OH-PLIF MEASUREMENTS AND ACCURACY INVESTIGATION IN HIGH PRESSURE GH 2 /GO 2 COMBUSTION

Transcription

1 OH-PLIF MEASUREMENTS AND ACCURACY INVESTIGATION IN HIGH PRESSURE GH 2 /GO 2 COMBUSTION By ARAVIND VAIDYANATHAN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA

2 2008 Aravind Vaidyanathan 2

3 To my Guru Sainath of Shirdi 3

4 ACKNOWLEDGMENTS I express my sincere gratitude to my advisor, Dr. Corin Segal, for giving me the opportunity to do research under his valuable guidance and providing me with moral support and encouragement during the ups and downs of my graduate studies. I am also grateful to all the members of the PhD advisory committee for their critical evaluation and valuable suggestions on my research work. I am indebted to Dr. Jonas Gustavsson for his continued patience and guidance like an elder brother. I thank all my colleagues in the Combustion and Propulsion Laboratory; moreover working with people of diverse cultural background is a memorable experience. I am grateful to all my friends and relatives for their continued support and encouragement. I also express my sincere gratitude to my Master of Science advisor Prof. Job Kurian of IIT Madras, India and all my teachers who have helped me push the limits of my thinking and imagination. Finally I am extremely thankful to my parents for their endless support to me in pursuing higher education. This work has been performed with the support from NASA Grant NCC3-994 with Claudia Meyer as the Program Manager. 4

5 TABLE OF CONTENTS ACKNOWLEDGMENTS...4 LIST OF TABLES...7 LIST OF FIGURES...8 NOMENCLATURE...12 ABSTRACT...16 CHAPTER 1 INTRODUCTION...17 page Hydroxyl Radical (OH) in Non-premixed Flames...25 Motivation for the Current Work OH PLANAR LASER INDUCED FLUORESCENCE - THEORY AND REVIEW...29 Fluorescence Modeling...29 Fluorescence and Interference Signals...37 Laser...38 Absorption and Excitation, Line Shape and Fluorescence Efficiency...38 Experimental Constants...38 Review of OH PLIF Diagnostic Studies...39 Fluorescence Strategy and Interference Signals...64 Laser...64 Absorption & Excitation, Line Shape and Fluorescence Efficiency...65 Experimental Constants EXPERIMENTAL FACILITY AND DIAGNOSTICS METHODS...68 Experimental Test Facility and Operating Conditions...68 OH-PLIF Diagnostics...72 Wall Boundary Conditions OH-PLIF IMAGE PROCESSING AND QUANTITATIVE ANALYSIS...77 Fluorescence and Interference Signals...77 Laser...82 Absorption and Excitation, Line Shape, and Fluorescence Efficiency...84 Experimental Constants

6 5 RESULTS AND UNCERTAINTY ANALYSIS...89 Chamber Pressure Measurements...89 OH-PLIF Measurements...92 Quantification of OH Concentration and Uncertainty at 10, 27, 37 and 53 bar CONCLUSIONS FUTURE WORK APPENDIX A B C MATLAB SCRIPTS USED FOR DATA PROCESSING PROPOSED NEW METHODOLOGY FOR PHOTON CALIBRATION OH ABSORPTION PROFILES OH Absorption Profiles at 10 bar and K Temperature Range OH Absorption Profiles at 27 bar and K Temperature Range OH Absorption Profiles at 37 bar and K Temperature Range OH Absorption Profiles at 53 bar and K Temperature Range D OH NUMBER DENSITY CONTOURS Thirteen Instantaneous OH Number Density Contours at 10 bar Thirteen instantaneous OH Number Density Contours at 27 bar Thirteen Instantaneous OH Number Density Contours at 37 bar Thirteen Instantaneous OH Number Density Contours at 53 bar E TEMPERTURE MEASUREMENTS AND BOUNDARY CONDITIONS LIST OF REFERENCES BIOGRAPHICAL SKETCH

7 LIST OF TABLES Table page 1-1 Previous Experimental Studies on Rocket Injectors Review of OH-PLIF Diagnostics Experimental Operating Conditions Colliding Species Cross Section for Collisional Quenching

8 LIST OF FIGURES Figure page 1-1 Chamber wall cracks due to local heating. Blanching indicates regions of insufficient wall cooling Comparison of CFD predicted wall heat flux measurements with experimental results Two-State Quasi-Steady Two-Step Modeling of Fluorescence Physical significance of the terms in OH number density expression Pressure range in the reviewed studies Combustion Chamber Cross Section Injector Details Laser spectral profile measured using Burleigh Wavemeter before doubling to 283 nm OH-PLIF Experimental Set-up Average of 13 instantaneous images obtained at near steady state for chamber pressure of 10 bar Average of 13 instantaneous images obtained at near steady state for chamber pressure of 27 bar Average of 13 instantaneous images obtained at near steady state for chamber pressure of 37 bar Average of 13 instantaneous images obtained at near steady state for chamber pressure of 53 bar Normalized laser sheet intensity profile variation obtained from acetone fluorescence images Camera calibration corresponding to the detection strategy employed in the OH- PLIF measurements and region of interest Chamber pressure versus time for GH 2 /GO 2 combustion for 10 bar and O/F mass flow of Chamber pressure versus time for GH 2 /GO 2 combustion for 27 bar and O/F mass flow of

9 5-3 Chamber pressure versus time for GH 2 /GO 2 combustion for 37 bar and O/F mass flow of Chamber pressure versus time for GH 2 /GO 2 combustion for 53 bar and O/F mass flow of Instantaneous image-processed OH-PLIF images at near steady state chamber pressure of (a) 10, (b) 27, (c) 37 and (d) 53 bar Average of thirteen instantaneous image-processed OH-PLIF images at near steady state chamber pressure of (a) 10, (b) 27, (c) 37 and (d) 53 bar Average of thirteen instantaneous image-processed OH-PLIF images at near steady state chamber pressure of (a) 35, (b) 36, and (c) 37 bar indicating the repeatability and reliability of OH-PLIF measurements for determination of OH concentration Mean position of reaction zone determined from the average OH-PLIF images at (a) 10, (b) 27, (c) 37 and (d) 53 bar Temperature and specie mole fraction variation based on equilibrium calculations with equivalence ratios of at (a) 10, (b) 27, (c) 37 and (d) 53 bar ' 5-10 Absorption coefficient ( f B B ) variation with equivalence ratio and temperature 12 1 ( K) at (a) 10, (b) 27, (c) 37 and (d) 53 bar showing that the variation with respect to mean is 12.4, 14.6, 14.5 and 15.1% respectively Absorption profile of OH at (a) 3017 K and 10 bar, (b) 3085 K and 27 bar, (c) 3103 K and 37 bar, and (d) 3125 K and 53 bar simulated using LIFBASE showing a complete overlap with the laser spectral profile at all pressures Overlap integral Φ Φ dν laser abs variation at (a) 10, (b) 27, (c) 37 and (d) 53 bar with temperature corresponding to equivalence ratio of 0.5 3, indicating that the variation with respect to mean is 1.3, 1, 0.8 and 0.5% respectively and can be assumed negligible Collisional quench rate Q 21 variation at (a) 10, (b) 27, (c) 37 and (d) 53 bar with temperature and colliding species mole fraction corresponding to equivalence ratio of indicating that the variation with respect to mean is 4.1, 3.9, 3.8 and 3.7 % respectively Instantaneous OH number density contours at near steady state chamber pressure of (a) 10, (b) 27, (c) 37 and (d) 53 bar Average of thirteen instantaneous OH number density contours at near steady state chamber pressure of (a) 10, (b) 27, (c) 37 and (d) 53 bar

10 5-16 OH-PLIF measurement uncertainties at (a) 10, (b) 27, (c) 37 and (d) 53 bar B-1 Calibration set-up for photon calibration B-2 A series of 900 images of 32x32 pixel size was obtained at each exposure B-3 A series of 900 images of 32x32 pixel size was obtained each exposure B-4 Counts vs exposure time at 532 nm B-5 Photons vs counts at 310 nm C-1 Absorption profile of OH simulated using LIFBASE at equivalence ratio of (a) 0.5, (b) 1, (c) 1.5, (d) 2, (e) 2.5 and (f) 3 corresponding to temperatures of K for gaseous H 2 -O 2 flame at 10 bar C-2 Absorption profile of OH simulated using LIFBASE at equivalence ratio of (a) 0.5, (b) 1, (c) 1.5, (d) 2, (e) 2.5 and (f) 3 corresponding to temperatures of K for gaseous H 2 -O 2 flame at 27 bar C-3 Absorption profile of OH simulated using LIFBASE at equivalence ratio of (a) 0.5, (b) 1, (c) 1.5, (d) 2, (e) 2.5 and (f) 3 corresponding to temperatures of K for gaseous H 2 -O 2 flame at 37 bar C-4 Absorption profile of OH simulated using LIFBASE at equivalence ratio of (a) 0.5, (b) 1, (c) 1.5, (d) 2, (e) 2.5 and (f) 3 corresponding to temperatures of K for gaseous H 2 -O 2 flame at 53 bar D-1 Thirteen instantaneous OH number density contours at near steady state chamber pressure of 10 bar D-2 Thirteen instantaneous OH number density contours at near steady state chamber pressure of 27 bar D-3 Thirteen instantaneous OH number density contours at near steady state chamber pressure of 37 bar D-4 Thirteen instantaneous OH number density contours at near steady state chamber pressure of 53 bar E-1 Chamber wall temperatures vs time at inner locations of 37, 47, 58, 70, 89 and 102 mm from the injector face E-2 Chamber wall temperatures vs time at middle locations of 37, 47, 58, 70, 89 and 102 mm from the injector face E-3 Chamber wall temperatures at inner and middle locations along the chamber wall at end of the 8 s

11 E-4 Exponential function assumed for heat flux evolution with time E-5 Experimental and computational temperatures at 37 mm axial location E-6 Experimental and computational temperatures at 47 mm axial location E-7 Experimental and computational temperatures at 58 mm axial location E-8 Experimental and computational temperatures at 70 mm axial location E-9 Experimental and computational temperatures at 89 mm axial location E-10 Experimental and computational temperatures at 102 mm axial location E-11 Chamber wall heat fluxes calculated based on 3D computations and linear + unsteady assumption at 37 bar E-12 Computational and Experimental Temperatures for 37 bar at the end of 8s

12 NOMENCLATURE A Electronic Excited State A laser Cross sectional area of the laser beam or sheet (cm 2 ) A Pixel Projection Pixel projection area (cm 2 ) A 21 Spontaneous emission rate (s -1 ) B 12 Einstein B coefficient for absorption (cm 3 J -1 s -2 ) B 21 Einstein B coefficient for emission (cm 3 J -1 s -2 ) ' B 12 2 B12 c (cm J -1 ) c Speed of light (cms -1 ) C Heat capacity (J kg -1 K -1 ) E E( v ) Laser energy per pulse (J) Laser spectral energy per pulse (Jcm) g 1 Degeneracy in the ground electronic state g 2 Degeneracy in the upper excited electronic state GO 2 GH 2 h I( v ) J Gaseous oxygen Gaseous hydrogen Planck s constant (Js) Laser spectral fluence (Wcm -2 cm) Jet momentum flux ratio k Thermal conductivity (W m -1 K -1 ) k B Boltzmann constant (J K -1 ) l Laser sheet thickness (cm) 12

13 LOx M Liquid Oxygen Molecular weight (g) n Total population density (cm -3 ) n 1 Population density in the ground state (cm -3 ) n 2 Population density in the excited state (cm -3 ) o n Total number density (cm -3 ) N p Number of photons OH-PLIF O/F P Hydroxyl Planar Laser-Induced Fluorescence Oxidizer / Fuel Pressure (bar) q A Heat flux per unit area (W m -2 ) Q 21 Collisional quench rate (s -1 ) Re D RET Reynolds number based on diameter Rotational energy transfer T Temperature (K, o C) T inner Temperature at 3.2 mm from inner wall (K, o C) T middle Temperature at 9.5 mm from inner wall (K, o C) U velocity (m/s) V Volume probed by the laser (cm 3 ) VET Vibrational energy transfer W 12 Stimulated absorption rate (s -1 ) W 21 Stimulated emission rate (s -1 ) 13

14 X Electronic ground state ΔT Temperature difference (K, o C) Δt Δx Time difference (s) Distance between temperature measurement locations Δ v c Collisional width (cm -1 ) Δ v D Doppler width (cm -1 ) Δ Collision induced shift (cm -1 ) v C shift Δ v D shift Doppler induced shift (cm -1 ) ν Wavenumber (cm -1 ) m i Reduced mass of OH and the colliding species s i Ω 4π τ l Colliding species cross section Fraction of solid angle Laser pulse duration (ns) f B Boltzmann factor f ( ) Normalized collisional line shape function (cm) c v f ( ) Normalized Doppler line shape function (cm) D v Φ () v Absorption line shape function (cm) abs Φ () v Laser spectral profile (cm) laser F F ( m m ) O2 H2 actual Fluorescence yield ( m m ) O2 H2 stoichiometric, equivalence ratio 14

15 ρ Density (kg m -3 ) c i Colliding species mole fraction 15

16 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy OH-PLIF MEASUREMENTS AND ACCURACY INVESTIGATION IN HIGH PRESSURE GH 2 /GO 2 COMBUSTION Chair: Corin Segal Major: Aerospace Engineering By Aravind Vaidyanathan August 2008 In-flow species concentration measurements in reacting flows at high pressures are needed both to improve the current understanding of the physical processes taking place and to validate predictive tools that are under development, for application to the design and optimization of a range of power plants from diesel to rocket engines. To date, non intrusive measurements have been based on calibrations determined from assumptions that were not sufficiently quantified to provide a clear understanding of the range of uncertainty associated with these measurements. The purpose of this work is to quantify the uncertainties associated with OH measurement in a oxygen-hydrogen system produced by a shear, coaxial injector typical of those used in rocket engines. Planar OH distributions are obtained providing instantaneous and averaged distribution that are required for both LES and RANS codes currently under development. This study has evaluated the uncertainties associated with OH measurement at 10, 27, 37 and 53 bar respectively. The total rms error for OH-PLIF measurements from eighteen different parameters was quantified and found as 21.9, 22.8, 22.5, and 22.9 % at 10, 27, 37 and 53 bar respectively. These results are used by collaborators at Georgia Institute of Technology (LES), Pennsylvania State University (LES), University of Michigan (RANS) and NASA Marshall (RANS). 16

17 CHAPTER 1 INTRODUCTION Over the past several decades, considerable effort has been dedicated for the development of rocket engine technology including the space shuttle main engine (SSME) which operates at pressures of 350 bar and a range of upper stage engines which operate with pressure ranges from several bars to fewer than 100 bar. Yet, considerable difficulties remain to develop a design tool that will adequately describe the physical processes occurring in the rocket engines. These predictive tools require validation through accurate experiments. An example of a current area of concern is illustrated by the photograph of the SSME injector face shown in Figure1-1 The cracks and blanching in the chamber wall near the outer row of the injectors is due to local uneven heating and must be corrected in future design. Figure 1-1. Chamber wall cracks due to local heating. Blanching indicates regions of insufficient wall cooling [Courtesy: Mr.Kevin Tucker, NASA Marshall Space Flight Center, Huntsville, AL] The consequences can be viewed as increased flight risk and maintenance costs and indicates that that there is still a need to better understand the combustion chamber dynamics. The most reliable method to accomplish this task is by the experimental study of the full scale 17

18 engines; however despite their reliability and robustness these experiments are costly. Hence, Computational Fluid Dynamics (CFD) is continuously being developed for future designs. The capabilities and limitations of CFD as a rocket injector design tool were addressed by Tucker et al. [1]. The major challenges currently faced in CFD are due to lack of adequate date base for the CFD validation. The expected performance of the CFD is such that the physical description of the problem will develop from a small scale simulation to near full prototype with continuously increased complexity and confidence [1, 2]. An example of the current status of the predictive capability is shown in Figure 1-2. CFD Comparison to Wall Heat Flux Measurements q" (MW/m^2) X (mm) Wall Heat Flux Measurements Team 1 Team 2; Calculation 1 Team 2; Calculation 2 Team 3 Team 4 Team 5 Team 6 Figure 1-2. Comparison of CFD predicted wall heat flux measurements with experimental results [Source: 3 rd International Workshop on Rocket Combustion and Modeling, Paris, March 2006]. The CFD predicted results of the six different groups are inconsistent with each other and quite inaccurate when compared to experiment. The plots in Figure 1-2 show the comparison of wall heat fluxes results obtained from various CFD groups with the experiments. The CFD predicted results of the six different groups 18

19 are inconsistent with each other and quite inaccurate when compared to experiments. This shows that considerable improvements need to be made in the predictive capabilities of the CFD tool. Tucker et al. [1] indicated the necessity to obtain experimental database for a single element gas-gas injector for code validation and optimization of the injector performance. According to the authors [1] the single element design, referred to as the baseline design, can be used to model performance and environmental indicators as function of the geometric variables like orifice sizes, post tip thickness and cup details of the injector. Moreover the simplicity to run a CFD code for a single element injector for code validation and subsequent improvement in the code before validating more complex configurations were also addressed in detail. In the study conducted by Calhoon et al. [3] a systematic approach to investigate and characterize high performance injectors are explained in detail. The importance of single element injector small scale testing, which gradually paved ways to multi element full scale testing of rocket engines was also emphasized. The importance and relevance of gas-gas injector for the development of gas-liquid injector technology was further discussed by Schley et al. [4] who indicated that the accurate prediction of gas-gas system using the CFD codes is necessary before applying the CFD codes to predict gas-liquid system. Clearly, the accurate prediction of the gas-gas system is not a sufficient condition to predict gas-liquid system but is a necessary preliminary step before the inclusion of additional complexities like accurate treatment of atomization and spray combustion. The gas-gas single element dataset consists of inflow measurements of species concentration, temperature and velocity; temperature boundary conditions at inlet and exit of the combustion chamber; wall heat transfer boundary conditions; 19

20 A brief review of the existing experimental data focused on the inflow species measurements for coaxial injector studies is tabulated in Table 1-1 and covers rocket injector studies in the past years. The reviews clearly indicate the lack of adequate inflow quantitative species measurement with a thorough uncertainty analysis. Furthermore, when evaluated, the uncertainties shown in Table 1-1 indicate that considerable work remains to be done to improve the existing accuracy so that the database may be useful to support code validation. 20

21 21 Table 1-1. Previous Experimental Studies on Rocket Injectors Ref. Injector type Chamber Parameters Experimental Pressure Method (bar) Foust et al. [5] Foust et al. [6] Brummund et al. [7] Mayer et al. [8] Yeralan et al. [9] Wehrmeyer et al. [10] Single element shear (GH 2 /GO 2 ) Single element shear, swirl (GH 2 /GO 2 ) Single element shear (LOx/GH 2 ) Single element shear (LOx/GH 2 ) Single element swirl (LOx/GH 2 ) Single element swirl (LOx/GH 2 ) 13 Inflow velocity and species concentration (H 2 O, H 2, O 2 ) Inflow species concentration (H 2 O, H 2, O 2 ) 20 Inflow species visualization (OH) Jet and flame visualization 28 Inflow species concentration (H 2 O, H 2, O 2 ) and temperature 60 Inflow species visualization (H 2 O, H 2, O 2 ) LDV for velocity and Raman spectroscopy for species Raman spectroscopy Planar Laser Induced Predissociation Fluorescence (PLIPF) Shadowgraph, Flame emissions Raman spectroscopy Raman spectroscopy Species Quantification Mole fraction of H 2 O,H 2 and O 2 Mole fraction of H 2 O,H 2 and O 2 Signal intensity (qualitative) Signal intensity (qualitative) Mole fraction of H 2 O, H 2 and O 2. Signal intensity (qualitative) Uncertainty Source (% error) (i) Non-linear temperature dependence of Stoke band factor (40) Rms error (%) 40 (i) Laser pulse energy fluctuation(5), 45 (ii) Non-linear temperature dependence of Stoke band factor (45) (i)calibration measurements (40), (ii)shot noise

22 22 Table 1-1. Continued. Ref. Injector type Chamber Pressure (bar) Herding et al. [11] Candel et al. [12] Ivancic et al. [13] Juniper et al. [14] Mayer et al. [15] Yeralan et al. [16] Single element shear (LOx/GH 2 ) Single element shear (LOx/GH 2 ) Single element shear (LOx/GH 2 ) Single element shear (LOx/GH 2 ) Single element shear (LOx/GH 2 ) Single element swirl (LOx/GH 2 ) Parameters 1 10 Inflow species visualization (OH) 10 Inflow species visualization (OH, O 2 ) Temperature 60 Inflow species visualization (OH), Temperature 70 Inflow species visualization (OH) Jet and flame visualization 28 Inflow species concentration (H 2 O, H 2, O 2 ) and temperature Experimental Method OH emissions # PLIF for OH and O 2. CARS for temperature OH emissions CARS for temperature OH emissions Shadowgraph, Flame emissions Raman spectroscopy Species Quantification Signal intensity (qualitative) Signal intensity (qualitative) Signal intensity (qualitative) Signal intensity (qualitative) Signal intensity (qualitative) Mole fraction of H 2 O, H 2 and O 2. Uncertainty Source (% error) (i)calibration measurements (19), (ii)shot noise(10) Rms error (%) 22

23 23 Table 1-1. Continued. Ref. Injector type Chamber Pressure (bar) Mayer et al. [17] Kalitan et al. [18] Singla et al. [19] Singla et al. [20] Single element shear (LOx/GH 2 ) Single element swirl (LOx/CH 4 ) Single element shear (LOx/CH 4 ) Single element shear (LOx/GH 2 ) Parameters 63 Jet and flame visualization 41 Inflow species (OH, CO 2 ) and jet visualization 1 70 Inflow species visualization (OH, CH) 63 Inflow species concentration (OH) and visualization (OH) Experimental Method Shadowgraph, OH emissions OH visualization by PLIF and emission images, CO 2 by emission images and jet visualization by shadowgraph and laser light scattering OH and CH emissions PLIF for OH concentration and OH emissions for flame visualization Species Quantification Signal intensity (qualitative) Signal intensity (qualitative) Signal intensity (qualitative) Signal intensity ( semi quantitative) * Uncertainty Source (% error) (i) Boltzmann fraction variation in K temperature range (10), (ii) laser beam absorption by OH(10 30) and (iii)variation in quench rate due to species and temperature variation Rms error (%) 32

24 24 Table 1-1. Continued. Ref. Injector type Chamber Pressure (bar) Singla et al. [21] Smith et al. [22] Vaidyanathan et al. [23] Single element shear (LOx/CH 4 ) Single element shear (LOx/GH 2 ) Single element shear (GO 2 /GH 2 ) Parameters Inflow species visualization (OH) Inflow species (OH) and jet visualization Inflow species concentration (OH) Experimental Method PLIF for OH visualization Shadowgraph, OH emissions PLIF for OH concentration Species Quantification Signal intensity (qualitative) Signal intensity (qualitative) Mole fraction of OH Uncertainty Source (% error) (i) UV PAH fluorescence and OH fluorescence are of same intensity at bar - - (i) Boltzmann fraction variation in K temperature range (15), (ii) laser beam absorption by OH over a distance of 3 mm(8) # PLIF Planar Laser Induced Fluorescence * Singla et al. [20] provided semi-quantitative OH distribution in signal intensities without converting them to the actual number densities. Additional error sources which typically originate from photon calibration, shot noise, spatial variation of camera sensitivity and spatial variation in laser sheet intensity profiles were not addressed. One of the main objectives of the study carried out by Singla et al. was to provide OH distribution for CFD validation. Rms error (%) - 17

