the model for flames

#1 and #2. Figure 5.16 shows LII data for flames #1 and #2 along with the best fit LII

curve from the model for a bi-modal normal and lognormal particle size distribution and

a single mode normal and lognormal distribution. Also indicated for each curve is its

respective χ value. The χ value for each fit is indicated in the table below:

Page 232: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

214

Figure 5.16 Least-squares fits to the experimental LII curves using a lognormal(1 and 2 mode) and normal (1 and 2 mode) particle size distribution for flame #1 andflame #2. Laser fluence = 0.09 J/cm2 for the LII curves of flame #1 and flame #2.

Least squares fit to LII data-flame#1laser fluence = 0.09 J/cm2

0

5

10

15

20

25

30

35

40

45

50

0 100 200 300 400 500 600time(ns)

LII

sig

nal (

arb.

uni

ts)

LII

sig

nal (

arb.

uni

ts)

fit- normal (2-modes)fit- normal (1-mode)

LII datafit- lognormal (1-mode)

fit- normal (2-modes)fit- normal (1-mode)

LII datafit- lognormal (1-mode)

Least-squares fit to LII data-flame#2laser fluence = 0.09 J/cm2

0

5

10

15

20

25

30

35

40

45

50

0 100 200 300 400 500 600

time(ns)

fit-lognormal (2-modes)

fit-lognormal (2-modes)

Page 233: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

215

Table 5.3 χ values for the fit to the LII data using various size distributions

fit normal-1 mode lognormal-1 mode normal-2 modes lognormal-2 modes

χ, flame #1 3.0 1.9 1.0 0.7

χ, flame #2 3.7 2.9 0.6 0.5

Clearly the bi-modal particle size (normal and lognormal) distribution gives a better fit

than the single mode normal and lognormal distribution for the overall curves.

Particle size distributions obtained from the best-fit curves for a bi-modal normal, bi-

modal lognormal, single mode normal, and single mode lognormal particle size

distribution are compared with particle- sampled size distributions in Figure 5.17 for

flames #1 and #2. For particle-sampled size distributions, the dependent axis of the

graphs in the figure represents the total particle volume ( n x Vp) while the independent

axis represents the volume for a single particle (Vp). There is good agreement of the bi-

modal lognormal and bi-modal normal particle distributions with the grid sampling

particle size distributions for flame #1 with the primary particle distribution mode at Vp =

33 -1436 nm3 (corresponding to a particle diameter range 2a = 4 -14 nm ). There is also

Page 234: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

216

Figure 5.17 Comparison of LII-derived particle size distributions with grid samplingparticle size distributions. The dependent axis values for the grid sampling represent thetotal particle volume (n x Vp). Also indicated is the total number of particles sampled forflame #1 and flame #2.

Size distributions for hematiteflame #1

0.E+00

1.E+05

2.E+05

3.E+05

4.E+05

5.E+05

6.E+05

7.E+05

8.E+05

1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06

individual primary particle volume(nm3)

Particles sampled1900

Particles sampled

960

tota

l vol

ume

(nm

3 )

grid samplingnormal (2 modes)normal (1 mode)lognormal (1 mode)

grid samplingnormal (2 modes)normal (1 mode)lognormal (1 mode)

Size distributions for hematiteflame#2

0.E+00

1.E+04

2.E+04

3.E+04

4.E+04

5.E+04

6.E+04

7.E+04

8.E+04

9.E+04

1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06

individual primary particle volume(nm3)

tota

l vol

ume

(nm

3 )

lognormal (2 modes)

lognormal (2 modes)

Page 235: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

217

agreement of the bi-modal lognormal and bi-modal normal particle size distributions

with the grid sampling particle size distributions for flame #1 for larger particle size

modes at Vp = 7238 - 590,000 nm3 (corresponding to a particle diameter range 2a = 24 -

104 nm). The best agreement with the grid-sampling distribution for the larger particle

size modes for flame #1 is with the bi-lognormal distribution. The bi-modal lognormal

and normal distributions for flame #2 and the grid sampling distribution show good

agreement for the primary particle distribution mode at Vp = 4 – 2460 nm3 (corresponding

to a particle diameter range 2a = 2 – 18 nm). The grid sampling particle size distribution

for flame #2 lacks the large particle size distribution as seen in the bi-modal particle size

distributions for flame #2. In the particle sampling distribution for flame #1, the particle

bins of intermediate and large particle sizes contain only a few particles, which yields a

large relative statistical uncertainty compared to the primary particle bin (which contains

1893 particles). For an ideal comparison of particle sampling and LII-derived particle

size distributions, the number of particles sampled should be equal to the number of

particles within the probe volume of the LII experiment. Using the known seeding and

flow conditions along with the known LII probe volume, the estimated number of

hematite particles within the LII probe volume is ~ 1014 (compared to 1900 particles

sampled for flame #1). Therefore a particle size distribution mode with a peak four orders

of magnitude smaller than the primary mode will have statistical significance in the LII-

derived particle distributions.

