Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Investigations of fuel and hydrodynamic effects in highly turbulent premixed jet flames
(USC Thesis Other)
Investigations of fuel and hydrodynamic effects in highly turbulent premixed jet flames
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Investigations of Fuel and Hydrodynamic Effects in Highly Turbulent Premixed Jet Flames By Laurel Paxton December 2019 A Dissertation Presented to the Faculty of the USC Graduate School University of Southern California Presented in Partial Fulfillment of the Requirements of the Degree Doctor of Philosophy (Mechanical Engineering) Acknowledgements This dissertation would not have been possible for the conversations, advice and collaborations with many in the AME department and my other walks of life. First and foremost, I have had the privilege to work with Professor Egolfopoulos from whom I have learned so much. I am thankful for his guidance and encouragement to think outside of the box which has made me a better engineer. I would like to thank Jennifer Smolke for teaching me everything I know about how to be a good experimental researcher. Some of my fondest moments of the PhD program are the times we spent working in the lab. I would also like to acknowledge Steven Luna and Hiba Kahouli for helping me run and build the turbulent burner rigs and for the countless hours of troubleshooting experimental and numerical results. Additional thanks to the rest of my labmates (Roe, D.J., Jaggan, Vyaas, Ashkan and Robert) for their guidance and support. I would like to thank my collaborators at UCLA, including Dr. Pineda, Chuyu Wei, Fabio Bendana, and Kevin Schwarm for wonderful discussions of combustion and life outside the PhD program. Thank you to the AME department sta for helping with the ins and outs of equipment acquisi- tion and the intricacies of the PhD program requirements as well as being willing to help whenever problems arose. I would like to acknowledge funding from the NDSEG fellowship for the past two years which allowed me to focus full time on this project. This work was also thanks to the sponsorship by Dr. Chiping Li at AFOSR for supporting our turbulent ame work and asking the tough questions that need to be asked. To the rest of my friends both in the PhD program and out, thank you for being there for me through this process. I could not have done this without you. Thank you to my family for always being in my corner. I am so incredibly glad that I can share this accomplishment with you. Finally, this dissertation is dedicated to Alex Valentino. Thank you for your patient support throughout the late nights and weekends. You are my rock through all of the good times and the bad. ii Contents Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi 1 Introduction 1 1.1 Combustion in Aircraft Engines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Literature Review: Fuel and Hydrodynamic Eects in Turbulent Premixed Combustion 2 1.3 Literature Review: Pressure Eects in Turbulent Premixed Combustion . . . . . . . 4 1.4 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2 Background 6 2.1 Laminar Flame Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.1.1 Premixed Flames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.1.2 The Eect of Fuel Type on Laminar Flame Structure and Chemical Kinetics 7 2.1.3 Molecular Transport and Heavy Hydrocarbons . . . . . . . . . . . . . . . . . 8 2.1.4 Flame Stretch and the Counter ow Conguration . . . . . . . . . . . . . . . 9 2.2 Turbulent Combustion Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2.1 Basics of Turbulence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2.2 Regimes of Premixed Turbulent Combustion . . . . . . . . . . . . . . . . . . 12 2.2.3 Eects of Pressure on Turbulent Flames . . . . . . . . . . . . . . . . . . . . 13 2.2.4 Canonical Turbulent Flame Congurations . . . . . . . . . . . . . . . . . . . 14 3 Experimental Methodology 15 3.1 Experimental Congurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.1.1 Atmospheric Piloted Premixed Jet Burner (PPJB) . . . . . . . . . . . . . . 15 3.1.2 Variable Pressure Burner Conguration . . . . . . . . . . . . . . . . . . . . . 17 3.2 Diagnostic Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.2.1 Diagnostic Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.2.2 CH* Chemiluminescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.2.3 Particle Image Velocimetry (PIV) . . . . . . . . . . . . . . . . . . . . . . . . 20 3.2.4 Formaldehyde (CH 2 O) PLIF . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.2.5 Simultaneous CH 2 O PLIF - 2-D PIV . . . . . . . . . . . . . . . . . . . . . . 22 iii 3.2.6 Laser Absoprtion Tomography . . . . . . . . . . . . . . . . . . . . . . . . . . 24 4 Numerical Methodology 26 4.1 PREMIX Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 4.2 Unsteady Opposed Jet Code (Unsteady OPPDIFF) . . . . . . . . . . . . . . . . . . 26 4.3 Reynolds-Averaged Navier Stokes (RANS) Calculations . . . . . . . . . . . . . . . . 27 4.3.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 4.4 Kinetic Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 5 Eects of Heat Release and Fuel Type on Global Observables 29 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 5.2 Experimental Parameter Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 5.3 Global CH* Observables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 5.3.1 Flame Height (H ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 5.4 Characteristics of the Shear Layer: u 0 Layer Thickness . . . . . . . . . . . . . . . . 31 5.4.1 Heat Release Eects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 5.4.2 Fuel Eects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 5.5 Turbulent Kinetic Energy and Turbulent Shear Stress in the Shear Layer . . . . . . 33 5.6 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 6 Preliminary Investigations of Global Flame Topology at Elevated Pressures 36 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 6.2 Experimental Parameter Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 6.3 Experimental Observables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 6.4 Fuel and Pressure Eects on Global Flame Observables . . . . . . . . . . . . . . . . 38 6.4.1 Global Flame Shape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 6.4.2 Flame Height (H ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 6.4.3 Maximum CH* Intensity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 6.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 7 Investigations of Fuel and Hydrodynamic Eects through Laser Absorption Spec- troscopy (LAS) 43 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 7.2 Hydrodynamic Eects on the Flame Thermochemical Structure . . . . . . . . . . . 44 7.2.1 Experimental Parameter Space . . . . . . . . . . . . . . . . . . . . . . . . . 44 7.2.2 Numerical Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 7.2.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 7.3 Fuel Eects on the Flame Thermochemical Structure . . . . . . . . . . . . . . . . . 48 7.4 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 iv 8 Simultaneous Measurement of the Local Velocity Field and Preheat Structure 51 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 8.2 Experimental Parameter Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 8.3 Eects of Fuel Type on C 2 HO Concentration and Structure . . . . . . . . . . . . . . 53 8.4 Conditional Velocity and Vorticity Statistics . . . . . . . . . . . . . . . . . . . . . . 54 8.4.1 Calculation of Conditional Statistics . . . . . . . . . . . . . . . . . . . . . . 54 8.4.2 Conditional Statistics: V 0 C;x . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 8.4.3 Conditional Statistics: ! 0 C;xy . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 8.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 9 Assessment of Experimental Observables for Local Extinction Through Unsteady Laminar Flame Calculations 59 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 9.2 Numerical Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 9.2.1 Unsteady OPPDIFF Conguration . . . . . . . . . . . . . . . . . . . . . . . 60 9.2.2 Conditions of Interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 9.3 Characterization of the Extinction Transient: Indicators of Heat Release . . . . . . 61 9.3.1 Characterization of the Extinction Transient: Behavior of CH 2 O Through Extinction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 9.4 Eect of Increased Pressure on Experimental Observables . . . . . . . . . . . . . . . 65 9.4.1 Eect of Pressure on X CH 2 O;max . . . . . . . . . . . . . . . . . . . . . . . . . 65 9.4.2 Correlation between the Product of X OH and X CH 2 O and _ q max . . . . . . . . 66 9.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 10 Conclusions and Recommendations 68 10.1 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 10.2 Recommendations for Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 v List of Figures 1.1 Typical aircraft engine combustion structure . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Borghi/Peters diagram demonstrating range of experimentally investigated fuels and conditions. Present work is indicated in red. . . . . . . . . . . . . . . . . . . . . . . 3 2.1 Laminar premixed ame structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.2 Dependence of S L on fuel type [1] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.3 Lumped mechanism for high temperature n C 12 H 26 oxidation . . . . . . . . . . . 9 2.4 Congurations of various stretched ames . . . . . . . . . . . . . . . . . . . . . . . 10 2.5 Regimes of turbulent combustion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 3.1 Piloted Premixed Jet Jurner (PPJB) cutaway. All measurements are in mm. . . . . 15 3.2 USC variable pressure chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 3.3 VP-PPJB cutaway. All measurements are in mm. . . . . . . . . . . . . . . . . . . . 18 3.4 Schematic of implemented diagnostics . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.5 H calculation from CH* centerline luminosity . . . . . . . . . . . . . . . . . . . . . 19 3.6 1-D calculated laminar ame structure for a lean premixed CH 4 -air ame . . . . . . 23 3.7 Sample Mie scattering and CH 2 O signal for a Jet-A/air ame,S L =15.3 cm/s,Re jet =25,000. 23 3.8 Hardware and representative data analysis for laser absorption tomography system [2] 24 4.1 Unsteady counter ow frequency regimes . . . . . . . . . . . . . . . . . . . . . . . . 27 5.1 H variation with S L , _ q max , and f at Re jet =25,000. . . . . . . . . . . . . . . . . . . 30 5.2 u 0 variation with x=D and _ q max for toluene ames at Re jet =25,000. . . . . . . . . . 31 5.3 u 0 variation with x=d for toluene ames at Re jet =25,000. . . . . . . . . . . . . . . 32 5.4 Fuel eects in d u 0=dx with changing _ q max and S L . Re jet =25,000. . . . . . . . . . . . 33 5.5 TKE and TSS variation with x=D for matching _ q max . . . . . . . . . . . . . . . . . . 34 6.1 Global ame shape at 1, 2 and 3 atm demonstrating tip-quenched and tip-ignited states for Re jet =25,000 and S L =24 cm/s. . . . . . . . . . . . . . . . . . . . . . . . . 38 6.2 Global ame appearance as a function of fuel forRe jet =25,000,P =2 atm andS L =24 cm/s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 6.3 H /D as a function of S L at 1, 2, and 3 atm. . . . . . . . . . . . . . . . . . . . . . . 40 vi 6.4 Maximum experimental CH* intensity and calculated CH* mole fraction for exam- ined C 1 through C 3 hydrocarbons at 1 atm. . . . . . . . . . . . . . . . . . . . . . . 41 6.5 Maximum CH* intensity for CH 4 and C 2 H 4 at 1, 2, and 3 atm. . . . . . . . . . . . . 41 6.6 Maximum calculated CH* mole fraction for CH 4 and C 2 H 4 at 1, 2, and 3 atm. . . . 42 7.1 Experimental setup: the regions of interest for the computation are indicated via the red dashed lines. A radial cut of interest (x=D=14) is shown as a white dashed line. 45 7.2 Radial proles for experimental CO, CO 2 mole fractions, and temperature forRe jet =25,000 and Re jet =50,000. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 7.3 Radial proles for simulated CO, CO 2 mole fractions, and temperature forRe jet =25,000 and Re jet =50,000. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 7.4 Normalized radial proles for simulated CO, CO 2 mole fractions, andk forRe jet =25,000 and Re jet =50,000. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 7.5 Predicted CO mole fraction from laminar ame calculations . . . . . . . . . . . . . 49 7.6 Experimental CO mole fractions for conditions considered . . . . . . . . . . . . . . 50 8.1 Eect of increasing Reynolds number on the instantaneous CH 2 O signal for an C 2 H 4 - air ame, S L =11.3 cm/s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 8.2 Radial proles of averaged CH 2 O signal at x=D=16, Re jet =25,000 for CH 4 , C 2 H 4 and Jet-A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 8.3 Example contours of local distance from the reaction zone (shown in red). Contours are shown only for the reactants to improve clarity. . . . . . . . . . . . . . . . . . . 55 8.4 Conditional and unconditioned velocity statistics for CH 4 and Jet-A,S L =15.3 cm/s, Re jet =25,000. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 8.5 Conditional and unconditioned vorticity statistics for CH 4 and Jet-A,S L =15.3 cm/s, Re jet =25,000. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 9.1 Extinction transient for premixed CH 4 ame (P =1 bar,T u =403K). . . . . . . . . . . 62 9.2 Reaction path analysis for premixed CH 4 and n-C 12 H 26 ames at P =1 bar. . . . . . 63 9.3 Zoomed snapshot of evolution of (X OH;max ) scaled , (X CH 2 O;max ) scaled and ( _ q max ) scaled with CH 2 O destruction timescale for premixed CH 4 and n-C 12 H 26 ames at P =1 bar. . 64 9.4 Dependence of key transient species on maximum temperature, P =1 bar, T u =403K. 64 9.5 Evolution of (X CH 2 O;max) scaled and (X OH;max) scaled for n-C 12 H 26 ame at P =1, 5, and 10 bar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 9.6 Evolution of OH x CH 2 O concentration and _ q max for premixed n-C 12 H 26 ame at P =1 and 5 bar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 vii Chapter 1 Introduction Combustion is the world's primary energy conversion method, producing over 80% of the world's total energy. Our transportation system is also almost entirely reliant on combustion. In the United States in 2013, ground vehicles and aircraft burned approximately two-thirds of all petroleum imported or produced in the United States. Commercial aviation has become an invaluable part of our global economy, facilitating both economic and cultural exchanges. Combustion research can help identify the physical and chemical processes that are responsible for the conversion of reactants to nal products. This will enable improved eciency and safety while reducing the emissions of harmful pollutants including carbon dioxide. As combustion behavior is application-dependent, this study will focus on the combustion envi- ronment expected for gas turbines, in particular aircraft engines. 1.1 Combustion in Aircraft Engines Most commercial and military aircraft are powered by gas turbine engines. After air intake, the ow passes through a series of compressors to create the pressure dierential required for thrust. Fuel is then added to the ow in the combustor and burned. Some of the energy in the heated exhaust is utilized to power the turbine (in turn assisting the compressor rotation), while the remaining exhaust produces thrust by exiting through a nozzle. The combustor is therefore a crucial part of any gas turbine engine. In the combustor, fuel is introduced using an injector. In order to ensure proper mixing of fuel and air prior to combustion, turbulence is generated, often using perpendicular injection of air into the main ow to produce a swirling oweld. In a premixed ame, the fuel and air is mixed together before ignition. Due to the high ow velocity, stabilization devices such as swirl stabilization, blu bodies or pilot ames are used to anchor the ame in the combustion zone. This stabilization allows the premixed ame to operate with leaner mixture ratios with lower temperatures. This reduces fuel consumption and pollutant formation. However, premixed ames can also be subject to ame blowo (extinction due to high velocity gradients in the ow) and ashback ( ame propagation backwards towards the inlet), which can cause unpredictable combustion and engine damage. Aircraft engines also operate 1 Figure 1.1: Typical aircraft engine combustion structure at high pressures (up to 30 atm for current engines and up to 60 atm for next-generation engines) to improve eciency and lower ame temperatures to reduce NOx emissions. Hydrocarbon fuels are of particular importance in aviation. As aircraft have limited weight to carry fuel, aviation fuels must have a high energy density, providing a large heat release for the amount burned. For this reason, commercial and military aviation have turned to kerosene- based liquid fuels, with high molecular weights (MW) in the 160-180 range with an average carbon number of 11.5-12.5. These fuels, such as Jet-A, JP-5 and JP-8, have a greater energy density per volume than gaseous low carbon number fuels, such as methane. These fuels are blends of dierent hydrocarbon types with varying paran, olen and aromatic content. Modeling the behavior of a combustor is therefore challenging due to the dynamics of combustion phenomenon. Turbulent combustion involves the nonlinear coupling of thermodynamics, chemical kinetics, uid mechanics and molecular transport. Due to the complex nature of these interactions, most combustion modeling and engine designs continue to rely on empirical data. Empirical data, such as the boundary conditions for canonical experiments, enables modelers to anchor development work to the real physical phenomena occuring. Experimental studies therefore have much to oer in the eort to improve engine performance and avoid catastrophic behavior. 1.2 Literature Review: Fuel and Hydrodynamic Eects in Turbulent Premixed Combustion Though turbulent premixed combustion has been studied extensively, relatively few investigations have assessed nite-rate chemistry or the eect of fuel chemistry, especially at high turbulence levels. In addition to the Hi-Pilot conguration developed by Driscoll [3], Bilger and colleagues developed the piloted premixed jet burner (the Sydney burner) in order to access intense turbulence and high 2 Karlovtiz numberKa environments [4]. Under these conditions, the smallest eddies of the ow can enter and subsequently broaden the preheat zone. This zone is referred to as the thin reaction- broadened preheat zone [5]. Increasing the turbulence level further produces even smaller eddies that are able to penetrate the reaction zone and potentially modify the ame structure. This regime is termed the broken reaction zone regime. Experimental studies have reported evidence of preheat broadening (e.g., [6, 7, 8, 9, 10]), however evidence of modied or broken reaction layers has not been observed at extreme levels of turbulence when removing eects of cold air entrainment [11]. The Borghi/Peters diagram attempts to dene the existence of these regimes by using key ame and turbulence properties such as laminar ame speed S L , integral length scale L int , laminar ame thickness f and turbulent velocity uctuationu 0 [12, 13]. These concepts will be discussed in greater length in Section 2.2. Figure 1.2 shows the Borghi/Peters diagram with a representative survey of turbulent premixed ame experiments. Figure 1.2: Borghi/Peters diagram demonstrating range of experimentally investigated fuels and conditions. Present work is indicated in red. Two points can be drawn from this experimental survey. First, while the lower end of the thin reaction zone regime has been explored, the upper limits of the thin reaction zone and broken reaction zone regime have not been well-studied. Second, most studies in these regions have focused on methane and hydrogen, and so there remains a scarcity of data for dierent fuels (especially high carbon number fuels), despite their importance in practical combustion engines. Such heavy hydrocarbons generally have a lower resistance to cracking and decompose into smaller fragments within the preheat zone [14, 15, 16, 17]. If the thermal structure of the preheat zone is modied by turbulence, the eects may be minor for methane ames, as methane is more resistant to fuel decomposition. However, this may or may not be the case for heavy hydrocarbons. Increased exposure to the high temperatures of the preheat zone due to preheat broadening could alter the composition of hydrocarbon fragments reaching the reaction zone, causing dierent extinction and 3 propagation behaviors [18]. Additionally, the diusivities of heavy fuels are substantially dierent compared to hydrogen and methane. Preferential diusion eects can further alter the local ame structure [19]. It is important to note here that in past turbulent combustion studies (e.g., [20, 7, 21, 22]), fuel eects were typically lumped under the laminar ame speed (S L ), as follows from Damkohler's original representation of S T =S L . This is indicative of assumptions also behind a large body of turbulent combustion models: the structure of a turbulent ame is the same as that of a one- dimension freely propagating ame. A previous study for various fuels by Carbone et. al [23] showed that in the presence of signicant heat loss, the global ame height derived from CH* measurements when plotted against S L , did suppress fuel eects among the liquid fuels, though methane remained an outlier, suggesting that the fuel chemistry may impact the turbulence- ame interaction. 