25 From the previous experimental studies tabulated in Table 1-1, it can be seen that only one third of them addressed the uncertainties associated with the measurements and only a limited number of factors have been included. A comprehensive and thorough investigation of the uncertainties associated with the inflow measurements is clearly needed. This is the primary motivation of the present work. Before discussing the motivation of the current work, the importance of hydroxyl radical measurement in non premixed flames is reviewed. Hydroxyl Radical (OH) in Non-premixed Flames In the injector vicinity of a non-premixed flame the OH radical is present in the reaction zone of the fuel-oxidizer shear layer jets and is, therefore, a good flame marker [24 32]. Seitzman et al. [25] characterized OH structures in turbulent non-premixed hydrogen flames and found that the OH was confined to the flame as a thin structure at the base of the flame and was also found in the diffuse regions near the tip of the flame where the hot product gases existed. According to Barlow et al. [27] OH concentration peaks near the stoichiometric condition in hydrogen flames. In this study [27] the equivalence ratio in the shear layer of supersonic and subsonic jets varied between The authors opined that since the stoichiometric contour is often separated from the centre of the shear layer in turbulent diffusion flames, the OH fluorescence can be a good reaction zone marker. In this study [27] the growth and relative widths of shear layer for both compressible and incompressible flow were determined based on the OH measurements. Clemens and Paul [28] also discussed the use of OH as reaction zone marker. According to the authors [28] the OH can also appear as a product in lower temperature regions due to its relatively slow three-body recombination reaction, H+OH+M H 2 O + M, M being the third 25

26 body. However, these regions appear as distributed and diffused OH zones when compared to the thin laminar like filament structures in the primary reaction zone. Thus the appearance of OH in the shear reaction zones represents the flame front and could be used to mark the reaction zones in the GH 2 /GO 2 combustion carried out in the current study. Similarly Ivancic et al. [13] in the study of time and length scales in LOx/GH 2 rocket combustors found out that the OH emissions present on the symmetry line in the near injector regions come from the OH radicals produced within the reaction zone. Donbar et al. [30] identified the reaction zone structures in a turbulent non-premixed methane jet flames based on CH-OH PLIF images. According to the authors [30] if the wrinkling in the flame is not severe, the fuel rich boundary of the OH zone can be identified and used as the stoichiometric contour. The stoichiometric contour in this study was identified as existing in a thin zone in the gap between CH and OH regions. The stoichiometric contours were used to determine the flame surface density and degree of flame wrinkling. The visualization of reaction zone from OH-PLIF images is mentioned in the work done by Pickett et al. [31]. According to these authors in non-premixed flames OH is consumed in the fuel rich region and hence the OH zone is confined to the flame whereas in the case of premixed flames, OH continues to exist in high temperature product regions. Singla et al. [21, 22] cites the importance of OH radical in high pressure cryogenic flames as representing the characteristics of combustion reactions, presence in high temperature stoichiometric regions and flame-front marker. Experimental investigation of the effects of heat release in a subsonic turbulent planer H 2 jet was done by Theron et al. [32]. In this study H 2 was injected through the central rectangular slot whereas air was supplied from the upper and lower channels above and below the slot 26

27 respectively. The OH radical was tracked by fluorescence technique and the mean position of the reaction zone was identified as the position of maximum OH fluorescence signal intensity from the centre line along the test section height. The axial evolution of the mean position of the reaction zone was represented as the stoichiometric contour of maximum temperature. These studies clearly identified the usefulness of tracking OH in non-premixed flames as a marker of the flame zone that is close to stoichiometric region; hence the continuous use of OH measurement for combustion applications. 27

28 Motivation for the Current Work Based on the existing information the present work is focused on providing OH measurement with a detailed uncertainty analysis. The flow field is generated by a shear coaxial H 2 /O 2 flame. This study was aimed at obtaining quantitative OH concentration at chamber pressures of bar range and oxygen/fuel (O/F) mass flow ratio of four using OH-PLIF diagnostic. The uncertainty sources and their respective contributions to the OH concentration measurements will be addressed and discussed in detail in Chapters 4 and 5. The data obtained here includes OH-PLIF measurements at pressures of 10, 27, 37 and 53 bar. Temperature measurements for boundary conditions are also included to compliment the information provided to the CFD modelers. The data corresponding to chamber pressure of 10, 27, 37 and 53 bar were post-processed in this work and the uncertainties associated with the OH measurements were identified and evaluated. Thus, the rest of the document includes the following discussions: theory and review of OH planar laser induced fluorescence experimental facility and diagnostic methods employed OH-PLIF image processing and quantitative analysis results and uncertainty analysis conclusions future work Equation Section 2 28

29 CHAPTER 2 OH PLANAR LASER INDUCED FLUORESCENCE - THEORY AND REVIEW A brief discussion of laser induced fluorescence (LIF) application to obtain the number density of the species being probed, in this case, OH is given below followed by a review of existing studies. Fluorescence Modeling Fluorescence modeling is based on a two level excitation / detection strategy within the linear regime. Detailed explanations are given in Eckberth [33] and others [34 39]. Excited State A ν Laser Excitation λ = 283 nm A 3 ν Fluorescence Emission Step 1 Step 2 Vibrational level Rotational level Ground State X ν X ν 0 λ = nm Figure 2-1. Two-State Quasi-Steady Two-Step Modeling of Fluorescence The laser induced fluorescence process is illustrated in Figure 2-1. It consists of a two step process: the first step is the excitation of the molecule/radical from the ground state (X) to the upper excited state (A) by laser absorption; the second step is the spontaneous emissions of photons when the molecule relaxes from the upper excited state to their ground states. Given the certain energy loss associated with the process, emission is at longer wavelength than the excitation. Emission occurs very close after absorption and is of the order of less than 10 ns in the case of OH in an atmospheric flame [38]. The quantification of the number of photons collected in this process can be used to determine the number density of the molecule/radical in 29

30 the region of interest provided all the processes involved in the fluorescence are properly accounted for and modeled. The processes involved in fluorescence can be more specifically termed as stimulated absorption-w 12, stimulated emission-w 21, spontaneous emission-a 21 and collisional quenching- Q 21. These four processes of energy transfer take place between the electronic states, in this case, the ground state (X) and the upper excited state (A). In the upper excited state the two processes of interest are the rotational energy transfer -RET and the vibrational energy transfer -VET. The excitation is provided by a monochromatic source from a pulse laser with short duration of less than 10 ns. This permits fluorescence detection time of less than 500 ns which helps in avoiding the interference from other background emissions during diagnostics. The rate of absorption by the molecule/radical is given by B12 W12 = I( v) 2 c (2-1) Here W 12 (s -1 ) is the stimulated absorption rate, B12 is the Einstein B coefficient for absorption (cm 3 J -1 s -2 ), c is the speed of light (cms -1 ), I( v) is the laser spectral fluence (Wcm -2 cm) E( v) given by A τ laser l, where E( v ) is the laser spectral energy per pulse (Jcm), A laser (cm 2 ) is the cross sectional area of the laser beam or sheet andτ l (s) is the laser pulse duration. Since the absorption process involves laser/molecule interaction it is called stimulated absorption rate. The molecule/radical will relax from the upper state to the ground state by the following three processes as described below. The first path constitutes of stimulated emission, in which the molecule/radical interacts with the laser and returns to the ground state. The stimulated emission rate, W 21 (s -1 ) is given by 30

31 B21 W21 = I( v) 2 c (2-2) where B21 is the Einstein B coefficient for emission (cm 3 J -1 s -2 ). The absorption and emission rates are related by g1w 12=g2W 21 (2-3) Here g 1 and g 2 are the degeneracies of the ground and the upper electronic states respectively. The second path constitutes of the spontaneous emission in which the molecules relax from the upper excited state to the ground state by emitting fluorescence. This is the main mechanism for LIF signal production. The spontaneous emission rate is dictated by Einstein coefficient for spontaneous emission A 21 (s -1 ). The spontaneous emission rate and the stimulated absorption rate are related by A B = 8π hν (2-4) where h (J.s) is the Planck s constant and ν is the wave number of the particular individual transition (cm -1 ). In the third process, the molecules in the upper excited electronic state can relax to the ground state by collisions with other molecules called collisional quenching. The quenching rate is modeled as, Q 21 P 8kT B = c is i kt B i π mi 1 2 (2-5) where P is the pressure, B colliding species mole fraction, k is the Boltzman constant, T is the temperature, s i, the colliding species cross section and c represents the i m i is the reduced mass of excited molecule/radical,in this case, OH and the colliding species. Quenching 31

32 represents the rate of non-radiative decay of the excited state molecule to the ground state. It can be noticed from Equation 2-5 that quenching linearly increases with pressure and hence at high pressures the fluorescence signal intensity due to spontaneous emission can be significantly reduced due to quenching. This is one of the major challenges in applying LIF techniques at high pressures. In RET the molecules in the upper excited rovibrational state can move to neighboring rotational levels in the same excited electronic state due to collisions with other molecules. Similarly in VET the molecules migrate to neighboring vibrational levels of the same upper excited state. The collisional quench model in Equation 2-5 needs to be modified to take into account the effect due to RET and VET. The modified model for collisional quench rate of OH which also takes into account the effect of RET and VET is discussed in Chapter 4 in detail. Other mechanisms involved in the energy transfer processes are predissociation and photoionization [33]. Predissociation is the process in which the excited molecule dissociates prior to the emission of the photon. In photoionization, the excited molecule gets ionized prior to the emission of the photon. Based on the two state two step model as shown in Figure 2-1 a mathematical formulation of all the processes involved in fluorescence is made to infer target species number density. The population density in the ground state, n 1 (cm -3 ) and in the excited state n 2 (cm -3 ) constitute the total population density of n = n 1 + n 2 (cm -3 ) for the specific robvibrational transition being excited. The rate of change of molecules in the upper excited state (A) per unit volume is then given by dn dt 2 ( ) = n W n W + Q + A (2-6)

33 In the current study fluorescence in the linear regime is considered, thus the fluorescence signal is linearly proportional to the input laser irradiance. In other words, the number of fluorescence signal photons collected is linearly proportional to the number of input laser photons supplied during the duration of the pulse. In contrast to linear regime, fluorescence signal photons become independent of both laser irradiance and collisional quenching in the case of saturation regime. The laser irradiance used in the current study which is x 10 6 W/cm 2 is nearly four-five orders of magnitude less than the laser irradiance employed for saturation LIF studies by Carter et al. [40]. Hence for the current study the pumping is weak and the fluorescence can be considered to be in the linear regime. dn2 At steady state, dt is zero and in the linear regime, as W 12 is negligible [33], n 2 is expressed as n = 2 n W 1 12 (Q + A ) (2-7) The fluorescence signal or the number of photons, N p can then be expressed as Ω = (2-8) π Np n2a21v τ l 4 where, V (cm 3 ) is the volume probed by the laser and 4π Ω is the fraction of the solid angle detected. Substituting the expression of n 2 from Equation 2-7 and rearranging Equation 2-8 A Ω = n τ l Q A 4π 21 Np 1W12 V ( ) (2-9) For weak pumping, n 2 << n 1 and total population density n ~ n 1. The population density n 1 (cm -3 ) in the ground state rovibrational energy level is related to the total number density of the 33

34 molecule/ radical by n 1 = n o f B. Here no is the total number density and fraction of the specific rovibrational energy level in the ground state. Thus, can be rewritten as fb is the Boltzmann Np in Equation 2-9 A Ω τ l Q A 4π o 21 Np = n fbw12 V ( ) (2-10) Substituting the expression of W12 from Equation 2-1 into Equation 2-10 B A Ω N = n f I( v) V τ l Q A 4π o p B 2 c ( + ) (2-11) Emitted and absorbed light has a finite bandwidth which is called the line broadening [33, 35]. This means that in reality, the energy of a dipole transition which is well defined by the energy difference between two quantum states is not monochromatic and has a certain spectral width and shape. The line broadening in a typical combustion environment is due to three main reasons, namely natural broadening, collisional/pressure broadening and Doppler broadening. Each is briefly discussed below. Natural broadening is due to the finite lifetime of the molecule/radicals in the excited state. If the molecule were to radiate energy for an infinite period, the line shape is a delta function. Since the lifetime is finite it represents a Lorentzian function [35]. In general the effect of natural broadening is much smaller compared to collisional and Doppler broadening; hence, it is often neglected [33]. Similarly in the case of collisional broadening, the lifetime of the molecule in radiating the energy is reduced if it collides with other molecules. The Doppler broadening occurs due to the Doppler shift caused by the relative motion of the molecule and the laser beam propagation. The collisional broadening represented by a Lorentzian function [35] is 34

35 1 Δvc f c( v) = (2-12) 2 2 π Δv ( ) c v vo + 2 where f c( v ) is the normalized line shape function, Δ vc is the spectral width associated with collisional broadening, v o is the central frequency of the transition involved. For OH the collisional width could be calculated from the empirical model provided by Davidson et al [41] based on spectroscopic measurements carried out in a shock tube at conditions of 60 bar and 1735 K P Δ vc = cm Po T (2-13) Similarly the Doppler broadening represented by Gaussian profile [35] is ln 2 v vo f ( v) = exp 4ln 2 D (2-14) ΔvD π Δv D where f ( v ) is the normalized line shape function, Δv D D is the spectral width associated with collisional broadening and vo is the central frequency of the transition involved. The Doppler width [35] is 2 8kTv B o ln(2) 2-7 T -1 Δ vd = = 7.16 x 10 v o cm mc M (2-15) where T is the temperature, k B, the Boltzmann factor, m, the mass of the molecule/radical and M is the molecular weight of the molecule/radical which is OH in the current study. The spectral distribution due to the line broadening is expressed as a normalized line shape function, Φ ( v) and is defined as Φ ( vdv ) = 1. The absorbing species line shape function, abs + abs 35

36 Φ abs () v is obtained as the convolution of collisional and Doppler line shape functions which is generally referred to as the Voigt profile [33, 35]. Moreover, the central frequency of the absorption profile gets shifted due to the collision with neighboring molecules and/or due to the Doppler effect [20, 37, 41]. The collision induced shift for OH is given [20, 41] by 0.45 ± 0.08 P Δ vc shift = cm Po T (2-16) and the Doppler shift [37] is given by v -1 Δ vd shift = v o cm c (2-17) Here Δ v C shift and ΔvD shift represent the collisional and Doppler shifts respectively, v o is the central frequency of the specific rovibrational transition, v, the velocity of the molecules and c, is the speed of light. In the current study, the absorption profile for OH is simulated using the commercially available software LIFBASE [37]. The laser profile used in this study is assumed to be well represented by the Gaussian profile. The laser line profiles and the absorption line profiles relevant to the current study will be discussed later in Chapters 3 to 5. Thus, to account for the spectral distribution of the laser profile and the absorption profile of the target species, the fluorescence signal in Equation 2-11 is modified as B A Ω N n f I() () V τ l (2-18) Q A 4π o p = B v Φ 2 abs v dv c ( + ) Substituting for E( v) I( v) = and E( v) = E Φlaser ( v) where E is the laser energy per pulse and A τ laser l Φ laser () v is the laser line shape, into Equation 2-18 and rearranging, 36

37 N n E f B A Ω V A (2-19) Q A 4π o B p = Φ 2 laserφabsdv c ( + ) n o OH (I) 1) Fluorescence (i) Detection Electronics (ii) Excitation / Detection Strategy (iii) Detection Environment 2) Interference Signals (iv) Laser internal scattering (ii) Background emission (iii) Mie / Rayleigh Scattering N p = E f B B 12 A 21 W 2 ( ) V A c Φ Φ dν laser abs A Q π (II) 1) Laser (i) Shot to shot power fluctuation (ii) Laser sheet / beam profile variation (iii) Laser absorption (OH & other molecules) OH-PLIF Measurement (III) 1) Absorption and Excitation (i) Boltzmann factor (Temperature) (ii) Absorption Coefficient (Spectroscopy) 2) Line Shape (iii) Overlap integral (line shape & laser center line shift) (iv) Model (Collisional & Doppler width/shift) 3) Fluorescence Efficiency (v) Quench rate (Collider species cross section/ mole fraction,pressure, Temperature ) (vi) Model for quantum yield (IV) 1) Experimental Constants (i) Probe volume (ii) Solid angle detected (iii) Transmission efficiency of filters (iv) Photon detection efficiency of camera Figure 2-2. Physical significance of the terms in OH number density expression Equation 2-19 can be rearranged in terms of OH number density. The physical significance of the terms from the experimental, modeling and quantifying point of view are shown in Figure 2-2. The four categories of OH-PLIF measurement mentioned in Figure 2-2 are discussed here. Fluorescence and Interference Signals The excitation and detection strategy of OH consists of A-X (0, 0), A-X (1,0), A-X (3,0) transitions of which A-X(1,0) is employed in the current study. The detection electronics employed to collect fluorescence could be an ICCD camera, photodiode or spectrograph. The detection environment of OH is typically a combustion zone. The interference signals refer to the 37

38 potential interferences from other species in the combustion environment, elastic scattering and the background emissions. Laser The laser pulse energy employed in PLIF measurements and the shot to shot power fluctuation needs to be monitored. The laser beam/sheet profile is non-uniform in space and needs to be corrected for quantitative measurements. The laser is absorbed by OH and other species in the combustion environment resulting in attenuation of the beam as it traverses through the flame. All these factors contribute to the measurement uncertainties. Absorption and Excitation, Line Shape and Fluorescence Efficiency The Boltzmann fraction, f B in the initial state population, n o f B varies with temperature and hence a careful selection of rovibrational transitions with minimum temperature dependence is recommended for PLIF diagnostics. The dependence of Φabs with temperature and pressure is to be accounted for species quantification. The determination of fluorescence yield from Equation 2-5 also requires the knowledge of colliding species mole fraction in addition to temperature and pressure fields. Experimental Constants The strength of the fluorescence signal detected depends on the intersection volume of laser beam/sheet with the flame known as the probe volume and the fraction of solid angle collected. To avoid the interference signals and elastic scattering, optical filters are employed while collecting fluorescence; however most of the optical filters have transmission efficiency of less than 60 % at 310 nm where the OH fluorescence is detected. In addition to this the photon detection efficiency at 310 nm for an ICCD camera is less than 25 %. All these reduce the strength of the detected fluorescence signal. 38

39 Review of OH PLIF Diagnostic Studies LIF techniques can be used for temperature, pressure, velocity, density or mole fraction measurements in wide range of environments [33, 35, 38, 42, 43]. Equation 2-19 helps determine the number density directly from the fluorescence signal. Moreover PLIF provides species measurements in various fluids including combustion environments. Hanson [42] provided a detailed review of the application of planar imaging of fluorescence, giving examples of PLIF application to obtain species concentration, 2D temperature fields, velocity and pressure imaging. In the following discussions, studies related to OH fluorescence and its planar imaging in combustion zones will be presented. A brief review of the OH-PLIF diagnostics is tabulated in Table 2-1. The table is set up to identify the four categories as (I) Fluorescence and interference signals, (II) Laser energy fluctuation, spatial profile non-uniformity and attenuation, (III) Absorption coefficient variation with temperature, overlap integral modeling and dependence on temperature and pressure, and fluorescence yield modeling and dependence on temperature and pressure and (IV) Experimental constants corresponding to Figure 2-2. The last column in the table indicates the main results from each study. 39

40 40 Table 2-1. Review of OH-PLIF Diagnostics Authors Target Species (I) Fluorescence strategy and interference signals Dieke & Crosswhite [44] Allen & Hanson [24] OH emissions in atmospheric flame Imaging OH in atmospheric heptane-air flame (II) Laser energy, spatial profile and attenuation (III) Absorption &excitation line shape and fluorescence efficiency (IV) Experimental constantstransmission & photon detection Observations fundamental study which provided ultraviolet bands of OH in nm Excitation Q 1 (6), A-X(1,0) Detection (1,1) at 310 nm camera Interference elastic scattering from droplets 10 mj per pulse The Q1(6) transition at 283 nm was devoid of temperature dependence across the field of view Interference filter with ε=15% at 310 nm was used to collect fluorescence. Signal collected at 90 o to laser OH fluorescence was used to comprehend the hydrodynamic flame structure and the combustion zones Jeffries et al. [45] OH,NH, CH, CN & NCO fluorescence spectrum in atmospheric CH 4 /N 2 O flame Excitation(OH) nm, A-X(0,0) Detection(OH) 350 nm, A-X(0,1) Monochromator, photomultiplier 0.2 mj per pulse - - Excitation specific to OH produced weak fluorescence emissions from NH and CN due to electronic energy transfer between molecules/ radicals

41 41 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Smith & Crosley [46] Garland & Crosley [47] (i) Quenching rate constants of OH with H 2, N 2 O & ten hydrocarbons at 1200 K (ii) OH is produced by thermal decomposition of H 2 O 2 Excitation(OH) nm, A-X(0,0) Detection(OH) 309 nm, A-X(0,1) Monochromator, photomultiplier (II) Laser energy, spatial profile and attenuation (III) Absorption &excitation line shape and fluorescence efficiency 2 mj per pulse (i) Measured time decay of the fluorescence with pressure was used to obtain quenching rate constants. (ii) The measured cross sections had 15% accuracy Temperature and species dependent quenching cross section of OH was predicted using a model based on attractive forces (IV) Experimental constantstransmission & photon detection Observations - Attractive forces between the molecules need to be properly taken into account in the case of quenching models for accurate prediction of quenching cross sections. - The predicted quenching cross sections of NH 3, H 2, NO, O 2, H 2 O, N 2 O,CH 4, CO and CO 2 agreed within +30 % of the experimental values

42 42 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Edwards et al. [48] OH-LIF in solid propellant flames at 35 bar Excitation(OH) nm, A-X(0,0) Detection(OH) 310 nm, A-X(0,1) Monochromator, Photomultiplier Interference elastic scattering from particulates (II) Laser energy, spatial profile and attenuation (i) 6 mj per pulse (ii) significant laser attenuation (iii) increase in optical thickness with pressure (III) Absorption &excitation line shape and fluorescence efficiency (i)quenching decreased the LIF signal with increasing pressure. (ii) Saturation LIF to avoid effects of quenching. (IV) Experimental constantstransmission & photon detection Fluorescence collected at 90 o to laser propagation Observations (i) Lack of availability of high pressure kinetic and spectroscopic data were addressed as the major challenges in LIF at high pressures

43 43 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Schefer et al. [26] Seitzman et al. [14] OH concentration in turbulent CH 4 -jet flame OH-PLIF in a turbulent nonpremixed H 2 /air jet atmospheric flame Excitation P 2 (7), A-X(1,0) Detection 312 nm vidicon camera Excitation Q 1 (3), A-X(0,0) Detection A-X(0,0),(1,1) CCD camera (II) Laser energy, spatial profile and attenuation (i) Laser attenuation was negligible. (ii) radiative trapping was < 5% (i) mj per pulse (ii) laser absorption is 3 20 % (III) Absorption &excitation line shape and fluorescence efficiency +5 % variation in initial state population in K temperature range +40 % variation in initial state population in K temperature range (IV) Experimental constantstransmission & photon detection 10 nm bandwidth filter centered at 312 nm Observations (i)oh concentration was obtained from flat flame calibration. (ii) +10 % error from calibration measurements, 7% due to photon statistics (iii) OH concentration was five times higher than equilibrium values in reaction zones - Spatial autocorrelation was used to determine flame angle and correlation lengths

44 44 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Kohse- Hoinghaus et al. [49] Line shape, temperature and estimated OH concentration from a CH 4 /air flat flame at 1 10 bar Excitation 283nm,A-X(1,0) Detection A-X(0,0),(1,1) Photomultiplier (II) Laser energy, spatial profile and attenuation 1.5 mj per pulse (III) Absorption &excitation line shape and fluorescence efficiency (i) There was loss of fluorescence signal due to quenching and absorption line shape broadening with increasing pressure and the estimated signal reduction was of the order of 100 in the bar range. (IV) Experimental constantstransmission & photon detection Interference filter centered at 315 nm with FWHM of 38 nm Observations (i) The simulated Voigt profile matched well with the measured one (ii) OH concentration from absorption measurements with 30% accuracy agreed well with numerical predictions (ii) Feasibility of applying numerical modeling to obtain effect of quenching and line broadening on fluorescence efficiency was mentioned

45 45 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Seitzmann & Hanson [50] Comparison of A-X (1,0), (0,0) and (3,0) schemes for quantitative fluorescence imaging. The A-X(1,0) scheme is highlighted here (II) Laser energy, spatial profile and attenuation - (i) 10 mj/cm 2 (4 mj for 80x0.5 mm sheet) is considered to ensure fluorescence in linear regime within + 5% down to zero energy (ii) Need to apply corrections for spatial laser profile variation (III) Absorption &excitation line shape and fluorescence efficiency (i) Need to choose rotational transition with low temperature dependence (ii) Overlap integral variation in a nonisobaric flow(1 5 bar) is 30 40% for lasers with line width of cm -1 (iii) Overlap integral variation with temperature ( K) is less then + 5% for line widths less than 0.5 cm -1 (iv) quench rates vary only by <10% in regions of OH concentration (IV) Experimental constantstransmission & photon detection (i) Assumption: Fluorescence is emitted equally into 4πSr (ii) Random noise in the detector(iccd) is contributed by shot noise, quantum efficiency, electron gain, dark, readout and digitization noise (iii) Pulse to pulse variation in laser bandwidth contributes to error in OH concentration measurement. Observations (i) Actual laser induced excitation and emission can deviate from the two state, two step quasi steady model leading to systematic errors (ii) Nonlinear responses to change in laser energy, population fraction and depletion are well within the range of A- X(1,0) excitation scheme.