Page 236: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

218

The particle-sampling size distributions for flame #1 are used to compute an LII curve

from the model. This curve is then compared to the LII curve for flame #1 by calculation

of the fit parameter χ. LII curves are computed using one, several, or all particle bins

from the particle sampling measurements of flame #1. Figure 5.18 shows the particle

sampling-computed LII curves compared with measured LII curve of flame #1 for a laser

fluence of 0.09 J/cm2. Bin 1 (1893 particles, 10 nm mean particle diam.) generates a

curve with χ = 5.0. Bin 2 (3 particles, 27.5 nm mean particle diam.) generates a curve

with χ = 7.2. Bins 1-2 generates a curve almost identical to the curve of Bin 1, with χ =

4.6. Adding the 2 particles of Bin 3 (42.5 mean particle diam.) to 1896 particles of Bins

1-2 yields a curve with χ = 3.2. Adding the single particle of Bin 4 (65 nm mean particle

diam.) to the 1898 particles of Bins 1-3 yields a curve with χ = 1.1. Adding the single

particle of Bin 5 to the 1899 particles of Bins 1-4 yields a curve with a relatively poor fit

(χ = 7.0).

The above analysis shows the extreme sensitivity of LII decay curves to large particles,

even if there are only a small amount relative to the relatively smaller particles. It is more

appropriate to consider the particle volume contribution within a given particle size range

rather than just the number of particles within a given size range. This makes sense

Page 237: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

219

Figure 5.18 LII curves generated from grid sampling data. The particle bins used togenerate each curve are indicated to the right of the graph. All curves are normalized tothe peak LII intensity.

particle bins included#1#2#1+#2#1+#2+#3

#1+#2+#3+#4all bins

lii data-flame #1

18933211

particle diam.bin range(nm)

1 4 - 1 42 24 -343 34 -444 64 -745 94-104

LII curves generated from sampling data vs. LII data-flame#1

0

10

20

30

40

50

60

0 100 200 300 400 500

time(ns)

Laser fluence = 0.09 J/cm2

Page 238: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

220

because at sufficient laser fluences it is known that ILII ~ ƒv, and since ƒv ~ Nda3,

then ILII ~ Nda3. It is therefore beneficial to produce particles that result in a single mode

size distribution that can be modeled with a normal or lognormal distribution.

5.6.5 Results of LLS

Figure 5.19 shows the comparison of the Mie-scattered signal from flame #1 and flame

#2 as the laser fluence is varied. The systematic error is indicated for each data point. At

low laser fluences (below 0.025 J/cm2) the scattering from the particles created in flame

#2 is 5 times as large as the scattering from the particles created in flame #1. The

scattering plot vs. laser fluence for flame #1 is linear over the region (0 - 0.23 J/cm2).

This is expected from Mie scattering theory since the scattering cross section is linear

with laser intensity. Above a laser fluence of 0.23 J/cm2, the curve for flame #1 deviates

from a linear curve, where the dependence on fluence is now less than linear. As

indicated from the integrated LII intensity vs. laser fluence curve (Fig. 5.14), this is the

region where mass loss due to vaporization begins to occur. The curve for flame #2 is

linear over the range of fluences (0 - 0.025 J/cm2). The curve begins to have a laser

fluence dependence less than linear at a fluence of 0.025 J/cm2, which is well below the

Page 239: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

221

Figure 5.19 LLS vs. fluence for flame #1 and flame #2. Indicated in the figure are linearregression fits to portions of the data and systematic error on the measurements. Theshaded rectangles represent linear fits to the data within the standard deviation of the fit.