1.3 Literature Review: Pressure Eects in Turbulent Pre- mixed Combustion Though turbulent premixed combustion has been studied extensively, relatively few investigations have assessed the importance of nite-rate chemistry in the context of thermodynamic conditions. While atmospheric experiments can reveal important features related to the interaction of turbulence and ame chemistry, the ame structure and dynamic behavior at engine-relevant conditions can vary substantially from those of atmospheric or near-atmospheric ames. Aircraft engines operate at high pressures to improve eciency and lower ame temperatures to reduce NOx emissions. In contrast, sub-atmospheric pressures are also relevant to high-altitude re-light and lean blowout in gas turbines and scramjets. However, the impact of pressure on the coupling of fuel chemistry and turbulence has yet to be well characterized. Out of 3300 published studies on turbulent ames that involve theory, modeling, and/or ex- periments, less than 2% of all turbulent ame studies have been carried out at non-atmospheric pressure, i.e. either for P < 1 atm or P > 1 atm, for all fuels. Only 4 have involved experiments with pre-vaporized fuels at non-atmospheric conditions [20, 24, 25, 26, 27, 28, 29, 30] In the existing P > 1 atm experimental studies of turbulent ames, CH 4 has been largely the fuel of choice and the limited C 2 H 4 and C 3 H 8 results are not sucient to derive substantial conclusions about fuel eects. Nonlinear pressure chemical eects will potentially result in complicated scaling laws for global observables such as ame height, blow-o limits, etc. It is essential that the response of global observables to pressure and fuel be rst experimentally measured so that scaling laws can be devel- oped. 4 1.4 Objectives Based on the considerations raised in the literature reviews, the main goal of this investigation is the experimental characterization of stability limits, global properties, and aspects of the local struc- ture of turbulent premixed ames through measurements of velocities and scalar properties. This goal can be broken down into two objectives. The rst objective is to generate well-characterized experimental data for shear-dominated turbulent jet ames for a wide range of fuel types (small to large alkanes/alkenes and aromatic fuels), equivalence ratios and pressures. The second objective is to characterize the eects of fuel chemistry while keepingS L constant among the fuels examined. These studies will provide a much-needed look at the role of fuel chemistry in the high Reynolds and Karlovitz number regimes. Experimental results will be obtained using a modied Sydney burner facility to consider a range of fuels including light gaseous fuels such as methane to heavy liquid fuels such as dodecane. A variable pressure facility will be used to extend capabilities to elevated pressures to asses the role of pressure in scaling global properties and local structure. Results will focus rst on global observables of the ame, including the ame height derived from CH* luminosity as well as the two-dimensional spatially resolved velocity elds through Particle Image Velocimetry (PIV). While PIV is crucial to understanding the behavior of a turbulent oweld, it is impossible in highly turbulent congurations to determine velocity vectors before and after the ame using the processed velocity eld alone and highly challenging in the pre-processed particle eld images. Therefore, it is necessary to use an additional diagnostic to identify the location of the ame. CH 2 O PLIF is used simultaneously with PIV to identify the location of the ame and thus identify velocity vectors as before or after the ame. The calculation of these velocity statistics conditioned on the ame surface allows the examination of the eect of the ame on the velocity eld. The eect of fuel chemistry will be examined through both the resultant CH 2 O intensity and the conditioned velocity statistics. Quantitative measurements of the thermochemical structure of the ame are also of great im- portance because they provide model validation targets. Potential fuel and hydrodynamic eects will be explored through collaboration with the Spearrin group at UCLA to produce quantitative 2-D spatially-averaged CO and CO 2 mole fraction and temperature elds. 1-D laminar ame speed calculations, opposed jet calculations and Reynolds-Averaged Navier- Stokes simulations have also been performed to add insight to the experiments presented above. The document is organized as follows: Chapter 2 establishes the necessary background to un- derstand laminar premixed ames and turbulent premixed combustion theory. Chapters 3 and 4 detail the experimental and numerical methodology respectively. Chapters 5 through 9 discuss the results that have been obtained through the course of the project. Chapter 10 presents potential future directions for the project and summarizes the impacts to the combustion community of the completed studies. 5 Chapter 2 Background 2.1 Laminar Flame Theory 2.1.1 Premixed Flames Combustion involves the combination of a fuel and an oxidizer. In premixed ames, the fuel and oxidizer are mixed together before being burned. This mixture can be more precisely described through the equivalence ratio that denes the proportions of oxidizer to fuel: = (F=A) actual = (F=A) stoichiometric (2.1.1) where F and A are the mass/molar quantities of the fuel and air respectively. Flames with less than 1 are referred to as fuel-lean or simply lean ames, as the fuel is the decient reactant. Regardless of equivalence ratio, the global structure of a premixed ame can be divided into three regions: the preheat zone, the reaction zone, and the downstream equilibrium region. Figure 2.1 shows this structure. Figure 2.1: Laminar premixed ame structure The preheat zone is the molecular transport zone where reactant transport and thermal diusion rst takes place. It is dened as the balance between convection and thermal diusion, as reactants are preheated up to a temperature at which reaction rates become signicant. The next zone is the reaction zone is where vigorous chemical reactions take place. As the chemical reactions are high 6 activation energy processes, the reaction zone is very thin, usually a factor of 10 smaller than the preheat zone. There are three fundamental parameters that characterize a laminar premixed ame: laminar ame speed S L , laminar ame thickness f , and laminar ame time t f . S L is dened as the propa- gation speed of the Ideal Flame Model (IFM) [1]. Under the IFM, ames must be steady, laminar, 1D, planar, and adiabatic. S L is a fundamental characteristic of a premixed mixture that describes its' exothermicity, diusivity and reactivity. The dependence ofS L on these parameters can be seen in Equation 2.1.2 whereP is the pressure, is the transport coecient,E a is the activation energy, and T 1 is the maximum temperature in the oweld: S L P (n=21) c p 1=2 exp E a 2R o T +1 (2.1.2) It is clear that while S L is dependent on transport properties, it is also strongly dependent on the chemical kinetics through the pressure and Arrhenius terms. In addition to S L , f characterizes the balance of heat and mass ux into the ame front. f is dened as: F = (T max T u )=jrTj max (2.1.3) A nal parameter t f characterizes the balance between thermal/mass diusion and convection in the ame front and is dened below: t f = f =S L (2.1.4) The ame time is of importance in turbulent combustion studies, as the relation of ame time to a characteristic timescale of the ow can inform the expected behavior of the ame. 2.1.2 The Eect of Fuel Type on Laminar Flame Structure and Chem- ical Kinetics As S L is dependent on the reaction rate in the ame, it is therefore highly sensitive to the ame temperature and fuel type. Dierent fuels have dierent reaction pathways due to dierences in molecular structure and bond type. Figure 2.2 shows the behavior ofS L for three main hydrocarbon groups. Alkanes have the lowestS L and level o with increasing carbon number. Both alkenes and alkynes exhibit a sharp drop-o in S L and similarly level out with larger carbon numbers. This is due to the dilution of the double or triple bond present with the increase in saturated bonds from more carbon atoms. The process of hydrocarbon breakdown is initiated when the fuel molecules are attacked through H-abstraction reactions by the H, O, and OH radicals. This produces a large radical that can then be broken down into smaller and smaller fragments. This breakdown process is governed by the - scission rule. For a radical, an unpaired electron at the radical site strengthens the adjacent bonds. For example, for alkanes, the C-C bond next to the radical-paired bond is the most susceptible to break. This process repeats over and over until the fuel molecule has been reduced to simple 7 Figure 2.2: Dependence of S L on fuel type [1] components, primarily the H radical, methyl (CH 3 ) radical, H 2 , CH 4 , C 2 H 4 , C 3 H 6 , and C 6 H 12 . Describing the pathways of these decomposition reactions is a major area of study in constructing and analyzing combustion mechanisms that describe the chemistry occurring during combustion. A sample mechanism is shown in Fig. 2.3. 2.1.3 Molecular Transport and Heavy Hydrocarbons For premixed ames, the Lewis number Le is an important parameter that describes the balance between thermal and mass diusivity: Le = D d (2.1.5) where is the thermal diusivity of the reactant mixture andD d is the mass diusivity of the de- cient reactant. For the consideration of lean ames, the mixture averaged Le can be approximated as: Le = N 2 D fuel = s MW fuel MW N 2 (2.1.6) The Lewis number plays a critical role in the presence of ame curvature, which will be discussed in the next section. Another consideration in the Lewis number formulation is the dierence in mass diusivity between the fuel and oxidizer, also known as preferential diusion. This is especially important with heavy hydrocarbons. As the carbon number increases, the mass diusivity decreases, shifting the ratio of oxygen and fuel diusivity. This can cause oxygen to be preferentially transported into the ame which can modify local reactivity. 8 Figure 2.3: Lumped mechanism for high temperature n C 12 H 26 oxidation 2.1.4 Flame Stretch and the Counter ow Conguration Flame Stretch The concept of ame stretch was rst introduced by Karlovitz to describe ame extinction in the presence of velocity gradients. In the past decades, signicant advances have been made in under- standing how ame stretch in uences ame structure, as all practical ames experience stretch [31]. This includes mathematical analyses of the structure and propagation of stretched ames as well as experimental and numerical verication of predicted properties. Flame stretch is dened as the rate of area generation along the ame surface: = 1 A dA dt =r t v s,t + (V F n)(rn) (2.1.7) In this equation,r t is the tangential gradient operator at the surface, v s,t is the tangential component of ow velocity at the ame, V F is the velocity of the ame, and n is the unit normal vector of the ame surface. A closer examination of the equation shows that is aected by aerodynamic straining (r t v s,t ), ame motion (V F ) and ame curvature (rn). Depending on the ame conguration, ames can experience either positive, zero or negative stretch. Positive ame stretch occurs when the boundary of a surface element expands to create more surface. Negative stretch on the other hand occurs when the ame undergoes compression. Figure 2.4 shows several classical stretched ame congurations. Dierential diusion can also interact with stretch to alter local reactivity. Such eects are strongly dependent on the ame geometry. 9 Figure 2.4: Congurations of various stretched ames The Counter ow Flame The laminar counter ow conguration is one of the simplest experimental techniques to obtain the laminar ame speed by utilizing a stretched ame. The opposed jet or counter ow conguration consists of two concentric nozzle impinging to create an axisymmetric ow with a stagnation plane between the two ows. Due to the symmetric nature of the conguration, two premixed ames can be stabilized on either side of the stagnation plane. This is termed the twin ame conguration. When using a twin ame conguration, the boundary conditions at both nozzle and stagnation plane are identical. Therefore, a stream function can be used to reduce the 2D axisymmetric ow conguration to a 1D problem. (x;r) =r 2 U(x);u = 1 r @ @r ;v = 1 r @ @x (2.1.8) Inserting this into the continuity equation produces: @ @x (ru) + @ @r (rv) = 0 (2.1.9) Finally, it is assumed that thermodynamic properties vary only in the axial direction, and through the low Mach number assumptions, thermodynamic pressure throughout the ow is con- stant. Therefore, there is one numerical solution, where the radial pressure gradient is a constant eigenvalue of the ow: 1 r @P @r =C (2.1.10) 10 Stretch in the counter ow conguration can be expressed at the centerline as: = du dx (2.1.11) The minimum axial centerline velocity just upstream of the ame is dened as the reference ame speed S u, ref , and the maximum absolute value of the velocity gradient in the hydrodynamic zone is dened as the strain rate K. Counter ow extinction can happen through several dierent mechanisms, depending on the Lewis number of the mixture. For Le<1, the enthalpy from mass transport is dominant over heat loss from the ame. Increases in stretch will increase the burning rate and push the ame closer to the stagnation plane. Proximity to the stagnation plane will decrease ame thickness and residence time, resulting in incomplete combustion and eventually extinction. ForLe>1, heat loss is dominant over the enthalpy from mass transport into the ame. Increases in stretch will reduce the burning rate. Eventually, the burning rate is no longer sucient to support complete combustion, and extinction will occur. 2.2 Turbulent Combustion Theory 2.2.1 Basics of Turbulence Turbulent ows form the basis for practical combustion devices, as the turbulence promotes rapid mixing and increased ame surface area. Turbulence contains uctuations in velocity, pressure and temperature. In particular, turbulence is characterized by identiable eddies that deform, rotate and divide. Turbulent motion is inherently random, therefore descriptions of turbulence are statistical in nature. First, several key turbulence parameters will be dened. Assuming that observations at a certain point are ensemble-averaged, U is the mean velocity at that point. The uctuating component of velocity due to turbulence is termed u 0 and is dened: u 0 = ( <u i u i > 3 ) 1=2 = (2k=3) 1=2 (2.2.1) where u i is the ith component of the 3D velocity vector and k is the averaged turbulent kinetic energy of the ow. Energy is transferred through a turbulent ow from the largest to the smallest scales, according to Richardson's energy cascade [32]. Vortex stretching transfers energy to smaller and smaller scales until it is dissipated due to viscosity. The dissipation rate describes how the kinetic energy of the ow is dissipated. This dissipation occurs predominantly at the smallest scales, where the velocity gradients are the largest: =h @u i @x j @u i @x j i (2.2.2) The largest and smallest length scales of the ow can be further characterized. The integral length scale L int is one of the largest scales in the ow and is considered the characteristic length 11 scale for that ow. It is the scale at which energy is injected into the ow. L int is dened statistically based on measured properties of the turbulent ow: L int = (u 0 ) 3 = (2.2.3) For turbulent jets, L int is proportional to the jet diameter. The smallest scales of the ow can also be numerically described. Kolmogorov's 1941 theory on turbulence further the concept of the energy cascade to state that for very high Reynolds number, the statistics of the smallest scales are uniquely described by the relationship between kinematic viscosity and energy dissipation, under the assumption of local isotropy [33]. This scale is known as the Kolmogorov scale : = ( 3 ) 1=4 (2.2.4) A Kolmogorov time scale can also be derived: = ( ) 1=2 (2.2.5) 2.2.2 Regimes of Premixed Turbulent Combustion When discussing turbulent combustion behavior, there are three important nondimensional num- bers: turbulent Reynolds number Re t , Karlovitz number Ka and Dahmkohler number Da. Re t provides a more local description of the turbulent uctuations, eddy size and viscosity. Re t is dened as: Re t = u 0 L int (2.2.6) In addition to Re t , Ka and Da relate the turbulent time scales to the ame time scales. Ka is a measure of the imposed strain rate on the ame surface: it relates the chemical time scale to the Kolmogorov time scale: Ka =Da 1 =t F =t = 2 F = 2 (2.2.7) Ka is also the inverse ofDa calculated at the Kolmogorov scale. LargeKa means that chemical reactions occur slower than some of the turbulent time scales. In this situation, turbulent eddies can enter the ame front and potentially disrupt the chemical reactions occurring. The Borghi/Peters Regime Diagram A regime diagram is a way to classify the dierent regimes of turbulent combustion by using simple relationships between turbulence and combustion parameters. In addition to Re t and Ka, two other relationships are used: u 0 /S L and L int / f . u 0 /S L relates the convective ow scales and ame propagation. L int / f relates the ame thickness to the size of eddies in the ow. These parameters can be related together using the following equation: u 0 =S L =Re t (L int = f ) 1 =Ka 2=3 (L int = f ) 1=3 (2.2.8) Using these relationships, a regime diagram can be created to distinguish regimes of turbulent 12 combustion. This diagram, shown in Fig. 2.5 was rst proposed by Borghi [12] and subsequently modied by Peters [13]. Figure 2.5: Regimes of turbulent combustion It should be noted that conditions for practical combustion applications place ames in the thin and broken reaction zones. The thin reaction zone is characterized by eddies that are small enough to penetrate the preheat zone of the ame: is smaller than the preheat zone but still larger than the reaction zone. At higher turbulence levels, becomes smaller than the thickness of the reaction layer and thus can potentially modify the ame structure. Though the diagram has been used as a core foundation in the study of turbulent combustion, it is based on the assumption of unity Lewis number and may therefore not be accurate nor appropriate for classifying the behavior of heavy fuels. 2.2.3 Eects of Pressure on Turbulent Flames Kobayashi, Williams, Niioka, Dinkelacker, and coworkers (e.g., [30, 25, 27, 29] have investigated the turbulent ame speeds and structure atP > 1 atm and have shown (among others) that as pressure increases, the turbulence scales and the rate of turbulence decay decrease due to the decrease of the kinematic viscosity for grid-generated turbulence. The ames strongly wrinkle due to the small- scale turbulent motions and/or the increased instability-forming propensity. In addition, S T /S u o increases faster with u 0 /S u o due to the enhancement of Darrieus-Landau instabilities as the ames get thinner. This increase however is less for C 3 H 8 /air compared to C 2 H 4 /air ames given that the Darrieus-Landau instabilities are restrained for C 3 H 8 ames due to diusive-thermal eects (Le > 1). Nevertheless, in the existing P > 1 atm experimental studies of turbulent ames, CH 4 has been largely the fuel of choice and the limited C 2 H 4 and C 3 H 8 results are not sucient to derive substantial conclusions about fuel eects, especially chemical. It is well established by numerous 13 studies, that as well as ame extinction/ignition limits depend non-linearly on the local density due to the competition between 2-body chain branching and 3-body chain termination reactions that control the oxidation of low carbon-number hydrocarbons [34]. Furthermore, the small-scale structures result in intense mixing and preheat zone broadening. This in turn could aect the decomposition process of high molecular weight fuels, given the large sensitivity of decomposition kinetics to temperature. Together, these factors make turbulent ame dynamics at engine-relevant conditions a complex problem that warrants a parametric experimental investigation to identify the eects of fuel chemistry and thermodynamic states. 