46 46 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Locke et al. [51] Carter & Barlow [52] Merits and demerits of PLIF applied to reactive flows OH & NO- PLIF in a turbulent nonpremixed H 2 /air jet atmospheric flame Obstacles include stray light interferences, quenching contributions and RET Excitation(OH) O 12 (8), A-X(1,0) Detection(OH) A-X(0,0),(1,1) photomultiplier tube photocathode (II) Laser energy, spatial profile and attenuation (III) Absorption &excitation line shape and fluorescence efficiency (IV) Experimental constantstransmission & photon detection Observations Merits include 2D imaging, multi species probing, identifying primary reaction zones, temperature field imaging and semiquantification - (i) The need for spectroscopic data for quenching correction was mentioned (ii) Colliding species and temperature field data was obtained from equilibrium calculations - (i) To obtain OH concentration an initial calibration was carried out in a lean H 2 /air flame in a Hencken burner

47 47 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Paul [53] Temperature dependent collisional model for OH in K range (II) Laser energy, spatial profile and attenuation (III) Absorption &excitation line shape and fluorescence efficiency (IV) Experimental constantstransmission & photon detection Observations (i) A function for predicting temperature dependent cross section for collisional quenching of OH by various molecules is provided (ii) A model for fluorescence yield in A-X(1,0) excitation scheme by incorporating effect of VET in the excited electronic state(a)

48 48 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Allen et al. [54] Imaging OH in 1 10 bar heptane, methanol and ethanol-air flame Excitation(OH) 283 nm, A-X(1,0) Detection(OH) nm ICCD camera Interference 100 ns gate time to avoid background luminosity and chemiluminescent gas emissions. (II) Laser energy, spatial profile and attenuation (i)3 mj per pulse. (ii) laser attenuation was estimated as ~30% due to absorption by hydrocarbons (III) Absorption &excitation line shape and fluorescence efficiency (i) Effect of pressure on fluorescence signal intensity in linear regime was analyzed based on steady state and multi level transient approach (IV) Experimental constantstransmission & photon detection A combination of filters transmitted fluorescence from nm Observations (i) As long as fluorescence was in linear regime, quasi steady state model used in deriving fluorescence yield was valid for the experimental conditions investigated.

49 49 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Battles and Hanson [55] LIF measurements of OH and NO in 1 10 bar methane flames: Fluorescence modeling and experimental validation Excitation(OH) P 1 (8), nm, A-X(1,0) Detection(OH) A-X(0,0) Photomultiplier tube(pmt) Interference No significant interference near 285 nm (II) Laser energy, spatial profile and attenuation (i)100 μj per pulse to ensure fluorescence in linear regime (ii) Judicious selection of absorption transition to avoid significant laser attenuation (III) Absorption &excitation line shape and fluorescence efficiency (i) Fluorescence signal was modeled as two state two step steady process in linear regime (ii) Use of laser with large bandwidths to minimize effect of pressure on overlap integral. (iii) Laser with large bandwidths provided more flexibility in tuning the centre line of the absorption profile (IV) Experimental constantstransmission & photon detection Observations - (i) The single point OH equilibrium concentration from LIF measurements agreed well with the calculated equilibrium values of OH. (ii) This implied that the effect due to overlap integral, absorption line strength variation due to temperature and fluorescence yield were well accounted by the model used to predict them

50 50 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Locke et al. [56, 57] Paul et al. [58] OH-PLIF imaging to lean burning JP-5 combustor at bar Collisional quenching of OH at high temperature measured in a shock tube in K temperature range Excitation(OH) One among R 1 (1), R 1 (10), Q 1 (1) at A-X(1,0) Detection(OH) A-X(0,0) ICCD camera Excitation(OH) Q 1 (2)/Q 1 (5), A-X(1,0) Detection(OH) 310 nm,a-x(0,0) Photomultiplier tube(pmt) (II) Laser energy, spatial profile and attenuation (i) 10 mj per pulse (ii) laser beam spatial nonuniformity was corrected by fluorescence imaging of R590 dye solution (III) Absorption &excitation line shape and fluorescence efficiency (IV) Experimental constantstransmission & photon detection - Interference filter centered at 315 nm with FWHM of 10.6 nm - (i) Rate coefficients from fluorescence life time was converted into quenching cross sections by dividing with average collisional velocity of the species pair (ii) Quenching model formulated by Paul 53 could predict the temperature dependent behavior observed from experiments Bandpass filterer, nm Observations The practical importance of applying PLIF to high pressure combustor was highlighted At 2300 K, the ratio of the measured quenching cross section to quenching model [53] predicted values for H 2 O and O 2 are 1.12 and respectively

51 51 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Nandula et al. [59] (i) Single point LIF measurement in turbulent lean premixed methane flame. (ii)temperatur e & species (H 2,H 2 O,O 2 ) from Raman/ Rayleigh measurements Excitation(OH) O 12 (8), A-X(1,0) Detection(OH) 310 nm,a-x(0,0),(1,1) Photomultiplier tube(pmt) (II) Laser energy, spatial profile and attenuation (III) Absorption &excitation line shape and fluorescence efficiency - (i) The species concentration and temperature obtained from STANJAN was used to calibrate the measurement from a H 2 -air and CH 4 flame. (ii) The fluorescence signal was corrected for the variation in Boltzmann fraction and collisional quench rate. (IV) Experimental constantstransmission & photon detection Observations - (i) Uncertainties in measurement were identified as 10.5% due to shot noise and 5% due to wavelength drift. (ii) Location and growth of shear layer were determined from the OH distribution. (iii) The super equilibrium OH concentration were nearly four times higher than the equilibrium counter parts

52 52 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Ngyuen et al. [60] OH concentration from LIF measurements in a methaneair Bunsen flame. Rayleigh /Raman measurements Excitation(OH) O 12 (8), A-X(1,0) Detection(OH) nm, A- X(0,0),(1,1) Photomultiplier tube(pmt) (II) Laser energy, spatial profile and attenuation (i)40 μj per pulse to ensure fluorescence in linear regime (ii) O 12 (8) transition was chosen to avoid significant laser attenuation (III) Absorption &excitation line shape and fluorescence efficiency (i) For electronic quenching corrections, the temperature and colliding species concentration data were obtained from Raman/Rayleigh measurements. (ii) The OH number density was calibrated against the equilibrium OH composition corresponding to the measured Rayleigh temperature in a lean CH 4 -air flame (IV) Experimental constantstransmission & photon detection Combination of color glass filters (WG-295 & Hoya U-340) Observations (i) In the study, it was observed that the temperature and OH concentrations at the inner flame zones could be well predicted using a one dimensional premixed laminar flame model incorporating finite rate chemistry.

53 53 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Arnold et al. [61] Atkan et al. [62] Quantitative measurements of OH by PLIF from a laminar premixed methane / air flat flame at pressures of 1, 5 and 20 bar OH LIF, 2D and spectroscopic measurements at 5 36 bar in a laminar premixed methane/air flames Excitation(OH) P 1 (8), A-X(1,0) Detection(OH) A-X(0,0),(1,1) ICCD Interference (i) 120 ns gating to suppress flame emissions. Excitation(OH) nm, A-X(1,0) Detection(OH) 310 nm, A-X(0,0),(1,1) CCD camera Interference No interference from other molecule (II) Laser energy, spatial profile and attenuation (i)1 mj per pulse (ii) Background subtraction to avoid light reflections (iii) laser spatial variation was corrected (i) 14 mj per pulse (ii) Estimated laser absorption was less than 10 % (III) Absorption &excitation line shape and fluorescence efficiency (i) Boltzmann fraction variation was 10 % in the range of K (ii) Absorption line shape was measured by careful scanning of P 1 (8) line. (iii) Temperature data was obtained from CARS and numerical simulation (i) Scanned excitation spectra and simulated excitation spectra matched very well and signified that there are no interferences from other molecules in the A-X(1,0) nm range (O 2 interference for A-X(3-0) scheme) (IV) Experimental constantstransmission & photon detection (i) WG295 filter was used to suppress the elastic scattering (ii) Spatial variation (pixel to pixel) of camera sensitivity was corrected Bandpass filter(wg305 and UG11) centered at 310 nm, FWHM-16 nm and peak transmission efficiency of 5.5% Observations (i) Absolute concentration was obtained from 1D absorption measurements (i) The advantages of the A-X(1,0) LIF detection scheme were identified as devoid of fluorescence trapping (A- X(0,0))

54 54 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Hicks et al. [63] Fluorescence imaging of combustion species in gas turbines up to 20 bar and associated complexities Excitation(OH) 283nm, A-X(1,0) Detection(OH) nm, A- X(0,0),(1,1) ICCD camera Interference Interferences from PAH as they are broadband absorbers and emitters in the emission spectrum of OH, flame emissions, laser light scattering & wall luminescence (II) Laser energy, spatial profile and attenuation (i) 16 mj per pulse (ii) laser sheet nonuniformity was corrected by obtaining the quartz reflected images of the laser sheet (iii) Background subtraction of the nonresonant images (III) Absorption &excitation line shape and fluorescence efficiency (i) Pressure induced line broadening and quenching effects which tend to decrease the fluorescence signal. (ii) Selection of line transitions with weak absorption coefficients to avoid considerable laser absorption and attenuation (IV) Experimental constantstransmission & photon detection (i) Combination of WG-305 & UG 11 filters (transmission efficiency ~56% in the nm range). (ii) Weak signals require pixel binning in the camera (iii) Determination of camera magnification and its accurate alignment. Observations Complexities in OH imaging (i) Test rig and optical system vibration and displacement (ii) Optical window cleanliness(soot formation on the windows), cooling and structural integrity (iii) laser wavelength drift (iv) Optical thickness of the medium (v) Noisy spikes in the collected signal due to abrupt rise in laser intensity

55 55 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Tamura et al. [64] Collisional quenching of OH measured in a premixed laminar methane flames at < 1 bar Excitation(OH) R 2 (6), A-X(0,0) Detection(OH) A-X(0,0) PMT (II) Laser energy, spatial profile and attenuation <0.5 μj per pulse (III) Absorption &excitation line shape and fluorescence efficiency (i) Flame temperatures were measured from the excitation scans. (ii) For quenching rate determination, equilibrium compositions of the colliding species concentration were calculated. (iii) Quenching contributions from individual colliding species was calculated based on temperature dependent rate expression (IV) Experimental constantstransmission & photon detection Observations - (i) The measured quench rate and the calculated quench rate based on temperature and species dependent quench rate model agreed very well. (ii) The excellent agreement between the calculations and the experiments showed that collisional quench rate could be well predicted from knowledge of gas temperature and colliding species concentration

56 56 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Candel et al. [12] Meier et al. [65] OH-PLIF to investigate shear coaxial cryogenic jet flames Species and temperature measurements from piston engine(10 bar) and aero engine test rig(6 bar) Excitation(OH) Q 1 (6), nm A-X(1,0) Detection(OH) A-X(0,0) ICCD Interference Raman signal at 296 nm from the liquid phase Excitation(OH) nm A-X(1,0) Detection(OH) 315 nm, A-X(0,0) ICCD camera Interference Interference from fuel fluorescence (II) Laser energy, spatial profile and attenuation 30 mj per pulse (i) 5 10 mj per pulse. (ii) Laser sheet spatial variation was corrected by normalized acetone fluorescence images on an average basis (iii) laser shot to shot energy fluctuation was monitored using a fast photodiode (III) Absorption &excitation line shape and fluorescence efficiency (i) The Q 1 (6) transition was selected to minimize the temperature dependence (i) Transition was selected to minimize the variation of state population in K temperature range (IV) Experimental constantstransmission & photon detection UG-5 and WG-305 fliter (i) Interference filters centered around 315 nm with FWHM 30 nm Observations (i) LO x jets scattered and dispersed the laser sheet thereby affecting OH fluorescence (i) Areas of OH concentration were used to identify zones of homogenous combustion and high heat realse.

57 57 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Frank et al. [66] Hicks et al. [67] OH-PLIF in heptane and Jet-A spray flames at 5, 7 and 11 bar OH-PLIF applied to combustors burning Jet-A fuel at pressures of 9 and 18 bar Excitation(OH) 283nm A-X(1,0) Detection(OH) 315 nm, A-X (0,0),(1,1) ICCD camera Interference Scattering from fuel droplets Excitation(OH) 282nm A-X(1,0) Detection(OH) 316 nm, A-X (0,0),(1,1) ICCD camera Interference Scattering from fuel droplets (II) Laser energy, spatial profile and attenuation (i)3 mj per pulse (ii) Laser attenuation across the flame at higher pressures was attributed to increased OH number density and hydrocarbons mj per pulse (III) Absorption &excitation line shape and fluorescence efficiency (i) At high pressures there was considerable decrease in fluorescence signal due to quenching and line broadening. (IV) Experimental constantstransmission & photon detection Interference filters - Interference filters centered at 316 nm with FWHM 2.6 nm and peak transmission of 16% Observations (i)the OH distribution was used to analyze the turbulent spray structure (i) OH-PLIF images were used to mark flame and recirculation zones. (ii) The use of OH-PLIF images in fuel injector design and kinetic modeling was highlighted.

58 58 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Stocker et al. [68] Thiele et al. [69] Identification of rotational lines of OH in H 2 /O 2 and methane/air flame OH-PLIF in spark ignited combustion of H 2 /air mixtures Excitation lines at A-X(0, 0), (1,0), (2,0) and (3,0) were recorded using spectrograph Excitation(OH) 283nm A-X(1,0) Detection(OH) 310 nm, ICCD camera (II) Laser energy, spatial profile and attenuation (III) Absorption &excitation line shape and fluorescence efficiency (IV) Experimental constantstransmission & photon detection Observations 5 mj per pulse - - (i) The entire rovibrational transitions in the nm range were excited, detected, identified and tabulated 0.2 mj per pulse - - (i) The raw gray scale images were filtered using a 2-D Gaussian filter to reduce noise. (ii) The flame front position was identified as the region of steepest gradient in the flame/oh image. (iii) Temporal evolution of the flame kernel was identified from the OH images.

59 59 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Schulz et al. [70] Santhanam et al. [71] Laser absorption by H 2 O at shock heated temperatures of K in nm range and pressures of 1 70 bar OH-PLIF visualization in actively forced swirlstabilized spray combustor Detection CCD camera Spectrograph Excitation(OH) nm A-X(1,0) Detection(OH) 315 nm, ICCD camera (II) Laser energy, spatial profile and attenuation Light from deuterium lamp (i)6 mj per pulse (ii) Neglected the variation of laser sheet intensity (III) Absorption &excitation line shape and fluorescence efficiency (IV) Experimental constantstransmission & photon detection Observations - - (i)laser absorption by H 2 O at 283 nm was negligible for pressures of 1 70 bar and K temperature range (i) For OH calibration, the effect due to variation in quenching cross section across the flame was neglected 10 nm narrow band pass interference filter at 315 nm (i) To calibrate OH, water vapor in atmospheric pressure at high temperatures was used as the calibration source

60 60 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Grisch et al. [72] OH-PLIF measurements in H 2 /air diffusion flame Excitation(OH) Q 1 (5) A-X(1,0) Detection(OH) A-X (0,0),(1,1) ICCD camera (II) Laser energy, spatial profile and attenuation (i)<10 μj per pulse (ii) Laser shot to shot power fluctuation was monitored (III) Absorption &excitation line shape and fluorescence efficiency (i) For calculation of collisional quench rate, the colliding species concentration and temperature were obtained from adiabatic equilibrium conditions (IV) Experimental constantstransmission & photon detection UG5 and WG 295 filters Observations (i) OH calibration was carried out in a H 2 /air flame of equivalence ratio 0.9. (ii) Fluorescence intensity of OH along the height was compared with the simulated OH profiles. (iii) The estimated uncertainty in absolute OH concentration was 20%

61 61 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Meyer et al. [73] Kalitan et al. [18] OH-PLIF in swirlstabilized spray flames OH-PLIF in LO x /methane flames at 41 bar Excitation(OH) Q 1 (9) A-X(1,0) Detection(OH) A-X (0,0),(1,1) ICCD camera Excitation(OH) Q 1 (9) A-X(1,0) Detection(OH) A-X (0,0),(1,1) ICCD camera (II) Laser energy, spatial profile and attenuation (i) 24 mj per pulse (ii) shot to shot power fluctuations was estimated as + 5% (i) Signal attenuation due to laser absorption by OH. (ii) Light scattering as it traversed through spray (III) Absorption &excitation line shape and fluorescence efficiency (i) Boltzmann fraction variation was % in the range of K (ii) The collisional quenching rate variation with species concentration and temperatures corresponding to equivalence ratios of 0.5 to 3 was estimated to be (IV) Experimental constantstransmission & photon detection WG 295 and UG % in that range - UG11 and WG305 filter Observations (i) Laser energy absorption due to OH and droplet scattering accounted to + 10 % uncertainty. (i) OH images were used as indicators of combustion zones of high temperatures

62 62 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Singla et al. [20] OH-PLIF in LO x /GH 2 jet flames up to 63 bar Excitation(OH) Q 11 (9.5) A-X(1,0) Detection(OH) nm, A-X (0,0),(1,1) ICCD camera, Spectrometer Interference Raman scattering from LOx jet (II) Laser energy, spatial profile and attenuation (i) 42 mj per pulse (ii) laser beam absorption by OH was estimated to be % (iii) Laser beam is absorbed and scattered by LO X core in the centre (III) Absorption &excitation line shape and fluorescence efficiency (i) The variation of Boltzmann fraction accounted to 10% in the range of K and was considered insignificant (ii) The collider species concentration and temperature field for quench rate at 63 bar was calculated based on the collider species mole fraction and temperature field of a counter flow LOx/GH 2 flame at 1 bar (IV) Experimental constantstransmission & photon detection Bandpass filter Observations (i) A detailed description of fluorescence modeling was provided. (ii)quenching did not strongly perturb the spatial fluorescence (iii) The collected fluorescence spectra matched well with the simulated spectra from LIFBASE (iv) The mean position of the flame, flame stabilization, corrugation and unsteadiness of the jet were observed from OH images

63 63 Table 2-1. Continued. Authors Target Species (I) Fluorescence strategy and interference signals Singla et al. [21] Feasibility of OH-PLIF in LO x /methane jet flames upto 30 bar Excitation(OH) Q 11 (9.5) A-X(1,0) Detection(OH) nm, A-X (0,0),(1,1) ICCD camera, Spectrometer Interference PAH fluorescence (II) Laser energy, spatial profile and attenuation 42 mj per pulse (III) Absorption &excitation line shape and fluorescence efficiency (IV) Experimental constantstransmission & photon detection - (i) Filter scheme 1: 56% transmission at nm (ii) Filter scheme 2: 25% at 308 nm with FWHM of 15 nm Observations (i) The limiting factor for OH- PLIF in oxygen/methane flame above 25 bar was PAH fluorescence interference when compared to laser absorption in hydrogen/oxygen flames (ii) The LOx/methane flame was less stabilized compared to LOx/H 2 flame

64 Based on these studies certain observations are useful as follows Fluorescence Strategy and Interference Signals The choice of excitation at A-X(1,0) and detection at A-X(0,0), (1,1) in the nm range has the advantage that the elastic and laser internal scattering can be effectively blocked. Moreover there are no interferences from molecules like H 2 O and O 2 in the combustion environment. The radiative trapping which is predominant in the A-X(0,0) scheme is negligible [62]. To suppress the flame emissions and to collect all the fluorescence, gate width of the order of ~150 ns for the detection system, in this case ICCD camera can be employed. The background emissions need to be corrected depending on the signal strength. Laser The laser pulse energy / area of 10 mj/cm 2 could ensure fluorescence in the linear regime within + 5% [50]. Hence laser energy typically of 2 3 mj with a sheet cross section of 40 mm x 0.5 mm could be effectively used to ensure fluorescence in the linear regime. The laser attenuation across flame/ combustion environment depends on the strength of the absorption transition excited, the number density of the molecule, unwanted absorption by molecules like hydrocarbons and the path length traversed by the laser. The laser absorption and attenuation can vary from less than 5 % to 100 % across the flame depending on the flame conditions. In a GH 2 /LOx flame the absorption of laser by other combustion species like H 2 O was found negligible at 283 nm [20, 70]. The use of a strong transition with a laser of large line widths and low energy to ensure fluorescence in the linear regime can be viewed as a potential method to obtain fluorescence of good signal strength. The use of lasers with relatively larger line widths can lead to excitation of rovibrational lines neighboring the strong transitions. Hence the excitation of these lines needs to be taken into account in the fluorescence modeling. The laser shot to shot power fluctuation 64

65 could be monitored and corrected or the mean value of the laser energy can be used provided the uncertainty from energy fluctuation is accounted in quantitative measurements. To correct for the laser sheet profile variation in space, acetone fluorescence images [65] from the same excitation wavelength, 283 nm in this case could be used. Absorption & Excitation, Line Shape and Fluorescence Efficiency In quantitative measurement, one of the major uncertainties is due to the initial state population variation with temperature and is normally 10 15% in the K range. This problem could be approached in the following ways. The transition could be selected such that the variation of the state population in the temperature range of interest is negligible so that it does not contribute to the uncertainty in the measurements. The second approach is to obtain the temperature field information in the region of interest either from calculations based on equilibrium conditions, detailed numerical simulation of the combustion field /reference flame or by calibration measurements via thermocouple measurements, Raman / Rayleigh measurements in actual combustion environment/ reference flame. However, each of these approaches has themselves a degree of uncertainty. The third approach is to determine the uncertainties in the measurements due to state population variation in the temperature region of interest and account for the uncertainties in the quantitative measurements [73]. The line broadening and shifting at higher pressures reduce the overlap integral and hence the fluorescence signal. The variation in the overlap integral needs to be, therefore taken into account in quantitative measurements. The reduction in fluorescence signal due to decrease in the overlap integral can be overcome by employing lasers with larger line widths [55]. The effect of collisional quenching in reducing the fluorescence efficiency can lead to signal reduction of the order of a factor of 100 in the bar range [49]. Most of the works which utilize OH-PLIF for applied spectroscopy uses collisional quench model given by 65