Scattering vs. Fluence

fluence(J/cm2)

scat

teri

ng s

igna

l (ar

b. u

nits

)

Scattering vs. Fluence

fluence(J/cm2)

flame#2

flame#1

standard deviationof fits, flame#2

linear fit to pts. 6-20, flame #2

linear fit to pts. 1-20,flame#1

linear fit to pts. 1-5,flame #2

scat

teri

ng s

igna

l (ar

b. u

nits

)

0

50100

150200

250300

350400

450

0 0.1 0.2 0.3 0.4

flame#2

flame#1

standard deviationof fits, flame#2

linear fit to pts. 6-20, flame #2

linear fit pts 1-20,flame#1

linear fit to pts. 1-5,flame #2

0

20

40

60

80

100

120

140

0 0.02 0.04 0.06 0.08 0.1

Page 240: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

222

fluence where vaporization is occurring (see Fig. 5.14). At higher fluences, the curve

again becomes linear, until a fluence of 0.23 J/cm2 is reached and vaporization effects

decrease the scattering dependence on laser fluence.

The factor of five difference in scattering intensities at low fluence between flames #1

and #2 can be explained from Rayleigh-Debye-Gans theory. For isolated spherical

particles the scattering signal is

I N Cpp

pνν νν θ= ( ) (5.5)

where Cpνν θ( ) is defined in (2.28) C

a F mk

pνν θ θ( ) ~

( )cos ( )

6

22 . The scattering signal for

an aggregate with aggregate number density Na is

I N Caa

aνν νν θ= ( ) (5.6)

where Caνν θ( ) is defined in (2.29) C n C qRa

pp

gνν νν= ƒ2 ( ). Since the particle volume

fraction should be conserved for equivalent seed levels in the flame,

Nanp = Np (5.7)

Combining (5.5), (5.6), and (5.7), the ratio of scattering signals for isolated particles and

aggregates is

II

N n qR

Nn qR

a

pa p g

pp g

νν

νν

= ƒ2 ( )

( ) (5.8)

Page 241: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

223

The form factor ƒ calculated from the results of the TEM particle images of flame#2

(where roughly, np~ 4 and Rg~31/2a0) is approximately equal to one. Therefore the ratio of

scattering signals should be approximately equal to np~ 4, which is the approximate ratio

of scattering signals from LLS at low laser fluences between flame #2 and flame #1.

The deviation in the curve of flame #2 from linearity (in Fig. 5.19) at a fluence of 0.025

J/cm2 cannot be explained by vaporization, since the integrated LII curve (Fig. 5.14)

shows vaporization effects only at higher fluences. ( i.e. above 0.23 J/cm2). A possible

explanation of this phenomenon is laser-induced aggregate breakup. Further experimental

work is needed to investigate aggregate breakup. A better experimental setup to study

aggregate breakup would involve the use of a pump and probe laser. Therefore, the pump

laser can be varied in fluence, while the probe laser (CW preferably) independently

monitors the scattering at a fixed laser fluence. This isolates the change in scattering

signal due to morphological changes in the system from the change in scattering due to

the change in laser fluence. This technique is used in the study of soot in Chapter 4.

5.7 Conclusion

Particles of hematite are effectively produced in a seeded premixed flame, as verified by

X-ray diffraction of particle material and TEM images of particles deposited on a grid.

Page 242: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

224

Variation in flame conditions produces similar primary particle size distributions.

However, certain conditions will allow for particle aggregation while other conditions do

not. The particle size distributions for grid sampling show good agreement with the

LII-derived bimodal (normal and lognormal) particle size distributions for both flame

conditions. LII-derived particle size distributions indicate a bi (or multi)-modal

distribution for both flame conditions, with a peak total volume distribution for the

smaller particle size mode one order of magnitude larger than the larger particle size

modes. Grid sampling measurements indicate a few relatively large particles in flame #1

that are not present in the TEM images of flame #2. A lack of statistics may be the cause

of this discrepancy. The origin of these sparse, larger particles is not well understood. For

the current experimental conditions (seed levels, flowrates, etc.), the maximum diameter

a hematite particle can grow to be is approximately 35 nm for a data acquisition height of

4 cm. A possible explanation for these large particles is particle growth on the burner

honeycomb, followed by periodic break away [Rosner 2001a].

A model based on the Melton model is developed and used with a novel search procedure

to back out particle distributions from time resolved LII data, for a known laser fluence,

within a known spectral region, and a known excitation wavelength. The primary particle

size distribution mode shows good agreement with TEM measurements. The fit of the LII

curve modeled from a bimodal distribution to the data gives a better fit than the fit of the

Page 243: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

225

LII curve modeled from single mode distributions. This technique is limited to particle

size distributions of at most two modes, laser fluences that produce negligible mass

vaporization, and particle sizes that have LII signal decay times sufficiently longer than

the decay time of the laser pulse. Also the time response of the detection system must be

fast enough to have sufficient number of sample points of the data.