2.2.4 Canonical Turbulent Flame Congurations It is generally desirable that the fundamental characteristics of a turbulent ame could be described by simple correlations that would enable the general application of these parameters to dierent systems. However, experiments and computations have shown that upstream wrinkling is coupled with the downstream wrinkling of the ame [6]. Thus, ame wrinkling is not constant across the ame front and depends on the experimental conguration. To accommodate this geometric dependence, turbulent ame studies typically utilize four general classications for experimental congurations: 1. Envelope ames (jet, Bunsen and other axisymmetric congurations) 2. Oblique ames (V-shaped, blu-body stabilized) 3. Flat ames (low-swirl, counter ow, diusion) 4. Spherical ames (including fan-stirred chambers) Canonical congurations attempt to provide an experimental setup that can be well-characterized in order to aid modeling eorts but can also be duplicated by others in the community to provide ad- ditional support to the scientic problem. Each conguration has its advantages and disadvantages to studying turbulent combustion phenomena. 14 Chapter 3 Experimental Methodology 3.1 Experimental Congurations 3.1.1 Atmospheric Piloted Premixed Jet Burner (PPJB) The current work utilizes a modied piloted premixed jet burner design as rst implemented by Bilger and colleagues at the University of Sydney [4]. The burner consists of three ames: the central jet ame, pilot ame and the co ow/shroud ame. The high velocity central jet contains the premixed reactants of interest. The pilot ame anchors the central jet to the burner exit in order to prevent blowo from the high shear present. The co ow ame surrounds both pilot and jet ame to provide a uniform temperature environment along the jet length to prevent quenching from outside (cold) air entrainment. A schematic of the burner conguration is shown in Fig. 3.1. Figure 3.1: Piloted Premixed Jet Jurner (PPJB) cutaway. All measurements are in mm. One of the main advantages of the PPJB is that the presence of the pilot and co ow allows the central jet to reach both high Reynolds number and high Karlovitz number conditions. The cong- 15 uration is also simple in design and has easy optical access. Lastly, the conguration is statistically stationary, so it is straightforward to obtain important turbulent oweld statistics. However, the presence of multiple streams could result in dilution eects, as the ame will mix with hot products from the co ow and pilot ames [35]. In addition, due to the high velocity of the central jet, the turbulence is shear-generated and thus not homogeneous and isotropic. However, shear-generated premixed ames are present in many practical situations, thus there is merit to studying properties of this class of ame. Driscoll argues that while homogeneous isotropic turbulence appears desirable, it has not been demonstrated that such conditions could be created nor that it would be of rele- vance to real environments [6]. Bilger and colleagues have studied the original PPJB conguration using temperature measurements, CH 2 O and OH PLIF to demonstrate that the conguration can produce ames in the thin reaction-broadened preheat zone regime. Therefore, this burner provides a strong framework to make a contribution to the knowledge base of fuel eects in the broadened preheat and broken reaction zone regimes. The main jet consists of a straight stainless steel tube with an internal diameter of D=5.84 mm and an external diameter of 6.35 mm. The main jet is surrounded by a pilot ow coming out of a co-annular tube with internal diameter of D pilot =22.9 mm and outer diameter of 24.1 mm. The nozzle edge of the jet is located 0.6mm above the pilot nozzle. A small step (0.5 mm thick) located 11 mm from the burner exit is used to hold a 5 mm thick brass perforated plate. This plate is composed of 52 holes (1.52 mm in diameter) and serves to stabilize the premixed pilot ame. The pilot tube has an unburnt exit velocity of 0.75 m/s before the step. The pilot tube is 70 mm in length and externally conical in shape with a 32.4 mm diameter base. The pilot ame is a premixed ethylene-air ame with an equivalence ratio =0.8. Finally, the outermost co-axial co ow uses hot products to thermally insulate the jet. A fuel mixture of 20% hydrogen and 80% ethylene (by mole) with air at = 0.51 was used to provide a temperature of T co ow = 1600 K. The co ow ames are stabilized on a perforated plate (outer diameter of 203 mm) with 1800 x 1.52 mm holes. The cold ow exit velocity before the plate is 0.7 m/s. All three streams are at the controlled laboratory air temperatures of 298K before being subsequently burned. The burner rig also utilizes a liquid fuel vaporization system that was developed over the years, initially for laminar ame investigations (e.g., [16, 17]). The current system allows for much higher owrates and is capable of vaporizing up to 20 mL/min of liquid fuel and air ow rates of up to 400 SLPM [23]. The system has been successfully utilized for liquid fuels of carbon number up to C 12 [36, 23]. This is a highly unique capability to a canonical turbulent premixed ame conguration and enables the following experimental studies to consider light gaseous fuels such as methane in addition to heavy liquid fuels such as dodecane. 16 3.1.2 Variable Pressure Burner Conguration Variable Pressure Chamber (VPC) Figure 3.2: USC variable pressure chamber The variable pressure chamber was present in the laboratory from a previous project in 2010 but had never been tested at elevated pressures. Figure 3.2 depicts the existing variable pressure chamber. It is 686 mm in height and 508 mm in diameter. It is constructed from 304 stainless steel, and is professionally certied to withstand wall temperatures up to 100 C and pressures up to 10 atm. The chamber has four Plexiglas windows around the chamber body that are each 180 mm in diameter. This provides excellent optical access to the proposed burner congurations and allows for the implementation of simultaneous laser diagnostics such as PIV and PLIF. The chamber exhaust is provided by four symmetric outlet ports in the upper body of the chamber. The chamber is cooled using dilution air provided below the outlets of the chamber and a water- cooled copper plate against the upper lid of the chamber. Chamber status is monitored using an electronic pressure gauge and three thermocouples placed to gauge thermal loading on critical chamber components. The chamber was pressure-tested to 4 atm and leak-checked before any combustion experiments were attempted. Variable Pressure Piloted Premixed Jet Burner (VP-PPJB) The variable pressure piloted premixed jet burner is a slightly modied version of the atmospheric PPJB: Figure 3.3 shows the 1D burner schematic. 17 Figure 3.3: VP-PPJB cutaway. All measurements are in mm. The main jet and pilot structure is unchanged from the atmospheric PPJB. The pilot ame is a premixed ethylene-air ame with an equivalence ratio of 1.0. This equivalence ratio is higher than the atmospheric conguration in order to stabilize the ame at constant mass ow conditions for elevated pressures. The co-axial co ow structure has been changed in several areas. First, the co ow is designed to only provide cold air ow as the pressure chamber cannot handle the thermal loading that would be produced in a ame-holding co ow. Therefore, the co ow structure does not include water- cooling infrastructure. The diameter of the co ow is also reduced due to the space constraints of the chamber. 3.2 Diagnostic Techniques 3.2.1 Diagnostic Setup The general diagnostics conguration is shown below: 3.2.2 CH* Chemiluminescence Chemiluminescence is often used for monitoring the presence of chemical reaction [37] and, under some conditions, the magnitude of the emitted light at particular wavelengths may be used as a semi-quantitative measure of the heat release rate, especially for fully premixed systems [38, 39]. In laminar ames, the ame chemiluminescence signal is a relatively good marker of the location of the maximum heat release rate [40]. Though there is some ambiguity regarding the interpreta- tion of chemiluminescence measurements in turbulent ames (due to complicated dependencies on curvature and ow history), the presence of ame chemiluminescence indicate vigorous burning. To determine global CH* behavior, 200 images of the averaged CH* luminosity were captured using an optical lter centered at 434 nm with a bandwith of 17 nm to capture the primary emittance band from CH* (431 nm). The images were taken using an Andor Zyla 5.5 camera with an exposure 18 Figure 3.4: Schematic of implemented diagnostics time of 100 ms. The background luminosity of the co ow and pilot were subtracted and the resulting images averaged together. The CH* images were used to calculate the turbulent ame height (H ). Flame Height (H ) Flame height derived from chemiluminescence (H ) is commonly used in jet and Bunsen burner congurations to compare between ames. Flame heights are calculated from the integrated line- of-sight CH* as Abel-inverting CH* images distorts the centerline data required for H calculation. Comparison against H derived from both line-of-sight integrated and planar CH* concentrations from LES of the PPJB showed no signicant dierence. Figure 3.5: H calculation from CH* centerline luminosity Figure 3.5 shows a typical ame centerline prole to illustrate H determination. H is dened as the location where the centerline prole of the ame brush chemiluminescence drops to a quarter of its maximum value. 19 3.2.3 Particle Image Velocimetry (PIV) Particle Image Velocimetry (PIV) is an optical technique that enables accurate resolution of a ow eld in time and/or space. PIV has been widely used in turbulent ows to characterize oweld behavior. Initially, the oweld is seeded with re ective tracer particles that are small enough to follow the ow. A thin laser sheet is then used to illuminate the tracer particles in an area of interest that is followed by a camera. The laser then res pulses at two discrete sequential intervals to produce two sequential images in time. The movement of the oweld can then be determined by correlating the movement of the particle eld between the two camera images. The two camera frames are divided into a number of interrogation windows. Using signal processing and cross- correlation techniques, a single velocity vector can be calculated for each window. In this manner, velocity vectors can be calculated for the area of interest. PIV provides advantages over other velocimetry techniques because it is a non-intrusive mea- surement and enables much greater ow resolution than point measurement techniques like probe or hotwire anemometry. This improved resolution allows for the visualization of vortical structures that may interact with the ame as well as the instantaneous wrinkling of the ame surface. How- ever, care must be taken to minimize the uncertainty in PIV measurements, such as particle dropout due to beam misalignment, 3D averaging over beam width, peak locking errors due to particle size and low particle density. To minimize these errors, the laser beam was carefully aligned, and the thickness and overlap of the beams was measured using burn paper. Given the small radial velocities of the jet (on the order of 4% of the exit velocity), particle dropout was considered to be negligible for the beam thickness used. Test images were taken before each data to ensure adequate particle and correlation could be achieved. In addition to the errors mentioned above, PIV in combusting ows can experience errors from the thermophoretic force on the seeding particles in the presence of thermal gradients as well as uncertainty in the index of refraction due to temperature uctua- tions [41]. A thorough quantication of the errors from the fast uctuations in refractive index from a propane gas ame show a systematic error of less than 5% of the exit velocity [42]. Stella et al. also concluded that refractive index variations are negligible on a laboratory scale. Thermophoretic forces are present however, which are proportional to particle diameter and temperature gradient. For small particles of 0.3 micron in size, the velocity lag measured in a ame was on the order of 0.3 m/s, which is small in comparison to the velocities experienced by the jet ame. Therefore these errors are neglected and rolled into an overall uncertainty. Green light at 532 nm and 25 Hz was generated using a high-power Q-switched laser (80 mJ/pulse) and expanded using plano-cylindrical lenses into a thin sheet through the jet center- line. The jet ow was seeded with 0.3-0.5 m aluminum oxide particles that survive through the ame. For the completed studies discussed in this paper, 1000 statistically independent coupled Mie scattering images were taken with an Andor Zyla 5.5 camera with an F28.5 mm or an F85 mm lens and a laserline lter to reject the ame luminosity signal. The inter-frame time was chosen to produce a maximum pixel shift of 10 pixels for all PIV studies (determined to be the optimum pixel 20 shift from prior experiments [43] and thus varied between 1.7 and 15s. Generation of velocity eld quantities is done using Davis Flowmaster software (LaVision Research Inc). The spatial resolution and interrogation window sizing for the full-eld velocity eld determination is detailed in the table below. Measurement Full-Field Studies Spatial Resolution (pixel/mm) 103 microns/pixel Interrogation Window Sizing 64x64 pixels decreasing to 16x16 with 50% window overlap Table 3.1: Spatial Resolution and PIV Processing Specications Vectors with a peak ratio of less than 1.05 were removed, and the universal outlier detection method was also used for spurious vector removal. The algorithm for processing the PIV vector elds is a modication of the correlation image velocimetry technique described by Fincham and Spedding [44]. The PIV uncertainty of mean ow properties of the central jet is dependent on the jet velocity U jet and is estimated to be0.05U jet . This error is due to the fact that velocity vectors are not well resolved below a pixel displacement of 0.5 pixels, as well as the additional sources of uncertainty discussed above. 3.2.4 Formaldehyde (CH 2 O) PLIF Planar Laser Induced Fluorescence (PLIF) is one of the most widely used and versatile diagnostic techniques in turbulent combustion. Laser-induced uorescence uses a high energy laser pulse at a specic wavelength to energize a selected molecule and cause it to release that energy as light [45]. In combustion studies, it is very desirable to observe the chemical structure of the ow and correlate species behavior to ame response. PLIF can provide an instantaneous 2D measurement of the local chemical structure. OH, formaldehyde (CH 2 O) and CH are the most widely accessible and used molecules in PLIF, though development is being conducted to observe HCO, CO and CO 2 . CH 2 O is formed in the preheat zone through the breakdown of the methyl radical (CH 3 ). As temperatures increase towards the reaction zone, CH 2 O is consumed through reaction with OH. Therefore CH 2 O concentration can be used as a reasonable indicator of the preheat zone [45]. In the turbulent combustion community, CH 2 O PLIF is used to observe preheat zone behavior and validate theories of ame response to extreme turbulence. In ames with complex and spatially varying structures, the direct interpretation of the presence of CH 2 O to a precise measurement of the preheat zone is not always possible. CH 2 O PLIF utilizes a laser beam tuned to 355 nm to excite formaldehyde to produce uores- cence. With CH 2 O has several excitation wavelengths, the 355 nm band was chosen because high excitation pulse energies can be obtained using an Nd-YAG laser. Similar to PIV, the laser beam is expanded to a sheet so that the resulting CH 2 O uorescence can be imaged over an area of interest. 21 However, unlike PIV, only a single laser pulse is required, allowing for high pulse energies needed to induce uorescence. For accurate spatial resolution, PLIF sheet thicknesses are below 1 mm in width. The CH 2 O uorescence that results from excitation at 355 nm occurs over the broadband range in the visible spectrum (450 to 600 nm). However, this signal is subject to competing signals from polycyclic aromatic hydrocarbons (PAH), chemiluminescence, soot radiation and cycloaro- matic/large hydrocarbons. Therefore care is required to eliminate these competing signals through the judicious use of optical lters, exposure time and operating conditions. Optical lters enable the camera capture system to isolate wavelength of bands of particular interest, eliminating signal contributions from other chemical species. To further isolate the CH 2 O uorescence, an image in- tensier is required. The CH 2 O uorescence lifetime is very short (< 200 s) compared to typical camera exposure times as well as weaker than the chemiluminescence signal. The use of an image intensier enables the camera to only capture the exposure of the uorescence signal. In this setup, blue light at 355 nm and 10 Hz was generated using a high-powered Q-switch laser (Continuum Surelite III) and expanded using plano-cylindrical lens into a thin sheet through the jet centerline. The CH 2 O uorescence was captured using an Andor Zyla 5.5 camera coupled with an image intensier (Lambert). Two colored glass lters (GG385 and BG-3) were used to focus on the primary band of CH 2 O uorescence from 385 nm to 435 nm. Laser energy per pulse was approximately 135 mJ, and the sheet thickness for the PLIF measurements was approximately 0.5 mm. 3.2.5 Simultaneous CH 2 O PLIF - 2-D PIV CH 2 O PLIF can not only be used to examine the preheat zone in a turbulent ame, it can also be an indicator of the location of the reaction zone. In turbulent ames, the location of maximum heat release is most often identied using the product of CH 2 O and OH. This conclusion was validated using laminar premixed simulations as well as vortex- ame experiments that were able to correlate CH 2 O x OH with the presence of HCO, a key marker of heat release [46]. Recently, Wabel et al. [11] demonstrated that the spatial derivative of the CH 2 O signal could be used as an acceptable indicator of the location of maximum heat release. This was independently veried using laminar ame calculations for a lean premixed CH 4 ame: Figure 3.6 shows the ame structure, focusing on the key chemical species CH 2 O, OH, the product of CH 2 O and OH, heat release rate _ q and the absolute value of the spatial derivative of CH 2 O. This method presents an attractive alternative to determine the presence of the ame in the absence of complimentary OH-PLIF. This signicant reduction in complexity facilitates the coupling of CH 2 O-PLIF and PIV in order to identify the ame surface and subsequently determine which velocity vectors are before the ame and after the ame. This is not possible using only the Mie scattering signal and the particle density shift present as a result of the ame, as the particle density is high enough post- ame to prevent 22 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22 0 0.2 0.4 0.6 0.8 1 X Axis (cm) Normalized Values CH 2 O OH CH 2 O x OH ˙ q | d[CH 2 O]/dx | Figure 3.6: 1-D calculated laminar ame structure for a lean premixed CH 4 -air ame . automated identication of the ame surface. Figure 3.7 shows a comparison between a sample Mie scattering image and a CH 2 O-PLIF image. The CH 2 O-PLIF image provides a clear edge that can be used to determine the ame surface. (a) Mie scattering (b) CH 2 O Figure 3.7: Sample Mie scattering and CH 2 O signal for a Jet-A/air ame, S L =15.3 cm/s, Re jet =25,000. The importance of this determination will be discussed later in Chapter 8. For the simultaneous CH 2 O PLIF and 2-D PIV studies, the PLIF and PIV laser beams were aligned concentrically, and the alignment was veried both before and after the selected imaging area. The PLIF image was captured in the inter-frame time between the two PIV laser pulses. 