66 Equation 2-5 for calculation of collisional quench rate. This requires the knowledge of the pressure, temperature, mole fraction and temperature dependent cross section of the colliding species. The expression for temperature dependent cross section from a complex collision model and the measured temperature dependent cross sections of the colliding species such as H 2, H 2 O, and O 2 are available in the literatures [53, 58, 64]. The knowledge of temperature and mole fraction of the colliding species can be obtained in the same way as the temperature field data is obtained to account for initial state population variation. The quenching rate variation across the combustion field is not significant [20] and could normally account for less than 10% variation [50]. Experimental Constants The fluorescence process is considered such that it is emitted equally in all directions and that the photons are Poisson distributed [35]. There is, thus an uncertainty in the exact number of photons detected and this uncertainty is called shot noise. The uncertainties in detection system also arise due to the quantum efficiency of the photocathode, the thermal current in the CCD chip known as dark current, readout noise in the A/D conversion and the digitization noise [35, 50]. The spatial variation of camera sensitivity across the chip also adds to the uncertainties. For weak fluorescence signal detection pixel binning at the cost of resolution is also recommended. Thus the challenges involved in applying OH-PLIF at high pressures and temperatures in the linear regime could be recognized as 1) rotational level population dependence on temperature, 2) reduced fluorescence efficiency due to absorption line shape broadening and collisional quenching, 3) laser beam attenuation, absorption and steering, 4) scattering interference from other molecules and 5) insufficient spectroscopic data. The thirty nine OH fluorescence studies described above were largely done in low pressure as shown in Figure

67 Sample of 39 OH Fluorescence Studies 25 Number of Studies bar bar bar bar bar bar Pressure Figure 2-3. Pressure range in the reviewed studies Of the studies conducted at 20 bar or higher, only four were directed towards OH-PLIF in cryogenic flames from a coaxial injector. No previous work other than the recent study done by Vaidyanathan et al. [23] involved OH-PLIF in gaseous shear injector studies at high pressures. Accurate measurement in gaseous environments is an important precursor to cryogenic studies to establish robust computational methods [1, 4]. In the current work, the temperature range will be selected to account for the variations in Boltzmann fraction and the OH concentration will be bracketed within the temperature range. The uncertainty sources and their contribution to the species concentration measurement is thus the major goal of this work. Furthermore the OH-PLIF measurements obtained as a part of this work, will compliment the very few existing data sets at high pressures. Equation Section (Next) 67

68 below. CHAPTER 3 EXPERIMENTAL FACILITY AND DIAGNOSTICS METHODS The experimental test facility, operating conditions and diagnostic methods are described Experimental Test Facility and Operating Conditions The experimental test facility consists of the combustion chamber, the injector and the propellant/purge feed system with valves and regulators. The schematic of the cross section of the combustion chamber with the injector assembly, windows for optical access and exhaust nozzle is shown in Figure 3-1. Quartz Windows (Uncooled) Segmented Chamber Wall Injector Assembly Exit Nozzle Figure 3-1. Combustion Chamber Cross Section The combustion chamber is made of oxygen free Copper. A detailed description along with the transient thermal analysis of the chamber could be found in Reference 74 and 76. The combustion chamber geometry cross section has an inner dimension of 25 mm x 25 mm and an outer dimension of 63.5 mm x 63.5 mm. The inner cross section has radius corners of 3 mm. The combustion chamber was equipped with UV grade fused silica windows for optical access. The windows are flush with the inner chamber wall and are not cooled. 68

69 The injector assembly houses a single element coaxial shear injector. The details of the injector assembly are shown in Figure 3-2 [74, 76]. Spacer D1, in(mm) D2, in(mm) D3, in(mm) (1.2) (2.2) (2.69) Oxidizer Tube Fuel Post Figure 3-2. Injector Details The injector and the fuel annulus were made of oxygen free copper and the injector housing was made of stainless steel. The oxidizer is injected straight into the chamber through the center tube while the fuel is injected through the annular region surrounding it. The injector is supported by a spacer to ensure that the oxidizer nozzle stays in the centre of the injector/fuel annulus assembly during the operation. The spacer also acts as a baffle to provide uniform distribution of fuel flow upstream to the chamber entrance. Other features in the combustion chamber including the exhaust nozzle assembly, segmented chamber extensions and igniter are described in detail in Reference 74. The exhaust nozzle is replaced with different areas to ensure the desired chamber pressure. The segmented chamber extensions are used to vary the chamber lengths and adjust the window locations relative to the injector. The combustion inside the chamber was initiated by spark ignition housed in one of the chamber extensions. The leads of the igniter were connected to a high voltage transformer capable of providing 10, 000 volts. 69

70 The propellant/purge system supplies the fuel and oxidizer for the experiments and nitrogen for purge. Both the propellants and the nitrogen are pressure fed from high pressure gas bottles through tubing that incorporate regulators and valves at their respective locations and are described in detail in Reference 74. The control of the propellant pressure and the mass flow rates for different test conditions and the opening and closing of the propellant lines at the beginning and end of the combustion tests are achieved using pressure regulators, regulating needle valves, solenoid valves and check valves. The DAQ/control system comprises of the power supply unit, DAQ/control hardware, DAQ/control software and the DAQ sensors. A detailed description of the DAQ system is given in Reference 74. Several temperatures are measured to provide boundary conditions for each measurement. Thermocouples are placed in chamber, injector face, exhaust nozzle and the heat flux thermocouples are housed in chamber walls. The chamber thermocouple is housed in the segmented chamber extension located immediately upstream of the exhaust nozzle and protrudes into the chamber and hence into the flame. An Omega K-type thermocouple is used as the chamber thermocouple with an inconel sheath of diameter 1.59 mm with an exposed tip and a response time of 15 ms. The chamber thermocouple is used to monitor the temperature rise during the runs which indicates the initiation and sustenance of combustion. The injector face thermocouples consist of two thermocouples housed in the injector face of the fuel annulus. An additional thermocouple is located behind the injector housing to detect possible backflow. The location of the injector face thermocouples are at 2.1 mm and 4.2 mm radially outwards from the center of the injector and a detailed schematic showing the thermocouple locations are given in Reference 74. The temperature measurements from the 70

71 injector face thermocouples are used to infer the temperature of the recirculation region formed at the injector face. The exhaust nozzle thermocouple measures outflow boundary conditions. The heat flux thermocouples are embedded in the side chamber walls. At each axial location there are two thermocouples. Depth location of the two heat flux thermocouples in side chamber walls at any chamber cross section is shown in Figure 3-3. Their axial location is given in detail in Reference 74 along with an analysis of wall heat fluxes. Heatflux thermocouple locations Figure 3-3. Location of Heat Flux Thermocouples, dimensions in mm The calculation of heat fluxes based on the temperature measurements from these two locations will be explained later in the section Wall Boundary Conditions section. The GH 2 /GO 2 experimental conditions investigated in the current study are tabulated and presented in Table 3-1. The nominal pressures were selected to cover the range from bars. The values indicated in the table are the actual measured values. 71

72 Table 3-1. Experimental Operating Conditions P bar O/F Massflow O/F Velocity F Hydrogen massflow g/s Hydrogen velocity m/s Exit nozzle ID mm Chamber Length mm The GH 2 /GO 2 combustion experiments lasted for 9.75 s following ignition for 10 bar case where as for all the other test cases the combustion run time was limited to 7.75 seconds following ignition. OH-PLIF Diagnostics For the PLIF experiments third harmonic output at 355 nm from a Nd-YAG (Continuum Surelite II) pulsed laser was used to pump the OPO (Continuum Panther). The FWHM spectral width of output beam measured using a Burliegh WA-4500 wavemeter was ~5 cm -1 and the centerline of the laser before doubling corresponded to as shown in the Figure 3-4. The measured spectral width was in agreement with the manufacturer s specification. The output from the OPO was frequency-doubled to obtain a UV beam at 283 nm. 72

73 Figure 3-4. Laser spectral profile measured using Burleigh Wavemeter before doubling to 283 nm The UV beam had a measured pulse energy of 0.89 mj and was used to excite the OH A-X (1,0) rotational transitions. The laser beam at 283 nm was formed into a sheet of 4 cm x 0.05 cm cross section using a series of fused silica lenses. The sheet was made 4 cm in height; however the central portion of 2 cm of the light sheet passed through the chamber to ensure that the wings of the Gaussian beam are not used for PLIF diagnostics. The schematic of the OH-PLIF diagnostic setup is shown in Figure 3-5. Fluorescence images were collected perpendicular to the direction of laser beam propagation using an ICCD camera (DiCam-Pro Cooke Corp.) equipped with 105mm/4 telephoto UV lens. The laser and the camera were synchronized using a pulse generator (DG 535 Stanford Research Systems) and were operating at 10 Hz. The camera was used in double shutter mode such that it collected fluorescence for 100ns in synchronization with the laser in the first image. The second image was collected 500ns after the first image for the same duration of 100ns to capture flame 73

74 emissions. The effective resolution of the camera was 66 micrometer/pixel in 4 x 4 binning mode. A combination of 3mm WG 305 Schott and 3 mm UG 11 filters were used to collect fluorescence from nm while effectively blocking elastic scattering. The combined transmission efficiency of the filters was about 55% between 306 and 320 nm. Figure 3-5. OH-PLIF Experimental Set-up OH-PLIF images were acquired for the entire run time at the rate of 10 Hz with 100 ns exposure time. Out of the instantaneous OH-PLIF images recorded, thirteen of the instantaneous images recorded at the near steady state at the end of the experiments were averaged and represented as averaged OH-PLIF image. Correspondingly, thirteen background emission images recorded by operating the camera in the double shutter mode were averaged. The instantaneous and averaged background emission images were then subtracted from the instantaneous and averaged OH-PLIF image respectively. 74

75 Wall Boundary Conditions The wall boundary conditions consist of wall heat fluxes determined from temperature measurements in the combustion chamber. Conley et al. [76] calculated the heat fluxes from the temperature measurements. In this and in the previous study [75, 76] the heat fluxes were calculated by solving the steady state one dimensional heat conduction equation and adding a correction term to compensate the heat absorption by the chamber as shown below k T T Δx ( i o ) r CΔx T 2 T o,2 o,1 q A =,2,2 + Δt (3-1) where q A, is the heat flux, Δx is the distance between the thermocouple pairs, the subscripts i is assigned for thermocouple close to the inner chamber wall, o represents the one farthest and 1 and 2 represents the initial and final times respectively. The very nature of heat transfer in the combustion chamber is three dimensional. Thus calculation of heat fluxes based on 1D assumption can introduce errors. Thus in the current work the heat flux was calculated by numerically solving the unsteady 3D heat conduction equation. The method will be discussed in detail in the following section. The chamber extensions shown in Figure 3-3 incorporate thermocouple pairs placed next to each other and separated by 7 mm in the transverse direction. For each thermocouple pair, the temperatures are measured at 3.2 and 9.5 mm from the inner chamber walls respectively. The temperature at location 3.2 mm from the inner wall is denoted as T inner and that measured at 9.5 mm is denoted as T middle. A 3D model of the central portion of the chamber from 37 to 102 mm from the injector face was chosen as the computational domain. The central portion of the chamber was selected since the temperature measured outside of this domain did not indicate an axial temperature 75

76 gradient. The outer wall was assumed to be insulated such that the heat released during the experiment was assumed to be accumulated in the chamber. The validity of the insulated wall assumption was checked by imposing forced convection at the outer walls, assuming outer wall temperature to be at 100 o C and ambient air temperature set at 27 o C. Forced convection was calculated by assuming an air velocity of 10 m/s. These conditions are considerably more dissipative than experienced during the experiments. The heat transfer for the case of a laminar forced convection past a flat plate with the prescribed values was calculated. The heat transfer thus determined was 0.1% of heat flux values in the chamber walls due to combustion and thus all the heat released during the transient process was assumed to accumulate in the walls. The computational temperatures which evolved over the period of 7.75 seconds were matched with the actual temperatures obtained from the experimental run at the inner and middle locations which are at 3.2 and 9.5 mm from the inner chamber walls, respectively. The imposed heat flux at the inner chamber wall was changed for different sets of computation so that the temperatures T inner and T middle obtained from the computations, matched the experimental results within 4 to5 o C. The discretized 3D heat conduction equation [77] is k ( ) ( T ) ( ) ( ) 2 i+ 1, j, l, t 2Ti, j, l, t+ Ti 1, j, l, t + T 2 i, j+ 1, l, t 2Ti, j, l, t+ Ti, j 1, l, t dx dy k r C + + ( ) ( T ) ( ) ( ) 2 i, j, l+ 1, t 2 Ti, j, l, t Ti, j, l 1, t = T 2 i, j, t+ dt Ti, j, t dz dt k (3-2) Here, the density, r, and heat capacity, C, for Copper 110 are 8700 kg/m 3 and 385 J/(kg K), respectively The computational domain consisted of a 51 x 51 x 51 grid and the time step was seconds. The heat flux obtained through this procedure is used to accompany the in-flow species concentration measurement in the process of code validation. The Matlab scripts used for data processing are detailed in Appendix A.Equation Section (Next) 76

77 CHAPTER 4 OH-PLIF IMAGE PROCESSING AND QUANTITATIVE ANALYSIS The OH-PLIF image processing, the methodology for determining the OH concentration and the uncertainties in the measurement analyses are discussed below. Fluorescence and Interference Signals The background emission was subtracted from the fluorescence + background images. The images for the four different test cases are shown in Figure 4-1 to 4-4. Figure 4-1(a) to 4-4(a) show the raw image which has been corrected by subtracting the background shown in Figure 4-1(b) to 4-4(b). The intensity levels of the background subtracted OH-PLIF images are shown in Figure 4-1(c) to 4-4(c). 77

78 H 2 O 2 H Height (mm) (a) Width (mm) H 2 O 2 H Height (mm) (b) Width (mm) H 2 O 2 H Height (mm) (c) Width (mm) Figure 4-1. Average of 13 instantaneous images obtained at near steady state for chamber pressure of 10 bar; (a) OH-PLIF + background emission image, (b) background emission image and (c) OH-PLIF image 78

79 H 2 O 2 H Height (mm) (a) Width (mm) H 2 O 2 H Height (mm) (b) Width (mm) H 2 O 2 H Height (mm) (c) Width (mm) Figure 4-2. Average of 13 instantaneous images obtained at near steady state for chamber pressure of 27 bar; (a) OH-PLIF + background emission image, (b) background emission image and (c) OH-PLIF image 79

80 H 2 O 2 H Height (mm) (a) Width (mm) H 2 O 2 H Height (mm) (b) Width (mm) H 2 O 2 H Height (mm) (c) Width (mm) Figure 4-3. Average of 13 instantaneous images obtained at near steady state for chamber pressure of 37 bar; (a) OH-PLIF + background emission image, (b) background emission image and (c) OH-PLIF image. 80

81 H 2 O 2 H Height (mm) (a) Width (mm) H 2 O 2 H Height (mm) (b) Width (mm) H 2 O 2 H Height (mm) (c) Width (mm) Figure 4-4. Average of 13 instantaneous images obtained at near steady state for chamber pressure of 53 bar; (a) OH-PLIF + background emission image, (b) background emission image and (c) OH-PLIF image 81

82 From Figure 4-1 to 4-4, it is evident that at higher pressures of 37 and 53 bar, the background emissions are stronger than at 10 bar. This shows that collection of fluorescence with a gate-width narrowed to 100 ns was not sufficient to suppress the flame emissions. The sources of the background emissions are recognized as typical flame emissions from OH and water molecules in a H 2 /O 2 flame [8]. The spectra of background emissions from a LOx/GH 2 flame in the range of 300 to1100 nm at 60 bar was measured by Mayer et al. [15] and found that the contributions from the OH and O 2 flame emissions lied in the 300 to 400 nm range, the predominant spectra being OH A-X (0,0) branch at 310 nm. The contributions from H 2 O could be found spanning the 400 to1000 nm range. It is noteworthy to note the transmission of UV filters used in this study to block the elastic scattering and transmitted light in the range of 300 to 400 nm and above 650 nm. The UV filters WG 305 & UG 11 served as the best combination considering the low laser energy of 0.89 mj /pulse and the large laser line width of 5cm -1 used here, which both tend to decrease the fluorescence signal strength. Thus considering the emissions from H 2 /O 2 flame and the transmission range of UV filters, the background emissions observed in the current study were identified as due to OH, O 2 and H 2 O. Laser The laser shot to shot power fluctuation was monitored for 290 pulses. The average of the 290 laser pulse energies accounted to 0.89 mj/ pulse with a standard deviation of 0.10 mj/ pulse. The fluctuation in the laser energy accounted for an uncertainty of 11 %. The laser sheet profile variation in space was corrected from calibration using acetone fluorescence. The laser sheet at 283 nm was passed through the chamber filled with acetone vapor and the 2D fluorescence was collected by effectively blocking the elastic scattering using the UV filters and ICCD camera. Ninety acetone fluorescence images were averaged and normalized with the maximum intensity/counts along the width to obtain the spatial variation of 82

83 the laser sheet in the region of interest. The normalized laser sheet profile variation in a percentage intensity scale is shown in Figure 4-5. The laser sheet had maximum intensity of above 90% at heights of 11 to 15 mm while it gradually decreased to 25% at heights of 1 and 19 mm on either side. Figure 4-5. Normalized laser sheet intensity profile variation obtained from acetone fluorescence images. The intensity is provided in percentage scales. The intensity is above 90 % at heights of 11 to 15 mm and gradually decreases to 25% at heights of 1 and 19 mm Based on the normalized acetone fluorescence images shown in Figure 4-5, the laser sheet intensity variation in space was corrected for all the OH-PLIF images acquired in the current study and the resultant uncertainty calculated as the ratio of the standard deviation to the average values of the 90 normalized fluorescence images was 5.9 %. The absorption of the laser sheet by OH and other molecules that interact with the laser beam as it passed through the combustion chamber need to be further discussed. In GH 2 /GO 2 combustion one of the main combustion products is water vapor. The absorption cross section of H 2 O between 190 and 320 nm at K temperature ranges and pressures up to 70 bar increases as a function of temperature [70]. However the effect of absorption is small enough for wavelengths above 280 nm such that the absorption by H 2 O can be neglected. The laser beam absorption, as it traversed through the region of interest was insignificant as seen in Figure 4-1 to 83

84 4-4. The laser beam absorption by OH will be estimated based on the Beer-Lambert s law, once the OH number density is determined. Absorption and Excitation, Line Shape, and Fluorescence Efficiency The fluorescence signal in Equation 2-19 is rearranged and the number density of OH, o noh is expressed as n o OH Np = E fbb12 A21 Ω 2 ( laser absdν ) V A c Φ Φ A21 + Q21 4π (4-1) f 12 The term B B 2 ( Φ Φ dν laser abs ) c represents the overlap between the laser spectral profile and the specific rovibrational absorption profile of the molecule under consideration. This is valid when the laser spectral width is small enough that it does not excite other rovibrational branches existing nearby. In the experiment carried out here, the OPO spectral width of 5 cm -1 was large enough to excite a series of nine rovibrational lines around 283 nm. In this case, the collected signal, Np expressed in Equation 4-1 needs to be modified so that the excitations of all the rovibrational transitions lying within the spectral bandwidth of the OPO are properly taken into account. Thus Equation 4-1 is modified to include the contributions of a series of rotational lines resulting in n o OH Np = 9 B12 E fb 1 A 21 Ω 2 ( ΦlaserΦabsdν ) V A c A21 + Q21 4π Before the interpretation of the concentration from Equation 4-2 the parameters in the (4-2) expression need to be examined in detail. The nine rotational transitions of OH A-X (1-0) lying 84

85 within the spectral width of the laser having a Gaussian profile centered at cm -1 with FWHM 5 cm -1 as shown in Figure 3-5 were identified as P (6) + Q (3) + R (3) + Q (6) + Q (1) + R (1) + Q (2) + R (2) + R (14) [37] The term dependent due to 9 1 f B c 2 B 12 9 ' denoted as f B B (cmj -1 ), where 12 1 f B. The line shape Φ abs in ( Φ Φ dν laser abs ) B ' 12 2 B = 12 c is temperature is both temperature and pressure dependent and the quench rate Q determined from Equation 2-5 also requires the knowledge of 21 temperature field and colliding species mole fraction as discussed in Chapter 2. The different approaches to circumvent this problem were identified from the review of OH-PLIF diagnostic studies as calibration measurements via Rayleigh/Raman measurements of temperature & species and calculations / numerical simulation based on equilibrium conditions. ' The other approach is to obtain the variation in f B B, 12 Φ Φ dν and Q with laser abs 21 temperature and colliding species mole fraction corresponding to a broad range of equivalence ' ratio. The approach used in this study is to use the average values of f B B, Φ Φ dν 12 laser abs and Q for the calculations, and determining the uncertainty in the OH concentration due to the 21 variation over a broad range of equivalence ratio. The resultant variation and corresponding uncertainties will be presented and discussed in chapter The term A A 21 + Q in Equation 4-2 known as the fluorescence yield needs to be further analyzed. The effect of quenching becomes predominant at high pressures when A 21 << Q 21. In 85

86 GO 2 /GH 2 combustion the colliding species are mainly H 2 O, O 2 and H 2 molecules. The corresponding colliding cross sections [53] are given in Table 4-1. Table 4-1. Colliding Species Cross Section for Collisional Quenching Species H 2 O O 2 H 2 Colliding species cross section (Å 2 ) The fluorescence yield based on the Equation 2-5 is well represented for A-X (0-0) transitions. For transitions also involving A-X (1-1) the expression for the fluorescence yield needs to be modified [20, 53] to A21(0,0) A21(1,1) s 0 s 10 s10 F = + 1+ Q 21 A 21(0,0) s1 s1 s1 1 (4-3) where A 21(0,0) and A 21(0,0) represent the spontaneous emission rates from A-X(0,0) and A-X(1,1), respectively. s 0 and s 1 represent the total effective cross sections for quenching from vibration levels ' u = 0 and ' u = 1 respectively. The vibrational energy transfer from ' u =1 to 0 is represented by s 10. The approximate value [53] for A A 21(1,1) 21(0,0), s s 0 s, 10 1 s 1 are 0.575, 1and 0.58 respectively. Experimental Constants The OH-PLIF diagnostic in this study is associated with 2D imaging of the fluorescence on a CCD chip. Thus the volume V (cm 3 ) in Equation 4-2 corresponding to the collected fluorescence signal intensity in each pixel in the camera is equal to the product of the pixel projection area A Pixel Projection (cm 2 ) and the laser sheet thickness l (cm). The uncertainty associated with the volume probed is due to the accurate determination of the pixel resolution. The pixel resolution was obtained by calibrating it against the accurately 86

87 known dimensions of a wire with constant diameter and length. The resultant uncertainty due to the variation in pixel resolution accounted to 2.8 % uncertainty in the probe volume. To obtain absolute OH concentration in number density, the arbitrary selected unit, counts of the camera, are to be converted to photometric units. This was done by calibration of the camera against a light source of known irradiance. The light source used in this study was a thousand watt, quartz halogen, tungsten filament lamp with designation of from Oriel Instruments. The uncertainty in the irradiance levels near the 310 nm wavelength was 2.3 % as mentioned in the lamp specifications. The camera calibration corresponding to the detection strategy employed in the OH-PLIF measurements and region of interest is shown in Figure 4-6. Number of Photons(N p ) Data Photons Linear fit N p = 1.59 *Counts Counts Figure 4-6. Camera calibration corresponding to the detection strategy employed in the OH-PLIF measurements and region of interest. The uncertainty in the calibration due to non linearity associated with the fit of 1.8 % together with the uncertainty in the lamp irradiance of 2.3 % amounted to a net uncertainty of 2.9 % in photon calibration. The uncertainty in the photon calibration due to the non-linearity associated with the fit is calculated and is 1.8 %. The net uncertainty in the photon calibration due to lamp irradiance uncertainty of 2.3 % and uncertainty of 1.8% due to non linearity in calibration fit accounted to 87