Laser light scattering shows a fivefold increase in signal for the two flame conditions

used. The degree of aggregation in the second flame conditions that is determined from

LLS agrees well with the observed aggregation from TEM images. For the aggregated

particles, there is a region where the scattering signal is non-linear with laser fluences

below the fluence needed for vaporization, which may be indicative of aggregate

breakup. Further experiments are needed along with an improved setup with a pump-

probe system to separate the effects of increase in scattering due to laser fluence, and

decrease in scattering due to aggregate break-up.

Page 244: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

226

Chapter 6Summary and Conclusions

Several axisymmetric laminar flame systems are successfully characterized in terms of

particle and species concentrations, temperature, and particle size distribution through the

use of Rayleigh and spontaneous Raman scattering, laser light scattering, flame emission,

laser-induced incandescence and laser absorption techniques. The quantities derived from

the optical measurements show good agreement with the quantities obtained by non-

optical measurements and computations.

An axisymmetric steady and time-varying laminar methane diffusion flame is

characterized in terms of two-dimensional profiles of species concentrations,

temperature, and mixture fraction using Rayleigh and spontaneous Raman scattering.

Images of temperature and mixture fraction derived from the two scalar technique show

good agreement with the temperature and mixture fraction obtained with the multi-

species Raman scattering technique away from stoichiometric. The discrepancy in the

images around stoichiometric is due to the inherent assumptions of the two scalar

technique, which does not account for flame intermediates and differential diffusion.

Difference Raman scattering effectively eliminates the C2 and PAH fluorescence

inteference on the species Raman signals of oxygen, carbon monoxide and carbon

Page 245: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

227

dioxide. An empirical correction effectively eliminates the remaining interference seen

just rich of the flame front, which may possibly be enhanced PAH Raman scattering.

Further work is needed to investigate the source of this weak remnant Raman signal.

For the steady flame, two-dimensional profiles of major species concentrations and

temperature derived from Rayleigh and multi-species difference Raman measurements

show good agreement with the computations in terms of flame structure, flame length,

and absolute concentrations and temperature. For the forced flame, measurements

indicate a larger modulation than indicated by the computations. For the experiments,

certain phases of the forcing indicate a distinct region where the flame "pinches off". The

computations indicate a significantly less defined "pinch off" region. The "pinch off"

phenomenon is more pronounced as the flow modulation increases from 30% to 50%.

The region downstream of the "pinch" has significant elastic scattering interference

where Rayleigh data cannot be interpreted. This indicates significant soot production in

the forced flame, as opposed to the insignificant amounts of soot produced in the steady

flame. Further work is needed to investigate the soot produced at these specific phases of

the forcing, possibly with the use of LII, LLS, and laser absorption.

An axisymmetric, laminar, sooting, ethylene diffusion flame is characterized in terms of

two-dimensional profiles of soot volume fraction and temperature, and point

Page 246: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

228

measurements of soot particle size distributions and aggregate information using

Rayleigh and fuel Raman scattering, LII, LLS, and laser absorption. A temperature image

derived from the two scalar technique shows good agreement with thermocouple probe

measurements and computations of temperature in terms of flame length, lift-off height

and flame structure. The peak flame temperature from the two scalar technique is 7%

lower than the computations and thermocouple measurement. A region of elastic

scattering interference from soot prevents optical measurements of temperature in this

region.

A two-dimensional profile of soot volume fraction obtained from time-integrated LII

shows similar structure near the peak signal region compared to probe measurements and

computations. The probe and LII measurements peak along the centerline while the

computations peak along the wings. The lack of detectable soot from LII in the wings of

the flame may indicate the soot in this region of the flame is transparent to the laser.

Soot particle size distributions obtained from time-resolved LII signals show good

agreement with particle sampling-derived particle size distributions in terms of peak

location and size distribution. Both measurements produce a peak soot particle size

distribution near 20-22 nm in diameter. Sensitivity analysis reveals that the particle size

prediction by the model is quite sensitive to parameters such as thermal accommodation

Page 247: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

229

coefficient and refractive index. In future work, The model used to interpret the LII

signals can be improved by including aggregate effects on the LII signal such as heat

shielding and enhanced energy absorption.

Time-resolved LLS and laser absorption experiments indicate a laser fluence region that

alters the soot aggregate structure while not causing significant mass vaporization.

Further work is needed to investigate the effect of the laser on soot aggregate structure.

One should be able to estimate the bond strength of the aggregates from this technique.