840 statistically independent image-pairs were captured for each dataset. The ame surface was determined by binarizing the CH 2 O-PLIF signal at the maximum gradient of the intensity. The resolutions for the PLIF and PIV images are shown in the table below, with the PIV processing fol- lowing the interrogation window sizing stated below and postprocessing as discussed in Section 3.2.3 23 Measurement CH 2 O-PLIF PIV Spatial Resolution (pixel/mm) 13.8 microns/pixel 16 microns/pixel Interrogation Window Sizing 64x64 pixels decreasing to 12x12 with 50% window over- lap Table 3.2: Spatial Resolution and PIV Processing Specications 3.2.6 Laser Absoprtion Tomography Laser absorption spectroscopy (LAS) is an experimental technique that can provide quantitative, calibration-free method of measuring major combustion species, particularly CO and CO 2 , and temperatures using compact low-power semiconductor mid-infared (IR) lasers [47]. Though weaker in spatial resolution capability (due to line-of-sight integration), the mobility and compactness of state-of-the-art Interband Cascade Lasers (ICLs) and Quantum Cascade Lasers (QCLs) enable spatially resolved measurements in ames that can be combined with tomographic reconstruction techniques [48, 49] to characterize non-uniform ows [50, 51]. A partnership was undertaken with the Spearrin lab at UCLA to adapt their laser absorption spectroscopy techniques to provide time-averaged proles of CO, CO 2 and temperature in the PPJB. The methods are detailed thoroughly in the literature [47] and are beyond the scope of this PhD work, so they will only be brie y reviewed here in the context of the experimental setup. The compact lasers and detectors were mounted to a dual mechanical translation stage system as shown in Fig. 3.8, to characterize the time-averaged thermochemical structure of the ames. Beam Splitters PV Detectors CO 2 Laser (4193 nm) CO Laser (4979 nm) Focusing Mirrors PPJB Stepper Motor Horizontal Translational Stage Manual Vertical Translational Stage H 2 /air coflow Focusing Lens and Spectral Filter Figure 3.8: Hardware and representative data analysis for laser absorption tomography system [2] . The concentric laser beams were focused to beam diameters of approximately 0.5 mm, represent- 24 ing a limiting factor on the spatial resolution of the measurements. During a jet ame experiment, the assembly (and therefore, the beams) is translated radially via an automatic translation stage, and the encoder signals of the stepper motor are used to resolve the spatial location of the measure- ments in time. A manual vertical stage translates the entire assembly to repeat the measurements at dierent heights downstream of the jet exit. Overall, the radial spatial resolution of the mea- surements was 0.5 mm and the vertical spatial resolution was 20 mm. More information about the diagnostic setup and methods|including details about wavelength selection, spectral tting, tomographic reconstruction, and uncertainty analysis|are provided in previous work spearheaded by UCLA [2]. 25 Chapter 4 Numerical Methodology 4.1 PREMIX Code S L is a key parameter used to characterize the fuel/air mixtures investigated as discussed earlier. S L is computed using the PREMIX code, developed by Kee et al. [52]. The code is designed to model steady laminar 1-D premixed ames. Assuming a freely propagating, isobaric, adiabatic 1D ame front, PREMIX calculates the species and temperature distribution in the ame front. It integrates with the CHEMKIN II and TRANSPORT subroutine packages to provide detailed chemistry, thermodynamic and transport databases. The governing equations used are given below. d _ m dt = 0 (4.1.1) _ m dY k dt + d(AV k Y k ) dx A _ w k W k = 0 (4.1.2) _ m dT dx + A c p K X k=1 V k Y k c p;k dT dx 1 c p d dx (A dT dx ) + A c p K X k=1 _ w k h k W k = 0 (4.1.3) The solution is found using a Newton solver and iterating towards the solution using a time- stepping algorithm. 4.2 Unsteady Opposed Jet Code (Unsteady OPPDIFF) Unsteady counter ow simulations provide a useful canonical conguration to study extinction rel- evant to the highly strained, unsteady nature of turbulence. Unsteadiness was introduced for the reactant velocity by imposing xed sinusoidal variations of a given amplitude around the mean exit velocity [53]. The opposed jet code is based o of the OPPDIFF FORTRAN routine [52]. Changing the frequency of the velocity variations produces three dierent extinction regimes [53, 54, 55] as shown in Fig. 4.1. At low frequencies (200 Hz), the initial increase in strain rate will cause a quasi-steady extinc- tion. At high frequencies (600 Hz), ame extinction is suppressed as the ame is unresponsive to 26 Figure 4.1: Unsteady counter ow frequency regimes strain rate forcing (e.g., [53]). At intermediate frequencies however (406, 412 Hz), ames exhibit an unsteady extinction response, characterized by a progression towards extinction, a recovery period and a nal extinction process. The frequency required for unsteady oscillation is dependent on the magnitude of the extinction strain rate (K ext ) as well as the magnitude of the velocity uctua- tion [56]; all reported strain rates (K) correspond to the maximum magnitude of the axial gradient of the axial velocity prole in the hydrodynamic zone (e.g., [53]). 4.3 Reynolds-Averaged Navier Stokes (RANS) Calculations Steady-state Reynolds-Averaged Navier Stokes (RANS) were performed in order to model the steady state behavior of the PPJB conguration. RANS provides quantities corresponding to averages over time for statistically stationary mean ows, making it well suited for application to a turbulent jet as well as a good counterpoint to the line-of sight averaged experimental measurements [57]. The ame was modeled with the open source Computational Fluid Dynamics (CFD) solver OpenFOAM using the solver reactingFOAM which is used for statistically steady-state combustion. The combustion was modeled assuming a partially-stirred reaction model which denes a transient reaction rate dependent on a hybrid of the turbulent and chemical mixing time scales, assuming only a portion of the mixed fraction reacts. 4.3.1 Governing Equations Assuming Reynolds-averaged quantities, the Reynolds-averaged Navier-Stokes equation is: U j @U i @x j = 1 @P @x i + @ @x j ( @U i @x j + @(u 0 i u 0 j ) @x j (4.3.1) The simulations use second-order accurate spatial discretization and rst-order accurate tempo- ral discretization. 27 4.4 Kinetic Models The kinetic models used with PREMIX and the Opposed Jet Code were USC-Mech II [58] for the CH 4 ames and JetSurf 2.0 [59] for the n-C 12 H 26 ames. Both models have been extensively validated against propagation and extinction data for laminar ames and include the CH*/OH* model developed by Nori and Seitzman [60]. Though the model has not been validated under extinction conditions, it has been validated against a similar range of fuels and pressures as utilized in this paper [61, 62, 63]. The RANS calculations used a 30 species reduced model based o the USC Mech II mechanism [58]. 28 Chapter 5 Eects of Heat Release and Fuel Type on Global Observables 5.1 Introduction One of the primary goals of the USC PPJB experimental conguration is to explore the global properties and local structure of piloted premixed turbulent ames. The preliminary characteriza- tions of the USC PPJB, with regards to fuel eects, by Carbone and Smolke [23, 36, 43] examined the ame chemiluminescence and the behavior of the mean and uctuating velocity components. These studies brie y explored the potential of laminar scaling parameters such as the laminar ame speed and the adiabatic ame temperature to scale the observables. Only the laminar ame speed showed reasonable success to scale the ame height derived from CH* chemiluminescence. This study seeks to further those previous investigations into candidate scaling parameters for various fuels and experimental observables. 5.2 Experimental Parameter Space Preliminary experiments were performed at two jet Reynolds numbers, Re jet u jet D= = 25,000 and 50,000, where U jet is the bulk ow velocity and is the kinematic viscosity at the burner exit. Mixtures of air with four fuels, namely methane (CH 4 ), ethylene (C 2 H 4 ), n-heptane (nC 7 H 16 ), and toluene (C 7 H 8 ), were used at=0.5, 0.6, 0.7, 0.9, and 1.0 at an unburned mixture temperature of 298 K. This range of and fuels allowed for a wide spread in S L from 3 cm/s to 68 cm/s. Table 7.1 provides a summary of the conditions investigated. S L as well as the ame thickness ( f ) and the ame time ( f ) were calculated using the PREMIX code [52]. Re t u 0 L int = was measured as in previous studies [36, 23]. It should be noted that although u 0 does change with , L int shows only a minor dependence on and fuel type [23]. As characterized by the high u 0 and Ka, these ames are expected to fall either in the thin reaction zone regime or in the broken reaction zone regime. 29 u jet (m/s) Re jet u 0 SL t = u 0 Lint Lint f = ( Lint u 03 ) 1=2 Ka Da = t f Re t 68.5 25000 12 - 274 0.5 2.5 - 21.9 0.0168 24 - 3998 0.01 - 2.67 3026 133 50000 23 - 532 0.28 2.3 - 20.4 0.006 67 - 11195 0.005 - 1.28 5483 Table 5.1: Estimated range of key turbulence characteristics for the performed experiments. All values were calculated based on the kinematic viscosity of the unburned mixture, but the turbulent properties of the ow L int and u 0 were measured in the shear layer at x=D=15. 5.3 Global CH* Observables The main goal of this study is to further previous investigations in fuel eects [23, 36] on highly tur- bulent turbulent premixed ames. More specically, the focus was on candidate scaling parameters for various fuels and experimental observables. Though S L has previously been used to examine fuel scalings, it is desirable to also examine other scalings derived from laminar ame properties. In addition to S L , _ q max , and f were used to scale H between the four fuels. 5.3.1 Flame Height (H ) The result of these scalings is rst considered using H . Figure 5.1 depicts the results of these scalings. The uncertainty in these measurements is1D, estimated from experimental repeatability. 5 10 15 20 25 30 35 40 25 30 35 40 45 S L (cm/s) H fl /D CH 4 C 2 H 4 n− C 7 H 16 C 7 H 8 (a) S L 0 2,000 4,000 6,000 25 30 35 40 45 ˙ q max (J/(cm 3 ·s)) H fl /D CH 4 C 2 H 4 n− C 7 H 16 C 7 H 8 (b) _ q max 0 0.5 1 1.5 2 2.5 25 30 35 40 45 δ f (mm) H fl /D CH 4 C 2 H 4 n− C 7 H 16 C 7 H 8 (c) f Figure 5.1: H variation with S L , _ q max , and f at Re jet =25,000. H initially decreases sharply with increasing S L and _ q max , and subsequently, the reduction in H is notably milder. It is evident that none of the scalings presented are completely sucient to suppress the dierences between the fuels. Higher S L (lower f ) does scale H for methane, n- heptane, and toluene, though ethylene remains unique in all scalings. Similar behavior was observed at Re jet =50,000. 30 5.4 Characteristics of the Shear Layer: u 0 Layer Thickness Though global ame observables based on CH* measurements are useful variables for model valida- tion and insight into fuel eects, velocity eld statistics provide a complimentary perspective. PIV measurements were made to characterize fuel eects on the local and global uid mechanics char- acteristics of the jet. The mean and uctuating velocity prole behavior have been examined in a previous study [23] and will be discussed in Section 5.5. This study endeavors to explore a dierent aspect of the shear-generated turbulence, namely the thickness of the shear layer as dened by u 0 ( u 0). Shear layer growth in the presence of heat release has been studied extensively, as knowl- edge of the ow processes responsible for mixing is crucial to understanding turbulent combustion [64, 65, 66]. First, the response of the shear layer growth to heat release will be explored. 5.4.1 Heat Release Eects u 0 was calculated as the horizontal full-width three-quarters max (FW3QM) of theu 0 distribution in the two shear layers. The FW3QM was chosen to allow determination of the layer thickness further along the jet height, as the full-width half max (FWHM) was not sucient due to higher turbulence intensity in the jet center. Both the FWHM and FW3QM were tested lower in the jet to conrm that the same behavior was observed. The calculated layer thicknesses are then averaged between the two shear layers in order to reduce measurement noise. u 0 was calculated for all the experimental conditions, though only the Re jet =25,000 case will be discussed here. Due to the merging of the shear layers at higherx=D, u 0 was calculated fromx=D=3 tox=D=25; measurements forx=D< 3 were not reliable due to low seeding density in the pilot ame. Figure 5.2 shows the response of u 0 for toluene ames to changing heat release _ q max through changing . Toluene was chosen as a fuel that has not been previously examined in these types of studies. Other fuels exhibited a similar response to changing _ q max . 5 10 15 20 25 2 4 6 8 10 12 14 x/D δ u 0 (mm) φ = 0.50 φ = 0.60 φ = 0.70 φ = 0.90 φ = 1.00 Figure 5.2: u 0 variation with x=D and _ q max for toluene ames at Re jet =25,000. Early on in the jet, no dierences are seen with _ q max due to the dominant eect of the pilot 31 ame. As the jet develops though, u 0 begins to grow dierently depending on _ q max . The change in from 0.5 to 1.0 results in a change of two orders of magnitude for _ q max . The lowest _ q max case has the largest u 0, and increasing _ q max causes a subsequent reduction in u 0. This reduction in u 0 with increased _ q max has been observed also by McMurtry et al. [65], who suggested that the reduction of shear thickness with increased heat release was caused by reduced mass entrainment rates. The linearity in the axial growth of u 0 has been reported also by Hermanson and Dimotakis [64] who conrmed a decrease in layer width with increasing heat release under moderate ow temperatures up to 940 K. The present study extends the worthwhile ndings of Hermanson and Dimotakis [64] and McMurtry et al. [65] to greater temperatures (1400 K to 2400 K) and conrms that they are still valid in the presence of intense heat release. Hermanson and Dimotakis [64] also posited that the increasing change in density with greater heat release was responsible for the decrease in u 0. Thring and Newby [67] rst proposed the momentum diameterd to account for the density dierences in a turbulent jet between the jet and ambient uid, which was expanded by Ricou and Spalding [68] in the case of isothermal turbulent jets. Tacina and Dahm [69] utilized the equivalence principle to extend the momentum diameter scaling in the case of heat release for turbulent jet diusion ames, though only for hydrogen and methane ames. This classical scaling was tested in the case of a premixed turbulent jet for a range of fuel types. d was calculated as: d = ( 0 = 1 ) (1=2) D (5.4.1) where 1 is the unburned jet uid density and 0 is the burned jet uid density. 5 10 15 20 25 1 2 3 4 5 x/d ∗ δ u 0 (mm) φ = 0.50 φ = 0.60 φ = 0.70 φ = 0.90 φ = 1.00 Figure 5.3: u 0 variation with x=d for toluene ames at Re jet =25,000. Figure 5.3 depicts the axial variation of u 0 usingd as the scaling parameter of the axial distance. The results indicate this scaling is appropriate in accounting for the eect of heat release on the growth of the layer thickness. Thus, it can be concluded that the reduction in shear layer thickness with increasing heat release is signicantly in uenced by the density dierences between the burned and unburned states. 32 5.4.2 Fuel Eects The eect of fuel type on u 0 was also examined. As S L and _ q max were seen as promising scaling parameters in Section 5.3.1, they were utilized here. Using the linear dependence of u 0 with axial distance, the u 0 growth rate, dened as d u 0=dx, was determined for each tested and fuel at Re jet = 25,000. The calculated d u 0=dx are plotted against _ q max and S L in Fig. 5.4. 0 2,000 4,000 6,000 0.04 0.05 0.06 0.07 0.08 0.09 ˙ q max (J/(cm 3 ·s)) dδ u 0/dx (mm/mm) CH 4 C 2 H 4 n−C 7 H 16 C 7 H 8 (a) d u 0=dx vs. _ q max 5 10 15 20 25 30 35 40 0.04 0.05 0.06 0.07 0.08 0.09 S L dδ u 0/dx (mm/mm) CH 4 C 2 H 4 n−C 7 H 16 C 7 H 8 (b) d u 0=dx vs. S L Figure 5.4: Fuel eects in d u 0=dx with changing _ q max and S L . Re jet =25,000. From Fig. 5.4a, it is clear that d u 0=dx decreases with increasing _ q max , as also demonstrated in section 5.4.1. However, this decrease in d u 0=dx is not the same for each fuel. Methane, n-heptane, and toluene exhibit similar decreases ind u 0=dx across the ranges of _ q max . Ethylene has consistently lower values of d u 0=dx. For _ q max > 3000, the fuels approach similar values of d u 0=dx. As _ q max decreases, ethylene begins to exhibit lower values of d u 0=dx than the other fuels for a similar _ q max . The dierences in d u 0=dx behavior are also apparent when scaling against S L in Fig. 5.4b. Though methane and toluene show the same decrease in d u 0=dx for increasingS L , both n-heptane and ethylene deviate for lowerS L . These trends are consistent with the observations onH . Though density played a role in the behavior of the u 0, the dierences in both burned and unburned density among the fuels is very small. The observed behavior therefore indicates that although heat release and density dierentials in uence the behavior of the shear layers, the fuel chemistry also appears to play a role in the shear layer structure. 5.5 Turbulent Kinetic Energy and Turbulent Shear Stress in the Shear Layer Flame eects on TKE have been explored in previous studies (see for example [70]), as the interaction between turbulence and combustion can be expressed by the balance between TKE production and destruction [71]. In a previous study [23], it was shown thatu jet andu 0 are only marginally aected by the fuel type whenS L was kept constant. However, investigations into fuel eects on higher order moments such as TKE and the turbulent shear stress (TSS) have not been conducted. A better 33 understanding of possible fuel eects would assist the turbulent combustion modeling community in validating closure models for ames at high Reynolds and Karlovitz numbers. TKE and TSS was calculated through the Davis software as: TKE = 2 X i=1 3 4 (V i V i ) 2 (5.5.1) TSS = ((Rey xx Rey rr ) 2 + (Rey rx ) 2 ) 1=2 (5.5.2) where R uv = 1 N1 P N i=1 (u i u)(v i v) where u and v are the respective spatial directions. The TKE calculation assumes that an invisible 3 rd component V z contains a similar turbulent contribution as the cumulative values calculated for V r and V x . The maximum value of TKE and TSS (located in the shear layer) was calculated along the axial length of the jet. As discussed earlier in this chapter, in order to assess fuel eects, matchingS L (13 cm/s) and _ q max (63J=(cm 3 s) cases were considered. These matching conditions were chosen to be close to those examined by Smolke et al. [36]. It also should be noted that at these lean conditions, S L scales with _ q max for all fuels. MatchingS L instead of _ q max did not change the conclusions that follow. Figure 5.5 shows the axial development of TKE and TSS for the matching _ q max case at Re jet =25,000. 5 10 15 20 25 30 35 20 25 30 35 40 45 x/D Turbulent Shear Stress (m/s) 2 CH 4 C 2 H 4 n−C 7 H 16 C 7 H 8 (a) TSS, Re jet =25,000 5 10 15 20 25 30 35 40 50 60 70 80 90 100 x/D Turbulent Kinetic Energy (m/s) 2 CH 4 C 2 H 4 n−C 7 H 16 C 7 H 8 (b) TKE, Re jet =25,000 5 10 15 20 25 30 35 40 60 80 100 120 140 x/D Turbulent Shear Stress (m/s) 2 CH 4 C 2 H 4 n−C 7 H 16 C 7 H 8 (c) TSS, Re jet =50,000 5 10 15 20 25 30 35 40 150 200 250 300 x/D Turbulent Kinetic Energy (m/s) 2 CH 4 C 2 H 4 n−C 7 H 16 C 7 H 8 (d) TKE, Re jet =50,000 Figure 5.5: TKE and TSS variation with x=D for matching _ q max . Below x=D=10, all fuels exhibit the same linear increase in both TSS (Fig. 5.5a) and TKE (Fig. 5.5b) again due to the dominant in uence of the pilot at lower x=D. Afterx=D=10, the TSS 34 and TKE magnitudes experience a steady period before the turbulence begins to decay. Fuel eects become apparent after the initial growth period. The results for ethylene, n-heptane, and toluene ames scale closely with _ q max , however methane ames exhibit a sharper turbulence decay after the growth period. This discrepancy is also seen in the decay of TKE. At Re jet =50,000, the dierences in TKE and TSS between the fuels are more pronounced, as seen in Figs. 5.5c and 5.5d. As with the Re jet =25,000 case, the maximum TKE for methane ames decreases more quickly than the other fuels. Though the n-heptane and toluene ames exhibit similar decay rates for TSS and TKE, the maximum TKE for ethylene ames is higher than all other fuels. Qualitatively, this re ects the behavior of H as the ethylene ames, with the shortest ames (an indicator of stronger burning) , have the highest TKE and TSS production while methane ames, with the longest H , have the lowest TKE and TSS production. 