88 2.9%. A proposed new methodology to calibrate the camera as part of the thesis study is explained in Appendix B. The calibration obtained from the new method is compared with the conventional calibration shown in Figure 4-6. As the photons are Poisson distributed [35], there is uncertainty in the exact number of photons detected; this is called shot noise. The uncertainty contribution from shot noise due to the Poisson distributed photon number was calculated as the ratio of the standard deviation to the average photon arrival from the OH-PLIF images at 10, 27, 37 and 53 bar. The uncertainty contribution due to shot noise at 10, 27, 37 and 53 bar accounted to 6.9, 7.05, 6.8 and 6.7% respectively. The camera has spatial variation of pixel intensities. The systematic and random spatial variation is eliminated by linear filtering in which the value of an output pixel in the image is computed as a weighted average of neighboring pixels [11, 78]. In the current study, each pixel value was computed as a weighted average of the neighboring 5 x 5 matrix of pixels with equal weights. The uncertainty due to systematic and random spatial variation of pixel intensities, minimized by linear filtering [11, 78] in which the value of an output pixel in the image is computed as a weighted average of neighboring pixels, was calculated as the ratio of the difference in pixel intensities before and after filtering to their corresponding averaged values. The uncertainty contribution due to pixel smoothening of the OH-PLIF images at 10, 27, 37 and 53 bar accounted to 7, 7, 6.3 and 6 % respectively. The Matlab scripts used for data processing are detailed in Appendix A. 88

89 CHAPTER 5 RESULTS AND UNCERTAINTY ANALYSIS Experiments at high pressure GH 2 /GO 2 combustion covered Oxygen to Fuel (O/F) mass flow ratio of 3.77 corresponding tof =2.15 and pressures of 10, 27, 37 and 53 bar. The results presented here include (i) Experimental Conditions and Chamber pressure measurements and (ii) Image processed OH-PLIF measurements. The uncertainty analysis includes determination of uncertainties for the OH-PLIF measurements at 37 bar. Chamber Pressure Measurements In the GH 2 /GO 2 experiments the chamber pressure was increased by increasing the propellant mass flow rates while keeping the O/F mass flow and velocity ratios constant for a constant exhaust nozzle area. The chamber pressure rise in time for the four experimental conditions of GH 2 /GO 2 combustion is shown in Figure 5-1 to 5-4. The sequence included nitrogen pressurization followed by fuel injection and ignition. It should be noted that ignition is identified in the figures by the high oscillation induced in the sensor recording. The pressure increases in time and for higher pressures, which are of interest here, attains a near steady state at 7 8 sec following ignition. To attain steady-state at lower pressures longer experimental time would have been required. 89

90 Ignition Chamber Pressure Combustion shut-off Pressure (bar) Nitrogen pre-pressurization OH-PLIF images range time(s) Figure 5-1. Chamber pressure versus time for GH 2 /GO 2 combustion for 10 bar and O/F mass flow of 3.7. The experiment continued for 10 sec following ignition Chamber Pressure Combustion shut-off 25 Pressure (bar) Ignition Nitrogen pre-pressurization OH-PLIF images range time(s) Figure 5-2. Chamber pressure versus time for GH 2 /GO 2 combustion for 27 bar and O/F mass flow of 3.7 The experiment continued for 8 s following ignition. 90

91 Chamber Pressure Combustion shut-off 35 Pressure (bar) Ignition OH-PLIF images at near steady state Nitrogen pre-pressurization time(s) Figure 5-3. Chamber pressure versus time for GH 2 /GO 2 combustion for 37 bar and O/F mass flow of 3.7. The pressure attains a near steady state at the end of 8 s following ignition Chamber Pressure Combustion shut-off 50 Pressure (bar) Ignition OH-PLIF images at near steady state 20 Nitrogen pre-pressurization time(s) Figure 5-4. Chamber pressure versus time for GH 2 /GO 2 combustion for 53 bar and O/F mass flow of 3.7. The pressure attains a near steady state at the end of 8 s following ignition. 91

92 OH-PLIF Measurements The OH-PLIF images acquired for the experiments include thirteen instantaneous images. At each pressure the instantaneous images were averaged. The instantaneous images, acquired over a period of 100 ns each are needed for validation of LES codes while the average is used for validation of RANS codes. The OH-PLIF images shown in Figure 5-5 to 5-6 were image processed to eliminate background emissions; correct for spatial variation in laser intensity; smoothen the images and minimize the spatial variation of pixel intensities; Figure 5-5 to 5-6 show an example of an instantaneous image and average image for each pressure case. 92

93 (a) (b) (c) Figure 5-5. Instantaneous image-processed OH-PLIF images at near steady state chamber pressure of (a) 10, (b) 27, (c) 37 and (d) 53 bar (d) 93

94 (a) (b) (c) (d) Figure 5-6. Average of thirteen instantaneous image-processed OH-PLIF images at near steady state chamber pressure of (a) 10, (b) 27, (c) 37 and (d) 53 bar. 94

95 To ascertain the repeatability of the OH-PLIF measurements the average of imageprocessed OH-PLIF images acquired at 35, 36 and 37 bar pressure cases for similar experimental conditions were compared and are shown in Figure 5-7 (a c). The average OH-PLIF images in Figure 5-7 (a c) shows that the OH-PLIF measurements were repeatable and can be used for determination of OH concentration with confidence. (a) (b) (c) Figure 5-7. Average of thirteen instantaneous image-processed OH-PLIF images at near steady state chamber pressure of (a) 35, (b) 36, and (c) 37 bar indicating the repeatability and reliability of OH-PLIF measurements for determination of OH concentration. 95

96 The intensity levels of the image processed OH-PLIF images shown in Figure 5-5 to 5-6 are related to the number density of OH by Equation 4-2. Theses image need to be processed as described in Chapter 4 and thus the OH number density is determined. The OH-PLIF images in Figure 5-5 to 5-6 show certain interesting features that are noteworthy. From the OH-PLIF image at 10 bar, it can be observed that the flame is smooth and less corrugated as seen from OH images at higher pressures. For the four experimental conditions, O/F velocity and density ratios governing the development of shear layer remain the same. The difference in the four experiments is the turbulence and the heat release. In addition to these there can be also Soret and Dufour cross diffusion effects arising from concentration and temperature gradients. These secondary effects need to be evaluated in complimentary CFD efforts. As noted, the OH radical in a non premixed flame is considered to be a good marker of the reaction zone. Similar to the study described in Reference 32, the stoichiometric contour was traced from the axial evolution of the location of maximum OH intensity in the flame, as indicative of the mean position of the reaction zone [32]. Thus from all four average OH PLIF images, shown in Figure 5-6(a d), the mean position of the reaction zone was quantitatively determined and is shown in Figure 5-8(a d). 96

97 (a) (b) (c) (d) Figure 5-8. Mean position of reaction zone determined from the average OH-PLIF images at (a) 10, (b) 27, (c) 37 and (d) 53 bar. The OH-PLIF signal at 27 and 37 bar would indicate a lifted flame, however this is an effect of strong background correction. 97

98 The mean position of the reaction zone at 10 bar shows that the flame is anchored at the lip of the oxidizer post and is typical of the coaxial shear flames [20]. For test cases at higher pressures of 27 and 37 bar the OH-PLIF signal would indicate a lifted flame, however this is an effect of strong background correction. In all cases the flame is anchored at the lip. The shear layers merge at mm from the injector face at pressures of 27, 37 and 53 bar. The location of the maximum OH concentration is similar for all cases, explaining the similar effect of same density and velocity ratio on shear layer development regardless of the difference in turbulence and heat release rates. To analyze the effect of turbulence the Reynolds number of GO 2 and GH 2 were calculated ρud1 as μ and ( ) ρu D D 3 2 μ where ρ is the density, U, the velocity and μ the dynamic viscosity of the gas, and D1, D2 and D3 are the dimensions of injector as shown in Figure 3-2 respectively. The Reynolds number, Re D for GO 2 was 38100, 75380, and and for GH 2, Re D was 5752, 11534, and at 10, 27, 37 and 53 bar respectively. The Re D of GH 2 and GO 2 clearly indicates that flow regime is turbulent. The momentum flux ratio was ( 2 ρu ) defined as J = 2 ( ρu ) GO2 GH2. For all the pressure cases the momentum flux ratio that governs the growth of the shear layer remained the same and was 2.7. In the study by Seitzman et al. [25] the OH structures in turbulent non-premixed hydrogen flame were characterized at Re D of 2300, 8600, and It was found that, as the flow transits from laminar to turbulent regime, there is significant change in the OH structures from low strain rate, thick filament zones to high strain rate, thin filament, more diffuse regions. Another notable observation was that at higher Reynolds number the OH structures became increasingly convoluted and similar behavior was observed in the current study also. 98

99 The studies which focused on shear coaxial cryogenic flames [11, 12, 13, 19, 20, 21] identified the wrinkling, corrugation and flapping of the flame to be caused by the combined effects of turbulence and instabilities in the flow field. Singla et al. [20, 21] proposed stability criteria based on the ratio of oxidizer lip thickness to the flame thickness and for the flame to be stable the ratio needed to be greater than one. As the flame anchors on the oxidizer lip, the size and dynamics of the recirculation region in the lip wake influences the flame stability. Thus, in the current study the wrinkling and corrugation of the flame, at higher pressures with higher Reynolds number, is attributed to the increased turbulence where as the flapping of the flame which is evident from the instantaneous OH distribution in Appendix D is attributed to the instability dictated by two factors: (i) the size of the recirculation zone in the wake of oxidizer lip and, (ii) the large scale flow fluctuation in the recirculation region formed on the injection face around the jet injectors. 99

100 Quantification of OH Concentration and Uncertainty at 10, 27, 37 and 53 bar To determine the number density from the image-processed OH-PLIF images in Figure ' to 5-6, the values of f B B, 12 Φ Φ dν and Q in Equation 4-2 are calculated as follows. laser abs 21 1 A broad range of equivalence ratio for GH 2 /GO 2 combustion was considered and covered range [73]. The OH radical probed in the shear reaction zone of the GH 2 /GO 2 combustion is known to exist mostly around the region of stoichiometry. Hence, the equivalence ratio of could be considered a very broad range of conditions in the flame. Therefore this assumption is quite conservative and is expected to yield a larger uncertainty than actually encountered in the experiment. However given the lack of data it was adopted here to bracket with confidence the possible experimental uncertainty. In a first approximation equilibrium conditions are assumed. The equilibrium conditions for the chemical reactions pertaining to the GH 2 /GO 2 experiments carried out at 10, 27, 37, and 53 bar was calculated using STANJAN [79]. The variation of temperature and mole fraction of species, H 2 O, H 2, and O2 with equivalence ratio of is shown in Figure 5-9(a d). It was found that, the temperature varied between 2500 and 3500 K, with the maximum at stoichiometry. 100

101 1 Mole fraction H 2 O 10bar Mole fraction H 2 10 bar Mole fraction O 2 10 bar Temperature 10 bar Mole fraction Temperature( o C) Equivalence ratio (φ) 1 (a) Mole fraction H 2 O 27 bar Mole fraction H 2 27 bar Mole fraction O 2 27 bar Temperature 27 bar Mole fraction Temperature( o C) Equivalence ratio (φ) (b) 101

102 1 Mole fraction H 2 O 37 bar Mole fraction H 2 37 bar Mole fraction O 2 37 bar Temperature 37 bar Mole fraction Temperature( o C) Equivalence ratio (φ) (c) 1 Mole fraction H 2 O 53 bar Mole fraction H 2 53 bar Mole fraction O 2 53 bar Temperature 53 bar Mole fraction Temperature( o C) Equivalence ratio (φ) Figure 5-9. Temperature and specie mole fraction variation based on equilibrium calculations with equivalence ratios of at (a) 10, (b) 27, (c) 37 and (d) 53 bar. The temperature has a maximum value of 3500 K at stoichiometry and decreases to a minimum of 2500 K at equivalence ratio of 3. (d) 102

103 ' The variation of f B B (cmj -1 ) with temperature which in turn varies with the equivalence ratio is shown in Figure 5-10(a d). Absorption Coefficient (cmj -1 ) Absorption Coefficient 10 bar Mean Absorption Coefficient 10 bar Absorption Coefficient (cmj -1 ) Equivalence ratio (φ) (a) Absorption Coefficient 27 bar Mean Absorption Coefficient 27 bar Equivalence ratio (φ) (b) 103

104 Absorption Coefficient (cmj -1 ) Absorption Coefficient 37 bar Mean Absorption Coefficient 37 bar Absorption Coefficient (cmj -1 ) Equivalence ratio (φ) (c) Absorption Coefficient 53 bar Mean Absorption Coefficient 53 bar Equivalence ratio (φ) (d) ' Figure Absorption coefficient ( f B B ) variation with equivalence ratio and temperature 12 ( K) at (a) 10, (b) 27, (c) 37 and (d) 53 bar showing that the variation with respect to mean is 12.4, 14.6, 14.5 and 15.1% respectively. ' The mean value of f B B is used for the calculation. The uncertainty due to the variation with temperature/equivalence ratio with respect to mean at 10, 27, 37 and 53 bar is 12.4, 14.6, 14.5 and 15.1 % respectively. 104

105 The absorption profiles of OH at 10, 27, 37 and 53 bar were simulated using LIFBASE [37]. To simulate the absorption profile Φ abs, the collisional and Doppler widths were obtained from Equation 2-13 and Equation 2-15 respectively. The absorption profiles at 3017 K and 10 bar, 3085 K and 27 bar, 3100 K and 37 bar, and 3125 K and 53 bar for an equivalence ratio of 2 along with the laser spectral profile are shown in Figure 5-11(a d) Laser Profile OH Absorption Profile at T =3017 and 10 bar (a.u) x ν (cm -1 ) (a) Laser Profile OH Absorption Profile at T =3085 and 27 bar (a.u) x 10 4 ν (cm -1 ) (b) 105

106 1 0.9 Laser Profile OH Absorption Profile at T =3103 and 37 bar (a.u) x ν (cm -1 ) (c) Laser Profile OH Absorption Profile at T =3125 and 53 bar (a.u) x 10 4 ν (cm -1 ) (d) Figure Absorption profile of OH at (a) 3017 K and 10 bar, (b) 3085 K and 27 bar, (c) 3103 K and 37 bar, and (d) 3125 K and 53 bar simulated using LIFBASE showing a complete overlap with the laser spectral profile at all pressures. The OH absorption profiles at 10, 27, 37 and 53 bar were simulated for a broad temperature range of K corresponding to the equivalence ratio of and are 106

107 provided in the Appendix C. The OH absorption profile broadens and the centre line of the OH absorption profile shifts with pressure and temperature. The overlap integral of the absorption profile of OH at 10, 27, 37 and 53 bar, and the Gaussian spectral laser profile, Φ is calculated by Φ Φ dν. As a result of the variation laser abs laser of the absorption profile with temperature, the determined overlap integral also varies for each pressure case over the broad range of temperature in the flame. To find out the variation the overlap integral is calculated for a temperature range of K corresponding to an equivalence ratio of at 10, 27, 37 and 53 bar. The results are shown in Figure 5-12(a d) Overlap Integral 10 bar Mean Overlap Integral 10 bar Overlap(cm) (a) φ 107

108 Overlap Integral 27 bar Mean Overlap Integral 27 bar Overlap(cm) (b) φ Overlap Integral 37 bar Mean Overlap Integral 37 bar Overlap(cm) (c) φ 108

109 Overlap Integral 53 bar Mean Overlap Integral 53 bar Overlap(cm) (d) Figure Overlap integral Φ Φ dν laser abs variation at (a) 10, (b) 27, (c) 37 and (d) 53 bar with temperature corresponding to equivalence ratio of 0.5 3, indicating that the variation with respect to mean is 1.3, 1, 0.8 and 0.5% respectively and can be assumed negligible. The uncertainty due to the variation in the overlap integral at 10, 27, 37 and 53 bar is determined as 1.3, 1, 0.8 and 0.5 % respectively over the broad temperature range of K and is therefore assumed negligible. The overlap integral could also vary due to the centre line shift of the laser profile. The center line of the laser profile was measured as from the Burleigh Wavemeter. The uncertainty in the overlap integral variation at 10, 27, 37 and 53 bar due to the laser center line shift accounts to 2.8, 1.6, 1 and 0.2 % respectively. The absorption profile broadens and gets shifted as pressure increases. Hence in most of the studies carried out using lasers with small spectral widths of less than 1 cm -1, the centre line wavelength of the laser needs to be shifted in order to overlap with the center line wavelength of the OH absorption profile. The most important area of concern is the pressure broadening. The overlap between the laser spectral profile and the pressure broadened absorption profile φ 109

110 decreases leading to a decrease in the strength of the collected fluorescence signal as the pressure increases. The spectral width of the laser employed in this study was larger than the spectral width of the broadened absorption profile even at elevated pressures such as 37 and 53 bars. This can be considered as an advantage since it was ensured that the laser profile overlapped with the absorption profile at all pressures thereby ensuring fluorescence with good signal strengths. The variation of quench rate, Q at 10, 27, 37 and 53 bar with temperature and the species 21 mole fraction corresponding to equivalence ratio of is shown in Figure 5-13(a d). Collisional Quench Rate(s -1 ) 1.22 x Collisional Quench rate 10 bar Mean Collisional Quench rate 10 bar Equivalence ratio (φ) (a) 110

111 Collisional Quench Rate(s -1 ) 3.3 x Collisional Quench rate 27 bar Mean Collisional Quench rate 27 bar x Equivalence ratio (φ) (b) Collisional Quench rate 37 bar Mean Collisional Quench rate 37 bar Collisional Quench Rate(s -1 ) Equivalence ratio (φ) (c) 111

112 Collisional Quench Rate(s -1 ) 6.5 x Collisional Quench rate 53 bar Mean Collisional Quench rate 53 bar Equivalence ratio (φ) (d) Figure Collisional quench rate Q 21 variation at (a) 10, (b) 27, (c) 37 and (d) 53 bar with temperature and colliding species mole fraction corresponding to equivalence ratio of indicating that the variation with respect to mean is 4.1, 3.9, 3.8 and 3.7 % respectively. The mean value of Q is used to calculate F in Equation 4-3. The uncertainty due to the 21 variation of Q at 10, 27, 37 and 53 bar with temperature and colliding species mole fraction 21 corresponding to equivalence ratio of with respect to mean is 4.1, 3.9, 3.8 and 3.7 %, respectively. All the parameters in Equation 4-2 required for determination of OH number density was calculated and the image-processed OH-PLIF images in Figure 5-6 and 5-7 were converted into absolute concentration. Figure 5-14(a d) and 5-15(a d) represents instantaneous and averaged OH concentration at 10, 27, 37 and 53 bar respectively. Appendix D includes complete set of instantaneous OH number density contours for all the pressure cases. 112

113 (a) (b) (c) (d) Figure Instantaneous OH number density contours at near steady state chamber pressure of (a) 10, (b) 27, (c) 37 and (d) 53 bar 113

114 (a) (b) (c) (d) Figure Average of thirteen instantaneous OH number density contours at near steady state chamber pressure of (a) 10, (b) 27, (c) 37 and (d) 53 bar. 114

115 The uncertainty due to laser absorption by OH creating a horizontal incident photon flux gradient, estimated from average OH number density using Beer-Lambert s law [35], I I o 9 f B B12 1 o = exp -hν 2 ( ΦlaserΦabsdν) noh z, for z= 1mm path length of the laser at 10, 27, c 37 and 53 bar was 2, 3.3, 3.8 and 4.7 % respectively. In summary, the uncertainties associated with OH quantitative measurement based on the conservative assumptions made here were quantified for 10, 27, 37 and 53 bar cases as shown in Figure 5-16(a d). n o p OH 9 B12 E f B A ( Φ Φ dν laser abs ) Camera Camera Calibration 2.9% Shot noise 6.9% Pixel Smoothening 7% Other Signals Laser scattering - blocked Background emission - corrected Fluorescence trapping - negligible for A-X(1,0) Laser Shot to shot power fluctuation - 11% Laser sheet spatial variation- 5.9% Laser absorption (OH) 2% Laser absorption(h 2 O) - negligible N = W V A c A Q 4 π Absorption Boltzmann factor (Temperature) Absorption Coefficient (Spectroscopy) Line Shape Overlap integral (line shape) Overlap integral (laser center line shift) Model (Collisional & Doppler width/shift) Fluorescence Efficiency Quench rate (Collider species cross section/ mole fraction,pressure, Temperature ) Model for quantum yield 12.4% Total uncertainties (rms error) = 21.4 % (a) Volume Pixel area 2.8% 1.3% 2.8% 4.1% 115

116 n o p OH 9 B12 E f B A ( Φ Φ dν laser abs ) Camera Camera Calibration 2.9% Shot noise 7% Pixel Smoothening 7% Other Signals Laser scattering - blocked Background emission - corrected Fluorescence trapping - negligible for A-X(1,0) Laser Shot to shot power fluctuation - 11% Laser sheet spatial variation- 5.9% Laser absorption (OH) 3.3% Laser absorption(h 2 O) - negligible n Other Signals Laser scattering - blocked Background emission - corrected Fluorescence trapping - negligible for A-X(1,0) Laser Shot to shot power fluctuation - 11% Laser sheet spatial variation- 5.9% Laser absorption (OH) 3.8% Laser absorption(h 2 O) - negligible N = W V A c A Q 4 π Absorption Boltzmann factor (Temperature) Absorption Coefficient (Spectroscopy) Line Shape Overlap integral (line shape) Overlap integral (laser center line shift) Model (Collisional & Doppler width/shift) Fluorescence Efficiency Quench rate (Collider species cross section/ mole fraction,pressure, Temperature ) Model for quantum yield 14.6% Total uncertainties (rms error) = 22.8 % (b) o p OH 9 B12 E f B A ( Φ Φ dν laser abs ) Camera Camera Calibration 2.9% Shot noise 6.8% Pixel Smoothening 6.3% Absorption Boltzmann factor (Temperature) Absorption Coefficient (Spectroscopy) Line Shape Overlap integral (line shape) Overlap integral (laser center line shift) Model (Collisional & Doppler width/shift) Fluorescence Efficiency Quench rate (Collider species cross section/ mole fraction,pressure, Temperature ) Model for quantum yield Volume Pixel area 2.8% 1% 1.6% 3.9% N = W V A c A Q 4 π 14.5% Total uncertainties (rms error) = 22.5 % (c) Volume Pixel area 2.8% 0.8% 1% 3.8% 116

117 n o p OH 9 B12 E f B A ( Φ Φ dν laser abs ) Camera Camera Calibration 2.9% Shot noise 6.7% Pixel Smoothening 6% Other Signals Laser scattering - blocked Background emission - corrected Fluorescence trapping - negligible for A-X(1,0) Laser Shot to shot power fluctuation - 11% Laser sheet spatial variation- 5.9% Laser absorption (OH) 4.7% Laser absorption(h 2 O) - negligible N = W V A c A Q 4 π Absorption Boltzmann factor (Temperature) Absorption Coefficient (Spectroscopy) Line Shape Overlap integral (line shape) Overlap integral (laser center line shift) Model (Collisional & Doppler width/shift) Fluorescence Efficiency Quench rate (Collider species cross section/ mole fraction,pressure, Temperature ) Model for quantum yield 15.1% Total uncertainties (rms error) = 22.9 % (d) Volume Pixel area 2.8% 0.5% 0.2% 3.7% Figure OH-PLIF measurement uncertainties at (a) 10, (b) 27, (c) 37 and (d) 53 bar. The rms error include the contributions from (i) camera calibration (ii) shot noise, (iii) pixel smoothening, (iv) laser power variation, (v) laser spatial variation, (vi) laser absorption by OH, (vii) absorption coefficient, (viii) overlap integral, (ix) quench rate variation and (x) pixel area accuracy and accounted to total rms error of 21.4, 22.8, 22.5 and 22.9 % at 10, 27, 37 and 53 bar respectively. The rms error includes the contributions from: i. Camera calibration ii. Shot noise iii. Pixel smoothening iv. Laser power variation v. Laser spatial variation vi. Laser absorption by OH vii. Absorption coefficient viii. Overlap integral ix. Quench rate variation x. Pixel area accuracy 117