Hematite nanoparticles produced in an axysymmetric premixed laminar flame are

characterized in terms of point measurements of particle size distributions and aggregate

information using time-resolved LII and LLS. Two flame conditions produce similar size

particles, with one flame producing isolated spherules and the other producing

aggregates, as observed in the TEM pictures of particle sampling measurements. Both the

LII-derived distribution and grid sampling distributions determine the major primary

particle size distribution mode to be 5-15 nm in diameter for both flames. The best fit to

the LII curve for both flames assumes a bi-modal lognormal distribution, where the

second mode has a particle size distribution of 20-50 nm in diameter for both flame

conditions, and is approximately an order of magnitude smaller in total particle volume

than the major size distribution mode. Grid sampling measurements indicate a relatively

Page 248: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

230

few large and intermediate particles in the samples for the conditions producing isolated

spherules, but these particle are not observed in the samples for conditions producing

aggregates.

The grid sampling particle size distribution for the flame conditions producing isolated

spherules is used to generate an LII curve to compare to the experimental LII curve.

Results indicate extreme sensitivity of the LII signal to the addition of a relatively few,

relatively large diameter particles. It is therefore recommended to find the right

conditions or compound to produce a single mode distribution since this will simplify the

analysis of the LII data.

Laser and optical diagnostic techniques will have economic, environmental, and scientific

importance in the current combustion technologies, and the new technologies that

develop every day. Large combustion driven generators and propulsion systems will

employ in-situ monitoring and feedback systems containing optical diagnostics to

increase efficiency. In contrast, the miniaturization of technology produces a increasing

need for laser and optical diagnostics on the microscale and nanoscale.

Page 249: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

231

References

Avallone, E.H., Baumeister III, T., Marks’ Standard Handbook for Mechanical

Engineers, McGraw-Hill, New York (1996).

Beretta, F., Cincotti, V., D’Alessio, A., Menna, P., Combust. Flame, 61:211-218 (1985).

Bilger, R.W., Stårner, S.H., and Kee, R.J., “On Reduced Mechanisms for Methane-AirCombustion in Nonpremixed Flames,” Combust. Flame, 80:135-149 (1990).

Bohren, C.F. and Huffman, D.R. Absorption and Scattering of Light by Small Particles,Wiley, New York, pp. 69-81 (1983).

Cetegen, B.M. and Ahmed, T.A., “Experiments on the periodic instability of buoyantplumes and pool fires,” Combust. Flame, 66: 81-86 (1993).

Chen, L.D. and Roquemore, W.M., “Visualization of jet flames,” Combust. Flame 66:81-86 (1986).

Dasch, C. J., “One-dimensional tomography: a comparison of Abel, onion-peeling, andfiltered backprojection methods,” Appled Optics 31: 1146-1152 (1992).

Dibble, R.W., Masri, A.R., and Bilger, R.W., “The Spontaneous Raman ScatteringTechnique Applied to Nonpremixed Flames of Methane,” Combust. Flame

67:189 (1987).

Dobbins, R.A. and Megaridis, C.M., “Absorption and Scattering of Light by Polydisperseaggregates,” Appl. Optics, 30:4747-4754 (1991).

Eckbreth, A. C., Laser Diagnostics for Combustion Temperature and Species. AbacusPress, Cambridge, Massachusetts (1996).

Page 250: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

232

Eckbreth, A.C., “Effects of Laser-Modulated Particulate Incandescence on RamanScattering Diagnostics,” J. Appl. Phys. 48: 4473 (1977).

Eisner, A.D. and Rosner, D.E, “ Experimental Studies of Soot ParticleThermophoresis in Non-Isothermal Combustion Gases Using Thermocouple

Response Techniques,” Comb. and Flame 61:153-166 (1985).

Farias et al., ASME Trans., J Heat Transfer 117:158 (1995).

Fielding, J., “ Two-dimensional scalar measurements for turbulent flamecharacterization,” Thesis, Yale University (2001).

Filippov, A. V., Markus, M. W., and Roth, P., “In-situ characterization of ultrafineparticles by laser-induced incandescence: sizing and particle structuredetermination,” Journal of Aerosol Science, 30:71-87 (1999).

Filippov, A.V., and Rosner, D.E., “Energy transfer between an aerosol particle and gas athigh temperature ratios in the Knudsen transition regime,” Internat, J. Heat and

Mass Transf. 43:127-138 (2000a).

Filippov, A V. Zurita, M. Rosner, D E., “Morphology and physical properties of fractal-like aggregates,” Journal of Aerosol Science. 31 SUPPL. 1:S92-S93 (2000b).