5.6 Concluding Remarks The present study examines the eects of heat release and fuel type on piloted premixed turbulent jet ames. The ame height was scaled with laminar ame speed, maximum heat release rate, and ame thickness for all fuels at Re jet = 25,000 and 50,000. None of the scalings were found to be adequate to suppress fuel eects at all conditions, though dierences were less pronounced for stronger burning ames. The shear layer thickness was measured using the distribution of the velocity uctuation in the shear layers and was used to conrm the in uence of heat release. The classical momentum diameter was used to scale the eects of changing heat release, as the shear layer thickness is driven by the density dierences between burned and unburned states. The shear layer thickness was found to exhibit similar trends between the fuels for high levels of heat release, but the slope of the shear layer thickness growth does not scale well between the fuels for lower heat release rates. Scaling with the laminar ame speed also did not suppress fuel dierences in the layer thickness growth slope. A more detailed examination of the shear layers was conducted using the turbulent kinetic energy and turbulent shear stress for matching maximum heat release rates and laminar ame speeds. Though all fuels have a similar production period for turbulent kinetic energy and turbulent shear stress atRe jet = 25,000, methane ames exhibit a steeper decrease than the other fuels. This discrepancy was found to be more evident for the higher Reynolds number. This is consistent with the behavior of the scaled ame height, as ethylene, with the shortest ame height, exhibited the greatest TKE and TSS production. 35 Chapter 6 Preliminary Investigations of Global Flame Topology at Elevated Pressures 6.1 Introduction Experimental data regarding combustion phenomena in high-pressure environments is particularly limited because of the diculty of the experimental procedures. The development of USC's variable pressure facility enables the exploration of the impact of the thermodynamic state, namely pressure, in the context of fuel specic eects. 6.2 Experimental Parameter Space The rst objective of the new variable pressure chamber was to establish the parameter space for initial studies. The four general space parameters are listed below. 1. Pressure (P ) 2. Equivalence ratio ( 1.0) 3. Fuel type (C 1 - C 3 neat compounds) 4. Turbulence (V jet , Re jet , Re T ) The primary parameter is pressure. In these experiments, results at elevated pressures will be discussed rst, as due to experimental constraints, elevated pressures were the most easily accessible. The next parameter is the equivalence ratio: these experiments will focus on on 1.0, as the accumulation of unburnt fuel in the chamber is a safety hazard. Thirdly, the fuel type explored will focus on C 1 - C 3 neat compounds. This will include three alkanes (methane, ethane and propane) and two alkenes (ethylene and propene), as previous studies indicated unusual behavior regarding the scaling of ethylene with n-alkane fuels. A vaporization system is in development which will enable the study of liquid fuels at non-atmospheric pressures. 36 The nal parameter is the nature of the turbulence scaling. As discussed in Section 2.2.3, it is not clear which turbulence parameters to hold constant when comparing to atmospheric results. Matching the Reynolds number at elevated pressures requires changing the ow bulk velocity which, for a turbulent jet ame, will change the u 0 . Multiple experimental studies (ex. [25, 27]) have ap- proached the problem by keeping the bulk velocity constant, which assumes that the relationship between the bulk velocity and the u 0 is not pressure-dependent (which Kobayashi showed is not necessarily the case). For a preliminary examination, the experimental conditions will use a con- stant Reynolds number between dierent pressures. Future studies will pursue constant velocity conditions. 6.3 Experimental Observables 2-D averaged, line-of-sight integrated CH* will be used to derive a number of global ame ob- servables. This rst includes the global ame shape as well as the ame height H (discussed in Section 3.2.2. The maximum CH* intensity along the centerline will be captured as well. Compli- mentary laminar 1-D ame calculations will be used as well to provide potential scaling parameters. The jet ame conditions for this study are summarized in the table below. V jet (m/s) P (atm) Re jet Fuel S L (cm/s) 68.5 1 25,000 CH 4 0.4 to 1.0 3 to 65 C 2 H 4 C 2 H 6 C 3 H 6 C 3 H 8 34.25 2 25,000 CH 4 0.4 to 1.0 3 to 55 C 2 H 4 C 2 H 6 C 3 H 6 C 3 H 8 22.83 3 25,000 CH 4 0.5 to 1.0 3 to 50 C 2 H 4 C 2 H 6 C 3 H 6 C 3 H 8 Table 6.1: Experimental parameter space for elevated pressure experiments The pilot was a stoichiometric ethylene-air ame at a Reynolds number of 1,135. The co ow was cold at air at a Reynolds number of 1,900. 37 6.4 Fuel and Pressure Eects on Global Flame Observables 6.4.1 Global Flame Shape The rst observable examined was the global ame shape derived from the averaged CH* measure- ments. The ame shape was observed to move between two primary shape states: tip-ignited and tip-quenched, as seen in Fig. 6.1. A sample set of ame images is also shown in Fig. 6.2. (a) 1 atm (b) 2 atm (c) 3 atm Figure 6.1: Global ame shape at 1, 2 and 3 atm demonstrating tip-quenched and tip-ignited states for Re jet =25,000 and S L =24 cm/s. In the tip-quenched state (shown in Figure 6.3a), the ame tip is not fully ignited. In the tip- ignited state (shown in Figure 6.3b), the ame tip is fully ignited, and the ame has the characteristic shape of a jet ame. In the atmospheric PPJB conguration, the co ow is utilized to keep ames in the tip-ignited regime.Previous velocity eld measurements observed that the survival of the jet centerline core velocity as a function of axial distance increased with increasing . The thermal expansion of the jet combined with the increasing heat provided at higher equivalence ratios acts to protect the jet centerline from mixing with the surrounding ambient air. Higher u 0 s have also been correlated with higher equivalence ratios, which may also serve to protect the jet core from quenching by the ambient air. 6.4.2 Flame Height (H ) The dierence in ame regime between tip-ignited and tip-quenched is clearly visible in H . Fig- ure 6.3 illustrates the dependence of H normalized by the jet diameter D as a function of S L . H /D exhibits a non-monotonic dependence on S L : H /D will increase withS L until a turning point is reached and then decrease as S L continues to increase. This turning point corresponds to 38 (a) CH 4 (b) C 2 H 4 (c) C 2 H 6 (d) C 3 H 6 (e) C 3 H 8 Figure 6.2: Global ame appearance as a function of fuel for Re jet =25,000, P =2 atm and S L =24 cm/s. the transition between the tip-ignited and tip-quenched regions and is labeled on the gures with a red dashed line. All fuels examined in this study exhibited a similar dependence on S L at 1 and 2 atm, though slight deviations are seen in the behavior of C 2 H 4 . This suggests that in the absence of the co ow, the tip-quenched region does not display a dependence on fuel as suggested by the tip-ignited studies of Chapter 5. H /D at 3 atm does not scale as well withS L . These ames however exhibited larger uctuations in ame height and intensity than the 1 and 2 atm cases, so these results are interpreted with caution. It is possible that at this higher pressure, acoustic eects due to the chamber geometry are aecting the ame. Future modications to the chamber are planned to modify its acoustic prole. WhileS L is an appropriate parameter to characterize ame behavior at a set pressure, increasing pressure aects both S L as well as the mixture density. For reaction orders < 2 (all conditions considered here have n < 2), increased pressure will reduce S L but increase the mixture density, thus resulting in an overall increase in the mass burning rate. Therefore, to compare ame quantities across dierent pressures with signicant changes in mixture density, the density-weighted laminar ame speed u S L should be used. The eect of pressure on H /D is explored more by examining the scaling of H /D vs. S L and u S L in Fig. ??. 6.4.3 Maximum CH* Intensity In addition to the ame height, the maximum CH* intensity of the ame can also inform the response of the ame to varying pressure. The maximum CH* intensity is determined as the maximum CH* intensity along the ame centerline. Complimentary laminar ame calculations to determine the 39 0 5 10 15 20 25 30 35 40 5 10 15 20 25 30 S L (cm/s) H fl /D CH 4 C 2 H 4 C 2 H 6 C 3 H 6 C 3 H 8 (a) 1 atm 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 S L (cm/s) H fl /D CH 4 C 2 H 4 C 2 H 6 C 3 H 6 C 3 H 8 (b) 2 atm 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 S L (cm/s) H fl /D CH 4 C 2 H 4 C 2 H 6 (c) 3 atm Figure 6.3: H /D as a function of S L at 1, 2, and 3 atm. predicted 1-D CH* mole fraction were also performed to compare against the experimental results. Though a quantitative comparison between CH* intensity and CH* mole fraction cannot be made, it can be assumed that an elevated CH* mole fraction would result in an increased CH* intensity captured in the experimental setup. The rst point of comparison was to examine the luminosity of the C 1 through C 3 hydrocarbons at 1 atm as shown in Fig. 6.4. The experimental and calculated results show close agreement in the observed CH* trends. As expected, CH* intensity increases exponentially with increasing S L . CH* is produced through the consumption of C 2 H, and the increase of reacting fuel concentration as the ame moves towards stoichiometric increases the availability of C 2 H to be excited to CH*. Similarly, CH* concentra- tion at the same ame conditions increases with increasing carbon number: CH 4 has the lowest concentrations and C 3 H 6 has the highest. The n-alkanes do not have a consistently higher CH* con- centration than the n-alkenes. The CH* mole fraction for C 2 H 4 (X CH;C2H4 ) is less thanX CH;C2H6 , but X CH;C3H6 is greater than X CH;C3H8 . A path analysis of the laminar ame calculations shows that the C 2 hydrocarbons form C 2 H from C 2 H 4 . The C 3 hydrocarbons exhibit an extension and shift in the formation path. C 3 H 8 produces C 2 H 4 from n-C 3 H 7 . C 3 H 6 however produces C 2 H 4 from C 3 H 6 directly as well as from C 2 H 5 . These additional precursor paths enable the increased production of the requisite C 2 H and thus CH*. Next, the eect of pressure on the CH* concentration was studied for CH 4 and C 2 H 4 ames. 40 0 10 20 30 40 50 0 500 1,000 1,500 2,000 2,500 3,000 S L (cm/s) Counts (A.U.) CH 4 C 2 H 4 C 2 H 6 C 3 H 6 C 3 H 8 (a) Maximum CH* Intensity, 1 atm 0 10 20 30 40 50 0 2·10 −10 4·10 −10 6·10 −10 8·10 −10 1·10 −9 S L (cm/s) CH* Mole Fraction CH 4 C 2 H 4 C 2 H 6 C 3 H 6 C 3 H 8 (b) Maximum CH* Mole Fraction, 1 atm Figure 6.4: Maximum experimental CH* intensity and calculated CH* mole fraction for examined C 1 through C 3 hydrocarbons at 1 atm. Figure 6.5 shows the maximum CH* intensity from the experiment and Fig. 6.6 shows the maximum calculated CH* mole fraction plotted against the density-weight ame speed. 0 0.02 0.04 0.06 0.08 0 500 1,000 1,500 ρ u S L (g/cm 2 s) Counts (A.U.) 1 atm 2 atm 3 atm (a) CH 4 0 0.03 0.06 0.09 0.12 0.15 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 ρ u S L (g/cm 2 s) Counts (A.U.) 1 atm 2 atm 3 atm (b) C 2 H 4 Figure 6.5: Maximum CH* intensity for CH 4 and C 2 H 4 at 1, 2, and 3 atm. As the pressure increases, the CH* concentration increases at higher u S L . This trend is clear for both CH 4 and C 2 H 4 ames. The reduction in CH* intensity at higher pressures suggests a reduction in reactivity. The observed trends are consistent between the experiments and simulations, which suggests that the ame chemistry is responsible for the reduction of reactivity, not a pressure- induced change in the turbulence parameters. 6.5 Concluding Remarks This chapter presents the introduction of the PPJB burner to a variable pressure facility. These studies represent the rst work on elevated pressure turbulent ames ever conducted in this lab. 41 0 0.02 0.04 0.06 0.08 0 2·10 −11 4·10 −11 6·10 −11 8·10 −11 1·10 −10 1.2·10 −10 ρ u S L (g/cm 2 s) CH* Mole Fraction 1 atm 2 atm 3 atm (a) CH 4 0 0.03 0.06 0.09 0.12 0.15 0.18 0 5·10 −10 1·10 −9 1.5·10 −9 2·10 −9 2.5·10 −9 ρ u S L (g/cm 2 s) CH* Mole Fraction 1 atm 2 atm 3 atm (b) C 2 H 4 Figure 6.6: Maximum calculated CH* mole fraction for CH 4 and C 2 H 4 at 1, 2, and 3 atm. The modied PPJB conguration was used to begin baseline studies of global observables as a function of fuel and pressure. Using averaged CH* chemiluminescence to calculate H , H for various fuels was found to scale moderately well withS L at 1 and 2 atm but less so at 3 atm. H /D is controlled by extinction and propagation eects with varying pressure, as seen in the tip-quenched and tip-ignited regimes. The experimental maximum CH* intensity and calculated PREMIX maximum CH* concen- tration show close agreement in the observed CH* trends. As expected, CH* intensity increases exponentially with increasingS L . IncreasingS L moves the ame towards stoichiometric conditions. CH* is produced through the consumption of C 2 H, and the increase of reacting fuel concentration increases the availability of C 2 H to be excited to CH*. Similarly, CH* concentration at the same ame conditions increases with increasing carbon number. 42 Chapter 7 Investigations of Fuel and Hydrodynamic Eects through Laser Absorption Spectroscopy (LAS) 7.1 Introduction As computational power increases, enabling higher delity simulations of turbulent ames with de- tailed chemistry, it is important to provide experimental measurements for model validation targets. Specically, quantitative measurements of the thermo-chemical structure, namely spatially resolved species and temperature, are needed for high Reynolds and Karlovitz number ames in the thin and broken reaction zone regimes. Traditional intrusive measurement techniques, such as gas chro- matographs and thermocouples, disturb the local ow eld which prevents a denitive interpretation of the results. Raman scattering methods have been used for point measurements of major species (N 2 , CO 2 for example) [72, 73], but this approach does not lend itself to the characterization of ame structure as it is dicult to produce multiple point measurements due to the size and power of the lasers required. Laser-induced uorescence (LIF) [74, 75] and chemiluminescence [36, 23] have been used extensively for species-specic imaging on an instantaneous and planar basis, but such methods do not easily yield quantitative species measurements due to a lack of temperature information. Rayleigh scattering has been a common approach for thermometry [4, 76], but such methods are limited to "clean" ows and require an estimation of gas composition that limits usage to low molecular weight fuel applications [77, 78]. An attractive alternative to the above methods is laser absorption spectroscopy (LAS). Though weaker in spatial resolution capability (due to line-of-sight integration), LAS provides a quantitative, calibration-free method of measuring major combustion species, particularly CO and CO 2 , and temperatures using compact low-power semiconductor mid-infared (IR) lasers [47]. The LAS system was developed and operated by UCLA, and more details on the system are provided are provided in Section 3.2.6. Following these observations, the goal is to investigate fuel and hydrodynamic eects on the ame 43 u jet (m/s) Re jet S L (cm/s) u 0 SL Lint f Ka Da = t f Re t 68.5 25000 20 41.1 9.55 222 0.25 3026 133 50000 20 80 10.23 624 0.12 5483 Table 7.1: Estimated key turbulence characteristics for the performed experiments. All values were calculated based on the kinematic viscosity of the unburned mixture, but the turbulent properties of the ow L int and u 0 were measured in the shear layer at x=D=15. thermochemical structure at high Karlovitz and Reynolds numbers. More precisely, the objectives are two-fold. The rst objective is to demonstrate the applicability of LAS techniques to provide high-quality major species data (CO, CO 2 and temperature) for turbulent ames in the thin and broken reaction zone regimes. Lastly, RANS calculations will be utilized to determine if simplied kinetics and ow approximations can capture the observed trends. 7.2 Hydrodynamic Eects on the Flame Thermochemical Structure 7.2.1 Experimental Parameter Space Experiments were performed at two jet Reynolds numbers, Re jet u jet D= = 25,000 and 50,000, where U jet is the bulk ow velocity and is the kinematic viscosity at the burner exit. The jet ame was an ethylene - air mixture at =0.55 at an unburned mixture temperature of 298 K. The co ow was a hydrogen/air ame at 1500 K. In contrast to previous experiments, a pure hydrogen co ow was required in order to remove any additional sources of CO or CO 2 . Table 7.1 provides a summary of the conditions investigated. The laminar ame speed, ame thickness and ame time were calculated using the PREMIX code [52]. The turbulent Reynolds number Re t u 0 L int = where L int is the integral length scale, was measured as in previous studies [36, 23]. The Karlovitz number (Ka) is dened asKa f , where is the Kolomogorov time scale determined as in previous studies [36, 23]. As characterized by the high u 0 and Ka, these ames are expected to fall in the thin reaction zone with broadened preheat layer regime. 7.2.2 Numerical Approach RANS calculations were used to provide a complimentary numerical analysis to the experimental measurements. Details of the principles regarding RANS are discussed in Section 4.3. 44 Flow Conguration The size of the computational domain was selected to be 234 mm x 100 mm in the axial and radial directions with a uniform radial grid spacing of 1000 points. In accordance with the radial symmetry of the jet, one radial wedge of the burner is simulated. Figure 7.1: Experimental setup: the regions of interest for the computation are indicated via the red dashed lines. A radial cut of interest (x=D=14) is shown as a white dashed line. The turbulent inlet velocity prole for the jet was established assuming fully developed pipe turbulence. The k-omega SST model was used to model the jet turbulence, as this model captures the gradients of a jet ow better than the typical k-epsilon model. Inlet conditions of k and ! at the jet exit were approximated by assuming a constant turbulence intensity I of 10%: k = 3=2(IU jet ) 2 (7.2.1) ! = k 0:5 C L int (7.2.2) where C =0.09. The pilot was simulated using hot ethylene/air combustion products at 1780K, and the co ow was simulated using hot hydrogen/air combustion products at 1500 K. Both streams had an exit velocity of 4.5 m/s corresponding to the burnt velocity of the original mixtures. RANS only captures the average behavior of the jet as it solves the Navier-Stokes equations averaged in time. RANS does not take into account the eects of preheat broadening as well as real turbulence, though approximations have been made. Therefore, due to the assumptions made in these RANS simulations, caution will be used in applying results, and thus discussion of the results in later sections will focus on a qualitative analysis of the observed trends. 