118 The uncertainty due to camera calibration of 2.9%, laser shot to shot power fluctuation of 11%, laser sheet spatial variation of 5.9 % and pixel area of 2.8 % remain the same for all the pressure cases. The uncertainties due to the laser shot-to-shot power fluctuation could be eliminated by monitoring the laser energy variation during the experiments. The uncertainty in laser sheet spatial variation in could be eliminated by monitoring the spatial profile during experiments from a separate test cell, uniformly filled with a fluorescing substance like acetone. The shot noise accounted to 6 7 % in all the pressure cases. The average number of photons collected in all the pressure cases was in the range. As the pressure increases the decrease in the OH signal strength is expected due to collisional quenching. But in the current study the increase in the pressure was achieved by increasing the propellant mass flow rate resulting in increased OH production at higher pressures. Thus as the pressure increased the strength of the OH signal collected primarily depended on the collisonal quenching and increased OH production. The uncertainty due to pixel smoothening used to minimize the contribution of camera sensor randomness was also 6 7% for all the pressure cases. The uncertainty due to the laser absorption by OH was estimated to increase from 2 to 4.7 % in the bar range. From the OH-PLIF images in Fig. 5 it could be recognized that the effect of laser absorption is negligible for all the pressure cases and as indicated by the uncertainty estimation too. Of all the uncertainties the variation of the absorption coefficient with temperature was the highest and was 12 to 15 % in the 10 to 53 bar range. The uncertainty due to this can be reduced, provided the 2D temperature field is available through measurements or calculations. For flames with wrinkling, corrugation and large fluctuations the use of temperature field data from 118

119 numerical simulation or a reference flame could lead to additional errors as the instantaneous temperature field of the actual flame and simulated/reference flame cannot be precisely matched. The uncertainty due to the variation of overlap integral due to line shape broadening decreased from 1.3 to 0.5 at bar pressure range. Similarly the uncertainty in the overlap integral due to the shift in the center line of the laser decreased from 2.8 to 0.2 % in the 10 to 53 bar range. The relatively low variations in the overlap integral is attributed to the use of large laser line width of 5cm -1 thereby obtaining a complete overlap between laser spectral and OH absorption profile at all pressures. Moreover the mean value of the overlap integral was reduced by only 10 % from 10 to 53 bar in the current study compared to the 30 40% reduction of overlap integral in other studies [50] due to the use of lasers with small line widths of the order of 0.5 1cm -1. The uncertainty due to variation in collisional quenching was nearly 4 % in all the pressure cases and is less significant compared to the absorption coefficient variation of 12 15% with temperature. The uncertainty contributions from spectroscopic constants and uncertainty in the mathematical model describing the fluorescence process, collisional/doppler widths and shifts, and quench rate are identified as negligible in this study. Thus the total rms uncertainty in the OH number density measurements for a GH 2 /GO 2 flame determined from a broad range of uncertainty sources accounted to of 21.4, 22.8, 22.5 and 22.9 % at 10, 27, 37 and 53 bar respectively. The improvements identified in the study includes elimination of uncertainties from laser shot to shot power fluctuation, laser spatial sheet variation, and minimizing the uncertainty due to temperature variation from simultaneous temperature measurements. The incorporation of the improvements suggested in the study could potentially reduce the uncertainty from the present 119

120 uncertainty of % to nearly 11 % for all the pressure cases thereby improving the quality of the data for CFD code validation. The boundary conditions included temperature measurements and evaluation of wall heat fluxes. These data are collected simultaneously with the OH concentration and are used in the computational studies that parallel this work. These results are included in Appendix E. 120

121 CHAPTER 6 CONCLUSIONS The purpose of this work was to provide a database of in-flow planar species concentration and to quantify the uncertainty associated with these measurements. The experimental conditions investigated used O/F mass flow ratio fixed at 3.77 and chamber pressures of 10, 27, 37 and 53 bar. Nine rovibrational lines at A-X(1,0) transition of OH excited at 283 nm was employed to obtain OH distribution in the shear reaction zone near the coaxial injector. The following are the conclusions: Benchmark inflow OH concentration data was generated in the same experimental facility with the same propellant system and instrumentation method for a range of pressures from bar. The instantaneous OH concentration and the averaged concentration in number densities which were inexistent in previous single injector studies over the bar range can be used to validate of LES and RANS CFD codes respectively. This is the first contribution of the current study. The wrinkling, corrugation and flapping of the flame at higher pressures of bar is due to the combined effects of turbulence due to increased Reynolds number and jet instability caused by size and dynamics of the recirculation region in the wake of oxidizer post lip. The quality of the benchmark inflow data was improved by a thorough and comprehensive uncertainty analysis and assessment, and this is the second contribution of the study. The systematic uncertainties, which remained the same irrespective of the experimental conditions, were evaluated at all pressures; uncertainty due to camera calibration, laser shot to shot power fluctuation, laser sheet spatial variation and pixel area accuracy was 2.9, 11, 5.9 and 2.3 % respectively. The uncertainties due to laser shot to shot power fluctuation and laser sheet spatial variation could be potentially minimized in future studies. The uncertainty due to shot noise and pixel smoothening were, each 6 7% for all the pressures cases. The laser elastic scattering was effectively blocked and contributions from the background flame emissions were eliminated for accurate quantitative measurements. The uncertainty due to absorption of laser across the flame by H 2 O was negligible and by OH was 2 5 % in bar range. 121

122 The uncertainty in absorption coefficient variation with temperature was % in bar range and was the maximum among all the uncertainties. The uncertainty could be potentially minimized provided there is availability of temperature field data from experiments/ computations. The uncertainty in overlap integral with temperature variation was % and % with laser centerline shift and the mean value of overlap integral was reduced by 10% in the bar range. The use of lasers with larger line widths is recommended for OH-PLIF measurements at high pressures for minimizing the uncertainty due to overlap integral. The uncertainty in collisional quench rate variation with temperature and colliding species mole fraction was nearly 4% at all pressures and is insignificant compared to the 12 15% variation of absorption coefficient with temperature. The uncertainty in the spectroscopic constants, mathematical model used to describe fluorescence process, collisional and Doppler widths, and collisional quenching are negligible. The total rms uncertainty contributions in OH number density analyzed and determined from 18 sources at 10, 27, 37 and 53 bar was 21.9, 22.8, 22.5 and 22.9 % respectively. The quality of the inflow data was improved from uncertainty assessment of two sources in previous studies to 18 sources; by quantifying 12, eliminating 2 and identifying as negligible the rest 4.The information is valuable for CFD validation as it brackets the reliability of the experimental data base. To reduce the uncertainty to nearly 11 % from the current 23%, potential areas for future improvements include elimination of the uncertainties due to laser power and spatial variation, and absorption coefficient variation with temperature. 122

123 CHAPTER 7 FUTURE WORK The future work should be directed towards improving the accuracy of the OH concentration in areas highlighted in the current study. The uncertainty in OH concentration measured today reaches nearly 23% for all the pressure cases. Some of the major uncertainties that were significant and that can be minimized include: i % from temperature dependence of absorption coefficient ii. 11% from laser shot to shot power fluctuations iii. 6 % from laser sheet spatial variation in intensity The uncertainty in the absorption coefficient of % can be minimized, if there is availability of the temperature field data either from simultaneous temperature measurements or from CFD simulations. This remains problematic for a number of reasons. Planar temperature measurements in a high pressure reacting flow has not been attempted today, given the difficulty to adapt the point wise absorption technique to the current flow field. CFD simulation either in time averaged or time accurate formulations have uncertainties that far exceed the 12 15% evaluated in this study. Hence while several CFD-experimental combined studies may improve this item the future work will require considerable effort. An experimental technique that may be attempted is a two-line OH thermometry. The advantage of using simultaneous temperature measurements is that both the OH and temperature field data can be spatially matched. The disadvantages results from the considerable complexity of the experimental setup and additional contributions to overall uncertainty from the temperature measurement errors. data; Thus the following procedure need to be adopted to revise both the experimental and CFD 123

124 Step 1: the OH number density data simulated from RANS simulations should be validated against the experimental data and the CFD code should be improved till it predicts the measured OH concentration within the current uncertainty limits. Step 2: the temperature field data from the improved CFD code can be used to refine the calculation of OH concentration from experimental measurements. The procedures in Step 1 and Step 2 should be followed iteratively till the uncertainty contribution due to unknown temperature field attains a minimum value. The second major uncertainty source is the laser shot to shot power fluctuation. To eliminate it, a fixed percentage of the total laser power could be monitored through out the experiments. Additional equipment would require a laser power meter than can be synchronized with OH-PLIF shot to shot images. Similarly the uncertainty from the averaged laser sheet spatial intensity could be reduced by monitoring the spatial profile throughout the experiments. This could be done by tracking the spatial intensity profile of the laser from a separate test cell, uniformly filled with a fluorescing substance like acetone and synchronized with the shot to shot OH-PLIF images. Additional experimental challenges include separate optical setup and detection electronics, and extraction of a part of total laser power at 283 nm for acetone fluorescence. It is estimated that incorporation of the improvements suggested here in future works would minimize the uncertainty in the OH concentration measurements from 23 % to nearly 11 % and this is the third contribution of the study. 124

125 APPENDIX A MATLAB SCRIPTS USED FOR DATA PROCESSING The Matlab scripts used for data processing are provided here: i. 3D heat flux processing -37 bar ii. Elimination of background emissions- 37 bar iii. Laser sheet spatial profile uncertainty iv. Conventional photon calibration v. Poisson photon calibration - photon count 300 ns vi. Poisson photon calibration - camera calibration vii. OH number density contours-37 bar viii. Mean reaction zone-37 bar 125

126 3D heat flux processing clear all; close all; warning off; d=0;c=0;dt=0.0001;k=1;i=0;ar=0;s=100; t=7.75;h=0.001;e=t/(dt*s); %number of grids g=51;gz=51; %centre point cp=(g-1)/2 +1 ; dx=63.5/(g-1); Ar=(dx)^2; L=1;L1=roundn(9.3/dx,0);L2=roundn(20.3/dx,0);L3=roundn(32.3/dx,0);L4=roundn(51.4/dx,0);L 5=g; %inner wall a=roundn(12.7/dx,0); %a=int32(12.7/dx); hi=roundn(0.8/dx,0); %hi=int32(0.8/dx); %ho=int16(2.4/dx); ho=3*hi; in=roundn(15.8/dx,0);m=roundn(22.2/dx,0);o=roundn(28.5/dx,0); %in=int16(15.8/dx);m=int16(22.2/dx);o=int16(28.5/dx); q=0.4*ones(g,g,g); qi=0.4*zeros(g,g,g); Temp=300*ones(g,g,m);Temp1=300*ones(g,g,m);Temp2=300*ones(g,g,m);Temp3=300*ones(g,g,m);Temp4=300*ones(g,g,m);Temp5=300*ones(g,g,m); T=300*ones(g,g,g); c=388/(8700*385)*(dx*1e-3)^-2; d=dx*1e-3*1*1e6/388; %qi=1.58; LT=300; P1=29; P2=30; %e=0.01/ %m1=1.92; n1=1; %m1(1:g)=(1.19e-6*(0:dx:63.5).^ *(0:dx:63.5).^ *(0:dx:63.5).^ *(0:dx:63.5) ); m1(1:l3)=( *(0:dx:(l3-1)*dx).^ *(0:dx:(l3-1)*dx) ); m1(l3:g)=( *((l3-1)*dx:dx:63.5).^ *((l3-1)*dx:dx:63.5) ); for i=dt:dt:t %qi= m1*i+ n1; for j=cp-a:cp+a qi(j,cp-a,1:g)=m1'*(1-exp(-i/n1)); 126

127 qi(j,cp+a,1:g)=m1*(1-exp(-i/n1)); qi(cp-a,j,1:g)=m1*(1-exp(-i/n1)); qi(cp+a,j,1:g)=m1*(1-exp(-i/n1)); end %T(2:g-1,2:g-1)=T(2:g-1,2:g-1) + %c*dt*(t(3:g,2:g-1)-2*t(2:g-1,2:g-1)+t(1:g-2,2:g-1)+ %T(2:g-1,3:g)-2*T(2:g-1,2:g-1)+T(2:g-1,1:g-2)) ; 2D unsteady T(2:g-1,2:g-1,2:g-1)=T(2:g-1,2:g-1,2:g-1) + c*dt*(t(3:g,2:g-1,2:g-1)-2*t(2:g-1,2:g-1,2:g- 1)+T(1:g-2,2:g-1,2:g-1)+ T(2:g-1,3:g,2:g-1)-2*T(2:g-1,2:g-1,2:g-1)+T(2:g-1,1:g-2,2:g-1) + T(2:g-1,2:g-1,3:g)-2*T(2:g-1,2:g-1,2:g-1)+T(2:g-1,2:g-1,1:g-2)) ; %bottom surface boundary condition T(2:g-1,2:g-1,1)=T(2:g-1,2:g-1,1) + c*dt*(t(3:g,2:g-1,1)-2*t(2:g-1,2:g-1,1)+t(1:g-2,2:g-1,1)+ T(2:g-1,3:g,1)-2*T(2:g-1,2:g-1,1)+T(2:g-1,1:g-2,1)); %Top surface boundary condition T(2:g-1,2:g-1,g)=T(2:g-1,2:g-1,g) + c*dt*(t(3:g,2:g-1,g)-2*t(2:g-1,2:g-1,g)+t(1:g-2,2:g-1,g)+ T(2:g-1,3:g,g)-2*T(2:g-1,2:g-1,g)+T(2:g-1,1:g-2,g)) ; %outer wall bc T(:,1,1:g) = T(:,2,1:g)- q(:,1,1:g)*dx*1e-3/388; T(:,g,1:g)= T(:,g-1,1:g)-q(:,g,1:g)*dx*1e-3/388; T(1,:,1:g)=T(2,:,1:g)-q(1,:,1:g)*dx*1e-3/388; T(g,:,1:g)=T(g-1,:,1:g)-q(g,:,1:g)*dx*1e-3/388; q(:,1,1:g)=388*(t(:,2,1:g)-t(:,1,1:g))/(dx*1e-3); q(:,g,1:g)=388*(t(:,g-1,1:g)-t(:,g,1:g))/(dx*1e-3); q(1,:,1:g)=388*(t(2,:,1:g)-t(1,:,1:g))/(dx*1e-3); q(g,:,1:g)=388*(t(g-1,:,1:g)-t(g,:,1:g))/(dx*1e-3); %inner wall bc T(cp-a:cp+a,cp-a,1:g)= T(cp-a:cp+a,cp-a-1,1:g)+d*qi(cp-a:cp+a,cp-a,1:g); T(cp-a:cp+a,cp+a,1:g)=T(cp-a:cp+a,cp+a+1,1:g) +d*qi(cp-a:cp+a,cp+a,1:g); T(cp-a,cp-a:cp+a,1:g)=T(cp-a-1,cp-a:cp+a,1:g)+ d*qi(cp-a,cp-a:cp+a,1:g); T(cp+a,cp-a:cp+a,1:g)=T(cp+a+1,cp-a:cp+a,1:g)+d*qi(cp+a,cp-a:cp+a,1:g); if roundn(i/dt,0)==k*s Temp(:,:,k)=T(:,:,L)-273; Temp1(:,:,k)=T(:,:,L1)-273; Temp2(:,:,k)=T(:,:,L2)-273; Temp3(:,:,k)=T(:,:,L3)-273; Temp4(:,:,k)=T(:,:,L4)-273; Temp5(:,:,k)=T(:,:,L5)-273; k=k+1; end; end; figure(1) plot(squeeze(temp(cp-hi,cp-in,1:end)),'r'); hold on plot(squeeze(temp(cp+hi,cp-m,1:end)),'b'); xlabel('time(ms)','fontsize',18); 127

128 ylabel('temperature(^oc)','fontsize',18); grid on; title('computaional Temperatures Inner & Middle 83 mm','fontsize',15); axis([1 e 20 LT]); set(gca,'fontsize',15) legend('inner','middle',2); %Temp=T-273; figure(2) [c,h]=contourf(temp(:,:,end)); colorbar; %line([cp-a cp-a cp+a cp+a cp-a],[cp-a cp+a cp+a cp-a cp-a],'color','w','linewidth',2); rectangle('position',[cp-a,cp-a,2*a,2*a],'facecolor','w') line([1 cp-in],[cp-hi cp-hi],'color',[1 1 1],'linewidth',2.5); line([1 cp-m],[cp+hi cp+hi],'color',[1 1 1],'linewidth',2.5); %line([1 cp-o],[cp+ho cp+ho],'color',[1 1 1],'linewidth',1.5); title('computational Temperatures 2D at t=7.75s and x=83 mm','fontsize',18); set(gca,'xtick',[ ]) set(gca,'ytick',[ ]) set(gca,'xticklabel',{'0.25';'0.5';'0.75';'1.00';'1.25';'1.50';'1.75';'2.00';'2.25';'2.50'}); set(gca,'yticklabel',{'0.25';'0.5';'0.75';'1.00';'1.25';'1.50';'1.75';'2.00';'2.25';'2.50'}); xlabel('length (inch)','fontsize',18); ylabel('breadth (inch)','fontsize',18); %input file [filename, pathname] = uigetfile('*.*', 'Select test data file.'); if isequal(filename,0) isequal(pathname,0) disp('user pressed cancel') %a = 0; else disp(['user selected ', fullfile(pathname, filename)]) data = load([pathname filename]); %a = 1; end %data put into different matrix b=231; c=457; t1=data(:,1); CT=data(:,2); cp=data(:,3); op=data(:,4); fp=data(:,5); omfr=data(:,6); fmfr=data(:,7); ofmr=data(:,8); ER=data(:,9); ITL=data(:,10); ITS=data(:,11); 128

129 BIT=data(:,12); WT1 = zeros(c-b+1,2); t2=(t1(b:c)-5250)/1000; %Data processing for heat flux %loop Len=[L L1 L2 L3 L4 L5];it=0; axial=[ ];%Tem=[Temp Temp1 Temp2 Temp3 Temp4 Temp5]; qcomp=zeros(1,6);qlinear=zeros(1,6);texpi=zeros(1,6);texpm=zeros(1,6);tcompi=zeros(1,6);t compm=zeros(1,6); for P1=23:2:33 P2=0; P2=P1+1; it=it+1; WT1 = zeros(c-b+1,2); t2=zeros(c-b+1,1); t2=(t1(b:c)-5250)/1000; WT1(:,1)=data(b:c,P1); WT1(:,2)=data(b:c,P2); T2=zeros(1,2);T1=zeros(1,2); T2=polyfit(t2,WT1(:,2),1);T1=polyfit(t2,WT1(:,1),1); WT(1:e,1:2)=0; t=dt:7.75/e:7.75; t=t'; WT(1:e,1)=polyval(T1,t); WT(1:e,2)=polyval(T2,t); x = [ ]; Texp=[WT(e,1) WT(e,2)]; T2a= T(cp+hi,cp-m,Len(it))-273;T1a=T(cp-hi,cp-in,Len(it))-273; Tcomp=[T1a T2a]; Texpi(it)=Texp(1);Texpm(it)=Texp(2); Tcompi(it)=Tcomp(1);Tcompm(it)=Tcomp(2); figure(it+2) plot(x,texp,'or',x,tcomp,'ob'); grid on xlabel('distance from inner wall (mm)','fontsize',18); ylabel('temperature (^oc)','fontsize',18); axis([ LT]); text(x(1),texp(1)+5,num2str(roundn(texp(1),0)),'fontsize',15,'color','r'); text(x(2),texp(2)-5,num2str(roundn(texp(2),0)),'fontsize',15,'color','r'); text(x(1),tcomp(1)-5,num2str(roundn(tcomp(1),0)),'fontsize',15,'color','b'); text(x(2),tcomp(2)+5,num2str(roundn(tcomp(2),0)),'fontsize',15,'color','b'); text(8,texp(1)+20,strcat('q=',num2str(qi(cp-a,cp-a,len(it))),'mw/m^2'),'fontsize',15,'color','k'); title(strcat('experimental and Computational Temperature Comparisons at ',num2str(axial(it)),' mm'),'fontsize',18); set(gca,'fontsize',15); legend('experiment','computation',2); saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(it+2),'.emf')); 129

130 saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(it+2),'.fig')); if it ==1 Te=Temp; end if it==2 Te=Temp1; end; if it ==3 Te=Temp2; end if it==4 Te=Temp3; end; if it ==5 Te=Temp4; end if it==6 Te=Temp5; end; compt(1:e,1:2)=0; compt(:,1)=squeeze(te(cp-hi,cp-in,1:end)); compt(:,2)=squeeze(te(cp+hi,cp-m,1:end)); figure(it+8) plot(t,compt(:,1),'r',t,wt(:,1),'--r',t,compt(:,2),'b',t,wt(:,2),'--b'); title(strcat('linear fit for temperatures at ',num2str(axial(it)),' mm'),'fontsize',18); xlabel('time(s)','fontsize',18); ylabel('temperature(^oc)','fontsize',18); %axis([ LT]); axis([ LT]); %axis tight; grid on; set(gca,'fontsize',15); legend('computation inner','experiment inner','computation Middle','Experiment Middle',2); saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(it+8),'.emf')); saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(it+8),'.fig')); slope88(1,1)= T1(1,1); slope88(1,2)=t2(1,1); slope88*1e3; qcomp(it)=qi(cp-a,cp-a,len(it)); %Linear Assumption HF881=zeros(length(WT(:,1)),1);HF88unsteady1=zeros(length(WT(:,1)),1); HF881=(388/ )*(WT(:,1)-WT(:,2))/ ; HF88unsteady1=HF e-6*(8700*385* *T2(1,1)); 130

131 figure(it+14) plot(t,[hf881 HF88unsteady1],'Linewidth',1); Title(strcat('Heat Flux Linear Assumption at ',num2str(axial(it)),' mm'),'fontsize',18); ylabel('heat Flux (MW/m^2)','Fontsize',18); xlabel('time(ms)','fontsize',18); set(gca,'fontsize',15); legend('hf881','hf88unsteady1',2); axis([min(t) max(t) 0 2]); grid on; saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(it+14),'.emf')); saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(it+14),'.fig')); p1=polyfit(t,hf88unsteady1,1) %disp(num2str(p1(1)),'*t',num2str(p1(2))) HF881(end); qlinear(it)= HF88unsteady1(end); figure(it+20) plot(t,wt(:,1),'--r',t,wt(:,2),'--b'); title(strcat('linear fit for temperatures at',num2str(axial(it)),' mm'),'fontsize',18); xlabel('time(s)','fontsize',18); ylabel('temperature(^oc)','fontsize',18); %axis([ LT]); axis([ LT]); grid on; set(gca,'fontsize',15); legend('experiment inner','experiment Middle',2); saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(it+20),'.emf')); saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(it+20),'.fig')); end; axial1=[ ]; qcomp2d=[ ]; axial=[ ]; figure(30) plot(axial,qcomp,'-dr',axial,qlinear,'-sb',axial,qcomp2d,'-*g'); grid on; xlabel('distance from Injector Face (mm)','fontsize',18); ylabel('heat Flux (MW/m^2)','FontSize',18); set(gca,'fontsize',15); axis([ ]); legend('heat Flux Computational 51x51x51 grid','heat Flux Linear','Heat Flux 2D',1); saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(30),'.emf')); 131