Frank, J.H., Lyons, K.M., Marran, D.F., Long, M.B., Stårner, S.H., and Bilger, R.W.(1994), “Mixture Fraction Imaging in Turbulent Nonpremixed HydrocarbonFlames,” Twenty-Fifth Symposium (International) on Combustion, TheCombustion Institute, pp. 1159-1166.

Gaydon, A.G., The Spectroscopy of Flames, 2ed., John Wiley & Sons, New York(1974).

Hamins, A., Yang, J.C., Kashiwagi, T., “An experimental investigation of the pulsationfrequency of flames,” Twenty-fourth Symposium (International) on Combustion,The Combustion Institute, 1998, pp. 1695-1702 (1992).

Page 251: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

233

Hassel, E. P., 1996, RAMSES Spectral Synthesis code, University of Darmstadt,[email protected]

Hertz, H. M., and Faris, G. W., “Emission Tomography of Flame Radicals,” Optics

Letters 13:351-353 (1988).

Holzer, W., Le Duff, Y., and Altmann, K., “J Dependence of the Depolarization Ratio ofthe Rotational Components of the Q branch of H2 and D2 Raman Band,” J. Chem.

Phys., 58:642 (1973).

Hsu, W.P., and Matijevic, “Optical properties of monodisperse hematite hydrosols,”Appl. Optics 24:1623-1630 (1985).

Hughey, B. J., and Santavicca, D. A., “Comparison of techniques for reconstructingaxisymmetric reacting flow fields from absorption measurements,” Combustion

Science and Technology 29:167-180 (1982).

JANAF Thermochemical tables, J. Phys Chem. Ref. Data, 14, Suppl 1 (1985).

Keane, R.D. and Adrian, R.J., “Theory of cross-correlation analysis of PIV images,”Appl. Scient. Research, 49: 191-215 (1992).

Kelman, J.B., Masri, A.R., Stårner, S.H., Bilger, R.W. (1994), “Wide-Field ConservedScalar Imaging in Turbulent Diffusion Flames by a Raman and RayleighMethod,” Twenty-Fifth Symposium (International) on Combustion, TheCombustion Institute, pp. 1141-1147.

Kent, J.H., and Wagner, H. Gg., “Soot particle measurements in diffusion

flames,”Combustion and Flame, 47:53-65 (1982).

Page 252: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

234

Köylu, Ü.Ö., “Quantitative analysis of in situ optical diagnostics for inferring particle/aggregate parameters in flames: implications for soot surface growth and totalemissivity,” Combust. Flame, 109:488-500 (1996).

Köylu, Ü.Ö., “Radiative properties of flame-generated soot,” J. Heat Transfer, 115:409

(1993).

Köylu et al, “Simultaneous measurements of soot volume fraction and particlesize/microstructure in flames using a thermophoretic sampling technique,”Combust. Flame, 110:494-507 (1997).

Lee, S. C. and Tien, C. L., Symposium (International) on Combustion. 1981, pp. 1159-1166 (1981).

Leroy, O., Perrin, J., Pealat, M., Jolly, J., “ Thermal accommodation of a gas on a surfaceand heat transfer in CVD and PECVD experiments,” J. Phys D: Appl. Phys., 30:

499-509 (1997).

Lin, P., “Quantitative Charactarization of a Laminar Axisymmetric Nitrogen DilutedMethane/Air Diffusion Flame,” Thesis, Yale University (1995).

Long, D.A., Raman Spectroscopy, McGraw-Hill, New York (1993).

Long, M.B., in Instrumentation for Flows with Combustion (A.M.K.P. Taylor, Ed.),467, Academic Press (1993).

Lynch, J.F, Ruderer, C.G., Duckworth, W.H. (eds.), Engineering Properties of Selected

Ceramic Materials, American Ceramic Society (1966).

Mackowski, D.W., Appl. Optics 34:3535-3545 (1995).

Page 253: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

235

Marran, D.F., “Quantitative Two-Dimensional Laser Diagnostics in Idealized and

Practical Combustion Systems,” Thesis, Yale University (1997).

Markowski, G.R. , “Improving Twomey’s algorithm for inversion of aerosol particle

measurement data,” Aerosol Sci. Technol., 7:127-141 (1987).

Masri, A. R., Bilger, R. W., and Dibble, R. W., “‘Fluorescence’ Interference with RamanMeasurements in Nonpremixed Flames of Methane,” Combust. Flame 68:109(1987).

Maxworthy, T., “The flickering candle: transition to a global oscillation in a thermalplume,” J. Fluid Mech., 390:297-323 (1999).