45 7.2.3 Results and Discussion Experimental Results For both Reynolds number cases, one radial cut at a height of x=D=14 was used to compare the behavior of the CO and CO 2 mole fractions and temperature. The experimental radial proles for CO, CO 2 and temperature and shown in Figure 7.2. For both species, typical mole fraction uncertainty was6% and the temperature uncertainty was130 K. 0 0.4 0.8 1.2 1.6 2 0 0.002 0.004 0.006 0.008 0.01 0.012 r/D Mole Fraction Re=25,000 Re=50,000 (a) X CO 0 0.4 0.8 1.2 1.6 2 0 0.01 0.02 0.03 0.04 0.05 0.06 r/D Mole Fraction Re=25,000 Re=50,000 (b) X CO2 0 0.4 0.8 1.2 1.6 2 200 400 600 800 1,000 1,200 1,400 1,600 1,800 r/D Temperature (K) Re=25,000 Re=50,000 (c) T (K) Figure 7.2: Radial proles for experimental CO, CO 2 mole fractions, and temperature for Re jet =25,000 and Re jet =50,000. The CO mole fraction remained largely unchanged in response to the increase in turbulence intensity. At the higher Reynolds number,X CO also had a wider prole, diusing more out towards the co ow. However, while the mole fractions of CO were similar at the chosen heights, the CO 2 mole fractions were signicantly lower for theRe jet =50,000 case. This could suggest greater entertainment of the co ow or more likely, a reduction in the oxidation of the CO associated with the increased Reynolds number and nite rate kinetics. The behavior of the temperature results suggested the latter. Temperature resolution below 900 K was not reliable, so those regions were not plotted. The radial temperatures for theRe jet =50,000 case were generally lower than the Re jet =25,000 case, despite both jet mixture conditions having the same adiabatic ame temperature of 1695 K. The lower temperature prole was consistent with a lesser degree of oxidation. Comparison with RANS Calculations As with the experimental results, for both simulated Reynolds number cases, one radial cut at a height of x=D=14 is used to compare the behavior of the CO and CO 2 mole fractions and temperature. The calculated radial proles for CO, CO 2 and temperature and shown in Figure 7.3. The RANS calculations successfully captured the overall trends and magnitudes observed in the experimental results. CO production is similar between the two Reynolds numbers, but the resultant oxidation to CO 2 was greatly reduced for the Re jet =50,000 case. As in the experimental case, RANS also captured the increased radial spread of CO towards the co ow at higher Reynolds 46 0 0.4 0.8 1.2 1.6 2 0 0.005 0.01 0.015 0.02 0.025 r/D Mole Fraction Re=25,000 Re=50,000 (a) X CO 0 0.4 0.8 1.2 1.6 2 0 0.01 0.02 0.03 0.04 0.05 0.06 r/D Mole Fraction Re=25,000 Re=50,000 (b) X CO2 0 0.4 0.8 1.2 1.6 2 200 400 600 800 1,000 1,200 1,400 1,600 1,800 r/D Temperature (K) Re=25,000 Re=50,000 (c) T (K) Figure 7.3: Radial proles for simulated CO, CO 2 mole fractions, and temperature forRe jet =25,000 and Re jet =50,000. numbers. As RANS cannot capture the more complicated eects of preheat broadening and local extinction, the observed results cannot be attributed to these phenomena. The calculated temperature proles shown in Fig. 7.3c also re ect the temperature reduction of the higher Reynolds number case, further suggesting the impact of turbulence intensity on the ame chemistry. This can be further observed by analyzing the relationship of the average ame structure and the jet shear layers. Fig. 7.4 shows the normalized radial proles for the simulated CO and CO 2 mole fractions the turbulent kinetic energy k. 0.4 0.8 1.2 1.6 2 0 0.2 0.4 0.6 0.8 1 1.2 r/D (X i ) norm and (k) norm CO CO 2 k (a) Re jet =25,000 0.4 0.8 1.2 1.6 2 0 0.2 0.4 0.6 0.8 1 1.2 r/D (X i ) norm and (k) norm CO CO 2 k (b) Re jet =50,000 Figure 7.4: Normalized radial proles for simulated CO, CO 2 mole fractions, andk forRe jet =25,000 and Re jet =50,000. The jet shear layers are indicated by the distributions of k. At Re jet =25,000, the ame is anchored towards the unburnt side of the shear layer (slightly left of the peak k). The CO 2 mole fraction also matches the distribution of k, suggesting that the behavior of the shear layers control the products of ame oxidation. However, asRe jet increases, the position of the ame products and the shear layer shifts. Figure 7.4b shows that the ame oxidation is occurring fully in the shear layer, and the peak turbulent kinetic energy occurs between the peak CO and the peak CO 2 . It should be noted that the increase in Reynolds number results in an increase of turbulent kinetic energy of more than 300%. The new ame positioning suggests that the ame oxidation process from CO to CO 2 is now vulnerable to disruption by the increased turbulence intensity and could 47 result in a reduction in produced CO 2 . The shift of the CO 2 concentration out towards the co ow at Re jet =50,000 suggested some entrainment of the co ow, though this was likely not a dominant contributor in the reduction of CO 2 at the higher Reynolds number. 7.3 Fuel Eects on the Flame Thermochemical Structure LAS is well-suited to the exploration of the eects of fuel chemistry on turbulent ames, as it is not reliant on calibration measurements that can be stymied by the complex decomposition breakdown of heavy liquid hydrocarbons. The LAS measurements were expanded to explore the role of fuel chemistry in the ame thermochemical structure, and the experimental conditions are listed in the table below. The fuels of interest were chosen to match the baseline fuels discussed previously in Chapter 5. The jet Reynolds number was selected as 50,000 as previous studies indicated that dierences between the fuels became more evident at elevated Reynolds numbers. The ame speed was selected as 20 cm/s: this is an increased ame speed from previous comparisons between fuels and was selected to reduced the ame height to be more easily accessible by the LAT equipment. Re jet Fuel S L (cm/s) 50,000 methane (CH 4 ) 20 ethylene (C 2 H 4 ) n-heptane (n-C 7 H 16 ) toluene (C 7 H 8 ) Table 7.2: Experimental Conditions for the Expanded LAS Measurements To start, an initial laminar ame calculation was performed to establish the chemical parameters predicted from laminar ame theory. The predicted CO mole fraction in a laminar ame for the fuels of interest is shown in Fig. 7.5. As expected, the dierent fuels produce dierent amounts of CO. The ordering of the maximum CO follows the ratio of carbon to hydrogen present in a fuel molecule with toluene (7 carbon to 8 hydrogen) having the greatest CO and methane (1 carbon to 4 hydrogen) having the lowest. The laminar ame studies suggested that an appropriate way to compare between the fuels at turbulent conditions would be to assess the maximum CO concentration present in the ame brush. The maximum CO concentration was located along the centerline of the ame and depending on the fuel, was located at a height of 140 to 200 mm above the burner exit. Once the location of the maximum CO concentration was located, a radial prole at that height was taken. This process is illustrated in Fig. 7.6a. To compare the maximum CO mole fractions, the radial proles for the four fuels considered are presented in Fig. 7.6b. Here ethylene has the largest CO mole fraction with methane the smallest, and toluene and n- heptane following closely together. This is not the trend predicted by the laminar ame calculations. 48 6·10 −2 9·10 −2 0.12 1·10 −2 1.5·10 −2 2·10 −2 2.5·10 −2 3·10 −2 3.5·10 −2 4·10 −2 4.5·10 −2 X Axis (cm) CO Mole Fraction CH 4 C 2 H 4 n−C 7 H 16 C 7 H 8 Figure 7.5: Predicted CO mole fraction from laminar ame calculations This suggests that the interaction of the turbulent ow eld and the fuel chemistry is signicant to alter the resultant ame products. Though the ordering of the fuels does not match the predicted laminar ordering, it is identical to the ordering seen in the ame height and shear layer studies discussed in Chapter 5. In addition to dierences in fuel ordering, the structure of the CO mole fraction eld is also dierent between the fuels. Methane, toluene and n-heptane converge to similar proles away from the ame centerline; however, ethylene shows a much steeper reduction in CO mole fraction for increasing r=D. Lastly, the maximum measured CO mole fractions are substantially lower than the values predicted from the laminar ame calculations. This is not unexpected, as turbulent mixing is likely to alter the distribution of CO in the ame. Work by Steinberg et al. [35] proposes that the equivalence ratio in the reaction zone of a lean turbulent ame with a richer pilot does not correspond to the jet equivalence ratio but rather an intermediate value between the jet and pilot equivalence ratios. However, the CO mole fractions in the turbulent case are signicantly lower than the laminar prediction, not higher as would be suggested by substantial support by the pilot. It is possible that the turbulence is both altering the mixing of the ame reactants as well stratifying the reaction zone structure, resulting in competing eects between the two. 7.4 Concluding Remarks This chapter demonstrates that mid-infrared laser absorption tomography, focusing on the funda- mental bands for CO and CO 2 can obtain quantitative, spatially resolved thermochemical data without requiring calibration or knowledge of gas composition. The experimental measurements of mole fraction and temperature showed sensitivity to the turbulence intensity and chemical kinetic progress. RANS calculations showed good trend agreement with the experimental results despite chemical kinetic and ow simplications. Extension of the LAT measurements to multiple fuels 49 (a) Sample CO Mole Frac- tion Image 0.5 1 1.5 2 2.5 0 5·10 −3 1·10 −2 1.5·10 −2 2·10 −2 r/D CO Mole Fraction CH 4 C 2 H 4 n−C 7 H 16 C 7 H 8 (b) CO mole fraction at point of max concentra- tion for fuels considered Figure 7.6: Experimental CO mole fractions for conditions considered indicated that the CO mole fractions measured in the turbulent ames did not match the laminar ame predictions, both in magnitude and fuel ordering. This suggests that the interaction between the ow turbulence and the fuel chemistry results in a signicant departure from laminar predic- tions and indicates the limited applicability of laminar ame parameters in scaling turbulent ame results. 50 Chapter 8 Simultaneous Measurement of the Local Velocity Field and Preheat Structure 8.1 Introduction As discussed in 3.2.4, CH 2 O PLIF can provide information regarding the presence of CH 2 O in a turbulent ame, which indicates the preliminary steps of fuel oxidation, which can be an indicator of the ame preheat zone. At conditions of extreme turbulence and Karlovitz number, the preheat zone of the ame can be modied by the turbulent vortices, which could alter the fuel decompo- sition process more signicantly for heavy liquid hydrocarbons. However, CH 2 O PLIF studies of heavy liquid hydrocarbons are limited by two factors. First, few turbulent ame congurations can successfully introduce liquid fuel to the experimental rig. Second, many heavy liquid hydrocarbons in their liquid form uoresce at the CH 2 O excitation wavelength, thus preventing the accurate determination of a pure CH 2 O signal (a common problem in the study of liquid spray ames). However, recent unpublished developments have determined that for lean premixed ames where the liquid fuel is pre-vaporized and mixed with air, fuel uorescence is negligible and therefore does not prohibit access to the CH 2 O signal. While measurements of single species trends can provide signicant insight on the ame behavior, the eect of the ame on the velocity eld can indicate the evolution of turbulence through the ame surface. An eective way to accomplish this is through the use of conditional statistics. Conditional statistics track the evolution of the velocity vector as a function of distance from the ame surface, rather than the laboratory frame of reference. The need to know conditioned velocities in premixed turbulent ames have been previously established by Bray et al. [79] and Shepard et al. [80]. Velocity uctuations at a point in the oweld have three contributions: one due to uctuations where only reactants are present, a second when only products are present and a third from amelet crossing (which is not turbulence). The use of conditioned velocity measurements prevents ctitious contributions to u 0 by eliminating the amelet crossing contribution. Previous calculations performed by Wabel et al. [11] demonstrated the use of conditional statistics to assess the potential decay of turbulence through the preheat zone. However such studies have limited 51 their explorations only to methane. It's important to extend these observations to heavy liquid hydrocarbons as the eect of preheat zone modication through turbulence may aect these ames dierently due to fuel decomposition. 8.2 Experimental Parameter Space In order to successfully apply edge detection techniques to a CH 2 O-PLIF image, a clear edge needed to be visible. Initial studies were performed on an ethylene-air ame at twoRe jet (25,000 and 50,000) to determine the signal and edge quality. These images are shown in Fig. 8.1. (a) Re jet =25,000 (b) Re jet =50,000 Figure 8.1: Eect of increasing Reynolds number on the instantaneous CH 2 O signal for an C 2 H 4 -air ame, S L =11.3 cm/s. The higher Reynolds number images demonstrated a broader and more diuse edge structure consistent with other studies. The lack of dened edge presented a major diculty for edge detec- tion, therefore it was decided to focus on the Re jet = 25,000 case. Once the Reynolds number condition was dened, three fuels were chosen based on previous stud- ies: methane, ethylene and Jet-A. Methane and ethylene were chosen as previous studies indicated that they demonstrated unique behavior despite scaling with S L . Jet-A was an ideal companion choice as a liquid jet fuel and a reduced aromatic makeup which improves compatibility with laser diagnostics. The measurements focused on a spatial region 20 mm wide reaching from x=D=13 tox=D=17. This was done to provide sucient resolution for the PIV and PLIF measurements, and it enabled focus on a region of sucient CH 2 O intensity. The PLIF-PIV experimental methodology and PIV processing follows as discussed in Section 3.2.3. The generation of conditional statistics will be discussed later in this section (Section 8.4.1). 52 Re jet Fuel S L (cm/s) u 0 =S L Re T Ka 25,000 Methane 15.3 110 5890 1244 Ethylene Jet-A Table 8.1: Estimated key turbulence parameters for the experiments performed. 8.3 Eects of Fuel Type onC 2 HO Concentration and Struc- ture It was of interest to rst compare the C 2 HO signal intensities between the dierent fuels. This was done by averaging the dataset images together to produce an average prole. To compare the averaged sets, a representative radial prole was taken at x=D=16. The mean of the C 2 HO signal is shown in Fig. 8.2a. −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 r/D Counts (A.U.) CH 4 C 2 H 4 Jet-A (a) Mean CH 2 O Signal −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 0 500 1,000 1,500 2,000 2,500 3,000 3,500 r/D Counts (A.U.) CH 4 C 2 H 4 Jet-A (b) Root Means Square (RMS) of CH 2 O Signal Figure 8.2: Radial proles of averaged CH 2 O signal at x=D=16, Re jet =25,000 for CH 4 , C 2 H 4 and Jet-A. The C 2 HO intensity is dierent for the dierent fuels: laminar ame calculations indicated that ethylene should have the highest C 2 HO concentration with methane the lowest and Jet-A similar to methane. However the Jet-A signal is signicantly higher than methane, being closer in magnitude to ethylene. Ethylene has previously been shown to be the primary product of the jet fuel decomposition process [81]. The similar concentration of Jet-A compared to ethylene in the turbulent ame suggests that the decomposition process may indeed by altered by the eects of preheat broadening. There are also small dierences observed in the radial distribution of the C 2 HO signal: ethylene exhibits a more narrow radial prole than methane or Jet-A. This is consistent with previous observations that ethylene has a more narrow ame structure than other fuels examined. 53 Further dierences between the fuels are present when the root mean square (RMS) of the mean CH 2 O signal is examined. As with the velocity eld, the RMS can provide insight to the uctuating component of the CH 2 O signal. The RMS of the mean radial prole at x=D=16 is shown in Fig. 8.2b. The RMS signal prole depends signicantly on the fuel. There is a substantial dierence of more than 20% in the maximum RMS for each fuel which suggests that the fuel type is a strong factor in the instantaneous behavior of the ame. It is also signicant to note that though the mean CH 2 O signal follows an ordering of the fuels as suggested by the laminar ame calculations, the RMS instead follows the order previously observed in the velocity RMS as well as the CH* ame height with methane forming the lower bound of the quantity in question, ethylene a maximum bound and the liquid fuels bounded in between. This consistent trend suggests that the fuel chemistry has a signicant and consistent in uence on the behavior of the ame despite eorts to scale results using laminar ame properties. 8.4 Conditional Velocity and Vorticity Statistics 8.4.1 Calculation of Conditional Statistics The primary goal is to explore the eect of fuel on the uctuating quantities of the ame conditioned on the ame surface. Unconditioned velocity statistics cannot account for the presence of the ame, and the magnitude can be aected by whether it is located before or after the ame. Conditioned statistics can correct for the presence of the ame and provide an unbiased assessment of the local velocity uctuations. The process of identifying the conditional statistics can be divided into six steps. 1. Map the ame surface identied through the CH 2 O-PLIF signal to the vector eld produced by the corresponding PIV image. 2. Identify the vectors before and after the ame and assign each vector a distance from the ame surface. 3. Discretize the distance from reaction surface into increments 0.05 mm from the reaction zone and bin vectors together that fall into those bin increments. 4. Average the vectors inside each distance bin to compute a mean quantity corresponding to that distance from the reaction zone. 5. Compute the conditional quantity at each distance from the reaction zone by subtracting the bin average from each in the bin. Preliminary image processing was performed using Davis to remove spotting due to the image intensier from the PLIF images. A MATLAB code was written to process the velocity elds from Davis with the PLIF images to successfully compute the desired conditional quantities. 54 Figure 8.3: Example contours of local distance from the reaction zone (shown in red). Contours are shown only for the reactants to improve clarity. . 8.4.2 Conditional Statistics: V 0 C;x The rst conditional statistics of interest were those in the axial (x) direction as the velocities in this direction are dominant in the jet ame conguration. To establish a baseline for the conditional statistics, the unconditioned statistics for V x and V 0 x were examined in Figs. 8.4b and 8.4c for methane and Jet-A with a radial prole at a height of x=D=16. The axial velocity proles are similar, in both the mean and uctuating proles. The prole of V 0 x displays the two prominent peaks resulting from the jet shear layers. As noted from previous studies, do not locally show the eects of the ame though globally exhibit the eects of the ame when compared to cold ow proles. The conditional axial velocity uctuation (the velocity uctuation conditioned on the distance from the ame surface, V 0 C;x however exhibits the clear in uence of ame-generated turbulence. Figure 8.5a showsV 0 C;x as a function of distance from the reaction zone. Negative distances indicate the pre- ame region, and positive distances indicate the post- ame region. V 0 C;x increases slowly at rst, exhibiting a sharp increase moving towards the reaction zone, peaking at the ame surface before decreasing in the post- ame region. This sharp peak is exhibited by both the methane and Jet-A ames. There is also a near constant increase inV 0 C;x both before and after the ame due to the presence of the shear layers. Similar behavior is seen in the conditional velocity statistics calculated by Wabel [11], thoughV 0 C;x calculated in the HI-Pilot conguration has fewer changes in the velocity uctuation in the broader pre and post- ame regions due to dierences in the ow conguration (the PPJB uses shear-generated turbulence and the HI-Pilot uses grid-generated turbulence). Both fuels behave similarly in the pre- ame region. At the ame surface, deviations begin to appear between the methane and Jet-A proles. Methane exhibits a slightly greater peak in the 55 −4 −2 0 2 4 6 5 10 15 20 25 Distance from Reaction Zone (mm) Conditional V 0 x (m/s) CH 4 Jet-A (a) Conditional Axial Velocity Fluctuation (V 0 C;x ) −1.5 −1 −0.5 0 0.5 1 1.5 20 40 60 80 100 120 r/D ¯ V x (m/s) CH 4 Jet-A (b) V x at x=D=16 −1.5 −1 −0.5 0 0.5 1 1.5 12 14 16 18 20 r/D V 0 x (m/s) CH 4 Jet-A (c) V 0 x at x=D=16 Figure 8.4: Conditional and unconditioned velocity statistics for CH 4 and Jet-A, S L =15.3 cm/s, Re jet =25,000. velocity uctuation: V 0 C;x continues to rise to a signicantly higher velocity uctuation than the Jet-A ame at a greater distance from the ame surface. These dierences may suggest that fuel chemistry entering the ame has an impact on the post- ame results. Modication of the preheat zone through extreme turbulence could aect the fuel pyrolysis process for Jet-A, thus modifying the fuel fragments reaching the reaction zone. 8.4.3 Conditional Statistics: ! 0 C;xy Another quantity of interest is the eect of the ame and fuel type on the 2-D vorticity in the xy direction (! xy ). As with the conditional statistics for V 0 x , the unconditioned mean and uctuating vorticity is calculated and shown in Figs. 8.5b and 8.5c. 56 −3 −2 −1 0 1 2 3 15,000 20,000 25,000 30,000 35,000 Distance from Reaction Zone (mm) Conditional ω 0 xy (1/s) CH 4 Jet-A (a) Conditional XY Vorticity Fluctuation −1.5 −1 −0.5 0 0.5 1 1.5 −20,000 −10,000 0 10,000 20,000 r/D ¯ ω xy (1/s) CH 4 Jet-A (b) ! xy at x=D=16 −1.5 −1 −0.5 0 0.5 1 1.5 20,000 25,000 30,000 35,000 40,000 45,000 50,000 r/D ω 0 xy (1/s) CH 4 Jet-A (c) ! 0 xy at x=D=16 Figure 8.5: Conditional and unconditioned vorticity statistics for CH 4 and Jet-A, S L =15.3 cm/s, Re jet =25,000. For the sameS L , the mean and uctuating vorticities were similar, showing no strong dierences as a result of the fuel type. Both proles are typical of shear-generated turbulent ames. The mean vorticity displays peak vorticities at r/D=0.5 due to the presence of the shear layers. The uctuating vorticity however remains constant between the two shear layer peaks, before dropping as the turbulence decays. The conditional vorticity uctuation (! 