132 saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(30),'.fig')); figure(31) plot(axial,texpi,'-or',axial,tcompi,'--db'); grid on; xlabel('distance from Injector Face (mm)','fontsize',18); ylabel('temperature (^oc)','fontsize',18); set(gca,'fontsize',15); axis([ ]); legend('t_i_n_n_e_r Exp','T_i_n_n_e_r Comp',1); saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(31),'.emf')); saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(31),'.fig')); figure(32) plot(axial,texpm,'-or',axial,tcompm,'--db'); grid on; xlabel('distance from Injector Face (mm)','fontsize',18); ylabel('temperature (^oc)','fontsize',18); set(gca,'fontsize',15); axis([ ]); legend('t_m_i_d_d_l_e Exp','T_m_i_d_d_l_e Comp',1); saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(32),'.emf')); saveas(gcf,strcat('e:\aravind7\combustiontests\heatflux Processing\37bar\37bar\',num2str(32),'.fig')); 132

133 Elimination of background emissions clear all close all x= [];y=[];s=[];sum1=0;z=0;sumlaser=0;sumnolaser=0;avglaser=0;avgnolaser=0;lasersubback=0;in j=[];suminj=0;avginj=0;z1=0;z2=0;r=[];rb=[];sumr=0;sumrb=0;avgr=0;avgrb=0;rc=0;no =0;a=0;a1=0;a2=0; c=63; b=75; d=b-c+1; for i=c:b x{i}= imread(strcat('e:\aravind7\ohplif\afterproposal\091807ouf1ip3ca3sat\091807ouf1ip3c A3SAT06\35barlasertunedon283nm_00',num2str(i),'A','.tif')); x{i}=double(x{i}); sumlaser=sumlaser +x{i}; y{i}= imread(strcat('e:\aravind7\ohplif\afterproposal\091807ouf1ip3ca3sat\091807ouf1ip3c A3SAT06\35barlasertunedon283nm_00',num2str(i),'B','.tif')); y{i}=double(y{i}); sumnolaser=sumnolaser +y{i}; S{i}=x{i}-y{i}; sum1=sum1+s{i}; end %avg image avglaser=sumlaser/d; avgnolaser=sumnolaser/d; z=avglaser-avgnolaser; z(find(z<0))=0; %Reference Picture information gives 1mm = pixels %Set Injector location pixel. ILX = 25; ILY = 101; %Create X and Y axis from reference picture information. PS = 1/15.05; XL = 0-ILX; XH = 319-ILX; YL = 0-ILY; YH = 175-ILY; Y=PS*YL:PS:PS*(YH-1); X=PS*XL:PS:PS*(XH-1); cmap=(0:20)'/20*[1 1 1]; k=7;l=95;m=25; X1=X(25:319); Y2=Y(56:148); 133

134 figure(1) avglaser=avglaser(56:148,25:319); image(x1,y2,squeeze(avglaser/k)); set(gca,'fontsize',18) axis([-1.5 max(x1) -3 3]); cmap1=(0:l)'/l*[1 1 1]; colormap(cmap1); h=colorbar('horiz'); set(h,'fontsize',18); set(gca,'yaxislocation','right'); %rectangle rectangle('position',[-1.5,-0.6,1.5,1.2],'facecolor',[ ]); rectangle('position',[-1.5, ,1.5,0.2435],'facecolor',[ ]); rectangle('position',[-1.5,1.1,1.5,0.2435],'facecolor',[ ]); %text text(-1.5,0,'o_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); xlabel('height (mm)','fontsize',18); ylabel('width (mm)','fontsize',18); axis equal axis manual set(gcf,'paperposition',[ ]) ; set(gcf, 'color', 'white'); saveas(gcf,'e:\aravind7\ohplif\afterproposal\091807ouf1ip3ca3sat\avgohimages06\av goh35bar1','emf'); figure(2) avgnolaser=avgnolaser(56:148,25:319); image(x1,y2,squeeze(avgnolaser/k)); set(gca,'fontsize',18) cmap1=(0:l)'/l*[1 1 1]; colormap(cmap1); h=colorbar('horiz'); set(h,'fontsize',18); set(gca,'yaxislocation','right'); %rectangle rectangle('position',[-1.5,-0.6,1.5,1.2],'facecolor',[ ]); rectangle('position',[-1.5, ,1.5,0.2435],'facecolor',[ ]); rectangle('position',[-1.5,1.1,1.5,0.2435],'facecolor',[ ]); %text text(-1.5,0,'o_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); xlabel('height (mm)','fontsize',18); ylabel('width (mm)','fontsize',18); axis([-1.5 max(x1) min(y2) max(y2)]); 134

135 axis equal axis manual set(gcf,'paperposition',[ ]) ; set(gcf, 'color', 'white'); saveas(gcf,'e:\aravind7\ohplif\afterproposal\091807ouf1ip3ca3sat\avgohimages06\av goh35bar2','emf'); figure(3) z1=z(56:148,25:319); image(x1,y2,squeeze(z1/k)); set(gca,'fontsize',18) cmap1=(0:m)'/m*[1 1 1]; colormap(cmap1); h=colorbar('horiz'); set(h,'fontsize',18); set(gca,'yaxislocation','right'); %rectangle rectangle('position',[-1.5,-0.6,1.5,1.2],'facecolor',[ ]); rectangle('position',[-1.5, ,1.5,0.2435],'facecolor',[ ]); rectangle('position',[-1.5,1.1,1.5,0.2435],'facecolor',[ ]); %text text(-1.5,0,'o_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); xlabel('height (mm)','fontsize',18); ylabel('width (mm)','fontsize',18); axis([-1.5 max(x1) -3 3]); axis equal axis manual set(gcf,'paperposition',[ ]) ; set(gcf, 'color', 'white'); saveas(gcf,'e:\aravind7\ohplif\afterproposal\091807ouf1ip3ca3sat\avgohimages06\av goh35bar3','emf'); 135

136 Laser sheet spatial profile uncertainty Ulaserfluc=0; for m=140:232 for n=25:319 laser= [];lasersum=0;laseravg=0;laseravgnorm=0;fluc=0; c1=10;b1=99; d=b1-c1; for o=10:99 laser=0; laser= imread(strcat('e:\aravind7\ohplif\afterproposal\092407laserprofile\laserprofile1acetone283n m\laserprofile1acetone283nm_00',num2str(o),'.tif')); laser=double(laser); fluc(o)=laser(m,n)/max(max(laser(m,:))); % lasersum=lasersum +laser; end fluc=fluc(10:99); fluc=reshape(fluc,1,prod(size(fluc))); Ulaserfluc(m-139,n-24)=100*std(fluc)/mean(fluc); end; end; mean(mean(ulaserfluc)) %Uncertainty in laser fluctuation is

137 Conventional phtoton calibration close all clear all Count=[ ]; Ex = [ ]; %2) 310 nm filter with FWHM 10 nm xi=0;yi=0;lamda=0;trans=0; lamda=[ ]; Trans=[ ]; lntrans =log(trans); xi=279:1:334; yi=exp(interp1(lamda,lntrans,xi,'spline')); figure(1) plot(lamda,trans,'*k', xi,yi,'--k'); grid on set(gca, 'Fontsize',18); xlabel('\lambda(nm)'); ylabel('transmission(%)'); axis([ ]); legend('data Transmission','Linear fit',2); saveas(gcf,'filtertransmission310nm','emf'); %3) Lamp Irradiance pixelarea = *1e-12; %in m^2 h=6.626*1e-34; %Js f=9.67*1e14; %frequency(s^-1) Irrad=0; %(mw/m^2 nm) Energy=0;Np=0; Cou=zeros(1,20);Np1=zeros(1,20);Calib=0; A= *1e1; B= *1e3; C= *1e-1; D= ; E= *1e5; F= *1e8; G= *1e10; H= *1e12; Irrad = (xi).^-5.* exp(a+b*(xi).^-1).*(c +D*(xi.^-1) + E*(xi.^-2) + F*(xi.^-3) + G*(xi.^-4) + H*(xi.^-5)); %Irrad = (lamda).^-5.* exp(a+b*(lamda).^-1).*(c +D*(lamda.^-1) + E*(lamda.^-2) + F*(lamda.^-3) + G*(lamda.^-4) + H*(lamda.^-5)); Energy = sum(irrad*1e-3.*yi*0.01)*ex*1e-9 * pixelarea/0.55 ; Np= Energy/(h*f); Np1=Np; Cou=Count; %Cou(1)=0; %Np1(2:20)=Np(1:19); %Np(1)=0; Cou1=0:20:2900; Calib=interp1(Cou,Np1,Cou1,'spline'); %Calibconst = mean(np(2:9)./cou(2:9)) 137

138 Calibconst = sum((np(2:9)./cou(2:9)).*cou(2:9))/sum(cou(2:9)); figure(2) plot(cou,np1,'*k',cou1,calib,'--k') grid on set(gca, 'Fontsize',18); xlabel('counts'); ylabel('number of Photons(N_p)'); legend('data Photons','Linear fit',2); saveas(gcf,'photoncalibration310nmfull','emf'); figure(3) plot(cou(1:9),np1(1:9),'*k') grid on set(gca, 'Fontsize',18); xlabel('counts'); ylabel('number of Photons(N_p)'); legend('data Photons',2); saveas(gcf,'photoncalibration310nm','emf'); figure(4) plot(cou(1:9),np1(1:9),'*k',cou(1:9), Calibconst*Cou(1:9),'--k'); grid on set(gca, 'Fontsize',18); xlabel('counts'); ylabel('number of Photons(N_p)'); text(100,650,['n_p = ', num2str(calibconst,3), ' *Counts '],'FontSize',18) legend('data Photons','Linear fit',2); saveas(gcf,'photoncalibration310nmeqn','emf'); Stdesti=0; errslope=0; yesti=(calibconst*cou(2:9)); Stdesti= (sum((np(2:9)-yesti).^2)/(length(yesti)-2))^0.5 errslope = sqrt(sum((cou(2:9)- mean(cou(2:9))).^2)^-1)*stdesti % no 95% confidence interval UPhotonCalib = (0.0286/1.59)*

139 Poisson photon calibration photon count 300ns clear all close all mint=[]; x= [];y=[];z=0;pc=[];sum1=0;sumlasersq=0;sumlaser=0;sumnolaser=0;avglaser=0;avgnolaser=0;la sersubback=0;inj=[];suminj=0;avginj=0;z1=0;z2=0;r=[];rb=[];sumr=0;sumrb=0;avgr=0;avg RB=0;RC=0;No=0;a=0;a1=0;a2=0; c=100; b=999; d=b-c+1; x=zeros(88,320,b-c+1); for i=c:b %i=77;j=43;k=35; x=[];y=[]; x= double(imread(strcat('e:\aravind7\ohplif\afterproposal\100207photoncalibration\300ns\300n s_0',num2str(i),'a','.tif'))); y= double(imread(strcat('e:\aravind7\ohplif\afterproposal\100207photoncalibration\b300\b300 _0',num2str(i),'A','.tif'))); z=x-y; z(find(z<=0))=0; sumlaser=sumlaser +z; sumlasersq=sumlasersq + z.^2; sumnolaser=sumnolaser+y; poiss(i-99)=z(15,15); mint=[mint; mean2(z)]; end; %PC=x-y; %avg image avg=0; V=0;S=0;KI=0; avg=sumlaser/d; V=(sumlasersq-avg.^2*d)/(d-1); S=V./avg; KI=avg./V; %S2=S(42:50,63:216); S2=KI; avg2=avg; %avg2=avg(42:50,63:216); S3=reshape(S2,prod(size(S2)),1); avg3=reshape(avg2,prod(size(s2)),1); figure(1) set(gca,'fontsize',15); hist(s3,0:0.01:0.5) xlabel('calibration Constant'); 139

140 ylabel('n'); text(0.35,240,['mean = ' num2str(mean(s3))],'fontsize',12); text(0.35,190,['\sigma = ' num2str(std(s3))],'fontsize',12); title('exposure time = 300 ns'); grid on saveas(gcf,'300ns','emf'); %mean(s3) %std(s3) figure(2) hist(avg3) mean(avg3) std(avg3) avgnolaser=sumnolaser/d; avgnolaser=reshape(avgnolaser,prod(size(s2)),1); mean(avgnolaser) std(avgnolaser) goodp=find(abs(poiss-mean(poiss)<(mean(poiss)+std(poiss)))); k=mean(poiss(goodp))/var(poiss(goodp)); poiss1=poiss*k; figure(3) set(gca,'fontsize',15); title('exposure time = 300 ns'); hist(poiss1,(min(poiss1(goodp)): 1: max(poiss1(goodp)))) text(191,31,['mean = ' num2str(mean(poiss1(goodp)))],'fontsize',15); text(191,21,['\sigma^2 = ' num2str(var(poiss1(goodp)))],'fontsize',15); xlabel('photons'); ylabel('n'); grid on hold on %figure(4) set(gca,'fontsize',15); %x1=min(poiss1(goodp)):1:max(poiss1(goodp)); %x1=x1+0.05; x1=134:1:233; y2=poisspdf(x1,175); y1=length(poiss1(goodp))*y2; plot(x1,y1,'+-r') xlabel('photons'); ylabel('n'); grid on legend('poisson Fit','Data'); %saveas(gcf,'poiss300a','emf'); %axis([ ]) mean(poiss); std(poiss); k 140

141 Poisson photon calibration camera calibration %clear all close all %Counts Counts=[ ]; StdCount=[ ]; Counts1=Counts+StdCount; Counts2=Counts-StdCount; Calconst= [ ]; %Calibration Constant StdCal=[ ]; Calconst1=Calconst+StdCal; Calconst2=Calconst-StdCal; %Calconst= [ ]*(0.7/0.55)*(0.5/0.12); Calconstavg= (sum(counts.*calconst))/sum(counts); Photon=Counts.*Calconst; Photon1=Counts.*mean(Calconstavg); %Photon1=Counts.*Calconst1; Photon2=Counts.*Calconst2; Ph=Photon*(.7/.55)*(.5/.12); Ph1=Photon1*(.7/.55)*(.5/.12); Ph2=Photon2*(.7/.55)*(.5/.12); figure(1) set(gca, 'fontsize', 15); plot(counts, Photon,'o'); xlabel('counts'); ylabel('photons') title('camera Calibration at 532 nm') grid on figure(2) set(gca, 'fontsize', 15); plot(counts, Photon*(.7/.55)*(.5/.12),'o'); xlabel('counts'); ylabel('photons') title('camera Calibration at 310 nm') grid on figure(3) set(gca, 'fontsize', 15); p = polyfit(counts,photon1*(.7/.55)*(.5/.12),1) ; %plot(counts,ph,'ob',counts,ph1,'o--k',counts,ph2,'og', Counts, polyval(p,counts),'r'); plot(counts,ph,'ob',counts, polyval(p,counts),'r'); xlabel('counts'); ylabel('photons') %title('camera Calibration at 310 nm') 141

142 text(300,900,['y = ', num2str(p(1)), ' *x '],'FontSize',15) legend('data Photons','Linear fit'); axis([ ]); grid on %saveas(gcf,'photoncalibration310nm','emf'); f=polyval(p,counts); mean(abs(f-ph)./ph) mean(ph) exposure=[ ]; p1 = polyfit(exposure, Counts,1) ; figure(4) set(gca, 'fontsize', 15); plot(exposure, Counts,'o',exposure,Counts1,'ok', exposure,counts2,'og', exposure, polyval(p1,exposure),'r'); xlabel('exposure(ns)'); ylabel('counts') %title('counts vs exposure at 532 nm') %text(51,1400,['y = ', num2str(p1(1)), ' *x + ', num2str(p1(2))],'fontsize',15); legend('mean Count','Mean Count + Std','Mean Count - Std', 'Linear fit'); axis([ ]); grid on %saveas(gcf,'cameracalibration','emf'); %error in estimate yesti=(calconstavg*counts*(.7/.55)*(.5/.12)); Stdesti= (sum((ph-yesti).^2)/(length(ph)-2))^0.5 %uncertainty in slope errslope=0; % errslope = sqrt(sum((counts- mean(counts)).^2)^-1)*stdesti*2.447; errslope = sqrt(sum((counts- mean(counts)).^2)^-1)*stdesti % no 95% confidence interval %uncertainty in intercept errintercept=0; %errintercept = 2.447*Stdesti*sqrt( mean(counts)^2/sum((counts- mean(counts)).^2)); errintercept = Stdesti*sqrt( mean(counts)^2/sum((counts- mean(counts)).^2));%no 95% confidence interval %uncertainty in photons errphotons = errslope*counts+ errintercept; 142

143 OH number density contours-37 bar clear all close all T=0;phi=0;fb=zeros(length(phi),9);Qo=0;FYield=zeros(1,length(phi)); phi=[ ]; RH2=2*phi;%mole of hydrogen RO2=ones(1,length(phi));%mole of Oxygen PH2Omf=[ ];% product mole fraction of H2O computed from Stanjan PH2mf=[ ];%product mole fraction of H2 compued from Stanjan PO2mf=[ e e-6 ];%product mole fraction of O2 computed from Stanjan PHmf=[ e-2 ];%product mole fraction of H computed from Stanjan POmf=[ e e-5 ];%product mole fraction of O computed from Stanjan POHmf=[ ];%product mole fraction of OH computed from Stanjan T=[ ];%Temperature corresponding to equivalnec ratio %fb(1,:)=[ ];% Boltzmann factor associated with excitation lines fb(1,:)=[ ]; fb(2,:)=[ ]; fb(3,:)=[ ]; fb(4,:)=[ ]; fb(5,:)=[ ]; fb(6,:)=[ ]; %fb(8,:)=[ ]; %fb(9,:)=[ ]; %Absorption coefficient of the individual lines (cmj^-1) B12=[ ]*1e12/3e10; %Lines [P21(6)_6.5 Q2(3)_2.5 R12(3)_2.5 Q1(6)_6.5 Q2(1)_0.5 R12(1)_0.5 %Q2(2)_1.5 R12(2)_1.5 R2(14)_13.5]; %Lines[ % ] % Pressure and collisional cross section of H2O, H2 and O2; P=36.1; sigmah2o=22; sigmah2=5; sigmao2=10; %quenching Qo= 1.229e5*P*1e5*((PH2Omf*sigmaH2O/2.96) + (PH2mf*sigmaH2/1.337)+(PO2mf*sigmaO2/3.33))./(T.^0.5); %FYield FYield=1.08e6./Qo; 143

144 %Absorption for i=1:6 fbb12(i,:)= fb(i,:).*b12; sumfbb12(1,i)=sum(fbb12(i,:)); end; figure(1) set(gca,'fontsize',15) ; plot(phi,qo,'--k',phi,mean(qo)*ones(1,length(phi)),'k') legend('collisional Quench rate 37 bar','mean Collisional Quench rate 37 bar'); xlabel('equivalence ratio (\phi)','fontsize',20); ylabel('collisional Quench Rate(s^-^1)','fontsize',20); ax=axis; axis([0 5 ax(3:4)]); ax=0; grid on saveas(gcf,'e:\aravind7\ohplif\afterproposal\36bar\absolutedensity36bar\quenchratevariatio n','emf'); figure(2) set(gca,'fontsize',15) ; plot(phi,fyield,'--k',phi,mean(fyield)*ones(1,length(phi)),'k') legend('fluorescence Yield 37 bar','mean Fluorescence Yield 37 bar'); xlabel('equivalence ratio (\phi)','fontsize',20); ylabel('fluorescence Yield','fontsize',20); ax=axis; axis([0 5 ax(3:4)]); ax=0; grid on saveas(gcf,'e:\aravind7\ohplif\afterproposal\36bar\absolutedensity36bar\fluoryieldvariation','emf'); figure(3) set(gca,'fontsize',15) ; plot(phi,sumfbb12,'--k',phi,mean(sumfbb12)*ones(1,length(phi)),'k') legend('absorption Coefficient 37 bar','mean Absorption Coefficient 37 bar'); xlabel('equivalence ratio (\phi)','fontsize',20); ylabel('absorption Coefficient (cmj^-^1)','fontsize',20); ax=axis; axis([0 5 ax(3:4)]); grid on saveas(gcf,'e:\aravind7\ohplif\afterproposal\36bar\absolutedensity36bar\absorcoeffvariatio n','emf'); j=3 L=0;x1=0;x=0;y=0;y1=0;yd=0; Ove=0; %k=[ ]; for i=1:6 L= dlmread(strcat('lif',num2str(t(i)),'k',num2str(i+1),'.mod'),','); x=l(:,1); 144

145 y1= L(:,2)/max(L(:,2)); %y=0.188*exp(-(x ).^2/9.016); y=0.188*exp(-(x ).^2/9.016); y2=y/max(y); figure(i+j) set(gca,'fontsize',15) ; plot(x,y2,'--k',x,y1,'k'); legend('laser Profile',strcat('OH Absorption Profile at T = ',num2str(t(i)),' and 37 bar')); xlabel('\nu (cm^-^1)','fontsize',20); ylabel('(a.u)','fontsize',20); axis([ ]); grid on saveas(gcf,strcat('e:\aravind7\ohplif\afterproposal\36bar\absolutedensity36bar\profilephi',nu m2str(phi(i)),'.emf')); dv=0; dx(1:79)=x(2:80)-x(1:79); %figure(3+i) %plot(dx) mean(dx); %data analysis/overlap ylaser=y/sum(y); yabs=l(:,2)/sum(l(:,2)); yabs1=yabs/mean(dx); Overlap=sum(y.*yabs1*mean(dx)); Ove(i)=Overlap; end figure(i+j+1) set(gca,'fontsize',15) ; plot(phi,ove,'--k',phi,mean(ove)*ones(1,length(phi)),'k'); legend('overlap Integral 37 bar','mean Overlap Integral 37 bar'); xlabel('\phi','fontsize',20); ylabel('overlap(cm)','fontsize',20); ax=axis; axis([0 5 ax(3:4)]); grid on saveas(gcf,'e:\aravind7\ohplif\afterproposal\36bar\absolutedensity36bar\overlapvariation','e mf'); Tempfactor= (sumfbb12.*ove.*fyield).^-1; figure(i+j+2) set(gca,'fontsize',15) ; plot(phi,tempfactor,'--k',phi,mean(tempfactor)*ones(1,length(phi)),'k'); legend('temperature Dependent Factors','Mean'); xlabel('equivalence ratio (\phi)','fontsize',20); ylabel('tempfactor(cm^2j^-^1)','fontsize',20); ax=axis; axis([0 5 ax(3:4)]); 145

146 grid on mftempfactor=tempfactor.*t; figure(i+j+3) set(gca,'fontsize',15) ; plot(phi,mftempfactor,'--k',phi,mean(mftempfactor)*ones(1,length(phi)),'k'); legend('temperature Dependent Factor mole fraction','mean'); xlabel('equivalence ratio (\phi)','fontsize',20); ylabel('tempfactor(cm^2j^-^1k)','fontsize',20); ax=axis; axis([0 5 ax(3:4)]); grid on NormTempfactor=Tempfactor/mean(Tempfactor); Variation=std(NormTempfactor); Percentagevariation=100*(Variation/mean(NormTempfactor)) figure(i+j+4) set(gca,'fontsize',15) ; plot(phi,normtempfactor,'--k',phi,mean(normtempfactor)*ones(1,length(phi)),'k'); legend('normalized Tempfactor','Mean'); xlabel('\phi','fontsize',20); ylabel('normtempfactor','fontsize',20); ax=axis; axis([0 5 ax(3:4)]); grid on NormmfTempfactor=mfTempfactor/mean(mfTempfactor); Variationmf=std(NormmfTempfactor); Percentagevariationmf=100*(Variationmf/mean(NormmfTempfactor)) figure(i+j+5) set(gca,'fontsize',15) ; plot(phi,normmftempfactor,'--k',phi,mean(normmftempfactor)*ones(1,length(phi)),'k'); legend('normalized Tempfactor','Mean'); xlabel('\phi','fontsize',20); ylabel('normmftempfactor','fontsize',20); ax=axis; axis([0 5 ax(3:4)]); grid on figure(i+j+6) set(gca,'fontsize',18) ; [AX,H1,H2] = plotyy(phi,[ph2omf' PH2mf' PO2mf'],phi,T,'plot'); set(get(ax(1),'ylabel'),'string','mole fraction') set(get(ax(2),'ylabel'),'string','temperature(^oc)') %plotyy(phi,[ph2omf PH2mf PO2mf],phi,T); xlabel('equivalence ratio (\phi)','fontsize',20); set(h1(1),'linestyle','-','color','b') set(h1(2),'linestyle','-','color','g') set(h1(3),'linestyle','-','color','k') %set(h1(4),'linestyle','-','color','c') 146