McEnally, C S. Schaffer, A M. Long, M B. Pfefferle, L D. Smooke, M D. Colket, M B.Hall, R J., “Computational and experimental study of soot formation in a coflow,

laminar ethylene diffusion flame,” Symposium (International) on Combustion. v 11998. pp. 1497-1505.

Melling, A.,”Tracer particles and seeding for particle image velocimetry,” Meas. Sci.

Technol., 8:1406-1416 (1997).

Melton, L.A., “Soot diagnostics based on laser heating,” Appl. Optics. 23:2201 (1984).

Mewes, B., and Seitzman, J. M., “Soot volume fraction and particle size measurementswith laser-induced incandescence,” Applied Optics, 36:709-717 (1997).

Miles, P., Carbon Dioxide Spectral Synthesis code, Sandia National Labs (1996).

Page 254: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

236

Murphy, W. F., “The Rayleigh Depolarization Ratio and Rotational Raman Spectrum ofWater Vapor and the Polarizability Components for the Water Molecule,” J.

Chem. Phys. 67:5877 (1977).

Najm, H.N., Paul, P.H., Mueller, C.J., and Wyckoff, P.S., “On the adequacy of certainexperimental observables as measurements of flame burning rate,”Combustion

and Flame, 113:312-332 (1998).

Ni, T., Pinson, J.A., Gupta, S., Santoro, R.J., “Two-dimensional imaging of soot volumefraction by the use of laser-induced incandescence,” Appl. Optics, 34: 7083(1995).

Osborne, R. J., Brown, T. M., Pitz, R. W., Tanoff, M. A., Smooke, M. D., “Study ofStructure and Emissions of Partially-Premixed Methane Flames ins LaminarCounterflow,” AIAA 96-0212 (1996).

Penney, C. M., Goldman, L. M., and Lapp, M., “Raman Scattering Cross Sections,”

Nature Phys. Sci. 235:110 (1972).

Petarca, L, and Marconi, F, “Fluorescence Spectra and Polycyclic Aromatic Species in aN-Heptane Diffusion Flame,” Combust. Flame 78:308 (1989).

Pinson, R.A., “Quantitative, planar soot measurements in a D.I. diesel engine using laser-induced incandescence and light scattering,” SAE paper 932650 (Society ofAutomotive Engineers, Warrensdale, PA, 1993).

Quay, B., Lee, T.W., Ni, T., Santoro, R.J., “Spatially-resolved measurements of sootvolume fraction using laser-induced incandescence,” Combust Flame 97:384-392

(1994).

Reckers, W., Huwel, L., Grunefeld, G., Andresen, P. Spatially-resolved multi-species andtemperature analysis in hydrogen flames. Applied Optics 32: pp. 907-18 (1993).

Page 255: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

237

Rosner, D.E. Private communication (2001a).

Rosner, D.E., Transport Processes in Chemically Reacting Flow Systems, DoverPublications (2000).

Rosner, D. E. , Schaffer, A.M., Long,, M.B. and La Mantia, B., “In-Situ, Real-TimeSizing Of Inorganic Nano-Particle Populations In Flames Using Laser-InducedIncandescence,” To be presented at 20th Annual AAAR Conference, October 15-19, 2001, Portland, Oregon USA (2001b)

Rosner, D.E., Mackowski D.W. and Garcia-Ybarra, P., “Size- and Structure-Insensitivityof the Thermophoretic Transport of Aggregated Soot Particles in Gases,”Combustion Science and Technology 80: 1-3, 87-101 (1991).

Rowell, R. L., Aval, G. M., Barrett, J. J., “Rayleigh-Raman Depolarization of Laser LightScattered by Gases,” J. Chem. Phys. 54:1960 (1971).

Santoro, R.J., Semerjian, H.G., and Dobbins, R.A. Combust. Flame 51:203-218 (1983).

Schrötter, H. W., and and Klöckner, H. W., “Raman Scattering Cross Sections in Gasesand Liquids” in Raman Spectroscopy of Gases and Liquids (A. Weber, ed),Springer-Verlag, Berlin (1979).

Skaggs, R.R. and Miller, J.H., “Tunable Diode Laser Absorption Measurements ofCarbon Monoxide and Temperature in a Time-Varying, Methane/Air, Non-premixed Flame,” Twenty-sixth Symposium (International) on Combustion, TheCombustion Institute, 1996, p. 1181.