0 C;xy ) is calculated in the same manner asV 0 C;x and shown in Fig. 8.5a. The upstream vorticity before the preheat zone is relatively constant and similar between the fuels. As the ame surface nears the ame surface, the local vorticity uctuation peaks sharply before decreasing sharply through the ame. Peak uctuating vorticity is reached not at the ame surface itself but rather slightly before. This trend was also observed in Wabel et al.[11] which also suggests that the observed behavior is independent of the method of turbulence generation. The post- ame vorticity uctuation is more elevated for the methane ame as compared 57 to the Jet-A ame, beginning after vorticity peak. As also experienced with V 0 C;x , this suggests a potential modication of the fuel fragments entering the reaction zone, thus modifying the post- ame products. 8.5 Concluding Remarks Simultaneous CH 2 O PLIF and 2-D PIV were undertaken to study the eect of fuel chemistry on the velocity eld statistics conditioned on the distance from the ame surface. The C 2 HO intensity is found to be dierent for the dierent fuels examined, though the Jet-A signal was higher than expected from the predictions made in laminar ame calculations. The RMS of the C 2 HO instead follows the order previously observed in the velocity RMS as well as the CH* ame height with methane forming the lower bound of the quantity in question, ethylene a maximum bound and the liquid fuels bounded in between. This consistent trend suggests that the fuel chemistry has a signicant and consistent in uence on the behavior of the ame despite eorts to scale results using laminar ame properties. The C 2 HO signal was used to identify vectors before and after the ame in order to produce ame-conditioned statistics. The conditional x velocity uctuation and xy vorticity uctuation showed the distinct presence of the ame front as well as indications that fuel chemistry may in uence the the post- ame velocity eld. 58 Chapter 9 Assessment of Experimental Observables for Local Extinction Through Unsteady Laminar Flame Calculations 9.1 Introduction The characterization of local extinction in turbulent combustion, a phenomenon that is not only a manifestation of nite-rate kinetic eects that have theoretical challenges but is also of practical importance for the operation of combustion devices, has been the focus of extensive research over many decades. For instance, the stability of premixed ames in afterburners has been one of the rst problems studied (e.g., [82, 83, 4, 84, 85, 86]). The Turbulent Non-Premixed Flame (TNF) Workshop has focused on such local extinction phenomena for years for piloted jet diusion ames and have provided experimental databases for model validation [87, 88, 89, 90, 91], with special emphasis on conditional data, with the conditioning done on mixture fraction. More recently, signicant activity on high Karlovitz number premixed ames [4, 92, 93, 94, 95, 23] is revealing local information on ame zone thickness but also on conditional statistics, with the conditioning done on progress variable. It would be helpful to supplement these one-time scalar data with information on the transient behavior of ames as they undergo extinction and how the various commonly-used scalar measurements behave through a ame extinction event. Chemiluminescence has been used often for monitoring the presence of chemical reaction [37]. Chemiluminescence emission forms as a result of key chemical processes in the ame, where excited radicals such as CH*, OH* and C 2 emit light at a characteristic wavelength as they return to a lower energy state [96]. Extensive studies have shown that under some conditions, the magnitude of the emitted light at particular wavelengths, namely CH* and OH*, may be used as a semi- quantitative measure of the heat release rate, especially for fully premixed systems [38, 39]. Further studies at high strain rates and increased pressures also showed a strong link between CH*, OH* and the heat release rate [96] though the CH* chemiluminescence did not appear to be sensitive changes in the local strain rate [97]. In the presence of local extinction, however, which involves 59 sharp transients in chemical species mass fractions, the way the chemiluminescence signal adapts is open to interpretation [98]. Najm et al. [99] argued that CH* is not an adequate indicator of local extinction, as it was observed that despite a breakage of the CH* ame surface, OH and HCO were still present in signicant quantities. However, such simulations did not continue in time past the moment of CH* breakage and did not capture the full extinction process. Recently, in a high Reynolds number experiment of hydrocarbon-air premixed ames, the absence of CH* was correlated with local extinction [23], but it would be advantageous to know exactly when the chemiluminescence signal is lost during the extinction transient. A similar question arises with laser diagnostic studies for ames close to extinction, where separate or simultaneous OH and CH 2 O-PLIF have been used to provide statistics of local extinc- tion [100] as well information on the behavior of the preheat layer at high turbulence levels (e.g., [101, 6]) and the local heat release rate [102, 95]. CH 4 -air lean turbulent premixed ame stabilized by a blu body close to the global blow-o operating condition [103] showed signicant build-up of CH 2 O inside the recirculation zone, attributed to the presence of incomplete combustion products due to local extinctions along the ame, while in a swirl CH 4 non-premixed ame, the absence of OH was attributed to ame lift-o from the blu body edge [104]. As with chemiluminescence, the evolution of OH and CH 2 O during an extinction transient are important to quantify so that the interpretation of experimental observables can be made unambiguously. This chapter seeks to evaluate the relevance and signicance of CH 2 O and chemiluminescence as observables under conditions close to extinction. Questions remain regarding the role of pressure and the eect of fuel under these conditions. These questions should be addressed quantitatively in order to reliably apply the experimental methods in complex ames and reacting environments. This work proposes to ll these gaps and discuss quantitatively the behavior of experimental observables as a function of ame conguration, pressure and duration of the extinction transient for fuels relevant to practical applications. As turbulent ame studies push to higher pressures, an understanding of species behavior under extinction conditions will facilitate interpretation of diagnostic results. Unsteady counter ow simulations provide a useful canonical conguration to study extinction relevant to the highly strained, unsteady nature of turbulence. Simulations were conducted using an unsteady premixed counter ow conguration. Turbulence cannot be represented either by a sine wave or by a pulse, nevertheless exploration of these canonical problems with laminar ames can help with building insights on how reaction zones in turbulence may respond to sudden and local excursions in strain rate. 9.2 Numerical Approach 9.2.1 Unsteady OPPDIFF Conguration Premixed back-to-back counter ow simulations were conducted using a modied opposed jet con- guration (OPPDIF) [105, 53] as discussed in Section 4.2. All premixed ames examined in this 60 Table 9.1: Cases investigated. Fuel Pressure (bar) T u (K) K ext (s 1 ) CH 4 1 403 1830 1 498 360 5 571 9674 n-C 12 H 26 1 403 474 5 571 3031 chapter are in the unsteady extinction regime and extinguish during the second period of oscillation. The domain size was 0.5 cm with a mesh resolution of 2.77 microns per grid point. Unsteadiness was introduced for the reactant velocity by imposing xed sinusoidal variation of a given amplitude around the mean exit velocity. For the cases discussed in this paper, the amplitude of the sinusoidal variations was set to produce a strain rate amplitude of approximately 10% of the starting strain rate near extinction. 9.2.2 Conditions of Interest Both the premixed and non-premixed ame congurations were studied with CH 4 and n-C 12 H 26 at dierent pressures. CH 4 is commonly used in turbulent ame studies and n-C 12 H 26 is represen- tative of complex practical liquid fuels. The investigated conditions are summarized in Table 9.1. The ame response in the premixed case was examined for CH 4 /air and n-C 12 H 26 /air ames at equivalence ratio =0.7 for 1 and 5 bar. The unburned mixture temperature (T u ) was 403 K at 1 bar. For higher pressures,T u was adjusted to account for n-C 12 H 26 vaporization requirements. The non-premixed ame conguration was investigated for three dierent pressures: 1, 5, and 10 bar. The temperature of the fuel and oxidizer was chosen to ensure the existence of pure fuel in vapor form at the respective pressure. The kinetic models used were USC-Mech II [58] for the CH 4 ames and JetSurf 2.0 [59] for the n-C 12 H 26 ames including the CH*/OH* model developed by Nori and Seitzman [60]. All quantities plotted have been scaled by the starting value of the computational cycle. 9.3 Characterization of the Extinction Transient: Indica- tors of Heat Release Extinction results will rst be discussed for the 1 bar extinction case. Figure 9.1 illustrates the extinction process for the CH 4 premixed ame. The results in Fig. 9.1 include the peak instantaneous values of species mole fractions and peak heat release rates scaled by the starting value of the computational cycle, that is (X i;max ) scaled for species i and ( _ q max) scaled respectively. 61 0 0.7 1.5 2 0 0.25 0.5 0.75 1 t × K ext (X i,max ) scaled and (˙ q max ) scaled OH ˙ qmax HCO OH × CH2O CH* CH2O Figure 9.1: Extinction transient for premixed CH 4 ame (P =1 bar,T u =403K). Figure 9.1 focuses on the behavior of six key quantities, that is OH, HCO, the product of mole fractions of OH and CH 2 O, CH*, and CH 2 O as well as the maximum heat release rate ( _ q max ). X HCO and the product of X OH and X CH 2 O are often used as experimental markers for heat re- lease in turbulent ames, while CH* is produced during vigorous burning and can indicate the location of the reaction zone as it is concentrated in regions of high temperature. The results of Fig. 9.1 indicate that the ame experiences a period of decreased reactivity followed by recovery and then complete extinction. Extinction is indicated by the steep loss of _ q max and species con- centrations. (X HCO;max ) scaled and (X OH;max ) scaled x (X CH 2 O;max ) scaled follow the temporal behavior of _ q max . (X CH;max ) scaled does closely follow the temporal uctuations of the heat release. It can be seen that Najm et al. [99] were initially correct in their hypothesis, as CH* does disappear while there is still OH and HCO present. Yet, as CH* disappears, OH and HCO (in addition to _ q max ) shortly follow. CH* scaled well with the temporal uctuations of _ q max and OH prior to extinction, which suggests that the general extinction potential can be inferred through the CH* behavior. In all simulations with unsteady extinction, no ame recovery was observed after the maximum CH* concentration dropped to negligible concentrations for both premixed and non-premixed ames. Therefore, it is perhaps more accurate to treat CH* as the rst indicator that the ame is on the path to extinction and will extinguish. Though the absence of CH* may not indicate immediate extinction, it should be treated as an indicator that extinction will occur. 62 9.3.1 Characterization of the Extinction Transient: Behavior of CH 2 O Through Extinction Of particular note is the behavior of (X CH 2 O;max ) scaled . As seen in Fig. 9.1, CH 2 O concentration remains high once the extinction process initiates and subsequently decays at a more gradual rate than the other key species. This behavior is observed in both the CH 4 and n-C 12 H 26 ames. (X CH 2 O;max ) scaled does not experience a signicant response to the oscillation in strain rate, unlike _ q max and other species. Reaction path analysis indicates that the dominant pathway for CH 2 O consumption throughout the extinction transient is through the CH 2 O + OH, HCO + H 2 O reaction, as can be seen in Fig. 9.2. Though previous studies have suggested the in uence of CH 2 O reaction of O and HO 2 , these reactions did not feature heavily under these conditions [106, 107]. 0 20 40 60 CH 2 O+OH⇔ HCO+H 2 O CH 2 O+H⇔ HCO+H 2 CH 2 O+O⇔ HCO+OH CH 2 O Consumption Pathway (%) CH 4 n-C 12 H 26 Figure 9.2: Reaction path analysis for premixed CH 4 and n-C 12 H 26 ames at P =1 bar. This dependency on OH suggests that a loss of OH concentration would results in a halt in CH 2 O consumption. This can be further elucidated by observing the characteristic destruction time of CH 2 O, [C CH 2 O ]= _ D CH 2 O , whereC CH 2 O and _ D CH 2 O stand for the molar concentration and molar destruction rate of CH 2 O respectively. is an indicator of CH 2 O consumption and the overall reaction intensity. Under vigorously burning conditions, remains small. When the consumption process slows and/or stops, this change can be observed through an exponential increase in . Figure 9.3 provides a closer look at the behavior of (X OH;max ) scaled , ( _ q max ) scaled , (X CH 2 O;max ) scaled , and . Reaction cessation is indicated by the notable reduction in ( _ q max) scaled and the attendant expo- nential increase in. Loss of (X OH;max ) scaled occurs concurrently with this exponential increase. As expected from the reaction path analysis, this conrms that the loss of OH concentration correlates with the cessation of CH 2 O consumption. Therefore, after extinction CH 2 O is no longer being con- sumed and it is merely transported away from the previously reacting region via convection and/or diusion. It is also worthwhile to mention that the some of the reaction processes that consume large hydrocarbon fragments to produce CH 2 O are also responsible for the consumption of OH. 63 (a) CH 4 ame (b) n-C 12 H 26 ame Figure 9.3: Zoomed snapshot of evolution of (X OH;max ) scaled , (X CH 2 O;max ) scaled and ( _ q max ) scaled with CH 2 O destruction timescale for premixed CH 4 and n-C 12 H 26 ames at P =1 bar. During the extinction process therefore, CH 2 O consumption may also be oset by its concurrent production, responsible in part for the reduction in OH. This relationship could also account for the relative insensitivity of CH 2 O to the ow oscillations of the premixed ame. Another contributing factor to the persistence of X CH 2 O is the insensitivity of X CH 2 O to the temperature uctuations present during extinction. Figure 9.4 illustrates the dependence on maxi- mum temperature of the ve key species and _ q max through the extinction transient of the CH 4 ame extinction. 1,100 1,200 1,300 1,400 1,500 1,600 1,700 0 0.2 0.4 0.6 0.8 1 Maximum Temperature (K) (X i,max ) scaled and (˙ q max ) scaled OH ˙ qmax HCO OH × CH2O CH2O CH* Figure 9.4: Dependence of key transient species on maximum temperature, P =1 bar, T u =403K. The behavior of X CH 2 O compared to the other species is particularly striking. (X OH;max ) scaled , (X HCO;max ) scaled , (X OH;max ) scaled x (X CH 2 O;max ) scaled , (X CH;max ) scaled show a sharp decrease in con- centration in response to small changes in temperature. This is to be expected with high activation 64 energy processes in which small decreases in temperature will cause large reductions in reaction rates and chemical activity. X CH 2 O however is relatively insensitive to the temperature uctuations present. X CH 2 O exhibits only small uctuations to the reduction of temperature during the extinc- tion transient. This could also account for the relative insensitivity of CH 2 O to the ow oscillations of the premixed ame. In Fig. 9.4, the initial reduction in X CH 2 O as the ame oscillates towards extinction is due to the continuation of reaction of CH 2 O with OH balanced by against CH 2 O pro- duction from fuel consumption reactions. However, after the notable reduction of OH concentration at the lower temperatures, incomplete combustion continues to produce some CH 2 O to compete with the outwards transport of species. 9.4 Eect of Increased Pressure on Experimental Observ- ables Pressure can notably aect burning characteristics [108, 109]. The vast majority of applications operate at high pressures, and thus turbulent ame studies at high pressure should address the changes in ame behavior from atmospheric conditions. In particular, it is desirable to understand how changes in pressures can aect experimental observables in the event of an unsteady ame extinction. 9.4.1 Eect of Pressure on X CH 2 O;max To explore the potential eects of pressure on CH 2 O concentration in an extinction event, premixed ame calculations at the pressures investigated in this work are discussed next. Figure 9.5 shows the evolution of CH 2 O and OH concentration in a n-C 12 H 26 ame with the time being scaled by the extinction strain rate; using a density-weighted strain rate instead did not change the conclusions. 0 2 4 6 8 10 12 0 0.2 0.4 0.6 0.8 1 0 2 4 6 8 10 12 0 0.2 0.4 0.6 0.8 1 1.2 t×K ext (X CH2O,max ) scaled (X OH,max ) scaled CH 2 O 1bar CH 2 O 5bar OH 1bar OH 5bar Figure 9.5: Evolution of (X CH 2 O;max) scaled and (X OH;max) scaled for n-C 12 H 26 ame at P =1, 5, and 10 bar 65 Increased pressure results in an extension of reactivity, clearly noted between the 1 bar and 5 bar cases, as is the increasing time-separation between the reductions of OH and CH 2 O. The increase in (X CH 2 O;max ) scaled is also slightly visible as a small peak at the tK ext =10 and 13. 9.4.2 Correlation between the Product of X OH and X CH 2 O and _ q max The product ofX OH andX CH 2 O is often used as an experimental marker for heat release in turbulent ames, as rst validated by Paul and Najm [99, 46]. Studies by Nikolaou and Swaminathan [110] showed that despite the drawbacks, X OH x X CH 2 O can still be used to indicate increased chemical activity in both methane and multi-component fuels. While this diagnostic is widely implemented in turbulent ames at atmospheric pressures, its implementation under high pressure conditions is far more limited. It is desirable to know if the correlation between the product of OH and CH 2 O and _ q max is still valid for increasing pressures. Figure 9.6 shows the evolution of OH x CH 2 O concentration and _ q max in a n-C 12 H 26 premixed ame at both 1 and 5 bar. 0 2 4 6 8 10 12 14 0 0.2 0.4 0.6 0.8 1 t × K ext (X i,max ) scaled and (˙ q max ) scaled (OH × CH 2 O) 1 bar (˙ q max ) 1 bar (OH × CH 2 O) 5 bar (˙ q max ) 5 bar Figure 9.6: Evolution of OH x CH 2 O concentration and _ q max for premixed n-C 12 H 26 ame at P =1 and 5 bar As seen in the CH 4 ame at 1 bar, (X OH;max ) scaled x (X CH 2 O;max ) scaled and _ q max both exhibit similar temporal behavior. When the pressure is increased to 5 bar, this correlation remains valid, as (X OH;max ) scaled x (X CH 2 O;max ) scaled tracks the extinction behavior of _ q max quite well. The results suggest rst that (X OH;max ) scaled x (X CH 2 O;max ) scaled is a suitable diagnostic to approximate the behavior of _ q max for multiple fuels. Furthermore, the close correlation between the product of X OH and X CH 2 O and _ q max is valid at both 1 bar and 5 bar throughout the extinction process. This indicates that this experimental approach for tracking the heat release rate can likely be utilized to provide insight to ame behavior for not only a range of fuels but also high pressure, highly turbulent conditions with the potential for extinction events. 66 9.5 Concluding Remarks In the present study, unsteady premixed ame calculations were performed to characterize species evolution during transient extinction. The results can be used towards improved interpretation of experimental diagnostics in turbulent combustion. CH 4 and n-C 12 H 26 ames were chosen for this analysis as CH 4 is typical of turbulent combustion studies, and n-C 12 H 26 is of interest in practical fuel studies. These fuels were studied under a variety of pressures with accompanying change in preheat temperature to ensure vaporization for n-C 12 H 26 . It was observed that CH 2 O concentration persists after the extinction transient. Through reac- tion path analysis and examination of the CH 2 O consumption time, it was found that in both CH 4 and n-C 12 H 26 premixed ames, reaction with OH is the dominant CH 2 O consumption pathway. When the extinction transient passes and OH dissipates, CH 2 O consumption also ceases and is transported away. Increased pressure extends the region of reactivity. Increased pressure also does not aect the good correlation of OH x CH 2 O and _ q max for n-C 12 H 26 premixed ames. The behavior of CH* and OH* chemiluminescence was also examined during the extinction transient, and it was concluded that CH* could serve as an indicator that ame extinction was imminent. In both the premixed and non-premixed ame congurations, it was determined that CH* and OH* loss precedes extinction, and no circumstances were observed when a ame remained burning shortly after a notable reduction in CH*/OH* concentration. It should be noted that while the present studies involved a simplied set of conditions controlling the behavior of unsteady laminar ames, the results strongly suggest that unsteady eects must be taken into consideration during the interpretation of experimental data obtained in complex turbulent ame experiments in which local extinction phenomena are highly probable at large Karlovitz numbers. 