147 set(h2,'linestyle','-','color','r') legend(' Mole fraction H_2O 37 bar', ' Mole fraction H_2 37 bar',' Mole fraction O_2 37 bar','temperature 37 bar'); grid on Factor=0; Factormf=0; Bk= e-23;%Boltzmann constant E=0.89*1e-3% energy of laser J/pulse; A=2.058%cm^2 V=3.46e-5 %cm^3 eta=0.12*11.5; % (Photon detection efficiency * Factor associated with Gain) epsilon=0.55; saf=5.4e-4% Solid angle fraction; MTempfactor=mean(Tempfactor) MTempfactormf=mean(mfTempfactor) Factor= MTempfactor*((E/A)*V*saf)^-1; %botlzmann fraction, overlap integral and quenching %Factor= MTempfactor Factormf= ((P*1e5/Bk)*1e-6)^-1*MTempfactormf*((E/A)*eta*epsilon*V*saf)^-1; AvgTemp=MTempfactormf/MTempfactor %Image Processing % 1) Laser Sheet Profile Variation and Subsequent Correction laser= [];lasersum=0;laseravg=0;laseravgnorm=0; c1=10;b1=99; d=b1-c1; %centre Position ILX1 = 25; ILY1=186; %Create X and Y axis from reference picture information. PS = 1/15.05; XL1 = 0-ILX1; XH1 = 319-ILX1; YL1 = 0-ILY1; YH1 = 256-ILY1; Y1a=PS*YL1:PS:PS*(YH1-1); X1a=PS*XL1:PS:PS*(XH1-1); for o=10:99 laser=0; laser= imread(strcat('e:\aravind7\ohplif\afterproposal\092407laserprofile\laserprofile1acetone283n m\laserprofile1acetone283nm_00',num2str(o),'.tif')); laser=double(laser); lasersum=lasersum +laser; end laseravg=(lasersum/90)-104; %k=2;d=b1-c1; h = ones(5,5) / 25; 147

148 laseravg = imfilter(laseravg,h,'conv'); %h = fspecial('gaussian',[5 5]); %laseravg = imfilter(laseravg,h); %z2=z; figure(i+j+7) laseravg=laseravg(140:232,25:319); laseravgnorm=laseravg/mean(mean(laseravg)); for i=1:93 laseravgnorm(i,1:295)=laseravg(i,1:295)/max(max(laseravg(i,1:295))); end X11=X1a(25:319); Y21=Y1a(140:232); %image(x11,y21,squeeze(45*laseravgnorm)); image(x11,y21,squeeze(100*laseravgnorm)); %[c,h]=contourf(x11,y21,laseravgnorm,[ ]); % l=1; cmap1=(0:100)'/100*[1 1 1]; colormap(cmap1); %colormap(jet); h=colorbar('horiz'); set(h,'fontsize',18); %colorbar; %clabel(c,'manual'); hold on; rectangle('position',[-1.5,-0.6,1.5,1.2],'facecolor',[ ]); rectangle('position',[-1.5, ,1.5,0.2435],'facecolor',[ ]); rectangle('position',[-1.5,1.1,1.5,0.2435],'facecolor',[ ]); %text text(-1.5,0,'o_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); set(gca,'fontsize',18) ; set(gca,'yaxislocation','right'); xlabel('height (mm)','fontsize',18); ylabel('width (mm)','fontsize',18); %axis square axis equal axis manual axis([-1.5 max(x11) -3 3]); grid on set(gcf,'paperposition',[ ]) ; grid on set(gcf, 'color', 'white'); saveas(gcf,'e:\aravind7\ohplif\afterproposal\36bar\absolutedensity36bar\lasersheetvariatio n','tif'); % 2) OH-PLIF image Processing 148

149 x= [];y=[];s=[];sum1=0;z=0;sumlaser=0;sumnolaser=0;avglaser=0;avgnolaser=0;lasersubback=0;in j=[];suminj=0;avginj=0;z1=0;z2=0;r=[];rb=[];sumr=0;sumrb=0;avgr=0;avgrb=0;rc=0;no =0;a=0;a1=0;a2=0;a3=0;a4=0; c1=63;b1=75; d=b1-c1+1; c2=100;b2=100; %centre Position ILX = 25; ILY=101; %Create X and Y axis from reference picture information. PS = 1/15.05; XL = 0-ILX; XH = 319-ILX; YL = 0-ILY; YH = 175-ILY; Y=PS*YL:PS:PS*(YH-1); X=PS*XL:PS:PS*(XH-1); for i=c1:b1 x=0;y=0;z=0;z2=0;sumr=0;sumrb=0;avgr=0;avgrb=0;rc=0;no=0;a=0;a1=0;a2=0;% OUF1IP3CA3SAT07% x= imread(strcat('e:\aravind7\ohplif\afterproposal\091807ouf1ip3ca3sat\091807ouf1ip3c A3SAT06\35barlasertunedon283nm_00',num2str(i),'A','.tif')); x=double(x); sumlaser=sumlaser +x; y= imread(strcat('e:\aravind7\ohplif\afterproposal\091807ouf1ip3ca3sat\091807ouf1ip3c A3SAT06\35barlasertunedon283nm_00',num2str(i),'B','.tif')); y=double(y); sumnolaser=sumnolaser +y; %S=x{i}-y{i}; %sum1=sum1+s{i}; z=x-y; z(find(z<0))=0; figure(i) h = ones(5,5) / 25; z2 = imfilter(z,h,'conv'); %h = fspecial('gaussian'); %z2 = imfilter(z,h); z2=z2(56:148,25:319); z2=z2./laseravgnorm; %laser sheet profile variation corrected, spatial variation in intensifier is also corrected here z3=z2; a3=(1.59*z2)*factor*1e-15; X1=X(25:319); 149

150 Y2=Y(56:148); [c,h]=contour(x1,y2,a3,[ ]); axis([-1.5 max(x1) min(y2) max(y2)]); %colormap(gray); colormap(jet); h=colorbar('horiz'); set(h,'fontsize',18); %colorbar; %clabel(c,'manual'); hold on; rectangle('position',[-1.5,-0.6,1.5,1.2],'facecolor',[ ]); rectangle('position',[-1.5, ,1.5,0.2435],'facecolor',[ ]); rectangle('position',[-1.5,1.1,1.5,0.2435],'facecolor',[ ]); %text text(-1.5,0,'o_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); set(gca,'fontsize',18) ; set(gca,'yaxislocation','right'); xlabel('height (mm)','fontsize',18); ylabel('width (mm)','fontsize',18); title('number Density of OH (10^1^5 molecules/cm^3)'); axis equal axis manual %axes('position',[-1.5,-4, ,8]) %axis tight %colormap gray set(gcf,'paperposition',[ ]) ; grid on set(gcf, 'color', 'white'); %set(gcf,'position',[ ]); %M(i)=getframe(gcf); %saveas(gcf,'molefractionoh7bar1','tif'); %saveas(gcf,'molefractionoh7bar1','fig'); %figure(i+1) saveas(gcf,strcat('e:\aravind7\ohplif\afterproposal\36bar\absolutedensity36bar\instoh35bar',num2str(i),'.tif')); end z=0;avglaser=0;avgnolaser=0;a4=0; avglaser=sumlaser/d; avgnolaser=sumnolaser/d; z=avglaser-avgnolaser; %z=sum1/d; z(find(z<0))=0; figure(i+1) h = ones(5,5) / 25; 150

151 z2 = imfilter(z,h,'conv'); %h = fspecial('gaussian',[0 0]); %z2 = imfilter(z,h); z2=z2(56:148,25:319); z2=z2./laseravgnorm;%laser sheet profile variation corrected, spatial variation in intensifier is also corrected here %z2=z2./laseravgwnorm; %a4=z2*factormf; a4=(1.59*z2)*factor*1e-15; %a4=(z2)*factor*1e-15; X1=X(25:319); Y2=Y(56:148); %Y1=Y(34:143); [c,h]=contour(x1,y2,a4,[ ]); %[c,h]=contour(x1,y2,a4,[ ]); axis([-1.5 max(x1) min(y2) max(y2)]); %colormap(gray); colormap(jet); h=colorbar('horiz'); set(h,'fontsize',18); %colorbar %clabel(c,'manual'); hold on; rectangle('position',[-1.5,-0.6,1.5,1.2],'facecolor',[ ]); rectangle('position',[-1.5, ,1.5,0.2435],'facecolor',[ ]); rectangle('position',[-1.5,1.1,1.5,0.2435],'facecolor',[ ]); %text text(-1.5,0,'o_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); set(gca,'fontsize',18) ; set(gca,'yaxislocation','right'); xlabel('height (mm)','fontsize',18); ylabel('width (mm)','fontsize',18); title('number Density of OH (10^1^5 molecules/cm^3)'); axis equal axis manual set(gcf,'paperposition',[ ]) ; grid on set(gcf, 'color', 'white'); saveas(gcf,'e:\aravind7\ohplif\afterproposal\36bar\absolutedensity36bar\avgoh35bar','tif'); %OH absorption a4re=0; a4re=reshape(a4,1,prod(size(a4))); levels=min(roundn(a4re,0)):1:max(roundn(a4re,0)); N=hist(a4re, levels); 151

152 avgdensity = sum(levels(20:length(n)).* N(20:length(N)))/sum(N(20:length(N))) figure(199) plot(levels,n,'o') grid on %(I/Io)=exp(-h*nu* B12*N*dy) OHabsorppercent = exp(-6.636e- 34*1.06e15*mean(sumfbB12)*mean(Ove)*avgdensity*1e15*0.1) % shotnoise a4r=z3; %a4r=x-y; %a4r=a4r(56:148,25:319); a4r(find(a4r<0))=0; a4r=a4r./laseravgnorm; %a4r=a4r./laseravgwnorm; a4r=(1.59*a4r ); a4r=reshape(a4r,1,prod(size(a4r))); levels1=min(roundn(a4r,0)):1:max(roundn(a4r,0)); N1=hist(a4r, levels1); avgphoton=sum(levels1(50:length(n1)).* N1(50:length(N1)))/sum(N1(50:length(N1))) figure(200) plot(levels1,n1,'o'); grid on %Uncertanities %shot noise Ushotnoise = (sqrt(avgphoton)/avgphoton)*100 %Photon Calibration UPhotonCalib = 2.9 %(0.0286/1.59)*100=1.8, irradiance= 2.3 %Shot to shot power %fluctuation(e:\aravind7\ohplif\afterproposal\092407laserprofile\laserenergy) UPowerfluc = (0.10/0.89)*100 %Laser absorption Ulaserabs=(1-OHabsorppercent)*100 %Absorption Coefficient Uabsorp= (std(sumfbb12)/mean(sumfbb12))*100 %Overalp Uoverlap = (std(ove)/mean(ove))*100 %Ovelap line shift Uoverlapshift =100*(mean(Ove) )/mean(Ove) %Collisonalquench rate UQo= (std(qo)/mean(qo))*100 %Pixel Area Upixarea= 2.8 UFilter=6.3 %Laser spatial homogenity all the points = ULaSpatial=

153 UTotalrms = sqrt(ushotnoise^2 + UPhotonCalib^2+ UPowerfluc^2 + Ulaserabs^2 + Uabsorp^2 + Uoverlap^2 + Uoverlapshift^2 + UQo^2 +Upixarea^2 +UFilter^2 +ULaSpatial^2) Mean position of reaction zone 37 bar for r=1:295 for c=1:47 if a3(c,r)== max(a3(1:47,r)) width(r)= Y2(c); end; end; end; for r=1:295 for c=47:93 if a3(c,r)== max(a3(47:93,r)) width1(r)= Y2(c); end; end; end; r=1:300; figure(12) plot(x1(12:295),width(12:295),'--k',x1(12:295),width1(12:295),'k'); hold on; rectangle('position',[-1.5,-0.6,1.5,1.2],'facecolor',[ ]); rectangle('position',[-1.5, ,1.5,0.2435],'facecolor',[ ]); rectangle('position',[-1.5,1.1,1.5,0.2435],'facecolor',[ ]); %text text(-1.5,0,'o_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); text(-1.5, ,'h_2\rightarrow','horizontalalignment','right','fontsize',18); set(gca,'fontsize',18) ; set(gca,'yaxislocation','right'); xlabel('height (mm)','fontsize',18); ylabel('width (mm)','fontsize',18); legend('mean reaction zone lower', 'Mean reaction zone upper'); axis([-1.5 max(x1) min(y2) max(y2)]); axis equal axis manual set(gcf,'paperposition',[ ]) ; grid on set(gcf, 'color', 'white'); saveas(gcf,'e:\aravind7\ohplif\afterproposal\091807ouf1ip3ca3sat\instoh35barmf06\l OS37bar','tif'); Equation Section 2 153

154 APPENDIX B PROPOSED NEW METHODOLOGY FOR PHOTON CALIBRATION As explained in the OH-PLIF diagnostics in Chapter 4, the photons from fluorescing OH are captured by ICCD camera which has a photon detection efficiency of 12 % at 310 nm. The ICCD camera provides the detected photons in counts which is an arbitrary unit. The arrival of photons on an average is Poisson distributed when a light source emits photons at a constant rate. In this case a 10 W dc Tungsten-halogen lamp was chosen for calibration. The calibration setup is shown in Figure B-1. A filter with transmission efficiency of 70% at 532 and FWHM of 10+2 nm was used to block all other radiations. The camera has photon detection efficiency of 50 % at 532 nm when compared to the photon detection efficiency of 12% at 310 nm. 10 W dc light source 532 nm filter ε = 70 % Photon detection(532nm) ε = 50 % N P lens N C N C = Counts Camera N P = Number of photons Photocathode CCD chip Phosphor Micro Channel Plate (MCP) Figure B-1. Calibration set-up for photon calibration The photocathode detects the photons and emits photoelectrons. The photoelectrons are accelerated and amplified in the micro channel plate (MCP), a process referred to as gain. The amplified photoelectrons bombard the phosphor emitting photons. In turn, theses photons are 154

155 detected by the CCD chip and are read out in arbitrary units called counts. The purpose of the photon/camera calibration is to obtain the number of photons that originally arrived at the photocathode from the arbitrary unit counts. At 532 nm, the number of photons, the number of counts N c by Np is related to N p x 0.7 transmission(532) x 0.5 photon detection(532) x k (MCP,Phosphor, CCD) = Nc (B-1) where k(mcp,phosphor, CCD) represent the constant which is unknown. The expression in Equation B-1 can be rearranged in terms of N C as N p N c = (B-2) ( 0.7 transmission x 0.5 photon detection x k(mcp,phosphor, CCD) ) Also Np can be expressed as where K 532 is the calibration constant at 532 nm given by Np = K532Nc (B-3) K 532 = 1 ( 0.7 transmission x 0.5 photon detection x k(mcp,phosphor, CCD) ) Since Np = K532Nc, Mean(N ) K x Mean (N ) = Variance(N ) K x Variance (N ) P 532 C 2 P 532 C (B-4) For Poisson distribution, the mean and the variance are equal. Since the photons are Poisson distributed, Mean(N P) = Variance(N P). Hence from Equation B-4 the expression for K532 can be derived and expressed as K Mean (N ) C 532 = (B-5) Variance (N C ) 155

156 Once K 532 is known, the calibration constant at 310 nm is calculated as 0.7 x 0.5 K = K (B-6) 0.55 x A 32x32 pixel area in the camera sensor was selected for the calibration. A set of 900 images were taken at exposure times of 20, 60, 100, 140, 180, 220, 260 and 300 ns. The 10 W dc lamp source emits photons at a constant rate. In order to calibrate the camera over a range of counts the exposure time of the camera was varied from ns thereby detecting more photons and hence higher counts Figure B-2. A series of 900 images of 32x32 pixel size was obtained at each exposure For a fixed exposure time corresponding to a value of fixed count, Nc the calibration constant K 532 was calculated out of the 900 images at each pixel location. One out of the 32x32 pixels (centre one), has been highlighted. The calibration constant is similarly obtained for all other pixel locations and the average of the K 532 obtained for a particular exposure time /particular counts from the 32x32 pixel matrix is represented as the corresponding average value. 156

157 Thus the average calibration constant K 532 for the series of ns exposure time (series of counts) were obtained. The distribution of the number of photons at the central pixel location highlighted in Figure B-2 for 900 images at exposure time of 300 ns is calculated from is shown in Figure B-3 N =K N p 532exp osure300 ns C and Poisson Fit Data N Mean = 175 σ 2 = Photons Figure B-3. A series of 900 images of 32x32 pixel size was obtained each exposure The photons that arrived over a set of 900 acquisitions are shown in Figure B-3. The mean and the variance of the 900 acquisitions are 175. The Poisson fit with a mean and variance of 175 is also shown in the plot. For each exposure time the average of the counts of 900 images and 32 x 32 pixels was calculated and is plotted against the corresponding exposure time (ns) as shown in Figure B

158 Counts vs exposure at 532 nm Mean Count Mean Count + Std Mean Count - Std Linear fit 1500 Counts exposure(ns) Figure B-4. Counts vs exposure time at 532 nm From the average K 532 for the series of counts, the corresponding average K 310 was found out. The corresponding number of photons at 310 nm were calculated and plotted against the number of counts and is shown in Figure B Camera Calibration at 310 nm Data Photons Linear fit 1200 Photons y = *x Counts Figure B-5. Photons vs counts at 310 nm 158

159 The photon calibration obtained from the proposed new methodology is N p = 0.663*N c. The uncertainty in the photon calibration which is due to the non-linearity associated with the fit and accounted to 5%. The photon calibration obtained from conventional calibration shown in Figure 4-6 was N p = 1.59 * N c and is higher by a factor of 2.4 when compared to the calibration obtained from the proposed new methodology. The relatively low value predicted by the new method is attributed to the systematic and random variation of pixel intensities from the camera sensor that could have potentially affected the mean and variance of the Poisson distribution leading to errors. But compared to the conventional method of photon calibration, the proposed new methodology does not require costlier equipment like a light source of known irradiance. 159

160 APPENDIX C OH ABSORPTION PROFILES OH Absorption Profiles at 10 bar and K Temperature Range Laser Profile OH Absorption Profile at T =3148 and 10 bar (a.u) x ν (cm -1 ) (a) Laser Profile OH Absorption Profile at T =3398 and 10 bar (a.u) x 10 4 ν (cm -1 ) (b) 160

161 1 0.9 Laser Profile OH Absorption Profile at T =3277 and 10 bar (a.u) x 10 4 ν (cm -1 ) (c) Laser Profile OH Absorption Profile at T =3017 and 10 bar (a.u) x 10 4 ν (cm -1 ) (d) 161

162 1 0.9 Laser Profile OH Absorption Profile at T =2738 and 10 bar (a.u) x 10 4 ν (cm -1 ) (e) Laser Profile OH Absorption Profile at T =2480 and 10 bar (a.u) x 10 4 ν (cm -1 ) (f) Figure C-1. Absorption profile of OH simulated using LIFBASE at equivalence ratio of (a) 0.5, (b) 1, (c) 1.5, (d) 2, (e) 2.5 and (f) 3 corresponding to temperatures of K for gaseous H 2 -O 2 flame at 10 bar. 162

163 OH Absorption Profiles at 27 bar and K Temperature Range Laser Profile OH Absorption Profile at T =3245 and 27 bar (a.u) x 10 4 ν (cm -1 ) (a) Laser Profile OH Absorption Profile at T =3544 and 27 bar (a.u) x 10 4 ν (cm -1 ) (b) 163

164 1 0.9 Laser Profile OH Absorption Profile at T =3393 and 27 bar (a.u) x 10 4 ν (cm -1 ) (c) Laser Profile OH Absorption Profile at T =3085 and 27 bar (a.u) x 10 4 ν (cm -1 ) (d) 164

165 1 0.9 Laser Profile OH Absorption Profile at T =2769 and 27 bar (a.u) x 10 4 ν (cm -1 ) (e) Laser Profile OH Absorption Profile at T =2492 and 27 bar (a.u) x 10 4 ν (cm -1 ) (f) Figure C-2. Absorption profile of OH simulated using LIFBASE at equivalence ratio of (a) 0.5, (b) 1, (c) 1.5, (d) 2, (e) 2.5 and (f) 3 corresponding to temperatures of K for gaseous H 2 -O 2 flame at 27 bar. 165

166 OH Absorption Profiles at 37 bar and K Temperature Range Laser Profile OH Absorption Profile at T =3272 and 37 bar (a.u) x 10 4 ν (cm -1 ) (a) Laser Profile OH Absorption Profile at T =3587 and 37 bar (a.u) x 10 4 ν (cm -1 ) (b) 166

167 1 0.9 Laser Profile OH Absorption Profile at T =3427 and 37 bar (a.u) x 10 4 ν (cm -1 ) (c) Laser Profile OH Absorption Profile at T =3103 and 37 bar (a.u) x 10 4 ν (cm -1 ) (d) 167

168 1 0.9 Laser Profile OH Absorption Profile at T =2777 and 37 bar (a.u) x 10 4 ν (cm -1 ) (e) Laser Profile OH Absorption Profile at T =2496 and 37 bar (a.u) x 10 4 ν (cm -1 ) (e) Figure C-3 Absorption profile of OH simulated using LIFBASE at equivalence ratio of (a) 0.5, (b) 1, (c) 1.5, (d) 2, (e) 2.5 and (f) 3 corresponding to temperatures of K for gaseous H 2 -O 2 flame at 37 bar. 168

169 OH Absorption Profiles at 53 bar and K Temperature Range Laser Profile OH Absorption Profile at T =3308 and 53 bar (a.u) x 10 4 ν (cm -1 ) (a) Laser Profile OH Absorption Profile at T =3646 and 53 bar (a.u) x 10 4 ν (cm -1 ) (b) 169

170 1 0.9 Laser Profile OH Absorption Profile at T =3470 and 53 bar (a.u) x 10 4 ν (cm -1 ) (c) Laser Profile OH Absorption Profile at T =3125 and 53 bar (a.u) x 10 4 ν (cm -1 ) (d) 170

171 1 0.9 Laser Profile OH Absorption Profile at T =2787 and 53 bar (a.u) x 10 4 ν (cm -1 ) (e) Laser Profile OH Absorption Profile at T =2500 and 53 bar (a.u) x 10 4 ν (cm -1 ) (f) Figure C-4. Absorption profile of OH simulated using LIFBASE at equivalence ratio of (a) 0.5, (b) 1, (c) 1.5, (d) 2, (e) 2.5 and (f) 3 corresponding to temperatures of K for gaseous H 2 -O 2 flame at 53 bar. 171

172 APPENDIX D OH NUMBER DENSITY CONTOURS Thirteen Instantaneous OH Number Density Contours at 10 bar (a) (b) (c) 172

173 (d) (e) (f) 173

174 (g) (h) (i) 174