Smooke, M. D., Ern, A., Tanoff, M. A., Valdati, B. A., Mohammed, R. K., Marran, D. F.,Long, M. B., “Computational and Experimental Study of Nitric Oxide in anAxisymmetric Laminar Diffusion Flame,” Twenty-sixth Symposium

(International) on Combustion, The Combustion Institute, 1996, pp 2161-2170.

Page 256: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

238

Smooke, M. D., Xu, Y, Zurn, R. M., Lin, P., Frank, J. H., Long, M. B., “Computationaland Experimental Study of OH and CH Radicals in Axisymmetric LaminarDiffusion Flames” Twenty-fourth Symposium (International) on Combustion, TheCombustion Institute, 1992, p. 813.

Smooke, M. D., Lin, P., Lam, J. K., Long, M. B., “Computational and ExperimentalStudy of a Laminar Axisymmetric Methane-Air Diffusion Flame” Twenty-third

Symposium (International) on Combustion, The Combustion Institute, 1990, p.575.

Smyth, K.C., Harrington,J.E., Johnson,E.L., and Pitts,W.M., “Greatly enhanced sootscattering in flickering CH4/Air Diffusion Flames,” Combust. Flame, 95:229-239(1993).

Smyth, K. C., Miller, J. H., Doorfman, R. C., Mallard, W. G., and R. J. Santoro, “SootInception in a Methane / Air Diffusion Flame as Characterized by Detailed

Species Profiles,” Combustion and Flame, 62:157 (1985).

Stårner, S.H., Bilger, R.W., Lyons, K.M., Frank, J.H., Long, M.B., “Conserved ScalarMeasurements in Turbulent Diffusion Flames by a Raman and Rayleigh RibbonImaging Method,” Combust. Flame, 99:347-354 (1994).

Stårner, S.H., Bilger, R.W., Frank, J.H., Marran, D.F., and Long, M.B., “MixtureFraction Imaging in a Lifted Methane Jet Flame,” Combust. Flame, 107:307-313(1996)

Thorn, R.J. and Winslow, G.H., “Vaporization Coefficient of Graphite and Computation

of Equilibrium Vaporization,” J. Chem. Phys., 26:186, (1957).

Twomey, S. Introduction to the Mathematics of Inversion in Remote Sensing and Indirect

Measurement. Elsevier, Amsterdam (1977).

Page 257: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

239

Vander Wal, R. L., and Weiland, K. J., “Laser-induced incandescence: Development andcharacterization towards a measurement of soot-volume fraction,” Applied

Physics B, 59:445-452 (1994).

Vander Wal, R. L., “Laser-induced incandescence: detection issues,” Applied Optics,

35:6548-6559 (1996).

Vander Wal, R.L et al., “Laser-induced incandescence applied to metal nanostructures,”Applied Optics, 38:5867 (1999).

Vander Wal, R. L., Ticich, T. M., and Stephens, A. B., “Optical and microscopyinvestigations of soot structure alterations by laser-induced incandescence,”Applied Physics B, 67:15-123 (1998).

Wainner, R.T. and Seitzman, J.M., “Soot diagnostics using laser-induced incancdescencein flames and exhaust flows,” 37th Aerospace Sciences Meeting, Reno, NV, Jan

11-14, 1999, paper AIAA 99-0640.

Walsh, K.T., “ Effect of light-collection geometry on reconstruction errors in Abelinversions,” Optics Letters 25:457-9 (2000).

Will, S., Schraml, S., Leipertz, A., “Two-dimensional soot particle-sizing by time-resolved laser-induced incandescence,” Optics Letters, 20:2342 (1995)

Woodward, L. A., “General Introduction” in Raman Spectroscopy, Theory and Practice

(H. A. Szymanski, ed), Plenum Press, New York (1967).

Xing, Y., Koylu, U.O. and Rosner, D.E., “In Situ Light ScatteringMeasurements of Morphologically Evolving Flame Synthesized OxideNano-aggregates,” Appl. Optics 38:2686-2697(1999).

Xu, Y., Smooke, M. D., Lin, P., and Long, M. B., “Primitive Variable Modeling ofMultidimensional Laminar Flames,” Comb. Sci. Tech., 90:289 (1993).

Page 258: ABSTRACT - Yale Universityguilford.eng.yale.edu/pdfs/as_thesis.pdf · ABSTRACT Quantitative Characterization of Species, Temperature, and Particles in Steady and Time-Varying Laminar

240

Xu, Y., Numerical Calculations of an Axisymmetric Laminar Diffusion Flame wih

Detailed and Reduced Reaction Mechanisms, Thesis, Yale University (1991).