67 Chapter 10 Conclusions and Recommendations 10.1 Concluding Remarks A Piloted Premixed Jet Burner (PPJB) was developed at the University of Southern California to investigate the behavior of heavy hydrocarbons in the presence of highly turbulent premixed jet ames. To compliment the atmospheric studies, a variable pressure chamber and a modied PPJB was developed to examine turbulent premixed ames at elevated pressures. These congurations were used to study the ame height derived from CH* luminosity as well as the two-dimensional spatially resolved velocity elds through Particle Image Velocimetry (PIV). First, the ame height in the atmospheric conguration was scaled with laminar ame speed, maximum heat release rate, and ame thickness for all fuels at Re jet = 25,000 and 50,000. None of the scalings were found to be adequate to suppress fuel eects at all conditions, though dierences were less pronounced for stronger burning ames. A more detailed examination of the shear layers was conducted using the turbulent kinetic energy and turbulent shear stress for matching maximum heat release rates and laminar ame speeds. Though all fuels have a similar production period for turbulent kinetic energy and turbulent shear stress atRe jet = 25,000, methane ames exhibit a steeper decrease than the other fuels. This discrepancy was found to be more evident for the higher Reynolds number, suggesting that the impact of fuel chemistry becomes more important with increasing turbulence. The modied PPJB conguration and pressure chamber were also used to begin baseline studies of global observables as a function of fuel and pressure. Using averaged CH* chemiluminescence to calculate H , H for various fuels was found to scale moderately well with S L at 1 and 2 atm but less so at 3 atm. H /D was found to be controlled by extinction and propagation eects with varying pressure, as seen in the tip-quenched and tip-ignited regimes. When H /D is scaled by the density-weighted laminar ame speed, the tip-quenched state can be seen to be strongly in uenced by density. The tip-ignited state however is dominated by velocity. For the same u S L , H /D decreases with increasing pressure (decreasing jet velocity). The experimental maximum CH* intensity and calculated PREMIX maximum CH* concentration showed close agreement in the observed CH* trends. 68 A partnership was undertaken with the Spearrin lab at UCLA to obtain quantitative, spatially resolved proles of the CO and CO 2 mole fractions and the temperature eld. The experimental measurements of mole fraction and temperature showed sensitivity to the turbulence intensity and chemical kinetic progress. RANS calculations showed good trend agreement with the experimental results despite chemical kinetic and ow simplications. Extension of the LAT measurements to multiple fuels indicated that the CO mole fractions measured in the turbulent ames did not match the laminar ame predictions, both in magnitude and fuel ordering. This suggests that the interaction between the ow turbulence and the fuel chemistry results in a signicant departure from laminar predictions and indicates the limited applicability of laminar ame parameters in scaling turbulent ame results. Simultaneous CH 2 O PLIF and 2-D PIV were undertaken to study the eect of fuel chemistry on the velocity eld statistics conditioned on the distance from the ame surface. The C 2 HO intensity is found to be dierent for the dierent fuels examined, though the Jet-A signal was higher than expected from the predictions made in laminar ame calculations. The RMS of the C 2 HO instead follows the order previously observed in the velocity RMS as well as the CH* ame height. The conditional x velocity uctuation and xy vorticity uctuation also showed the distinct presence of the ame front as well as indications that fuel chemistry may in uence the the post- ame velocity eld. Lastly, unsteady premixed ame calculations were performed to characterize species evolution during transient extinction. It was observed that CH 2 O concentration persists after the extinction transient. Through reaction path analysis and examination of the CH 2 O consumption time, it was found that in both CH 4 and n-C 12 H 26 premixed ames, reaction with OH is the dominant CH 2 O consumption pathway. When the extinction transient passes and OH dissipates, CH 2 O consumption also ceases and is transported away. Though the unsteady calculations involved a simplied set of conditions controlling the behavior of unsteady laminar ames, the results strongly suggest that unsteady eects must be taken into consideration during the interpretation of experimental data obtained in complex turbulent ame experiments in which local extinction phenomena are highly probable at large Karlovitz numbers. In conclusion, the consistent trends observed across a variety of experimental diagnostics suggests that the fuel chemistry has a signicant and consistent in uence on the behavior of the ame despite eorts to scale results using laminar ame properties. 10.2 Recommendations for Future Work Future work exists to further the CH 2 O PLIF-PIV investigations in the atmospheric pressure PPJB facility to additional fuels and turbulence levels. The receipt of DURIP grant in Spring 2019 has enabled the development of OH PLIF and Filtered Rayleigh Scattering (FRS) capabilities for the measurement of OH radicals and temperature in a turbulent oweld. Combing OH PLIF and 69 CH 2 O PLIF will allow more accurate determination of the preheat and reaction zone structures which can then be expanded to study heavy molecular weight fuels. PIV capabilities and scalar measurements (when appropriate) will be expanded to the variable pressure chamber to study the eects of pressure and fuel chemistry on the oweld. Future collaborations with UCLA are planned to expand the Laser Absorption Tomography measurements to additional fuels and higher speeds in the atmospheric pressure PPJB facility. As the LAT measurements perform well at non-atmospheric pressures, these diagnostics will also be integrated with the VPPJB facility to explore both low and high pressure regimes. The VPPJB facility will also be upgraded with a vaporization system in order to study heavy molecular weight fuels at both high and low pressures. Companion eorts with Stanford are underway to use LES to duplicate the results discussed in Chapter 7. 70 Bibliography [1] C. K. Law, Combustion physics, Cambridge university press, 2010. [2] C. Wei, D. Pineda, L. Paxton, F. Egolfopoulos, R. Spearrin, Appl. Phys. B 124 (6). [3] J. Temme, T. Wabel, A. Skiba, J. Driscoll, in: 53rd AIAA Aerospace Sciences Meeting, 2015, p. 0168. [4] M. J. Dunn, A. R. Masri, R. W. Bilger, Combust. Flame 151 (1-2) (2007) 46{60. [5] T. M. Wabel, A. W. Skiba, J. E. Temme, J. F. Driscoll, Proc. Combust. Inst. 36 (2) (2017) 1809{1816. [6] J. F. Driscoll, Prog. Energy Combust. Sci. 34 (1) (2008) 91{134. [7] M. Lawes, M. P. Ormsby, C. G. W. Sheppard, R. Woolley, Combust. Flame 159 (5) (2012) 1949{1959. [8] A. Y. Poludnenko, E. S. Oran, Combust. Flame 157 (5) (2010) 995{1011. [9] Y. Chen, R. W. Bilger, Combust. Flame 131 (2002) (2006) 400{435. [10] F. Dinkelacker, A. Soika, D. Most, D. Hofmann, A. Leipertz, W. Polifke, K. D obbeling, Proc. Combust. Inst. 27 (1) (1998) 857{865. [11] T. M. Wabel, A. W. Skiba, J. F. Driscoll, Combust. Flame 188 (2018) 13{27. [12] R. Borghi, Plenum Pub. Corp (1985) 117. [13] N. Peters, Turbulent combustion, Cambridge university press, 2000. [14] S. G. Davis, C. Law, Combust. Sci. Technol. 140 (1-6) (1998) 427{449. [15] X. You, F. N. Egolfopoulos, H. Wang, Proc. Combust. Inst. 32 (1) (2009) 403{410. [16] A. T. Holley, X. Q. You, E. Dames, H. Wang, F. N. Egolfopoulos, Proc. Combust. Inst. 32 (1) (2009) 1157{1163. 71 [17] C. Ji, E. Dames, Y. L. Wang, H. Wang, F. N. Egolfopoulos, Combust. Flame 157 (2) (2010) 277{287. [18] J. Smolke, F. Carbone, F. N. Egolfopoulos, H. Wang, In press Combust. Flame. [19] A. Lipatnikov, J. Chomiak, Progress in Energy and Combustion Science 31 (1) (2005) 1{73. [20] D. Bradley, M. Lawes, K. Liu, R. Woolley, Proc. Combust. Inst. 31 (1) (2007) 1393 { 1400. [21] K. H. Goh, P. Geipel, F. Hampp, R. P. Lindstedt, Proc. Combust. Inst. 34 (2) (2013) 3311{ 3318. [22] C. Mouna m-Rousselle, L. Landry, F. Halter, F. Foucher, Proc. Combust. Inst. 34 (2) (2013) 2941{2949. [23] F. Carbone, J. L. Smolke, A. M. Fincham, F. N. Egolfopoulos, Combust. Flame 180 (2017) 88{101. [24] C. Coh e, C. Chauveau, I. G okalp, D. Kurtulu s, Proc. Combust. Inst. 32 (1) (2009) 180{1810. [25] H. Kobayashi, T. Nakashima, T. Tamura, K. Maruta, T. Niioka, Combust. Flame 108 (1) (1997) 104{117. [26] A. Ratzke, T. Sch oer, K. Kuppa, F. Dinkelacker, Combust. Flame 162 (1) (2015) 2778{2787. [27] H. Kobayashi, H. Kawazoe, Proc. Combust. Inst. 28 (1) (2000) 375{382. [28] H. Kobayashi, Y. Otawara, J. Wang, F. Matsuno, Y. Ogami, M. Okuyama, S. Kadowaki, Proc. Combust. Inst. 34 (1) (2013) 1437{1445. [29] H. Kobayashi, K. Seyama, H. Hagiwara, Y. Ogami, Proc. Combust. Inst. 30 (1) (2005) 827{834. [30] H. Kobayashi, T. Tamura, K. Maruta, T. Niioka, F. Williams, Proc. Combust. Inst. 26 (1) (1996) 389{396. [31] C. Law, D. Zhu, G. Yu, in: Symposium (International) on Combustion, Vol. 21, Elsevier, 1988, pp. 1419{1426. [32] L. Richardson. [33] A. N. Kolmogorov, Journal of Fluid Mechanics 13 (1) (1962) 82{85. [34] U. Niemann, K. Seshadri, F. Williams, Proc. Combust. Inst. 34 (1) (2013) 881{886. [35] T. Wabel, R. S. Barlow, A. M. Steinberg, Combust. Flame 208 (2019) 327 { 329. 72 [36] J. Smolke, S. Lapointe, L. Paxton, G. Blanquart, F. Carbone, A. M. Fincham, F. N. Egol- fopoulos, Proc. Combust. Inst. 36 (2) (2017) 1877{1884. [37] A. G. Gaydon, H. G. Wolfhard, Flames, their structure, radiation, and temperature, Halsted Press, 1979. [38] J. G. Lee, D. A. Santavicca, J. Prop. Power 19 (5) (2003) 735{750. [39] C. S. Panoutsos, Y. Hardalupas, A. Taylor, Combust. Flame 156 (2) (2009) 273 { 291. [40] T. Kathrotia, U. Riedel, A. Seipel, K. Moshammer, A. Brockhinke, Appl. Phys. B 107 (3) (2012) 571{584. [41] A. Stella, G. Guj, J. Kompenhans, M. Rael, H. Richard, Experiments in Fluids 30 (2) (2001) 167{180. [42] R. Schl uler, J. Czarske, A. Fischer, Optics and Lasers in Engineering 54 (2014) 93{104. [43] J. Smolke, Investigations of Fuel Eects on Turbulent Premixed Jet Flames, Ph.D. thesis, University of Southern California (2017). [44] A. Fincham, G. Spedding, Experiments in Fluids 23 (6) (1997) 449{462. [45] J. E. Harrington, K. C. Smyth, Chemical Physics Letters 202 (3-4) (1993) 196{202. [46] P. Paul, H. Najm 27 (1) (1998) 43{50. [47] C. S. Goldenstein, R. M. Spearrin, J. B. Jeries, R. K. Hanson, Prog. Energy Combust. Sci. 60 (2017) 132{176. [48] C. J. Dasch, Appl. Opt. 31 (1992) 1146{1152. [49] K. J. Daun, K. A. Thomson, F. Liu, G. J. Smallwood, Appl. Opt. 45 (1) (2006) 4638{4646. [50] M. Ravichandran, F. C. Gouldin, Combust. Sci. Technol. 60 (1-3) (1988) 231{248. [51] P. Nau, J. Koppmann, A. Lackner, K. Kohse-H oinghaus, A. Brockhinke, Appl. Phys. B Lasers Opt. 118 (3) (2015) 361{368. [52] R. Kee, J. Grcar, M. Smooke, J. Miller, Sandia National Laboratories. [53] F. N. Egolfopoulos, C. S. Campbell, J. Fluid Mech. 318 (1996) 1 { 29. [54] F. N. Egolfopoulos, Proc. Combust. Inst. 25 (1) (1994) 1365 { 1373. [55] F. N. Egolfopoulos, Proc. Combust. Inst. 25 (1) (1994) 1375 { 1381. 73 [56] C. Sung, C. Law, Combust. Flame 123 (3) (2000) 375 { 388. [57] T. Poinsot, D. Veynante, Theoretical and numerical combustion, RT Edwards, Inc., 2005. [58] H. Wang, X. You, A. Joshi, A. L. S. Davis, F. N. Egolfopoulos, C. K. Law. [59] H. Wang, E. Dames, B. Sirjean, D. A. Sheen, R. Tango, A. Violi, J. Y. W. Lai, F. N. Egolfopoulos, a. R. K. H. D. F. Davidson, C. T. Bowman, C. K. Law, W. Tsang, N. P. Cernansky, D. L. Miller, R. P. Lindstedt. [60] V. Nori, J. Seitzman, 46th AIAA Aerospace Sciences Meeting and Exhibit. [61] V. Nori, J. Seitzman, Proc. Combust. Inst. 32 (1) (2009) 895{903. [62] E. Petersen, M. Knopp, N. Donato, F. G uthe, ASME. J. Eng. Gas Turbines Power 134 (5). [63] A. Hossain, Y. Nakamura, Combust. Flame 161 (1) (2014) 162 { 172. [64] J. C. Hermanson, P. E. Dimotakis, J. Fluid Mech. 199 (1989) (1989) 333{375. [65] P. A. McMurtry, W. Jou, J. Riley, R. W. Metcalfe, AIAA J. 24 (6) (1986) 962{970. [66] W. J. A. Dahm, P. E. Dimotakis, AIAA J. 25 (9) (1987) 1216{1223. [67] M. W. Thring, M. P. Newby, Proc. Combust. Inst. 4 (1) (1953) 789{796. [68] F. P. Ricou, D. B. Spalding, J. Fluid Mech. 11 (01) (1961) 21. [69] K. M. Tacina, W. J. A. Dahm, J. Fluid Mech. 415 (2000) 23{44. [70] A. N. Lipatnikov, J. Chomiak, Prog. Energy Combust. Sci. 36 (1) (2010) 1{102. [71] R. K. Cheng, Combust. Sci. Technol. 41 (3-4) (1984) 109{142. [72] V. Bergmann, W. Meier, D. Wol, W. Stricker, Appl. Phys. B Lasers Opt. 66 (4) (1998) 489{502. [73] R. Cabra, J.-Y. Chen, R. Dibble, A. Karpetis, R. Barlow, Combust. Flame 143 (4) (2005) 491{506. [74] M. J. Dunn, A. R. Masri, R. W. Bilger, R. S. Barlow, Flow, Turbul. Combust. 85 (3-4) (2010) 621{648. [75] B. R. Halls, J. R. Gord, T. R. Meyer, D. J. Thul, M. Slipchenko, S. Roy, Proc. Combust. Inst. 36 (3) (2017) 4611{4618. [76] M. J. Dunn, A. R. Masri, R. W. Bilger, R. S. Barlow, G. H. Wang, Proc. Combust. Inst. 32 (2) (2009) 1779{1786. 74 [77] F. Fuest, R. S. Barlow, G. Magnotti, J. A. Sutton, Combust. Flame 188 (2018) 41{65. [78] I. Trueba Monje, J. A. Sutton, in: 2018 AIAA Aerosp. Sci. Meet., American Institute of Aeronautics and Astronautics, Reston, Virginia, 2018. [79] K. Bray, P. A. Libby, G. Masuya, J. Moss. [80] I. Shepherd, R. Cheng, T. Plessing, C. Kortschik, N. Peters, Proceedings of the Combustion Institute 29 (2) (2002) 1833{1840. [81] H. Wang, R. Xu, K. Wang, C. T. Bowman, R. K. Hanson, D. F. Davidson, K. Brezinsky, F. N. Egolfopoulos, Combust. Flame 193 (2018) 502{519. [82] M. Russi, I. Cornet, R. Cornog, Symp. (Int). Combust. 4 (1) (1953) 743 { 748. [83] F. Wright, Combust. Flame 3 (1959) 319 { 337. [84] S. J. Shanbhogue, S. Husain, T. Lieuwen, Prog. Energy Combust. Sci. 35 (1) (2009) 98 { 120. [85] S. Chaudhuri, S. Kostka, M. W. Renfro, B. M. Cetegen, Combust. Flame 157 (4) (2010) 790 { 802. [86] J. Kariuki, J. R. Dawson, E. Mastorakos, Combust. Flame 159 (8) (2012) 2589 { 2607. [87] A. Masri, R. Dibble, R. Barlow, Prog. Energy Combust. Sci. 22 (4) (1996) 307 { 362. [88] R. Barlow, J. Frank, Symp. (Int). Combust. 27 (1) (1998) 1087 { 1095. [89] M. Ihme, H. Pitsch, Combust. Flame 155 (1) (2008) 90 { 107. [90] A. Garmory, E. Mastorakos, Proc. Combust. Inst. 33 (1) (2011) 1673 { 1680. [91] V. N. Prasad, M. Juddoo, A. R. Masri, W. P. Jones, K. H. Luo, Combust. Theory Model. 17 (3) (2013) 483{503. [92] B. Coriton, J. H. Frank, A. G. Hsu, M. D. Smooke, A. Gomez, Proc. Combust. Inst. 33 (1) (2011) 1647 { 1654. [93] B. Zhou, C. Brackmann, Q. Li, Z. Wang, P. Petersson, Z. Li, M. Ald en, X. S. Bai, Combust. Flame 162 (7) (2015) 2937 { 2953. [94] H. Wang, E. R. Hawkes, J. H. Chen, Combust. Flame 180 (2017) 110 { 123. [95] A. W. Skiba, T. M. Wabel, C. D. Carter, S. D. Hammack, J. E. Temme, J. F. Driscoll, Combust. Flame 189 (2018) 407 { 432. [96] Y. Hardalupas, M. Orain, Combust. Flame 139 (3) (2004) 188{207. 75 [97] B. Higgins, M. McQuay, F. Lacas, S. Candel, Fuel 80 (11) (2011) 1583{1591. [98] J. Rehm, P. Paul, Proc. Combust. Inst. 28 (2) (2000) 1775{1782. [99] H. N. Najm, P. H. Paul, C. J. Mueller, P. S. Wycko, Combust. Flame 113 (3) (1998) 312 { 332. [100] R. Yuan, J. Kariuki, A. Dowlut, R. Balachandran, E. Mastorakos, Proc. Combust. Inst. 35 (2) (2015) 1649 { 1656. [101] A. Ratner, J. M. D. J. F. Driscoll, C. D. Carter, J. A. Mullin, Proc. Combust. Inst. 22 (2000) 245{251. [102] S. B ockle, J. Kazenwadel, D. Shin, C. Schulz, J. Wolfrum, Proc. Combust. Inst. 28 (1) (2000) 279{286. [103] J. Kariuki, A. Dowlut, R. Yuan, R. Balachandran, E. Mastorakos, Proc. Combust. Inst. 35 (2) (2015) 1443 { 1450. [104] D. E. Cavaliere, J. Kariuki, E. Mastorakos, Flow, Turbul. Combust. 91 (2) (2013) 347{372. [105] A. E. Lutz, R. J. Kee, J. F. Grcar, F. M. Rupley, OPPDIF: A Fortran program for computing Opposed-Flow diusion ames, Tech. Rep. SAND96-8243, UC-1409, Sandia (1997). [106] S. Hochgreb, F. Dryer, Combust. Flame 91 (3) (1992) 257{284. [107] V. Dias, C. Duynslaegher, F. Contino, J. Vandooren, H. Jeanmart, Combust. Flame 159 (5) (2012) 1814{1820. [108] C. K. Law, Combust. Sci. Technol. 178 (1-3) (2006) 335{360. [109] H. K. Chelliah, C. K. Law, T. Ueda, M. D. Smooke, F. A. Williams, Proc. Combust. Inst. 23 (1) (1991) 503{511. [110] Z. Nikolaou, N. Swaminathan, Combust. Flame 161 (12) (2014) 3073{3084. 76
Abstract (if available)
Abstract
Combustion is the world's primary energy conversion method and is responsible for powering commercial and military aviation. Combustion research can help identify the physical and chemical processes that are responsible for the conversion of reactants to final products, which will enable improved efficiency and safety. Turbulent jet flames are a canonical way to study the complex turbulent and chemical environments seen in real combustors. This dissertation will discuss the experimental approach for studying such flames and discuss the recent advances made in understanding these complex phenomena.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Investigations of fuel effects on turbulent premixed jet flames
PDF
Flame characteristics in quasi-2D channels: stability, rates and scaling
PDF
CFD design of jet-stirred chambers for turbulent flame and chemical kinetics experiments
PDF
Accuracy and feasibility of combustion studies under engine relevant conditions
PDF
Flame ignition studies of conventional and alternative jet fuels and surrogate components
PDF
Direct numerical simulation of mixing and combustion under canonical shock turbulence interaction
PDF
Studies on the flame dynamics and kinetics of alcohols and liquid hydrocarbon fuels
PDF
Studies of methane counterflow flames at low pressures
PDF
Re-assessing local structures of turbulent flames via vortex-flame interaction
PDF
End-gas autoignition investigations using confined spherically expanding flames
PDF
Model based design of porous and patterned surfaces for passive turbulence control
PDF
Modeling and analysis of parallel and spatially-evolving wall-bounded shear flows
PDF
Modeling investigations of fundamental combustion phenomena in low-dimensional configurations
PDF
Pressure effects on C₁-C₂ hydrocarbon laminar flames
PDF
Experimental and kinetic modeling studies of flames of H₂, CO, and C₁-C₄ hydrocarbons
PDF
Studies of combustion characteristics of heavy hydrocarbons in simple and complex flows
PDF
Experimental studies of high pressure combustion using spherically expanding flames
PDF
Experimental investigation of the propagation and extinction of edge-flames
PDF
Experimental investigation of soot nucleation and growth in premixed flat flames
PDF
Developing tuned anisotropic porous substrates for passive turbulence control
Asset Metadata
Creator
Paxton, Laurel
(author)
Core Title
Investigations of fuel and hydrodynamic effects in highly turbulent premixed jet flames
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Mechanical Engineering
Publication Date
12/03/2019
Defense Date
10/17/2019
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
chemiluminescence,counterflow flames,fuel effects,OAI-PMH Harvest,shear layers,turbulent flames
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Egolfopolous, Fokion (
committee chair
), Bermejo-Moreno, Ivan (
committee member
), Ronney, Paul (
committee member
), Shing, Katherine (
committee member
), Spedding, Geoff (
committee member
)
Creator Email
lpaxton@usc.edu,ltpaxton0@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-243801
Unique identifier
UC11673983
Identifier
etd-PaxtonLaur-7975.pdf (filename),usctheses-c89-243801 (legacy record id)
Legacy Identifier
etd-PaxtonLaur-7975.pdf
Dmrecord
243801
Document Type
Dissertation
Rights
Paxton, Laurel
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
chemiluminescence
counterflow flames
fuel effects
shear layers
turbulent flames